Interlanguage

Language Learning & Language Teaching (LL&LT)

The LL&LT monograph series publishes monographs, edited volumes and
text books on applied and methodological issues in the field of language
pedagogy. The focus of the series is on subjects such as classroom discourse
and interaction; language diversity in educational settings; bilingual
education; language testing and language assessment; teaching methods and
teaching performance; learning trajectories in second language acquisition;
and written language learning in educational settings.
For an overview of all books published in this series, please see
http://benjamins.com/catalog/lllt

Editors
Nina Spada

Ontario Institute for Studies in Education

University of Toronto

Nelleke Van Deusen-Scholl

Center for Language Study
Yale University

Volume 39
Interlanguage. Forty years later
Edited by ZhaoHong Han and Elaine Tarone

Interlanguage
Forty years later
Edited by

ZhaoHong Han
Columbia University

Elaine Tarone
University of Minnesota

John Benjamins Publishing Company

Amsterdam/Philadelphia

TM

The paper used in this publication meets the minimum requirements of

theAmerican National Standard for Information Sciences Permanence
of Paper for Printed Library Materials, ansi z39.48-1984.

Library of Congress Cataloging-in-Publication Data

Interlanguage : forty years later / Edited by ZhaoHong Han and Elaine Tarone.
p. cm. (Language Learning & Language Teaching, issn 1569-9471 ; v. 39)
Includes bibliographical references and index.
1. Interlanguage (Language learning) 2. Second language acquisition. I. Han, Zhaohong,
1962- editor of compilation. II. Tarone, Elaine, 1945- editor of compilation.
P118.2.I575 2014
418.0071--dc23
2013050659
isbn 978 90 272 1319 8 (Hb ; alk. paper)
isbn 978 90 272 1320 4 (Pb ; alk. paper)
isbn 978 90 272 7049 8 (Eb)

2014 John Benjamins B.V.

No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any
other means, without written permission from the publisher.
John Benjamins Publishing Co. P.O. Box 36224 1020 me Amsterdam The Netherlands
John Benjamins North America P.O. Box 27519 Philadelphia pa 19118-0519 usa

Table of contents
Acknowledgements
Introduction
ZhaoHong Han and Elaine Tarone
chapter 1
Enduring questions from the Interlanguage Hypothesis
Elaine Tarone

vii
1

chapter 2
Rediscovering prediction
Terence Odlin

27

chapter 3
From Julie to Wes to Alberto: Revisiting the construct of fossilization
ZhaoHong Han

47

chapter 4
Interlanguage, transfer and fossilization: Beyond second language
acquisition
Silvina Montrul

75

chapter 5
The limits of instruction: 40 years after Interlanguage
Bill VanPatten

105

chapter 6
Documenting interlanguage development
Kathleen Bardovi-Harlig

127

chapter 7
Methodological influences of Interlanguage (1972): Data then
and data now
Susan Gass and Charlene Polio

147

Interlanguage Forty Years Later

chapter 8
Trying out theories on interlanguage: Description and explanation
over 40 years of L2 negation research
Lourdes Ortega
chapter 9
Another step to be taken Rethinking the end point of the interlanguage
continuum
Diane Larsen-Freeman

173

203

chapter 10
Interlanguage 40 years on: Three themes from here
Larry Selinker

221

Name index
Subject index

247
253

Acknowledgements
This book itself an acknowledgement of the impact of Larry Selinkers seminal
1972 paper Interlanguage would not have been possible without the most enthusiastic and thoughtful participation of its contributors. To them we owe a
heartfelt thank you. All the chapters herein underwent a rigorous review process,
and here we would particularly like to thank our external reviewers for their critical assistance. In alphabetical order, they are: Robert Bley-Vroman, Rod Ellis, Jan
Hulstijn, Scott Jarvis, Carol Klee, Patsy Lightbown, Leila Ranta, Peter Robinson,
Rafael Salaberry, and John Schumann. Equally, we want to express our appreciation to the organizers of The 2nd TCCRISLS (http://www.tc.columbia.edu/
tccrisls/), in particular: Timothy Hall, Farah Akbar, Waiman Adrienne Lew,
Hiromi Noguchi, Mi Sun Park, Ji-Yung Jung, Eun-Young Kang, and Cheng-ling
Alice Chen. It is this forum that provided the inspiration and much of the content
of this book. Eun Young Kang deserves an additional word of gratitude for her
bibliographic assistance. Lastly, our appreciation goes to Kees Vaes, Nina Spada,
and Nelleke Van Deusen-Scholl at John Benjamins who have kindly given us valuable assistance in the production of this volume.
ZhaoHong Han and Elaine Tarone

Introduction
ZhaoHong Han and Elaine Tarone

Few works in our field second language acquisition (SLA) can endure multiple reads, but Selinkers (1972) Interlanguage is a clear exception. Written at
the inception of the field, this paper delineates a disciplinary scope, asks penetrating questions, advances daring hypotheses, and proposes a first-ever coherent conceptual and empirical framework that inspired, and has continued to
inspire, SLA research.
Sparked by a heightened interest in this founding text on its 40th anniversary,
in 2012 a group of scholars convened at The 2nd Teachers College, Columbia
University Roundtable in Second Language Studies (TCCRISLS) to examine and
contemplate extrapolations of the seminal text for the past, the present, and the
future of SLA research. The papers delivered at that forum, written out, reviewed,
and then revised, are contained in the present volume.
This book is as much about history as it is about the present and the future of
the field of SLA. The themes covered herein run the gamut from the theoretical to
the empirical to the pedagogical. In ten chapters, scholars confirm, expand, and/or
challenge the original conceptions articulated in Selinker (1972). Each chapter offers a prism through which one can view the evolution both of a specific domain
and of the field writ large.
Chapter 1, by Elaine Tarone (Distinguished Teaching Professor, University of
Minnesota, and author [with Bonnie Swierzbin] of Exploring Learner Language
[2009, Oxford University Press]), provides a historical context for the Interlanguage paper. Recalling her perspective as a student of Selinkers, first at the
University of Edinburgh in 1968 and later at the University of Washington in the
early 1970s, Tarone deftly traces conceptual origins, explicates and highlights enduring themes, and explores influences on the 1972 paper. Many of the enduring
themes she identifies are explored in depth in the ensuing chapters. Tarone contends that the best way to read the Interlanguage paper is as a series of foundational questions and possible answers, rather than as a proposed theory. These
foundational questions include: What if the language produced by adult L2 learners has its own underlying system? What if adult interlanguages inevitably fossilize

ZhaoHong Han and Elaine Tarone

(stabilize short of full identity with a target language)? She poses two of her own
questions for future research: How do learners with little or no alphabetic literacy
develop their interlanguages? And how do sociolinguistic forces within the speech
community affect an individuals interlanguage (IL) development?
One prominent theme in the Interlanguage paper is language transfer,
which Selinker suggests is one of five central cognitive processes making up a
latent psychological structure, largely responsible for the variable outcome of
adult second language acquisition. In Chapter 2, Terence Odlin (Professor, Ohio
State University, and author of Language Transfer [1989, Cambridge University
Press]) picks up on Selinkers assertion of a need to predict IL development, especially in relation to language transfer. Reflecting on recent studies that seemingly
enable predictions to be made for SLA in novel contexts, Odlin observes
that Selinkers Latent Psychological Structure often takes meaning-based concerns into account in trying to create interlingual identifications between the
grammatical systems of source and target languages. Odlin advances five sets of
testable hypotheses for future research, cautioning, nevertheless, that group tendencies often seem predictable but individual behavior is usually much less so,
particularly in light of three factors: multilingualism, idiosyncrasy in forms, and
idiosyncrasy in meanings.
A considerable portion of the Interlanguage paper is dedicated to proposing and presenting arguments for the construct of fossilization, framed as a
mechanism as well as a behavioral artifact of second language (L2) acquisition.
Chapter 3, by ZhaoHong Han (Professor, Teachers College, Columbia University,
and author of Fossilization in Adult Second Language Acquisition [2004, Multilingual Matters]), revisits the construct, in light of two widely observed phenomena
of SLA: inter-learner and intra-learner variable success. Han claims, inter alia,
that fossilization occurs in juxtaposition with acquisition, underscoring that
fossilization research is less about revealing deviances from the presumed
norm than about addressing a dual cognitive problem: Why is it that in spite of
propitious conditions, development is cut short in some areas? And why is the
developmental interruption made most apparent when learners attempt self-expressions in the target language?
The issues of transfer and fossilization, both inter-related, are also the focus
in Chapter 4 by Silvina Montrul (Professor, University of Illinois at Urbana-
Champaign, and Editor of Second Language Research), but this time, creatively in
the context of bilingualism and multilingualism. Montrul expounds on the impact
of the Interlanguage Hypothesis on contemporary SLA research within a Universal Grammar framework and discusses how this research, in turn, helps refine and
elucidate much of Selinkers (1972) original proposal. The author argues that interlanguage constructs such as transfer, simplification, and fossilization are neither

Introduction

unique to SLA nor defective processes, behaviors, or states of knowledge; rather,

these are natural characteristics of any type of bilingualism and multilingualism,
found in both individuals (e.g., heritage speakers acquisition of their first language) and communities (e.g., Spanish immigrants in the U.S.), and can even be
drivers of language change.
Selinker (1972) argued that successful acquisition (i.e., achieving native-like
competence) cannot come about through instruction. In Chapter 5, Bill VanPatten
(Professor, Michigan State University, and author of Input Processing and Grammar Instruction [1996, Ablex]) supports Selinkers position, contending that formal instruction focused on explicit rules has a limited role to play in shaping L2
learners underlying competence (their mental representation of language). Rather, growth in competence results from complex interplays among input, language
specific mechanisms, and input processing mechanisms. Reviewing research on
staged development (including U-shaped behavior) and on poverty of the stimulus, VanPatten argues that there is no evidence that explicit rule instruction (in
contrast to input processing) affects the L2 learners underlying representation of
language at any stage, and that interlanguages are constructed independently of
rule instruction all along the way.
Following Selinkers principle that learners should be observed while engaging
in meaningful performance in a second language (1972: 210), in Chapter 6
Kathleen Bardovi-Harlig (Professor, Indiana University at Bloomington, and author of Tense and Aspect in SLA [2000, Blackwell]) uses a functional approach to
interlanguage analysis that investigates form and meaning relationships, and the
way these change over time in the developing interlanguage. She identifies two
ways to track these relationships: 1) A form-to-function analysis tracks a linguistic
form (e.g., morpheme, or tense-aspect form) and its changing meanings, and
2) a function-to form, or concept-oriented analysis tracks a function or meaning
(e.g. expression of past time), and the changing linguistic forms used to express it.
Taking a primarily concept-oriented approach, Bardovi-Harlig explores (a) the expression of temporality, (b) pragmatics, and (c) conventional expressions, showing
that interlanguage exhibits both systematicity and innovation in all three areas.
Selinkers (1972) principle that the best data for study of interlanguage is that
produced during meaningful performance in a second language is contested in
Chapter 7 by Susan Gass (University Distinguished Professor, Michigan State
University, and author [with Alison Mackey] of the Handbook of Second Language
Acquisition [2012, Routledge]), and Charlene Polio (Professor, Michigan State
University, and Associate Editor of Modern Language Journal). Gass and Polio
provide a historical and dynamic perspective on the question of what counts
as relevant data, noting alternative views to Selinkers (e.g. Corder, 1973). The
authors make the case for multiple data sources in the study of interlanguage,

ZhaoHong Han and Elaine Tarone

arguing that data that were disallowed in the 1970s should not be in todays research environment. Reviewing research studies that used grammaticality judgments and artificial language both before and after Selinker (1972), they show the
value of such data types in exploring theoretical constructs, such as indeterminacy
and ultimate attainment, which cannot be studied solely using the type of data
Selinker argued for 40 years ago.
In Chapter 8, Lourdes Ortega (Professor, Georgetown University, and Editor of
Language Learning) explores and questions Selinkers (1972) admonition to pursue
description prior to explanation in second language study. Through a chronicle of
SLA studies on negation from a variety of theoretical perspectives, Ortega frames
the relationship between description and explanation as a function of the theoretical or methodological approach adopted by the given study. Building on her historical review, Ortega calls for a rethinking of several topics: the definition and
scope of research on variation, native language versus universal influences on interlanguage, and what she terms an ambivalent relationship in the field among
constructs like accuracy, development, and native speaker competence.
Chapter 9, by Diane Larsen-Freeman (Professor, University of Michigan, and
author [with Lynne Cameron] of Complex Systems and Applied Linguistics [2008,
Oxford University Press]), challenges the idea of defining successful learning as
the reorganization of linguistic material from an IL to identity with a particular
TL (Selinker, 1972, p. 22). She contests the assumption of such a normative, telic
endpoint, arguing that if interlanguage is a complex adaptive system, there is no
endpoint to its development; it is an open system that is always changing, with
no goal to its development. The author acknowledges that there are conflicts between this view and the unavoidably normative aim of pedagogy and language
assessment. She ends her chapter by exploring activities that teachers can engage
with to reconcile the non-telic nature of language learning with the normative
demands of teaching.
Finally, coming full circle, in Chapter 10 Larry Selinker (Professor, New York
University) explores and exemplifies three major themes related to the Interlanguage paper: its place in the intellectual history of scholars studying second language acquisition; the present implications of viewing interlanguage as a linguistic
system in its own right; and the need in future scholarship for a deep interlanguage semantics. Selinker ends his chapter with advice for young scholars of interlanguage: He urges them (and all of us) to cultivate an active and independent
intellectual curiosity by constantly testing, or doubting currently accepted orthodoxies in interlanguage scholarship, even those he himself propounds. He
stresses the value of finding good questions, and (in keeping with Talmudic logic)
the importance of not ever discarding the full range of possible arguments and
answers to the questions one asks.

Introduction

As this book amply shows, over a span of four decades the field of SLA has
evolved into a complex web of competing conceptions and practices. Yet, a strong
consensus can be seen across the chapters in support of the foundational postulate
in the Interlanguage paper, namely that learners build an independent linguistic
system driven by a powerful internal, interfacing mechanism that, in turn, interacts with the environment in which learning occurs. Where researchers appear to
part company is in the vantage points they choose for their pursuit of the common
goal of understanding that system. The differences cannot be understated; in keeping with Selinkers advice (this volume) to cultivate open-mindedness, they exert
an influence over fundamental questions such as: What constitutes SLA research?
What should this research attempt to describe and explain? Questions that have
emerged from within this volume are many, out of which ten practical ones stand
out for us as editors:
1. What goals do second language learners have for their developing linguistic
systems?
2. How should interlanguage research take any such goals into account?
3. Should native speakers of the target language provide baseline data in L2 research? From which speech communities should they be sampled, and why?
4. Should interlanguage development be judged in terms of success or failure?
5. Does explicit rule-based instruction facilitate, hinder, or fail to affect interlanguage development?
6. How does (lack of) alphabetic print literacy affect processes of interlanguage
development?
7. How do sociolinguistic forces within the learners speech community variably
affect interlanguage development?
8. Should SLA researchers borrow theories from other disciplines, or build their
own?
9. Is it possible or desirable to have a uniform theory of SLA, and agreed-upon
definitions of acquisition? Why?
10. Should SLA research benefit from a conceptual separation between the theoretical and the applied?
We would like to leave these questions for the reader to contemplate while considering at his or her own pace the insights and perspectives provided by each
chapter.
This book offers a resource to novices and experts alike in and beyond the field
of SLA. By virtue of revisiting the Interlanguage paper, a founding text, in the
company of 10 current scholars, the book sets a precedent for helping advance
disciplinary inquiry through identifying and mining its scholarly heritage and
legacy. It is our hope that this exercise will be continued as the field advances.

ZhaoHong Han and Elaine Tarone

Foto 1. Speakers at the Interlanguage 40 Years Later Symposium

Foto 2. Elaine Tarone, Susan Gass, Miriam Eisenstein, Larry Selinker, Jan Hulstijn

chapter 1

Enduring questions from the Interlanguage

Hypothesis
Elaine Tarone

University of Minnesota
This chapter claims that the Interlanguage Hypothesis is best understood, not
as a theory of second language acquisition (SLA), but as a set of questions that
motivate divergent answers and research programs. Selinkers (1972) basic
question is whether there is a linguistic system that underlies the output of
second language learners. Related questions that continue to stimulate research
focus on: the relationship between first and second language acquisition,
whether and how the linguistic systems formed in SLA fossilize, and whether
and how learners interlanguage use varies in different social situations. The
chapter also considers related questions not addressed in Selinker (1972): the
impact of alphabetic print literacy on interlanguage development, and whether
interlanguages are features of speech communities.

Introduction
In the Fall of 1968, as a graduate student in Edinburgh Universitys Applied
Linguistics Department, I was privileged to witness the early stages of creation of
the Interlanguage Hypothesis. As a Berkeley grad who had been transformed by
the civil rights movement, Id not been particularly successful in efforts to teach
standard English to high school dropouts in an Urban League Street Academy in
Harlem, and so I had set out to learn more about applied linguistics. That quest led
me in the Fall of 1968 to be the only American in a cohort of 30 graduate students
in Edinburgh, all of us studying applied linguistics to become better English as a
Second Language (or English as a Second Dialect) professionals. The only other
American in the program was a professor on Fulbright leave from the University
of Washington: Larry Selinker.
As it turned out, I had left a social revolution in the States and walked into
an intellectual revolution in the field of applied linguistics. Larry Selinker was
in Edinburgh to work on a new construct he called interlanguage along with

Elaine Tarone

Prof S.P. Corder, who had proposed something related called transitional competence (Corder, 1967). Together these ideas were going to revolutionize our understanding of the way adults learn second languages, and free second-language pedagogy from the chains of behaviorism, aligning it more with Chomskys (then-new)
theory of the way first languages are acquired. The central hypothesis of their work
was that learner language was not a hodgepodge of errors all caused by the learners native language (NL), but an autonomous linguistic system in its own right
that evolved according to innate and probably universal processes.
Selinker had recently completed his Ph.D. at Georgetown under the direction
of Robert Lado, who was still defending his (1957) behaviorist claim that all second-language (L2) learning difficulties were the result of the transfer of patterns
from one linguistic system (the native language, or NL) to another (the target language (TL, or L21). According to Lado, a contrastive analysis of linguistic differences between these two linguistic systems, the NL and the TL, could predict all
such learning difficulties and all errors. But Selinker s dissertation, designed to
test Lados claim, had found that many errors in the language produced by learners
of Hebrew could not in fact be explained by transfer from English NL. In other
words, the behaviorist explanation for second-language acquisition did not account for all learner language data.
The evolving consensus in discussions at Edinburgh 196869 was that in the
process of second-language acquisition, there were not two, but three linguistic
systems of interest: native language, target language, and learner language. This
third system, called a transitional competence by Corder, and interlanguage by
Selinker, should be considered a linguistic system in its own right, and the object
of as much serious linguistic study as the NL and the TL systems. Cognitive
processes other than transfer might be at work in shaping this third linguistic
system; researchers should be able to describe that interlanguage (IL), and
through that research, the underlying cognitive processes that created that system. Almost no research of this type existed in 196869. At the end of the year,
I was accepted into a graduate program in applied linguistics at the University of
Washington in Seattle, where during the next three years, Larry Selinker continued to work on drafts of a publication (the interlanguage paper) that was ultimately to be published in 1972. He continued to discuss the key concepts of
interlanguage at length. In his graduate courses, as he led discussions he would
read from and add to the manuscript he was drafting, presenting one or another
of the themes he was working on and inviting questions: transfer (what other
1. In this chapter, the terms target language (TL) and second language (L2) are used interchangeably to refer to any language that is the target of acquisition after the learner has acquired
a native language.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

cognitive processes might there be?), fossilization (surely not every learners language fossilized! Did the learner language systems of children stop developing
the way adults seemed to?) interlingual identifications (how does the learner
understand the 3 linguistic systems she is struggling with?), the data to be used
to research IL (do learners grammaticality judgments accurately access the rules
of interlanguage? Why or why not?). In class, I was privileged to join Larrys
students2 in discussing these questions, asking our own, arguing over possible
answers. Now and then Larry would scribble a student comment on a little torn
off piece of paper and tape that scrap to the growing collage, turning the whole
thing sidewise, or upside down, to read his draft to us. Later the collage would be
tidied up and published as Selinker (1972), the interlanguage paper, and those
scraps of paper would become footnotes framed as personal communications
from named students3.
In other words, the interlanguage paper was not so much written as it was
gradually assembled over a period of several years as a set of possible hypotheses and questions, emerging gradually out of extensive co-construction and collaboration, arguments and debates, between Larry and anyone he could get to
join the discussion. And although the resulting publication, Selinker (1972),
may not be a tightly organized piece of prose, I hope to show that the themes
that were identified in that paper as a result of that process have been enduring
and lively ones that have continued to resurface for the last 40 years in research
on second-language acquisition. I would argue that the essential role of the interlanguage paper in our field has been to identify core issues and questions to
interest, stimulate or goad researchers to go out to gather empirical data, to try
to find out what adult language learners were really doing when they learned
another language.
The interlanguage (IL) hypothesis is not really a theory, I would argue, if we
think of a theory as a proposal that converges upon a single set of answers to some
question. Rather, the IL hypothesis asserts a foundational question, namely: What
if learner language is a linguistic system? From this implied question, more questions arise, as well as different possible answers to those questions. In this way, we
can understand the IL hypothesis as a proposal asking us to think as widely as possible about what the nature of a possible IL linguistic system might be, and how we
can gain insight into that by gathering empirical data. Conceived of as questions,

2. Among others: Chris Adjmian, Hal Edwards, Tom Huckin, and Robert Vroman (later
Bley-Vroman).
3. Few scholars today are as conscientious about documenting and acknowledging their
students contributions to their thinking.

Elaine Tarone

not answers4, the themes advanced in Selinker (1972) have served to stimulate a
respectable amount of research and theory-building.
Many of the original themes in the IL paper can certainly be recognized as
related to some of the answers proposed in the various theories of second-language
acquisition that have been proposed and have generated research since 1972. But
while those various SLA theories have risen or fallen in popularity over that period, the central questions of the interlanguage paper, generating different possible
answers, seem to continue to generate interest, research, and theory-building.
Those enduring themes were explored in some depth in discussions at the Teachers College, Columbia University Interlanguage Symposium of 2012 and are reflected in the different chapters in this volume, in light of some of the research data
and theoretical ruminations they have generated. In the present chapter, I provide
a brief overview of five interrelated themes three themes explicitly laid out in the
original interlanguage paper, and two that have surfaced in subsequent exploration of the Interlanguage Hypothesis. In keeping with their origins in Selinker
(1972), I have presented each theme as a question rather than an answer.
Selected, interrelated themes originating in Selinker (1972)
1. What if adult L2 acquisition is different from child L1 acquisition? How,
and why?
As noted earlier, the IL hypothesis began to develop very soon after Chomsky
(1965) had proposed a powerful theory about the way children learn to speak their
native language: born with an innate knowledge of language (UG) that engages
innate acquisition processes, children are (with rare exceptions) completely successful in acquiring the ability to speak their native language. One of the central
questions raised by Selinker (1972) suggests that the processes of adult second
language acquisition must be different from the child L1 acquisition described by
Chomsky. This is because the outcomes of child first language acquisition and
adult second language acquisition are dramatically different: very few adults acquire a second language perfectly, an outcome that children seem to effortlessly
and universally achieve in acquiring L1.
The IL hypothesis postulates that one major reason for this could be that
adults, unlike children, do not have access to Universal Grammar (UG)5, and so
they resort to other kinds of cognitive processes to internalize a TL. Selinker (1972)
4. As Selinker (this volume) makes clear, this focus on good questions together with the variety of possible answers to them, has deep roots in the Talmudic tradition.
5.

It was then referred to as the Language Acquisition Device, or LAD.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

referred to this set of general cognitive processes as a latent psychological structure (LPS). But another major reason could also be that adult L2 learners, by definition, already have acquired one language, and that early-learned language system could influence, or interfere with, the formation of subsequent language
systems (see Odlin, this volume). In other words, both LPS and NL transfer6 are
cognitive processes engaged in adult SLA that are absent in child first language
acquisition (see Selinker & Lakshmanan, 1992).
An early debate about IL that followed from these possibilities had to do with
whether IL is a natural language, one that is acquired by processes of UG, and
whose constructions follow language universals. Corder (1967) felt that adult L2
learners were capable of beginning all over again with UG to acquire L2s de novo,
engaging exactly the same universal processes used in acquiring their L1s. Selinker
(1972) suggested a different possibility: that L2 learners could be beginning the
acquisition process not with UG but with their LPS. If so, IL was not a natural
language and its constructions did not have to follow language universals. Adjmian
(1976) proposed a compromise position, arguing that ILs might be natural languages, but ones whose grammars possessed a distinctive characteristic. He proposed that IL grammars were peculiar in being permeable; this permeability was
what allowed learners to incorporate forms from their native language and to generalize or otherwise distort target language constructions when communicating.
Adjmian argued that because ILs possess this unique characteristic of permeability, learner judgments of the grammaticality of IL utterances could not be trusted,
and needed to be supplemented with large amounts of specific data from learner
performance. An ongoing question related to UG and IL is whether adults have
full, partial, or no access at all to UG in second language acquisition. This debate
obviously has continued over the subsequent decades in a range of different ways
Is there a critical period for language acquisition, after which adults have no access
at all to UG (see Birdsong, 1999, 2006; Hakuta et al., 2003)? Or, in acquiring a
language after the first, do adults have full, partial or no access to UG (see Montrul,
this volume)? And do the circumstances affect degree of access (Eckman, Bell
& Nelson, 1984; Eubank, 1991; Gass, 1989; Schachter, 1996; Sharwood Smith
& Truscott, 2005; White, 2003)?
As Selinker (1972) focused on similarities and differences between adult L2
acquisition and child L1 acquisition, it also set up a dialectic between innatist UG
explanations and cognitivist ones. That dialectic has been a deep and productive
one. Are universal, language-specific processes of the sort specified in UG required
6. An anonymous reviewer notes that UG access and NL transfer were considered related in
early debates about UG in SLA, and that the so-called no-access and indirect access positions
reflect this thinking.

Elaine Tarone

to account for adult SLA, or are general cognitive strategies adequate to explain
adult SLA? While research on SLA that relies on UG assumptions is ongoing, many
SLA researchers7 now believe that emergentist cognitive mechanisms are sufficient
to account for SLA. As Ellis (1998: 657) puts it:
... simple learning mechanisms, operating in and across the human systems for
perception, motor-action and cognition as they are exposed to language data as
part of a communicatively-rich human social environment by an organism eager
to exploit the functionality of language suffice to drive the emergence of complex
language representations.

A last question related to the issue of age of second language acquisition is this:
Selinker (1972) restricted the IL hypothesis to learners who are adults, assuming
that children are able to access UG to acquire a second language. But does the IL
hypothesis sometimes apply to children as well? And if so, under what circumstances? Within 3 years of the publication of the IL paper, Selinker, Swain and
Dumas (1975) published data gathered from older children in French language
immersion programs whose learner language seemed to fossilize in much the
same way that adults did. What could have caused this? Did these children no
longer have access to UG? Or, is their fossilization less a result of a cognitive
critical period and more a result of the immersion context in which they are
learning, where they are seldom pushed to produce the TL? In immersion classrooms, Swain (1995) argued, TL input alone was not enough: the children needed
to be pushed to produce output in order for successful second language acquisition to occur.
2. What if adult ILs inevitably fossilize at some point short of the learners goal?
The first theme, age of acquisition, is closely interrelated with the second: fossilization. Is the end state of an IL rule system inevitably different from the TL
rule system? Selinker (1972) posited that it is that some constructions in the
ILs of adults virtually always fossilize, or stop developing somewhere short of
the learners TL goal. Fossilization may be the most enduring characteristic of
interlanguage. And, according to Selinker, fossilization is related to backsliding,
which is: the regular reappearance in second language performance of linguistic phenomena which were thought to be eradicated in the performance of the
learner (1972, p. 211). Selinker (1972) proposes fossilization as a kind of defining characteristic of adult interlanguage: at least some adult interlanguage constructions always fossilize, while this does not seem to occur in child language.
As the 1972 paper was being written, this claim was one of the most hotly
7. See, for example, Robinson and Ellis (2008) for a research compilation, and MacWhinney
(2008, 2012) for a proposed unified model.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

debated. For one thing, was fossilization supposed to be categorical? Did 100%
of all adult learner ILs contain fossilized constructions? Were there no exceptions? Everyone seemed to be acquainted with a possible exception, though no
one produced an actual empirical example. Under pressure, Selinker conceded
that it might be possible for a very few8 adults to have ILs with no fossilized
constructions, but his crucial point was that the fossilization question really was
an empirical question. Research was needed to find out whether the idea of IL
fossilization had empirical support. Research questions multiplied: Do adult IL
systems always contain fossilized constructions? If not always, how often? What
might be the characteristics of the good language learner with no evidence of
fossilization? Was fossilization selective in its impact across all the linguistic
levels, and all linguistic units? If just some levels or units fossilized, what made
these more fossilizable than others? These and other questions were later explored in depth in Selinker (1992), Selinker and Lakshmanan (1992) and of
course Han (2004, this volume) and Han and Odlin (2006).
Of course, once the fossilization hypothesis is framed as an empirical matter,
to test the hypothesis that fossilization occurs at some point short of the target
goal, we must have a pretty clear idea of what the goal of IL development is. Is the
goal of IL development identical with the competence of a NS of the TL? Should
we define that goal by collecting the utterances of monolingual native speakers of
the TL, produced as they focus on meaning? Do we ask monolingual native speakers of the TL for their grammaticality judgments? Or should we ask the adult L2
learner what he or she thinks that goal is? (see Gass and Polio, this volume) Isnt
what matters for learning what the learner thinks the TL goal is? In his 1972 paper,
Selinker gives us two contradictory answers to the question of how a researcher
should empirically identify the goal of IL development. First, in describing the
three linguistic systems the learner is dealing with (NL, IL and TL), he says the
only possible data to be used to characterize each of these three systems is language produced by speakers of each one so in the case of the TL, utterances
produced by NSs of the TL. The measure of a learners success, in this view, would
be productive performance in the TL by the second language learner which is
identical to that produced by the native speaker of that TL (Selinker, 1972: 223).
In this view, the learners target, or successful end state grammar, is in fact identical
with that of a NS of the TL.
But Selinker also offers a second contradictory answer to this question on
the very next page. There he characterizes this ideal notion of the TL as identical
8. Generally few want to believe that fossilization is categorical and inevitable. Selinkers estimate that 5% of the adult population might be capable of completely successful L2 acquisition
was just that: an estimate, born in the debate, and in need of empirical testing.

Elaine Tarone

to the grammar of a NS as a teaching perspective, not a learning perspective. In

a learning perspective, we are interested in where the learner will tend to end
up, not where we would like him to end up (Selinker, 1972: 224). From the
learning perspective, it is the learner who chooses the target, and this target is
always one norm of one dialect within the interlingual focus of attention of the
learner (p. 213). In other words, a second answer to the question is that what
matters in defining the goal of acquisition is the psychological reality of the
learners perception of the TL norm in relationship to his NL norm. It is the
learners perception that creates interlingual identifications across NL, IL and
TL linguistic systems. The idea of interlingual focus of attention and of interlingual identifications in the mind of the learner suggests that the target norm
that matters most is the learners perception of that norm... and not the TL norm
as performed by native speakers of the TL (see Larsen-Freeman, this volume).
So we are left with two possible, but conflicting answers to the question of defining the goal of IL development. This way of conceptualizing the teleological goal
of IL development was taken up later by Bley-Vroman (1983), and presages
Cooks (1991, 1992, 1996) construct of multi-competence9 in the mind of the
bilingual, where NL and TL norms cannot be assumed to be the same as a monolinguals norms for the NL and TL. And crucially, we really cannot empirically
test the reality of fossilization without opting for one or another conceptualization of the goal of IL development, and the kind of data we need to characterize
that goal.
3. What if the language a learner produces when focused on meaning is quite
different from that produced when s/he is focused on formal accuracy?
This third theme is related to the first two. The idea that there are two kinds of
language knowledge at work in the mind of the adult L2 learner derives from
Selinkers (1972) extensive discussion of the kind of data that should be used to
study interlanguage. More than once, Selinker argues that the best data to use to
infer the structure of IL are utterances (implied to be oral utterances) that the
learner produces when focused on meaning. We might say that these utterances
are the product of the learners competence or knowledge system. Data specifically ruled out in Selinker (1972) as being of doubtful relevance to a theory of
second language learning are data produced when learners are focused on form,
such as the learners performance of drills in the classroom, repetition of nonsense
syllables in experiments (1972: 210), or grammaticality judgments. Utterances
produced by the learner in doing grammar drills or repetitions, or grammaticality
judgments produced by the learner, will only reveal the learners conscious
9. I am grateful to an anonymous reviewer for pointing out this parallel.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

perception of the structure of the TL10. Such activities do not access the implicit
underlying IL system; they access the learners explicit knowledge of TL form.
Selinker (1972: 214) construes IL to be the implicit systematic linguistic system
that produces learner language, and not so much the explicit system of learned
grammar rules a learner turns to when focused on form (VanPatten, this volume,
argues that formal explicit instruction is unrelated to the implicit IL system).
Of course, this proposal that there are two L2 knowledge systems, one conscious and one unconscious, has turned out to be a very enduring theme, one that
has regularly resurfaced. It is embodied in Krashens (1976, 1981, 1982) Monitor
Model as the difference between acquired and learned knowledge of L2; and Ellis
(2002, 2006) refers to these as the learners implicit and explicit L2 knowledge. This
theme may be one of the least controversial in the field of SLA: it is widely believed
that there are two versions of L2 knowledge in the mind of the learner, one implicit and one explicit.
Whether, and how these two knowledge bases might be related to one another
is more controversial than whether the two exist at all. A number of troubling
questions arise from the statement that there are two knowledge systems. For example: How do we explain the differences between implicit and explicit knowledge
in the same individual? Are the knowledge bases related, and if so how? Is transfer
from one to the other possible (i.e., is there an interface between the two systems)?
Do we adopt, with Krashen (1985) and VanPatten (this volume), a non-interface
position, believing that L2 acquisition has to be unconscious, and that no amount
of conscious knowledge about the grammar of the language will affect that unconscious acquisition process? Or do we adopt, with Schmidt (2001), an interface
position, asserting that learners have to consciously notice a new form or construction in order to internalize it, and incorporate it into their implicit knowledge
system? At present, I dont think either position has actually been supported
(though VanPatten, this volume, might argue otherwise), since we do not have
direct access into the minds of L2 learners at a level that can differentiate conscious
and unconscious knowledge bases, and the movements of language items or constructions between them.11

10. We have seen that Adjmian (1976) also described the doubtful utility of learner grammaticality judgments about IL because of what he called the unique permeability of the IL
system to invasion by both NL rules and TL rules as explicitly presented in class.
11. Nina Spada (personal communication) points out that some researchers working with L1
and L2 learners in brain imaging studies can certainly point to evidence that implicit and explicit knowledge systems are anatomically separate (Paradis, 2004; Ullman, 2004). I would still
argue that the movement of language constructions from explicit to implicit storage has yet to
be empirically documented in brain imaging studies.

Elaine Tarone

If we move beyond Selinkers (1972) view that IL only refers to the learners
unconscious knowledge system, and decide that IL also encompasses the learners
explicit knowledge, then we would have to use other forms of data: grammaticality judgments, cloze passages, metalinguistic statements and written production
(see Gass & Polio, this volume). But then we would also need to account for the
systematic variation we observe in those data. We would need a rigorous analytical
variationist framework to account for a learners variable performance in systematically producing one or another speech style, drawn from one or another body of
language knowledge, one implicit and one explicit. And again, how are those
knowledge systems related to one another? What is the impact of cognitive focus
on form, of interlocutor and other sociolinguistic variables, on the learners implementation of different speech styles drawn from an interlanguage system containing both implicit and explicit knowledge?
Labovs (1970) Observers Paradox should be useful to us in wrestling with
these questions. The Observers Paradox is a set of axioms claimed to apply to all
human languages, and developed to explain the complex interrelationship between speech styles and speaker attention to form. Tarone (1979) explicitly related
Labovs axioms to interlanguage, claiming that if IL is a natural language, it must
be expected to obey these same axioms, displayed in Table 1.
One cannot help but notice the similarity of the Observers Paradox to the
paradox identified in Selinker (1972) with regard to getting good data on IL: if we
ask the learner to speak while explicitly focusing attention on accuracy, we will
access the learners explicit knowledge of TL grammar, but not necessarily learn
about the implicit linguistic system used when the learner speaks when focused
on meaning.
Table 1. Axioms of the Observers Paradox (Labov 1970)
Axiom One: Style-shifting. There are no single style speakers. Every speaker shifts
linguistic and phonetic variables as the situation and topic change.
Axiom Two: Attention. It is possible to range the styles of a speaker along a continuous
dimension defined by the amount of attention paid to speech.
Axiom Three: Vernacular. In the vernacular style, where the minimum amount
of attention is given to speech, the most regular and systematic phonological and grammatical patterns are evidenced. Other styles tend to show more variability.
Axiom Four: Formality. When a speaker is systematically observed, a formal context is
thereby defined, and the speaker pays more than the minimum amount of attention to speech.
Axiom Five: Good Data. The best way to obtain enough good data on any one speaker is
through an individual tape-recorded interview: a formal context.
The Observers Paradox: if we get good recorded data, we get bad data on the vernacular style
because recording makes the speaker focus attention on form and style-shift away from
(=incorporate features of the formal style into) the vernacular, which is what we want to study.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

By 1975, several research studies had already been published documenting

systematic variability in IL speech. These variationist research studies (e.g.
Beebe,1980; Dickerson, 1975; Gatbonton,1975) provided empirical support for
Tarones view that IL style-shifting occurred along the lines described in Labov
(1970). Subsequent studies explored the theoretical implications of such a variable interlanguage (Bayley & Preston, 1996; Beebe & Zuengler, 1983; Tarone,
1983). Bayley and Tarone (2011) summarize the most recent research in this
vein, which has by and large supported the position that IL is systematically and
predictably variable in response to the same sociolinguistic variables as other
natural languages. In most studies, L2 learners produced more formal (usually
more TL-like) speech styles when paying attention to speech or editing writing,
and less formal (more IL-like) speech styles when paying attention to meaning.
But theres more going on than just attention to speech, or noticing in todays
parlance: the cognitive process of attention is just an intermediary variable; other things, namely interactional sociolinguistic variables, are what causes learner
attention to shift. In other words, L2 learners set their own targets for interlanguage development based on language norms in the social groups they identify
with and want to participate in. Bell (1984) demonstrated that the focus of attention itself, in natural language use, is determined or directed by such sociolinguistic variables as the speakers relationship with the interlocutor, and group
affiliation and identity. Rampton (1995) showed how these same variables affect
the language use of bilinguals, describing the agency of bilingual Pakistani teenagers who deliberately chose to produce a nonnative negative construction
(me no (verb)) to affirm their outgroup identities while flouting their English
teachers efforts to get them to use a standard English. In this way, they asserted
their right to select their own norms for their IL.
Fasold and Preston (2007) propose a model of the individual adult secondlanguage learners systematically variable NL and IL knowledge systems, based on
decades of variationist research on learner language. They show how both the NL
and IL systems are simultaneously systematic and variably responsive to speaker
agency and such situational factors as interlocutor and topic. The IL speakers attention fluidly mediates not just style-shifting but also language-shifting in response to shifts in social and contextual variables. Such systematic style-shifting
has, almost since the IL hypothesis was published, been meticulously detailed using multiple regression tools such as VARBRUL.12 It has also been detailed in longitudinal case studies of single learners like Bob (Tarone & Liu, 1995) that show
the impact of different interlocutors and social settings on the rate and route of the
learners acquisition of English L2 questions. This detailed longitudinal study of
12. See Young and Bayley (1996) for a description of VARBRUL in SLA research.

Elaine Tarone

SLA across three distinct social settings showed that Bobs pattern of language
change began in an informal context and spread over time to deskwork interactions with his peers, and was last to appear in his interactions with his classroom
teacher. Social setting also impacted Bobs order of acquisition of English questions,
with Stage 4 and 5 questions emerging in the informal context before Stage 3 questions. These findings could only have been discovered by a variationist longitudinal case study of this design.
Focus on Form as a pedagogical approach seems to be directly based on twin
assumptions directly related to issues of attention and task demand similar to
those just discussed. One assumption is variationist, stating that learner language
production varies, possibly between a target norm and an IL norm, as a result of
shifts of learner attention between form and meaning, where attention is mediated
by such variables as task demand or audience (Bayley & Preston, 1996; Bayley &
Tarone, 2011; Tarone, 1988). The second assumption is the interface position (as
Krashen [1985] termed it) regarding the relationship between explicit and implicit knowledge. In this position, explicit knowledge of grammar rules is believed to
evolve into implicit IL knowledge that can be used to produce constructions during meaning-focused interactions. Focus on Form pedagogy (Doughty & Williams,
1998; Fotos & Nassaji, 2006; Long, 1991) is set up to get the learner to shift attention from meaning to form and back again, so that explicit constructions will interface with, and become part of, the implicit knowledge system. (Here again it
must be noted that VanPatten, this volume, does not believe explicit reference to
formal rules affects implicit knowledge.) Variationist SLA researchers (e.g., Bayley
& Tarone 2011: 43) might see formal instruction as input that fosters just one kind
of language change, namely change from above. Language change within a speech
community may occur either from above (that is, begin with the most formal,
superordinate style that is explicitly taught in a classroom, and over time spread
downward into less formal styles) but it can also change from below (begin in a
relatively implicit, informal style and spread upward to become part of a more
formal style, as was the case in Tarone & Liu [1995]). They might question whether pedagogical techniques relying only on change from above always have the
desired direct impact on the learners implicit knowledge base, particularly when
formal, rule-based techniques tend to be implemented without regard for the influence of strong contextual and sociolinguistic variables known to directly impact
style-shifting, such as the learners relationship with the interlocutor, the topic of
the interaction, and the learners identity, group membership, and roles played in
the group.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

Two questions that follow from Selinker (1972)

Two future topics in SLA follow from the three just discussed in the interlanguage
paper, though they were not explicitly mentioned there; I think an update to the
Interlanguage Hypothesis might well include them.13
1. Do alphabetically illiterate adults construct their interlanguages the same
way that literate L2 learners do?
To answer questions raised in Selinker (1972), university researchers have accumulated a considerable database consisting of samples of oral and written language produced by second language learners. Unfortunately (but perhaps because
the researchers have been primarily university professors and graduate students),
that database has been rather narrowly focused on a particular type of L2 learner,
namely, an educated and (mostly alphabetically) literate one. Almost all the second language learners we have studied over the last 40 years have been literate in
their native language, and usually they have been alphabetically literate. As a consequence, we know next to nothing about how the alphabetically illiterate learner
constructs an interlanguage. In the Interlanguage paper, the acquisition of oral
skills14 was assumed to necessarily be primary and come first, and perhaps it does,
but there is nothing in the IL hypothesis to problematize the difference between
oral and written IL skills, or the way literacy in either L1 or L2 might influence L2
oral skills.
This issue has come into particular focus in the 21st century as more and more
adult L2 learners who are not alphabetically literate immigrate to virtually every
major urban area in the world. This participant group is almost completely unstudied by second language acquisition researchers, a particularly shocking fact
given their disproportionately large numbers globally. Bigelow, et al. (2006), and
Tarone, Bigelow and Hanson (2009) summarize research in cognitive psychology
showing that alphabetic print literacy significantly improves individuals ability to
explicitly process and manipulate units of oral language. They report a comparative study of two groups of adolescent English L2 learners, one significantly less
alphabetically literate than the other, and demonstrate that the less-literate L2
learners encountered significantly more difficulty in two oral L2 processing tasks
focused on English question formation: in taking up oral recasts to their errors,
13. Selinker has periodically updated the Interlanguage Hypothesis (cf. Selinker, 1992; Selinker
& Lakshmanan, 1992; Selinker & Douglas, 1985; Selinker, Swain, & Dumas, 1975).
14. As noted by an anonymous reviewer, it is interesting that oral skills in IL discussions have
focused primarily on speaking and not listening. A full understanding of interlanguage will
surely require more attention to learners receptive skills, both listening and reading.

Elaine Tarone

and in doing oral elicited imitation tasks. The researchers conclusions are consistent with those of cognitive psychologists working with monolingual illiterate
populations: alphabetically illiterate adults do not have literacy-based tools
(e.g. phonological awareness and visual formal lexical representations) for working memory that their literate counterparts have for parsing oral L2 input. The
implications of this conclusion for research on SLA may be profound.
For example, while many IL researchers propose the use of multiple data
sources in addition to oral utterances produced with a focus on meaning (e.g. Gass
& Polio, this volume), many such data sources (fill-in-the-blank, grammaticality
judgments, oral linguistic awareness tasks, response to formal corrective feedback)
cannot be used, or work very differently with learners who are alphabetically illiterate and who lack cognitive language processing skills that we tend to assume all
adults have, but which in fact are known to be conferred by alphabetic literacy.
We need to diversify our database of L2 learners in the future, to address questions such as these: Does alphabetic print literacy have an impact on oral IL development? If so, what is that impact, and how does it take place? Is formal rule-based
instruction reliant on learners alphabetic literacy skills, and if so how? For example, what kind of grammaticality judgments do non-literate adults produce? Over
time, as they become literate and develop a written IL, does this new knowledge
base affect their oral IL, and if so how? A number of researchers, including Martha
Bigelow, Kit Hansen, Martha Young-Scholten, Anna Vainikka, Jeanne Kurvers,
Ineke van de Craats and the author have wrestled with such issues for a decade
now, in Europe and North America through a web-based professional organization called Low Educated Second Language and Literacy Acquisition, by Adults
(LESLLA).15 But we have barely begun to scratch this surface; more SLA researchers need to look at these issues (for beginning work, see Bigelow et al., 2006;
Tarone, Bigelow, & Hansen, 2009; Tarone, Swierzbin & Bigelow, 2006).
2. Is IL a feature of an individual, or of a speech community?
The IL paper (Selinker, 1972) can easily be interpreted as focusing on individuals
and their cognitive processes; IL appears to be modeled as a product of individual construction. But no individual uses language for meaning in a social vacuum,
even in foreign language contexts where the only exposure to the language is in a
classroom. How much commonality is there between different individuals ILs?
Individual variation clearly exists from one learner to the next, but how much is
there from one IL grammar to another? And if we think of second language learners as belonging to a speech community (a group with shared language norms),
15. Proceedings of annual meetings of LESLLA are downloadable at http://www.leslla.org/
workshops.htm

Chapter 1. Enduring questions from the Interlanguage Hypothesis

can we also think of interlanguage as the construction of that community rather

than of an isolated individual? How do we come to understand the shared norms
of a speech community such as one formed within the walls of a language classroom? These questions, initially touched on by Dennis Preston (1989), have
become increasingly important as interactionist and sociocultural SLA theories
have gained ground, and influence L2 pedagogy in the classroom, where learners
in pairs and groups produce learner language in meaning-focused interactions.
We need to recognize the possibility that language classes, including intensive
English classes for international students from around the world, can be very
similar to speech communities, in having standard and non-standard language
norms, prestige and vernacular speech styles, and in-group and out-group community members who have constructed identities that impact their language use:
identities like geek, player or hipster. If so, any researcher (or teacher) who sets
up pair work needs to be aware of those identities as impacting interactional
dynamics. They should also know about research on language change within
speech communities, such as change from above (such as change initiated by the
teacher, the textbook or the geek groups) and change from below (such as
change initiated by the player or hipster groups). How do such social dynamics
affect our view of IL as a linguistic system, and the social and cognitive forces that
help it evolve?
Relevance of the interlanguage hypothesis to the present
So what is the relevance of the Interlanguage Hypothesis for the present? What is
the role of a divergent, question-asking approach to the study of learner language
in a field that has several strong theories that offer answers? I would attempt to
answer these final questions by describing the way we teach teachers and new researchers about the field of second-language acquisition.
Since 1972, I have taught many courses on second language acquisition, as part
of educating language teachers and applied linguistics graduate students. At first,
those courses focused on the Interlanguage Hypothesis, and explored questions like
those raised in this chapter. Then the courses evolved and started to focus on theories the Monitor Theory, UG, Processability Theory, Input and Interaction Theory,
Noticing Hypothesis, Focus on Form, Sociocultural Theory, Complexity Theory, Input Processing. Each theory had different founders and different answers to those
questions raised in the IL paper. My courses focused less and less on asking questions about learner language, and more and more on proposed answers to those
questions; I moved farther away from the Interlanguage papers approach of asking
questions. Fat books were now available that summarized and provided an overview

Elaine Tarone

of SLA research, with chapters on each theory with descriptions of research studies
supporting it. Learn this, seemed to be the message. Second language acquisition
had become a body of knowledge, and not a mode of inquiry. Like my colleagues, I
increasingly used a one-way transmission model to teach teachers and graduate
students about theories and theorists, their seminal publications and possible implications. And increasingly my students seemed less excited to investigate learner language and how it forms, and started asking me what I thought The Answer was:
Whose Theory is Right?
Freeman and Johnson (1998) argued that such courses on SLA research second language acquisition research were not relevant to language teaching and
should not be required coursework for teachers. And I realized that Freeman and
Johnson might have a point... not just about preparing language teachers, but
perhaps also about the preparation of SLA researchers. Knowing the names and
content of SLA theories does not in and of itself prepare people to teach language,
or to do research on learner language either, for that matter. What, I asked myself,
in the study of SLA is useful for teachers and for beginning researchers who both
might need to analyze learner language for themselves? What I had found most
interesting (and also most useful) was the approach used in the Interlanguage
paper: looking at examples of learner language and wondering about them, asking questions and framing hypotheses about interlanguage and how it developed
... and then learning to use tools and develop skills to seek answers for the questions. Selinker and Gass (1984) provided a start, but was focused more on researcher preparation than teacher preparation. Patterning Tarone and Swierzbin
(2009) on Selinker and Gass (1984), we created a lab book with video samples
of learner language for language teachers to explore: samples produced by six
adult English language learners, all doing the same meaning-focused communication tasks. We asked questions like those that the Interlanguage paper raised
for us: Look at this piece of learner language: Is this an error? Why? What patterns do you see here? What patterns of acquisition do you see? What do you see
happening to learner language in interaction? Why do you think this learner
didnt notice corrective feedback? We tried to use a divergent approach, and tried
not to give one answer to each question at the back of the book. We wanted our
students to think: about possible patterns of development, possible theoretical
perspectives and multiple ways to understand what learners do in acquiring second languages as adults. In this way, we are trying, for ourselves and the language
teachers and researchers we work with, to come full circle, back to the approach
taken in 1972 when the Interlanguage (IL) Hypothesis was formulated: prioritizing the discussion of learner language data, probing the possibility that learner
language has an underlying linguistic system, and that systematic processes are
used in second language acquisition.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

Conclusion
In this chapter I have claimed that the Interlanguage Hypothesis is not really a
theory, in that it is not a proposal that converges upon a single set of answers to
some question (cf. VanPatten & Williams, 2007). Rather, the IL hypothesis asserts
a foundational question, namely: What if the language produced by second-
language learners is systematic? From this implied question, more questions arise,
as well as different possible answers to those questions. The underlying logical
structure of the Interlanguage Hypothesis, then, takes the shape of a questionasking model with divergent possible answers, keeping an open door to several
different theoretical perspectives.
This chapter has identified three enduring questions posed in Selinker (1972)
that continue to generate research:
1. What if adult L2 acquisition is different from child L1 acquisition? How, and
why?
2. What if adult ILs inevitably fossilize at some point short of the learners goal?
3. What if the language a learner produces when focused on meaning is quite
different from that produced when s/he is focused on formal accuracy?
The chapter has also explored two additional questions that have arisen since
Selinker (1972) but that follow directly from the issues explored therein:
1. Do alphabetically illiterate adults construct their interlanguages the same way
that literate L2 learners do?
2. Is interlanguage a feature of individuals or of speech communities?
I have claimed in this chapter that viewing the IL hypothesis as a question-asking
model offers the most productive approach to really engaging teachers and researchers alike in exploring learner language, showing them multiple tools for
understanding it, and inviting new insights from a widening community of researchers and teachers. In this way, we can understand the IL hypothesis as a proposal asking us to think as widely as possible about what the nature of a possible
IL linguistic system might be, and how we can gain insight into that by gathering
empirical data. Conceived of as questions, not answers16, the themes advanced in
Selinker (1972) have served to stimulate a respectable amount of research and
theory-building, and should continue to do so into the future.

16. As Selinker (this volume) makes clear, this focus on good questions together with the variety of possible answers to them has deep roots in the Talmudic tradition a connection of which
I have only recently become aware.

Elaine Tarone

References
Adjmian, C. (1976). On the nature of interlanguage systems. Language Learning, 26, 297320.
Bayley, R., & Preston, D. (Eds.) (1996). Second Language Acquisition and Linguistic Variation.
Amsterdam: John Benjamins.
Bayley, R., & Tarone, E. (2011). Variationist perspectives. In A. Mackey & S. Gass (Eds.), Routledge Handbook of Second Language Acquisition (pp. 4156). New York, NY: Routledge.
Beebe, L. (1980). Sociolinguistic variation and style shifting in second language acquisition.
Language Learning, 30, 433447.
Beebe, L., & Zuengler, J. (1983). Accommodation theory: An explanation for style shifting in
second language dialects. In N. Wolfson & E. Judd (Eds.). Sociolinguistics and Language
Acquisition (pp. 195213). Rowley, MA: Newbury House.
Bell, A. (1984). Language style as audience design. Language in Society, 13, 145204.
Bigelow, M., delMas, R., Hansen, K., & Tarone, E. (2006). Literacy and the processing of oral
recasts in SLA, TESOL Quarterly, 40(4), 665689.
Birdsong, D. (Ed.). (1999). Second Language Acquisition and the Critical Period Hypothesis.
Mahwah, NJ: Lawrence Erlbaum Associates.
Birdsong, D. (2006). Age and second language acquisition and processing: A selective overview.
In M. Gullberg & P. Indefrey (Eds.), The Cognitive Neuroscience of Second Language Acquisition. Malden, MA: Blackwell.
Bley-Vroman, R. (1983). The comparative fallacy in interlanguage studies: The case of systematicity. Language Learning, 33, 117.
Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Cook, V.J. (1991). The poverty-of-the-stimulus argument and multi-competence. Second Language Research, 7(2), 103117.
Cook, V.J. (1992). Evidence for multi-competence. Language Learning, 42(4), 557591.
Cook, V.J. (1996). Competence and multi-competence. In G. Brown, K. Malmkjaer & J. Williams
(Eds.), Performance and Competence in Second Language Acquisition (pp. 5769).
Cambridge: CUP.
Corder, S.P. (1967). The significance of learners errors. International Review of Applied
Linguistics, 5, 160170.
Dickerson, L. (1975). The learners interlanguage as a system of variable rules. TESOL Quarterly,
9, 401407.
Doughty, C. & Williams, J. (1998). Focus-on-form in Classroom Second Language Acquisition.
Cambridge: CUP.
Eckman, F., Bell, L., & Nelson, D. (Eds.). (1984). Universals of Second Language Acquisition.
Rowley, MA: Newbury House.
Ellis, N. (1998). Emergentism, connectionism and language learning. Language Learning, 48,
631664.
Ellis, R. (2002). Does form-focused instruction affect the acquisition of implicit knowledge?
Studies in Second Language Acquisition, 24, 223236.
Ellis, R. (2006). Modeling learning difficulty and second language proficiency: The differential
contributions of implicit and explicit knowledge. Applied Linguistics, 27, 431463.
Eubank, L. (Ed.). (1991). Point Counterpoint: Universal Grammar in the Second Language.
Amsterdam: John Benjamins.

Chapter 1. Enduring questions from the Interlanguage Hypothesis

Fasold, R., & Preston, D. (2007). The psycholinguistic unity of inherent variability: Old Occam
whips out his razor. In R. Bayley & C. Lucas (Eds), Sociolinguistic Variation: Theory, Methods,
and Applications (pp. 4569). Cambridge: CUP.
Fotos, S., & Nassaji, H. (2006). Form-focused Instruction and Teacher Education: Studies in
Honour of Rod Ellis. Oxford: OUP.
Freeman, D., & Johnson, K. (1998). Reconceptualizing the knowledge-base of language teacher
education. TESOL Quarterly, 32, 397418.
Gass, S. (1989). Language universals and second language acquisition. Language Learning, 39,
497534.
Gatbonton, D. (1975). Systematic Variation in Second Language Speech: A Sociolinguistic Study.
Unpublished Ph.D. dissertation, McGill University.
Hakuta, K., Bialystok, E., & Wiley, E. (2003). Critical evidence: A test of the critical period hypothesis for second language acquisition. Psychological Science, 14, 3138.
Han, Z.-H. (2004). Fossilization in Adult Second Language Acquisition. Clevedon: Multilingual
Matters.
Han, Z.-H., & Odlin, T. (Eds.). (2006). Studies of Fossilization in Second Language Acquisition.
Clevedon: Multilingual Matters.
Krashen, S. (1976). Formal and informal environments in language learning and language acquisition. TESOL Quarterly, 10, 157168.
Krashen, S. (1981). Second Language Acquisition and Learning. Oxford: Pergamon.
Krashen, S. (1982). Principles and Practice in Second Language Acquisition. Oxford: Pergamon.
Krashen, S. (1985). The Input Hypothesis: Issues and Implications. London: Longman.
Labov, W. (1970). The study of language in its social context. Studium Generale, 23, 3087.
Long, M. (1991). Focus on form: A design feature in language teaching methodology. In K. de
Bot, R. Ginsberg, & C. Kramsch (Eds.), Foreign Language Research in Cross-cultural
Perspective (pp. 3952). Amsterdam: John Benjamins.
MacWhinney, B. (2008). A unified model. In P. Robinson & N. Ellis (Eds.), Handbook of Cognitive Linguistics and Second Language Acquisition (pp. 341371). New York, NY: Routledge.
MacWhinney, B. (2012). The logic of the unified model. In S. Gass & A. Mackey (Eds.), The Routledge Handbook of Second Language Acquisition (pp. 211227). New York, NY: Routledge.
Paradis, M. (2004). A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins.
Preston, D. (1989). Sociolinguistics and Second Language Acquisition. Oxford: Blackwell.
Rampton, B. (1995). Crossing: Language and Ethnicity among Adolescents. London: Longman.
Robinson, P., & Ellis, N. (2008). Handbook of Cognitive Linguistics and Second Language Acquisition. New York, NY: Routledge.
Schachter, J. (1996). Maturation and the issue of universal grammar in second language acquisition. In W.C. Ritchie & T.K. Bhatia (Eds.), Handbook of Second Language Acquisition
(pp. 159194). San Diego, CA: Academic Press.
Schmidt, R.W. (2001). Attention. In P. Robinson (Ed.), Cognition and Second Language Instruction. Cambridge: CUP.
Selinker, L. 1972. Interlanguage. IRAL: International Review of Applied Linguistics in Language
Teaching, 10, 209241.
Selinker, L. (1992). Rediscovering Interlanguage. London: Longman.
Selinker, L., & Douglas, D. (1985). Wrestling with context in interlanguage theory. Applied Linguistics, 6, 190204.
Selinker, L., & Gass, S. (1984). Workbook in Second Language Acquisition. Rowley, MA: Newbury
House.

Elaine Tarone
Selinker, L., & Lakshmanan, U. (1992). Language transfer and fossilization: The multiple effects
principle. In S. Gass & L. Selinker (Eds.), Language Transfer in Language Learning, 2nd ed.
(pp. 4770). Amsterdam: John Benjamins.
Selinker, L, Swain, M., & Dumas, G. (1975). The interlanguage hypothesis extended to children.
Language Learning, 25, 139152.
Sharwood Smith, M., & Truscott, J. (2005). Stages or continua in second language acquisition: A
MOGUL solution. Applied Linguistics, 2, 219240.
Swain, M. (1995). Three functions of output in second language learning. In G. Cook & B. Seidlhofer (Eds.) Principle and Practice in Applied Linguistics: Studies in Honour of H.G.
Widdowson (pp. 125144). Oxford: OUP.
Tarone, E. (1979). Interlanguage as chameleon. Language Learning, 29, 181191.
Tarone, E. (1983). On the variability of interlanguage systems. Applied Linguistics, 4, 143163.
Tarone, E. (1988). Variation in Interlanguage. London: Edward Arnold.
Tarone, E., & Liu, G-Q. (1995). Situational context, variation and SLA theory. In G. Cook & B.
Seidlhofer (Eds.), Principle and Practice in Applied Linguistics: Studies in Honour of H.G.
Widdowson (pp. 107124). Oxford: OUP.
Tarone, E., Swierzbin, B., & Bigelow, M. (2006). The impact of literacy level on features of interlanguage in oral narratives. Rivista di Psicolinguistica Applicata, VI(3), 6577. (Thomas Baldwin & Larry Selinker, Eds.) Special issue on Interlanguage: Current thought and Practices.
Tarone, E., Bigelow, M. & Hansen, K. (2009). Literacy and Second Language Oracy. Oxford: OUP.
Ullman, M.T. (2004) Contributions of memory circuits to language: The declarative/procedural
model. Cognition 92: 231270.
VanPatten, B., & Williams, J. (2007). Introduction: The nature of theories. In B. VanPatten &
J. Williams (Eds.), Theories in Second Language Acquisition (pp. 116). Mahwah, NJ:
Lawrence Erlbaum Associates.
White, L. (2003). Second Language Acquisition and Universal Grammar. Cambridge: CUP.
Young, R., & Bayley, R. (1996). VARBRUL analysis for second language acquisition research. In
R. Bayley & D. Preston (Eds.), Second Language Acquisition and Linguistic Variation
(pp. 253306). Amsterdam: John Benjamins.

chapter 2

Rediscovering prediction
Terence Odlin

Ohio State University

This chapter focuses on predictions involving language transfer, which are
among the predictions that Selinker (1972) regarded as feasible. Recent studies
(e.g., Jarvis 2002) show results that make plausible some predictions for novel
SLA contexts (e.g., Speakers of Finnish as a group will have a greater difficulty
with the articles of Portuguese than will speakers of Swedish as a group).
Although predictions involving structures such as articles seem viable (along
with some inferences about their significance for what Selinker called the Latent
Psychological Structure), there are limits on predictability. Group tendencies
often seem predictable but individual behavior much less so, particularly in light
of three phenomena considered in the chapter: multilingualism, idiosyncrasy in
IL forms, and idiosyncrasy in IL meanings.

Introduction
The words predict, prediction, etc. occur at various points in Selinkers (1972) article on interlanguage (e.g., pp. 213 and 222), and the concluding section stresses
the need for predictions involving language transfer, transfer-of-training, overgeneralization, learning strategies, and communication strategies (p. 229, points
6, 8, & 9). While the desirability of predictions is thus clear in the 1972 interlanguage (IL) agenda, the feasibility of predictions of language transfer has been
controversial in second language acquisition (SLA). This chapter reconsiders
the issue.
Any such reconsideration naturally requires a look at the history of SLA, but
some caveats about that history should be noted. First, the study of transfer did not
begin with Robert Lado or his mentor Charles Fries, even though the point of
departure in the discussion to follow will be Lado. Second, interest in transfer as a
psycholinguistic phenomenon goes back at least to the 1830s and thus clearly predates behaviorism (Odlin 2008). Lado is, however, a key figure in matters related
to prediction, as seen in one of his frequently cited assertions:

Terence Odlin

The plan of this book rests on the assumption that we can predict and describe
the patterns that will cause difficulty in learning, and those that will not cause difficulty, by comparing systematically the language and culture to be learned with
the native language and culture of the student. (Lado 1957, p. v)

It is well known that Lados position eventually came to be viewed quite skeptically, as in the following assessment:
Ideally, the psychological aspect of the Contrastive Analysis Hypothesis should
deal with the conditions under which interference takes place. That is, it should
account for instances when linguistic differences between the first and second
languages lead to transfer errors and instances when they do not. It is because it is
not possible to predict or explain the presence of transfer errors solely in terms of
linguistic differences between the first and second languages that a psychological
explanation is necessary. What are the non-linguistic variables that help determine whether and when interference occurs? (Ellis 1985: 24)

Elliss and other textbooks use the expression Contrastive Analysis Hypothesis, a
term coined by Wardhaugh (1970), although there is no uniform characterization
in these books of what the CAH actually is (Odlin 2012a). Even so, the Ellis quote
is not unusual in its pessimism about the feasibility of contrastive predictions.
Schachter (1974) challenged a growing conviction about the unpredictability
of contrastive analysis the title of an article by Whitman & Jackson (1972) and
viewed a priori contrastive analysis (i.e., predictions based on cross-linguistic
comparisons) as viable. She used a study of relative clauses to support her position,
though others have questioned the specific conclusions she drew about avoidance
strategies (Kamimoto, Shimura, & Kellerman 1992). Yet whatever interpretation
of her empirical findings is best, Schachter was one of the first researchers to compare the performances of different L1 groups on the same task, and thereby employed a powerful method for demonstrating the occurrence of transfer.
In the last four decades, numerous studies have used research designs like
Schachters in comparing different L1 groups, and from the many findings there
has emerged strong evidence of the major role of transfer in SLA. One might wonder, pace Schachter, whether her evidence really supports an a priori approach
since some results seem consistent with what professionals might have already
observed language learners doing. In such cases, then, the supposed predictions of
a priori approaches would essentially be expectations that already observed behavior will reappear in data more systematically collected indeed, the notion of
retrodiction (Larsen-Freeman & Cameron 2008: 231) may be applicable. In any
case, enough SLA research now exists to warrant some kinds of predictions about
undeniably new situations. The examples in the next section accordingly make a
case for the viability of some contrastive predictions.

Chapter 2. Rediscovering prediction

Predictions in light of some empirical work

There now exist several studies that compare two or more groups of learners with
different native languages, and Table 1 identifies key characteristics of four such
studies, each of which will be the basis for a set of predictions to follow. The first,
Jarvis (2002), investigated structures used in English to help maintain topic continuity, with definite and indefinite articles being the most frequently used structures. His results clearly show that speakers of Swedish, a language which (like
English) has both types of articles, were far more successful in using English articles than were speakers of Finnish, a language that does not have articles. The
second study (Sabourin et al. 2006) focused on grammatical gender in Dutch. The
most successful learners were speakers of German, a language quite similar to
Dutch, but even speakers of French, Spanish, and Italian did better than speakers
of English, a result consistent with the fact that all three Romance languages have
grammatical gender while English does not. The third study (Orr 1987) focused
on the bound morphology of Chichewa, a Bantu language, for speakers of Ngoni,
another Bantu language, and for speakers of Gujarati, an Indo-European language
(some examples of how nominal morphology works in Bantu languages appear
below). In the three studies described so far, speakers of the language similar to the
targets proved more successful than did speakers of the dissimilar languages, with
such success constituting strong evidence for positive transfer, which is typically a
facilitating influence due to cross-linguistic similarity, as when a native speaker of
English recognizes the meaning of the Spanish word honesto (=honest). In the
fourth study, however, the focus is on negative transfer, which is most typically
manifested in errors. Indeed, Helms-Park (2003) found that speakers of Vietnamese, a language that regularly uses serial verbs, often produced serial verb constructions on a picture task, as in the following example: She has managed to rise
the kite fly over the tallest building. No such constructions appeared in the English
of speakers of Hindi and Urdu.
Table 1. Four studies of transfer showing different outcomes for different L1 groups
Study

Structure

Target language Dissimilar NL Similar NL

1. Jarvis (2002)
2. Sabourin et al.(2006)
3. Orr (1987)
4. Helms-Park (2003)

Articles
grammatical gender
noun inflections
serial verbs

English
Dutch
Chichewa
English

Finnish
English
Gujarati
Vietnamese**

Swedish
French*
Ngoni
Hindi/Urdu

*In this group there were also languages besides French having grammatical gender.
**Unlike the other three, this study focuses on negative transfer, specifically on the use of serial verbs in L2
English.

Terence Odlin

The discussion will now turn to twelve predictions, each one being part of a set of
three, with four sets related to the studies in Table 1. The first set thus involves
predictions following from the results of the Jarvis study:

(1) a. Speakers of Finnish as a group will have greater difficulty with the articles of Portuguese than will speakers of Swedish as a group.
b. Speakers of Estonian as a group will have greater difficulty with the
articles of Portuguese than will speakers of Danish as a group.
c. Speakers of Korean as a group will have greater difficulty with the articles of Portuguese than will speakers of Albanian as a group.
In (1a) the only change from the Jarvis investigation is in the target language,
which is now Portuguese, a language like English in having both definite and indefinite articles (e.g., o barco the boat and um barco a boat). Prediction (1b)
differs from (1a) in that the two L1s are different. Even so, the results of (1a) and
(1b) should not be very different, because Estonian is a Uralic language similar to
Finnish, and Danish is a North Germanic language closely related to Swedish.
Prediction (1c) is theoretically the most interesting since it involves two L1s not
closely related to those in (1a) and (1b). Albanian, however, does have definite and
indefinite articles, whereas Korean has neither, and so if accurate, this prediction
would strongly support a possible generalization that the absence or presence of
articles in the L1 is a good predictor of the ease or difficulty that a particular group
of learners will have with articles in any target language.
The next set follows a similar pattern but concerns grammatical gender:

(2) a. Speakers of English as a group will have greater difficulty with the
grammatical gender markings of German than will speakers of French
as a group.
b. Speakers of Persian as a group will have greater difficulty with the
grammatical gender markings of German than will speakers of Pashto
as a group.
c. Speakers of Persian as a group will have greater difficulty with the
grammatical gender markings of Polish than will speakers of Pashto as
a group.
As with (1a), prediction (2a) shows only one change from Table 1: now the target
is German instead of Dutch. As noted before, these languages are closely related,
and so it seems logical to predict that if speakers of French do relatively well with
gender in Dutch, they will likewise fare better in German than will speakers of
English (even though there are some complications in the German system that will
require attention below). As with prediction (1b), the native languages in (2b) are
different, but in this case both are Indo-European languages in the Iranian branch.

Chapter 2. Rediscovering prediction

Despite their close historical relation, Persian and Pashto differ in that the latter
has grammatical gender but the former does not. In prediction (2c), the target
language is changed. Polish, an Indo-European language in the Slavic branch, has
grammatical gender, and so it makes an appropriate site for testing a possible generalization about positive transfer in any situation where a native language has
grammatical gender and where the target also does.
With regard to articles, it should be noted that unlike English, Portuguese has
articles that vary according to the number and gender of the head noun in the
noun phrase. Like French and Spanish, Portuguese distinguishes two genders,
masculine and feminine, so that while barco (boat) requires the masculine definite
article o, the noun janela (window) requires the feminine definite article a, thus a
janela the window. Because articles often combine the marking of (in)definiteness with the marking of grammatical gender in many languages, it seems credible
to envision another advantage that speakers of Swedish might have with Portuguese
articles since Swedish also has grammatical gender even though its gender categories are different from those of the Romance languages. Like English, Finnish does
not use grammatical gender, and so it seems possible to predict difficulty for speakers of either language learning German, Portuguese, or Polish in the area of gender
marking on adjectives, for example, even while English speakers would likely have
an advantage over Finns in the use of articles whether or not they were correctly
marked in gender.
Predictions (3a3c) concern transfer when the target is, as in Orrs study, a
Bantu language with a complex nominal inflection system:

(3) a. Speakers of Guajarati as a group will have greater difficulty with the
nominal inflection system of Swahili than will speakers of Ngoni as a
group.
b. Speakers of Swedish as a group will have greater difficulty with the
nominal inflection system of Swahili than will speakers of Finnish as a
group.
c. Speakers of L2 Swedish as a group will have greater difficulty with the
nominal inflection system of Swahili than will speakers of L2 Finnish
as a group.
It will help to consider two examples of noun phrases in the target language (Givn
1984: 373):

Ki-le ki-kapu ch-angu ki-dogo amba-cho ki-me-vunjika...

the basket mine small rel
it-perf-break
that small basket of mine that broke...

Terence Odlin

Yu-le m-toto y-angu m-dogo amba-ye u-me-kufa...

the child mine small rel
he-perf-die
that small child of mine who died...
In both examples there is a head noun preceded by a definite determiner and followed by a personal modifier, an adjective, and a relative clause, the head of the NP
being kapu (basket) in the first and toto (child) in the second. The structures indicated in boldface are bound morphemes that are obligatory class markers. The
markers take different forms where, for instance, ki is required in the first example
but yu in the second, which is because kapu and toto belong to different noun
classes. (There are also within-class forms such as yu-, m-, and u- in the second
example which vary for reasons other than lexical class.) Swahili and other Bantu
languages employ several noun classes that have a semantic basis, with kapu and
toto being prototypical members of their respective classes. The boldface forms
mark agreement patterns not altogether different from gender markers found in
Romance languages, for example, but with important differences such as serial order, where the bound morphemes in the two examples usually precede the stems.
Ngoni has patterns similar to those of Swahili (Orr 1987, pp. 1329), whereas
Gujarati does not, and so prediction (3a) is like predictions (1a) and (2a). Prediction
(3b) is similar to the other b predictions, but (3c) differs from all others in focusing
as it does on speakers of a second language learning a third. There are good reasons
to wish that (3b) and (3c) might be tested in coordinated studies since their results
would speak to a prediction discussed by Ringbom (2007): If you know Finnish as
L2, there will be no major problem in learning Swahili (p. 121). The logic here is
that adult L2 learners often enhance their metalinguistic awareness of structures that
native speakers of that language usually take for granted. Thus, if Ringbom is correct, prediction (3b) should be false but prediction (3c) accurate. The structures of
Finnish and Swahili are certainly different in some important ways, since the inflections of Finnish are suffixes, not prefixes, and since Finnish does not have a complex
system of noun classes triggering widely differing classes of bound morphemes as in
Swahili. Even so, both languages are considered agglutinative in their morphology,
and both show many agreement patterns between head nouns and other elements.
The fourth set focuses on serial verbs:

(4) a. Speakers of Hindi as a group will have greater difficulty with the serial
verbs of Mandarin Chinese than will speakers of Vietnamese as a
group.
b. Speakers of Bengali as a group will have greater difficulty with the serial verbs of Mandarin Chinese than will speakers of Thai as a group.
c. Speakers of English as a group will have greater difficulty with the serial verbs of Mandarin Chinese than will speakers of Gbe as a group.

Chapter 2. Rediscovering prediction

Like many other languages on the mainland of East Asia, Mandarin Chinese makes
heavy use of serial verbs as in the following example (Li & Thompson 1987: 825):
wo3 jiao4 ta1 mai3 2juzi1 chi1
I tell s/he buy orange eat
I told him/her to buy oranges to eat.
Serial verb constructions typically involve predicates having direct objects or other
arguments as with mai3 2juzi1 (the numbers in superscript indicate lexical tones),
but verbs without arguments are also common as with chi1. Vietnamese, Thai, and
Gbe are all languages that use many serial verbs, whereas Hindi, Bengali, and
English do not. Once again the (4c) prediction is the most interesting theoretically
because it involves a language quite distant (structurally, geographically, and culturally) from the Mandarin target, as Gbe is a Kwa language of West Africa. Despite any structural differences in the serial verb constructions of the languages
involved, positive transfer seems quite plausible. The findings of Helms-Park
(2003) indicate frequent attempts by Vietnamese speakers to import their L1 serial verb syntax into a language (English) quite different from the L1, and so it
seems probable that Vietnamese speakers would be even more prone to experiment with transfer when the target is Mandarin.
To my knowledge, none of the twelve predictions has been tested, yet although
some of them may never be, all are in principle falsifiable, and in some cases, SLA
researchers might find the right opportunities. For example, the 2016 Olympics in
Brazil offer many possibilities to study IL varieties of Portuguese. It should also be
noted that apart from the studies in Table 1, there is other evidence to warrant the
predictions, most especially the numerous studies of article use that point to the
importance of positive transfer (Luk & Shirai 2009; Odlin 1989: 34; 2003: 461).
There is likewise a detailed investigation by Alhawary (2009) of grammatical gender agreement in L2 Arabic with two L1 groups, English and French; consistent
with prediction (2a), Alhawarys results indicate that the French speakers outperformed the English speakers. Furthermore, in a close look at serial verb constructions in language contact, Migge (2003) found considerable evidence of transfer
from Gbe in the development of serial verbs in the Eastern Maroon Creoles
of Suriname.
How predictions get phrased is crucial, and so some metalinguistic details of
the twelve merit attention. Most importantly, the predictions focus on group behaviors of adults. Individual behaviors do matter, but group tendencies are clear in
the studies in Table 1, and if the phrase as a group were missing, the predictions
would be ambiguous and thus more problematic. Some phrasing could be changed,
however. Contrasting as they do the performance of two groups, the predictions
address possible successes as well as difficulties. Indeed, the predictions could be

Terence Odlin

reformulated to foreground success: e.g., Speakers of Swedish as a group will have

less difficulty with the articles of Portuguese than will speakers of Finnish as a
group. The difficulty referred to can, of course, be measured in various ways,
such as comparisons of the frequency of zero articles, in which case the group
predicted to have more difficulty would show more omissions of articles (Jarvis
2002; Odlin 2012b). It also seems possible to measure difficulty in terms of avoidance patterns, as Schachter and others have maintained.
The twelve predictions combine elements of both a priori and a posteriori assertions (the latter type involving something similar to the notion of retrodiction
discussed earlier). On the one hand, they state a priori expectations, but they differ
from the approach of Lado and colleagues in being based on empirical findings in
analogous language contact settings. In that sense, the twelve predictions mark a
real departure from earlier thinking. They are also relevant to a longstanding question in SLA research about what, if any, constraints there are on language transfer.
This concern is seen in some claims about constraints, either absolute or conditional, on the transfer of bound morphology, articles (a.k.a. functional projections), and other structures. However, as the Jarvis and Orr studies show, these
supposed constraints fail to account for verified cases of positive transfer. Accordingly, predicting the non-occurrence of transfer can be as perilous as any a priori
prediction of transfer (Odlin 2003: 454457, 2006: 2425).
One complication in any attempt at prediction is the problem of bidirectionality. In Swedish indefinite articles precede nouns whereas definite articles follow
them: thus, en flicka (a girl) but flickan (the girl). This difference in serial order
does not seem to hinder Swedish speakers very much in their acquisition of English articles. However, it may well be that the difference of serial order could make
acquiring L2 Swedish articles more difficult for speakers of English or speakers of
Portuguese. By the same token, it seems possible that speakers of French will experience greater difficulty with the three-gender system of German (masculine,
feminine, neuter) than German speakers will in learning French, which has a bipartite system (masculine, feminine). It is also noteworthy that even when languages have only two genders, they may not be as similar as the mere number
suggests. Dutch and Swedish have a neuter gender along with a common gender,
whereas French has, as noted, a masculine/feminine system. Clearly, the implications of such facts for SLA remain an empirical question.
A recent volume edited by Jarvis and Crossley (2012) indicates that the frequency of particular words (or word-forms) can prove to be very robust statistical
predictors (their term) of the first language of writers of L2 English texts. For example, Jarvis, Castaeda-Jimnez, and Nielsen (2012) found that the presence or
absence of articles in a large corpus helped distinguish English texts written by
Finnish speakers from those written by speakers of four other languages (Spanish,

Chapter 2. Rediscovering prediction

Portuguese, Danish, and Swedish). Although the predictions of the computer programs which Jarvis et al. (2012) used were not always accurate in attributing a particular L1 background to a particular text, the accuracy rates were beyond the
chance level and usually high (around 75%). Such programs, which operate on
straightforward information about forms, are certainly useful for making predictions about group membership in effect, using L2 behaviors to infer L1 backgrounds. However, to understand why any particular formal variable will contribute
to an accurate statistical model will require investigations into semantic and pragmatic concerns. Some of those concerns will come into focus in the next section.

Relevance to the latent psychological structure

The predictions discussed in the preceding section are not only interesting in their
own right but can also help, to the extent that they are eventually tested, to shed
light on what Selinker called the Latent Psychological Structure (LPS). With regard
to transfer, overgeneralization, and other phenomena central to understanding
the LPS, it seems clear that the psycholinguistic foundations of IL are embedded
in cognitive systems that include but also transcend the specializations of bilingual and multilingual systems. Accordingly, a full understanding of transfer
(as well as other phenomena) will require investigating issues of wider concern.
The focus in this section will be on two of those issues: 1) the communicative
value of the semantics and pragmatics of the structures in the twelve predictions
just discussed; 2) the extent to which linguistic processing is language-specific or
language-neutral.
The successes of learners aided by positive transfer are no less substantial than
the difficulties occasioned by negative transfer, and the former may ultimately
prove to be more important, as Ringbom (2007) believes, for understanding crosslinguistic influence. The problems of communicating in a new language are obviously great, and it is thus reasonable to consider what semantic or pragmatic factors
prove especially useful to learners in any successful search for cross-linguistic similarity to help meet communicative challenges. For example, if Swedish speakers
normally have an advantage (relative to Finnish speakers) in using articles, researchers may well wonder what exactly induces these learners to look for similarities and to cope with dissimilarities between the articles of their native Swedish
and those in a new language. Some meaning-based concerns seem the most likely
motivators, and indeed a plausible explanation comes from the thinking-forspeaking analyses of Slobin (1996), who has elaborated on reflections of Boas
(1938) on linguistic typology. Among the typological traits that Boas foregrounded

Terence Odlin

is the tendency in English to mark the epistemic status of nominal referents with
an obligatory article: i.e., in terms of whether the reference is definite or indefinite
and whether the entity denoted by the noun is a known topic in discourse. Applying the insights of Boas and Slobin to the case of L1 Swedish speakers, one may
infer that the processing of meaning-related concerns proves more important than
formal differences such as those in serial order noted above. A corollary of this
inference is a surmise about what to expect when the L1 has no obligatory category such as articles. Thus, although Finns can indeed signal definite or indefinite
reference in their L1 with structures such as demonstratives or even case-marking
(Chesterman 1991), the lack of a need to mark every noun for its epistemic status
leaves Finnish speakers at a great disadvantage relative to Swedes when coming to
English, Portuguese, or other languages with obligatory articles.
By the same token, grammatical gender may serve important communicative
goals despite the skepticism of some linguists about the utility of agreement systems (e.g., Jespersen 1922: 352355). Analyses vary as to the precise functions of
those systems, among which grammatical gender constitutes one particular type,
but a pragmatic function that Corbett (1999) emphasizes is also acknowledged by
other theorists (e.g., Ferguson & Barlow 1988: 1617). Corbett stresses the role of
agreement systems in allowing the speaker to keep track of referents in a discourse by means of the agreement categories (p. 17). One might add that listeners
and readers also need to keep track of referents, and so agreement systems probably contribute to comprehension as well as to production. Such an approach to
agreement makes it plausible to view the results of work on positive transfer by
Orr on Chichewa morphology and by Sabourin et al. on Dutch gender in a similar
perspective. In fact, Givn (1984: 362380) offers parallels between gender agreement in Hebrew and the nominal prefixation system of Swahili with regard to
topic continuity.
As noted above, Jarvis (2002) did not confine his analysis to articles alone but
described their role within a larger framework of topic continuity. A common
thread thus is evident in three distinct structural domains: article use, gender
marking, and noun class prefixation. Different though they are in some aspects of
form and meaning, each of these domains plays a significant part in offering cohesive clues that can help listeners and speakers construct coherent discourse
representations. The three structural domains discussed so far in this paragraph
involve noun phrases, but serial verbs constructions have likewise been analyzed
as a way to promote discourse coherence by Lehmann (1988: 190192), who
deems such constructions to lie about midway on a continuum between clauselevel cohesive devices (especially subordinate clauses) and those at the word level
or within words (such as with causative constructions signaled by verb inflections
in Quechua and other languages).

Chapter 2. Rediscovering prediction

The studies in Table 1 accordingly allow for a generalization about learner

motivations to use their native language when they seek to construct coherent
discourse representations. The transfer is largely positive for L1 speakers of
Swedish, French, and Ngoni, while largely negative for Vietnamese speakers. Yet
whether learner attempts are successful or not, the main implication for the Latent
Psychological Structure is that meaning-based concerns often seem to be involved
in attempts to create interlingual identifications between the grammatical systems
of source and target languages (see Han this volume; Tarone this volume).
Important differences are of course evident in the forms and functions of articles, gender marking, noun class morphemes, and serial verbs, and the crosslinguistic differences in each domain likely have implications for acquisition, as in
the cases of difference between gender-marked articles in Portuguese and the genderless articles in English noted in the preceding section. Still another consideration involving language-specific facts is that the cohesive function which gender
normally seems to serve does not rule out particular relations between the denotatum of a word and the gender marking it gets in any particular language. There
are, after all, many cross-linguistic mismatches such as between the German word
for the moon (Mond, masculine), and the French (lune, feminine), as well as some
interesting language-specific associations between a denotatum and its gender
that affect performance on some cognitive tasks (Jarvis and Pavlenko 2008).
The psycholinguistic literature on native speaker performance also gives reasons to believe that processing routines are to some extent language-specific, and
the routines may vary even between highly similar languages. Caramazza et al.
(2001) found that native speakers of French, Italian, and Spanish showed considerable group differences in how they selected appropriate determiners in picturenaming tasks and that there were comparable group differences between speakers
of Dutch and German. Caramazza et al. maintain that a primary source of such
performance differences is cross-linguistic variation in how categories such as
gender and number interact in the determiner selection process. Understanding
transfer and other concerns of the LPS thus requires clarity about where language-specific processing begins and ends along with clarity about universals of
language processing.
The problem of relating the specific to the universal is still more complex because of the interaction between linguistic and non-linguistic cognition, and the
domain of numeral classifiers has proven quite important for exploring that interaction. Classifier systems play a major systemic role in several languages, as in
Chinese, Thai, and Vietnamese; most typically, a noun phrase does not pluralize
nouns but instead employs a classifier to distinguish nouns according to various
semantic properties. For instance, to refer to two books in Mandarin Chinese, a
speaker will typically employ a three-word noun phrase: liang3 ben3 shu1, where

Terence Odlin

the final word is the head noun and the first form is the numeral (with the Arabic
superscripts specifying tones). The middle form, ben3, is the classifier and is obligatory. A different classifier, zhang1, is needed for certain other semantic categories,
however, as seen in liang3 zhang1 piao4 (two tickets); because of the typically different physical characteristics of tickets as opposed to books, the classifier zhang1 is
required before the head noun piao4 (ticket).
The morphemes ben3 and zhang1 are just two of a substantial list of classifiers
in Mandarin, and other varieties of Chinese also have classifiers not so different in
function as in form and, as mentioned above, Thai and Vietnamese (along with
some other languages of Southeast Asia) have similar constructions. The systemic
importance of numeral classifiers is seen in certain other regions as well including
Central America, where, for instance, classifiers somewhat like those in Chinese
are found, as in Yucatec, a Mayan language of Mexico (Lucy 1992) and in Jacaltec,
a Mayan language of Guatemala (Craig 1977).
Focusing on the significance of classifiers, Lucy has provided detailed support
for a relativist analysis of a contrast between English and Yucatec. Like Chinese,
the Mayan language does not routinely pluralize nouns, but its system of numeral
classifiers does distinguish nouns according to various semantic properties
(e.g., distinguishing nouns as to whether the denotata take the form of slices, edges, or strands). The grammatical contrast between Yucatec and English has, Lucy
finds, consequences for non-linguistic cognition. On picture recall tasks, for example, Yucatec-speaking and English-speaking subjects differed significantly in
how they remembered specific pictorial details.
Han (2010) has offered intriguing evidence that the classifiers of Chinese affect native speakers when they attempt to use English articles. Still, the possible
cognitive implications of classifier languages for SLA remain largely unexplored,
including possible implications involving transfer when the target is a classifier
language. The following predictions accordingly raise relevant issues:

(5) x. Speakers of English as a group will have greater difficulty with the numeral classifier system of Yucatec than will speakers of Jacaltec as a
group.
y. Speakers of English as a group will have greater difficulty with the numeral classifier system of Mandarin Chinese than will speakers of
Vietnamese as a group.
z. Speakers of English as a group will have greater difficulty with the numeral classifier system of Mandarin Chinese than will speakers of
Jacaltec as a group.
Because these predictions do not have studies like those in Table 1 as precedents
for predictions such as (1a), (2a), etc., the letters x, y, and z are used instead of a, b,

Chapter 2. Rediscovering prediction

and c. However, the logic of the predictions is still similar to the logic in the twelve
already discussed. In (5x), Yucatec is the target language and the L1s are English
(dissimilar to the target in the domain of classifiers) and Jacaltec (similar to the
target). In (5y) the target is different and also one of the L1s (Vietnamese), and in
(5z) the target is the same as in (5y) (Chinese) but one L1 (Jacaltec) is the same as
in (5x), with the theoretical issue being whether speakers of a classifier language
geographically and culturally far removed from the target can nevertheless capitalize on the structural similarity with their L1.
Testing predictions (5x5z) would obviously contribute much to the literature
on grammatical transfer, but it could also help in understanding just how deeply
any language-specific routines of processing may be entrenched in conceptual systems and in understanding how much any routines might affect non-linguistic
memory or categorization (provided that methods similar to those of Lucy are
employed). Such work could thus link in new ways some longstanding concerns
about relativity with longstanding concerns about fossilization (Han this volume)
and other problems involving the LPS.
Limits on predictability
Although group tendencies may often be predictable, individual behavior is much
less so for a variety of reasons, three of which will be considered here: multilingualism, idiosyncrasy in forms, and idiosyncrasy in meanings. The examples,
which describe events in the Charlie Chaplin movie Modern Times, come from a
corpus developed by Jarvis (1998), whose research design combines strengths of
methodologies employed by Selinker (1969) and Ringbom (1987).
Multilingualism
While there is abundant evidence of cross-linguistic influence in multilingual
studies, complications arise in predicting, for example, whether the L2 is more
likely than the L1 to influence IL structures in L3. Such predictions are further
complicated by individual exceptions, as in the following example where an L1
Finnish speaker defies a group tendency of L2 Swedish influence in prepositional
use and instead shows a transfer pattern from the L1 in L3 English:

(1) Butikman springed girls after. (=The shopkeeper ran after the girl.) (F9B 18)

Swedish influence is obvious in the forms butikman (shopkeeper) and springed

(ran), yet the learner used after as a postposition (indicating L1 Finnish influence)
despite having studied L2 Swedish for six years and despite a strong tendency of

Terence Odlin

learners with a similar background to use after as a preposition with, accordingly,

positive transfer from the L2 (Odlin 2012c). Swedish, which has a cognate preposition efter, probably influenced the group tendency since many more individuals in
the group (16 out of 35) used the form after in contrast to another group with the
same number of years of English (two) but with no L2 Swedish; in the latter group,
only five individuals out of 35 used after. Even so, learner F9B 18 shows no such
grammatical influence even while the use of after in the example may constitute a
case of L2 lexical influence. Such behavior is consistent with other cases of Finnish
speakers using English words as postpositions (Odlin 2012b) and with L1 Finnish
influence in subordinate clause patterns even when Swedish influence would have
been more helpful (Odlin 2013).
Idiosyncrasy in forms
In their accounts of events in Modern Times, many learners referred to the same
characters but with different forms (or no forms) before the referring nouns. In the
cases below, two more individuals along with learner F9B 18 all have a common
background of two years of L3 English and six years of L2 Swedish. Yet Table 2
shows considerable variation not only from one individual to the next but also in
the pattern used for different characters. For example, the learner F9B 09 uses an
to introduce Paulette Goddard (An woman steals a bread and run away) yet uses
the form a to introduce three other characters, the determiner another to introduce an older lady, and the first reference to a baker has no article (There was another woman and she tells to breadman what happened). Neither F9B 18 nor F9B 20
followed this pattern, and indeed only one other Finn resembled F9B 09 in using
the article an before a noun beginning with a consonant. This probably reflects
Swedish influence, as several L1 Swedish participants did use an in a similar way,
with the form an likely identified with the Swedish indefinite article en.
The great variability in Table 2 no doubt reflects different individual impressions of how the English article system works, and the error with an suggests that
L2 Swedish may occasionally play a role. The accurate use of articles in some cases
may well reflect positive transfer from the L2, as in the use of a and the by F9B 09
to introduce and continue referring to a husband and wife seen in a later part of the
film. Even so, it is difficult to assess the extent of any L2 influence, let alone predict
it. Some apparent patterns may or may not offer clues to formal choices, as where
both F9B 18 and F9B 20 use only the to refer to Chaplin in a subsequent mention
yet both individuals alternate between zero and the in referring to Goddard. Does
the inconsistency reflect simple inattention to the form of the NPs or perhaps
something else? Moreover, why does F9B 20 correctly introduce the reference to
the housewife with a yet omit an article in subsequent references? Answers to these

Chapter 2. Rediscovering prediction

Table 2. Individual cases of zero and articles in noun phrases indicating certain human
referents in scenes in Modern Times

F9B 09
1st
Sbs
F9B 18
1st
Sbs
F9B 20
1st
Sbs

PG

CC

OL

BA

HU

WI

An
/a

a
a

another
NSM

/the

a
the

a
the

The
/the

the
the

/the

NM
NSM

NM
NSM

The
/the

PN
the

some

The
The

PG = Paulette Goddard, CC = Charlie Chaplin, OL = older lady, BA = baker, HU = husband, WI = wife; 1st
= first reference, sbs = subsequent references; NM = not mentioned, NSM = no subsequent mention, PN =
proper noun, = no article or determiner used in the NP which had a singular countable common noun.

and other questions are not possible for these individuals, but some light might be
shed by conducting longitudinal research with multiple individuals to see what
does and does not change in article choices (e.g., Han 2010).
Idiosyncrasy in meanings
The prediction problem is all the greater if one tries to predict just what every individual may mean in using any particular form or no form. When native speakers of the target language use articles, a researcher can normally understand such
uses as either definite or indefinite reference. The following instances are the first
sentences in Modern Times accounts written by native speakers of English:

(2) The woman stole the bread and she ran into a man. (ex 20)
(3) A woman was walking and she was hungry... (ex 21, emphases added)

While similar in meaning, these sentences suggest different pragmatic assumptions: the first writer (ex 20) seems to assume that readers can identify the referent
of woman (i.e., the character played by Goddard), whereas the second writer
(ex 21) makes no such assumption. Native speakers of Swedish also vary in their
choice of articles:
(4) Flickan var hungrig. (sx 07)
Girl-the was hungry
(5) En flicka stal ett brd... (sy 11)
A girl stole a bread

Terence Odlin

In L1 Finnish writing, however, there is often indeterminacy in the same context:

(6) Nainen varasti leivn ja lhti karkuun .... . (fw 13)
Woman stole bread and left running
A/the woman stole bread and ran away.
As the English translation indicates, either a or the could work. There does exist in
Finnish a strong tendency for a NP before a verb to be definite, but Chesterman
(1991: 144, 160) notes important exceptions, and so caution is advisable in interpreting sentences such as 6. A similar indeterminacy is seen in some Finns L2
English:

(7) Woman take a bread. (F5 08)

Although the tendency noted by Chesterman might argue for interpreting woman
as a definite NP, there is no certainty, especially since there is no assurance that any
putative influence from Finnish syntactic patterns played a role. In other cases,
though, context reduces or eliminates the indeterminacy:

(8) There is a woman who is alone and hungry. The woman takes a bread from
the bakery....They bring the woman, who took the bread, in. (F9A 24, emphases added)

This excerpt illustrates the target-language pattern for article use with woman,
with definite articles used in all subsequent mentions of Goddard. In still other
cases, however, the articles have indeterminate or idiosyncratic meanings:

(9) A woman walked in the street and she was hungry.... A man walked in the
street and the woman run towards him....The police man was in the street
and the man knocked him with a stick. Then a man and a woman escaped.
(F9B 14)

This passage comes from a learner who had studied L2 Swedish for six years. Although the earlier uses of indefinite and definite articles for Goddard and Chaplin
are target-like, the indefinite articles in the last sentence are contextually misused
since the referents in the film are still Chaplin and Goddard. Perhaps the first uses
of articles by this learner mean more or less the same thing they would for a native
speaker. However, the problems of the misused ones should prompt some doubt
about that optimistic interpretation, and the whole passage supports Selinkers assertion that every sentence of the IL is to be regarded as idiosyncratic until shown
to be otherwise (1992: 159).
Not all uses of articles in learner English prove as problematic as those in
9, but such cases argue for caution. Bley-Vroman (1983) and others have emphasized that because ILs are natural languages in their own right, they need to be

Chapter 2. Rediscovering prediction

analyzed on their own terms without unwarranted assumptions that IL users invariably do what native speakers of the target do in particular pairings of forms
with meanings. How analysts understand IL productions (spoken or written)
should thus be compared with how literary translators approach their work, as
with Example 7, since how to interpret Woman in the sentence is problematic.
Translation research by Tabakowska (1993) has found that translators of literature vary in how they construe a source text whose NPs have no articles as in
Polish. While translators of literary texts usually enjoy some poetic license in how
they may understand the source text meaning, SLA researchers must exercise
greater circumspection in their inferences about what an IL production actually
means. Such caution will perforce imply limits on what can be predicted about
pairings of form and meaning.
Conclusion
This chapter reconsiders the feasibility of predictions of language transfer, the desirability of which was made clear in Selinkers (1972) agenda. The predictions
considered in the chapter come from a wide range of evidence in the last four decades for the major role of transfer in SLA, especially in studies that have compared the IL productions of different L1 groups. The evidence warrants some kinds
of predictions about undeniably new situations. Such predictions are interesting in
their own right, but they are also potentially helpful for shedding more light on
what Selinker termed the Latent Psychological Structure.
Acknowledgments
I would like to thank the anonymous readers of an earlier draft. Any shortcomings
of the chapter are, of course, mine alone.
References
Alhawary, M. (2009). Speech processing prerequisite or language transfer? Evidence from
English and French L2 learners of Arabic. Foreign Language Annals, 42, 367390.
Bley-Vroman, R. (1983). The comparative fallacy in interlanguage studies: The case of systematicity. Language Learning, 33, 117.
Boas, F. (1938). Language. In F. Boas. (Ed.), General Anthropology (pp.124145). Boston, MA:
Heath.

Terence Odlin
Caramazza, A., Miozzo, M., Costa, A., Schiller, N., & Alario, F. (2001). A crosslinguistic investigation of determiner production. In E. Dupoux (Ed.), Language, Brain, and Cognitive
Development (pp.209226). Cambridge, MA: The MIT Press.
Chesterman, A. (1991). On Definiteness: A Study with Special Reference to English and Finnish.
Cambridge: CUP.
Corbett, G. (1999). Agreement. In J. Miller (Ed.), Concise Encyclopedia of Grammatical Categories (pp.1218). Amsterdam: Elsevier.
Craig, C. (1977). Jacaltec: The Structure of Jacaltec. Austin, TX: University of Texas Press.
Ellis, R. (1985). Understanding Second Language Acquisition. Oxford: OUP.
Ferguson, C. & Barlow, M. (1988). Introduction. In M. Barlow & C. Ferguson (Eds.), Agreement
in Natural Language (pp.122). Stanford, CA: CSLI.
Givn, T. (1984). Syntax: A Functional-Typological Introduction, Volume I. Amsterdam: John
Benjamins.
Han, Z-H. (2010). Grammatical inadequacy as a function of linguistic relativity: A longitudinal
case study. In Z.-H. Han & T. Cadierno (Eds.), Linguistic Relativity in Second Language
Acquisition: Evidence of First Language Thinking for Speaking (pp.154182). Clevedon:
Multilingual Matters.
Helms-Park, R. (2003). Transfer in SLA and creoles: The implications of causative serial verbs in
the interlanguage of Vietnamese ESL learners Studies in Second Language Acquisition, 25,
211244.
Jarvis, S. (1998). Conceptual Transfer in the Interlanguage Lexicon. Bloomington, IN: IULC Publications.
Jarvis, S. (2002). Topic continuity in L2 English article use. Studies in Second Language Acquisition, 24, 387418.
Jarvis, S., & Pavlenko, A. (2008). Ccross-linguistic Influence in Language and Cognition. New
York, NY: Routledge.
Jarvis, S., & Crossley, S. (Eds.) (2012). Approaching language Transfer through Text Classification.
Bristol: Multilingual Matters.
Jarvis, S., Castaeda-Jimnez, G., & Nielsen, R. (2012). Detecting L2 writers L1s on the basis of
their lexical styles. In S. Jarvis & S. Crossley (Eds.), Approaching Language Transfer through
Text Classification (pp.3470). Bristol: Multilingual Matters.
Jespersen, O. (1922). Language: Its Nature, Development, and Origin. London: Allen and Unwin.
Kamimoto, T., Shimura, A., & Kellerman, E. (1992). A second language classic reconsidered
the case of Schachters avoidance. Second Language Research, 8, 251277.
Lado, R. (1957). Linguistics across Cultures. Ann Arbor, MI: University of Michigan Press.
Larsen-Freeman, D., & Cameron, L. (2008). Complex Systems and Applied Linguistics. Oxford:
OUP.
Lehmann, C. (1988). Towards a typology of clause linkage. In J. Haiman & S. Thompson (Eds.),
Clause Combining in Grammar and Discourse (pp.181225). Amsterdam: John Benjamins.
Li, C. & Thompson, S. (1987). Chinese. In B. Comrie (Ed.), The Worlds Major Languages
(pp. 811833). Oxford: OUP.
Lucy, J. (1992). Grammatical Categories and Cognition. Cambridge: CUP.
Luk, Z., & Shirai, Y. (2009). Is the acquisition order of grammatical morphemes impervious to
L1 knowledge? Evidence from the acquisition of plural -s, articles, and possessive s.
Language Learning, 59, 72154.
Migge, B. (2003). Creole Formation as Language Contact. Amsterdam: John Benjamins.

Chapter 2. Rediscovering prediction

Odlin, T. (1989). Language transfer: Cross-linguistic Influence in Language Learning. Cambridge:
CUP.
Odlin, T. (2003). Cross-linguistic influence. In C. Doughty & M. Long (Eds.), Handbook on
Second Language Acquisition (pp. 436486). Oxford: Blackwell.
Odlin, T. (2006). Could a contrastive analysis ever be complete? In J. Arabski (Ed.), Cross-
linguistic Influences in the Second Language Lexicon (pp. 2235). Clevedon: Multilingual
Matters.
Odlin, T. (2008). Conceptual transfer and meaning extensions. In P. Robinson & N. Ellis (Eds.),
Handbook of Cognitive Linguistics and Second Language Acquisition (pp. 306340). New
York, NY: Routledge.
Odlin, T. (2012a). The Contrastive Analysis Hypothesis. In P. Robinson (Ed.), The Routledge
Encyclopedia of Second Language Acquisition (pp. 129131). New York, NY: Routledge.
Odlin, T. (2012b). Nothing will come of nothing. In B. Kortmann & B. Szmrecsanyi (Eds.),
Linguistic Complexity in Interlanguage Varieties, L2 Varieties, and Contact Languages
(pp.6289). Berlin: De Gruyter.
Odlin, T. (2012c). Reconciling group tendencies and individual variation in the acquisition of L2
and L3. In D. Gabry-Barker (Ed.), Cross-linguistic Influences in Multilingual Language Acquisition, Second Language Learning and Teaching. (pp. 8197). Berlin: Springer.
Odlin, T. (2013). Accelerator or inhibitor? On the role of substrate influence in interlanguage
development. In D. Schreier & M. Hundt (Eds.), English as a Contact Language (pp. 298
313). Cambridge: CUP.
Orr, G. (1987). Aspects of the Second Language Acquisition of Chichewa Noun Class Morphology.
Unpublished Ph.D. dissertation.University of California, Los Angeles.
Ringbom, H. (1987). The Role of the First Language in Foreign Language Learning. Clevedon:
Multilingual Matters.
Ringbom, H. (2007). Cross-linguistic Similarity in Foreign Language Learning. Clevedon: Multilingual Matters.
Sabourin, L., Stowe, L., & de Haan, G. (2006). Transfer effects in learning a second language
grammatical gender system. Second Language Research, 22, 129.
Schachter, J. (1974). An error in error analysis. Language Learning, 24, 205214.
Selinker, L. (1969). Language transfer. General Linguistics, 9, 6792.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10, 209231.
Selinker, L. (1992). Rediscovering Interlanguage. London: Longman.
Slobin, D. (1996). From thought and language to thinking for speaking. In J. Gumperz &
S. Levinson (Eds.), Rethinking Linguistic Relativity (pp. 97114). Cambridge: CUP.
Tabakowska, E. (1993). Articles in translation: An exercise in cognitive linguistics. In R. Geiger
& B. Rudzka-Ostyn (Eds.), Conceptualizations and Mental Processing in Language (pp.
785800). Amsterdam: John Benjamins.
Wardhaugh, R. (1970). The contrastive analysis hypothesis. TESOL Quarterly, 4, 12330.
Whitman, R., & Jackson, K. (1972). The unpredictability of contrastive analysis Language
Learning, 22, 2942.

chapter 3

From Julie to Wes to Alberto

Revisiting the construct of fossilization*
ZhaoHong Han

Teachers College, Columbia University

This chapter revisits the construct of fossilization, the bedrock of Selinkers
(1972) Interlanguage Hypothesis. After reviewing the early conception of
fossilization, I focus my discussion on intra-learner, and to a lesser extent,
inter-learner differential success or failure, arguing that fossilization is selective,
idiosyncratic, and contingent. I end the discussion by underscoring that the
study of fossilization is less about revealing deviances from the presumed norm
than about resolving a dual cognitive conflict, namely, why is it that in spite of
propitious conditions, development is cut short in some areas? And why is the
developmental interruption made most apparent when learners attempt selfexpressions (i.e., meaning-based production) in the target language?

Introduction
A central debate ensuing from the publication of Selinker (1972) concerns the
question of why most L2 learners fail to acquire target language competence (Ellis,
2007). Bley-Vroman (1989, 2009) refers to it as the non-convergence property of
second language acquisition (SLA). Observations such as the following abound in
the 40 years of SLA literature:
The outcome of first language acquisition is success: normal children acquire the
grammar of the ambient language. Adult second language acquisition, on the
other hand, results in varying degrees of success. Failure to acquire the target language is typical. (Birdsong, 1992, p. 706)
* The title of this chapter takes inspiration from Bley-Vromans (2009) article where, updating his (1989) Fundamental Difference Hypothesis and discussing the explanatory burden of
SLA, he insightfully notes that an adequate theory must permit everything from the so-called
near-native cases like Julie (Ioup et al., 1994) to cases in which the acquired grammar is dramatically different from that of the input, like Schmidts (1983) Wes or Schumanns (1978)
Alberto (p. 178).

ZhaoHong Han

Few [adult second language learners] are completely successful; many fail miserably, and many achieve very high level of proficiency, given enough time,
input, effort and given the right attitude, motivation, and learning environment. (Bley-Vroman, 1989, p. 49)

Kellerman (1995) claims:

One of the most enduring and fascinating problems confronting researchers of
second language acquisition (SLA) is whether adults can ever acquire native-like
competence in a second language (L2), or whether this is an accomplishment reserved for children who start learning at a relatively early age. As a secondary issue,
there is the question of whether those rare cases of native-like success reported
amongst adult learners are indeed what they seem, and if they are, how it is that
such people can be successful when the vast majority are palpably not. (p. 219)

Boiled down, this debate has pointed to two key phenomena in SLA, inter-learner
and intra-learner differential success (or failure, for that matter). In their book titled Theories in Second Language Acquisition, VanPatten and Williams (2007)
make the two phenomena part of their list of 10 categorical observations in SLA
that call for theoretical explanation (see also Towell & Hawkins, 1994):
Observation #5:
Second language learning is variable in its outcome. (inter-learner differential
success)
Observation #6:
Second language learning is variable across linguistic subsystems. (intra-learner
differential success)

In their view, theories ought to accomplish two functions: a) explain observable

phenomena and b) unify explanations of various phenomena where possible. As it
turns out, four of the nine theories reviewed in that book the associative cognitive creed, skill acquisition theory, autonomous induction theory, and sociocultural theory deal with both observations 5 and 6, two processability theory
and input, interaction, and output address only one, observation #5, and the
rest neither. As of today, the field has not developed a full and coherent understanding of inter-learner differential success, much less of intra-learner differential
success. It is to the latter I will mostly turn in this chapter.
The remainder of this chapter proceeds as follows: I will first present case studies illustrating inter- and intra-learner differential success, and then review
Selinkers (1972) construct of fossilization as the earliest theoretical attempt to explain the phenomena. After that, I will turn to the current literature in order to (a)
update our general understanding of fossilization and (b) elucidate current conceptions of intra-learner differential success and, only where fitting, inter-learner

Chapter 3. Revisiting the construct of fossilization

differential success. The chapter will conclude with a discussion of current controversies surrounding the construct of fossilization and its future prospect.
Inter-learner and intra-learner differential success
An apt point of departure for understanding inter- and intra-learner differential
success lies in SLA case studies, which typically are longitudinal in nature. Some
of these studies have shed critical light on differential achievement between and
within learners. For example, a study by Ioup et al. (1994) focused on an adult
English-speaking learner, named Julie, of Egyptian Arabic in an untutored setting. Extensive testing via a speech production task, a grammaticality judgment
task, a translation task, an anaphoric interpretation task, and an accent recognition task showed that Julie apparently acquired native proficiency in Egyptian
Arabic (p. 73).
Julie has no noticeable foreign accent, makes few mistakes in morphology and
syntax, has good control of the lexicon, including conventionalized forms, and
appears to have sophisticated competence. (Ioup et al., 1994, p. 79)

The researchers attributed Julies overall success to her possession of language talent, a putative innate trait associated with unusual brain organization where a
greater portion of the cortex is devoted to language (Novoa et al., 1988; Obler,
1989; Schneiderman & Desmarais, 1988, cited in Ioup et al., 1994). Julie reportedly shared many generally accepted traits of language talent, including, but not
limited to, a superior associative memory, ability to master new codes, a sensitive
ear for phonetic cues, and precocious L1 development. Her language talent supposedly had two effects on Julies L2 acquisition giving her the ability to perceive
linguistically significant contrasts in L2 input, including those that were only implicitly noted, and enabling her to organize the information obtained into a nativelike L2 grammar, independent of the L1 grammar.
Even so, the staggering success of Julie in morpho-syntax appeared to be tinted with a lack of comparable success in the domain of discourse semantics. Ioup
et al. note that on the anaphoric interpretation task, which dealt with discourse
semantics, Julies performance fell short of that of native speakers (cf. Coppieters,
1987). The preferred NS interpretation was The girl angered the lady, whereas
Julie answered that The lady angered the girl (Ioup et al., 1994, p. 90).
Julie is, nevertheless, widely hailed as an outlier, one of the few adults
(hovering around 5% according to Selinker) who are highly successful second language learners, and allegedly counter-evidence to the Critical Period Hypothesis
(Lenneberg, 1967). In contrast, most learners fall well below that level of attainment.

ZhaoHong Han

Alberto, a Spanish-speaking adult learner of English reported in Schumann

(1978), appears to be one of those who fail miserably (Bley-Vroman, 1989, p. 49).
Schumann (1978) reports that:
When Alberto was compared with the other five subjects in terms of negative,
interrogative, and auxiliary development, he was found to have considerably less
growth in these structures than the other subjects. (p. 113)

This slow progression of learning, in Schumanns view, had to do with Albertos

lack of acculturation (i.e., keeping a psychological and social distance from speakers of English).
Between Julie and Alberto on the two ends of the success-failure continuum
are the vast majority of L2 learners who probably are like Wes, an adult Japanesespeaking learner of English, reported on in Schmidt (1983), who are able to achieve
success in some domains but not in others. For example, Wess grammatical competence was described as follows:
Over a three-year period characterized by extensive and intensive interaction
with native speakers, Wess ... acquisition of productive grammatical rules has
been minimal and almost insignificant. (Schmidt, 1983, p. 150)

In spite of his lack of grammatical development, Wes showed much progress in his
discourse competence. For example, progressive forms were no longer used for
declarative function with any frequency, while the use of imperatives increased
(e.g., Please next month send orders more quick); shall we? and lets were used
productively as patterns for a great many different requests; and in general Wess
directives showed a great deal more elaboration (shall we maybe go out coffee now,
or you want later?;OK, if you have time please send two handbag, but if youre too
busy, forget it) (Schmidt, 1983, p. 154).
Wes thus illustrates intra-learner differential success par excellence, and,
flanked by Alberto and Julie, presents a compelling scenario of inter-learner differential success. This, however, only gives us a macro idea of inter- and intralearner differential attainment. In fact, micro-level evidence abounds as well in the
literature. For example, VanBuren (2001) offers the following anecdote:
I have a highly intelligent Scandinavian friend who has resided in Britain for 42
years and who keeps saying The man which I saw ... He said it when I first met
him 41 years ago, and last month he was still saying it. Why? After I first asked
that question all those many years ago I consulted various works on structural
linguistics and married a Scandinavian. The answers I received in my quest for
a satisfactory answer made it clear that, in contrast to English, relative pronouns
in Scandinavian languages do not carry, what we would now call, the feature
[ animate] (sem and som the invariant forms). So is that it? I wondered. If one

Chapter 3. Revisiting the construct of fossilization

was to believe the proponents of structural contrastive analysis Trager, Smith

et al (1951), Fries (1952), Hill (1955), and Lado (1957), all splendid scholars
then that was indeed it: clearly, my friend was experiencing interference from his
mother tongue. But then, I asked on reflection, why did he not ever use the seemingly perfect equivalent in English, the invariant form that? And why did his wife
sometimes use which but most times who in similar circumstances, and why did
my own domestic informant never put a foot wrong in this regard? (p. 457)

Here, we see three learners displaying differential command of the English relative
pronouns, VanBurens wife reminiscent in a way of Julie; his friends wife reminiscent of Wes; and his friend of Alberto, all resonating with Bley-Vromans assertion
on the mix of ultimate attainment on the success-failure continuum.
By placing specific linguistic constructions under scrutiny, a number of case
studies, including Lardiere (1998a, 1998b, 2007), Han (2000, 2006, 2010), White
(2003), and Wang (2012), have also found clear evidence of intra-learner differential success. By way of illustration, Lardiere (1998a, 1998b, 2007), in her case study
spanning 18 years, found, inter alia, that her subject, Patty, an adult Chinesespeaking learner of English, had differential success with the English article system (cf. Han, 2010; White, 2003). Pattys production of definite articles appeared
to be more target-like than her production of indefinite articles. In fact, such differential success was pervasive in Pattys system. Consider the following utterances
from Patty:
1. China also send a lot of boat to the refugee who want to go back to China
2. So there is seven #seven opera you can only listen to
3. There are book club in Hawaii you may like to join
These utterances are highly revealing. While each of them is flawed, lacking in
grammatical markers like the plural -s, the 3rd person singular -s, etc., together
they provide robust evidence of successful acquisition of the English relative clause
construction: Each sentence contains a (different type of) relative clause that looks
completely target-like. The data thus illuminate another facet of intra-learner differential success: the intra-learner variation can occur in the same linguistic domain, i.e., morpho-syntax.
With the exception of Alberto, the learners discussed above can all be considered endstate learners1 by the common yardstick (see, e.g., Johnson et al., 1996).
They all had had far more than five years of immersion in the target language living and working in a country where the target language (TL) is spoken. Patty, for
1. For evidence of inter- and intra-learner differential success in learners en route, see LarsenFreeman (2006). The study examined the oral and written production data from five Chinesespeaking learners of English, demonstrating the waxing and waning of patterns in relation to
their accuracy, fluency, and complexity development.

ZhaoHong Han

instance, had lived and worked in the U.S. for 10 years prior to the onset of the
study. She was married to a native speaker of American English and spoke only
English with her husband (and later, also her daughter), had completed undergraduate and masters degrees in American universities and was thus highly literate in English (Lardiere, 2012, p. 258). The question then arises: Why did these
learners, in spite of the favorable learning conditions available to them, still wind
up with variable success? This question is the lynchpin for much of the discussion
in Selinker (1972).
Fossilization (Selinker, 1972)
Selinker (1972) coins and invokes the construct of fossilization to both describe
and explain the general failure in SLA. In particular, he calls attention to a crucial
fact, perhaps the most crucial fact, which any adequate theory of second-language
learning will have to explain, namely, the regular reappearance or reemergence
in IL productive performance of linguistic structures which were thought to be
eradicated (p. 216).
Thus, as a descriptor, fossilization refers to linguistic items, rules, and subsystems which speakers of a particular NL [native language] will tend to keep in their
IL [interlanguage] relative to a particular TL [target language], no matter what the
age of the learner or amount of explanation and instruction he receives in the TL
(p. 215). Simply put, fossilization concerns interlanguage-particular features that
are impermeable to such environmental influences as instruction or exposure to
the target language (for discussion of the nature of both influences, see Van Patten,
this volume). The metaphoric flavor of the term suggests that not only are these
features resistant, but they are also persistent and even permanent.
Fossilization is endemic. By Selinkers reckoning, it is an inescapable reality for
95% of L2 learners.2 As such, the term fossilization has also become a shorthand,
and even an explanation, for the general lack of success of SLA relative to first
language acquisition.
As an explanation, Selinker hypothesizes that a fossilization mechanism exists
in a latent psychological structure, which is made up of five central processes:
language transfer, transfer of training, learning strategies, communication strategies, and overgeneralization. These processes are deemed central to second-
language learning and ... each process forces fossilizable material upon surface IL
utterances, controlling to a very large extent the surface structures of these
2. The construct of fossilization was extended to child second language learners in Selinker,
Dumas, and Swain (1975).

Chapter 3. Revisiting the construct of fossilization

utterances (p. 217; emphasis in the original), and combinations of these processes produce what we might term entirely fossilized competences (ibid.) Selinker
further speculates that this mechanism is activated whenever [learners] attempt
to produce a sentence in the second-language, that is whenever they attempt to
express meanings, which they may already have, in a language which they are in
the process of learning (p. 212). (For discussion of the latent psychological structure, see Han, 2013.)
The dual referents of fossilization it being a mechanism and a phenomenon
have a number of implications. First of all, they implicate a causal relationship
between the mechanism and the phenomenon. The underlying argument goes as
follows: L2 learning is in large measure driven by a latent psychological structure
which engineers, among other things, surface structural deviances from the target
language that are stubbornly resistant and persistent. However, insofar as the five
central processes may function differently within the latent psychological structure, individual learners may produce different levels and extent of deviance. A key
process in the latent psychological structure is native language transfer, a probabilistic process induced by both learner external and internal factors. Consequently,
transfer (and fossilization, for that matter) happens both universally across
learners of different L1 backgrounds and idiosyncratically within learners of
the same L1 background (cf. Montrul, this volume; Odlin, this volume).
Second, the dual use of the construct of fossilization serves to underscore the
nature of fossilization as a psycholinguistic and neuro-cognitive phenomenon.
Selinker (1972) writes:
What seems to be most promising for study is the observation concerning fossilization. Many IL linguistic structures are never really eradicated for most second-language learners; manifestations of these structures regularly reappear in
IL productive performance, especially under conditions of anxiety, shifting attention, and second-language performance on subject matter which is new to the
learner. It is this observation which allows us to claim that these psycholinguistic structures, even when seemingly eradicated, are still somehow present in the
brain, stored by a fossilization mechanism (primarily through one of these five
processes) in an IL. (p. 221; emphasis in the original)

Third, the dual view of fossilization evokes ambivalence about the intra-learner
scope of fossilization, raising the question of whether fossilization is local or global. It seems that when tied to grammatical properties, fossilization is local. But
when tied to a mechanism, it projects a sense of global.
Overall, Selinkers construct of fossilization can be viewed as a theoretical
explanation for the general lack of success of SLA relative to first language acquisition. Empirically, the construct has a concrete reference to deviant IL structures

ZhaoHong Han

persisting in defiance of environmental influence including pedagogical intervention. Tied to possible variations within the latent psychological structure, it
speaks, indirectly rather than directly, to inter-learner differential failure. What
seems to have eluded this line of theorizing is intra-learner differential failure,
specifically, selective fossilization: what actually fossilizes and why fossilization is
selective within individual learners (cf. Han, 2009). As will be shown in the remainder of this chapter, present-day SLA research has shed substantive light on
these questions.
Intra-learner differential failure: Selective fossilization
The crux of intra-learner differential failure or success, as explicated earlier, is that
within any given L2 learners interlanguage, success and failure co-exist. This is
tantamount to saying that some parts of the system may develop fully, and other
parts may develop only partially. Focusing on the linguistic domain of morphosyntax, Hawkins (2000) defines selective fossilization as such:
I will refer to cases ... where morphosyntactic properties of the target language are
not used by L2 speakers in the same way as native speakers (but others are), even
after long immersion, as selective fossilization. (p. 76)

As shown earlier, there is converging evidence from longitudinal case studies that
attests to fossilization being local and selective. Local here means that fossilization does not affect individuals entire IL system; rather, it occurs only in its subsystems. Selective suggests that only certain linguistic properties are prone to
fossilization (e.g., Han, 2004, 2009, 2011, 2013; Han & Odlin, 2006; Hawkins, 2000;
Lardiere, 2012; Sorace, 2011). Additionally, fossilization is found to be idiosyncratic. As discussed in Han (2013), the idiosyncrasy can manifest itself in several
ways (see also Odlin this volume). First, fossilization can occur in learners under
different circumstances. Second, it can vary in its target and scope across learners.
Third, the factors leading to fossilization may not all be the same for individual
learners. Fourth, the timing of fossilization can be varied for individuals. Finally,
fossilization can differentially affect the interlanguage systems of learners who are
under similarly propitious learning conditions. All these, in effect, point to a more
general property, that fossilization is highly contingent on the interaction between
learner internal and external factors.
Speculations on fossilization abound (Han, 2004), but systematic explanatory
endeavors are fairly uncommon, and integrative attempts are even fewer (see,
however, Sorace, 2011). Most of the current explanations are discrete, confined to
the linguistic properties in question. Regarding the selective attainment of Patty,

Chapter 3. Revisiting the construct of fossilization

for example, three explanations have been proposed from the generativist paradigm (Hawkins, 2000). The first is Lardieres (1998b) Morphological Misreading
Hypothesis, which postulates that selective fossilization stems from a failure of the
language faculty to convert the fully feature-specified output of the syntacticcomputational component to morphological forms (p. 20). The second is Prvost
and Whites (2000) Missing Surface Inflection Hypothesis, according to which
computational deficits may prohibit retrieval of the appropriate variant of a lexical
entry. The third is Hawkins and Chans (1997) Failed Functional Features Hypothesis, which attributes lack of attainment to permanent representational deficits. It
is claimed that if certain grammatical features are not instantiated during first language acquisition or in the early years of life, they will never be available to enter
the L2 grammar.
All three hypotheses are predicated on the premise that Universal Grammar is
still available in SLA, directly or indirectly, but they, nevertheless, differ in whether they attribute selective fossilization to processing or representational deficits. In
other words, they view fossilization as a result of lack of processing or representation of a given morphological element. For the purposes of this chapter, it is relevant to point out also that all three hypotheses are aimed at accounting for a lack
of inflectional morphology in IL grammars, a rather limited focus. Selective
fossilization, in fact, goes beyond inflectional morphology. Indeed, what is challenging for any explanatory and/or predictive attempt is its rather broad scope,
including not only morpho-syntactic properties but also discourse properties such
as information structure.
Selective fossilization of discourse properties
One discourse element that has received increasing attention in L2 research in
recent years is topicalization. Topicalization is the process by which a speaker
signals that a constituent or segment of an utterance constitutes its topic
(Donaldson, 2012a, p. 651). Different languages may invoke different syntactic,
lexical, morphological, or phonetic means to realize topicalization in discourse. As
such, topicalization is essentially a syntax-discourse or syntax-pragmatics interface phenomenon. Theoretically, it is prone to fossilization. According to the Interface Hypothesis3 (Sorace, 2011; Sorace & Filiaci, 2006), language structures
involving an interface between syntax and other cognitive domains are less likely
to be acquired completely than structures that do not involve this interface (p. 1).
It has also been claimed that grammar-external interfaces (e.g., syntax-discourse)
3. Soraces recent work has extended this hypothesis to both bilingual first language acquisition and the early stages of L1 attrition.

ZhaoHong Han

are harder to acquire than grammar-internal interfaces (e.g., syntax-semantics)

(White, 2009; see, however, Montrul, 2011; Slabakova & Ivanov, 2011). Syntaxdiscourse features are context-dependent and highly susceptible to cross-linguistic
influence (Sorace & Keller, 2005). Empirical studies have attested to (a) their
fossilizability in adult L2 acquisition (see, e.g., Belletti et al., 2007; Sorace, 2005;
Sorace & Filiaci, 2006); (b) their permeability4 in L1 attrition (Wilson, Keller, &
Sorace, 2009); and (c) their protracted indeterminacy in bilingual L1 acquisition
(Sorace & Serratrice, 2009; Tsimpli et al., 2004). After reviewing the aggregate evidence from a number of empirical studies, Sorace (2011) concludes that there is
sufficient evidence for important developmental differences between linguistic
structures that require conditions of a formal feature within the grammar, and
structures that require the integration of contextual factors (p. 9).
L2 studies on topicalization have confirmed its fossilizability. Hans (2000)
longitudinal case study shows that in spite of extended interaction with English
as the target language, the two Chinese-speaking participants persisted in using
IL pseudo-passives (e.g., The letter about graphic file has received) and overusing
the English passive construction (e.g., My reply will be sent to you following this
mail) to fulfill an L1-inspired topic-comment discourse function (see also
Schachter & Rutherford, 1979; Yip, 1995). By Li and Thompsons (1976) typology,
Chinese is a topic prominent language (e.g. Book, I read), whereas English is a
subject prominent language (e.g., I read the book). What was important about the
findings of Han (2000) was that with the typological disparity between English
and Chinese, transfer occurred implicitly and indirectly (cf. Hendriks, 2000;
Trvise, 1986). Hendriks (2000) reports similar findings in Chinese-French interlanguage, noting that transfer of clearly Chinese constructions is so very rare in
L2 French (p. 387).
In an investigation of French topic marking, in particular, the use of left dislocation, in (a) child L1 learners, (b) adult Chinese-speaking learners, and (c) adult
native speakers of French, Hendriks (2000) shows that adult learners were more
like child L1 learners than adult native speakers. For example, they both used dislocation for reintroducing referents and making new information old. However,
unlike the child L1 learners or the adult native speakers, the adult learners used
other forms as well to promote referents to the topic position (i.e., sentence-initial).
This latter pattern reflects the influence of topic expression in Chinese in which
topics consistently occupy sentence initial position. Hendriks concludes from his
data as well as from previous studies (e.g., Hickmann et al., 1996; Hickmann
& Hendriks, 1999; Hendriks, 1998) that functions of topic-promoting devices are
of a universal kind, whereas the forms are language-specific and that, functions
4. See Adjmian (1976) for discussion on the notion of permeability.

Chapter 3. Revisiting the construct of fossilization

being universal, adult learners can use their knowledge of these functions when
acquiring an L2, whereas children do not have access to this kind of functional
background information (p. 392). This study suggests that when the target construction is one of typological similarity, such as topic-marking in French and
Chinese, adult learners acquisition load can be reduced to the extent that it entails
only acquisition of forms and form-function distributions, since the functions of
the target construction are part of the learners prior knowledge. Hendriks notes:
In contrast to functions, forms used by adult learners of French do not all coincide with the target-language dislocated forms. The forms have clearly been taken
from the French input (no transfer of Chinese constructions occurs), but, given
the cluster of forms overlapping in discourse-organizational patterns, adults in
the learner varieties examined here switch between a number of available target language constructions, which are more or less accepted and more or less
standard. (p. 393)

It is interesting to observe that selectivity here manifests itself as success with the
functions but failure with form-function mapping, due, according to Hendriks
(2000), to there being multiple forms encoding similar functions in the target
language. Selectivity of this nature has similarly been reported for Japanese-speaking learners of French (see Trvisiol, 1996, cited in Hendriks, 2000). Similar to
Chinese, Japanese is topic-sensitive, as well as subject-sensitive (Li & Thompson,
1976). The particles wa and ga serve respectively as topic and subject markers.
Like Hendriks Chinese speakers, the Japanese speakers in Trvisiol (1996) had no
trouble with the functions of topicalization devices, yet they struggled with the
constraints on the various forms encoding the function in the TL.
Lack of acquisition of syntax-discourse features often shows up as a lack of
command of their distributional properties, with overuse and/or underuse as its
hallmarks (see, e.g., Belletti et al., 2007; Han, 2000; Montrul, 2006). Belletti et al.
(2007) show that their near-native participants, who were native-speakers of
English, overused overt-subject pronouns while underusing post-verbal subjects
in their L2 Italian oral production. Likewise, invoking production data, Bohnacker
and Rosn (2008) reveal that L1-Swedish advanced learners of German were not
quite able to grasp the discourse function of the German prefield (vorfeld) first
clausal position that links a main declarative clause to the prior discourse as a locus of focus information. While crosslinguistic similarity (both Swedish and
German are V2 languages) obviously conferred some advantage on the learners, as
seen in their acquisition of the target language V2 syntax, it did not seem to help
much with the way they used the prefield. Again, the difference between the learners and the natives lay in the distribution of constituent types in the prefield.
Not all learners appear to equally have trouble with syntax-pragmatics features. Even where topicalization is concerned, full attainment does seem possible.

ZhaoHong Han

Donaldson (2012b), for example, reports that L1-English near-native speakers of

French performed within the native range in using left dislocation (e.g., Mariei,
ellei vient cet aprs-midi Mariei, sheis coming this afternoon) as a topic marker,
and in some cases, even surpassed the native controls, a phenomenon in and of
itself deserving independent investigation (see Larsen-Freeman this volume for
discussion of external norms). A closer look at these learners profiles shows longterm immersion in the target language (mean length of residence in France = 18.61
years). The participants (N = 10) all self-claimed very high-level mastery of
French, and, according to Donaldson (2012b), they were the most accomplished
speakers from his initial pool of approximately 20 candidates. Their L1, English,
is typologically similar to French with regard to the linguistic structure in question. Moreover, the structure is of high frequency in French, especially in informal
use. All these may have contributed to the reported high attainment. The overall
success notwithstanding, close inspection of the results at the level of individuals
yielded local and idiosyncratic fossilization. For instance, participant A5, in
spite of more than 14 years of immersion in France, still used left dislocation to
mark a brand new unanchored referent as a topic (see Table 10 in Donaldson,
2012b).5 This, however, does not change the fact that success at the syntax-discourse interface is possible.
Additional evidence that syntax-discourse features are not universally challenging for L2 learners is found in Hopp (2009). The study, focusing on scrambling
in sentence-medial position, a feature of German that serves a distinct discourse
function, shows that L2 learners and native speakers converged in their performance, as measured by an acceptability judgment task and a self-paced reading
task. But then it can be readily noted that the study employed receptive tests to
probe judgments on grammaticality and acceptability or comprehension. It is,
therefore, unknown if the convergence would have held up, had the measurement
tasks been production-based and fully contextualized.
In sum, while, in Selinkers conception, fossilization concerns surface linguistic materials, longitudinal case studies and non-longitudinal studies on near-
natives appear to have pinned down as particularly vulnerable to fossilization two
types of constructions: grammatical functors and syntax-pragmatics interface features. In Han (2009, 2010, 2011), it is argued that the nature of the two types of
structures overlaps in two ways: a) in their connection to discourse pragmatic
constraints (fossilizable structures tend to be those that encode complex form,

5. This observation apparently was deemed negligible when the author of the study concluded
that Hearer-new information (brand new anchored and unanchored) was virtually absent from
the LDs in the corpus (p. 420).

Chapter 3. Revisiting the construct of fossilization

meaning, and function relations6) and b) in their susceptibility to crosslinguistic

interference. The two overlapping features often go hand-in-hand (Han, 2013;
Sorace & Serratrice, 2006), suggesting that fossilizable structures tend to be L1-L2
(polarized) contrasts or language-specific features (see Odlin this volume). Thus,
L1 transfer appears to be a major factor leading to fossilization (cf. Selinker &
Lakshmanan, 1992).
Selinkers (1972) belief based on his informal observations, now corroborated
by much of extant empirical research, is that fossilization shows up under an intersection of two sets of conditions: (a) when the learner is attempting meaningful
production, and (b) when s/he is extremely relaxed or nervous, the psychological
states correlating with lack of attention to form. (Sociolinguistic research has
identified these same conditions under which the vernacular speech variety is
used, a point to which I will return.) According to the Single Resource Model of
Attention (Skehan, 1998), there is a trade-off between attention to form and meaning. When attention is allocated to form, learner output will be more accurate but
less fluent; conversely, when attention is allocated to meaning, the output will be
more fluent but less accurate. Still, the question remains: Why under those two
sets of conditions do learners typically fall back on an L1-based linguistic expression? What does this say about the nature of interlanguage? I will return to this
issue later on.
Idiosyncrasy and theoretical paucity
As noted earlier, extant research has found fossilization to be idiosyncratic. Not
every learner is equally susceptible to fossilization; by the same token, not all constructions allegedly fossilizable are of equal fossilizability. Thus, while English
near-native speakers may fossilize in assigning an anaphor to ambiguous subject
pronouns in their L2 Italian (Sorace & Filiaci, 2006), they may not in using French
left dislocation as a topic marker (Donaldson, 2012b). It follows that a theory of
fossilization (and SLA, for that matter) ought to be able to predict and account for
such ostensibly incoherent facts. Relatedly, it would be desirable for such a theory
to provide a concrete analytic tool, one that can be used both in a posteriori analysis of learner data as well as a priori analysis of target constructions. An additional
desirable feature of such a theory would be a capacity to explain both success and
failure, or both acquisition and fossilization.

6. Not all grammatical morphemes encode complex form, meaning, and function relations.
An example is the third person singular -s in English, a purely formal feature. There is no evidence in the literature that this grammatical morpheme is fossilizable.

 ZhaoHong Han

SLA theories of such scope and capacity are scant. Current theories, many of
them outsourced from other fields and disciplines, are typically long on explaining
success, but short on explaining failure. One notable exception, however, is the
emergentist approach (Ellis, 2007), according to which learning is constructionbased, rational, exemplar-driven, emergent, and dialectic (e.g., Ellis, 2006, 2007,
2008), but, essentially, regulated by attributes of input in the environment. The
learners mind is likened to that of a statistician, implicitly counting the tokens of
constructions (i.e., units of form-meaning mapping). In Elliss (2012) words,
frequency is a key determinant of acquisition because rules of language, at all
levels of analysis from phonology, through syntax, to discourse, are structural
regularities which emerge from learners lifetime unconscious analysis of the distributional characteristics of the language input (p. 261). Frequency leads to perceptual salience, strength of mental representation, retention, and ease of access.
Frequency induces learner-internal processes such as comparison, categorization,
abstracting generalities in form-meaning relationships, and the strengthening of
associations between forms and meanings (Ellis, 2002).
In this view, success of child L1 acquisition is fundamentally a function of
extensive exposure to large amounts of input. Yet, recognizing a qualitative difference in ultimate attainment between child L1 and adult L2 acquisition, the emergentist approach posits that in SLA, the input-driven process can be compromised
by learned attention, overshadowing, and blocking (Ellis, 2006, 2008). Ellis (2006)
explicates how learned attention or entrenchment of first language experience may
overshadow an L2 learners perception of input and block associative learning to
the extent that the learner becomes insensitive to the input cues and/or consistently misanalyzes the input, guided by L1 form-meaning mappings. Simply put,
L1 transfer may get in the way of an otherwise robust process of learning from
input powered by associative learning mechanisms.
The emergentist approach, thus, singles out two main factors in SLA, underscoring input as the driver of SLA and L1 the source of hindrance. However, this
theoretical approach does not provide specific explanations for inter-learner and
intra-learner differential success, a gap the Selective Fossilization Hypothesis
(Han, 2009) began to fill.
The Selective Fossilization Hypothesis
Similar to the emergentist approach, the Selective Fossilization Hypothesis (SFH;
Han, 2009) posits that input and L1 are the driving factors in SLA. Yet, unlike the
emergentist approach, the SFH casts L1 both as a facilitating and a debilitating
factor. Essentially, the SFH seeks to capture the interaction between input and L1

Chapter 3. Revisiting the construct of fossilization

(cf. Andersen, 1983; Kellerman, 1995) and hence the contingent nature of SLA, the
goal being to account for and predict intra-learner (and inter-learner, to a lesser
extent)7 differential success.
In the SFH, the input variable is expressed in terms of robustness, which is, in
turn, determined by frequency and variability (i.e., consistency). While frequency
refers to the number of times a given form appears in the input, variability in this
context8 concerns the form-meaning-function relation intrinsic to that form.
Thus, robust input would be [+frequent] and [variable], whereas non-robust input would be [frequent] and [+variable]. Figure 1 illustrates the variability dimension of input robustness, and Table 1 gives an example of non-robust input,
wherein multiple forms (i.e., disappeared, have been disappeared, were disappeared, and would have been disappeared) were used to encode the same sense,
at least from a target, prescriptive perspective.9

One form encoding one

meaning

Robust

One form encoding multiple

meanings in multiple
contexts
One form encoding multiple
meanings in a single
context
Multiple forms encoding the
same or similar meanings

Non-robust

Figure 1. The variability dimension of input robustness

7. To date, inter-learner differential success has been explained mostly in terms of individual
differences in cognitive and socio-psychological terms. The SFH, on the other hand, explains
them in terms of interaction between input and L1. Following the SFH, the learning outcome
differs, depending on the interactive configuration of the two.
8. The notion of variability as employed in the context of the SFH is different from how it is
employed in variationist SLA and sociolinguistic research where variability is discussed in quantitative terms (see Ortega, this volume; Tarone, this volume).
9. Carroll (2013) differentiates meaning into reference (i.e., the meaning of a form in isolation derived from its association with a tangible referent, as in a word) and sense (i.e., meaning
derived from the form in context in concert with its surrounding elements, as in a phrase or a
sentence).

ZhaoHong Han

Table 1. Multiple forms encoding the same meaning

Source: The New York Times 1/23/2010
Title: A Serene Advocate for Chiles Disappeared
...
She never again saw her husband, or her sons and daughter-in-law, nor did she hear a word
about their fates. All four are believed to have been disappeared by the Pinochet regime,
which came to power in a bloody 1973 coup that claimed the life of Chiles Socialist president,
Salvador Allende.
In the 34 years that followed, Mrs. Gonzlez transformed her outrage and grief into a tireless
advocacy for answers about the estimated 3,000 people who were killed or disappeared under
the Pinochet dictatorship from 1973 to 1990.
The day after her husband disappeared, Mrs. Gonzlez found an anonymous note at her home
that left little doubt that he had been seized by the regime.
To this day Mrs. Gonzlez feels blessed but sad to consider what might have been. If the child
and I had left with my husband that day, I also would have been disappeared, she said,
dragging deeply on a cigarette.
SHE was on the front lines, showing tremendous courage, said Jos Miguel Vivanco, the
Americas director for Human Rights Watch. Without her courage, more people probably
would have been disappeared, and the national attention to this would have been close to
zero.

Turning now to the other of the two cardinal variables in the SFH, the L1 variable is formulated in terms of markedness, which, similar to the input robustness variable, subsumes two sub-variables: frequency and variability. Thus, a
marked L1 construction would be one that is [frequent] and [+variable], while
an unmarked L1 construction would be one that is [+frequent] and [variable].10
Intersecting the two cardinal variables, each being a continuum, results in four
zones, among them an acquisition zone (IV) and a fossilization zone (II), as illustrated in Figure 2.
Zones I and III are less clean-cut, where it is hypothesized that input and L1
factors can be overridden by individual difference variables such as memory
and sensitivity.11 Thus, in these two zones, either acquisition or fossilization
may prevail for a given interlanguage construction. The concentric circles in
Figure 1 represent gradation, with the inner circles connoting lesser than the
outer ones.
10. A synonym to variability in the SFH context is consistency. Thus, a marked L1 construction would be one that is infrequent and inconsistent, while an unmarked construction is one
that is frequent and consistent.
11. Long (2003) argues that individual learners sensitivity interacting with the perceptual
salience of input (i.e., input attributes such as frequency, communicative value, and semantic
weight) may lead to stabilization or fossilization of some structures.

Chapter 3. Revisiting the construct of fossilization

Robust (L2)

The acquisition zone

II

Marked (L1)

Unmarked (L1)
III

IV
The fossilization zone
Non-robust (L2)

Figure 2. The Selective Fossilization Hypothesis (Han, 2009)

Importantly, the SFH does not presume a one-to-one correspondence cross-linguistically. Rather, the determination of what is marked, unmarked, or anything in between for the L1 markedness variable is made according to the nature of the L1
counterpart expression of the target construction (i.e., the presence or absence
thereof) along with its frequency and variability. By way of an example, consider
English and Chinese with respect to the article system: English is a [+article] language, while Chinese is [article]. In English, the articles are frequent but pragmatically variable to a large extent. The counterpart of the English articles in Chinese is
zero article, and its usage is unmarked: frequent and invariant. Based on these properties, the SFH predicts that the English articles would border Zone 4, and are, therefore, susceptible to fossilization in Chinese-English interlanguage. This prediction
was born out in longitudinal case studies (e.g., Han, 2009; Lardiere, 2007). Tracking
the natural written production of an endstate Chinese-speaking learner of English
over a span of five years, Han (2009) documents the informants incomplete article
marking stabilizing at around 84% for the indefinite article and 67% for the definite
article (see, however, Lardiere [2007]). The discrepancy between the definite and
indefinite articles suggests intra-learner differential success. Consider as another example the adverb placement feature of French against English. French allows an
adverb wedged between a verb and its direct object and hence the word order of
SVAO (e.g., Jean embrasse souvent Marie *John kisses often Mary), but English
does not. English, however, allows SAV, which French does not (*Jean souvent embrasse Marie John often kisses Mary). Both constructions are unmarked respectively. Thus, for francophones learning English, the input for SAV would be robust,

ZhaoHong Han

and its L1 counterpart would be marked since it does not exist in French. As such,
SAV would fall in the acquisition zone (see Figure 1). SVAO, on the other hand,
would fall in the fossilization zone, since there is no input for it (i.e., the input is nonrobust) and it is unmarked in the L1. Research indeed shows that while SAV is learnable, SVAO is both persistent and resistant in French-English interlanguage (Sheen,
1980; Trahey & White, 1993; White, 1989, 1991).
Both examples above illustrate that fossilization is (a) construction-specific,
(b) learner-specific (i.e., it happens in some learners but not others), and (c) language-specific (L1-L2 pairing). Such high contingency stems fundamentally from
an interaction between the strength of L2 input and that of L1, which varies in (a)
through (c). A critical assumption of the SFH is that input is not isomorphic with
the target language. The SFH holds that the target language is relatively stable
(see, however, Larsen-Freeman, 2006, this volume), but input is precarious (for
discussion, see Han, 2011, 2013).
Applying the SFH to a host of reportedly fossilizable constructions from different L1-L2 pairings, Han (2013) reveals two facets of a symbiotic relation, subtle
yet significant, between L1 markedness and L2 input robustness, mediated by the
nature of the target structure. First, when the target structure is variable in formmeaning-function mapping, such as syntax-discourse interface constructions, the
input tends not to be robust and transfer of an L1 unmarked usage is likely to
sneak in, in which case L1 is resorted to as a solution to a problem (p. 161). Second, non-robust input may induce L1 transfer, but, by the same token, L1 influence may skew the perception of otherwise robust input. In brief, the interaction
between L2 input and L1 transfer can be initiated either way, from L2 input to L1
transfer or vice versa (for discussion, see Han, 2013).
The SFHs proposition, empirically backed, that an unmarked L1 usage may
bear on L2 acquisition implies that its effects should be apparent when learners
attempt L2 production, especially in spontaneous production, wherein they are
preoccupied with creating and expressing their own meanings. Under those circumstances, it is conceivable that learners would have little time to monitor their
production; instead, they would rely on their thinking for speaking (Slobin,
1996), a form of cognition mobilized for and during communication. In Slobins
(1987) terms, In the evanescent timeframe of constructing utterances in discourse, one fits ones thoughts into available linguistic forms (p. 435). In L2 production, the most readily available linguistic forms are highly likely to be the L1
unmarked constructions because of their frequent and habitual use resulting in
neural entrenchment. The pseudo-passives discussed earlier from Han (2000) occurred more frequently in L2 informal than in formal writing. Likewise, the although/but construction in Chinese-English interlanguage (e.g., Though Im not a
Chinese major, but my Chinese is excellent), a direct copy of a construction in the

Chapter 3. Revisiting the construct of fossilization

L1 Chinese, appears mostly in spontaneous speech (Han & Lew, 2012). The s/he
pronominal conflation never occurs in writing, but often in spontaneous speech,
in native speakers of Chinese.
Unmarked usages have high accessibility, lending themselves well to spontaneous production. According to Slobin (2007), in each language there are semantic
elements that are habitually encoded either by grammatical means (morphological
elements, construction types) or obligatory lexemes (or non-encoded). Habitually
encoded semantics tends to have higher codability (ease of expression of the relevant categories), which often means higher accessibility, and transferability as well
in the context of second language use. Slobin (1996) insightfully notes:
Each native language has trained its speakers to pay different kinds of attention to
events and experiences when talking about them. This training is carried out in
childhood and is exceptionally resistant to restructuring in adult second-language
acquisition. (p. 89)

The phenomenon of subconsciously deploying, out of necessity, L1 thinking for

speaking during L2 production is discussed in Han and Lew (2012) under L1
thinking for L2 speaking. Adapting from Levelts (1989) speech production model,
Han and Lew posit that when L1 thinking for L2 speaking occurs, the operations
within the Conceptualizer are carried out and framed in the L1, resulting in an
L1-packaged message. This message will then be encoded using L2 grammatical
and phonological means, hence L1 thinking for L2 speaking.
That interlanguage may exhibit very different features when the learner is engaged in a communicative task versus a non-communicative task has long been
observed and theorized. Selinker (1972) was the first to note the discrepancy, arguing that meaningful interlanguage data could only come from learners communicative production. Selinker and Lamendella (1978) assert that fossilization is
set in motion whenever the learner attempts to express his own meanings in the
L2. Selinker and Douglas (1985) demonstrate that interlanguage varies according
to the discourse context. Yet, it was Tarone (1979, 1983) who first theorized about
IL variability. Relating Labovs Observers Paradox to interlanguage, Tarone (1979,
this volume) posits a number of axioms, including one that states in the vernacular style, where the minimum amount of attention is given to speech, the
most regular and systematic phonological and grammatical patterns are evidenced.
Other styles tend to show more variability (emphasis added). This implies that
when learners are engaged in spontaneous self-expression, they are subserved by
a systematic, if sometimes systematically variable, interlanguage system, which is,
in part, based in the L1. This interpretation is bolstered by extensive evidence
from fossilization research (L1 transfer is embedded in each and every instance of
fossilization documented) and by the L1-thinking-for-L2-speaking argument. If

ZhaoHong Han

we were to take the axiom seriously, it would raise a host of theoretical and
empirical questions about the nature of interlanguage knowledge, including but
not limited to the nature of explicit knowledge and its relation to second language
use, and most profoundly, the competence versus performance dichotomy which
has dominated SLA research for decades. Learners spontaneous production
would likely resume its central status in SLA research (see, however, Gass, 2009;
Gass & Polio, this volume), and, accordingly, researchers would need to be mindful of the Observers Paradox when pursuing an understanding of the default
interlanguage system.
Theoretical controversies and the future of fossilization research
Fossilization has become widely accepted as a psychologically real phenomenon
of considerable theoretical and practical importance (Long, 2003, p. 487). Reviewing four decades of research since Selinker (1972), Han (2013) concludes with
four robust findings:
1. Fossilization is selective.
2. Fossilization affects the acquisition of TL structures encoding variable form,
meaning, and function (discourse pragmatics) relations.
3. Fossilization is inspired by an L1-relativized mind,12 induced or reinforced by
L2 input attributes.
4. Fossilization is most evident in spontaneous production in which the learner
engages in manufacturing his own meaning and linguistic expression. (p. 165)
Still, it must be noted that the construct of fossilization is not uncontroversial. From various perspectives, critics have challenged its validity and utility
and, in some cases, put forth alternatives. Birdsong (2006), for one, claims that
fossilization has for far too long served as a lynchpin in SLA research, suggesting that it is time to consider the complementary question of learner potential in
late SLA, and more specifically, to ascertain the upper limits of SLA. Without
a clear mapping-out of the upper limits of attainment, researchers are deprived
of key points of reference in their exploration of constraints on learning
(Birdsong & Paik, 2008, p. 425). Birdsong and Paik adduce evidence from more
than 20 behavioral studies of a higher rate of success, 3 to 45 percent, at the L2
endstate than has traditionally been estimated (see, e.g., Long, 1990), thus mitigating the standard claim that late L2A is failure-ridden (Bley-Vroman, 1989;
Long, 1990; Selinker, 1972, among others).
12. The mind is tuned to the L1.

Chapter 3. Revisiting the construct of fossilization

Larsen-Freeman (2006, this volume; see also Ortega, this volume), on the
other hand, takes exception to a target perspective in second language research,
arguing against the static view of language and language learning underlying the
concept of fossilization and conception of success as conformity to (monolingual) native speaker norms. Instead, she advocates the view that language,
interlanguage included, is forever fluid, asserting that there is no end and there
is no state (2006). Success, it follows, should not be measured against any idealized target; rather, it should be gauged in its own terms (cf. Bley-Vroman, 1989;
Cook, 1999).
Delving into the initial conceptualization of fossilization as laid out in Selinker
(1972) and tracing its evolution in SLA research, Long (2003) points out its
ambiguity, noting that fossilization is alternately explanandum and explanans
(p. 487). Moreover, concerning fossilization as explanandum, Long takes issue
with variability, rejecting the possibility that there can be fossilized variation
(see, however, Han, 2004; Lardiere, 2007; Schachter, 1996).13 In view of these perceived problems, along with the methodological shortcomings he observes in
previous empirical research, Long recommends a shift of attention away from fossilization to the well-attested phenomenon of stabilization (p. 487; see, however,
discussion in Han, 2004, 2011; Han & Finneran, 2013).
Recently, writing about the phenomenon of fossilization, Lardiere (2012)
claims that the term is redundant in the light of the body of research on ultimate
attainment. Her reasoning is that fossilization, by virtue of its presumed permanence, is a form of non-nativelike endstate, and, as such, it is synonymous with
ultimate attainment.
Criticisms such as the ones noted above raise concerns that merit the fields
attention. And yet, it should be pointed out that some of the concerns are in themselves controversial (Han, 2004, 2011). For example, the view that identifies fossilization with ultimate attainment does not seem to be predicated on an adequate
understanding of fossilization, nor of ultimate attainment, each having its own set
of concerns. Conflating fossilization and ultimate attainment, in effect, obfuscates
their respective heuristic value, given that they are essentially different, albeit related, beasts. For one thing, while both terms denote a form of endstate, fossilization as a phenomenon is local, whereas ultimate attainment speaks to a global state
(Han, 2004, 2011, 2013). As Long (2003) surmises, and as Lardieres own study has
amply attested, if fossilization occurs, it operates locally, not globally throughout
an IL. Fossilization would not simply be the same thing as global non-nativelike L2
13. Fossilization was tied from the beginning (Selinker 1972) to variability, by virtue of it being
isomorphic with backsliding, i.e., occasional reappearance of errors thought to have been
eradicated.

ZhaoHong Han

attainment by adult starters, in other words (Long, 2003, p. 512). Ultimate attainment, as advised in Birdsong and Paik (2008), is properly used in a neutral sense
in reference to the outcome of second language acquisition (L2A), irrespective of
whether this outcome is similar to or different from nativelikeness. ... ultimate
attainment, endstate attainment, and asymptotic attainment are often freely substituted (p. 424). In a similar vein, White (2003) delineates three broad scenarios
of L2 ultimate attainment: native-like, non-native-like, and partially native-like.
In terms of their timing, fossilization and ultimate attainment are again at odds:
while ultimate attainment is putatively the terminal asymptotic state of L2A, fossilization arguably can occur at any point throughout the developmental process.
Hence, it is something expected in learners who are ab initio, en route, or al fine.
Importantly, fossilization occurs alongside acquisition. Therefore, as Long has justifiably argued, empirical proof of fossilization as a local phenomenon ought to
entail both failure and success. In other words, just as it is necessary to show that
certain constructions have ceased to develop, so is it to demonstrate that other constructions are simultaneously converging on the intended target (cf. Han, 2011).
A third difference between ultimate attainment and fossilization is that, epistemologically, fossilization research emanates from the Interlanguage Hypothesis
(Selinker, 1972), while research on ultimate attainment originates in concerns
with maturational (e.g., Johnson & Newport, 1989, 1991) and Universal Grammar (e.g., Hopp, 2004, 2010; Sorace & Filiaci, 2006; Sorace & Serratrice, 2006)
effects in SLA (Coppieters, 1987; Birdsong, 1992). Relatedly, the methodologies
employed in the respective domains are largely different: typically case studies on
fossilization versus cross-sectional studies on ultimate attainment, though recent
research has seen a crossover, as longitudinal case studies have been undertaken
in learners with long-term immersion in the target language (Han, 2000, 2006,
2011; Lardiere, 1998a, 1998b, 2007; White, 2003). Last but not least, research on
ultimate attainment is in the main concerned with inter-learner differences (Birdsong & Paik, 2008), whereas research on fossilization is also concerned with intra-learner differences.
In a recent handbook of SLA (Gass & Mackey, 2011), fossilization is, for the
first time, classified as an individual learner variable. Indeed, as argued throughout
this chapter and elsewhere (Han, 2009, 2011, 2013), fossilization is largely idiosyncratic, tied to intra-learner differential success or failure wherein lies also the heuristic value of the construct. Continued, systematic research on fossilization will
hold much promise of illuminating two long-standing conundrums, one being
intra-learner variability and the other the widely noted yet poorly understood discrepancy between received (as from classroom instruction though see Van
Patten, this volume) and receptive knowledge (i.e., knowledge that drives comprehension), on the one hand, and productive knowledge (i.e., knowledge that drives

Chapter 3. Revisiting the construct of fossilization

communicative production), on the other, even in advanced L2 users (for a recent

discussion, see Han & Finneran, 2013). Equally, by its longitudinal nature, research
on fossilization, not on ultimate attainment, is likely to benefit second language
instruction by providing for a robust understanding of developmental artifacts
and their etiology. Armed with that understanding, teachers would likely be more
able to properly allocate their instructional resources than without such knowledge, thereby increasing instructional efficacy.
Conclusion
This chapter revisited the construct of fossilization as theorized in Selinker (1972),
focusing, in particular, on the issue of selective fossilization. While early research
on fossilization essentially neglected intra-learner success or failure, recent decades have seen the void being gradually filled, both conceptually and empirically.
As a result, the general understanding of fossilization has grown to be more nuanced and sophisticated. The tension, however, appears to have lingered on, between a failure-driven and a success-driven approach to SLA research. On this
note, it is important to point out that fossilization research is less about revealing
deviances from the presumed norm than about resolving a dual cognitive conflict,
namely, why is it that in spite of propitious conditions, development is cut short in
some areas? And why is the developmental interruption made most apparent
when learners attempt self-expressions (i.e., meaning-based production) in the
target language? Ultimately, what does it say about the nature of interlanguage, in
mind, brain, and behavior?
Acknowledgments
I am grateful to Larry Selinker for his guidance in my doctoral study and for instilling in me the importance of pursuing a deep understanding of interlanguage.
I thank the reviewers for their very helpful comments on an earlier draft. Any errors that remain are solely my responsibility.
References
Adjmian, C. (1976). On the nature of interlanguage systems. Language Learning, 26, 297320.
Andersen, R.W. (1983). Transfer to somewhere. In S. Gass & L. Selinker (Eds.), Language
Transfer in Language Learning (pp. 177201). Rowley, MA: Newbury House.

ZhaoHong Han
Belletti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax
of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory, 25,
657689.
Birdsong, D. (1992). Ultimate attainment in second language acquisition. Language, 68, 706755.
Birdsong, D. (2006). Why not fossilization. In Z.-H. Han (Ed.), Studies of Fossilization in Second
Language Acquisition (pp. 173188). Clevedon: Multilingual Matters.
Birdsong, D., & Paik, J. (2008). Second language acquisition and ultimate attainment. In B.
Spolsky & F. Hult (Eds.), Handbook of Educational Linguistics (pp. 424436). Oxford:
Blackwell.
Bley-Vroman, R. (1989). The logical problem of second language learning. In S. Gass & J. Schachter
(Eds.), Linguistic Perspectives on Second Language Acquisition (pp. 4168). Cambridge: CUP.
Bley-Vroman, R. (2009). The evolving context of the fundamental difference hypothesis. Studies
in Second Language Acquisition, 31(Special Issue 02), 175198.
Bohnacker, U., & Rosn, C. (2008). The clause-initial position in L2 German declaratives. Studies in Second Language Acquisition, 30, 511538.
Carroll, S.E. (2013). Introduction to the special issue: Aspects of word learning on first exposure
to a second language. Second Language Research, 29(2), 131144.
Cook, V. (1999). Going beyond the native speaker in language teaching. TESOL Quarterly, 33(2),
185209.
Coppieters, R. (1987). Competence differences between native and non-native speakers.
Language, 63, 544573.
Donaldson, B. (2012a). Topicalization. In P. Robinson (Ed.), The Routledge Encyclopedia of
Second Language Acquisition (pp. 651652). London: Routledge.
Donaldson, B. (2012b). Syntax and discourse in near-native French: Clefts and focus. Language
Learning, 62(3), 902930.
Ellis, N.C. (2002). Frequency effects in language acquisition: A review with implications for
theories of implicit and explicit language acquisition. Studies in Second Language Acquisition, 24(2), 143188.
Ellis, N.C. (2006). Selective attention and transfer phenomena in SLA: Contingency, cue competition, salience, interference, overshadowing, blocking and perceptual learning. Applied
Linguistics, 27, 164194.
Ellis, N. C. (2007). SLA: The associative cognitive CREED. In B. VanPatten & J. Williams (Eds.),
Theories in Second Language Acquisition: An Introduction. . Cambridge: CUP.
Ellis, N.C. (2008). The dynamics of second language emergence: Cycles of language use, language change, and language acquisition. The Modern Language Journal, 92(2), 232249.
Ellis, N.C. (2012). Frequency. In P. Robinson (Ed.), The Routledge Handbook of Second Language
Acquisition (pp. 260265). New York, NY: Routledge.
Fries, C. (1952). The Structure of English: An Introduction to the Construction of English Sentences. New York, NY: Harcourt Brace.
Gass, S. (2009). A survey of SLA research. In B. Ritchie & T. Bhatia (Eds.), The New Handbook of
Second Language Acquisition (pp. 328). Bingley, UK: Emerald.
Gass, S., & Mackey, A. (Eds.). (2011). The Routledge Handbook of Second Language Acquisition.
New York, NY: Routledge.
Han, Z.-H. (2000). Persistence of the implicit influence of NL: The case of the pseudo-passive.
Applied Linguistics, 21(1), 78105.
Han, Z.H. (2004). Fossilization in Adult Second Language Acquisition. Clevedon: Multilingual
Matters.

Chapter 3. Revisiting the construct of fossilization

Han, Z.-H. (2006). Fossilization: Can grammaticality judgment be a reliable source of evidence?
In Z.-H. Han & T. Odlin (Eds.), Studies of Fossilization in Second Language Acquisition
(pp. 5682). Clevedon: Multilingual Matters.
Han, Z.-H. (2009). Interlanguage and fossilization: Towards an analytic model. In V. Cook &
L. Wei (Eds.), Contemporary Applied Linguistics, Vol. I: Language Teaching and Learning
(pp. 137162). London: Continuum.
Han, Z.-H. (2010). Grammatical morpheme inadequacy as a function of linguistic relativity: A
longitudinal study. In Z.-H. Han & T. Cadierno (Eds.), Linguistic Relativity in Second Language Acquisition: Thinking for Speaking (pp. 154182). Clevedon: Multilingual Matters.
Han, Z.-H. (2011). Fossilization -A classic concern of SLA research. In S. Gass & A. Mackey
(Eds.), The Routledge Handbook of Second Language Acquisition (pp. 476490). New York,
NY: Routledge.
Han, Z.-H. (2013). Forty years later: Updating the Fossilization Hypothesis. Language Teaching,
46(2), 133171.
Han, Z.-H., & Finneran, R. (2013). Re-engaging the interface debate in SLA: Strong, weak, none
or all? The International Journal of Applied Linguistics. DOI: 10.1111/ijal.12034
Han, Z.-H., & Lew, W.M. (2012). Acquisitional complexity: What defies complete acquisition in
SLA. In B. Szmrecsanyi & B. Kortmann (Eds.), Linguistic Complexity in Interlanguage Varieties, L2 Varieties, and Contact Languages (pp. 192217). Berlin: Walter de Gruyter.
Han, Z.-H., & Odlin, T. (2006). Introduction. In Z.-H. Han & T. Odlin (Eds.), Studies of
Fossilization in Second Language Acquisition (pp. 120). Clevedon: Multilingual Matters
Hawkins, R. (2000). Persistent selective fossilization in second language acquisition and the
optimal design of the language faculty. Essex Research Reports in Linguistics, 34, 7590.
Hawkins, R., & Chan, C. (1997). The partial availability of Universal Grammar in second language acquisition: The failed functional features hypothesis. Second Language Research, 13,
187226.
Hickmann, M., Hendricks, H., & Roland, F. (1996). The marking of new information in childrens narratives: A comparison of English, French, German, and Mandarin Chinese.
Journal of Child Language, 23, 591619.
Hickmann, M., & Hendricks, H. (1999). Reference maintenance in childrens narrative discourse. Journal of Child Language, 26, 419452.
Hendriks, H. (1998). Reference to person and space in narrative discourse: A comparison of
adult second language and child first language acquisition. Studi Italiani di Linguistica
Teoretica e Applicata, 27, 6786.
Hendriks, H. (2000). The acquisition of topic marking in L1 Chinese and L1 an L2 French.
Studies in Second Language Acquisition, 22, 369397.
Hill, A. (1955). Introduction to Linguistic Structures. New York: Brace Harcourt.
Hopp, H. (2004). Constraining L2 word order optionality: Scrambling in advanced English German and Japanese-German interlanguage. Second Language Research, 25(1), 3471.
Hopp, H. (2009). The syntax-discourse interface in near-native SLA: Off-line and on-line performance. Bilingualism: Language and Cognition, 12, 463483.
Hopp, H. (2010). Ultimate attainment in L2 inflection: Performance similarities between nonnative and native speakers. Lingua, 120, 901931.
Ioup, G., Boustagui, E., El Tigi, M., & Moselle, M. (1994). Reexamining the critical period hypothesis: A case study of successful adult SLA in a naturalistic environment. Studies in
Second Language Acquisition, 16(1), 7398.

ZhaoHong Han
Johnson, J. S., & Newport, E. L. (1989). Critical period effects in second language learning: The
influence of maturational state on the acquisition of English as a second language. Cognitive
Psychology, 21(1), 6099.
Johnson, J., & Newport, E. (1991). Critical period effects on universal properties. Cognition, 39,
215268.
Johnson, J., Shenkman, K., Newport, E., & Medin, D. (1996). Indeterminacy in the grammar of
adult language learners. Journal of Memory and Language, 35, 335352.
Kellerman, E. (1995). Crosslinguistic influence: Transfer to nowhere? Annual Review of Applied
Linguistics, 15, 125150.
Lado, R. (1957). Linguistics across Cultures: Applied Linguistics for Language Teachers. Ann
Arbor, MI: University of Michigan.
Lardiere, D. (1998a). Dissociating syntax from morphology in a divergent L2 end-state grammar. Second Language Research, 14(4), 359375.
Lardiere, D. (1998b). Case and tense in the fossilized steady state. Second Language Research,
14, 126.
Lardiere, D. (2007). Ultimate Attainment in Second Language Acquisition. Mahwah, NJ: Lawrence
Erlbaum Associates.
Lardiere, D. (2012). Fossilization. In P. Robinson (Ed.), The Routledge Encyclopedia of Second
Language Acquisition (pp. 256260). New York, NY: Routledge.
Larsen-Freeman, D. (2006). Second language acquisition and the issue of fossilization: There is
no end, and there is no state. In Z.-H. Han & T. Odlin (Eds.), Studies of Fossilization in
Second Language Acquisition (pp. 189200). Clevedon: Multilingual Matters.
Lenneberg, E. (1967). Biological Foundations of Language. New York, NY: Wiley.
Levelt, W. J. M. (1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press.
Li, C., & Thompson, S. (1976). Subject and Topic: A New Typology of Language. New York, NY:
Academic Press.
Long, M. (1990). Maturational constraints on language development. Studies in Second Language
Acquisition, 12, 251285.
Long, M.H. (2003). Stabilization and fossilization in interlanguage development. In C. Doughty
& M. H. Long (Eds.), The Handbook of Second Language Acquisition (pp. 487535). Malden
MA: Blackwell.
Montrul, S. (2006). Incomplete acquisition as a feature of L2 and bilingual grammars. In R.
Slabakova, S. Montrul & P. Prvost (Eds.), Inquiries in Language Development. Studies in
Honor of Lydia White (pp. 335359). Amsterdam: John Benjamins.
Montrul, S. (2011). Multiple interfaces and incomplete acquisition. Lingua, 121(4), 591604.
Novoa, L., Fein, D., & Obler, L. (1988). Talent in foreign languages: A case study. In L. Obler &
D. Fein (Eds.), The Exceptional Brain: The Neuropsychology of Talent and Special Abilities
(pp. 294302). New York, NY: Guilford Press.
Obler, L. (1989). Exceptional second language learners. In S. Gass, C. Madden, D. Preston & L.
Selinker (Eds.), Variation in Second Language Acquisition: Psycholinguistic Issues. Clevedon,
UK: Multilingual Matters.
Prvost, P., & White, L. (2000). Missing surface inflection or impairment in second language
acquisition? Evidence from tense and agreement. Second Language Research, 16, 103134.
Schachter, J. (1996). Learning and triggering in adult L2 acquisition. In G. Brown, K. Malmkjaer & J. Williams (Eds.), Performance and Competence in SLA (pp. 7088). Cambridge:
CUP.

Chapter 3. Revisiting the construct of fossilization

Schachter, J., & Rutherford, W. (1979). Discourse function and language transfer. In B. Robinett
& J. Schachter (Eds.), Second Language Learning: Contrastive Analysis, Error Analysis, and
Related Aspects (pp. 111). Ann Arbor, MI: University of Michigan Press.
Schmidt, R. (1983). Interaction, acculturation, and the acquisition of communicative competence. In N. Wolfson & J. Manes (Eds.), Sociolinguistics and Second Language Acquisition
(pp. 137174). Rowley, MA: Newbury House.
Schneiderman, E., & Desmarais, C. (1988). The talented language learner: Some preliminary
findings. Second Language Research, 4, 91109.
Schumann, J. (1978). The Pidginization Process: A Model for Second Language Acquisition.
Rowley, MA: Newbury House.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10(2), 209231.
Selinker, L., & Douglas, D. (1985). Wrestling with context in interlanguage studies. Applied
Linguistics 6(2), 190204.
Selinker, L., Dumas, & Swain, M. (1975). The interlanguage hypothesis extended to children.
Language Learning, 25(1), 139152.
Selinker, L., & Lakshmanan, U. (1992). Language transfer and fossilization: The multiple effects
principle. In S. Gass & L. Selinker (Eds.), Language Transfer in Language Learning
(pp. 190216). Amsterdam: John Benjamins.
Selinker, L., & Lamendella, J. (1978). Two perspectives on fossilization in interlanguage learning. Interlanguage Studies Bulletin, 3(2), 143191.
Sheen, R. (1980). The importance of negative transfer in the speech of near-bilinguals. International Review of Applied Linguistics, 18(2), 105119.
Skehan, P. (1998). A Cognitive Approach to Language Learning. Oxford: OUP.
Slabakova, R., & Ivanov, I. (2011). A more careful look at the syntax-discourse interface. Lingua,
121(4), 637651.
Slobin, D. (1987). Thinking for speaking. Proceedings of the Berkeley Linguistics Society, 13, 435444.
Slobin, D.I. (1996). From thought and language to thinking for speaking. In J. Gumperz & S.
Levinson (Eds.), Rethinking Linguistic Relativity (pp. 7096). Cambridge: CUP.
Slobin, D. (2007). Language and thought online: Cognitive consequences of linguistic relativity.
In V. Evans, B. Bergen, & J. Zinken (Eds.), The Cognitive Linguistics Reader (pp. 902928).
London: Equinox.
Sorace, A. (2005). Selective optionality in language development. In L. Cornips & K. Corrigan
(Eds.), Syntax and Variation Reconciling the Biological and the Social (pp. 111160).
Amsterdam: John Benjamins.
Sorace, A. (2011). Pinning down the concept of interface in bilingualism. Linguistic Approaches
to Bilingualism, 1, 133.
Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second
Language Research, 22(3), 339368.
Sorace, A., & Keller, F. (2005). Gradience in linguistic data. Lingua, 115, 14971524.
Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language development: Beyond structural overlap. International Journal of Bilingualism, 13, 195210.
Sorace, A., & L. Serratrice (2006). Anaphora resolution in near-native speakers of Italian. Second
Language Research 22(3), 339368.
Tarone, E. (1979). Interlanguage as chameleon. Language Learning, 29, 181191.
Tarone, E. (1983). On the variability of interlanguage systems. Applied Linguistics, 4, 142163.
Towell, R., & Hawkins, R. (1994). Approaches to Second Language Acquisition. Clevedon, UK:
Multilingual Matters.

ZhaoHong Han
Trager, G., & Smith, H. L. (1951). An Outline of English Structure. Norman, OK: Battenburg
Press.
Trahey, M., & White, L. (1993). Positive evidence and preemption in the second language classroom. Studies in Second Language Acquisition, 15, 181204.
Trvise, A. (1986). Is it transferable, topicalization? In E. Kellerman & M. Sharwood Smith
(Eds.), Crosslinguistic Influence in Second Language Acquisition. New York, NY: Pergamon.
Trvisiol, P. (1996, May). Marques des processus de thmatisation en franais L2 dans le discours
narratif dapprenants japonais [Thematization process markers in L2 French in the narrative
discourse of Japanese learners]. Paper presented at the Sixth International EuroSLA Conference, Nijmegen.
Tsimpli, T., Sorace, A., Heycock, C., & Filiaci, F. (2004). First language attrition and syntactic
subjects: A study of Greek and Italian near-native speakers of English. . International Journal of Bilingualism, 8(3), 257277.
VanBuren, P. (2001). Brief apologies and thanks. Second Language Research, 17(4), 457462.
VanPatten, B., & Williams, J. (Eds.). (2007). Theories in Second Language Acquisition: An Introduction. Mahwah, NJ: Lawrence Erlbaum Associates.
Wang, T. Y. (2012). Selective Fossilization in Advanced L2 Speakers of Japanese. Unpublished
M.A. thesis.Teachers College, Columbia University.
White, L. (1989). The principle of adjacency in second language acquisition: Do L2 learners
observe the subset principle? In S. Gass & J. Schachter (Eds.), Linguistic Perspectives on
Second Language Acquisition (pp. 134158). Cambridge: CUP.
White, L. (1991). The verb-movement parameter in second language acquisition. Language Acquisition, 1, 337360.
White, L. (2003). Fossilization in steady state L2 grammars: Persistent problems with inflectional morphology. Bilingualism: Language and Cognition, 6, 129141.
White, L. (2009). Grammatical theory: Interfaces and L2 knowledge. In T.K. Bhatia & W.C.
Ritchie (Eds.), The New Handbook of Second Language Acquisition (pp. 4968). Bingley,
UK: Emerald.
Wilson, F., Keller, F., & Sorace, A. (2009). Antecedent preferences for anaphoric demonstratives
in L2 German Proceedings of BUCID 33. Somerville, MA: Cascadilla Press.
Yip, V. (1995). Interlanguage and Learnability: From Chinese to English. Amsterdam: John
Benjamins.

chapter 4

Interlanguage, transfer and fossilization

Beyond second language acquisition
Silvina Montrul

University of Illinois at Urbana-Champaign

Selinkers (1972) Interlanguage proposal lies at the heart of linguistic approaches
to adult second language acquisition, especially subsumed under the theory
of Universal Grammar. Much research in recent years has directly addressed
the what, how, and why of transfer and fossilization, yet these are not unique
to adult L2 learners, but can also be found in other bilingual situations. This
chapter discusses first how the generative linguistics perspective has contributed
to refining and clarifying Selinkers idea, while the second part takes issue
with Selinkers characterization of bilingual varieties and illustrates transfer
and fossilization in individual bilingualism more broadly defined. The chapter
concludes with questions for future directions to pursue a deeper understanding
of the interlanguage phenomenon beyond second language acquisition.

Introduction
Cognitive, mentalistic perspectives on second language acquisition (SLA) have
had a dominant role in the theoretical landscape in the last fifty years, where concepts like nonnative speaker, learner and interlanguage are pivotal (see, however,
Larsen-Freeman this volume, Ortega this volume). Most of the constructs derive
their meaning from the epistemology and methodology of the Chomskyan paradigm the study of language as a cognitive system. Leading experts in the SLA
field, such as Corder (1967, 1973), Selinker (1972) and Dulay, Burt and Krashen
(1982), have embraced this perspective uncritically, and many of their views have
stayed with us since then.
Corders (1967) article The significance of learners errors and Selinkers (1972)
article Interlanguage form a symbolic point of origin for contemporary L2 acquisition studies (Thomas 2013, p. 37). The conceptualization of SLA research
since these fundamental publications has been that learning takes place in the
individual minds of learners as they learn the second language at different ages,

Silvina Montrul

in different contexts, and for different reasons. One major role of SLA research
from this perspective is to understand the nature of the second language learning process and the product of that process, that is, the development of the
grammatical system of second language learners from initial state to ultimate
attainment, and how the final state system differs from that of native language
acquisition. Hence, dichotomies such as native speaker/non-native speaker or
adult L2 learner/L1 learner are at the core of the theoretical debates. Corder
(1967) argued that even when they were very different from the grammars of
native speakers of the same language, second language learners grammars
exhibited consistency and systematicity, and this idea formed the basis of
Selinkers (1972) influential Interlanguage Hypothesis. In asking what are the
psycholinguistically relevant data of second language learning (p. 210), Selinker
(1972) was concerned with cognitive aspects of second language learning for
theory construction. His answer was that the relevant data were behavioral
events (i.e., attempted meaningful performance in a second language (p. 210),
which would lead to an understanding of their underlying psycholinguistic
structures and processes.
In this chapter I specifically focus on illustrating how the Interlanguage
Hypothesis and the research agenda it set in motion lies at the heart of contemporary linguistic and psycholinguistic approaches to adult SLA, especially subsumed
under the theory of Universal Grammar (White 1989, 2003a). Much research in
the Chomskyan tradition in recent years has directly addressed the what, how, and
why of transfer and fossilization, two defining features of interlanguages as cognitive and linguistic systems (Schwartz & Sprouse 1994, Lardiere 2007, White 2003a;
see also Han this volume, Odlin this volume, Tarone this volume). I discuss how
several concepts of the original interlanguage proposal have been incorporated
into contemporary linguistic approaches to SLA and how this theoretical perspective, together with advances in research methods, has contributed over the years to
refining and clarifying much of Selinkers (1972) original proposal. I further stress
the far-reaching significance of Selinkers ideas for bilingualism/multilingualism,
and language change. For example, in the second part of this chapter I argue that
many of the characterizing features of interlanguage more specifically transfer,
simplification, and fossilization are not defective behavior unique to L2 learners,
as Selinker (1972) suggests. In fact, they are normal psycholinguistics processes,
linguistic behavior, and cognitive states of knowledge that emerge naturally in any
bilingual and multilingual setting. I take issue with Selinkers characterization of
bilingual varieties as espoused in the original article and illustrate how the characterizing features of interlanguage are also found in individual bilingualism more
broadly defined, as in the formation of new language varieties through the acquisition of a second language at non-native levels or through incomplete acquisition

Chapter 4. Interlanguage, transfer and fossilization

of a first language by bilingual speakers of minority languages (also called heritage

speakers). I conclude by pointing out unresolved issues and future directions to
advance a deeper understanding of the interlanguage phenomenon beyond second language acquisition.
Generative approaches to second language acquisition
Linguistic competence and interlanguage
Many of the ideas laid out by Selinker in 1972 remain central to generative theory
applied to second language acquisition (Clahsen & Muysken 1986, 1989; Liceras
1986; White 1989, 2003a). The object of investigation of generative linguistics is
linguistic competence, that is, learners unconscious and implicit knowledge of
language, which is only partially reflected in their use of language or performance.
According to Chomsky (1981), Universal Grammar constrains first language acquisition by children and the resulting adult native language grammars. The main
concern of this approach in its beginning was whether Universal Grammar, assumed to operate in first language acquisition, was also accessible in second language acquisition (Liceras 1986; White 1989).
Reading Selinkers (1972) article suggests that the seeds of the debate on adult
second language learners access to Universal Grammar (Clahsen & Muysken
1986; Flynn 1987) appear to have been planted there. Selinker (1972) was concerned with a theory of language learning, of what was going on in the learners
minds as they tried to express themselves in meaningful performance or spontaneous language use (i.e., not in the context of classroom grammatical drills). He
assumed there was a psychological structure latent in the brain, activated when
one attempts to learn a second language (p. 210) in the sense of the latent language structure that is the biological counterpart of universal grammar (p. 211)
referred to by Lenneberg (1967) for the acquisition of a first language. The latent
language structure was assumed to be transformed by the infant into the realized
structure of a particular grammar in accordance with maturational stages. Selinker
(1972, p. 212) wrote: I shall further assume that there exists in the brain an already formulated arrangement which for most people is different from and exists
in addition to Lennebergs latent language structure . This is what Selinker called
the latent psychological structure. Stressing fundamental differences between
first and second language acquisition, Selinkers position was that the assumed
latent psychological structure underlying second language performance and acquisition had no direct counterpart to the concept of Universal Grammar assumed
for children:

Silvina Montrul

It is important to state that with the latent structure described in this paper as
compared to Lennebergs, there is no genetic time table; there is no direct counterpart to any grammatical concept such as universal grammar; there is no
guarantee that this latent structure will be activated at all; there is no guarantee
that the latent structure will be realized into the actual structure of any natural
language (i.e., there is no guarantee that the attempted learning will prove successful), and there is every possibility that an overlapping exists between this
latent language acquisition structure and other intellectual structures. (Selinker
1972, p. 212)

As I will discuss in a later section, this idea was later incorporated by Bley-Vroman in
hiss Fundamental Difference Hypothesis (Bley-Vroman 1989, 1990) (Bley-Vromans
domain-specific vs. domain general cognitive mechanisms would be the counterparts of Selinkers latent language structure and latent psychological structure.)
The competence/performance distinction is central to the Chomskyan paradigm. Selinker also drew attention to a focus on linguistic competence through the
empirical study of linguistic performance, and made several claims about SLA that
are still relevant today in debates about the nature of L2 knowledge. The term interlanguage was coined to refer to a separate linguistic system based on the observable output which results from a learners attempted production of the target
language norm (p. 214). Very often, second language learners attempted production of the target language is not identical to the target language (p. 213). In
fact, Selinker confidently suggested that about 95% of second language learners
are unsuccessful and that the second language learner who actually achieves
native-speaker competence cannot possibly have been taught this competence.
Successful learners, in order to achieve native-speaker competence, must have acquired these facts without having explicitly been taught them (p. 213).
The Interlanguage Hypothesis is indeed the main object of study in the application of Chomskyan generative theory to the logical problem of second language
acquisition (White 1989, 2003a). The logical problem refers to the idea that the
input and the environment underdetermine the complexity of language knowledge. Unlike usage-based models that claim that language is a product of the environment and can be exclusively explained by it (see, e.g., Ellis 2007), the logical
problem states that the environment cannot explain the entire capacity and a great
deal of language knowledge must be innate. Like children learning their native
language, second language learners must build an abstract grammatical representation of the target language, and not everything that second language learners
know comes from general cognition and direct instruction (see also VanPatten
this volume). The goal of generative SLA research is to understand the mental
linguistic system that second language learners build in their language learning
process, which may interact and interface with some aspects of cognition, but is

Chapter 4. Interlanguage, transfer and fossilization

L1

IL1

Initial state

IL2

Intermediate stages
(development)

IL3

L2

End state

Figure 1. Schematic representation of interlanguage

not entirely subsumed by it. As schematized in Figure 1, the interlanguage is a

developing but grammatically constrained linguistic system, which has an initial
state, intermediate stages of development, and a final or end state.
On one conceptualization of interlanguage, the initial state of the L2 acquisition process is the linguistic system of the native language (L1), as schematized
in Figure 1. As the learner is exposed to input in the target language (the L2),
s/he restructures the linguistic system throughout the process of development,
creating evolving linguistic representations. The hybrid linguistic system created
in the process is interlanguage, and it is assumed to incorporate formal linguistic
properties from the L1 and from the L2 to different degrees. That is, the closer
the learner is to the initial state, the more constrained by the L1 his/her interlanguage is going to be, and the more advanced in development the more similar
the interlanguage is to the L2. Note that the end state may or may not converge
on the target grammar of native speakers because native-like attainment in second language acquisition is not guaranteed. As Selinker (1972, p. 214) proposed,
investigating the nature of the interlanguage system(s) calls for methodological
comparisons between adult native speakers of the target language, child native
speakers of the target language if one is to understand the natural development
of a native language, the adult nonnative grammar of interlanguage at different
stages of development, and adult native grammars of the learners L1 to understand the extent to which the interlanguage system incorporates elements from
all these sources. Indeed, the vast majority of experimental studies investigating
the role of Universal Grammar in second language acquisition favor this methodological approach.
Childrens language acquisition process is putatively guided by Universal
Grammar (see, however, the usage-based approach as discussed in Ellis, 2007), in
the sense that during the course of development the childs linguistic hypotheses
fall within the range of grammatical possibilities allowed by Universal Grammar.
For example, English-speaking children produce sentences like *What do you
think what is in here? with a wh-copy for the intended question What do you think
is in here? (with a gap between think and is). Although the question with the
wh-copy is not adult-like in English, this is one way in which questions are asked

Silvina Montrul

in German (Was glaubst du, was gibt es hier drinnen?), so the childs grammar
generates a question that is not targetlike in English but is perfectly acceptable in
German, another natural language. The developmental errors that children make
are systematic, observe linguistic constraints, and fall within the range of variation
evident in natural languages. Most importantly, they are overcome in due time
without the help of instruction.
White (1989, 2003a) and many others have argued that interlanguage grammars are also constrained by Universal Grammar. There are many properties of
the target language that L2 learners acquire that are not obvious from the input
or explicitly instructed in the classroom. One example is the Overt Pronoun
Constraint (Montalbetti 1984), which states that in null subject languages like
Spanish or Japanese, an overt pronoun cannot be co-referential with a quantifier.
That is, if you say in Spanish Nadie cree que l es inteligente Nobody thinks he is
intelligent, the pronoun l cannot corefer with nadie; it has to refer to somebody else. Only a null pronoun, as in Nadie cree que es inteligente can corefer
with nadie in Spanish. Note that in English, a non-null subject language, the
overt pronoun he can corefer with nobody. L2 learners of Spanish and Japanese
learn in the classroom that in these languages you can omit subjects in many
cases, but they do not learn the interpretation of null and overt pronouns in
these constructions. There are several experimental studies testing the Overt
Pronoun Constraint with L2 learners of Spanish (Lozano 2002; Prez-Leroux &
Glass 1999) and Japanese (Kanno 1997). These studies show that intermediate
level learners of these languages are able to acquire, without instruction, the correct interpretation of null and overt pronouns in these null subject languages.
Interlanguage grammars are also characterized by systematic developmental errors arising during the process of acquisition among L2 learners of different linguistic backgrounds. One example is the null prep phenomenon studied by Klein
(1993). In English and in many languages there are verbs that subcategorize for a
preposition (e.g., think about, depend on). L2 learners of different languages have
been shown to omit the required preposition with these verbs in prepositional or
oblique relative clauses, as in (1c).
(1) a. The book about which you talked is a bestseller. (pied piping)
b. The book which you talked about is a bestseller. (preposition stranding)
c. *the book which you talked is a bestseller. (null preposition)
More recently, Perpin (2010) conducted an experiment on null prep in Englishand Arabic-speaking learners of Spanish. In an oral task where the participants
were asked to produce oblique relative clauses like those in (2a), they also produced relative clauses with no prepositions (), as in (2b):

Chapter 4. Interlanguage, transfer and fossilization

(2) a. El seor del que depende Mara es muy rico.

the man of whom depends Maria is very rich
The man on whom Maria depends is very rich.
b. *El seor que depende Mara es muy rico.
Thus, the omission of required prepositions in relative clauses and in questions
can be characterized as a genuine interlanguage phenomenon. It arises in learners
of different language backgrounds and in the acquisition of more than one language. Like L1 learners, L2 learners make errors that are typical of development.
L1 transfer
Many characteristic errors in second language acquisition are due to influence
from the first language, or language transfer. Transfer is a defining feature of the
cognitive process of second language acquisition: the native language plays a
prominent role in the behavioral and cognitive outcomes of second language acquisition. Questions posed by Selinker (1972), which have generated substantial
research in generative approaches to second language acquisition in the last three
decades, are what transfers, how, and why. Corder (1983) proposed that transfer
operates differently at different levels of linguistic structure, such as phonology
and syntax. Corder considered that the initial state in the acquisition of L2 phonology is the phonology of the L1, whereas it is not always the case that the native
language is the initial state for syntax; rather, there is a universal core. The study of
transfer or interference lacked theoretical rigor and predictive power under the
Contrastive Analysis Hypothesis advocated by Lado (1957). With the advent of the
Principles and Parameters approach to second language acquisition, the study of
transfer gained theoretical interest (Flynn 1987, Liceras 1986, White 1989). This
version of Universal Grammar held that human language is constrained by universal abstract principles and crosslinguistic differences are realized by parameters set
on the basis of experience (e.g. null subject languages vs. non-null subject languages; verb-raising languages vs. non verb-raising languages; local vs. long distance binding). The initial state of first language acquisition is Universal Grammar
and the child sets the parameters of her native language through exposure to input. Whereas L1 acquisition is parameter setting, L2 acquisition is parameter resetting: parametric values transfer and second language learners need to reset the
L2 values in their interlanguage development.
The question of whether Universal Grammar is the initial state of L2 acquisition in morphosyntax, and whether and how it interacts with transfer in interlanguage development gave rise to a lively debate and different theoretical positions
in the 1990s, in the context of the acquisition of lexical and functional categories.
Three main theoretical perspectives are displayed in Figure 2: The Full Access

Silvina Montrul
L1 Acquisition
L1
INPUT

L2
INPUT

UNIVERSAL GRAMMAR

G1

L2
INPUT

L2 Acquisition: Full access

G2

G3

UNIVERSAL GRAMMAR

Steady
state L1
Grammar

Steady state
of L1
Grammar

ILG2

ILG3

Steady
state ILG

L2 Acquisition: No access

L2 Acquisition: Full transfer/full access

UNIVERSAL GRAMMAR

UNIVERSAL GRAMMAR

Steady
state L1
Grammar

ILG2

ILG3

Steady
state ILG

L2
INPUT

Steady
state L1
Grammar

ILG2 ILG3 Steady

state ILG

Figure 2. The initial state and transfer in interlanguage development

Hypothesis (Epstein, Flynn & Martohardjono 1996), the No Access Hypothesis

(Bley-Vroman 1990; Clahsen & Muysken 1986, 1989; Meisel 1997) and the Full
Transfer/ Full Access Hypothesis (Schwartz & Sprouse 1996).1
The upper left panel of Figure 2 shows the initial state of L1 acquisition. Input
is filtered through Universal Grammar and the child builds interim grammars until reaching the adult final state. In L2 acquisition there are two possible sources of
knowledge at the initial state: Universal Grammar and the L1 of the learners. The
Full Access Model displayed in the upper right panel shows that L2 acquisition is
very much like L1 acquisition: Universal Grammar is the initial state in L2 acquisition. Although L1 knowledge is available, it does not play much of a role in interlanguage development. In the two other models, however, previous knowledge of
the L1 is seen as crucial. In the No Access model (bottom left panel), the L1 is the
initial state of L2 acquisition and there is no access to Universal Grammar. Universal Grammar is implicated in building the native language, but it is no longer available in adult L2 acquisition (see severed links). Interlanguage development results
from the interaction of input, the L1, and other learning mechanisms. Finally, the
Full Transfer/Full Access Hypothesis states that the L1 is the initial state of L2 acquisition, although Universal Grammar remains available throughout interlanguage development as well (see direct connections between UG and ILGs in the
figure). When the cognitive representation provided by the native language can no
1. In addition to these, there were at least two other theories: The Minimal Trees Hypothesis
(Vainikka & Young-Scholten 1996) and the Valueless Features Hypothesis (Eubank 1996).
See White (2003a) for details and discussion.

Chapter 4. Interlanguage, transfer and fossilization

longer accommodate L2 input for subsequent interlanguage restructuring, L2

learners resort to Universal Grammar during development, as Slabakovas (2001)
study described below illustrates. A prediction this theory makes is that the effects
of the native language will be stronger at initial stages of development and less so
at advanced levels of development. Furthermore, this theory states that native-like
competence is possible but not guaranteed (although this is not captured in the
corresponding figure).
Experimental studies including L2 learners of different proficiency levels provide the best testing case to address what the initial state is in L2 acquisition and
the question of whether the initial state determines to some extent the degree of
ultimate attainment. A good example of initial L1 transfer and progressive abandonment of this blueprint in response to input and grammar restructuring is
Slabakovas (2001) investigation of knowledge of telicity marking (a semantic notion related to aspect and endpoints) in English. A telic sentence like Antonia
baked a cake is interpreted as completed (telic) (i.e., Antonia finished the cake),
whereas the same sentence with a bare plural (Antonia baked cakes) is interpreted
as habitual and incomplete (atelic), where there is no indication that a cake was
finished. When one compares the morphological make-up of Romance and Slavic
languages to that of English, one notices that English does not mark aspectual
distinctions like telic/atelic overtly, that is, with morphology on the verb. By contrast, Bulgarian uses prefixes.
Slabakova conducted two studies to assess how Bulgarian and Spanish speakers would interpret telic and atelic sentences in English: one study included
Bulgarian-speaking learners of English ranging from low to high-advanced proficiency, and a follow-up study focused on low proficiency Spanish-speaking learners of English. Spanish marks telicity with count and mass nouns in object position
as in English. In one of the tasks, the L2 learners were asked to judge the felicity of
two clauses, which only differed in whether the object in the second clause was
count or mass (one way to indicate telicity in English). As can be seen, (3a) is a
logical combination while (3b) is odd. Learners were asked to judge how well these
sentences combined on a scale ranging from -3 (illogical) to +3 (logical).

(3) a. Antonia worked in a bakery and made cakes. (habitual 1st clause +
atelic 2nd clause)
b. #Antonia worked in a bakery and made a cake. (habitual 1st clause +
telic 2nd clause)

If the L1 constrains L2 initial representations, Slabakova predicted that Bulgarianspeakers would have significant difficulty distinguishing telic from atelic sentences
in English, due to the fact that the verb is morphologically unmarked, providing

Silvina Montrul

no obvious clues to aspect. Eventually, to overcome L1 influence, Bulgarian-speaking learners would have to realize that English marks telicity in the object (count
vs. mass objects). Since Spanish also marks telicity on the object (and grammatical
aspect with preterite-imperfect suffixes on the verb), Spanish-speaking L2 learners
were expected to perform like the English native speaker controls. The results
showed that the low proficiency Bulgarian-speaking group performed less accurately than all other groups, including the proficiency-matched Spanish-speaking
group, as predicted. This difference was statistically significant with telic sentences
(which require an overt verbal prefix in Bulgarian but not in Spanish or English).
Secondly, intermediate and advanced Bulgarian speaking learners did not differ
statistically from the native speakers of English. Slabakovas cross-sectional study
provides an excellent example of how low-intermediate learners transfer their L1
parametric values to the L2 (Full Transfer); yet, with development, intermediate
and advanced Bulgarian speakers overcome their L1 influence and reset the parameter to the target value (Full Access). Slabakovas study shows that, at least for
the phenomenon she investigated, L2 learners are able to overcome the influence
of their native language.
Selinker (1972, p. 219220) describes a tendency for simplification in many
interlingual situations as another feature of interlanguage and strategy of second
language learning. Many second language learners of English avoid grammatical
formatives such as articles, plural forms, and past tense forms as a strategy of
second language communication. Selinker cites Coulters (1968) examples from
two elderly Russian learners of English who omit articles in English. However, the
misuse of articles by Russian speakers learning English has a clear L1 transfer explanation because Russian is a language that does not have articles (Ionin 2003).
How do we know whether this apparent simplification or strategy of second language learning is due to transfer or something else?
An interesting question that has arisen in the context of transfer and grammatical theory is if there is nothing similar in the L1 to transfer from, what do L2
learners do at initial and intermediate levels of development? Are errors random
or systematic? Ionin, Wexler and Ko (2004) investigated the types of interlanguage
errors produced by L2 learners whose native languages do not have articles, such
as Russian and Korean learners, when learning articles in English. In addition to
article omission, once these learners learn articles, a common interlanguage pattern is misuse of the definite article the with indefinite NPs and the misuse of the
indefinite article a with definite NPs. Definiteness and specificity are two semantic
universals that constrain the distribution of articles, and the English article system
is based on definiteness, not specificity. Ionin et al. (2004) hypothesized that
semantic universals are available to L2 learners whose L1 does not have articles.
As a result, they may associate the definite article with [+specific] or [+partitive]

Chapter 4. Interlanguage, transfer and fossilization

instead of [+definite], for example. Ionin et al. (2004) found in written narratives
that Russian- and Korean-speaking learners of English relied on [+specificity]
rather than definiteness when using articles in English, producing examples such
as the following:

(4) a. If it is happen Ill spend money for the trip to California or Florida. Im
tired for winter this year.
b. I would go for example to the Grand Canyon and use the chance to fly
over it in the helicopter.
c. I really want to visit the famous city in a European country which has
old history. The exact place will be depended on the time which I have
or the flight price.

(5) a. 2 days before thanksgiving, my baby was bourn. We have to spend a

holiday at the hospital.
b. When I got the Syracuse airport, NY, I was very nervous about a new
environment, even I couldnt able to speak English well.
Analysis of written narrative data showed that the L2-English learners did indeed
overuse the in [definite, +specific] contexts to a much greater degree than in [definite, -specific] contexts. They also used the correctly in [+definite, +specific] contexts. As for article omission, both L1-Russian and L1-Korean speakers were significantly more likely to omit articles with indefinites than with definites. Ionin et
al. (2004) thus showed that article omission in L2-English, like article use, is not
random. While factors such as performance pressure may cause L2-learners to omit
articles more in oral or written narrative data, as Selinker suggested, than in a controlled formal written elicitation study, Ionin et al. found that the learners did not
omit articles to the same extent across all categories. They omitted a, which carries
little semantic information, significantly more than the, which encodes uniqueness
and presupposition. Thus, article omission, like article use, is systematic, and indicates that L2-English learners are aware of article semantics. It would be difficult,
however, to characterize this phenomenon as simplification, as Selinker suggests,
since the explanation based on transfer and universals is quite compelling.
Another question related to transfer that the study of articles raises is the following: Do L2 learners whose first language has no articles have more difficulty
acquiring article semantics than L2 learners whose first language has articles?
Ionin and Montrul (2010) looked at the interpretation of generic reference in
English by Spanish and Korean-speaking learners. Spanish has definite and indefinite articles like English, but there is a crucial difference between the two languages with respect to the interpretation of bare plurals. For example, bare plurals
in English have generic reference, as in Lions are dangerous, and definite plurals
have only specific reference, as in The lions are dangerous. In Spanish, bare plurals

Silvina Montrul

are ungrammatical in subject position (*Leones son peligrosos), and sentences with
definite plurals like Los leones son peligrosos have both generic and specific readings. Korean has no articles, and generic reference is expressed with bare plurals
(Saja-tul-i wihumha-ta Lion-plural-SUBJ dangerous-DECL). Ionin and Montrul
(2010) found that 58% of the Korean speakers in their study tested with a truth
value judgment task had target-like interpretation of definite plurals with specific
reference, and these included learners from low intermediate to advanced proficiency. However, only 19% of the Spanish-speaking learners showed target interpretation of definite plurals in English and all of them were advanced. Most of the
Spanish-speaking learners (64%) interpreted definite plurals as generic, following
the semantic tendency of Spanish. Thus, even though the Spanish-speaking learners come from a language with articles whereas the Korean learners do not, the
Korean learners acquired the meaning of English articles with specific reference
faster than the Spanish-speaking learners, because their L1 did not interfere.
To summarize thus far, transfer is a very salient component of interlanguage
grammars and with deeper and broader linguistic analysis it is more possible to
predict today than it was in the 1970s patterns of what, when and how it will occur
(cf. Odlin this volume). It is also possible to tease apart errors resulting from L1
transfer or from other universals, depending on the grammatical properties tested.
Transfer is prevalent in initial stages, but depending on the native language, the
target language, and the linguistic level of analysis, it can persist into ultimate attainment, as we see next, leading to localized fossilization (cf. Han this volume).
Fossilization
It is not uncommon for many L2 speakers to exhibit clearly non-native like features in their linguistic behavior, most notably in pronunciation and morphosyntax. Many adult second language learners display synchronic variability (cf. Ortega
this volume, Tarone this volume) and premature stability in some linguistic areas.
This is the phenomenon of fossilization, a term first introduced by Selinker (1972,
p. 215): Fossilizable linguistic phenomena are linguistic items, rules, and subsystems which speakers of a particular NL will tend to keep in their IL relative to a
particular TL, no matter what the age of the learner or amount of explanation and
instruction he receives in the TL. Fossilization is a pervasive phenomenon in L2
acquisition, and Selinker (1972, p. 217) claimed that the entire IL competence
could be fossilized in individual learners: Not only can entire IL competence be
fossilized in individual learners performing in their own linguistic situation (emphasis mine), but also in whole groups of individuals, resulting in the emergence of
a new dialect (here Indian English), where fossilized IL competences may be
the normal situation. However, what we know today through careful linguistic

Chapter 4. Interlanguage, transfer and fossilization

analyses of interlanguage grammars is that fossilization is localized, in the sense

that it occurs in some areas of grammar, especially phonology and inflectional
morphology, of an individual rather than their entire linguistic competence as
Selinker suggested. Furthermore, many if not most of the fossilizable items are due
to L1 transfer. That is, L1 influence is not just a temporary feature of interlanguage
development but it can persist in localized aspects of the interlanguage from the
initial state into ultimate attainment, leading to fossilization (see also Han 2013,
this volume).
In a recent longitudinal study of the initial state in child L2 acquisition, Li
(2012) studied six 79 year old Chinese-speaking children who were learning
English in an immersion context in the United States (English only school). The
childrens spontaneous productions were elicited through play and stories once a
month for eight months in interactions with the researcher. The recordings were
transcribed and coded for past tense, agreement and nominative subjects. Li found
that after six months of intensive exposure to English, the children were 100% accurate producing overt subjects with finite verbs, but had great difficulty with 3rd
person singular agreement (13% accuracy), regular past tense (16% accuracy), and
irregular past tense (38% accuracy). Since Chinese is assumed not to have Tense as
an abstract grammatical category, one can conclude that the childrens slow progress with tense and agreement can be related to the structure of Chinese. However,
Chinese also has null subjects and subject expression in English was not affected.
Since this study focused on the initial state, it is possible that given time and abundant input, these children will eventually converge on the grammar of tense and
agreement in English. Interestingly, similar accuracy rates with agreement and
tense are reported by Lardiere (1998a,b, 2007) in her exemplary longitudinal case
study of Patty, whose ultimate attainment in English diverges from native-like
competence in some respects and completely converges in others (see also Han
this volume). An essential difference between the data from Li and the data from
Lardiere is that Lis study involves children while Lardieres study is about an adult,
and recent research in child L2 acquisition suggests that child L2 learners can
be affected by L1 transfer as much as adults at initial stages of development
(Herschensohn 2007).
Lardiere followed the linguistic performance of Patty, an English L2 speaker
whose first language is Chinese, over a period of almost 20 years. Patty was recorded after she had lived in the USA for 10 years and then again almost 9 years
later. Lardiere shows that there were no changes in Pattys grammar between the
three recording sessions conducted 9 years apart. Lardiere (1998a,b, 2007) examined Pattys use of past tense and agreement morphology, as well as other syntactic
evidence of grammatical morphology in wh-questions and the nominal domain.
Results show a clear dissociation between morphological and syntactic aspects of

Silvina Montrul

verbal inflection: Patty only supplied 4.41% of 3rd person agreement morphology
and 34.66% of past tense morphology in obligatory contexts, while production of
overt subjects and nominative case was above 98% accurate. Thus, while fossilization in Patty was evident in her morphophonology (affixal inflection), her syntax
was nativelike. She had nativelike knowledge of pronominal case marking, case
marking in subjects as a function of finiteness, placement of verbs and adverbs,
relative clause formation, wh-movement, preposition stranding, subject/auxiliary
inversion, and do-support in questions. Therefore, one can conclude from this
study that fossilization did not affect Pattys entire competence, contra Selinkers
claim about fossilized competence.
There is another study that reinforces the conclusion that the problems Patty
exhibited with tense in English are L1-related. White (2003b) conducted a longitudinal study with a Turkish speaker living in Montreal for several years SD
who also seemed to have reached a level of stability in localized features of her
interlanguage development and was short of nativelike ability in tense, agreement
and determiners. Like Patty, SD was also recorded, one and a half years apart, and
her speech was analyzed for evidence of non-native features. SD showed about
80% accuracy on 3rd person agreement and past tense in English. Her level of fossilization (assessed longitudinally one and a half years apart) was much lower than
for Patty with tense and agreement presumably because Turkish has tense and
agreement instantiated in its grammar. SD was most inaccurate with determiners,
omitting about 40% of definite determiners in English, since Turkish has no determiners except for the numeral bir one. The conclusion one draws from these
studies is that transfer and fossilization can be related in that one (transfer) can
become the other (fossilization), a point that Selinker did not make clear in his
article (for discussion, see Han this volume). In other words, it seems that the L1
can actually have a deterministic role in ultimate attainment.
Related to fossilization, another major claim made by Selinker (1972) that has
generated substantial amount of research over the years concerns the level of success in achieving nativelike competence among second language learners. Selinker
states that at most only 5% of second language learners can converge on the target
language and 95% are typically non-convergent, showing fossilization, transfer errors, and other errors caused by instruction (see Rothman). Selinker (1972, p. 212)
wrote: Those adults who succeed in learning a second language so that they
achieve native-speaker competence have somehow reactivated the latent language
structure which Lenneberg describes, the counterpart of Universal Grammar. As
mentioned earlier, Selinkers ideas underlie the debate between the full access versus the no access views of Universal Grammar in L2 acquisition. Selinker clearly
holds a deficit view of second language acquisition, resonated in Bley-Vromans
(1989) Fundamental Difference Hypothesis, according to which the vast majority

Chapter 4. Interlanguage, transfer and fossilization

of second language learners cannot achieve native like competence and performance in a second language. Although L2 learners are faced with the same logical
problem as L1 learners, in the sense that they have to build a mental representation
of the target language based on input, Bley-Vroman suggests that the gap between
abstract knowledge and experience in L2 acquisition is not bridged by Universal
Grammar, or a domain-specific linguistic system. The main claim of Bley-Vromans
(1990) version of the Fundamental Difference Hypothesis (FDH) is that the innate
linguistic system that operates in childhood (Lennebergs latent language structure) is no longer efficient in adult foreign language learning, an idea consistent
with the Critical Period Hypothesis. Because L2 learners have already acquired
their L1 through the domain-specific innate linguistic system (UG), they can rely
on their L1 knowledge (a particular instantiation of UG, but not the full spectrum
of options). With L2 acquisition starting after puberty, L2 learners no longer have
access to the domain-specific linguistic mechanism. Instead, they deploy the domain-general cognitive system unrelated to UG (referred to by Selinker 1972 as
the latent psychological structure). Similar views are shared by other researchers
within the generative framework (Bley-Vroman, Felix & Ioup 1988; Clahsen &
Muysken 1986; Hawkins & Chan 1997; Meisel 1997, 2011; Schachter 1990; Sorace
1993; Tsimpli & Roussou 1991) and by researchers who work with other cognitive
approaches to second language acquisition that eschew the concept of Universal
Grammar (Abrahamsson & Hyltenstam 2009; DeKeyser 2000, 2003; Long 2007;
Paradis 2004; Ullman 2001, among others).
To summarize, I have shown how the powerful notions of interlanguage, developmental errors, language transfer, fossilization, access to Universal Grammar
and fundamental differences between native speakers and second language learners all laid out in Selinker(1972) have formed the basis for much contemporary
research in the cognitive and linguistic dimensions of second language acquisition. In the next section I show that these notions are not unique to second language acquisition, and many of the linguistic processes and behaviors typical of
second language acquisition extend to bilingualism more generally. Although
mentioned in passing, these ideas were actually present in Selinker (1972).
Beyond L2 acquisition: Language contact and change
From interlanguage to new language varieties
The study of second language acquisition from a cognitive and linguistic perspective emphasizes language knowledge and development in individuals, and the
focus of study is individual interlanguages. Studies of languages in contact and

 Silvina Montrul

language change as a sociolinguistic phenomenon focus on characteristics of

speech communities and their language varieties. Yet, speech communities are
made up of groups of individuals, and it seems that what happens at the individual level can be generalized to tendencies of groups. A group of interlanguages becomes a language variety. In situations of large scale immigration (Spanish in the
United States) or colonization (English in India), native and second language varieties can co-exist, eventually leading to language change.
If interlanguages and second language varieties are related, it thus follows that
key features of second language acquisition, such as simplification and overgeneralization errors, transfer, and fossilization, should also be found in situations of
language contact in general (Montrul 2008, Silva-Corvaln 1994, Thomason 2001).
Selinker (1972) was indeed aware of the relationship between interlanguage and
non-native varieties, since one of the purposes of his article was to lay a foundation for a psychological and mentalistic explanation of what Weinreich (1953, p. 7)
referred to as interlingual identifications (i.e., the incorporation of linguistic features from one language into another with consequent restructuring) in the context of language contact. Indeed, Selinkers article opens with the observations
made in the language contact and change literature, which he tries to pursue from
a psycholinguistic perspective.
Incomplete acquisition in second language learners, with remnants of language transfer and fossilization, and incomplete acquisition in a bilingual situation
have been identified as two of many potential mechanisms for language change at
the sociolinguistic and historical level by Thomason (2001) and McWhorter
(2007). Curiously, none of these researchers cite Selinker (1972) or any other central study in second language acquisition. This has been, unfortunately, typical of
related fields that deal with very similar, if not sometimes the same, phenomena
but due to theoretical and empirical differences do not relate to each other, or even
know of the existence of each other (cf. Han 2012). In referring to Indian English,
Selinker (1972, p. 217) wrote: Not only can entire IL competence be fossilized in
individual learners performing in their own linguistic situation, but also in whole
groups of individuals, resulting in the emergence of a new dialect (emphasis mine)
(here Indian English), where fossilized IL competences may be the normal situation. The phenomenon of fossilization discussed in the previous section is predicated on the completeness hypothesis (cf. Schachter 1990). Completeness entails
the availability of Universal Grammar or, as argued in non-UG models, reaching
(idealized) native speaker competence (Abrahamsson & Hyltenstam 2009; see,
however, Larsen-Freeman this volume; Ortega this volume). Universal Grammar,
Schachter argues, is not available in its entirety to adult second language learners,
which results in the incomplete growth of their second language. More specifically,
an L2 learner will not develop some linguistic property x completely if L2 input is

Chapter 4. Interlanguage, transfer and fossilization

received after puberty, when the L1 is fully formed. Bhatt (2002) argues convincingly, against Selinkers ideology underlying his statement, that Indian English is
the result of global fossilization, while Sharmas (2005a,b) empirical studies put
Selinkers claims to the test by teasing apart interlanguage features from dialectal
features in non-native varieties of English (NNVE).
Sharma (2005a, b) recognizes that indigenized NNVEs such as Indian English
represent an unusual sociolinguistic challenge because they can neither be straightforwardly subsumed under models of individual second language acquisition as
espoused by Selinker (1972) nor under sociolinguistic models of native variation.
It appears that the hybrid status of English in these cases derives from its functional status as a second language, on the one hand, and more native-like patterns
of indigenous transmission and use, on the other. To tease apart individual features related to errors typical of interlanguage development from features of dialectal stabilization, Sharma (2005a,b) conducted a study with 12 speakers of South
Asian languages (Gujarati, Punjabi, Tamil, Kannada and Hindi) living in the United States. Their ages ranged from 2467 and their years of residence in the United
States ranged from 5 months to 40 years. Three speakers had not received education in English, three had higher education in English, and the rest had received
their entire education in English. The four Hindi speakers, who also received their
entire education in English, reported the highest use of English at home, work and
with friends. The linguistic variables investigated were representative of L2 learning features (copula, past tense marking, agreement) and of stabilizing features
(use of definite and indefinite articles). Sharma found that eight of the 12 speakers,
the ones with fewer years of education in English and less daily use of English,
made errors with copulas, past tense, and agreement, while the four Hindi speakers with higher education in English and more frequent use of English (i.e., estimated higher proficiency) did not make any errors of this sort. However, all of the
South Asian speakers, regardless of level of education, length of residence in the
United States, and of daily use of English, used definite and indefinite determiners
in a non-target way. That is, they frequently omitted definite and indefinite articles, or misused them. This is reminiscent of the errors with Russian and Korean
speakers of English discussed earlier.
Transfer of the L1-specificity markings seems to be the main cause for divergence in article use in NNVE given that Hindi and related South Asian languages
do not have articles. Hindi marks indefinites with the numeral ek (one). Definiteness in Hindi is marked by case and word order. Because the speakers estimated
proficiency through education and language use was not correlated with error
rates with articles (it was correlated with tense, agreement and copula), Sharmas
results suggest that not all the divergences found in Indian English are limited to
transient interlanguage features. It is not that these speakers do not use English

Silvina Montrul

articles, but the rate of omission in required contexts was significantly different
with respect to Standard English. Because South Asian languages have agreement
and past tense, the errors with agreement and past tense in English cannot be
related to L1 influence, and Sharma describes them as interlanguage features,
probably meaning developmental errors that will eventually go away with more
advanced proficiency. Sharma (2005b) looked in more detail at the distribution of
articles in the 12 speakers in search of evidence of systematic and principled behavior. She found that the speakers use of articles appeared to be constrained by
discourse factors. While omission of articles increased relative to the overall familiarity status of the NP, overt articles were used for the purpose of discourse disambiguation. Sharma concluded that article use in NNVE results not from entire
fossilization of L2 competence but from a combination of transfer (specificity
marking in this case) and universal pragmatic factors of discourse identifiability
(i.e., givenness). As discussed earlier, transfer and fossilization are not unrelated:
some errors that have their source in L1 transfer can become fossilized, as we saw
in the case of Patty, the Chinese speaker studied by Lardiere (2007).
Incomplete acquisition and age effects
Not all fossilized features have their origin in language transfer when it comes
to interlanguages and non-native varieties (see, however, Han this volume).
Thomason and Kaufman (1988) note that when a population shifts to a new language and their rendition of the language ousts the original native one, as when
the Scandinavians invaded England, transfer effects occur alongside incomplete
acquisition. Even when the languages in contact have parallel or cognate structures, there can be reduction rather than language transfer. McWhorters (2007)
study traces signs of non-native acquisition in standard language grammars and
his basic thesis is that present-day standard languages are simpler than they used
to be and simpler than other related languages because they were broken down
by large numbers of adult second language learners who did not acquire the language completely. He recounts that the Danes and Norwegians were largely illiterate and did not impose their language in writing or in government when they occupied England. Therefore, it is likely that the loss of inflections in Old English
resulted from Old Norse speakers non-native and incomplete acquisition of English. The English timeline was decisively influenced by what Trudgill (2001) has
termed in apt and savory fashion the lousy language-learning abilities of the
human adult (McWhorter 2007, p. 103). Without a doubt, Selinkers (1972) conceptualization of second language acquisition is also at the heart of McWhorters
argument. That is, the native speaker has a privileged status, and adult second
language learners are in some sense deficient who do not achieve nativelike

Chapter 4. Interlanguage, transfer and fossilization

competence in a second language (see, however, Larsen-Freeman this volume,

Ortega this volume). The suggestion is that incomplete acquisition and fossilization at the individual level become a language variety when there are large groups
or intergenerational groups of second language learners who do not reach nativelike ability in a second language. A powerful reason for the lack of success of second language learners is linked to maturational effects.
In their groundbreaking study Johnson and Newport (1989) addressed the inability of second language learners to reach nativelike competence, by focusing on
the effects of age, and more precisely the potential existence of a critical period, for
second language acquisition. This study, and their numerous replications, partial
replications, and extensions by other researchers, confirmed that age is a defining
factor in the degree of ultimate attainment, and that the younger the age of arrival
and immersion in the second language environment for immigrants, the more nativelike their acquisition of the second language is likely to be as adults. Many proposals within generative approaches to second language acquisition have embraced
this position: the Fundamental Differences Hypothesis (Bley-Vroman 1989),
Hawkins and Chans (1997) Failed Functional Features Hypothesis and Tsimpli and
Dimitrakopoulous (2007) Interpretability Hypotheses, among others. Other, nongenerative supporters of age effects in second language acquisition have stressed
the role of domain-general learning factors, such as aptitude (Abrahamsson &
Hyltenstam 2009; DeKeyser 2000; Long 2007; Granena & Long 2013).
It turns out that incomplete acquisition or imperfect learning is not uniquely
characteristic of L2 acquisition. Montrul (2008), Silva-Corvaln (1994), OGrady
et al. (2011), Polinsky (2006), and many others have shown that it is possible to
have incomplete knowledge of a first, native language acquired in childhood. Although children acquiring a second language in a bilingual environment in an
immigrant situation can reach native-like levels of proficiency in the second language, as shown by the Johnson and Newport study, successful L2 acquisition
brings a downsize in these children. Many of these same children fail to receive
adequate input in their native language throughout childhood to reach native-like
competence in their own native language. These speakers of minority languages
are typically referred to in the United States as heritage speakers. Heritage speakers
from immigrant background are individuals who emigrated in early childhood
with their parents and other family members, or children of immigrants from
other countries. While the parents are either monolingual or dominant in a variety
of their languages (e.g., Spanish, Russian, Arabic, Korean, Mandarin, Tagalog,
Hindi in the United States), the children grow up in a context where the majority
language and the immigrant language of the family are spoken. Many of these
children are either monolingual or dominant in the heritage language in early
childhood, but as bilingualism progresses and changes during elementary school,

Silvina Montrul

they can go through a period of balance in the two languages (Kohnert & Bates
2002) and eventually become dominant in the majority language (Kohnert, Bates
& Hernndez 1999). Typically, when they reach adolescence, minority language
speakers are already dominant in the majority language, and by the time they are
adults, the majority language is both stronger and dominant in their overall proficiency and in all domains of use. The study of heritage speakers has challenged the
idea that native language acquisition is always successful and that age of acquisition is primarily responsible for fossilization and incomplete learning of a language. Because early bilinguals were exposed to the target language since infancy
and before puberty when UG was presumably still available, their knowledge of
the L1 should have been acquired with the specific-linguistic learning and processing mechanisms (i.e., latent language structure) assumed to operate in childhood. Hence, non-convergent acquisition in the case of heritage speakers could
not be due to lack of access to UG. Instead, it must be strictly due to insufficient
input and use. By contrast, on the deficit view of second language acquisition espoused by Selinker and others, non-convergence in L2 learners may arise due to
less efficient learning mechanisms than those deployed in childhood, although
insufficient input and use of the L2 may also be involved.
Although heritage speakers and second language learners differ in age of acquisition, timing and modality of input, and context of acquisition (Montrul 2008,
2012), many recent studies have shown that the linguistic systems of heritage
speakers and second language learners share many characteristics. Like second
language learners, heritage speakers display non-native or non-targetlike features
in their heritage language most notably in morphology, complex syntax, semantics and discourse pragmatics that result from transfer from the majority language, simplification, and even fossilization (Benmamoun, Montrul & Polinsky,
2013a, b).
Montrul and Ionins (2010, 2012) studies exemplify the role of dominant language transfer in heritage speakers of Spanish in the United States in the area of
the semantic interpretation of definite plural articles in Spanish and in English.
Montrul and Ionin (2010) investigated the two languages of twenty-three Spanish
heritage speakers in order to assess both the dominance relationship between the
two languages, and whether there is transfer from the stronger language onto the
weaker language in the meaning of generic reference. Spanish plural NPs with
definite articles can express generic reference (Los elefantes tienen colmillos de
marfil), or specific reference (Los elefantes de este zoolgico son marrones). English
plurals with definite articles can only have specific reference (The elephants in
this zoo are brown), while generic reference is expressed with bare plural NPs
(Elephants have ivory tusks). The Spanish heritage speakers completed three tasks
in Spanish (acceptability judgment, truth value judgment, and picture-sentence

Chapter 4. Interlanguage, transfer and fossilization

matching tasks) and the same three tasks in English. The heritage speakers showed
nativelike ability in English, as judged by a proficiency test, and their ceiling performance in the three tasks as compared to native speakers of English. Even though
the heritage speakers were very fluent in Spanish, their proficiency in Spanish was
lower than that of age-matched native speakers of Spanish and lower than their
own proficiency in English. The experimental tasks showed significant transfer
from English into Spanish with the interpretation of definite articles in generic
contexts. The heritage speakers accepted ungrammatical bare plurals in subject
position in the acceptability judgment task (e.g., *Leones son peligrosos) and interpreted sentences with definite articles as having more of a specific rather than generic interpretation. The results showed persistent transfer from the dominant
language, even in heritage speakers with quite advanced command of their heritage language. Montrul and Ionin (2012) compared 30 heritage speakers (23 were
the same subjects in Montrul and Ionin 2010) and 30 L2 learners of Spanish
matched in Spanish proficiency as measured by a cloze test and a vocabulary test.
The participants completed the three tasks employed in Montrul and Ionin (2010).
The results showed that the heritage speakers and the L2 learners exhibited the
same degree of dominant language transfer from English (the L1 of the L2 learners
but the L2 of the heritage speakers) with the acceptability of ungrammatical bare
plurals in subject position and the interpretation of definite articles in generic contexts. These findings strongly indicate that transfer is a powerful mechanism in
early bilingualism as well, which can lead to fossilization.
Simplification is another linguistic process identified in second language acquisition that plays a role in incomplete acquisition by heritage speakers. Polinsky
(2008) investigated gender agreement in Russian heritage speakers of low and intermediate proficiency in Russian. Russian has three genders: masculine, feminine, and neuter. Masculine nouns constitute about 46% of the nominal lexicon,
feminine nouns 41%, and neuter nouns 13%. Grammatical gender is correlated
with semantic gender in the usual way with people and animals, but there are
many exceptions as a function of both declensional case type and phonological
factors (especially noun endings). The system of gender assignment is intrinsically
linked to declensional class, requiring access to the endings in the unmarked
(nominative) case as well as additional declensional information. Heritage speakers of languages that mark case overtly have been shown to omit case or misuse it
(Montrul, Bhatt & Bhatia 2012; Polinsky 2006; OGrady et al. 2011). Polinsky found
that the loss of declensional classes led to a significant reanalysis of gender assignment in Russian heritage speakers. Low proficiency heritage speakers exhibited
two cases (nominative, accusative) rather than six. At the same time, despite significant changes in case marking, it was not the case that gender assignment and
agreement disappeared completely from heritage speakers with low proficiency in

Silvina Montrul

the language. Polinsky asked twelve heritage speakers who did not read Cyrillic
and who had not taken Russian classes to complete an oral task in which they were
presented with a noun and they had to provide an agreeing adjective or possessive
pronoun. What Polinsky found was that heritage speakers with lower proficiency
in the language had a two-way gender system (masculine, feminine) whereas those
with higher proficiency in the language, and who still knew declensional classes,
retained a three-way system (masculine, feminine, neuter). It seems that the loss
of case marking led to reanalysis and simplification of the gender classification and
agreement system in heritage speakers with low proficiency in the language.
Another study showing simplification and reanalysis in heritage speakers is
Kim, Montrul and Yoon (2009). The focus of this study was on the interpretation
of three local and long-distance Korean anaphors (caki, casin and caki-casin) in
heritage speakers of Korean. In English the reflexive pronoun himself/herself is an
anaphor that can only be bound locally, within the sentence. The study tested 51
Korean-English bilinguals raised in Korean-speaking families residing in the
United States (22 early bilinguals and 29 late bilinguals) and a group of 34 Korean
monolinguals residing in Korea with a written truth value judgment task. Overall
results indicated that the bilinguals maintained the distinction between local and
long-distance anaphors, though not to the same degree as monolinguals. There
was a tendency among early bilinguals to choose more local binding overall compared to the late bilinguals and Korean monolinguals. At the individual level, many
early bilinguals failed to differentiate between caki-casin and casin in terms of
binding distance, treating both as local anaphors, whereas monolinguals and late
bilinguals tended to collapse caki and casin, treating both as long distance anaphors. Whereas the monolinguals had a three-way anaphor system, the Korean
heritage speakers had a two-way system, suggesting simplification and reanalysis
in the domain of anaphor binding. Hence, simplification and reanalysis are prominent processes in early bilingualism as well, not just in L2 acquisition.
Incomplete L1 acquisition coupled with transfer-induced fossilization can also
lead to what amounts to stabilized dialectal features in some immigrant communities. An example of this scenario can be found in a recent study by Montrul and
Snchez-Walker (2013) testing Spanish-English bilingual children (ages 717) in
the United States, young adult heritage speakers, adult Mexican immigrants, and
age and SES-matched native speaker baseline groups in Mexico on their knowledge and use of differential object marking (DOM) in Spanish (the marking of
animate, specific direct objects with the preposition a, as in Juan vi a Mara
Juan saw Maria).
Young heritage speakers often report their highest and most frequent use of
the heritage language with their parents and relatives. Several studies have documented that native language knowledge can become unstable after many years

Chapter 4. Interlanguage, transfer and fossilization

of exposure to and use of a second language, and this is possible in adult immigrants who may be undergoing attrition (Schmid & Kpke 2013, Montrul 2008,
Schmid 2011).
First language attrition refers to the loss or regression of language skills in native speakers. Attrition is language loss at the individual level. It implies that a
given property of the language reaches a stable endpoint of acquisition at a given
age, but is subsequently lost, again due to reduced exposure to speakers of the
language and to written texts in the language after the offset of schooling. Attrition
is much easier to document and measure in adults than in children, although the
effects of attrition in adulthood seem to be relatively minor for the structural integrity of the grammar as compared to the effects in childhood (Kpke 2007;
Montrul 2008). In some reported cases, the effects of L1 attrition have been minimal. For example, Schmid (2002) found that, after more than 30 years of language
disuse, German Jews living in the United States exhibited some transfer from English but very few actual morphosyntactic errors that could be attributed to L1 attrition. No adult undergoing attrition in a bilingual environment has been shown
to regress in their language to such an extent as to forget how to conjugate verbs,
ask questions, or produce and discriminate native sounds (Keijzer 2007). Yet, in
many documented cases there are measurable changes when the immigrant group
is compared to a group of recent arrivals or native speakers in the country of origin
(Grel & Ylmaz 2013; Perpin 2013; Sorace 2000).
Intergenerational language loss due to incomplete acquisition and attrition
can lead to the emergence of a new dialectal feature in languages in contact. Since
adult immigrants are akin to the parents of the young adult heritage speakers at
the time of testing, this would suggest that heritage speakers may receive at present
qualitatively different (attrited) input from their parents, which would further
contribute to the apparent, arrested development or loss of localized grammatical
features in their Spanish. Montrul and Snchez-Walker (2013) sought to establish
whether some of the linguistic gaps, changes, or errors exhibited by heritage speakers are not only the result of interrupted development but also the result of linguistic patterns transmitted, reinforced, or modeled by the parental generation.
Montrul and Snchez-Walker administered an oral narrative task and an elicited
production task to eight groups: simultaneous bilingual children, sequential bilingual children, monolingual children from Mexico, simultaneous bilingual heritage
speakers, sequential bilingual heritage speakers, adult Mexican immigrants, young
native speakers from Mexico and older native speakers from Mexico. The results
showed that the children and adults tested in Mexico produced the required a
marker more than 95% of the time, while the bilingual children and adults living
in the United States omitted the marker significantly, more than 20% of the time.
Because a good number of adult immigrants who are supposedly the main source

Silvina Montrul

of input to the heritage speakers also omitted the marker, this study suggests that
the linguistic representation of DOM in heritage speakers may not be that different from that of the parental generation. Whereas DOM with animate objects in
monolingual varieties is not only obligatory but is also extending to inanimate
objects, in US Spanish the marker is on the verge of disappearing with animate
objects, probably due to convergence with English. Omission of DOM may actually be on the way to becoming a dialectal feature of US Spanish that is different
from other monolingual varieties. A similar situation is found with Spanish immigrants living in Switzerland (Grosjean & Py 1991), where Spanish is in contact
with French, a language that does not mark DOM. Grosjean and Py (1991) reported that first and second generation Spanish immigrants in a French-speaking
region of Switzerland also accepted ungrammatical sentences without Differential
Object Marking in Spanish, unlike Spanish speakers from Spain who were not living abroad. Therefore, in order to understand the precise source of these grammatical changes from one generation to the next, it is important to investigate the
potential relationship between the grammars of different types of bilingual heritage speakers longitudinally and that of the first generation, which may be the
main source of input to the second generation.
To summarize, transfer, simplification, and fossilization are not just quintessential aberrant features of second language acquisition that make interlanguage
grammars different and inferior to native monolingual grammars (see, however,
Larsen-Freeman, this volume; Ortega this volume). In fact, these mechanisms are
also common and emerge naturally in a first language that did not have a chance
to develop fully in a bilingual environment. When these individual features are
present in members of a speech community, they can be transmitted and reinforced, giving rise to what sociolinguists may call a language variety. Therefore,
transfer, simplification, and fossilization, among others, are not only psycholinguistic phenomena that arise in individual grammars (idiolects) as a natural effect
of bilingualism, as Selinker has suggested; they can also be important mechanisms
of language change at the sociolinguistic and historical level.
Conclusion
More than forty years after the publication of Selinker (1972), the Interlanguage
Hypothesis is still alive and doing great. Selinker called then for a research program that would focus on the psycholinguistic mechanisms underlying the principled nature of the grammatical systems constructed by second language learners
in the language learning process. I have shown in this chapter that, adopting the
Interlanguage Hypothesis as central to its enterprise, generative linguistics research

Chapter 4. Interlanguage, transfer and fossilization

applied to the logical problem of first and second language acquisition the idea
that input and the environment cannot explain everything about the complexity of
linguistic knowledge and how it is acquired has made significant strides in
understanding the principled variability (during development and in ultimate attainment) and cognitive underpinnings of adult second language acquisition. Interlanguage grammars are not wild, rogue grammars, but display systematicity
that can be traced back to influence from the native language and/or the operation
of universal principles that subsume natural languages. Although fossilization as
arrested development has been characterized as a defining feature of child and
adult second language acquisition (Selinker, Dumas & Swain 1975), it can also be
found in the acquisition of native languages acquired early in a bilingual environment (see also discussion in Han, 2013). In many cases, fossilization involves
transfer errors that have not been overcome, so transfer and fossilization can often
be related (cf. Han this volume). It is possible that fossilization in adult second
language learners arises from the operation of learning principles or cognitive
constraints that are different in adults and children, or as Selinker and many researchers within generative approaches see it, because adults no longer have access
to Universal Grammar. In the case of bilingual children who fail to acquire their
family language fully in a situation of societal bilingualism, fossilization may be
brought about by insufficient input and use of the language during the critical period for native language development. Beyond second language acquisition, when
fossilization affects many speakers of a speech community, it can lead to stabilization of a dialectal feature, as with article use in non-native varieties of English and
the loss of Differential Object Marking in Spanish in the United States. This is different from Selinkers original claim that fossilization affects the entire interlanguage competence of entire groups of people.
Still, fossilization and interlanguage in general hold many mysteries that we have
not yet unraveled. Even if fossilization is mainly a psycholinguistic phenomenon,
why does it occur and how does it occur? Is fossilization related to general cognitive
mechanisms of inhibition and control in the bilingual mind? Or is it related to the
executive functions of the mind? If so, why is fossilization more likely to occur in
adults acquiring a second language but also in children losing their first language?
What is the psycholinguistic connection? If language transfer and eventual fossilization are powerful mechanisms of language change that spread from individuals to a
speech community, how do they spread? In addition to deep and detailed linguistic
analyses, which have contributed significantly to understanding the linguistic domains that are more vulnerable to fossilization or incomplete acquisition, we need to
know more about cognitive phenomena and processes, such as implicit/explicit
memory and learning, inhibition and control, and to understand their potential relationship to transfer and fossilization of particular linguistic phenomena.

 Silvina Montrul

To conclude, I have shown that many linguistic features and processes of interlanguage (transfer, fossilization, simplification, universal tendencies) go beyond
adult L2 acquisition and language acquisition in the classroom, because they are
also key factors in real speakers in situations of language contact and change at the
socio-historical level. They are natural psycholinguistic processes and products of
bilingualism and multilingualism. The interlanguages arising from bilingualism
and multilingualism are normal and expected, and since the majority of the worlds
population is multilingual rather than monolingual, it follows that what is actually
considered abnormal today, forty years later, is monolingual behavior. In fact, according to McWhorter (2007), languages which have not had contact with other
languages such as the Nakh-Daghestanian language Chechen are exceedingly
complex, overspecified, and with many quirks. These monolingual languages are
rare and weird! Interlanguages are definitely not.
References
Abrahamsson, N. & Hyltenstam, K. (2009). Age of onset and nativelikeness in a second language: Listener perception versus linguistic scrutiny. Language Learning, 59(2), 249306.
Benmamoun, E., Montrul, S., & Polinsky, M. (2013a). Keynote article. Heritage languages and their
speakers: Opportunities and challenges for linguistics. Theoretical Linguistics, 39, 129181.
Benmamoun, E., Montrul, S., & Polinsky, M. (2013b). Defining the ideal heritage speaker:
Theoretical and methodological challenges. Rely to peer commentaries. Theoretical
Linguistics, 39, 259294.
Bhatt, R. (2002). Experts, dialects, and discourse. International Journal of Applied Linguistics, 12,
74109.
Bley-Vroman, R. (1989). The logical problem of second language learning. In S. Gass & J.
Schachter (Eds.) Linguistic Perspectives on Second Language Acquisition (pp. 4168).
Cambridge: CUP.
Bley-Vroman, R. (1990). The logical problem of foreign language learning. Linguistic Analysis,
20, 349.
Bley-Vroman, R. Felix, S., & Ioup, G. (1988). The accessibility of Universal Grammar in adult
second language learning. Second Language Research, 4, 132.
Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris.
Clahsen, H. & Muysken, P. (1986). The availability of Universal Grammar to adult and child
learners: A study of the acquisition of German word order. Second Language Research, 2,
93109.
Clahsen, H. & Muysken, P. (1989). The UG paradox in L2 acquisition. Second Language Research,
5, 129.
Coulter, K. (1968). Linguistic Error-Analysis of the Spoken English of Two Native Russians.
Unpublished M.A. thesis, University of Washington.
Corder, S. P. (1967). The significance of learners errors. International Review of Applied
Linguistics, 5, 160170.
Corder, S.P. (1973). Introducing Applied Linguistics. Harmondsworth: Penguin.

Chapter 4. Interlanguage, transfer and fossilization

Corder, S.P. (1983). A role for the mother tongue. In S. Gass & L. Selinker (Eds.), Language
Transfer in Language Learning (pp. 1831). Amsterdam: John Benjamins.
DeKeyser, R. (2000). The robustness of critical period effects in second language acquisition.
Studies in Second Language Acquisition, 22, 499534.
DeKeyser, R. (2003). Implicit and explicit learning. In C. Doughty & M. Long (Eds.), The Handbook of Second Language Acquisition (pp. 313348). Malden, MA: Blackwell.
Dulay, H., Burt, M. & Krashen, S. (1982). Language Two. Oxford: OUP.
Ellis, N.C. (2007). SLA: The AssociativeCognitive CREED. In B. VanPatten & J. Williams (Eds.),
Theories in Second Language Acquisition: An Introduction. Cambridge: CUP.
Epstein, S., Flynn, S., & Martohardjono, G. (1996). Second language acquisition: Theoretical and
experimental issues in contemporary research. Brain and Behavioral Sciences, 19, 677758.
Eubank, L. (1996). Negation in early German-English interlanguage: More valueless features in
the L2 initial state. Second Language Research, 12, 73106.
Flynn, S. (1987). A Parameter-setting Model of L2 Acquisition. Dordrecht: Reidel.
Granena, G. & Long, M. (2013). Sensitive Periods, Language Aptitude and L2 Ultimate Attainment. Amsterdam: John Benjamins.
Grosjean, F. & Py, B. (1991). La restructuration dune premire langue: Lintgration de variantes
de contact dans la comptence de migrants bilingues. La Linguistique, 27, 3560.
Grel, A. & Glmaz, G. (2013). Restructuring in the L1 Turkish grammar: Effects of L2 English
and L2 Dutch. In M. Schmid & B. Kpke, B. (Eds.), First Language Attrition (pp. 3766).
Amsterdam: John Benjamins.
Han, Z-H. (2012). Second language acquisition. In J. Banks (Ed.), The Encyclopedia of Diversity
in Education (pp. 19101915). Thousand Oaks, CA: Sage.
Han, Z-H. (2013). State-of-the-art article: Forty years later Updating the Fossilization Hypothesis. Language Teaching, 46(2), 133171.
Hawkins, R. & Chan, C. (1997). The partial availability of Universal Grammar in second language acquisition. The failed functional features hypothesis. Second Language Research, 13,
187226.
Herschensohn, J. (2007). Language Development and Age. Cambridge: CUP.
Ionin, T. (2003). Article Semantics in Second Language Acquisition. Unpublished PhD dissertation, MIT.
Ionin, T., Ko, H., & Wexler, K. (2004). Article semantics in L2acquisition: The role of specificity. Language Acquisition, 12, 369.
Ionin, T., & Montrul, S. (2010). The role of L1transfer in the interpretation of articles with
definite plurals in L2English. Language Learning, 60(4), 877925.
Johnson, J., & Newport, E. (1989). Critical period effects in second language learning. The influence of maturational state on the acquisition of English as a second language. Cognitive
Psychology, 21, 6099.
Kanno, K. (1997). The acquisition of null and overt pronouns in Japanese by English speakers.
Second Language Research, 13, 26587.
Keijzer, M. (2007). Last in First out? An Investigation of the Regression Hypothesis in Dutch Emigrants
in Anglophone Canada. Amsterdam: LOT, Netherlands Graduate School of Linguistics.
Kim, J.-H., Montrul, S., & Yoon, J. (2009). Binding interpretation of anaphors in Korean heritage
speakers. Language Acquisition, 16(1), 335.
Klein, E. (1993). Toward Second Language Acquisition: A Study of Null Prep. Dordrecht: Kluwer.
Kohnert, K. & Bates, E. (2002). Balancing bilinguals II. Lexical comprehension and cognitive
processing in children learning Spanish and English. Journal of Speech, Language, and
Hearing Research, 45, 347359.

 Silvina Montrul
Kohnert, K., Bates, E., & Hernndez, A. (1999). Balancing bilinguals, Lexical-semantic production and cognitive processing in children learning Spanish and English. Journal of Speech,
Language and Hearing Research, 42, 14001413.
Kpke, B. (2007). Language attrition at the crossroads of brain, mind, and society. In B. Kpke,
M. Schmid, M. Keijzer, & S. Dosterst (Eds.), Language Attrition. Theoretical Perspectives
(pp. 937). Amsterdam: John Benjamins.
Lado, R. (1957). Linguistics across Cultures. Ann Arbor MI: University of Michigan Press.
Lardiere, D. (1998a). Case and tense in the fossilized steady state. Second Language Research, 14,
126.
Lardiere, D. (1998b). Dissociating syntax from morphology in a divergent end-state grammar.
Second Language Research, 14, 359375.
Lardiere, D. (2007). Ultimate Attainment in Second Language Acquisition. A Case Study. Mahwah,
NJ: Lawrence Erlbaum Associates.
Lenneberg, E. (1967). Biological Foundations of Language. New York, NY: Wiley.
Li, M.-C. (2012). The Acquisition of Tense and Agreement by Early Child Second Language
Learners. Unpublished PhD dissertation, University of Illinois at Urbana-Champaign.
Liceras, J. (1986). Linguistic Theory and Second Language Acquisition. Tbigen: Gunter Narr.
Long, M. (2007). Problems in SLA. Mahwah, NJ: Lawrence Erlbaum Associates.
Lozano, C. (2002). The interpretation of overt and null pronouns in non-native Spanish.
Durham Working Papers in Linguistics, 8, 5366.
McWhorter, J. (2007). Language Interrupted. Signs of Non-native Acquisition in Standard
Language Grammars. Oxford: OUP.
Meisel, J. (1997). The acquisition of the syntax of negation in French and German: Contrasting
first and second language development. Second Language Research, 13, 227263.
Meisel, J. (2011). First and Second Language Acquisition. Cambridge: CUP.
Montalbetti, M. (1984). After Binding: On the Interpretation of Pronouns. Unpublished PhD
dissertation, MIT.
Montrul, S. (2008). Incomplete Acquisition in Bilingualism. Re-examining the Age Factor.
Amsterdam: John Benjamins.
Montrul, S. (2012). Is the heritage language like a second language? EUROSLA Yearbook, 12,
129.
Montrul, S., & Ionin, T. (2010). Transfer effects in the interpretation of definite articles by
Spanish heritage speakers. Bilingualism: Language and Cognition, 13(4), 449473.
Montrul, S., & Ionin, T. (2012). Dominant language transfer in Spanish heritage speakers and L2
learners in the interpretation of definite articles. The Modern Language Journal, 96(1), 7094.
Montrul, S., & Snchez-Walker, N. (2013). Differential object marking in child and adult Spanish heritage speakers. Language Acquisition, 20, 124.
Montrul, S., Bhatt, R., & Bhatia, A. (2012). Erosion of case and agreement in Hindi heritage
Speakers. Linguistic Approaches to Bilingualism, 2, 141176.
OGrady, W., Kwak, H-K, Lee, M., & Lee, O-S. (2011). An emergentist perspective on partial
language acquisition. Studies in Second Language Acquisition, 33, 223246.
Paradis, M. (2004). A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins.
Prez-Leroux, A.T., & Glass, W. (1999). Null anaphora in Spanish second language acquisition:
Probabilistic versus generative approaches. Second Language Research, 15, 220249.
Perpin, S. (2010). On L2 Grammar and Processing: The Case of Oblique Relative Clauses and
the Null-prep Phenomenon. Unpublished PhD dissertation, University of Illinois at UrbanaChampaign.

Chapter 4. Interlanguage, transfer and fossilization

Perpin, S. (2013). Optionality in bilingual native grammars. In M. Schmid & B. Kpke (Eds.)
First Language Attrition (pp. 127 156). Amsterdam: John Benjamins.
Polinsky, M. (2006). Incomplete acquisition: American Russian. Journal of Slavic Linguistics, 14,
191262.
Polinsky, M. (2008). Russian gender under incomplete acquisition. Heritage Language Journal,
6(1). <http://www.heritagelanguages.org/>
Rothman, J. (2008). Aspectual Selection in Adult L2 Spanish and The Competing Systems Hypothesis: When Pedagogical and Linguistic Rules Conflict. Languages in Contrast, 8(1),
74106.
Schachter, J. (1990). On the issue of completeness in second language acquisition. Second Language Research, 6, 93124.
Schmid, M. (2002). First Language Attrition, Use and Maintenance. The Case of German Jews in
Anglophone Countries. Amsterdam: John Benjamins.
Schmid., M. (2011). First Language Attrition. Cambridge: CUP.
Schmid, M. & Kpke, B. (2013). First Language Attrition. Amsterdam: John Benjamins.
Schwartz, B. D., & Sprouse, R. (1994). Word order and nominative case in non-native language
acquisition: A longitudinal study of (L1 Turkish) German interlanguage. In T. Hoekstra &
B. D. Schwartz (Eds.), Language acquisition studies in generative grammar (pp. 317368).
Amsterdam: Benjamins.
Schwartz, B., & Sprouse, R. (1996). L2 cognitive states and the full transfer/full access hypothesis. Second Language Research, 12, 4072.
Sharma, D. (2005a). Dialect stabilization and speaker awareness in non-native varieties of
English. Journal of Sociolinguistics, 9(2), 194224.
Sharma, D. (2005b). Language transfer and discourse universals in Indian English article use.
Studies in Second Language Acquisition, 27(5), 535566.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics and Language
Teaching (IRAL), 10(3), 209231.
Selinker, L., Dumas, & Swain, M. (1975). The Interlanguage Hypothesis extended to children.
Language Learning, 25(1), 139152.
Silva-Corvaln, C. (1994). Language Contact and Change. Spanish in Los Angeles. Oxford: OUP.
Sorace, A. (2000). Differential effects of attrition in the L1 syntax of near-native L2 speakers.
Proceedings of the 24th Boston University Conference on Language Development (pp.719
725). Somerville, MA: Cascadilla Press.
Slabakova, R. (2001). Telicity in the Second Language. Amsterdam: John Benjamins.
Sorace, A. (1993). Incomplete vs. divergent representations of unaccusativity in native and nonnative Italian grammars of Italian. Second Language Research, 9, 2247.
Thomas, M. (2013). History of the study of second language acquisition. In J. Herschensohn
& M. Young-Scholten (Eds.), The Cambridge Handbook of Second Language Acquisition
(pp. 2645). Cambridge: CUP.
Thomason, S. (2001). Language Contact. Washington, DC: Georgetown University Press.
Thomason, S. & Kaufman, T. (1988). Language Contact, Creolization, and Genetic Linguistics.
Berkeley, CA: University of California Press.
Trudgill, P. (2001). Contact and simplification. Historical baggage and directionality in linguistic change. Linguistic Typology, 5, 37174.
Tsimpli, I., & Roussou, A. (1991). Parameter resetting in L2? University College London Working
Papers in Linguistics, 3, 14969.

 Silvina Montrul
Tsimpli, I., & Dimitrakopoulou, M. (2007). The interpretability hypothesis: Evidence from whinterrogatives in second language acquisition. Second Language Research, 23, 215242.
Ullman, M. (2001). The neural basis of lexicon and grammar in first and second language, the
declarative/procedural model. Bilingualism: Language and Cognition, 4, 105122.
Vainnika, A. & Young-Scholten, M. (1996). Gradual development of L2 phrase structure. Second
Language Research, 12, 739.
Weinreich, U. (1953). Languages in Contact. The Hague: Mouton.
White, L. (1989). Universal Grammar and Second Language Acquisition. Amsterdam, John
Benjamins.
White, L. (2003a). Second Language Acquisition and Universal Grammar. Cambridge: CUP.
White, L. (2003b). Fossilization in steady state L2 grammars, persistent problems with inflectional
morphology. Bilingualism: Language and Cognition, 6, 129141.

chapter 5

The limits of instruction

40 years after Interlanguage
Bill VanPatten

Michigan State University

This chapter argues that Selinkers (1972) claim that instruction does not
significantly affect interlanguage development is essentially correct. Reviewing
general research on instructed second language acquisition as well as some
recent research of my own, I argue that instructed SLA to date has failed to
consider underlying constraints and processes in interlanguage development. In
addition, I argue that the fundamental problem in instructed SLA is its overall
focus on the acquisition of rules; that is, rules are not acquired from the input.
Instead, learners process morpho-phonological units in the speech stream and
assemble language over time. Rules, if they exist, evolve; they cannot be the
object of instruction or input processing.

Introduction
For some 30 years, a central concern of the research in instructed SLA has been
this question: To what extent does instruction in the formal properties of language
lead to their acquisition and/or move learners toward native-like linguistic representation and ability? In anticipation of this question, Selinker (1972) made the
following claim in his seminal paper, Interlanguage.
The second-language learner who actually achieves native-like competence cannot possibly have been taught this competence, since linguists are daily in almost
every generative study discovering new and fundamental facts about particular
languages. Successful learners, in order to achieve this native-speaker competence, must have acquired these facts (and most probably important principles of
language organization) without having explicitly been taught them. (p. 212213)

The purpose of the present chapter is to argue that Selinkers claim was, and still is,
essentially correct. More importantly, I will extend Selinkers claim to make the
following point: at any stage of acquisition, mental representation (defined below)

 Bill VanPatten

of language is largely unaffected by instructional efforts directed at formal properties of language.1 In so doing, I acknowledge here the theoretical and empirical
work offered by Stephen Krashen, Bonnie Schwartz, Nick Ellis, John Truscott,
Michael Sharwood Smith, and others who have grappled with the role of instruction, most of whom have come to similar conclusions as those outlined here. As is
the case in most science, ideas flow from other ideas. My hope is that I can put a
few new thoughts into the flow while staying true to the purpose of the present
volume and the symposium on which it was based: namely, linking where we are
today with the claims made in Selinker (1972).
First, some definitions
Language
Given Selinkers reference to generative research (see quote in introduction), I take
his notion of competence to be Chomskyan in nature: what I call underlying mental
representation of language (VanPatten, 2010, 2011, 2013a; VanPatten & Rothman,
in press). This representation has three basic characteristics. First, this representation is implicit in that we have difficulty articulating its content. Although we may
be aware of its existence (i.e., we all know we have language in our heads), we are
unaware of its exact nature. The non-linguist, for example, cannot offer a rule for
why (1) sounds fine but (2) sounds funny if not awful (i.e., not all verbs can combine with re-).

(1) Sam painted the office but didnt like it, so he repainted it several shades
lighter.
(2) Sam slept for an hour but was still tired so he reslept for another hour.

Likewise, the non-linguist might know that (3) is how you make a yes/no question
about Bill and vodka but that in Spanish this is impossible and instead the typical
yes/no question appears in (4).

(3) Does Bill drink vodka?

(4) Toma Bill vodka?

That person also knows that trying to make English like Spanish is disallowed:
*Drinks Bill vodka? But that same non-linguist cannot tell you why English does
one thing and Spanish does another. (I will refrain from discussion about what
1. One could make a similar argument for skill or proficiency, but for the purpose of the
present argument I limit my remarks to linguistic representation.

Chapter 5. The limits of instruction

linguists might say is the difference between English and Spanish for yes/no questions, leaving that for later.)
The second characteristic of the mental representation of language is that it is
abstract. Unlike simple textbook rules of thumb and laypersons explanations of a
rule, the explanation for why languages behave the way they do involves abstract
notions such as features, syntactic operations (e.g., movement, agreement), principles and constraints, and parametric variation, among others. For example, the
deceptively simple yes/no question in English such as Does Bill drink vodka? can be
described by the layperson (and teacher) as insert the auxiliary verb do, but the
reasons why do has to be inserted in the first place and why it winds up higher in
the sentence compared to the main verb is explained by referencing abstract features such as Q and T. (Again, I will return to yes/no questions later so I will not
explicate here.) Thus, what the layperson or teacher offers as a description of what
happens is a description from the outside and not a description of actual underlying competence or representation. What the teacher offers as a rule is not what
winds up in the learners mind/brain.2
The third characteristic of competence as it is used here is that it is interactive. Interactivity means that different parts of the underlying representation are
in communication with each other to create, accept, comprehend, and otherwise deal with sentences and utterances in a language. For example, the syntax
and the lexicon (where morphology resides3) of a language need to be in communication so that agreement occurs and particular operations produce appropriate sentences. Similarly, syntax communicates with discourse so that such
things as anaphoric interpretations (that are not strictly governed by syntax) can
be achieved. An example of this latter situation is found in Spanish with null and
overt subject pronouns. In (5a) and (5b), ambiguity is present in that either the
null (pro) or the overt (l) subject pronoun could refer to either antecedent in the
previous clause.
(5) a. Juan vio a Roberto mientras pro caminaba en la playa.
b. Juan vio a Roberto mientras l caminaba en la playa.
John saw Robert while he was walking on the beach
2. One reviewer suggested I insert a qualifier (is not what necessarily winds up) because we
do not categorically know what is in peoples heads/minds. What is clear from this chapter is that
I take a generative perspective on underlying representation. However, the relationship between
rules provided to learners and what winds up in their heads is untenable under just about any
theoretical position about underlying knowledge (e.g., Emergentism, functionalist approaches,
Processability Theory). See also note 4.
3. In the version of generative grammar I use, this is the case. See Sproat (1998) and Harley
and Noyer (1999) for alternative accounts of morphology.

 Bill VanPatten

There is nothing in the syntax of Spanish that requires linking one pronoun type
to one antecedent and the other pronoun type to another antecedent; that is,
either pronoun (pro/l) is (technically) free to select either antecedent (Juan/
Roberto), yet Spanish speakers tend to link null subjects with previous subject
nouns at the rate of about 70%, while allowing overt subjects to take either antecedent (the rate is about 50%) (see Alonso-Ovalle et al., 2002, as well as Keating,
VanPatten & Jegerski, 2011). Discourse constraints and preferences, then, provide information to the syntax about which pronouns are more likely to link to
which antecedents. This is different from a strictly syntactic constraint, such as
the Overt Pronoun Constraint (OPC), which blocks overt subject pronouns from
taking quantified and wh-antecedents such as no one, not +Noun, which + Noun,
and so on (Montalbetti, 1984). (Examples will appear later under the discussion
about poverty of the stimulus.) Discourse preferences do not enter into such
a constraint.
The point to be underscored here is that my claims about the limited role of
instruction are claims about instructional efforts at altering mental representation. Because mental representation is what allows and disallows the actual language we use (i.e., utterances we speak), mental representation, then, is central
to SLA, just as it is central to first language acquisition and use, and by central, I
do not mean central to all investigators research agenda. Clearly, someone interested in something like the social aspects of SLA would not be looking to see
whether English L1 learners of Spanish L2 have reset the parametric value of
null subject to + null subject (e.g., Block, 2003, and some of the chapters in
Atkinson, 2011). What I mean by central is that no matter what we investigate in
L2 research, mental representation is either the object of investigation or it is
assumed in some way.4
Formal Instruction
Formal instruction is one of those terms that can mean a variety of efforts. Here I
am going to restrict the definition to include only the following:
4. To be sure, emergentists and those working in something like skill theory do not conceptualize representation the way it is conceptualized here. But, in those paradigms, there is some
kind of representation. For example, emergentists view what looks like rules as a series of contingencies based on frequency distributions and like generativists they might use a term such as
agreement to talk about something that is not really a rule in the classic sense. In skill theory,
it is procedural knowledge (although procedural knowledge is not well articulated in skill theory compared to representation in generative and emergentist accounts). It is not clear to me how
those working within social frameworks conceptualize language, but presumably there is something in the learners mind/brain called language assumed in these frameworks as well.

Chapter 5. The limits of instruction

any purposeful effort to induce new knowledge in the learner;

attempts that may be external to the learner (e.g., from a teacher, from the
environment) or may come from the learner him/herself (e.g., self-guided
learning);
typically, explicit pedagogical or descriptive rules about language (including
formal features and paradigms);
typically, explanation of/reading about a target rule or formal features coupled
with some kind of practice to induce the new knowledge.
Excluded from this definition is something like processing instruction, an intervention that is not targeted at rule induction or a change in learner knowledge per
se. Instead, the focus of processing instruction is on altering processing behaviors, which of course should affect intake for the developing system (for more
detailed information on this, see VanPatten & Rothman, in press, as well as
VanPatten, forthcoming, although I will touch upon this issue again later in this
chapter).
The evidence for limited instructional effects
There are two major categories of evidence pointing toward the limited effects
of instruction. The first involves stages of acquisition. The second involves
what is traditionally referred to as the poverty of the stimulus. I will take each
in turn.
Staged development
For over four decades we have been aware that there is staged development in the
acquisition of various structures (e.g., Ellis, 2008; Lightbown & Spada, 2013). For
example, we have research on the acquisition of negation and wh-questions in
English, negation in German, copular verbs in Spanish, gender agreement in
Spanish, clitic object placement in French, and many others in various languages.
What all such research suggests is that learners construct an interlanguage from
the outset that is (largely) independent of any attempts to directly affect the formal
properties of the structure in question. We will remind ourselves here with an example from English negation, with research conducted in and out of classroom
settings. The well-known stages of the acquisition of negation in English appear
below (see also Ortega, this volume).

 Bill VanPatten

Stage 1: no + predicate
No drink beer No beer
Stage 2: subject + no + predicate
I no drink beer
Stage 3: unanalyzed dont alternates with no I dont/no drink beer He dont/no
drink beer
Stage 4: modals appear
I wont/will not drink beer
Stage 5: analyzed dont emerges
I dont/do not drink beer

He doesnt/does not drink beer
What is immediately clear from examining these stages is that only the final two
stages resemble anything that is explicitly taught to and practiced by learners. That
is, stages 13 seemingly ignore any instructional efforts. What is more, no instructor or curriculum purposefully teaches stage 1 and 2 structures; that is, no
teacher tells students about no + VP sentences and then has students practice such
structures. This just isnt English. Yet, learners do it anyway. This stage cannot be
traced to instructional efforts, then.5 If we examined any staged development for
any structure, we would see the same. What is more, a good number of these stages resemble those in English L1 acquisition, also suggesting a limited effect or no
effect of formal instruction.
A similar example is found in so-called U-shaped behavior. In U-shaped behavior, learners seemingly start off doing something correctly, only to go through
a phase where they do it incorrectly, and then later show evidence of re-acquisition of the structure. The most well known example comes from past tense. The
stages look something like this:
Stage 1: early acquisition of highly frequent irregulars (e.g., went, ate)
Stage 2: a cquisition of regulars (e.g., talked, studied) and simultaneous regularization of irregulars (e.g., wented/goed, ated/eated)
Stage 3: re-emergence of irregulars (e.g., went, ate)
In this scenario, the U-shaped behavior refers to irregulars: learners start off fine,
lose the irregulars, and then get them back. Again, no instructor or curriculum
teaches learners to regularize irregulars, yet learners do it anyway (just as child L1
learners do). Such patterns suggest that the instruction aimed at altering these
stages is not effective.
Both staged development and U-shaped behavior, then, suggest that Interlanguage is not directly shaped by instruction on formal elements. Although Selinker
focused on the end product (native-like competence), research on staged development and U-shaped behaviors suggests that we dont have to wait until the end
5. The reader is directed to Ellis (2008) as well as Kessler, Liebner, and Mansouri (2011) for
particular research that demonstrates that formal instruction does not alter developmental
stages.

Chapter 5. The limits of instruction

product to see whether or not instruction affects an underlying competence. My

position is that the effects of instruction and explicit learning do not affect underlying representation of language at any stage. We turn our attention now to other
evidence.
Poverty of the stimulus
From the generative perspective, the second major piece of evidence that points
toward a limited role for instruction comes from poverty of the stimulus (POS)
situations. POS situations refer to those in which people (language learners) come
to know more about a language than they could have learned from exposure to
input or from some kind of instruction and practice. POS situations abound in the
literature (e.g., Schwartz, 1998; White, 2003), and I offer a classic example of a POS
situation: the Overt Pronoun Constraint (OPC). In null subject languages like
Spanish, null and overt pronouns are both possible in simple declarative sentences
such as (6a) and (6b), while overt pronouns are disallowed in sentences for which
the null subject has no antecedent or referent as in (7), (8), and (9).
(6) a. Le el manuscrito.
I read the manuscript
b. Yo le el manuscrito.
(7) Est lloviendo/*Ello est lloviendo.6
Its raining.
(8) Es probable que venga/*Ello es probable que venga.
Its probable hell come.
(9) Me han robado/*Ellos me han robado.
They robbed me (with no referent for they, as in I was robbed).
In languages like Spanish, the OPC comes into play in situations in which a quantified or wh-antecedent precedes an overt pronoun. While the null pronoun is free
to take any antecedent that matches in number and gender, overt pronouns are
barred from taking quantified and wh-antecedents as exemplified in (11) and (12)
below (where i refers to the antecedent indicated and j refers to an antecedent
outside the sentence).
(10) El policai admiti que proi,j /l i,j acept el soborno.
The cop admitted he took the bribe.
6. There is at least one dialect of Spanish for which some speakers produce and accept an overt
pronoun in these contexts, suggesting that this one dialect may be moving toward being a nonnull subject language (see, for example, Toribio, 2000).

 Bill VanPatten

(11) Nadiei admiti que proi,j /l *i,j acept el soborno.

Nobody admitted that he took the bribe.
(12) Quini admiti que proi,j /l *i,j acept el soborno?
Who admitted that he took the bribe?
This constraint on what an overt pronoun may take as an antecedent cannot be
gleaned from the input, nor is it something taught to child L1 learners or, for that
matter, L2 learners. Yet this constraint is operative in the grammars of both L1
and L2 speakers of null subject languages (e.g., Kanno, 1998; Prez-Leroux &
Glass, 1999).
Another type of POS situation occurs when learners reset what is called a parameter. In this scenario, learners do indeed acquire something from the input,
but show evidence of extending it to aspects of the grammar they have not yet
encountered in the input. However, a good deal of POS research examines end
product, much as Selinker discussed. Given the claim here is that interlanguages
are constructed independently of instruction all along the way, is there evidence
for POS situations at the outset of acquisition? There is. I will illustrate with a recent example from my own research.
In Smith and VanPatten (2013), we were interested in what English L1 learners
of Japanese L2 do in the earliest stages of learning a head-final language. English
and Japanese are markedly different from each other in terms of phrase structure:
English is rigidly head-initial and Japanese is rigidly head-final. This means that in
English we get verb-object, preposition-noun, complementizer-clause, and similar
orders in phrases where verbs, prepositions, and complementizers are the heads of
the phrases. The opposite is true in Japanese: object-verb, noun-postposition, and
clause-complementizer. Classically, this kind of difference is referred to as parametric variation, although the nature and meaning of parameters has changed over
the last 20 years (see Boeckx, 2011, and Gallego, 2011, for some discussion). In our
study, then, we wanted to see if learners of Japanese L2 with English L1 could reset
the parametric value of head directionality early on in acquisition. We tested this
by teaching Japanese to people who knew no Japanese or any other head-final language. Instruction occurred in a controlled laboratory situation using a simple
computer tutorial. After instruction, we surprised them with a reading test that
contained phrases and structures they had not been exposed to but tested whether
the participants had projected beyond the rather limited input data.
Sixty learners participated in the experiment. During an input treatment, they
simultaneously listened to and read simple Japanese sentences with SOV and
SPPOV word orders such as Taro-ga ringo-o taberu Taro is eating an apple and
Taro-ga kichin-de ringo-o taberu Taro is eating an apple in the kitchen. The treatment consisted of 100 sentences, and most learners completed the treatment in

Chapter 5. The limits of instruction

about 4045 minutes. Immediately after treatment, participants completed a surprise posttest to measure their sensitivity to grammatical violations. During the
test, participants read sentences on a computer screen and then advanced to a
question about the content of the sentence. This test included sentences with both
grammatical SOV word order and ungrammatical SVO word order. To be sure,
these were sentences similar to those they were trained on. However, the test also
contained grammatical and ungrammatical yes/no questions that in Japanese are
made with the particle ka. So, participants saw both SOV-ka and *ka-SOV sentences (e.g., Taro-ga ringo-o taberu ka? and *Ka taro-ga ringo-o taberu). Important
here is that the participants were not exposed to questions during the treatment.
The test also included grammatical and ungrammatical sentences with embedded
clauses, using the compementizer toh. These sentences were either S [SOV-toh] V
or S [*toh-SOV] V (e.g., John-wa Taro-ga ringo-o taberu toh itta John said Taro is
eating an apple and *John-wa toh Taro-ga ringo-o taberu itta). Again, these were
novel structures to which the learners had not been exposed during treatment. We
used the reading times for the stimulus sentences as a measure of grammatical
sensitivity; that is, if learners were sensitive to grammatical violations, they should
read ungrammatical sentences slower as they attempted to read for meaning.
All participants were sensitive to violations of basic word order; that is, they
read *SVO sentences slower than SOV sentences. So, we at least know that they
picked up basic order from the input. The striking part of the results was found with
the novel sentences. All participants demonstrated sensitivity to violations of yes/no
word order, reading *Ka-SVO more slowly than SVO-ka sentences. That is, something was bugging them about ka (which is the head of its own phrase, CP, and
therefore follows the verb in Japanese) being in the wrong position: they preferred
the sentences where the word order was clause-Comp. As for sentences with embedded clauses, about 50% of the participants demonstrated sensitivity to word order
violations with complementizers, reading S [*toh-SOV] V sentences more slowly
than S [SoV-toh] V sentences. For this group of participants, again, something was
bugging them about toh not being in final position (i.e., clause-toh). The reason
these findings are so interesting is that these are structures the learners had not been
exposed to. After only 100 input sentences with basic word order, the participants in
our study demonstrated sensitivity to novel phrase structure, suggesting they had
already begun to reset head-directionality in Japanese. In short, they were projecting
beyond the data and extending head-finality to all parts of the grammar.
Our study was not about instruction per se (i.e., we were not testing whether
instruction made a difference or not); instead, its focus was the nature of parameter resetting and the status of parameters in L2 research. I cite it here because I
think it suggests something about the power of the internal language-making
mechanism. The participants in our study had not received any instruction (as

 Bill VanPatten

defined at the outset of this chapter) on phrases or phrase structure, had not been
exposed to questions and embedded clauses during the input treatment, and certainly hadnt been taught how yes/no questions and embedded clauses are formed.
Yet they showed evidence of sensitivity to grammatical violations of word order
with these structures. So, after basically 45 minutes of exposure to a language they
had never seen before (and where they had no knowledge of a like language), these
learners demonstrated a POS situation regarding phrase structure. The conclusion
I draw here based on these results is that non-instructional forces for language
creation in the minds of learners start working on input from the get go, extending
Selinkers original claim to the earliest stages of acquisition.
During the review of this chapter, one issue that surfaced a number of times
concerns the nature of instruction. Different reviewers made queries such as Isnt
an input treatment instruction? and Can you address deductive vs. inductive
approaches? Such queries suggest to me that readers may miss an important
point: one has to define instruction in some way, as I have done in this chapter.
Otherwise, instruction can be anything one wants it to be and the construct becomes meaningless. Does interaction with native speakers count as instruction,
then, especially when those speakers provide interactive feedback? Does the teacher merely talking to students in class constitute instruction? Such questions are
reminiscent of certain claims made in the L1 literature that caretaker speech to
children is a kind of instruction, which the vast majority of L1 researchers find to
be a forced claim at best. What is more, the kinds of comments made by the reviewers regarding the nature of instruction suggest to me that they may also be
missing a fundamental point of this chapter: that there are no rules to be instructed. This issue will emerge after the next section.
Yes, but...
In laying out the case for a very limited or non-role for formal instruction using
staged development and POS situations, a natural question or objection comes to
mind: Instruction was never meant to be about universals or parameters. Instruction was meant to help learners with language specific surface properties, such as
inflections, case markings, and so on. And this is largely true.7 But there are two
7. Megan Smith (personal communication) reminds me that surface properties of language
are often manifestations of underlying representation, so the idea that instruction is not meant
to affect representation is probably not true. That is, even though instructors and those in instructed SLA research might make the claim that instruction is about surface properties, the fact
that surface properties are often tied to underlying representation means that whether instruction intends to or not, it is trying to deal with representation at some point.

Chapter 5. The limits of instruction

issues underlying this objection. The first is that in staged development, the kinds
of things studied by acquisitionists are actually the targets of instruction. That is,
learners get explicit instruction and practice on negation, question formation, past
tense formation, copular verbs in Spanish, and so on. So, in this case, the argument
that instruction was never meant to affect universals or parameters is a non-
argument. The very thing studied in staged development is the object of instruction, and yet instruction shows no effects, even in the early stages.8
The second issue involves the effects of instruction on so-called language specific properties. Does instruction indeed make a difference? Is learner knowledge
about language specific properties due to instruction? To address this, I would like
to examine another recent study, this one involving syntax and verb morphology
in Spanish as L2.
In VanPatten, Keating, and Leeser (2012), we tested native speakers and
intermediate-level (3rd-year university students beginning their formal study of
literature and culture) on three structures in Spanish. All were related to verb
movement in some way: (1) wh-question formation; (2) adverb placement with
no ms (no longer); and person-number endings on simple present tense verbs.
All participants were tested for sensitivity to grammaticality via self-paced reading. In self-paced reading (SPR), participants read a sentence fragment by fragment on a computer screen, controlling what they see and how long they read by
pressing a button. All reading is for meaning; after each sentence, SPR participants advance to a content question about what they have just read. We used a
non-cumulative reading test, which means that fragments of the sentence appeared and disappeared as participants pushed a button to advance their way toward the end of the sentence. In this way, readers are required to keep what they
have just read in working memory as they move from fragment to fragment.
Comprehension questions appear on a separate screen after reading a sentence.
Reading times of target fragments and the spillover region right after the targets
are measured (other reading times can be measured as well, but for the present
purpose, reading times of other fragments are not required). Because the sentences are paired so that there are grammatical and ungrammatical versions of
each (randomized, counterbalanced, blocked, surrounded by distractors and
8. One reviewer made the following point: ...eventually, some learners manage to get the
structure right. Would that end-product speak against your central claim that instruction has
little impact on acquisition? The answer is no. Learners who get to the end product get there
in spite of or independently of instruction. This has to be the case, or no non-classroom/uninstructed learner would ever get to such stages, and yet some do. To make instruction the causative factor begs the question of how non-instructed learners develop representation. What is
more, a good deal of later staged learning happens well after formal instruction (e.g., during
study abroad, after one moves to a new culture and begins a life there).

 Bill VanPatten

fillers, and so on), we are interested in how long it takes participants to read the
target and spillover regions of the same grammatical and ungrammatical sentences. Even though participants are reading for meaning, the expectation is that
they will slow down slightly on ungrammatical segments in one of the two regions
examined as their internal processors detect something wrong (see the previous
discussion on the Smith & VanPatten study, which also measured reading times
but for the entire sentence).
For the wh-questions, grammaticality had to do with subject-verb inversion
(grammatical sentences had the inversion, ungrammatical sentences did not as in
Dnde cenan tus padres cuando van a Chicago? Where do your parents eat when
they go to Chicago? vs. *Dnde tus padres cenan cuando van a Chicago?). For the
placement of the adverb no ms, grammaticality had to do with whether the verb
had moved out of its VP or not (e.g., Juan no viaja ms a Francia porque no tiene
dinero John no longer travels to France because he doesnt have money vs. *Juan
no ms viaja a Francia porque no tiene dinero). For person-number, the grammaticality had to do with subject-verb morphological agreement (e.g., Ahora yo
tomo un refresco en la cafetera Right now Im drinking a soda in the cafeteria vs.
*Ahora yo toma un refresco en la cafetera).
The results were clear. The native speakers showed significant reading time
differences on all sentences. That is, they slowed down on the ungrammatical sentences for all three structures, thus demonstrating sensitivity to grammaticality on
all structures. The L2 learners slowed down on the ungrammatical wh-questions
and the ungrammatical adverb sentences. They did not slow down on the personnumber sentences. In short, they demonstrated grammatical sensitivity on two
structures but not on the third.
What is interesting about these results and why they are relevant to the present
discussion is that wh-question formation is not taught in most Spanish classes and
adverb placement never is. These structures are consequences of underlying feature specification briefly described at the outset of this chapter. Clearly, the L2
participants were demonstrating some kind of underlying representation for this
feature and for the parameterized consequences of it in Spanish. It is true that
wh-questions are readily encountered in the input, and so the results could be due
to simple exposure in the case of this structure (and this would be true in or out of
instructional settings). However, the particular adverb type we selected for the
study is rare in the input, and generally not present in input to early stage learners,
especially classroom learners. In a real sense, the learners in VanPatten, Keating
and Leesers study behaved like the Japanese L2 participants in the Smith and
VanPatten study described in the previous section; they evidenced a POS situation
for adverbs that could not be traced to instruction or to input.

Chapter 5. The limits of instruction

More importantly for the Yes, but... aspect of the present discussion, the
L2 participants did not show underlying lexical representation for the morphophonological units related to person-number on verbs. These are a particular aspect of Spanish surface grammar that are taught and practiced from day one. Yet,
after three years of college-level Spanish, these learners did not show evidence of
any sensitivity to the very thing that has been present in their formal instruction
from the first day of Spanish classes.
Once again, this particular study was not intended as an instructional study, but
I cite it here because its results are relevant to the present discussion. First, learners
demonstrated sensitivity to violations for those things that were not taught or practiced. Second, learners failed to demonstrate any sensitivity to violations for those
things that were taught and practiced, and for which they received feedback and for
which they were explicitly tested over the course of three years of study. And, by all
accounts for teachers of Spanish, person-number agreement on present tense verbs
is pretty basic. So, one could rightly ask the questions What the heck happened to
all that instruction and practice (and testing) on verb forms? And if instruction is
not helping learners get basic person-number agreement in their grammatical systems, how does such a thing come to be represented in the grammar of the learner?
What are the processes and mechanisms involved? What we concluded at the end
of our study was that things like person-number endings on verbs must be learned
from the input like anything else; they cant be taught and practiced in order to build
a mental representation of them. When we examined the input of typical classrooms and textbook materials, we discovered how relatively poor the input is in
terms of providing lots of samples of the various person-number endings. Thirdperson singular (and plural) tend to dominate the input, and overwhelmingly so.
So, why these non-effects? Against rules
If the argument here is correct, namely, that (1) Selinker was right about competence not being traceable to instruction, and (2) his argument can be extended to
every stage of acquisition, even initial stages, then a natural question is why we do
not see the effects of formal instruction on underlying competence. For me, the
answer is quite simple: instruction attempts to teach something that is not part of
that competence, namely rules and in the case of languages like Spanish and
Russian, also verbal and nominal paradigms. In most instructed SLA research,
the central idea is that learners internalize or learn some kind of rule/paradigm
from exposure. Researchers either state this explicitly or implicitly. For example,
Hulstijn (2005) says, Explicit learning is input processing with the conscious intention to find out whether the input information contains regularities and, if so,
to work out the concepts and rules with which these regularities can be captured

 Bill VanPatten

(p. 131, emphasis added). Robinson (1995) researched training on what he called
easy and hard rules (p. 303) and based his target structures on pedagogical rules.
Other researchers are less direct about what is learned, referring to knowledge or
structures. However, a careful reading suggests they are interested in rules in the
traditional sense. For example, R. Elliss (2005) study on testing explicit and implicit knowledge is clearly about rules that are the focus of English language
teaching as exemplified in his Table 3 (e.g., third person -s agreement, question
tags, yes/no questions, use of modals with bare verbs). Many more examples
abound in the literature.
But what if acquisition is not the internalization of rules? In VanPatten (2010,
2011, 2013a & b), and VanPatten and Rothman (2013), I have argued just this
point. Central to the thesis is that the development of formal properties of a grammar involves (1) input (contextualized language that learners hear), (2) internal
mechanisms that organize and/or constrain language (e.g., Universal Grammar,
general learning architecture), and (3) processors that mediate between input and
the internal mechanisms. Under this scenario, learners dont acquire rules from
the input; instead they are processors of linguistic data. What kind of data do they
process? The answer is morpho-phonological units such as words (including free
standing morphemes). These in turn receive coding from the inventory of features
provided by Universal Grammar (e.g., N, V, Adj, Tense, Case, Gender) once they
are tagged for meaning. These morpho-phonological units are stored in the lexicon and enter into a complex relationship with the computational component
known as syntax (including interfaces, as suggested earlier in this chapter). The
result of this relationship yields what we call language or better yet, sentences
and utterances.9 We can illustrate with a concrete example: polar (yes/no) questions in English.
In pedagogical accounts, the rule for yes/no questions in English is to insert
auxiliary do at the beginning of the sentence, making sure it is inflected correctly
for person-number and/or tense. However, what learners process and pick up
from the input is not this rule but the following morpho-phonological units along
with their underlying features (Im excluding entire sentences for ease of illustration as well as limiting modals to just one):

9. One reviewer repeatedly queried why I do not address other perspectives on language, such
as systemic-functional linguistics, or why I do not consider that instruction just might be bad in
some cases. I believe that the answer to the second question is evident in the very premise of this
chapter: no formal instruction as defined here can affect underlying representation. As for the
first question, scholars are free to choose perspectives from which to work. Mine happen to be
generative linguistics and language processing. Other scholars are free to address these issues
from their perspectives and to demonstrate similar or counter-arguments.

Chapter 5. The limits of instruction

do: <+V>, <N>, <+T>, <+present>, <past>

does: <+V>, <N>, <+T>, <+present>, <past>, <+3rd person>, <+sing>
did: <+V>, <N>, <+T>, <present>, <+past>
will: <+V>, <N>, <+T>, <present>, <past>
drink: <drink>, <+V>, <N>,
drinks: <drink>, <+V>, <N>, <+present>, <past>, <3rd person>, <sing>
These units are stored in the lexicon and are inserted when a yes/no question is
required. Some auxiliary or modal is inserted into the Tense Phrase (TP) of English because English is a language that obligatorily encodes Tense grammatically.
At the same time, the English CP contains a feature called Q that needs to be satisfied, and so the auxiliary/modal rises to the CP to satisfy the feature needs of
Comp. What is more, lexical verbs such as drink, eat, sleep, and die do not contain
T features, which is why they never move out of the VP in English (whereas they
do in a language like Spanish). Thus, the learner of Spanish processes and internalizes morpho-phonological units such as these:
tomo: <drink>, <+V>, <N>, <+T>, <+present>, <past>, <1st person>, <sing>
tomas: <drink>, <+V>, <N>, <+T>, <+present>, <past>, <2nd person>, <sing>
toman: <drink>, <+V>, <N>, <+T>, <+present>, <past>, <3rd person>, <pl>
It is this complex interplay of units with syntactic computation that yields Does Bill
drink vodka?, while simultaneously ruling out what would be a perfectly fine
sentence in Spanish: *Drinks Bill vodka? These syntactic reflexes are captured in
Figure 1 (simplified for illustration here).
CP
C

CP
TP

does Spec
Bill

C
T

T
tr

toma Spec
VP

Spec
tr

TP

Bill
V

T
T
tr

DP

drink

vodka

VP
Spec
tr

V
V

DP

tr

vodka

Figure 1. Differences in underlying movement for yes/no questions in English and Spanish

 Bill VanPatten

Under this scenario, acquisition becomes an interplay between input (morpho

-phonological units in the speech stream), internal learning mechanisms (in a
generative scenario, this would be UG and perhaps some general learning architecture that aids in the mapping of meaning), and a processing mechanism (which
mediates between input and the internal learning mechanisms). What learners
internalize from the environment, then, are pieces of a speech stream that subsequently get encoded with meaning and features. Rules, if they exist at all, are not
internalized but evolve over time.10
We can contrast this scenario with what typically happens during instruction consisting of the teaching and practicing of rules. Instruction attempts to
instill a product (in terms of a pedagogical rule) in the learners mind/brain
that doesnt match what will wind up in the competence. Here I come back to
processing instruction. At least one reviewer of this chapter queried something
like But doesnt PI purport to alter learner knowledge? The answer is, quite
simply, no. Processing instruction is quite unlike any other pedagogical intervention in its underlying assumptions, its execution, and its expectations. First,
underlying processing instruction is a model of input processing related to how
learners connect meaning and form (both morpho-lexical and sentential) and
how that processing relates to a mental representation. In short, processing instruction does not assume that learners internalize rules from the input or that
rules are learned from the input. Instead, learners internalize morpho-phonological units along with their meanings and underlying features (see above on
the discussion of question formation with do).11 Second, processing instruction
does not engage learners in practicing rules, practicing forms, or any other typical instructional behaviors. Instead it uses structured input activities to promote
correct processing of meaning vis vis formal features. Third, processing instruction does not expect to alter learner knowledge, and in research on processing instruction assessment focuses largely on interpretation of sentences and the
linking of form and meaning. For this reason, one does not see grammaticality
judgment tests, fill-in-the-blank, or any other knowledge-related tests in processing instruction research.

10. Principles such as the OPC described earlier do not evolve; they are there from the beginning and are relevant or not relevant depending on the language being learned. Thus, principles
are not acquired, offering one more piece of evidence that there are aspects of language that are
not explicitly taught or learned.
11. It is probably worth pointing out that input processing as defined here and elsewhere is not
the same as noticing. There is a tendency on the part of some researchers/readers to confuse the
two, which may cause those researchers/readers to think PI is informed by noticing. It is not. For
more detail, see VanPatten (2004, forthcoming).

Chapter 5. The limits of instruction

One more yes, but...

More than one reader is likely asking him or herself What about Norris and
Ortega (2000)? What about Spada and Tomita (2010)? These publications represent important meta-analyses on studies on the effects of form-focused instruction. The typical conclusion that readers draw from these publications is that
instruction makes a difference, and also that so-called explicit instruction is better than implicit instruction (but not that implicit instruction doesnt work).
These conclusions are certainly valid for the research reviewed in the meta-analyses as well as the way in which the authors conducted their analyses, so I hope the
following comments are not misconstrued as a commentary on these two publications. My comments instead are about how we interpret the conclusions and what
the analyses might really mean.
First and foremost, as pointed out by others (including Norris and Ortega
themselves), there is a bias toward explicit learning and explicit testing in instructed SLA (e.g., Doughty, 2003; Truscott, 2004). That is, researchers in this sub-discipline tend to use assessment measures that favor explicit learning. Of particular
importance, these measures favor explicit rule learning. These kinds of assessments stand in stark contrast to, say, the more implicit measures of sensitivity to
violations using self-paced reading found in Smith and VanPatten (in press) and
VanPatten, Keating, and Leeser (2012) to note just two recent examples.
Another important consideration in interpreting the conclusions of research
on instructed SLA is that the effects of instruction are not shown to be durable, if
we consider durability to be longer than several weeks or a month. In VanPatten
and Fernndez (2004), we could only find four studies that spoke to research on
true long-term effects: namely, research that returned to learners nine months to
one year later to see if the effects of instruction held. In no study were such effects
found; however, in our study on processing instruction, such effects were found. In
instructed SLA research, typical post-treatment measures occur immediately to
several weeks later. When effects are still found several weeks later, some cite the
research as showing durable effects. However, given what we report in VanPatten
and Fernndez, it is overly generous to conclude that instructional effects are durable under such limited time spans.
A third and final consideration on the conclusion that instruction makes a
difference and that explicit instruction is better than implicit is that no research to
date has been able to circumvent or otherwise overcome staged development.
That is, the stages found for the acquisition of negation, wh-questions, gender,
copular verbs in Spanish, and other structures are found in and out of classrooms,
with and without instruction. In short, staged development asserts itself in spite of
instruction. So just what does instruction affect and how does it affect it? It is not

 Bill VanPatten

at all clear to me that such studies demonstrate any change in underlying mental
representation of language caused by the instruction itself.
These three considerations lead me to argue that the claims for the benefits
of instruction on the formal properties of language are overstated (see the arguments also in Doughty, 2003; Kessler, Liebner, & Mansouri, 2011; Truscott,
2004; Truscott & Sharwood Smith, 2011). A more reasonable (i.e., more accurate) conclusion would be something like this: research on instructed SLA suggests that form-focused instruction benefits the explicit learning of pedagogical
rules as measured by tests that bias for explicit learning. Here I return to a comment I made in the introductory section of this chapter, namely that this conclusion is not entirely new. Indeed, Krashen made similar comments thirty years
ago (Krashen, 1982), and similar conclusions surface in various guises as attested in some of the citations in this chapter. If I have added something here, it
is one explanation of why formal instruction doesnt make much of a difference,
by focusing on the what of both language and instruction, and underlying assumptions about language.
Returning to Selinkers original claim, he was not (and I am not) equating
pedagogical rules with underlying competence. He was (and I am) using the concept of linguistic knowledge and acquisition quite differently from what is
generally used in instructed SLA. This leads me to a final point before concluding
this chapter.
At the symposium from which the present volume receives its motivation, I
was taken to task by another presenter who said I was sweeping form-focused research under the rug, and that, from her perspective, she did not care about mental
representation (or underlying competence). Aside from the fact that the research
on form-focused instruction speaks for itself, I find it odd that any L2 researcher
would claim not to care about underlying competence. For me, this is akin to a
physicist saying he or she does not care about the structure of atoms. I understand
L2 research is approached from a variety of perspectives (see the discussion in
Rothman & VanPatten, 2013, for example, as well as VanPatten & Williams, 2007),
but it seems to me that the sub-area of focus-on-form research has not grappled in
any real way with the nature of language. The result has been conflicting accounts
on the effects of instruction, some of which is due to methodological issues, but
some of which is also due to how researchers operationalize target structures.
Given the research described in this chapter and the way in which it points toward
an independence between acquisition and explicit learning/formal instruction, it
seems to me that the nature of language and, consequently, the nature of mental
representation is and should be at the heart of a lot of what we do. In fact, one
could argue that statements such as I dont care about mental representation
sweep the nature of language under the rug.

Chapter 5. The limits of instruction

Conclusion
To summarize my main points, they are these:
Selinker was correct that underlying competence is not derived from explicit
instruction/learning. As argued here, all underlying knowledge is the result of
complex interplays between input, language specific mechanisms, and input
processing mechanisms.
My argument is that Selinkers claim holds true for all learners and all stages of
development, not just those who have reached native-like or super-advanced
levels of knowledge and proficiency.
In short, the effects of instruction, particularly form-focused instruction, are
marginally related (at best) to how an internal grammar grows over time. In case
the reader has forgotten or skipped over parts of this chapter, this conclusion
probably seems odd given my work on processing instruction. The long-time research agenda on processing instruction that began with the initial publication of
VanPatten and Cadierno (1993) and my affiliation with that agenda would suggest that I would be a strong advocate of focus on form given the robust findings
of processing instruction research over the last two decades. But as discussed
earlier, at no point have I ever claimed that processing instruction is focused on
rule learning or rule internalization. To be sure, I have couched processing instruction in the context of altering processing behaviors and have not specified
anything about what learners actually internalize. In our recent research, we
are clarifying this issue, especially given that we are showing no link between
aptitude and the outcomes of processing instruction (e.g., VanPatten, 2013;
VanPatten et al., 2013), which stands in contrast to other research on instructed
SLA and aptitude (e.g., de Graaff, 1997; Robinson, 1995; Skehan, 2012). In this
research, I argue that aptitude is not a major factor in processing instruction because unlike other studies, processing instruction does not center on rule learning and rule testing. In short, for me it is perfectly congruent within a generative
model of acquisition to advocate for something like processing instruction while
rejecting the notion of rule learning and instructed SLA that purports to induce
rule learning.
I conclude by stating the obvious, namely that we have made much headway
in empirical research and theory since Selinkers 1972 publication. The amount of
research on SLA is, at times, overwhelming. But we must always be cognizant of
that which preceded us and not lose sight of our history. In the case of instructed
SLA, Selinker made a strong claim about the limits of instruction forty years ago.
I do not believe that we have proven him wrong on that account.

 Bill VanPatten

Acknowledgements
I would like to thank Megan Smith, Elaine Tarone, ZhaoHong Han, and the anonymous reviewers for comments on the first draft of this chapter. I would also like
to thank the students and organizers of the October 2012 symposium on Interlanguage: 40 Years Later for inviting me to talk and share my ideas. Finally, I offer
my thanks to Larry Selinker for his pioneering work in establishing second language acquisition as a contemporary research field.
References
Alonso-Ovalle, L., Fernndez-Solera, S., Frazier, L., & Clifton Jr., C. (2002). Null vs. overt pronouns and the topic-focus articulation in Spanish. Rivista Di Linguistica, 14(2), 151169.
Atkinson, D. (Ed.) (2011). Alternative Approaches to Second Language Acquisition. New York,
NY: Routledge.
Block, D. (2003). The Social Turn in Second Language Acquisition. Washington, DC: Georgetown
University Press.
Boeckx, C. (2011). Approaching parameters from below. In A. Di Sciullo & C. Boeckx (Eds.),
The Biolinguistic Enterprise: New Perspectives on the Evolution and Nature of the Human
Language Faculty (pp. 205221). Oxford: OUP.
de Graaff, R. (1997). Differential Effects of Explicit Instruction on Second Language Acquisition.
The Hague: HAG.
Doughty, C. (2003). Instructed SLA: Constraints, compensation, and enhancement. In C.
Doughty & M.H. Long (Eds.), The Handbook of Second Language Acquisition (pp. 256310).
Oxford: Blackwell.
Ellis, R. (2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies in Second Language Acquisition, 27, 141172.
Ellis, R. (2008). The Study of Second Language Acquisition. Oxford: OUP.
Harley, H. & Noyer, R. (1999). Distributed morphology. Glot International, 4(4), 39.
Hulstijn, J. (2005). Theoretical and empirical issues in the study of implicit and explicit secondlanguage learning. Studies in Second Language Acquisition, 27, 129140.
Kanno, K. (1998). Consistency and variation in second language acquisition. Second Language
Research, 14, 376388.
Keating, G.D., VanPatten, B., & Jegerski, J. (2011.) Who was walking on the beach? Anaphora
resolution in Spanish heritage speakers and adult second language learners. Studies in
Second Language Acquisition, 33, 193221.
Kessler, J., Liebner, M. & Mansouri, F. (2011.) Teaching. In M. Pienemann & J. Kessler (Eds.),
Studying Processability Theory (pp. 149156). Amsterdam: John Benjamins.
Krashen, S. D. (1982). Principles and Practice in Second Language Acquisition. New York, NY:
Pergamon Press.
Lightbown, P.M., & Spada, N. (2013). How Languages Are Learned. Oxford: OUP.
Montalbetti, M. (1984). After Binding: On the Interpretation of Pronouns. Unpublished PhD
dissertation, MIT.

Chapter 5. The limits of instruction

Norris, J.M., & Ortega, L. (2000). Effectiveness of L2 instruction: A research synthesis and
quantitative meta-analysis. Language Learning, 50, 417528.
Prez-Leroux, A.T., & Glass, W. (1999). Null anaphora in Spanish second language acquisition:
Probabilistic versus generative approaches. Second Language Research, 15, 220249.
Robinson, P. (1995). Aptitude, awareness and the fundamental similarity of implicit and explicit second language learning. In R. Schmidt (Ed.), Attention and awareness in foreign
language learning (pp. 303358). Honolulu, HI: University of Hawaii at Manoa.
Rothman, J., & VanPatten, B. (2013). On multiplicity and mutual exclusivity: The case for different SLA theories. In M.P. Garca-Mayo, M.J. Gutirrez-Mangado, & M. Martnez Adrin
(Eds.), Contemporary Approaches to Second Language Acquisition (pp. 243256).
Amsterdam: John Benjamins.
Schwartz, B. (1998.) The second language instinct. Lingua, 106, 133160.
Selinker, L. (1972.) Interlanguage. International Review of Applied Linguistics, 10, 209230.
Skehan, P. (2012). Language aptitude. In S.M. Gass & A. Mackey (Eds.), The Routledge Handbook
of Second Language Acquisition (pp. 381395). New York, NY: Routledge.
Smith, M. C. & VanPatten, B. (2013). Instructed SLA as parameter setting: Evidence from earlieststage learners of Japanese as L2. In C. Laval & M.J. Arche (Eds.), The Grammar Dimension in
Instructed Second Language Acquisition: Theory, Research, and Practice (pp. 127146). London:
Bloomsbury Academic.
Spada, N., & Tomita, Y. (2010). Interactions between type of instruction and type of language
feature: A meta-analysis. Language Learning, 60, 263308.
Sproat, R. (1998.) Morphology as component of module: Mapping principle approaches. In A.
Spencer & A.M. Zwicky (Eds.), The Handbook of Morphology (pp. 335348). Oxford:
Blackwell.
Toribio, A.J. (2000). Setting parametric limits on dialectal variation in Spanish. Lingua, 10,
315341.
Truscott, J., & Sharwood Smith, M. (2011). Input, intake, and consciousness: The quest for a
theoretical foundation. Studies in Second Language Acquisition, 33, 497528.
Truscott, J. (2004.) The effectiveness of grammar instruction: Analysis of a meta-analysis. English
Language & Teaching, 28, 1729.
VanPatten, B. (2004). Input processing in SLA. In B.VanPatten (Ed.), Processing Instruction: Theory, Research, and Commentary (pp. 131). Mahwah, NJ: Lawrence Erlbaum Associates.
VanPatten, B. (2010). The two faces of SLA: Mental representation and skill. International
Journal of English Language Studies 10, 118.
VanPatten, B. (2011). Stubborn syntax: How it resists explicit teaching and learning. In C. Sanz
& R. Leow (Eds.), Implicit and Explicit Language Learning: Conditions, Processing, and
Knowledge (pp. 921). Washington, DC: Georgetown University Press.
VanPatten, B. (2013a). Mental representation and skill in instructed SLA. In J. Schwieter (Ed.),
Innovations in SLA, Bilingualism, and Cognition: Research and Practice. Amsterdam: John
Benjamins.
VanPatten, B. (2013b) Aptitude as grammatical sensitivity: Recent research on processing instruction. In C. Sanz & B. Lado (Eds.), Individual Differences, L2 Development and Language Program Administration: From Theory to Application (pp. 1331). Washington DC:
Georgetown University Press.
VanPatten, B. (forthcoming). Input processing by novice learners: Issues in the nature of processing and in research methods. In Z-H. Han & R. Rast (Eds.), First Exposure to a Second
Language: Learners Initial Input Processing. Cambridge: CUP.

 Bill VanPatten
VanPatten, B., & Cadierno, T. (1993). Explicit instruction and input processing. Studies in Second Language Acquisition, 15, 225243.
VanPatten, B., & Fernndez, C. (2004). The long-term effects of processing instruction. In
B.VanPatten (Ed.), Processing Instruction: Theory, Research, and Commentary (pp. 273289).
Mahwah, NJ: Lawrence Erlbaum Associates.
VanPatten, B. & Rothman, J (2013). Against rules. In A. Benati, C. Laval, & M. J. Arche (Eds.),
The Grammar Dimension Instructed Second Language Learning (pp. 1535). London:
Bloomsbury.
VanPatten, B., Borst, S., Collopy, E., Qualin, A., & Price, J. (2013). Explicit information, grammatical sensitivity, and the First-noun Principle: A cross-linguistic study in processing instruction. The Modern Language Journal, 97, 504525.
VanPatten B, Keating G, & Leeser M (2012). Missing verbal inflections as a representational problem: Evidence from self-paced reading. Linguistic Approaches to Bilingualism, 2(2): 109140.
VanPatten, B., & Williams, J. (Eds.). (2007). Theories in second language acquisition: An introduction. Mahwah, NJ: Lawrence Erlbaum.
White, L. (2003). Second Language Acquisition and Universal Grammar. Cambridge: CUP.

chapter 6

Documenting interlanguage development

Kathleen Bardovi-Harlig

Indiana University at Bloomington

This chapter illustrates Selinkers (1972) claim that interlanguage is a linguistic
system in its own right by examining research in three different areas of L2
development: interlanguage temporality, L2 pragmatics, and conventional
expressions. The chapter begins with a review of functional approaches to
interlanguage analysis. It then reviews a longitudinal study of interlanguage
temporality illustrating the development of form-meaning mappings as learners
acquire temporal expression in L2 English. The importance of interlanguage
development in the pragmatics of L2 learners is emphasized in the next section
which demonstrates that the interpretation of learner forms in conversation
is dependent on the inventory of available linguistic devices. The chapter
closes by considering the role of interlanguage grammar in the production of
formulaic expressions.

Introduction
This chapter illustrates Selinkers (1972) claim that interlanguage is a linguistic
system in its own right by examining the findings of studies in three different
areas of L2 development. The research that I will consider follows the Interlanguage principle of observing learners as they are engaged in meaningful performance in a second language (Selinker, 1972, p. 210). I adopt a functional
approach to interlanguage analysis that promotes the investigation of formmeaning (and/or form-function) mappings. Functional approaches to second
language acquisition investigate such mappings in interlanguage and especially
how these change over time in the developing interlanguage system (BardoviHarlig, 2000; Dietrich, Klein, & Noyau, 1995). Functionalist approaches to language hold that language is primarily used for communication and does not
exist without language users. Because of its focus on communication, a functionalist perspective is quite compatible with the spirit of interlanguage analysis
laid out by Selinker (1972).

 Kathleen Bardovi-Harlig

The following sections demonstrate how examining learners communicative

production reveals interlanguage systems, many stages of which seem to reflect
what Selinker (1972, p. 212) called attempted learning enroute to successful
learning. The systematicity of interlanguage is illustrated by data from two longitudinal studies and one large cross-sectional study that shed light on how learners
employ emergent interlanguage tense-aspect morphology to establish temporal
contrasts; how access to interlanguage contrasts is necessary for a hearer or analyst
to determine pragmatic value; and how interlanguage grammar is visible in learner attempts at conventional expressions. The evidence in the following sections is
based on my published work on the L2 acquisition of temporality, pragmatics, and
conventional expressions.1 The learners in the studies reported here were instructed ESL learners who lived in the host environment.
Functional approaches to interlanguage analysis
Within a functional approach to second language acquisition, there are two main
paths to tracking development. One way is to identify a form such as a morpheme
or a syntactic construction, and then study the meanings associated with it over
time. For example, one could set up a longitudinal study that tracks the present
perfect (a tense-aspect form) and the meanings that learners associate with it over
time (e.g., Bardovi-Harlig, 1997). The second way is to identify a concept (or a
meaning) and learn what forms are associated with it. Not surprisingly, this is
called the concept-oriented or meaning-oriented approach (Bardovi-Harlig, 2000;
Klein, 1995; von Stutterheim & Klein, 1987). A basic tenet of the concept-oriented
approach to second language acquisition is that adult learners of second or foreign
languages have access to the full range of semantic concepts from their previous
linguistic and cognitive experience. Von Stutterheim and Klein argue that a second language learner, in contrast to a child learning his first language, does not
have to acquire the underlying concepts. What he has to acquire is a specific way
and a specific means of expressing them (1987, p. 194).
Determining how to approach the study of form and meaning is similar to
determining which of two different approaches to the photo safari is most revealing for a particular purpose. Taking the form-oriented approach, participants in the safari may decide to photograph a form; in this case, let us imagine
that it is a hippo. Photographers will follow the hippo to various locales to learn
about his environments, or what in interlanguage we might call distributions.
1. The sketches of the studies are necessarily abbreviated here. See the original articles for full
discussion.

Chapter 6. Documenting interlanguage development

On an alternative safari, the photographers concentrate on a concept. This

concept will be the flooded Okavango Delta of the Kalahari Desert, and the challenge will be to see what animals come there. According to the Discovery Channels Planet Earth (http://dsc.discovery.com/tv-shows/planet-earth), we will first
see cranes, then, in turn, black storks, water buffalos, hippos, giraffes, zebras,
lechwe, baboons, and finally predators such as hyenas. Although we could track
the hippo and learn a significant amount about the Okavango, we learn something entirely different when we look at all the animals who come to the Okavango during the flood. Hippos, the form identified as the object of the photo
shoot for the first safari, are only one of many types of inhabitants. By considering the delta as a habitat for many animals, we discover the order in which they
arrive and their relationship to each other.
Returning to interlanguage, the concept-oriented approach begins with a
learners need to express a certain concept, such as time, space, reference, modality, or a meaning within a larger concept (such as past or future time, within the
more general concept of time). From that starting point, it then investigates the
means that a learner uses to express the concept. So for example, futurity can be
expressed by adverbs, modals, will, going to, and lexical futures (i.e., future oriented verbs, such as want to or need to).
Consistent with other functional approaches, the concept-oriented approach
embraces a multi-level analysis, including lexical devices, morphology, syntax,
discourse, and pragmatics, which are the linguistic equivalent of the cranes, black
storks, water buffalos, hippos, giraffes, zebras, lechwe, baboons, and hyenas who
come to the Okavango Delta. In other words, the concept-oriented approach includes all means of expression used by learners. As Long and Sato (1984, p. 271)
note: function to form analysis automatically commits one to multi-level analysis,
since the entire repertoire of devices and strategies used by learners must be examined. Cooreman and Kilborn further observe, there is no formal separation of
the traditionally recognized subcomponents in language, i.e. morphosyntax, semantics, and pragmatics (1991, p. 196).
In addition to documenting the range of linguistic devices that speakers use to
express a particular concept (von Stutterheim & Klein, 1987), a functional analysis
also seeks to understand the relationship among the linguistic devices themselves
and between linguistic devices and contextual information (Klein, 1995). As Klein
observes, from the concept-oriented perspective, a substantial part of language
acquisition is the permanent reorganization of the balance among means of expression. Thus, a functional analysis is concerned with how a concept is expressed,
and how the expression of the concept changes over time. Returning to the photo
safari, we would expect to document the orders in which the animals arrive, and
we would expect to document how the animals co-inhabit the delta.

 Kathleen Bardovi-Harlig

Interlanguage temporality
As an example of interplay among means of expression and the changing balance,
consider a learners expression of past time. The earliest resource that learners have
is their interlocutors turns which may provide a time frame (or scaffolding) on
which a learner can build and universal principles such as chronological order by
which listeners assume that events in narratives are told in the same order in which
they happened. This is called the pragmatic stage (Meisel, 1987). In the next stage,
the lexical stage, learners use temporal and locative adverbials as well as connectives (e.g., and then) to indicate time. Finally, learners may move to the morphological stage in which tense indicates temporal relations. The inventory changes as
new forms are added: first lexical markers, then verbal morphology. Then the balance changes: As tense-aspect use becomes more consistent, the number of adverbials declines (Bardovi-Harlig, 2000).
I will illustrate these principles more concretely with the expression of reverse-chronological order, or what I have called reverse-order reports (BardoviHarlig, 1994). The concept of reverse-order is meaningful in light of the fact
that chronological order is the expected norm in certain contexts, the most notable being the narrative context. Chronological order is the defining characteristic of the narrative: Following Dahl, a narrative is considered to be a text in
which the speaker relates a series of real or fictive events in the order in which
they took place (Dahl, 1984, p. 116). In spite of the strong tendency for narrators to organize their narratives chronologically, they may also leave the sequence of events to predict a coming event, to evaluate an event in the sequence,
or to supply an event that happened previously. When they do, they must indicate their departure from chronological order, as Klein (1986) observes in his
formulation of the principle of natural order (PNO), which he describes as a
discourse principle: Unless marked otherwise, the sequence of events mentioned in an utterance corresponds to their real sequence (p. 127).2 Speakers
seem to follow two maxims:
Principle 1: Tell a story in chronological order
Principle 2: If you cant tell the story in chronological order, warn the listener.
Example (1) reports events in chronological order; Examples (2) and (3) report
events in reverse order. Example (2) utilizes the pluperfect to mark the event;
Example (3) uses adverbials. The first event is labeled [1] and the second [2].

2. This should not be confused with Krashens (1985) Natural Order Hypothesis.

Chapter 6. Documenting interlanguage development

(1) John graduated from high school in 1975. [1] He went to college five years
later. [2]
(2) John entered college in 1980. [2] He had graduated from high school five
years earlier. [1]
(3) I ate my lunch [2] after my wife came back from her shopping [1].

(Leech, 1971, p. 43)
Close examination of interlanguage development shows that learners use a range
of interlanguage contrasts to indicate departures from chronological order (reverse-order reports) that can be found in a longitudinal study of 16 learners of
English over the course of 15 months (Bardovi-Harlig, 1994). All learners were
beginners at the start of the longitudinal study, placed in Level 1 of a six-level program by a 3.5-hour placement exam. The written corpus from this study includes
1,576 written texts. Three complete texts from each month period were sampled
wherever possible. Each text that was selected recounted a past-time event.
Wherever there were multiple past-time texts available, the first three were chosen.
This procedure yielded a total of 430 texts. Journal entries comprised 87% of the
sample, 376 texts, and elicited narratives comprised the remaining 13% of the sample, 54 texts. Film retells accounted for 37 of the elicited texts (9% of the total
sample) and class assignments and essay examinations accounted for the remaining 17 texts (4% of the sample). One hundred and three reverse-order reports were
identified in the data. The reverse-order reports were coded for verbal morphology
and presence of other linguistic markers within the same sentence, namely adverbials, relative clauses, complements, and causal constructions. The date of each
reverse-order report was recorded, and then converted into ordinal months for
reporting purposes.
Examining each learners production separately, the longitudinal data show
that reverse-order reports emerge not at a given time, but in a given sequence.
Reverse-order reports emerge when a learners use of past in past-time contexts
has stabilized at 80% or above (see Bardovi-Harlig, 2000, Table 2.7). The emergence of pluperfect follows some time later (more rapidly for some learners than
others), which leaves a period of time when learners use reverse order reports and
must mark them in ways other than the pluperfect.
The analysis of the longitudinal data show that learners follow the principle of
natural order, and overtly signal nonchronological sequences; 100 (94.2%) of the
103 reverse-order reports showed an explicit marker of reverse order, whereas
only three (or 5.8%) did not, using what Meisel (1987) called implicit marking in
the pragmatic stage as in (4) which established event order through the implicit
contrast of the lexical items breakfast, lunch, and dinner.

 Kathleen Bardovi-Harlig

(4) I told her Im very hungry Can you eat dinner? She said to me Yes. [2]
She didnt eat breakfast, lunch [1] so also she was hungry. After dinner, we
took a walk Bloomington around.
[Noriko, Month 7.5]

Failure to overtly signal a reverse-order report may be a communicatively risky

strategy since the learner does not alert the reader/listener to the departure from
chronological order. As a result, learners do a pretty good job of keeping their listeners or readers informed about the time sequence even when verbal morphology
is not involved.
The explicitly marked reverse-order reports exhibited a variety of linguistic
devices: morphological contrast (tense-aspect usage), adverbials (single and dual),
and syntactic devices including causal constructions (especially the use of because), complementation (especially reported speech or thought), and relative
clauses (Table 1).
Of the 103 reverse-order reports, 63 employed a contrast in verbal morphology and 40 did not (37 of which marked the reverse-order report explicitly
and 3 did not). In the examples with no morphological contrasts, temporal reference is generally carried by lexical devices and syntax (Table 1). Nearly half of
the reverse-order reports with no morphological contrast are marked by time
adverbials; just over half of those show a single adverbial, as in (5) and (6) and
21.2% employ two adverbials, as in (7), where the contrast is between Level 2
and before [Level 2]. Recall these data were written; the original learner spelling
is retained.
Table 1. Past-Time Reverse-Order Reports
No Morphological Contrast
40 occurrences (52 markers)

No Marking
Single Adverb
Dual Adverbs
Relative Clause
Because
Complement
Total

3
13
11
7
12
6
52

5.7
25.0
21.2
13.5
23.1
11.5
100

Morphological Contrast
63 occurrences (68 markers)

No Additional Marking
Single Adverb
Dual Adverbs
Relative Clause
Because
Complement
Total

14
23
7
11
7
6
68

20.6
33.8
10.3
16.3
10.3
8.8
100

Note. From Reverse-Order Reports and the Acquisition of Tense: Beyond the Principle of Chronological
Order, by K. Bardovi-Harlig, 1994, Language Learning, 44, p. 253. Copyright 2006 by Wiley and Sons. Used
with permission.

Chapter 6. Documenting interlanguage development

(5) I went back to my apartment [2] after I finished washing [1].


[Hamad, Month 9.5]
(6) My sister played piano very well. Before she played [2], we were very nervous [1].
[Ji-An, Month 6.5]
(7) In level two I studied many new things for me [2], I didnt study befor in
the another school [1].
[Zayed, Month 3.5]
Causal constructions signaled by because form the next most common group of
reverse-order reports, as illustrated in (8) and (9).

(8) Yesterday was a pusy [busy] day because I had to go to Indiana bell [2]
because I didnt biad [paid] my bell [1] so I did. [Hamad, Month 11.0]
(9) I awake with a sore throat and temperature. Therefore I bought medicine
and took it. But I didnt become to feel better [2], because I made big mistake [1].
[Kazuhiro, Month 7.5]

Relative clauses and complements form 25.0% of the reverse-order reports without morphological contrasts. A reverse-order report with a relative clause is found
in (10) and with a complement in (11) and (12).
(10) Then the bolice [police] but [put] the girl [2] which stole the bread [1] on
the lory with Charlie.
[Zayed, Month 5.5]
(11) One of the women is cooking. He also stole foods from women, Then
Some people were playing magic show. The thief was watching magic rope.
And then, the women found [2] that thief man stole her food [1].

[Satoru, Month 9.0]
Complements often, though not always, report speech or thought, as in (12).
(12) He thought [2] that I said Coming [1].

[Noriko, Month 6.0]

Some reverse-order reports employ two markers (beyond dual adverbials) and include examples such as (13) in which the learner used dual adverbials and a relative clause as markers.
(13) at the moumint I was so happy because I saw my old teachers [2] who
tought me befor [1]
[Hamad, Month 5.5]
Of the 103 reverse-order reports in the group corpus, 63 showed a contrast in verbal
morphology. Two main points are worthy of note: The use of morphological contrasts does not presuppose targetlike contrasts and the presence of a contrast in tenseaspect morphology does not completely obviate the need for additional overt signals
of departures from chronological order. Nearly 80% of the reverse-order reports that
exhibit morphological contrast also employ other markers. However, there are nearly

 Kathleen Bardovi-Harlig

four times as many reverse-order reports with contrasts in tense-aspect morphology

that do not utilize lexical markers (20.6%) as reverse-order reports with no
morphological contrast (5.6%, Table 1). In addition, fewer reverse-order reports with
a tense-aspect contrast employ two markers (only 5 out of 63) than those without a
tense-aspect contrast (12 out of 40). Even once tense contrasts emerge, time adverbials continue to play an important role, with 44.1% of the markers falling into the adverbial category. There appears to be a shift from almost equal use of single and dual
adverbials in the reverse-order reports with no contrast in verbal morphology to a
preference for a single adverbial in the presence of a morphological contrast.
The findings show that reverse-order reports are indeed marked as Klein predicted. Reverse-order reports are marked by a variety of devices; fewer lexical and
syntactic devices are used when specialized verbal morphology is used; and, as
shown by the longitudinal data (see Bardovi-Harlig, 2000, Table 2.7), reverse-order
reports seem to emerge when expression of the past has stabilized, at the rate of at
least 80% use in past-time contexts.
Next let us consider the morphological contrasts themselves. Learner reverseorder reports employed a contrast between past and pluperfect in 66.7% of the cases
showing morphological contrast. The remaining reverse-order reports are divided
among other contrasts (Table 2). This suggests that learners recognize reverse-order
reports as an environment for a morphological contrast, and that the interlanguage
system may set up contrasts first, before determining what contrast fits best with the
emerging system. As the longitudinal data show (see Table 2.7, Bardovi-Harlig,
2000), reverse-order reports generally appear before the pluperfect does. The contrast between the simple past and pluperfect is illustrated in (14) and (15).
(14) John and I went to her building [2]. She had invited her friends [1].

[Eduardo, Month 1.0]
(15) There were many different shops bagages [2] maybe she had gone shopping many times and different shops [1].
[Hiromi, Month 6.5]
Table 2. Type of Morphological Contrasts in Past-Time Reverse-Order Reports
Morphological Contrast

Past Past Perfect

Past Present Perfect
Past Past Progressive
Present/Base Past
Misc

42
7
3
7
4
63

66.7
11.1
4.8
11.1
6.3
100

Note. From Reverse-Order Reports and the Acquisition of Tense: Beyond the Principle of Chronological
Order, by K. Bardovi-Harlig, 1994, Language Learning, 44, p. 253. Copyright 2006 by Wiley and Sons. Used
with permission.

Chapter 6. Documenting interlanguage development

The remaining 33.3% of the reverse-order reports with morphological contrasts

are divided among other contrasts, such as past and present perfect, as in (16), and
past and past progressive, as in (17).
(16) We went over to Steak and Shake for lunch. Then he asked me [2] if I have
been in spring Mill Park [1].
[Guillermo, Month 8.0]
(17) I found my conversation partner [2] and she was so happy because she was
trying to find me from three weeks [1].
[Saleh, Month 10.0]
Another common contrast is between a base form and the simple past. These cases are particularly interesting because the learners generally begin their narratives
in the past, then create a contrast by switching to the base form to relate the more
recent event (event 2) and using the simple past for the earlier event (event 1) as in
(18) and (19).
(18) After that the policeman caught him. . . In the car, there are many people
[2] who did bad things [1].
[Noriko, Month 11.0]
(19) She met charlie the nice man who was trying to helpe her when he said
she didnt steel the bread I did that, No body belive that because [2] the
lady saw the garil [1].
[Saleh, Month 13.0]
This contrast between the base and past illustrates the genius of interlanguage on
two levels. The principle of natural order requires the marking of events that do
not follow chronological order, and the interlanguage is able to meet that principle.
Moreover, the interlanguage marks the events in the expected direction: The more
distant event carries the tense marker whereas the more recent event does not.
This is the type of pattern that would be lost if the analysis were focused on either
correct form or errors. When approached from an interlanguage perspective, the
expression of temporality provides a rich view of many developmental stages in
second language acquisition.
Interlanguage development, choice, and interpretation in L2 pragmatics
Research in L2 pragmatics has more often investigated learners development of
rules of use (sociopragmatics) than development of language used in pragmatics
(pragmalinguistics). Given the focus on the relation of language to social context,
it is hard to imagine studying pragmatics without sociopragmatics. It should be
equally hard to imagine studying pragmatics without studying the linguistic resources for realizing the social rules (the pragmalinguistics). Since the publication
of The Interlanguage of Interlanguage Pragmatics (Bardovi-Harlig, 1999), I have
been particularly interested in how interlanguage development interacts with and

 Kathleen Bardovi-Harlig

underpins L2 pragmatic development. This section explores two types of development in pragmatics. The first concerns the internal structure of a speech act and
shows how examining language used for communication reveals structure even in
the expression of a speech act over time, not generally an environment selected for
the observation of linguistic development. The second concerns the development
of pragmalinguistic resources and how interlanguage development should constrain an analysts interpretations of learner production. This is a different approach from the first one. The second approach shows what an analyst or hearer
might miss in a communicative exchange if the developmental history or inventory were not taken into account, and emphasizes further the importance of the
concept of an interlanguage system for analysis.
The internal structure of speech acts emerges in sequences. I will illustrate this
order in disagreements. Consider first an interlanguage disagreement which exhibits the preferred form of disagreements in American English which positions
an agreement before a disagreement (Pomerantz, 1984) as in (20).
(20) Takako: You know cows smells so bad!
Kristen: Not as bad as pigs though
Takako: Yeah, but sometimes cows like more than pigs

(Bardovi-Harlig & Salsbury, 2004)
The agree-before-disagree structure in disagreements is a relatively late development in interlanguage following a series of steps. Tom Salsbury and I collected
data from 12 ESL learners in monthly interviews for 9 months, recording
both student-student and student-interviewer conversations (Bardovi-Harlig &
Salsbury, 2004; Salsbury & Bardovi-Harlig, 2000, 2001) and we extracted all the
exchanges that exhibited oppositional talk, including disagreements. The longitudinal data show that learners begin with what we call a bare but that is, but alone
that marks the disagreement explicitly as shown in (21)
(21) Samantha: right now you are already studying hard, and I think youre an
excellent student...
Eun Hui: but, ah, IEP course, course and ah, actually university lectures
different, right, so maybe, maybe university, this lecture is, I
cant understand sometimes.
Samantha: I cant either, so youre not alone
[Eun Hui: Month 7]
The next stage is the explicit acknowledgement of the interlocutors position by the
learner as seen in (22) in I know your mean which expresses the agreement in interlanguage form.

Chapter 6. Documenting interlanguage development

(22) Tom: if you dont take any IEP classes, then you have no connection
to the university, cause IU has to accept you first as a student,
do you see what I mean?
Eun Hui: I know your mean, but I dont think so
Tom:
no? ok.
[Eun Hui: Month 10]
In the last stage learners demonstrate the use of the conventional abbreviated
agree-before-disagree marker, yeah but in (20) and (23).
(23) Tom: its a cultural difference, do you think...but I know in Asia, you
can agree or disagree, its more of a written culture...
Eun Hui: yeah, but, in Korea, in Korea culture, during our class, we dont
say many things, but even though I know about that, just we
have to polite attitude during class..., but that is not, not helpful for us, but we have many classmate, may, ah, about 60 student, so
[Eun Hui: Month 11]
Following the emergence of yeah but learners have recourse to all the forms, full
agreement, yeah but, and the bare but. Once learners have a full tool kit, they can
make choices among their linguistic resources.
The concept of choice is one of the basic principles of pragmatics: all speakers
native speakers, nonnative speakers, and learnersmake choices among available
linguistic forms to convey social meanings. There are many examples in the literature of choice between address terms (usted versus tu), request strategies (would
you versus I was wondering if you would...), or an aggravator rather than a mitigator (I just decided that I will take syntax versus I was thinking about taking syntax).
The use of one form in place of another has meaning because there are other possible alternatives.
As interlanguage begins to include alternative forms, learners must expand
their form-meaning-use associations. On one end of the developmental continuum is a range of alternative means of expressions. On the other end of the
continuum is a stage in which a learner has only one way of expressing a concept,
following the one-to-one principle (one form for one concept; Andersen, 1984),
which means that there are no alternatives from which a learner can choose. Because pragmatic value is derived from the choice of available linguistic devices to
signal relationships among speakers, if learners have only one linguistic form
available to them, then the use of a particular form does not signal pragmatic information (that is, it does not have pragmatic value) within the learners linguistic
system. It only reveals the learners level of interlanguage development. Thus, the
study of acquisition of form in pragmatics, including grammar, lexicon, and formulaic language, is the study of the development of alternatives. The study of use

 Kathleen Bardovi-Harlig

in pragmatics must be understood in light of the forms available to the learner at

any given stage of interlanguage development. And although Kasper (2001, p. 51)
observed that Pragmatics is never only form, within an interlanguage approach
to pragmatics, the study of grammatical development is not only about form, it is
about how form develops in contexts, in form-function-use associations, and the
choice among alternatives that new forms present to learners.
To consider a case in which a learner has only one form, but target-language
interlocutors (and eavesdroppers such as analysts) have two, consider the use of
will and be going to. As semantic typologists note, all uses of future are also modal
because no one can know the future. The difference between will and going to is
often said to be one of pragmatics. Although will and be going to are truth functionally equivalent (meaning that if one is true the other is true, and vice versa), be
going to seems to invoke present relevance whereas will is oriented toward the future (Haegeman, 1989).
Learners often use will where native speakers use going to, and this seems to
have a pragmatic impact on the interpretation of their utterances. More specifically, when learners use will in the environment of a native speakers going to, it
appears to native speakers that learners may be rather insistent or making a strong
assertion. However, for a hearer to attribute insistence to a learners use of will in
the absence of other means of expressing the future is to apply a native-speaker
interpretation of the learners system. In the case of negative interpretation this can
be especially unfortunate because learners acquire and use going to later than will;
the difference in native-speaker and learner production is more likely due to the
learners stage of grammatical development than their pragmatics.
To illustrate these claims, I draw on data which come from the same longitudinal study as the data in the preceding section, but in this study all 1576 written
texts plus an additional 175 oral texts (102 conversational interviews and 73 elicited
narratives) were included (Bardovi-Harlig, 2004a, 2004b, 2005). In the examples
below native-speaker uses of going to illustrate both proximity and intentionality:
What are you gonna do this summer? Are you gonna...go to class this summer? (24)
What are you going to do for Christmas break? (25), What are you going to do? (26),
and What are you gonna do during the break? (27). However, the learners respond
using will, a form used to indicate future which emerges substantially earlier than
going to. In fact, of the 16 learners studied, only 8 used more than 10 tokens of
going to spontaneously during the 15 months of the study.
(24) Interviewer: What are you gonna do this summer? Are you gonna...go to
class this summer?
Kazuhiro:
I will get, mm, IEP class.
[L1 Japanese, Month 8.0]

Chapter 6. Documenting interlanguage development

(25) Interviewer: What are you going to do for Christmas Break?

Carlos: I will go back to my country on December 8.

[L1 Spanish, Month 3.0]
(26) Interviewer: Are you planning a vacation during, um...Spring Break?
Guillermo: I will go to Mexico in spring break with a friend of mine
[4 turns]
Interviewer: What are you going to do?
Guillermo: Well, I will go with that, with, that one group from, from the
/cher/ [church] and:: well well go about 35 people, and we
will go to see the...
[L1 Spanish, Month 6.0]
(27) Interviewer: Um, h-uh what are you gonna DO after ENGLISH after [ ]
BREAK,

I mean DURING the break. ((uppercase indicates stress)
Noriko:
Uh, I will go back to Japan.
Interviewer: Are you? Gonna go back to Japan?
Noriko:
Yeah. Just for, Christmas.
Interviewer: Mm hm. What are you gonna do?
Noriko: Mmm, I will meet my friend. I will go, I will, [ ], [ ], uh go
shopping 
[L1 Japanese, Month 8.5]
Examples (24)(27) show that the learners stage of development constrains their
responses since they have acquired just one grammatical form for future reference:
will. From a sociolinguistic and discourse perspective, this means that the learners
cannot accommodate to the speech of their interlocutors because they have not
acquired going to.3 From a psycholinguistic perspective, the interviewers use of
going to does not prime the learners use of going to for the same reason: going to
has not yet emerged in the learners interlanguage. In contrast, the learner in (28)
and (29) is a going to user, producing the reduced form gonna both early and late
in the longitudinal oral data, with will also appearing early and going to following
in written production at Month Three.
(28)

Interviewer:
Hamad:
Interviewer:
Hamad:

What are you going to do?

Were gonna go to, Daytona Beach.
To Daytona? Ah! When are you going?
[ ] spring break, on the first I think.

3. Although it appears that the learners in Examples 2427 may understand going to it is
equally possible that they are responding to their understanding of the time adverbials this summer, for Christmas break, during spring break, and during the break. See the discussion of time
adverbials in processing studies in Bardovi-Harlig (2000, Chapter 2).

 Kathleen Bardovi-Harlig

Interviewer:
Hamad:
Interviewer:
Hamad S:


On the first of March?

Yeah
Uh huh. So, what [I], how long are you gonna stay?
If we enjoy there we gonna stay, one week ten days.
[L1 Arabic, Month 1.5]

(29) Interviewer: Are you gonna come back for Jan=

Hamad:
=yeah sure sure sure sure
Interviewer: I what, what are you gonna do are you gonna, what are you
gonna do during the break?=
Hamad:
=Christmastime? Christmas?
Interviewer: yeah
Hamad:
Oh yeah, Im gonna go back.
Interviewer: You gonna go back.
Hamad S:
Yeah sure
[Month 10.5]
When this learner accommodates to the interviewers use of be going to in Examples (28) and (29) the result is a neutral report of his plans. In Examples (24) (27),
however, the learners use will to answer the interviewers going to, and because of
the apparent contrast, the learner responses seem to native speakers to be much
more determined or insistent about their plans than the situations warrant. However, because the learners use the only productive grammatical marker they have
for the future, this does not represent an actual contrast. Without a choice between
will and going to in the interlanguage grammar, the use of will conveys no more
than the future. As a result, what may sound insistent to the interviewer or outside
observers has neutral pragmatic value within the learners interlanguage.
Eight of the 16 learners did not use going to productively (showing fewer than
10 uses over the course of a year); this means that half of the learners in the corpus
lack the grammatical means to accommodate to the utterances of their interlocutor
if they involve future expression, regardless of how simple future expressions appear
on the surface (Bardovi-Harlig, 2004a). These examples show two things of relevance to our interlanguage inquiry: The importance of referring to the interlanguage
grammatical value in making pragmatic assessments, and the strength of interlanguage constraints on production. Whether a learners repetition of the interlocutors
turns is considered to be priming by the original utterance or speech accommodation by the learner, both are blocked by the learners grammatical development.
Conventional expressions: Interlanguage grammar in so-called formulas
The previous sections considered spontaneous written and oral production
data. This section explores timed oral-production data elicited by computerized

Chapter 6. Documenting interlanguage development

conversation simulations (Bardovi-Harlig, 2009). The conversation simulations

were designed to elicit language used in communication, the central criterion for
inclusion of language samples in this chapter, while at the same time providing
carefully controlled contexts for the use of conventional expressions in oral production. Because the contexts were developed to capitalize on field observations
that native speakers used very specific expressions in some contexts and not others, the task yielded a high density of expressions produced by learners. The density of production of 123 learners responding to the same scenarios is considerably
different from a data set consisting of production of formulaic language that is
incidental and diffuse. Previous studies that have identified interlanguage forms of
formulaic sequences such as take advantages of (Yorio, 1989) and I very appreciate (Eisenstein & Bodman, 1986) had freer tasks that resulted in language samples where formulaic use was individual and quite varied.
Although my original intention in constructing the conversational simulation
task and a corresponding recognition task was the investigation of the acquisition
of conventional expressions as a pragmatic resource, my attention was quickly
drawn to interlanguage forms of conventional expressions such as I just look and
Ill just looking for Im just looking.
Previous discussions of formulas in pragmatics have not often distinguished
formulas that arise in the process of acquisition and that cannot be analyzed by
the learner grammar (acquisitional formulas) from formulas that are the target of
acquisition and learning (social formulas; Bardovi-Harlig, 2006). The former include examples from Schmidts informant Wes Do you have time? and Are you
busy? used at the same time as his grammar generated You has keys? and When
Tim is coming? (Schmidt, 1983). Regarding these types of formulas, Krashen,
Dulay, and Burt (1982, pp. 232233) noted, Routines are whole utterances that
are unusually error-free and show no transitional stages of development or systematic order of acquisition. They are learned as unanalyzed wholes, much as one
learns a single word. The latter group includes social formulas. In contrast to
formulas that arise in early acquisition, intentionally learned, later acquired expressions do not provide evidence for processing claims in pragmatics studies. In
some cases it is even clear that the speakers grammars can or do analyze the
strings. To assume that all multi-word units are resistant to analysis by the interlanguage grammar does not take into account interlanguage data now available
from a large cross-sectional study.
In the cross-sectional study that I conducted (N = 179), an oral conversation
simulation elicited conventional expressions from learners at four levels of English
language study and native-speaker peers and native-speaker teachers. Learners
heard a scenario over computer headphones and read it on the screen. Following
the description of the scene, learners then heard a conversation turn addressed to

 Kathleen Bardovi-Harlig

them and were given seven seconds to respond. The seven seconds were to simulate a turn exchange and were purposely speeded to increase the chances that respondents would use conventional expressions due to reduced planning time. The
scenarios were derived from observation of spontaneous conversation, then piloted twice to verify preferred use of the conventional expressions. Two examples
are provided here.
(30) Shopping, no help
You go to a clothing store and you need to find a new shirt. A salesperson
approaches you. You dont want the salespersons assistance.
(Audio Only): Can I help you?
(next screen, visual only) You say:
[7 seconds to respond orally]
(31) Five minutes late
You made an appointment with your teacher. Unfortunately you arrive
five minutes late for the meeting. Your teacher says,
(Audio Only): Hello. Come on in.
(next screen, visual only) You say:
[7 seconds to respond orally]
Although the study was originally designed to study the use of conventional expressions in specific pragmatic contexts, learner responses to these scenarios raise
the issue of expression building and interlanguage grammar. A common assertion
is that learners learn formulaic language from input; that is, they simply repeat
formulaic sequences or acquire them from sheer frequency. The frequency of the
target expressions used in this study was established by their high use by native
speakers in the same context. In 22 contexts, native speakers used a single expression above 60% of the time. There are in addition claims that learners use such
sequences to derive or motivate grammar. The data show that the learner grammar
is evident even in the learning of conventional expressions and that it is stronger
than the influence of input. It is also clear that learners often know what the target
expression is, even when the string produced by the interlanguage grammar is
distinct from the target structure, as evidenced by the use of the key words in an
expression, but not the target grammar.
The expressions under consideration are short and relatively simple. For example, consider the string Im just looking which co-occurs with No thanks, Im just
looking (Example 30). Forty-four learners out of 122 attempted the expression.
Thirty of those produced the form Im just looking, but roughly one-third of the
learners who attempted the expression produced an interlanguage form with the
key words just and look, but with a range of interlanguage morphosyntax, including I just look, I just looking, Ill just looking, Just Im looking, and the unreduced
form, characteristic of learner language, I am just looking.

Chapter 6. Documenting interlanguage development

The conventional expression Sorry Im late (elicited by the context in Example

31) also yielded a variety of interlanguage attempts, including Im sorry for late,
Im sorry for Im late, Im sorry about late, Im so sorry about my late, Im so sorry to
being so late, Im sorry because I late. In addition to the attempts at the conventional expressions which, when compared to standard English, are not grammatical, learners also produced strings which are grammatical, when compared to
standard English, but which nevertheless do not tune into the conventional expression itself and at the same time reveal the interlanguage grammar. These
include Sorry for being late, Im sorry to be late, Im sorry to come late, and Sorry for
coming late.
Other interlanguage influences are seen when learners use transparent rather
than opaque forms and full forms rather than reduced forms. The learners who
participated in the study preferred the transparent Ill call you later with an adverbial to Ill call you back, with a particle (native speakers did the opposite), and
they produced more instances of You are welcome than Youre welcome and
more tokens of Thank you than Thanks (see Bardovi-Harlig, 2012 for discussion of variation in this data set).
There also appear to be contexts with which native speakers and learners associate a single word but not a specific expression. When faced with an unexpected visitor the night after a party in the apartment, 43% of the NS undergraduates said sorry for/about the mess and 36% of the NS teachers tested said,excuse
the mess. However 80% of the undergraduates and 93% of the teachers produced
the word mess. Only a very small percentage of the learners attempted an expression with mess (13% at the intermediate levels and 28% at the low-advanced
level), and of those who did, only four learners used Sorry about the mess, which
undergraduates also used. A variety of interlanguage forms are found, most involving the nominal and adjectival form of the key word mess/messy, requisite
articles, and optional modifiers; these include the interlanguage strings My apartment is mess, My apartment is so mess, My apartment is a messy, and Sorry for
messy. In addition, learners produce a variety of well-formed sentences including
My apartment is messy, My apartment is so messy, and (?) My apartment is too
messy. Learners produced the adjectival form messy, as in, My apartment is messy
and My apartment is a messy in 10 of the 16 cases (63%), a much higher proportion
than used by NS (8% by teachers and 18% by undergraduates).
Learner attempts at conventional expressions are influenced by both input
(learners know the key words which can only be learned from input) and interlanguage grammar which structures their form. Even in cases of very clear input, in
which native speaker agreement on the preferred conventional expressions is very
high, the strength of the interlanguage grammar shows through, revealing
interlanguage development in the attempts at the expressions. In addition, as the

 Kathleen Bardovi-Harlig

learner production of equivalents to Sorry Im late show, the interlanguage grammar allows learners to overshoot the target as well as to exhibit later learned grammatical but not entirely conventional developmental forms. What remains is
for learners to develop their ability to distinguish conventional-grammatical expressions from nonconventional-grammatical ones, an ability of native speakers
which Pawley and Syder (1983) called nativelike selection. Although nativelike
selection may eventually rule out grammatical but not conventional forms for second language learners, in the meantime the interlanguage forms argue that conventional expressions may be creatively produced by the learner grammar, using
the key words from input structured by the interlanguage grammar at its current
developmental stage.

Conclusion
This chapter has illustrated the results of following Selinkers principle of observing learners as they are engaged in meaningful performance in a second language
by providing three perspectives on interlanguage development. Although the
vantage points are distinct, including the expression of temporality, pragmatics,
and conventional expressions, a close analysis of the developing interlanguage
shows it to exhibit both systematicity and innovation. Interlanguage narrative sequences follow the universal principle of chronological order with innovative
contrasts. Learner attempts at conventional expressions show evidence that key
words may be associated with contexts but that interlanguage grammar determines the form of the strings that link them. Speech act realization also proceeds
from least linguistically marked, most direct forms to postponement of the dispreferred act, as shown by the longitudinal development of disagreements. Finally,
analyses of learner production of future forms show that sociolinguistic processes
such as speech accommodation are constrained by the current stage of grammatical development.
Thick sampling from longitudinal studies allows the tracking of interlanguage
development and the use of the corpus more generally to help interpret specific
points of production. Large cross-sectional studies that simulate oral communication also provide evidence that allows comparison across learners. The documentation of language in use by learners yields a rich portrait of interlanguage. The
documentation of interlanguage in communication is not a fast or easy process,
but the resulting corpora help us to understand the principles that drive language
acquisition and use.

Chapter 6. Documenting interlanguage development

References
Andersen, R.W. (1984). The one to one principle of interlanguage construction. Language
Learning, 34, 7795.
Bardovi-Harlig, K. (1994). Reverse-order reports and the acquisition of tense: Beyond the principle of chronological order. Language Learning, 44, 243282.
Bardovi-Harlig, K. (1997). Another piece of the puzzle: The emergence of the present perfect.
Language Learning, 47, 375422.
Bardovi-Harlig, K. (1999). The interlanguage of interlanguage pragmatics: A research agenda for
acquisitional pragmatics. Language Learning, 49, 677713.
Bardovi-Harlig, K. (2000). Tense and Aspect in Second Language Acquisition: Form, Meaning,
and Use. Oxford: Blackwell.
Bardovi-Harlig, K. (2004a). Monopolizing the future: How the go-future breaks into wills territory and what it tells us about SLA. EuroSLA Yearbook, 4, 177201.
Bardovi-Harlig, K. (2004b). The emergence of grammaticalized future expression in longitudinal production data. In M. Overstreet, S. Rott, B. VanPatten, & J. Williams (Eds.) Form and
Meaning in Second Language Acquisition (pp. 115137). Mahwah, NJ: Lawrence Erlbaum
Associates.
Bardovi-Harlig, K. (2005). The future of desire: Lexical futures and modality in L2 English future
expression. In L. Dekydtspotter & R. Sprouse (Eds.), 7th Generative Approaches to Second
Language Acquisition, (pp. 112). Somerville, MA: Cascadilla Proceedings Project. <http://
www.lingref.com/cpp/gasla/7/paper1151.pdf>
Bardovi-Harlig, K. (2006). On the role of formulas in the acquisition of L2 pragmatics. In
K. Bardovi-Harlig, C. Flix-Brasdefer, & A.S. Omar (Eds.), Pragmatics and Language Learning, (Vol. 11, pp. 128). Honolulu: University of Hawaii, National Foreign Language
Resource Center.
Bardovi-Harlig, K. (2009). Conventional expressions as a pragmalinguistic resource: Recognition and production of conventional expressions in L2 pragmatics. Language Learning, 59,
755795.
Bardovi-Harlig, K. (2012). Pragmatic variation and conventional expressions. In J. C. FlixBrasdefer & D. Koike (Eds.), Pragmatic Variation in First and Second Language Contexts:
Methodological Issues (pp. 141173). Amsterdam: John Benjamins.
Bardovi-Harlig, K., & Salsbury, T. (2004). The organization of turns in the disagreements of L2
learners: A longitudinal perspective. In D. Boxer & A.D. Cohen (Eds.), Studying Speaking to
Inform Second Language Learning (pp. 199227). Clevedon: Multilingual Matters.
Cooreman, A., & Kilborn, K. (1991). Functionalist linguistics: Discourse structure and language
processing in second language acquisition. In T. Huebner & C.A. Ferguson (Eds.), Cross
Currents in Second Language Acquisition and Linguistic Theory (pp. 195224). Amsterdam:
John Benjamins.
Dahl, . (1984). Temporal distance: Remoteness distinctions in tenseaspect systems. In B. Butterworth, B. Comrie, & . Dahl (Eds.), Explanations for Language Universals (pp. 105122).
Berlin: Mouton.
Dietrich, R., Klein, W., & Noyau, C. (1995). The Acquisition of Temporality in a Second Language.
Amsterdam: John Benjamins.
Eisenstein, M., & Bodman, J. W. (1986). I very appreciate: Expressions of gratitude by native
and non-native speakers of American English. Applied Linguistics, 7, 167185.

 Kathleen Bardovi-Harlig
Haegeman, L. (1989). Be Going to and Will: A pragmatic account. Journal of Linguistics, 25,
291317.
Kasper, G. (2001). Classroom research on interlanguage pragmatics. In K. Rose & G. Kasper
(Eds.), Pragmatics in Language Teaching (pp. 3360). Cambridge: CUP.
Klein, W. (1986). Second Language Acquisition. (Rev. ed., Bohuslaw Jankowski, Trans.)
Cambridge: CUP. (Original work published 1984)
Klein, W. (1995). The acquisition of English. In R. Dietrich, W. Klein, & C. Noyau (Eds.), The
Acquisition of Temporality in a Second Language (pp. 3170). Amsterdam: John Benjamins.
Krashen, S. (1985). The Input Hypothesis: Issues and Implications. London: Longman.
Krashen, S., Dulay, H., & Burt, M. (1982). Language Two. Oxford: OUP.
Leech, G.N. (1971). Meaning and the English Verb. Harlow, Essex: Longman.
Long, M., & Sato, C.J. (1984). Methodological issues in interlanguage studies: An interactionist
perspective. In A. Davies, C. Criper, & A.P.R. Howatt (Eds.), Interlanguage (pp. 253279).
Edinburgh: Edinburgh University Press.
Meisel, J.M. (1987). Reference to past events and actions in the development of natural language
acquisition. In C.W. Pfaff (Ed.), First and Second Language Acquisition Processes (pp. 206
224). Cambridge, MA: Newbury House.
Pawley, A., & Syder, F.H. (1983). Two puzzles for linguistic theory: Nativelike selection and
nativelike fluency. In J.C. Richards & R.W. Schmidt (Eds.), Language and Communication
(pp. 191226). London: Longman.
Pomerantz, A. (1984). Agreeing and disagreeing with assessments: Some features of preferred/
dispreferred turn shapes. In J.M. Atkinson & J. Heritage (Eds.), Structures of Social Action:
Studies in Conversation Analysis (pp. 57101). Cambridge: CUP.
Salsbury, T., & Bardovi-Harlig, K. (2001). I know your mean, but I dont think so Disagreements in L2 English. In L. Bouton (Ed.), Pragmatics and Language Learning (Vol. 10)
(pp. 131151). Urbana-Champaign IL: University of Illinois, Division of English as an International Language.
Salsbury, T., & Bardovi-Harlig, K. (2000). Oppositional talk and the acquisition of modality in
L2 English. In B. Swierzbin, F. Morris, M.E. Anderson, C.A. Klee, & E. Tarone (Eds.), Social
and Cognitive Factors in Second Language Acquisition: Selected Proceedings of the 1999
Second Language Research Forum (pp. 5776). Somerville, MA: Cascadilla Press.
Schmidt, R. (1983). Interaction, acculturation and the acquisition of communicative competence. In N. Wolfson & E. Judd (Eds.), Sociolinguistics and Second Language Acquisition
(pp. 137174). Rowley, MA: Newbury House.
Selinker, L. (1972). Interlanguage. IRAL, 10, 20931.
von Stutterheim, C., & Klein, W. (1987). A concept-oriented approach to second language studies. In C.W. Pfaff (Ed.), First and Second Language Acquisition Processes (pp. 191205).
Cambridge, MA: Newbury House.
Yorio, C. (1989). Idiomaticity as an indicator of second language proficiency. In K. Hyltenstam
& L.K. Obler (Eds.), Bilingualism across the Lifespan (pp. 5572). Cambridge: CUP.

chapter 7

Methodological influences of Interlanguage

(1972)
Data then and data now
Susan Gass and Charlene Polio
Michigan State University

This chapter examines the impact of Selinkers claim that certain data were
inappropriate for SLA research: grammaticality judgments and nonsense
syllables. The field has gradually come to understand SLA through multiple data
types, so that the current view is that data are not appropriate or inappropriate
in a vacuum, but rather need to be understood in the context of the research
questions asked. The chapter describes the data used in studies prior to 1972,
and then focuses on grammaticality judgments and nonsense data, including
artificial languages, in studies after 1972. Although both data types are common
in current SLA research, Selinker problematized their use and made us consider
what we can learn from data other than elicited speech.

Introduction
This volume celebrates the 1972 publication of Larry Selinkers Interlanguage
paper. This seminal article was a culmination of many ideas that Selinker had been
developing over the preceding years. It was influential in many ways and problematized numerous areas of second language research, some of which had already
been dealt with in some of Selinkers earlier works as well as earlier works of others
(e.g., Brire, 1968; Crothers & Suppes, 1967; Jakobovits, 1970; Lawler & Selinker,
1971; Nemser, 1971). One such example concerns the relationship between language learning and language teaching, the latter field serving as a springboard for
the former. In the years leading up to the 1972(a) publication (and even beyond),
these two disciplines were synonymous in the minds of many. In fact, Selinker
notes this confusion when he talks about the

 Susan Gass and Charlene Polio

confusion between learning a second language and teaching a second language.

[See also Mackey in Jakobovits, 1970, p. IX]. This confusion applies as well to
almost all discussions on the topic one hears. For example, one might hear the
term psychology of second language teaching and not know whether the speaker
is referring to what the teacher should do, what the learner should do, or both.
This terminological confusion makes one regularly uncertain as to what is being
claimed. (1972a, p. 210)

Selinker (and others) made what was then an important (if not novel at that time)
point that the two needed to be kept separate and in this way set the stage for an
independent discipline now known as second language acquisition (SLA) (see also
Larsen-Freeman, this volume; Tarone, this volume).
Other areas of concern that remain current even today (such as controlled
versus automatic production, and the role of individual differences, which he argues forcefully is the sine qua non of any theory of language learning) were also
hinted at in Selinker (1972a), as well as in some of his earlier works. In his words,
[a] theory of second-language learning that does not provide a central place for
individual differences among learners cannot be considered acceptable (p. 213).
Still other areas that point the way to future research are the role of rule (quote
in original) learning (1972b, p. 293), the need to look at groups of learners versus
single learners in depth, learning in relation to variables of pedagogical presentation (1972b, p. 294), the relationship of input to output (1972b, p. 296), and the
role of memory (1972b, p. 291). Selinker was perspicacious enough to recognize
the deep issues that were of significance to the field and, even though he did not
solve them, he understood that these constituted the shape of the field to come
(see VanPatten, this volume, on how rule learning, pedagogical presentation, and
input relate to interlanguage).
The goal of our chapter takes one specific perspective, namely, that of methodology. In particular, we examine the impact of claims made about appropriate data
for analysis. As Selinker notes (1972a p. 210), one of our greatest difficulties in
establishing a psychology of second language learning which is relevant to the way
people actually learn second languages, has been our inability to identify unambiguously the phenomena we wish to study. This in fact is the point of departure
for this chapter what are the relevant data and how do we get them?
What counts as data?
To address the issue of data, we first take into account a central question about
SLA: What counts as acquisition? As Norris and Ortega (2003) point out, this
question cannot be addressed without considering the crucial underlying issue of

Chapter 7. Methodological influences of Interlanguage (1972)

the theoretical approach of researchers. SLA is not monolithic and there is lack of
agreement about what constitutes acquisition. Nonetheless, whatever perspective
we take, measurement is essential in that it feeds into interpretations about linguistic systems, changes (or not) in learners systems, and factors that contribute
to or hinder development. Selinker made a similar observation saying that we
need theoretical preliminaries because without them it is virtually impossible to
decide what data are relevant to a psycholinguistic theory of second-language
learning (1972a p. 209).
One could argue that allowable data in the 1970s cannot be meaningfully
compared to allowable data in todays research environment because the ultimate
goal of SLA in the 1970s, namely, to account for how non-primary languages are
learned, are different from todays goals. However, even though the discipline of
SLA is far more sophisticated in terms of subareas investigated and methods used
to investigate those subareas, the goals have not changed substantially. Selinker
states 1972 goals clearly: we need to understand psycholinguistic structures and
processes underlying attempted meaningful performance in a second language
(1972a, p. 210). This is similar to Norris & Ortegas (2003, p. 717) statement that
[a]lthough by no means in a state of theoretical accord, the field of SLA is, on the
whole, interested in describing and understanding the dynamic processes of language learning (learning used here in its broadest sense) under conditions other
than natural first language acquisition. Where they part company is in the realm
of the data that are appropriate for use. This is not surprising given that in the
1960s and 1970s, methods were not a central focus of discussion and furthermore, there is greater diversity today with methods of data elicitation being
limited only by ones imagination (see Gass & Mackey, 2007). Books on second
language research methodology abound today (cf. Drnyei, 2007; Larson-Hall,
2009; Mackey & Gass, 2005 for recent examples) and there is a greater sophistication today than could ever have been imagined 4050 years ago (cf. recent books
on specific methodologies such as Bowles, 2010; Duff, 2008; Drnyei & Taguchi,
2009; Gass & Mackey, 2000; Jiang, 2012; McDonough & Trofimovich, 2009, to
name a few).
In this chapter, we take a step back and consider not just measurement, but the
data we need to elicit in order to measure variables. We consider the development
of data elicitation through the lens of the 1972 paper Interlanguage emphasizing
the important and oft-cited statement that we need to focus our analytical attention upon the only observable data to which we can relate theoretical predictions: the
utterances which are produced when the learner attempts to say sentences of a TL
(1972a, pp. 213214, emphasis in original). We relate this statement to todays SLA
emphasis and consider the types of data that have been and are currently being
used in SLA.

 Susan Gass and Charlene Polio

Selinker explicitly claimed that two types of data were inappropriate for second language research: 1) grammatical judgments because researchers will gain
information about another system, the one the learner is struggling with, i.e., the
TL (1972a, p. 213) and 2) behavior which occurs in experiments using nonsense
syllables (1972a, p. 210). He claimed that both data types are irrelevant because
neither represents meaningful performance where meaningful performance is defined (1972a, p. 210) as:
the situation where an adult attempts to express meanings, which he may already
have, in a language which he is in the process of learning...Thus, data resulting
from these ... behavioral situations are of doubtful relevancy to meaningful performance situations, and thus to a theory of second-language learning.

If these two data types are not relevant, what are relevant data in Selinkers Interlanguage paper? As noted, they are data that come from meaningful performance
situations and from three sources: 1) utterances in the learners native language
produced by the learner, 2) IL utterances produced by the learner, and 3) TL utterances produced by native speakers of that TL. It is these three data types that are
the psychologically-relevant data of second-language learning, and theoretical
predictions in a relevant psychology of second-language learning will be the surface structures of IL sentences (p. 214). We explore the impact of these thoughts in
early SLA studies emphasizing the gradual recognition of the importance of understanding SLA through a multiplicity of data types and the common view 40 years
later that data are not to be deemed appropriate or inappropriate in a vacuum, but
rather need to be understood only in the context of the research questions asked.
Data in historical context
As a way of understanding the impact of this methodological perspective, we turn to
an overview of what data were being used in years leading up to and soon after 1972,
focusing specifically on the two data types that Selinker argued were not relevant to
SLA. If there were an impact of Selinkers (1972a) claims, we would expect that prior
to the early to mid-70s we would see studies using the types of data that Selinker
argued against and after that time, a tapering off. In other words, how pervasive was
the use of these so-called not allowable data? To do this, we selected one journal,
Language Learning, and examined its content between the years 19671979.1
1. We selected Language Learning primarily because it was the only journal in existence in the
1960s that had as its specific focus language learning (viz. the title of the journal) and had been
in existence for some time prior to the years in question. We recognize that there were other
journals in existence at that time, for example, TESOL Quarterly. However, we opted not to

Chapter 7. Methodological influences of Interlanguage (1972)

50
45
40
35
30
25
20
15
10
5
0

er
Ot
h

n
io
sit
Po

g
in
st
Te

is
ys
al
an
ta

Da

rip
tio
sc
De

go
g
da
Pe

19671972
19731979

Number of articles

We considered approximately 250 articles and divided them into six categories. Pedagogy included articles with a focus on teaching methodology; Descriptions simply described some aspect of a particular language. Data analysis, our
area of interest, consisted of empirical studies containing original data for which
there was some sort of data analysis. The last three included Language testing, Position papers, and Other. Figure 1 displays these categories in two time periods: 1)
the years leading up to the publication of the Interlanguage paper (19671972)
and 2) the years after (19731979). 2
As can be seen from Figure 1, in the late 60s-early 70s, most articles were
about pedagogy or dealt with a linguistic description of a language or a comparison of languages. In the second period, the tide turned with most articles being in
the data analysis category. Within the 196779 period, of the 57 data analysis articles, 39% dealt with spontaneous data (oral and written) and the remainder dealt
with some sort of elicited data (e.g., from the Bilingual Syntax Measure). Only
seven of these articles used either nonsense data or judgment data. In other words,
at that time, there were not many scholars who used either of the two data types
that Selinker claimed to be irrelevant.

Figure 1. Categories of articles in Language Learning: 19671979

include articles from this journal for two reasons: (1) it was a new journal at the time with
Volume 1 corresponding to our first year of investigation. We did not want to include a newlyfounded journal that was just finding its voice and (2) its focus was on language teaching rather
than on language learning. Because Selinker was making the point that teaching and learning
foci were different, we wanted to take as conservative a position as possible and utilize the article
count only from Language Learning, the journal with a focus on learning.
2. A reviewer made the important point that the actual dissemination of the paper predates its
publication.

 Susan Gass and Charlene Polio

12

Number of articles

10
8
1970
1971
1972

6
4
2

r
Ot
he

on
Po
sit
i

g
st
in

na
ta
a
Da

Te

lys
is

tio
n
sc
rip

De

Pe

da
g

og
y

Figure 2. Categories of articles in Language Learning: 19701972

We next looked more closely at the period immediately prior to the 1972 publication date (19701972). Coincidentally (or not),3 it is only in 1972 that scholars
began to deal seriously with second language data as reflected in Figure 2. It is
precisely in 1972 that the picture changed with the preponderance of articles appearing in Language Learning falling into the data analysis category.
In the next two sections, we deal more specifically with the two data types at
issue. We first discuss grammaticality judgments and then nonsense syllables/
words, both from an historical perspective.
Grammaticality judgments
We turn to a discussion of the role of grammaticality judgments4 in the history of
second language research. To do this, we first look at some direct challenges to the
suggestion of the inappropriateness of grammaticality judgments in second language research; we then look at some milestone studies5 utilizing grammaticality
3. Elaine Tarone, one of the original participants in early discussions of SLA in Edinburgh,
argues that the discussions that occurred in those days leading up to the publication of the 1972
paper in fact stimulated the serious look at learner-language. She also speculates that one of the
reasons Corder invited Selinker to Edinburgh was precisely to discuss these important issues.
4. In this chapter we use the common term grammaticality judgment rather than the more
accurate term acceptability judgment.
5. An anonymous reviewer made the point that some key studies from the 1970s (e.g., WagnerGough & Hatch, 1975; Huebner, 1979) seemed to heed the call made by Selinker by using only
naturalistic data. One, of course, does not know why they used only naturalistic data; was it a

Chapter 7. Methodological influences of Interlanguage (1972)

judgments, and finally we present data reflecting the use of grammaticality judgments in recent second language studies.
We will not deal with the history of judgment data outside of SLA in this
chapter (see Schtze, 1996 for a relevant treatise), but suffice it to say that the
tradition of their use comes from the field of linguistics, which used elicited
judgment data as a way of determining acceptable and unacceptable utterances
in a language in order to make inferences about what was grammatical and what
was ungrammatical in that language. The basic argument used by linguists is
that production does not equal the totality of ones knowledge, and some of the
same arguments were used in early second language research.6 Despite the decidedly linguistic orientation of the field of SLA at the time, as noted earlier,
there was a lack of use of judgment data. This is quite possibly because of another influence on the emerging field of SLA, namely, from the field of child
language acquisition where because of age, judgment data were not a frequent
method of elicitation.
Taking a view opposite to that of Selinker, Corder (1973), writing at about the
same time as Selinker, argued that intuitional data are an important part of understanding L2 knowledge. He viewed learners as native speakers (NS) of their own
language (and perhaps each learner is the only NS of that language). In this view,
each interlanguage system is an error-free entity. And, importantly for the purposes of understanding the role of judgment data, he pointed out that a learner has
intuitions about the grammaticality of his language which are potentially investigable (p. 36). Corder was early in the recognition of the importance of multiple
data sources. He argued that it was important to have textual output, namely those
data that a learner produces (probably similar to Selinkers meaningful performance data). One then creates hypotheses that emanate from those data. Finally,
there is a need for what he calls elicitation procedures. These are data that are
needed to validate or invalidate the hypotheses created by the analyst. Descriptively adequate accounts of interlanguage must correspond to the intuitions of
learners and judgment data are based on ones interlanguage grammar. As Corder
puts it: To suggest otherwise is to suggest that a learner might say That is the
form a native speaker would use, but I use this form instead. (p. 41). Thus, at the

deliberate attempt to follow Selinkers directive? We point out, however, that only using naturalistic data can also limit what we learn about production.
6. There are many other areas of L2 knowledge and L1-L2 comparisons that can be made, such
as processing differences (including speed). They are beyond the scope of this chapter and will,
consequently, not be discussed here.

 Susan Gass and Charlene Polio

same time as the appearance of Selinkers (1972a) paper, Corders article expressed
a different opinion regarding the use of intuitional data.7
Other scholars wrote in the years following with specific arguments in favor of
the use of grammaticality judgment data, at times citing Selinkers comments
against them. One such example comes from Schachter, Tyson, and Diffley (1976).
A basic assumption that they emphasize is the need to characterize the learners
interlanguage, at different stages, in the acquisition of the target language (p. 67).
They argued [w]e are interested in characterizing learner knowledge of his language not simply learner production (p. 67). In arguing against Selinkers position, they provided the following hypothetical case:8
Suppose some linguist was an adult learner of Hebrew, and that linguist was interested in characterizing his own interlanguage. It surely would be legitimate for him
to consult his own intuitions (use introspective evidence) in his attempt to characterize his own interlanguage. And just as it would be legitimate for him to use his
own intuitive judgments as data, so also would it be legitimate for him to use the
intuitive judgments that he could elicit from other adult speakers of their interlanguages, as long as he used appropriate caution in interpreting the data. (pp. 68-69)

They supported their argument with judgment data from learners who were able
to distinguish between sentences about which they had no judgment and those
about which they did have judgments, the former the authors referred to as indeterminate sentences. Thus, through the use of judgment data, a theoretical construct of indeterminacy emerged.
Hyltenstam (1977) was also an early proponent of multiple sources of data,
including grammaticality judgments. His main concern was with the need to capture a large enough sample size with enough room to show what he called maximal variation (p. 385), something that is unlikely to occur with production data
alone. Further, data need to tell us something of the linguistic competence of the
learner (p. 385). The problem, then, was how to collect a large enough sample size
without forced production. His solution, not dissimilar to that of Corder discussed
earlier, was to collect freely produced data with the idea that these data provide
information on problematic parts of language. From this, one develops hypotheses
about what is problematic and why with the next step being to test these hypotheses using formally elicited data which in turn leads to greater detailed hypotheses
about the route of acquisition which can then be tested with more freely produced
data. This is illustrated in Figure 3.
7. Tarone (personal communication) reports that these papers evolved from discussions between Selinker and Corder at Edinburgh.
8. One can assume that this was not entirely a fabricated example given that Selinker himself
was an adult learner of Hebrew!!

Chapter 7. Methodological influences of Interlanguage (1972)

MORE
FREELYPRODUCED
DATA

FREELYPRODUCED
DATA

MORE
DETAILED
HYPOTHESES

HYPOTHESIS
GENERATION

DATA
ELICITATION

GJs

Figure 3. A schematic of Hyltenstams use of multiple data sources

Gass (1979) also argued for the use of multiple data sources, but with a slightly
different perspective. Both Corder and Hyltenstam had made the argument for
sequential data sources, with one building on the other. Gass, who investigated the
acquisition of L2 relative clauses, used judgment data not in response to any other
data source, but as one data type simultaneously collected with other types
(sentence combining and written production). Grammaticality judgments were an
important source of data primarily because of the linguistic tradition on which her
study was based; the other data sources were used as corroborative with the recognition of task influences on outcomes. Tarone (1979) recognized the importance
of multiple data sources in her discussion of sociolinguistic variation.
We briefly mention two sets of papers that appeared 1520 years following
the appearance of the Interlanguage paper, all of which are oft-cited and provided important results in our understanding of second language learning and
second language knowledge through the use of intuitional data. The first set addresses the issue of ultimate attainment (Birdsong, 1992; Coppieters, 1987) and
the second deals with age of arrival and ultimate attainment (Johnson & Newport, 1989, 1991).

 Susan Gass and Charlene Polio

The first set of papers addressed differences between competence and performance in the technical sense. Coppieters (1987) notes that language use is both
too rich and too incomplete to serve as the only or even the primary clue to the
underlying grammar of a speaker. Thus observed equivalence in language use
between NSs and NNSs, rather than providing an answer, raises a question concerning the underlying grammars of such speakers (p. 547). After a substantial
discussion about intuitional data (but no apology or justification for their use),
Coppieters presents data from near-native speakers and native speakers of
French in which he probed their intuitions and interpretations about complex
French sentences showing that despite production levels that were near equivalent, underlying grammars were not. This led him to consider the internal organization of language and, in particular, to consider formal versus functional
aspects of language. Birdsongs study was similar in orientation and in the use of
intuitional data.9 In his case, however, the data were scalar (15). Birdsong
pointed out numerous shortcomings of Coppieters study (including participant
selection and categorization of data), coming to the conclusion that there was
less divergence between native and non-native grammars than was the case in
Coppieters work. The major factor in understanding divergences was the age of
arrival in France. The important point is that it is only through the use of something other than production data that this discussion could have taken place.
The Johnson and Newport studies were similar in scope to the Coppieters and
Birdsong studies and similar in the data used. Johnson and Newport directly tested the critical period hypothesis through the use of judgments on oral sentences.
They did justify the use of grammaticality judgment data in the 1989 study by arguing that even six-seven year old children can do the same task so that the metalinguistic skills needed are minimally demanding. For whatever reason, in the
1991 study, there was no justification provided for the use of grammaticality judgments. This data source was important in their determination of the advantage (up
until puberty) on learning for those with an early arrival in the target language
country over those who arrived later in age. They concluded that there was a critical period effect for second language learning. The second study (1991) focused on
a Universal Grammar principle (subjacency). Their results suggested a deterioration of language learning abilities as a function of age. Thus, judgment data figured
prominently in discussions of theoretical issues relating to the human capacity to
learn a second language.

9. Other tasks formed part of Birdsongs study (e.g., think-alouds). A thorough discussion
goes beyond the scope of this historical overview.

Chapter 7. Methodological influences of Interlanguage (1972)

In the final part of this section on judgment data, we present data that reflect
the scope of articles published in recent second language research10 that made use
of judgment data. To do this, we examine publications from three prominent journals: Second Language Research, Studies in Second Language Acquisition, and
Language Learning. Because of our selection of journals with a greater linguistic
orientation, we had expected that they would be more likely to include data analysis studies that used judgment data, but this turned out not to be the case, as can
be seen in Figure 4, where only 85 of 418 articles used judgment data.
What is different in the then-and-now picture is that in the earlier days, grammaticality judgments, when used, were generally the only data source whereas in
todays research environment, they are combined with other information, such as
reaction times, or other means of data elicitation, such as oral production, EEGs,
eye-tracking, and measures of aptitude.
To provide a sampling of the type of questions being addressed today using
judgment data, we mention four studies:11 Roberts, Gullberg, and Indefrey (2008),
Goo (2012), Abrahamsson (2012), and Schulz (2011). We intend for this selection
to show the range of use of judgment data in current second language research; it
is, of course, by no means exhaustive.
350

Number of articles

300
250
200
150
100
50
0

Did not use GJs

Used GJs

Figure 4. Articles in Second Language Research, Studies in Second Language Acquisition

and Language Learning: 20072012

10. We are not including in these examples the large number of studies that look at judgment
data as an object of inquiry, in particular those that look at various modes of presentation
(e.g., timed and untimed) and their construct validity (e.g., Ellis, 1991, 2004, 2005, 2009; Ellis &
Loewen, 2007; Loewen, 2009; Gutirrez, 2013.)
11. Quite clearly, these are not an exhaustive list of studies using judgment data; we selected
these because they seemed to represent a diversity of questions asked, different models for the
use of judgment data, and a range of data sources beyond judgment data.

 Susan Gass and Charlene Polio

Roberts et al. (2008) examined the interpretation of subject pronouns in discourse

from L1 Turkish and German and L2 Dutch. Both German and Dutch are nonnull subject languages whereas Turkish allows null subjects. Their study used
eye-movement data in addition to judgment data. In the eye-tracking part, participants read a text for meaning and then selected the preferred referent for the
pronoun; in the second (a paper and pencil task judgment task), participants rated
texts on a scale from one to six for acceptability. Following the judgment task, a
comprehension test was administered to further probe pronoun preferences. Only
with the online processing eye-tracking task could one see differences between the
two non-native speaking groups and the native speaking (control) group. Thus,
this study was used as a way of looking at multiple data sources and differences in
results based on those different elicitation measures.
Goo (2012) had an entirely different focus and used the results from judgment
data in a more traditional way, namely to assess linguistic knowledge following an
intervention. In his case, he considered the efficacy of using recasts and metalinguistic feedback to Korean learners of English on sentences with that-trace violations (Who do they think that loved Jennifer?) with a moderator variable of working
memory. In addition to grammaticality judgments was a written production test.
In many of the early studies, one would find a justification (with or without an
apology) for the use of grammaticality judgments; in Goos study, there was none.
His methodology differed from that of Roberts et al. (2008). Rather than simply
asking for an acceptability rating, Goo asked learners to mark sentences as grammatical/ungrammatical and after the task were asked to return to those sentences
they had marked ungrammatical and make the appropriate correction (see also
Gass & Alvarez-Torres, 2005). Thus, judgment data were used in this study as a
dependent variable to determine efficacy of feedback.
Abrahamsson (2012) used judgment data in a study investigating age of arrival and ultimate attainment in the acquisition of morphosyntax (subject-verb
inversion, reflexive possessive pronouns, adverb placement in restrictive relative
clauses, and adjective agreement [gender and number] in predicate position) and
phonetics (voice onset time) by L1 Spanish/L2 Swedish speakers with an age of
arrival between one and 30 years. Grammaticality judgments were given to auditory presentation of sentences with participants responding grammatical or
ungrammatical with no correction being required. The results indicated negative
correlations between age of arrival and ultimate attainment for grammatical and
phonetic intuitions for early learners; native-like intuitions of both grammatical
and phonetic aspects ceased to occur at age 13 with the most likely chance of performing like native speakers being for those with an age of arrival from about
one to six years old. Only early L2 learners developed both grammatical and phonetic aspects of the L2 simultaneously. As mentioned earlier, there was often an

Chapter 7. Methodological influences of Interlanguage (1972)

accompanying apology/justification in earlier days (perhaps in reaction to Selinkers

article) when judgment data were used. Even in 2012, Abrahamsson stated:
...this study aimed to investigate aspects of L2 speakers passive, unconscious, and
implicit knowledge of Swedish grammar and pronunciation, without the need to
draw indirect conclusions from performance data. Implicit language knowledge...
is equated with acquired nonverbalized linguistic intuition, whereas explicit language knowledge corresponds to learned and, to a large extent, metalinguistic
competence that can be verbally expressed by speakers. (p. 197)

Later, he states, the focus in studies of very advanced or near-native L2 speakers

should not be on what they can do but rather on what they cannot do (p. 197). In
other words, only through some sort of probing of intuitions is this information
available. This does not mean that naturalistic production can never reveal information about what a learner does not know; it does mean that it is virtually impossible to make that determination with confidence from naturalistic data alone.
We have not dealt with the important issue of the knowledge base that
grammaticality judgments are tapping. For example, are they tapping implicit
knowledge, as is mentioned by Abrahamsson above? Or, are they tapping explicit
knowledge? This may, of course, depend on any time limits imposed on responses
that learners give. In research in the 70s and even 80s, judgments were generally
paper and pencil tests; more recently, however, computer-delivered judgment data
are able to impose time limits on judgments, perhaps impacting the knowledge
base being tapped. Methodological issues of judgment data go beyond the scope of
this chapter, but recent SLA research is focused on precisely this issue (cf. Ellis, et
al., 2009 and Rebuschat, in press).
Schulz (2011) investigated a syntactic structure (wh-scope marking) that does
not exist in the L1 or the L2. Even as late as 2011 and even in Second Language
Research, a journal with a heavy linguistics orientation, she justified the use of
grammaticality judgments in her study saying:
Even though acceptability judgment experiments are often criticized for being
somewhat unnatural linguistic tasks, they do overcome certain problems inherent in most production tasks, such as that of interpreting the non-occurrence of a
particular phenomenon....For this reason, acceptability judgments can serve as an
insightful addition to production data. (p. 326)

What was novel about her use of judgment data was the inclusion of an offline and
an online task. In the former, participants were provided with context for a sentence about which they had to make an intuitional judgment and, if the judgment
was that the sentence was incorrect, make suggestions for improvement. In the
online task, she made use of a self-paced reading task. In particular, she investigated the extent to which simplification was involved when processing complex

 Susan Gass and Charlene Polio

wh-questions. If simplification were involved when the learner is engaged in actual processing, then one would expect there to be more acceptance of wh-scope
marking in an online than in off-line task. What was novel, if not unique to judgment data, was her ability to determine how processing limitations might interact
with language development through a comparison of the on-line and off-line judgment results. Her results allowed her to make the argument that the non-L1/L2
form is a result of simplification when faced with a difficult processing task.
Nonsense data
We turn now to nonsense syllables and words, the second data type that Selinker
referred to as unacceptable. Nonsense data, in his view, were irrelevant because
they are not part of meaningful performance, as noted earlier. In this section,
we consider studies that do not use real language, namely studies that use nonsense (also called nonce) syllables and words as well as studies that use artificial
languages. It is not clear from Selinkers paper that he would have been opposed
to the latter, but we include such studies here given that they are not part of
meaningful performance especially when considering that most are done in
confined laboratory contexts. It is also not clear if Selinker was reacting to published research at the time or if he made the comment in anticipation in SLA
research of the use of nonsense syllables (common at the time in psycholinguistics studies).
Early research using nonsense syllable and words
As with our discussion of grammaticality judgments, we asked a question regarding the pervasiveness of nonsense data and artificial languages before and after the
publication of Selinkers (1972a) paper. We chose the time period 1960 to 1980,12
12 years before the Interlanguage paper and 8 years after. Because there was only
a small number of studies (3) published in what we would consider SLA journals,
we broadened our search for articles using nonsense words to Linguistics and Language Behavior Abstracts (LLBA), which covers studies outside of SLA. We found
that there were over 500 in LLBA, but interestingly, none before 1970. To give a
sense of what researchers outside of SLA journals were using nonsense syllables
and words for, we provide three examples of some of these 500 studies. Although
12. We took a slightly different time period than what we had looked at with grammaticality
judgments (beginning in 1960 rather than 1967). We wanted to take a slightly broader scope
given the realization that so few studies had been published using nonsense data.

Chapter 7. Methodological influences of Interlanguage (1972)

these are not SLA studies, we believe they would have been relevant to the then
emerging field of SLA.13
Many of the 500 studies were in the area of speech pathology and seem very
much related to the learning of pronunciation. As one example, in a study of articulation training, McReynolds (1972) published a study in Language and Speech
that examined children who had difficulty pronouncing /s/. She studied the use of
isolated sounds, nonsense syllables, and words to see if children transferred the
sounds they made in the training sessions to new words, finding that they did
transfer from nonsense syllables to real words but not from isolated sounds.
An article published in the Journal of Learning Disabilities in 1975 by Lasky,
Jay, and Hanz-Ehrman studied differences in the auditory processing of meaningful and non-meaningful sounds (words and nonsense words, respectively) and
meaningful and non-meaningful nonlinguistic sounds (e.g., a telephone ring or
animal noise for the meaningful and tones for the non-meaningful). The authors
compared the processing of these sounds by children with and without learning
disabilities, checked by their ability to repeat a sequence of five words or sounds.
They found a similar pattern with the two groups of children, but with the nondisabled children performing better across all tasks. What motivated this study was
the fact that some auditory training programs began with nonlinguistic meaningful sounds such as animal sounds or vacuum cleaners and nonsense syllables because they were considered easier, but in fact, they were not and the children paid
less attention to the task when it was harder. Thus, in a sense, the authors were
arguing against the use of nonsense syllables in auditory training for learners. To
be clear, the authors do not state that one should not use nonsense syllables for
training, just that they are harder.
Matlin and Stang (1975) published a study in the journal Perceptual and Motor
Skills on word frequency estimations. The authors used nonsense words, telling
the students that they were Turkish words, to determine how participants estimated word frequency. They found that participants overestimated low frequency
words and underestimated high frequency words. Furthermore, they overestimated the frequency of positively evaluated words and underestimated the frequency
of negatively evaluated words. (It seems that that was done on the basis of how
pleasant sounding the word was.) They also found that words were estimated to be
more frequent when presented in a distributed manner as opposed to in a mass.
Although the authors used word lists and not connected discourse, this study
could be particularly relevant given the emphasis on frequency in current usagebased theories of SLA.
13. Many of these early studies were fraught with design problems, and we do not critique the
studies here. The conclusions given here should not be accepted as fact.

 Susan Gass and Charlene Polio

Language Learning, which, as mentioned earlier, was the journal publishing SLA
research at the time, published three studies that used nonsense syllables, but all
three studies involved child L1 learners of English, not L2 learners. Two of them
used Berkos (1958) wug test, where children are shown pictures with nonsense
names and then the plural or some other form is elicited to see if the children follow
morphophonemic rules when using novel items. Solomon (1972) found that childrens ability to produce the plural allomorph varied according to the final sound of
the nonsense word. For example, , f, , and were more difficult. Baird (1973)
studied childrens use of the phonemes /s/, /z/, and /z/ as allomorphs for plurals,
possessives, and subject verb agreement. He wanted to see if childrens use of these
sounds varied within and across morphological contexts. For example, was the /z/
sound easier to use in plural than in SV agreement? Was the /z/ sound easier than /s/
in plural? The results showed that children seemed to acquire the phonemes according to their morphological use and not by sound. In other words, if a child could
form a plural /s/, he or she was likely to form a plural /z/, but not necessarily a subject-verb /s/. Finally, in a complex study of childrens speech perception, Read (1973)
studied how children, compared to adults, perceived various sounds by having participants compare real and nonsense words and say if they were alike or not. He
found that childrens perceptions differed from adults. The point of this study was to
relate the childrens perception errors to childrens spelling errors.
Although these three studies were not SLA studies, they could have been conducted with L2 learners, but, in fact, we could not find any studies with L2 learners
and nonsense words conducted prior to 1980. Although some of the studies
discussed above investigated perception, others investigated production or performance. We return to Selinkers objection to these studies not constituting meaningful performance. This raises the question: Can an experimenter create meaningful performance in a lab setting? And if the wug studies, for example, had been
conducted with adult L2 learners, would those results have informed us about the
system that they were learning? At the time of Selinkers (1972a) paper, there was
not frequent use of nonsense syllables words in second language scholarship. We
can only conclude that his comments were a preemptive strike against what was a
common methodological tool in related disciplines. We turn below to current
studies using nonsense syllables and words.
Current studies
To consider the impact of nonsense data in current research, we again did a search
in LLBA for articles from 20002011 and found over 1000, clearly an increase over
the number of nonsense word studies conducted in the 1970s. In order to focus
on SLA studies, we confined the search to Studies in Second Language, Applied

Chapter 7. Methodological influences of Interlanguage (1972)

Linguistics, Applied Psycholinguistics, and Language Learning and found eight articles (these in Studies in Second Language Acquisition, Language Learning, and
Applied Linguistics combined) and another 22 in Applied Psycholinguistics (the latter is not surprising given that Applied Psycholinguistics has a heavier focus on
experimental studies whereas the others often include studies based on naturalistic data or classroom language). In what follows we present a sampling of four
studies, two from Language Learning and two from Applied Psycholinguistics to
show the kinds of research questions being addressed in current SLA research
through the use of nonsense data.
Strapp, Helmick, Tonkovich, and Bleakney (2011) used nonsense forms to
study the effect of positive and negative evidence on the learning of irregular plurals and past tense verbs. They found that negative evidence was more beneficial
and, perhaps somewhat counter-intuitively, that verbs were learned better than
nouns. They refer to Berkos wug study with no mention of the use of nonsense
words being problematic, so clearly this technique has persisted for 40 years.
Kovacs and Racsmany (2008) conducted a study of phononological shortterm memory. Participants heard a word and repeated it. There were four types of
words: high probability L1-sounding-nonwords; low probability L1-sounding
non-words (low frequency consonant clusters), non-words with illegal phoneme
sequences, and non-words with non-L1 sounds. They found that the illegal sounds
were more difficult than the illegal sequences, which caused difficulty only when
they increased to six syllables.
Webb (2007) studied how much incidental exposure was needed to learn a
word. He used nonsense words to control the number of times learners heard a
new word. To do this, he replaced words with nonsense words in studies of EFL
learners. For example: He was not ill, and of course the beds in the ancon are for ill
people. And in addition to repetition, the context was manipulated: One day in
1994, I saw a picture in the window of a shop near the ancon. Here, the meaning is
not obvious from the context. To test learning, ten measures were used including
translation and sentence construction. The results showed greater gains per repetition, but even seeing the word ten times in context did not result in full learning. Only nonsense words can provide us with this kind of information because
they are the only words that researchers can be certain have never been encountered by participants.
Finally, Bird and Williams (2002) studied the presentation of words and nonwords either aurally or simultaneously with the written word or non-word. They
were interested in the effects of bimodal presentation as well as the effect on both
implicit memory, as measured by a lexical decision task, and explicit memory, as
measured by a test where participants are asked if they had seen the word in the
presentation. They found that bimodal input boosted implicit memory only in the

 Susan Gass and Charlene Polio

nonword condition. For real words, aural input alone was sufficient. For explicit
memory, bimodal presentation improved recall of words and non-words. What is
particularly interesting about this study is that there were two experiments done,
but in the first, they used very low frequency words that native speakers did not
know including: pavis, pongid, and mazard. In the second study, they used true
nonsense words. They did not discuss this decision suggesting that there is no difference, essentially, between a nonsense word and a word that is new to a second
language learner.
If we take these ten studies together, six from the 1970s and four more current
studies, we find that nonsense words are sometimes used in isolation, and sometimes embedded in another language (in all cases English). The studies may involve repetition, recognition, production, or lexical decision. To be more specific,
Read (1973) studied perception via a judgment task, McReynolds (1972) studied
training via a word production task, Matlin and Stang (1975) studied perception
by a frequency estimation task, Lasky et al. (1975) studied perception via a repetition task, and Solomon (1972) and Baird (1973) studied production via a wug test.
In more recent studies, Kovacs and Racsmany (2008) studied phonological shortterm memory through a repetition task, Strapp et al. (2011) conducted an intervention study on morphology, still using the wug test, and two studies were
concerned with vocabulary learning, one incidental exposure (Webb) and one
uni- vs. bimodal learning using a lexical-decision task (Bird & Williams).
Although the focus of the studies varies greatly, it is clear that nonsense words
are consistently still being used to answer questions that cannot be answered with
meaningful performance data.14 One interesting point is that in these articles,
there is little explicit justification for the use of nonsense words; only Strapp et al.
(2011) said This approach eliminates confounding variables such as previous experience with forms, readiness, and practice outside of the experimental setting.
(p. 514). And, of course, this is the strength of these studies. But how many of them
looked at what Selinker would consider meaningful performance? The studies reviewed above did have participants produce more than a single word (either one
lexical item or an item plus an inflectional morpheme). If we want to look at
14. One could argue that the questions being addressed through these studies do not relate to
interlanguage as originally conceived. We acknowledge that in some cases this is a valid argument (e.g., measuring learning from different quantities of exposure), but in other cases, it is not
valid, such as is the case with the wug studies where information is being gathered about
individuals interlanguage systems. In general, Selinker refers to a psycholinguistic theory of
second-language learning (p. 211) and the need to understand the major features of the
psychological structure of an adult whenever he attempts to understand second-language sentences or to produce them (p. 211). The studies in this chapter, for the most part, fit into these
categories.

Chapter 7. Methodological influences of Interlanguage (1972)

anything beyond the phonological, lexical, or morphological level, we have to turn

to studies using artificial languages.
Artificial language studies
Artificial language studies involve teaching participants a linguistic system in
which they must try to understand, produce, or judge sentences. These are different than those discussed above in that usually the learners try to learn a whole new
system or a subset of a linguistic system, not just a word. Thus, any study using an
artificial language is a second language study (assuming that the artificial language
in question obeys universal constraints) and does constitute meaningful performance using that new system.
In a search of journals from 1960 to 1980, we found 11 studies that used artificial languages. Four were in the Journal of Verbal Learning and Behavior and
all had the same first author, Shannon Moeser. From 20002011, we found 65
studies using artificial languages with human learners (i.e., as opposed to computers). Again, the searches were in LLBA and were limited to journals only.
Most were in psychology journals but two were in Language Learning. Below we
have selected to discuss two studies from psychology journals and the two from
Language Learning.
Jones and Kaschak (2009) completed a partial replication of a previous study
by Kashcak and Saffron (2006), which found that when an idiomatic phrase that
violated the core rules of an artificial grammar was presented with different prosody, the presence of the idiomatic phrase facilitated learning the structure of the
core grammar. The partial replication (with adults rather than children) included
presenting the language with visuals, namely, a shape associated with each word.
The authors were not able to replicate the results of the previous study, which did
not use visuals, and proposed that the different results might be related to the fact
that the mapping between visual reference worlds and linguistic input can affect
learning. (p. 87).
Kersten and Earles (2001) used an artificial language to test Newports less-ismore hypothesis (1988), which attempts to account for child/adult differences. She
argued that because of childrens less developed cognitive systems they can perceive language only in smaller amounts and only then map this language on to
external events. This helps them learn morphological relationships and vocabulary better. They can then learn to combine the language faster into complex structures. There is some evidence from American Sign Language (ASL) studies to
support this, but Kersten and Earles wondered about the visual medium of ASL
and wanted to see if the results could be extended to aural-medium (non-signed)
languages. They designed a study to examine whether or not there were benefits in

 Susan Gass and Charlene Polio

forcing adults to process smaller pieces of language. Adults viewed animated

events in which they heard either full sentences throughout the learning process
or first heard one word, then two words, then three words which ultimately made
up full sentences/phrases. A series of experiments supported this hypothesis in
that learners presented with smaller segments of language were able to learn vocabulary and morphology better.
In a Language Learning study, Cornish, Tamariz, and Kirby (2009) studied
language as a complex adaptive system. Through an experimental study, they
showed that languages evolve to become easier and more structured. This is not
because there is some innate capacity for learning language but because language
is a complex adaptive system. Participants were taught an artificial language and
then asked to produce what they believed was the correct language. This new language was then taught by researchers to a new group of learners in an attempt to
simulate a new generation of learners. Cornish et al. found changes in the systems
over generations particularly in the area of compositionality, meaning that a string
of morphemes in which the parts had no distinct meanings evolved over time so
that the parts did become meaningful. For example, in the system given to the
learners, they were taught a system with no compositionality, as in example (1).
(1)

Language with no compositionality

Kapihu: black triangle horizontal
Luki:
black triangle spiral
Humo: blue triangle horizontal
Namola: blue triangle spiral

But these learners created a system with compositionality, as in example (2).

(2)

Nekiki: black triangle horizontal

Nekipilu: black triangle spiral
Lakeki: blue triangle horizontal
Lakipilu: blue triangle spiral

They go into more detail but in summary, they were able to observe language
change over a very short period of time by using an artificial language.
Also in Language Learning, Morgan-Short, Sanz, Steinhauer, and Ullman
(2010) studied adults learning an artificial language in either an explicit or implicit condition. The researchers looked at how adjective and noun gender agreement were processed, considering both comprehension and production at two
levels of proficiency. As part of their study, they collected data on electrophysiological brain response by collecting event-related potentials (ERPs) from the participants. While wearing a cap designed specifically to capture ERP data, learners
heard or saw sentences containing grammatical violations. On a judgment test,

Chapter 7. Methodological influences of Interlanguage (1972)

there was an interaction effect for group and proficiency level on noun-article
agreement in which the implicit group did better than the explicit group. As for
the ERP findings, the researchers also found some differences between the groups
and there was an interaction for group and type of violation (noun-adjective or
noun-article).
In artificial language studies, one obvious question is: how much did these
languages look like real languages? Did they simply include a subject and a verb,
and maybe an object? Did they include dependent clauses and recursivity? Did
they follow the rules of Universal Grammar? Delving into the design of the studies
a bit more reveals some differences. Jones and Kaschak detailed their language,
taken from Kaschak and Saffron (2006), which includes a set of phrase-structure
rules. They presented all the sentences used in the study, making it easy for other
researchers to replicate it. Clearly, however, the language was a simple one compared to real languages.
The Kersten and Earles (2001) study used an artificial language created from a
grammar based on ASL. They wanted to keep the grammar the same but remove
the visual linguistic input; participants heard sentences with accompanying pictures. Simple sentences that included a morpheme expressing manner of motion,
something that exists in ASL, were used. It appears that no complex sentences
were used. The language discussed in the Cornish et al. (2009) study included a
language that had shapes, colors, and movement. For example, a blue triangle
could move horizontally or a red square could bounce. But again, it did not seem
to be a complex language with complex rules. The Morgan-Short et al. (2010) study
used a language called BROCANTO2 which was designed to be similar to
Romance languages and diverge from English in specific ways. The language included two articles, two adjectives, four nouns, four verbs, and four adverbs, but
there appeared to be no complex sentences.15
The other point to be considered here is not just the language itself but how it
is taught or learned. The amount of exposure and learning tasks varied, of course,
among the studies. As Cornish et al. (2009) said in their study Clearly, this experimental setup represents a highly simplified idealization of the real process of
linguistic transmission (p. 190).
In summary, the issues related to studies using non-real languages, studies use
nonsense words and artificial languages primarily to control for variables (such as
exposure). They also have the benefit of allowing for easier replication. The studies
15. During the question and answer period at the TCCRISLS symposium, Selinker stated that
he was not opposed to the use of artificial languages if they were very sophisticated systems that
resembled natural languages. This is not the case in the artificial languages used in L2 studies
leading us to conclude that artificial languages would be a part of Selinkers disallowable data.

 Susan Gass and Charlene Polio

using nonsense words include single words embedded in English. They could be
argued to have ecological validity in that they are simply new words that anyone
might encounter when learning a new language. As for the artificial language
studies, one can imagine among researchers with certain theoretical orientations
would be considered objectionable. These studies are by definition laboratory
studies where everything is tightly controlled. While in our opinion these studies
are valuable, it would be difficult to say that they resemble the real language learning process. Was Selinker suggesting that such studies lack ecological validity? In
other words, artificial language learning experiments can never, for a variety of
reasons, approximate naturalistic learning or classroom learning; the conditions
are not the same.
Conclusion
We have considered the impact of two so-called prohibited data types identified in
Selinkers (1972a) article, taking an historical perspective looking first at the types
of articles published prior to 1972 as well as after. We then reported grammaticality judgment studies and nonsense data studies that were published following the
publication of his article. In general, we found that both data types are alive and
well in current SLA research. What the Interlanguage paper did was to problematize the use of grammaticality judgments and nonsense data. It did not stop
SLA researchers from using these data types; it did make them think twice with
the path for each of the data types being different. For grammaticality judgment
data, for many years, their use was accompanied by some sort of justification, although to claim a cause/result relationship would be to step beyond appropriate
bounds. Second, in more recent years, judgment data were not often used as a sole
method of elicitation; rather, they were used as one of many data sources. The history of nonsense data took a somewhat different path, possibly because nonsense
data were never as prevalent as judgment data. This was likely the case because of
the main disciplinary impact on the field, namely, linguistics, which relied heavily
on intuitional data.
In todays research climate, the issue of methodology is more forgiving in the
sense that there are few prohibited data types. The more important issue is that
researchers carefully align their data type with their research questions. In the case
of experimental SLA, where there is increased influence from psychology and
psycholinguistics in terms of the research questions being asked, it is not surprising to see an increase in data emanating from those disciplines, nonsense data
being one type.

Chapter 7. Methodological influences of Interlanguage (1972)

To conclude, the impact of Selinkers Interlanguage paper in the realm of

methodology was felt not so much in the use or non-use of a particular data type
by SLA researchers, but it did have an impact on the attention SLA researchers
paid to issues of methodology. It didnt make researchers in the field of SLA stop
the use of any data type, but it did give the field occasion to more fully reflect on
the research methods used for data elicitation.
References
Abrahamsson, A. (2012). Age of onset and nativelike L2 ultimate attainment of morphosyntactic
and phonetic intuition. Studies in Second Language Acquisition, 34, 187 214.
Baird, R. (1973). Childrens phonological rules: A failure to replicate. Language Learning, 23,
223230.
Berko, J. (1958). The childs learning of English morphology. Word, 14, 150 -77.
Bird, S. & Williams, J. (2002). The effect of bimodal input on implicit and explicit memory: An
investigation into the benefits of within-language subtitling. Applied Psycholinguistics, 23,
509533.
Birdsong, D. (1992). Ultimate attainment in second language acquisition. Language, 68, 706755.
Bowles, M. (2010). The Think-aloud Controversy in Second Language Research. New York, NY:
Routledge
Brire, E. (1968). A Psycholinguistic Study of Phonological Interference. The Hague: Mouton.
Coppieters, R. (1987). Competence differences between native and nonnative speakers.
Language, 63, 544573.
Corder, S.P. (1973). The elicitation of interlanguage. In J. Svartvik (Ed.), Errata: Papers in Error
Analysis (pp. 3648). Lund: CWK Gleerup.
Crothers, E. & Suppes, P. (1967). Experiments in Second Language Learning. New York, NY: Academic Press.
Cornish, H., Tamariz, M., & Kirby, S. (2009). Complex adaptive systems and the origins of
adaptive structure: what experiments can tell us. Language Learning, 59 s1, 187205.
Drnyei, Z. (2007). Research Methods in Applied Linguistics. Oxford: OUP.
Drnyei, Z. & Taguchi, T. (2009). Questionnaires in Second Language Research: Construction,
Administration, and Processing. New York, NY: Routledge.
Duff, P. (2008). Case Study Research in Applied Linguistics. New York, NY: Taylor & Francis.
Ellis, R. (1991). Grammaticality judgments and second language acquisition. Studies in Second
Language Acquisition, 13, 161186.
Ellis, R. (2004). The denition and measurement of L2 explicit knowledge. Language Learning,
54, 227275.
Ellis, R. (2005). Measuring implicit and explicit knowledge of a second language: A psychometric study. Studies in Second Language Acquisition, 27, 141172.
Ellis, R. (2009). Measuring implicit and explicit knowledge of a second language. In R. Ellis, S.
Loewen, C. Elder, R. Erlam, J. Philp, & H. Reinders (Eds.), Implicit and Explicit Knowledge in
Second Language Learning, Testing and Teaching (pp. 3164). Bristol: Multilingual Matters.

 Susan Gass and Charlene Polio

Ellis, R., & Loewen, S. (2007). Conrming the operational denitions of explicit and implicit
knowledge in Ellis (2005) Responding to Isemonger. Studies in Second Language Acquisition, 29, 119126.
Ellis, R., Loewen, S., Elder, C., Erlam, R., Philp, J., & Reinders, H. (2009). Implicit and Explicit
Knowledge in Second Language Learning, Testing and Teaching. Bristol: Multilingual Matters.
Gass, S. (1979). Language transfer and universal grammatical relations. Language Learning, 29,
327344.
Gass, S., & Alvarez Torres, M. (2005). Attention when?: An investigation of the ordering effect
of input and interaction. Studies in Second Language Acquisition, 27, 131.
Gass, S., & Mackey, A. (2000). Stimulated Recall Methodology in Second Language Research.
Mahwah, NJ: Lawrence Erlbaum Associates.
Gass, S., & Mackey, A. (2007). Data Elicitation for Second and Foreign Language Research.
Mahwah, NJ: Lawrence Erlbaum Associates.
Goo, J. (2012). Corrective feedback and working memory capacity in interaction-driven L2
learning. Studies in Second Language Acquisition, 34, 445474.
Gutirrez, X. (2013). The construct validity of grammaticality judgment tests as measures of
implicit and explicit knowledge. Studies in Second Language Acquisition, 35, 423449.
Huebner, T. (1979). Order-of-acquisition vs. dynamic paradigm: A comparison of method in
interlanguage research. TESOL Quarterly, 13, 2128.
Hyltenstam, K. (1977). Implicational patterns in interlanguage syntax variation. Language
Learning, 27, 383411.
Jakobovits, L. (1970). Foreign Language Learning: A Psycholinguistic Analysis of the Issues.
Rowley, MA: Newbury House.
Jiang, N. (2012). Conducting Reaction Time Research in Second Language Studies. New York, NY:
Routledge.
Johnson, J., & Newport, E. (1989). Critical period effects in second language learning: The influence of maturational state on the acquisition of ESL. Cognitive Psychology, 21, 6099.
Johnson, J., & Newport, E. (1991). Critical period effects on universal properties of language:
The status of subjacency in the acquisition of a second language. Cognition, 39, 215258.
Jones, J.L., & Kaschak, M. P. (2009). Do idiomatic constructions always aid language learning?
Language Learning and Development, 5, 6993.
Kaschak, M.P., & Saffran, J.R. (2006). Idiomatic syntactic constructions and language learning.
Cognitive Science, 30, 4363.
Kersten, A., & Earles, J. (2001). Less really is more for adults learning a miniature artificial
language. Journal of Memory and Language, 44, 250273.
Kovacs, G., & Racsmany, M. (2008). Handling L2 input in phonological STM: The effect of nonL1 phonetic segments and non-L1 phonotactics on nonword repetition. Language Learning, 58, 597624.
Larson-Hall, J. (2009). A Guide to Doing Statistics in Second Language Research Using SPSS. New
York, NY: Routledge.
Lasky, E., Jay, B., & Hanz-Ehrman, M. (1975). Meaningful and linguistic variables in auditory
processing. Journal of Learning Disabilities, 8, 570577.
Lawler, J. & Selinker, L. (1971). On paradoxes, rules, and research in second-language learning,
Language Learning, 21, 2743.

Chapter 7. Methodological influences of Interlanguage (1972)

Loewen, S . (2009). Grammaticality judgment tests and the measurement of implicit and explicit L2 knowledge. In R. Ellis, S. Loewen, C. Elder, R. Erlam, J. Philp, & H. Reinders (Eds.),
Implicit and Explicit Knowledge in Second Language Learning, Testing and Teaching (pp.
94112). Bristol: Multilingual Matters.
Mackey, A., & Gass, S. (2005). Second Language Research: Methodology and Design. Mahwah, NJ:
Lawrence Erlbaum Associates.
Matlin, M. & Stang, D. (1975). Some determinants of word-frequency estimates. Perceptual and
Motor Skills, 40, 923929.
McDonough, K., & Trofimovich, P. (2009). Using Priming Methods in Second Language Research.
New York, NY: Routledge.
McReynolds, L. (1972). Generalization during speech sound discrimination training. Language
and Speech, 15, 385392.
Morgan-Short, K. & Sanz, C., & Steinhauer, K., & Ullman, M. (2010). Second language acquisition of gender agreement in explicit and implicit training conditions: An event-related
potential study. Language Learning, 60, 154193.
Nemser, W. (1971). Approximative systems of foreign learners. International Review of Applied
Linguistics, 9, 115124.
Newport, E. (1988). Constraints on learning and their role in language acquisition: Studies of
the acquisition of American Sign Language. Language Sciences, 10, 147172.
Norris, J.M., & Ortega, L. (2003). Defining and measuring SLA. In C. J. Doughty, & M. H. Long
(Eds.). Handbook of Second Language Acquisition (pp. 717761). Malden, MA: Blackwell.
Read, C. (1973). Childrens judgments of phonetic similarities in relation to English spelling.
Language Learning, 23, 1738.
Rebuschat, P. (in press). Measuring implicit and explicit knowledge in second language research.
Language Learning.
Roberts, L., Gullberg, M., & Indefey, P. (2008). Online pronoun resolution in L2 discourse: L1
influence and general learner effects. Studies in Second Language Acquisition, 30, 333357.
Schachter, J., Tyson, A., & Diffley, F. (1976). Learner intuitions of grammaticality. Language
Learning, 26, 6776.
Schulz, B. (2011). Syntactic creativity in second language English: Wh-scope marking in
Japanese-English interlanguage. Second Language Research, 27, 313341.
Schtze, C.T. (1996). The Empirical Base of Linguistics: Grammaticality Judgments and Linguistics
Methodology. Chicago IL: The University of Chicago Press.
Selinker, L. (1972a). Interlanguage. International Review of Applied Linguistics, 10, 209231.
Selinker, L. (1972b). Second-language learning experiments and mathematical learning theory.
Language Learning, 22, 291299.
Solomon, M. (1972). Stem endings and the acquisition of inflections. Language Learning, 22,
4350.
Strapp, C., Helmick, A., Tonkovich, H., & Bleakney, D. (2011). Effects of negative and positive
evidence on adult word learning. Language Learning, 61, 506532.
Tarone, E. (1979). Interlanguage as chameleon. Language Learning, 29, 181191.
WagnerGough, K., & Hatch, E. (1975). The importance of input in second language acquisition
studies. Language Learning, 25, 297308.
Webb, S. (2007). The effects of repetition on vocabulary knowledge. Applied Linguistics, 28, 4665.

chapter 8

Trying out theories on interlanguage

Description and explanation over 40 years of L2
negation research
Lourdes Ortega

Georgetown University
This chapter examines the development of English negation through the diverse
theoretical lenses that have been applied to this phenomenon over 40 years of
interlanguage research. Depending on the theory, L2 learners are imagined to
have different learning tasks: from traversing negation stages or adding negation
strategies in a sequence, in the foundational years of the field, to learning from
affordances experienced as contingent and structured by social practices, in
recent usage-based theories. Each lens contributes theory-specific analytical
tools and explanations that in turn can advance description by triangulation and
falsification. The chapter closes with the identification of three pending issues:
accounting for variation and context, disentangling crosslinguistic influences
vis--vis universal patterns, and clarifying the ambivalent role of accuracy
in development.

Introduction
Of the many questions and topics one can investigate in the field of second language acquisition (SLA), I have always found myself attracted to the study of
language development by adults who are learning to become competent in a new
language. I still remember the exhilarating feeling when, as a graduate student, I
learned about the concept of interlanguage and read Selinker (1972). What adult
learners do in building a new language, Selinker said, is build a natural language
a second time around, and the process is a systematic one deserving rigorous
investigation. I also still remember how I marveled at the fact that Selinker could
not have known for sure in 1972 the lasting influence his article would have in
the field.
Many years later, I re-read Selinker (1972) in preparation for the anniversary
conference that forms the basis for this collection. This other reading was of course

 Lourdes Ortega

very different, and one most certainly influenced (shall I say biased?) by the benefit of some twenty years of learning and doing SLA. This time a theme in the
seminal article stood out in my imagination as striking a truly contemporary
chord: Selinkers admonition to pursue description prior to explanation in the new
field of interlanguage studies he envisioned. Many other issues also sparked my
renewed interest and admiration, and most of them are taken up by other authors
in this celebratory collection. It is undoubtedly the quality of present-time relevance that has enabled both the research program and the constructs Selinker proposed in this article to inspire the sustained interest of SLA researchers for over
four decades now.
If it could not have been easy for Selinker (1972) to anticipate the tremendous
influence the article would exert on SLA, I suspect even less easy to foresee in 1972
must have been the many theoretical worlds that the field of SLA would travel. At
the time, the intellectual influences of Chomsky (1959) and Lenneberg (1967)
seemed to be undisputed and unrivaled, in the article as in the field. Since the mid
1990s, however, the theoretical and epistemological expansion witnessed in SLA
has been formidable (e.g., compare VanPatten & Williams, 2007, and Atkinson,
2011). These developments have surely not left interlanguage studies untouched. I
began wondering: What main theories have been tried on by interlanguage researchers in four decades, since 1972? And what might one learn about the virtues
and limits of description prior to explanation with the hindsight of 40 years of research history?
In this chapter I pursue these two questions and reflect on the relationship
between description and explanation, probing possible relationships between the
two by examining the different theoretical lenses that have been applied to interlanguage over the years. As a way of illustration, I will focus on the phenomenon
of negation mostly in English as the target language. This area of development has
been investigated fruitfully and repeatedly under different theoretical premises
and thus lends itself as an ideal case study for the purposes at hand.
The theme of description vs. explanation in Selinker (1972)
Even today, I find Selinkers (1972) admonition to prioritize description before
explanation intellectually appealing: we must have a clear idea of what is in [interlanguage], even if we cannot explain why it is there (p. 204). His rationale at the
time was that a pre-condition for any explanation is accumulated descriptions of
the phenomena in question, and such descriptions were unavailable for interlanguage in 1972. Furthermore, the task of describing learner language was uncharted
and would yield new, analytically unfamiliar data; therefore, as part and parcel of

Chapter 8. Trying out theories on interlanguage

trying to make sense of this uncharted field of inquiry and the new unfamiliar
evidence, Selinker anticipated traditional linguistic theory would be unfit to guide
research on learner language development because there is no necessary connection between relevant units of linguistic theory and linguistically-relevant units of
a psychology of second-language learning (p. 204). This did not, however, mean
advocating an a-theoretical research program. Instead, he reasoned, new datadriven units of analysis were needed, which in turn also called for new constructs.
In this seminal paper, he placed the stage light on several key constructs that were
arising in the SLA landscape at the time: interlingual identification, fossilization,
and the three systems of native language (NL), target language (TL), and, of course,
Interlanguage (IL). He speculated that relevant data in the new research domain of
interlanguage studies would come from neither exceptionally good nor particularly bad but average learners striving to make meaning in their L2, supplemented
by two baselines, one by the same learners in their L1 and another by native users
in the target language, and everything to be analyzed in a non-normative, selfreferenced spirit.
Selinker (1972) could have called for the adoption of theory-available constructs, or he could have urged others interested in joining him in the new research
program to proceed with description a-theoretically. Instead, he offered a specification of the new constructs and the type of data that were to guide the task of
description and would eventually prepare the ground for explanation. In his call
for description before explanation aided by new constructs and new kinds of data
that satisfied certain requirements, then, he demonstrated a firm empirical predisposition coupled with a theoretical prescience. The benefits of these qualities
would be reaped by others over the next 40 years.
A forewarning: The complexity of description-theory relationships
Aligning with Selinkers (1972) desideratum for prioritizing description over explanation, description has been at times ahead of explanation in interlanguage
studies. However, the relationship between the two has been far from straightforward. Before proceeding to the examination of the study of English negation as a
test case, a forewarning that underscores the complexity of possible relationships
between description and explanation is in order.
A first observation is that even the very initial efforts at describing interlanguage did not entirely sideline explanations. A well-known case in point is that of
natural orders for morpheme accuracy, synthesized by Dulay, Burt, and Krashen
(1982). This line of inquiry was purely descriptive in motivation at the outset,
aiming as it did at replicating the corresponding morpheme accuracy research

 Lourdes Ortega

pioneered by Roger Brown (1973) for children growing up monolingually in English. Nevertheless, very early on Larsen-Freeman (1976) pursued the challenge of
explaining the descriptive findings, in a correlational study where she tentatively
suggested that frequency of occurrence in native speaker input might be the best
explanation, noting that the more frequently a stimulus is encountered, the more
rapidly it will be acquired (p. 133). It was not until almost two decades later that
Zobl and Liceras (1994) took up the explanatory challenge again, this time making
use of generative linguistics as the theoretical lens. They reinterpreted the accumulated descriptive evidence on morpheme accuracy orders as proof that functional
categories transfer from the first language to the second, whereas the surface inflectional morphology that instantiates them must be acquired anew by learners.
A few years later, Goldschneider and DeKeyser (2001) made a third and oft-cited
attempt at adding explanation to description. They were able to show that a combination of five factors that amounted to the two underlying constructs of frequency and salience could account for the morpheme findings; they explained the
degree and source of difficulty that learners are likely to encounter for each morpheme at a remarkable level of 71% shared variance. This post-1990s explanation
essentially elaborated on Larsen-Freemans (1976), but was greatly enabled by
ideas and methods not available to Larsen-Freeman in her very early explanatory
quest. The new ideas were inspired by the family of emergentist and usage-based
approaches (N. Ellis, 1998) and the methodology of meta-analysis (Norris &
Ortega, 2000). Clearly, interlanguage researchers have always felt the need to go
beyond description.
A second, more general observation, is that the available interlanguage descriptions are varyingly robust, useful, and satisfactory, often for a variety of reasons that may or may not have to do with explanatory success. Some descriptive
facts have proven amazingly robust, regardless of their putative explanations.
One such robust fact is the morpheme accuracy findings just mentioned. Another
one is the patterned emergence of word order in L2 German. The developmental
sequence was initially gleaned by Meisel, Clahsen, and Pienemann (1981) and by
Clahsen and Muysken (1986), with samples of immigrants to Germany from
Romance L1 and Turkish L1 backgrounds, respectively. The same sequence was
later replicated longitudinally for classroom learners from English L1 background
in the UK (R. Ellis, 1989), Australia (Jansen, 2008), and the US (Byrnes, Maxim, &
Norris, 2010). The explanation for the word order sequence has evolved over the
years, from rather global arguments pertaining to the processing of surface structures in the original work to highly technical formulations in Pienemanns Processability Theory (Pienemann & Keler, 2011). Both its pedagogical usefulness and its
educational significance have also been open to varying interpretations and critiques (e.g., Hudson, 1993). But in the end, the empirical replication of the patterned

Chapter 8. Trying out theories on interlanguage

emergence of word order in adult learners of German, in both minority and majority, naturalistic and formal language contexts, makes it count as robust knowledge
about one area of interlanguage, albeit open to alternative explanations.
My third observation is less optimistic: At least for some phenomena, it
might be that energetic empirical activity reflects admirable descriptive and explanatory efforts, yet these simply do not add up to an increasingly more satisfying state of knowledge accumulation. In other words, in some areas at least,
interlanguage efforts have led to heaps of studies that have nevertheless yielded
only diffused knowledge. The abundant findings do not seem to add up to any
firm closure. The study of the acquisition of articles in L2 English might be one
of these areas: Little synthetic understanding has been reached, despite the plethora of studies (readers can be the judge of this, for example, by perusing Garca
Mayo & Hawkins, 2009, and Master, 1997). The acquisition of L2 tense and aspect probably figures in this group as well, considering that ever since the watershed synthesis by Bardovi-Harlig (2000), multiple new findings have been
reported but do not seem to build on previous ones; the keen empirical activity
in this domain does not seem to push knowledge to the next needed level of either descriptive or explanatory power, or both. It is unclear why in these areas of
interlanguage and others, many additional studies continue to be published, to
no apparent avail in terms of making individual study findings cohere and become well-integrated pieces of a single puzzle. An anonymous reviewer speculated that this state of affairs may be found mostly in interlanguage areas for
which structural complexity is particularly high and where a host of interrelated
factors are at play, suggesting the need for more and robust descriptive studies,
which is precisely why so many studies continue to be published. This may well
be true, but it means that the puzzle, for some interlanguage areas at least, is one
whose pieces lie around while its meaning fails to emerge.
Finally, it is noteworthy that the fit between data and theory has been approached from many angles in interlanguage studies. Some interlanguage phenomena have been investigated from multiple theoretical perspectives, in a
theoretical polyphony that is daunting. A good illustration of this is the L2 development of relative clauses, which has been amply studied through the alternating prisms of typological universals (e.g., Gass, 1979), formal Chomskyan
linguistics (e.g., Flynn, 1989), and emergentism (e.g., Mellow, 2006), among
other theories. This is also true of the negation research to be examined next in
this chapter. But the reverse is true as well, in that a given single theory has created sets of empirical phenomena defined and understood only by reference to
that theory and practically hermetically sealed from other theoretical worlds.
For example, it is only within the world of formal linguistic theory that researchers can give meaning to phenomena in interlanguage such as verb-raising of

 Lourdes Ortega

thematic and non-thematic verbs, noun raising and concord in determiner

phrases, or interrogatives and in-situ Q-words or quantification at a distance in
complementizer phrases (e.g., Slabakova, Montrul, & Prvost, 2006). These
notions are difficult if not impossible to employ as tools for the explanation of
phenomena outside that theory. A third and seemingly rare case is when a
researcher attempts theoretical syncretism in the study of learner language development, as in Carrolls (2007) autonomous induction theory, which blends
theoretical insights from Universal Grammar and key concepts from psycholinguistic processing, or OGradys (2012) theory of L1 and L2 acquisition, which
preserves the theoretically informed descriptive apparatus of formal linguistics
but infuses it with new meanings from emergentist theory.
In sum, it is important to recognize that in the domain of interlanguage studies, as in any domain of scholarly inquiry, description and explanation can have
their own independent lives in research. Both can be useful in their own ways and
both can withstand the test of time independently from each other. With this qualification, let us now proceed to the examination of the study of English negation as
a test case that can help illustrate the complex relationship between description
and explanation in the quest to understand interlanguage.
The development of L2 negation: Stages and sequences in the 1970s and 1980s
Almost every textbook of SLA will mention an influential study of the interlanguage development of negation in English, conducted by Cancino, Rosansky, and
Schumann (1978) with six participants originally from Puerto Rico living in the
Boston area (occasionally the data for only five participants was reported, as in
Cancino, Rosansky, & Schumann, 1975). The interpretations were further elaborated theoretically by Stauble (1978) for the two upper-class pre-adolescent children, Jorge and Juan (13 and 11 years old, respectively), and by Schumann (1978)
for one of the adult participants, Alberto, a 33-year-old Puerto Rican immigrant
(see also Han, this volume). The research team concluded that their L1 Spanish-L2
English learner participants showed consistent longitudinal evidence for a fourstage developmental pattern, shown in Table 1 (cf. VanPatten, this volume). The
stages were thought to be universal, and particularly the first stage of preverbal
negation was proclaimed as the starting point of L2 acquisition of negation, since
it was found by other researchers working with English and other target languages
around the same period. For example, Hyltenstam (1977) found that beginning
learners of L2 Swedish started off with preverbal negation, even those from L1s
that, like English, have postverbal negator placement.

Chapter 8. Trying out theories on interlanguage

Table 1. Four stages in the L2 development of English negation

Timing

Stages

Description and illustration

1.

negator + X

2.

unanalyzed dont

3.

aux/modal + negator

4.

post-verbal negation with

thematic verbs

no or not is used as an anaphoric marker placed in

front of any lexical category: Not today, I no can see, I
no use television, No this week
dont is supplied in obligatory contexts at first blush,
but in fact it is used as a memorized chunk only and
thus it reflects preverbal placement still: He dont like
it, I dont can explain
Postverbal negation emerges as a productive rule, but
only in auxiliary and modal verb contexts: He is not
hungry, I will dont see you tomorrow, And she cannot
sleep, Canno rain [sic]
Postverbal negation as a productive rule spreads to
all the relevant contexts, including lexical verbs: It
doesnt spin, I didnt went to Costa Rica, Because you
didnt bring, He doesnt laugh like us

Note. The illustrations are attested learner utterances reported in Schumann (1976) and Stauble (1978).

The next decade of the 1980s brought into the center stage of interlanguage research the notion of developmental sequences, rather than stages. This was done
through the work of Meisel, Clahsen, and Pienemann (1981) and Clahsen and
Muysken (1986), briefly alluded to earlier. They described the L2 acquisition of
German word order by immigrants to Germany from Romance and Turkish L1
backgrounds, and later Pienemann elaborated on the same findings and created
his own theory of Processability, which he applied successfully across a good number of target languages (see Pienemann & Keler, 2011). Table 2 shows the proposed sequence for word order of English, including negation.
Does anything change whether one conceives of negation development facts
in terms of stages or sequences? I have argued elsewhere (Ortega, 2009, Chapter 6)
one difference is that stages suggest progress by which a present stage becomes
outgrown and left behind whenever a new, more advanced stage is entered; sequences, on the other hand, entail the cumulative addition of ever more advanced
linguistic resources, without necessarily shedding previous ones. Old stages are
overcome in favor of new ones, whereas new steps on a given sequence are added on top of old ones.
More of an epistemological difference between the negation stages of the 1970s
and the negation and word order sequences of the 1980s can be postulated, perhaps, when the respective positions regarding the relationship between accuracy
and development are more closely inspected. Cancino et al. (1978) were doctoral

 Lourdes Ortega

Table 2. Developmental sequence for English word order, including English negation,
according to Processability Theory
Timing Word order steps

Illustration

1.
2.

Single word formulae

NEG + SVO
(external, anaphoric)
Fronting (Neg + X, Do + X,
Adv + X)

How are you? Hello

No me live here

4.

Inversion (with y/n, copula)

5.

Inversion elsewhere
(dont + V)
Cancel inversion

Have you seen him?

Where is he?
Why did she eat that?
3rd sing -s
Why didnt you tell me?
I wonder what he wants to eat

3.

6.

He no stay here

Corresponding
morpheme sequence
past -ed
plural -s possessive -s
plural agreement -s

Note. Based on Pienemann (2011), Table 4.1 on p. 51.

students at Harvard University and followed the methods established just a few
years earlier for child language acquisition of monolingual English by Roger Brown
(1973), also at Harvard. They thus coded dichotomously for suppliance (or absence) of the negator not in postverbal position, the normative placement in
English. Suppliance in obligatory contexts analysis is open to the criticism that it
suffers from the comparative fallacy (Bley-Vroman, 1983) by privileging native
speaking norms. By comparison, the teams behind the new work on sequences
(Meisel, Clahsen, & Pienemann, 1981; Pienemann, Johnston, & Brindley, 1988)
were vocal on what they viewed as a methodological and epistemological innovation: Instead of calculating percentages of accurate use as supplied in obligatory
contexts, which would amount to confusing development with mastery, they
championed the notion of emergence, or first productive use of a rule; that is, in
the present case, one productive, non-formulaic utterance where the negator is
used postverbally would suffice to proclaim acquisition. Selinker (1972) predicted
that new units of analysis would be needed, and the debate surrounding suppliance in obligatory contexts versus emergence can be understood as a contribution
to their development. The usefulness of this debate notwithstanding, it must be
recognized that neither the stages nor the sequences demanded targetlike uniformity as a prerequisite for proclaiming progress in interlanguage. In actuality, both
grammatical and ungrammatical learner utterances can be classified as belonging
to the same stage or to the same step in a sequence, as the illustrations in Tables 1
and 2 clearly show. I will return at the end of the chapter to the question whether
or not this shared perspective on the optionality of non-targetlikeness is sufficient

Chapter 8. Trying out theories on interlanguage

to veer stages and sequences away from presuming a teleological view of interlanguage development with a desirable end goal.
From a historical perspective, it can be argued the developmental sequence
work represents an improvement over the stages work essentially at the level of
description. In other words, the sequences work, particularly in its much developed and theorized iteration in Processability Theory (Pienemann, 2011), offered
a framework for locating negation itself along the sequences for syntax and morphology (this is shown in the right-hand column of Table 2). Negation starts off
being external or anaphoric (at step 2) in consonance with a pragmatic strategy for
word order, it progresses to preverbal status (at step 3) when fronting occurs and
number agreement emerges, and it finally becomes postverbal (at step 5) at the
time when inversion occurs hand in hand with the emergence of third person -s.
Nevertheless, surprisingly little can be said to be strictly different in the descriptive
facts depicted in Table 2, when compared to Cancino et al.s stages in Table 1.
(Likewise, the sequence of morpheme emergence in the last column of Table 2 is
also strikingly similar to the morpheme accuracy order reported in Dulay, Burt,
and Krashen [1982] and related work.) This is perhaps because the sequences in
the 1980s and the stages in the 1970s shared a similar Piagetian orientation: Both
are inspired in the unfolding over time of psycholinguistic and linguistic abilities
from within an individuals grammar.
The variationist approach (1980s1990s)
Chomsky (1959) and Lenneberg (1967) were greatly felt influences in Selinker
(1972), and in the field of SLA, more generally, at the time. But sociolinguistic
work by Hymes (1972), Labov (1966), and others also made strong inroads into
the study of learner language in the 1980s and 1990s (Beebe, 1988). In interlanguage studies, the variationist approach inspired particularly by Labovs work
sought to disentangle variation associated with change over time from variation
associated with significantly differentiated syntactic and discourse contexts. This
perspective thus sought and used more information than the stages and sequences
approaches, by examining the linguistic and nonlinguistic environment where
forms were attested. Thus, whereas the previous approaches considered development as isolatable from context, the variationist approach claimed a systematic
role for linguistic and non-linguistic context. This theoretical claim in turn had
consequences for what to accept as a valid explanation.
While Adamson and Kova (1981) were the first to apply the variationist analysis to English negation, it is a later study by Berdan (1996) that is best known in
SLA. The study employed VARBRUL (a logistic regression software developed at

 Lourdes Ortega

the University of Pennsylvania in the mid 1970s and widely used by sociolinguists
since then). Using the raw data available in an Appendix in Schumann (1978),
Berdan re-analyzed dont + V as it entered and spread in Albertos interlanguage
production and traced the probability of it occurring vis--vis no + V across linguistic contexts and over time.
Berdan (1996, pp. 1011) noted one virtue of the interlanguage variationist
program was that clear hypotheses could be formulated and tested. Variation in
mature L1 use is characteristically contextual (with context defined across linguistic and non-linguistic parameters), which is another way to say that L1 users
possess variable rules (in Labovs [1969] sense). Conversely, no time variation is
expected, since no language learning is assumed to happen in mature grammars.
For learner language, three alternative scenarios are plausible. If evidence for neither patterned contextual differentiation nor patterned change over time were
found, Berdan reasoned, this would mean interlanguages operate in free, nonsystematic variation constrained by neither context nor time. Note that if this were
the case, sociolinguists would have to conclude not only that fossilization is unavoidable but also that interlanguages are not natural grammars (i.e., they dont
exhibit variable rules). A pattern of no contextual differentiation coupled with
change over time would indicate development that is linear. This is, in fact, the
theoretical world assumed by empirical approaches that look at development over
time in isolation from context (including the stages and sequences work reviewed
in the previous section). For Berdan and other sociolinguistic SLA researchers, on
the other hand, time- and context- changes are likely to be simultaneously at play
in interlanguage data. If so, the traditional look at time changes alone would be
insufficient, because variation related to context may well mask variation related
to time (p. 209). In this last (and most plausible) scenario, the sociolinguistic
analysis would be most useful, as it would allow for systematic, statistical modeling of the interaction between time and context, thus helping disentangle the two
sources of interlanguage variation and ultimately helping understand their patterned contribution to interlanguage development.
Berdans (1996) empirical conclusions substantiated a role for systematic differentiation in the probabilistic occurrence of dont + V both over time and across
contexts. The predictive accountability of the analyses was impressive, with 80%
of tokens produced by Alberto over 10 months (out of a total of 186 dont +V
cases plus 265 no + V cases, so k = 451) well predicted. The level of detail in the
predictions was equally impressive. On the time axis, dont + V increased on average 1% per week over 40 weeks. Across contexts, dont + V was more likely to
occur in the presence of one of five contexts, four of them linguistic and the fifth
non-linguistic:

Chapter 8. Trying out theories on interlanguage

1.
2.
3.
4.
5.

first person singular pronoun I

recent dont V use
more varied verb choice (i.e., less dependence on formulae)
modal can
(at the end of study) elicited style

Berdans (1996) findings are usually acknowledged to have overturned Schumanns

(1978) conclusion that negation had fossilized in Albertos grammar (e.g., Selinker
& Han, 2001, p. 283). The empirical prowess of the reanalysis notwithstanding,
some caveats should be voiced as well. First, as Berdan himself acknowledged
(pp. 236237), statistically the VARBRUL approach assumes linearity in time
changes. This assumption is most certainly wrong because development of any
kind is known to be nonlinear. For example, the 1% weekly increase modeled in
the longitudinal data is clearly only an idealization. A practical limitation is that
the quantitative sociolinguistic variationist approach requires large data sets,
which are not always available particularly in SLA. A theoretical limitation is that,
most typically, a group approach is based on statistical notions such as central
tendencies (i.e., averages) and dispersion (i.e., standard deviations) that is problematic for other theories discussed later, such as dynamic/complex systems theories of language development.
These caveats do not detract from the value of the variationist interlanguage
work more generally. For one, it helps find learning that is difficult to detect. In
addition, it does so while illuminating relevant sources of variation and variability
other than sheer time. Sociolinguistic inquiry into language learning has shown
continuity in SLA (Preston & Bailey, 2009; Tarone, 2007). In recent work a favorable moment of expansion has been seen in this line of interlanguage research that,
if sustained, may help advance interlanguage studies theoretically in powerful
ways. Specifically, new variationist sociolinguistic studies (e.g., Gudmestad, 2012,
for Spanish L2; Hansen Edwards, 2011, for English L2) call for a contemporary
rethinking of the scope of what is acquired by L2 learners, which they argue must
naturally include the variable rules of the target speech community. While this
point was already made by sociolinguistically oriented researchers early in the history of the field (e.g., Adamson & Regan, 1991; see discussion of Type II variation
by Bayley & Tarone, 2013), this contemporary empirical line sharpens the lens
onto the very nature of what is being learned when interlanguage develops.
Negation in universal grammar perspective since the 1990s
Formal linguistic SLA researchers have also found the study of negation a fruitful
site for the advancement of their program of inquiry since the 1990s, turning
to this phenomenon motivated by the theoretically central question of whether

 Lourdes Ortega

Universal Grammar constrains L2 acquisition in the same ways as it is thought to

constrain L1 acquisition. Indeed, negation research in formal linguistic SLA is a
good illustration of a theory-first approach to both description and explanation in
interlanguage studies. The quest for relevant data and the problematization of locus of analysis are two salient features in this tradition.
In one of the best known studies on negation in this tradition, Meisel (1997)
declared he was concerned not with surface word order but with the internal
structure of the negative element (NEG) (p. 229), which in Chomskyan theory
means a concern with basic phrase structure (layers of functional projections,
headedness of these projections, etc.), the status of the negative element (head of
the Negative Phrase or adjoined maximal projection) and finiteness (triggering
verb-raising and possibly also the movement of NEG) (p. 241). It is as a consequence of this theory-first stance that he was able to radically change the locus of
analysis, and with it essentially the nature of the phenomenon: For the first time in
his study, the finiteness of the verb was propelled to key evidentiary warrant for
indexing the development of the abstract syntactic operations involved in negation. Meisel argued his data revealed fundamental competence differences between L1 and L2 acquisition. Specifically, he interpreted his evidence as suggesting
that adults apply surface strategies to the placement of negation, and he posited a
global impairment in L2 acquisition to acquire the abstract syntactic operations of
negation in Universal Grammar. A different interpretation within the same
theoretical framework was later put forth by Hawkins (2001), who concluded L2
acquisition is reflective of selective rather than global impairment. Under this alternative interpretation of the accumulated empirical evidence, some L1-non-instantiated functional features will be permanently non-acquirable, whereas other
functional features will be acquired. An intermediate compromise was reached by
Perales, Garca Mayo, and Liceras (2009) in a study of the acquisition of negation
in English as a third language by 78 L1 Spanish-Basque bilingual children in an
English-as-a-foreign-language context. These authors interpreted their evidence
as compatible with Meisels proposal that limited sensitivity to functional features
in non-native acquisition leads adult grammars to rely on linear strategies. In addition, however, the linear strategies are guided or constrained by existing L1
knowledge and thus result in Universal Grammar-sanctioned interlanguage solutions in many cases, as Hawkins would predict.
A conundrum in the theory-first approach seen in formal linguistic SLA work
on interlanguage is that, while a given description may be undisputed, it may give
rise to competing explanations, such as the ones just recounted for the three negation studies reviewed, all within the same theoretical world. And while there is
continuity in the key questions initially posed by formal linguistic SLA researchers
about interlanguage grammar, it always seems possible to make the claim that the

Chapter 8. Trying out theories on interlanguage

data specifically relevant to support a given theoretical explanation have yet to be

found. Indeed, the relevant interlanguage data have progressively been sought in
increasingly new places in this line of work: in the grammar knowledge exhibited
at the initial state or the final state, in Poverty-of-the-Stimulus phenomena
(see VanPatten, this volume), and in linguistic interfaces. For all these themes negation has nevertheless proven to be a fruitful interlanguage area to investigate.
In a more recent study, Grter, Lieberman, and Gualmini (2010) investigated
a subtle semantic-syntax interface phenomenon: the scope of negation. An utterance like They dont speak English or German can mean, depending on the language, that they speak neither English nor German, or it may mean: they dont
speak English... or is it maybe German that they dont speak? The latter either or,
cant remember interpretation, which represents a wide scope of negation and is
a superset or a weak interpretation, is preferred by mature (L1) users of Japanese.
On the other hand, the neither interpretation, which represents a narrow scope
of negation and a subset or a strong interpretation, is preferred by mature (L1) users of English. Grter et al. note the narrow scope of interpretation is also known
to be the default in child language acquisition, as monolingual children of a number of target languages have been found to favor the neither interpretation. The
initial default, moreover, is observed in L1 acquisition data up until around age
five. This means the phenomenon of scope of negation develops slowly and its
learning is protracted in L1 and thus likely to be at work in L2 acquisition at up to
intermediate levels of proficiency (p. 144).
What would it mean to find evidence from L1 English L2 Japanese lowproficiency learners showing a preference for the narrow scope of negation (the
neither interpretation) and a concomitant difficulty in acquiring the Japanese
wider scope? While it is important to empirically ascertain the pattern, the researchers reasoned theoretically it would be impossible to explain the observation.
Namely, this same descriptive fact could mean (a) that L2 learners start off with the
same acquisitional default as all monolingual L1 children do, or (b) that L2 learners start off with the L1 setting as their default. However, evidence collected from
L1 Japanese-L2 English learners, and therefore the reverse learning direction,
would render theoretically unequivocal explanations. If low-proficiency L1
Japanese learners of L2 English had no trouble accepting the narrow scope neither interpretation, it would mean the Universal Grammar default still constrains
adult language learning, supporting the position that L1 and L2 acquisition are
fundamentally similar. If on the other hand they exhibited trouble with the narrow
scope neither interpretation, it would mean that the L1-set values are the starting
point of learner grammars, supporting the position of Hawkins (2001) and others.
Grter et al. found evidence for L1 effects, which would support the theoretical

 Lourdes Ortega

position that Universal Grammar constrains L2 acquisition but only indirectly, via
the already acquired knowledge of L1.
I see two key differences that cast doubt on whether, ultimately, formal linguistic investigations of L2 development can be considered to be part of the same interlanguage research tradition that Selinker (1972) laid the ground for or might
need to be considered a distinct research domain (see, however, Montrul, this volume). One difference has to do with what counts as relevant data. Namely, the
properties of formal grammars are so very abstract that experimental data are usually needed, and much of the elicitation in studies of Universal Grammar in an L2
can hardly be considered to target the kinds of data that Selinker (1972) thought
ideal for the study of interlanguage: behavioral events revealing of attempted
meaningful performance (p. 210). The other difference has to do with the overarching goals of inquiry in the two traditions and is intricately connected to theoretical choices. The formal linguistic study of L2 development puts theory first and
is driven by the quest to understand the role that Universal Grammar or abstract
linguistic knowledge plays in the acquisition of human language across the life
span. Interlanguage researchers do not necessarily adhere to such theoretical commitment and instead, as this chapter shows, can take a number of approaches, all
the way from the pre-theoretical (e.g., in the stages work), to the formal but nongenerative (e.g., in Pienemanns 2011 Processability theory), to several functional
linguistic approaches (e.g., in Berdans 1996 reanalysis of Schumanns 1978 data;
and see later sections for several other functional approaches). This theoretical
choice difference also carries over into an important question: Are interlanguages
natural linguistic systems, or are they wild/rogue systems? It seems to me Selinker
(1972), and with him probably most interlanguage researchers, hoped to glean
evidence in favor of the revolutionary proposal that interlanguages are natural
grammars. Within the formal linguistic SLA perspective, however, answers vary
and at least some believe the jury is still out (see Meisel, 2011).
Also during the 1990s: Italian negation in the basic variety
Research on interlanguage has also drawn from functional linguistic theories, as
evident in the well-known work by Klein and Perdue (1997; Perdue, 1993a &
1993b). In this tradition, learning a language is learning how to mean via expanding information structuring resources, initially constrained by discourse-pragmatic focus-topic strategies and aided increasingly by more grammaticalized resources. Klein and Perdue coined the term Basic Variety to refer to the profiled
starting point of linguistic competence that was typical of the immigrant learners

Chapter 8. Trying out theories on interlanguage

they studied, characterized along the lines of phrasal, semantic, and pragmatic
functional principles (see also Bardovi-Harlig, this volume).
An interesting and unfairly forgotten study by Bernini (2000) recruited the
theoretical and analytical insights of the Basic Variety in order to both describe
and explain the acquisition of negation in L2 Italian. Bernini charted the same
phrasal, semantic, and pragmatic functional principles presented by Klein and
Perdue (1997) in a longitudinal corpus of 5 partially untutored L2 Italian learners.
Diffusion was the central construct in Berninis developmental description of negation: Negation is shown to progress in a cumulative process across four stages,
whereby the adjunction of new structures results in the quantitative reduction of
previous structures in terms of frequency of use and in the qualitative adjustment
of their functional range, until the target system is reached and negation in native
and nonnative varieties is indistinguishable but for minor details (p. 400). In
principle, this description of the interlanguage development of negation is highly
compatible with the stage, sequence, and variationist approaches discussed earlier, and this is true at the level of descriptive findings as well as developmental
conceptualizations: cumulative emergence, along increasing and decreasing frequency patterns, with diffusion of form-function distributions along increasingly
more varied contexts for negation, progressively spanning a wider functional
range. These functional-linguistic conceptualizations, however, subtly move explanations closer to more contemporary (post-1990s) theories to be reviewed in
the next section.
Table 3. Diffusion of structural types to express negation in L2 Italian found
by Bernini (2000)
Timing Structural types
1.
2.

no
no no(n) + X
X+ no

3.

no no/non + V
no(n) + X X + no

4.

no non + V
non + X X + no
non + V + mica

Illustration

Learner Variety
Pre-BV
Pre-BV

come soldi no [like money no = as for the

money, I didnt] natro no, solo questo [sic]
[other no, only this]
tembo una settimana io no abito no qua [time, Pre-BV &
one week, I do not live not here = I havent
post-BV
lived here for a week now] anche, il bambini,
non c la forsa [also, the children, there isnt
the strength = the children are not strong]
non ci ho mica pensato in tedesco [no I have
post-BV
thought in German = I wasnt thinking of it
in German]

Note. BV = Basic Variety as described in Klein & Perdue (1997). Bolding is used to indicate new
structural types where the negators no and non emerge over time, constituting a diffusion pattern.

 Lourdes Ortega

As shown in Table 3, at first the only resource to express negation that Bernini
found in his data is a holophrastic no, which stands alone as an anaphoric response
negating some entity or utterance in previous discourse. Next in time, an additional resource emerges: no or non pre- or postposed to negated elements, still at a
pragmatic-discourse level. If preposed, no(n) + X, the strategy is one of scope
marking. If postposed, X + no, the information structure is marked pragmatically by intonation, in what is a strategy to handle topicalization in the topic-focus
structure of utterances. At time 3, diffusion is observed of non, which now is preposed to verbal material, no(n) + V (this is the targetlike rendition in Italian, which
is a language with preverbal negation). This signals the first time verbal utterances
are negated, and in such cases the (preverbal) negator has now the status of a
grammatical constituent rather than just a pragmatic marker. This expansion towards constituency is observed while the learner develops more generally from
the pre-Basic Variety (no morphology, basic syntax) to the Post-Basic Variety
(emergence of at least some morphology and some subordination). In what can be
considered a functionalist recast of Berdans (1996) variationist logic, Bernini argues the process of diffusion of the preverbal negator non to increasingly more
verbal contexts is favored by the existential predicate (non c, there isnt) in what
he interprets as a syntactic trigger, and the copula (non , it isnt), which he takes
to be the result of a bias towards phonotactic preservation. Echoing the additive
spirit of the sequences work is Berninis observation that the old resources attested
at any given point remain in the learner grammar (this is shown through bolding
and unbolding in Table 3), although over time they show a decrease in frequency
and an increase in functional restriction. For example, X + no is new at time 2
and becomes specialized in time 3 as a contrastive negation strategy only. At the
very end of the development of negation investigated, at time 4, a discontinuous,
pragmatically marked emphatic negator mica is also attested: non + V + mica.
In sum, Bernini (2000) describes the development of negation as a process of
gradual refinement of unfolding, cumulative, increasingly functionally differentiated communicative repertoires. While the functional details are better fleshed
out than in previous studies, the descriptive facts of negation seen in stages, sequences, and variationist work are preserved in the Basic Variety analysis. One
difference is worthy of note, however: The end point of development captured in
the functionalist description is one isomorphic with negation resources in native
Italian: in the area of negation [in Italian], learners and native Italian speakers
share the same array of constructions serving the same range of functions
(p. 430). On this count, the interpretation of the functionalist approach clashes
directly with some of the interpretations of the formal linguistic approach reviewed in the previous section, which posit a fundamental impossibility for L2
grammars to converge with L1 grammars.

Chapter 8. Trying out theories on interlanguage

Onto the future? Usage-based approaches to English negation

This final historical stop brings us to the present time, as I review three different
contemporary investigations of L2 negation that share a general usage-based
stance. They each offer an application to the description of negation of a particular
usage-based theoretical approach: dynamic systems theory, constructionist grammar, and conversation analysis for SLA.
Employing the conceptual and analytical tools of dynamic systems theory
(DST), Verspoor and colleagues (van Dijk, Verspoor, & Lowie, 2011; Verspoor,
Lowie, & van Dijk, 2008) reanalyzed the negation data from Cancino et al. (1978).
They inspected the data individually and considered age differences for the two
five-year old learners, the two pre-adolescent children, and the two adults in the
original study. Their goal was to demonstrate how DST may afford new insights
into interlanguage that previous approaches cannot reveal. At the theoretical level,
DST posits variation is a precursor and a driving force of development. At the
empirical level, this translates into accounting for any instances in the data of enlarged variability in the vicinity of developmental jumps (Verspoor et al., p. 222).
Indeed, the data revealed great intra-learner, system-internal, self-organizing variability, supporting the theoretical tenet that development is self-adaptive and, in
sum, as the theory posits, dynamic. More specifically, Verspoor and her colleagues
argued that the four so-called stages (see Table 1) lost any stage-like quality once
the evidence was inspected separately for each learner. In addition, the inspection
of the individual data uncovered different developmental paces for different ages:
the two five-year olds were slow, the two pre-adolescents were fast, and by comparison the two adults exhibited very little change. This was inferred based on the
DST variability peaks in non-targetlike forms, which were considered pre-
announcements of restructuring or development. Namely, the data for the two
older learners (Alberto and Dolores) showed fewer peaks than the adolescents or
the five-year olds. For the two pre-adolescent learners only (Juan and Jorge), no +
V disappeared after month three, following a period of greatest fluctuation, and as
analyzed do + neg began to consolidate; for the remaining seven months of observation, (non-linear) moderately upward development was seen in both learner
profiles of negation. Verspoor and her colleagues granted some commonalities
that did seem to occur across the six data sets but argued that these were general
in nature. For example, the dont construction was overused by all learners to some
extent. They further maintained that the shared common patterns do not support
the idea of stages or sequences in and of themselves. For example, all of the six
learners were observed to start off with tokens from all four stages (Table 1) in
highly variable frequency. In sum, once disaggregated at the individual level, there
were no real stage cut-offs to be seen in any of the individual data over time.

 Lourdes Ortega

From the reanalyses and insights reported by van Dijk, Verspoor, and Lowie
(2011) and Verspoor, Lowie, and van Dijk (2008), the main lesson DST asks interlanguage researchers to learn is that the relevant data and the relevant units of
analysis cannot be found in averages, and instead individually disaggregated evidence is necessary. Other benefits stem from the new theoretical and analytical
constructs introduced by this contemporary perspective. They represent nearly
untested innovation in interlanguage studies: DST researchers describe fluctuations, progress-regress, acceleration-deceleration within general upward trends,
highly variable peaks and wells, competitive or mutually supportive relations.
Much of the same epistemological staples discussed here for DST are echoed in the
complexity theory perspective espoused by Larsen-Freeman (2011; this volume).
The second application to the description of negation from a usage-based perspective that I will discuss comes constructionist grammar (e.g., N. Ellis et al.,
2013). Constructionism has been applied to L2 English negation recently by
Eskildsen (2012), who conducted a longitudinal analysis of two Spanish L1 adult
learners. Their histories of English usage were captured in a rich classroom corpus
(the Multimedia Adult English Learner Corpus, Reder et al., 2003) spanning two
years for one learner and 3.5 for the other. Following the cognitive linguistic and
acquisition theory of constructionism, Eskildsen set out to map construction-
specific inventories of L2 linguistic resources for doing the work of negating in
English interlanguage.
The development of construction-based grammars, in L2 as in L1, calls for the
investigation of co-existing degrees of specificity and schematicity in families of
learner utterances, along a specificity-schematicity cline ranging from formulas
(strictly item-based) to low-scope patterns (less schematic) to schemas (increasingly more abstract). Schematicity is predicted to increase over a history of sufficient usage simply by virtue of analogy and abstraction processes supported by
general basic mechanisms of human cognition. Eskildsen (2012) concludes that
this continuum of schematicity (i.e., the exemplar-based developmental trajectory of negation) found in his data can be best described as follows. From the very
beginning, there were concrete instantiations of negative constructions, called
fixed recurring multi-word expressions (MWEs) (e.g., I dunno). With more use in
natural communication in this classroom context, partially schematic, partially
concrete patterns called utterance schemas or item-based patterns were also
additionally produced (e.g., I dont Verb). With more use yet, increasingly generalizable schematic constructions arose from systematic commonalities among patterns (e.g. NP do NEG VERB). The inventory of negation constructions generally
increased cumulatively, eventually leading to redundant, multiple levels of schematicity for negation coexisting psycholinguistically. In other words, all three

Chapter 8. Trying out theories on interlanguage

kinds of constructions were attested and coexisted by the end of the longitudinal
observation period.
At first blush, the notion of schematicity may appear similar to the notion of
formulaicity vs. analysis in stages work (Cancino et al., 1978). For example, it
appears similar to the observation that dont is initially an un-analyzed negator
(at stage 2, see Table 1) and only later exhibits the behavior of abstract preverbal
grammatical negation (and only gradually, over stages 3 and 4, cf. Table 1). Why,
then, cant we simply assume that the new constructionist parlance the construction inventory evolves along increasing degrees of schematicity is the same traditional description, just differently put, that says from stage 2 to stages 3 and 4 a
change happens from a highly specific knowledge of dont to a more abstract
knowledge of dont? What stops us from equating previous interlanguage approaches and usage-based constructionist developmental theories of interlanguage? The differences are subtle but important.
One difference goes to the very ontology of what is meant by development.
Namely, the traditional degrees of analysis perspective rests on the metaphor of
abstract rules as autonomous knowledge entities that spread in application over
increasingly more complex utterances until each rule reaches maturity, that is, until conformity is reached with the target rule. The issue of degree of schematicity
in developmental construction inventories sets up a very different metaphor, one
in which rules have no ontological and hence no explanatory status: what the
linguist or the analyst calls negation does not seem to be learned as a rule-governed syntactic phenomenon to be deployed across diverse linguistic patterns in a
broad-sweeping manner, but seems to emerge in different patterns in different
ways at different points in time along, rather than across, constructional lines
(Eskildsen 2012, p. 365) (cf. VanPatten, this volume, on the impossibility of rulebased learning).
In constructionist views of language development, there is no traversing of
stages (Cancino et al., 1978), no adding of steps in a sequence (Meisel et al., 1981),
and no accretion of processing strategies (Pienemann, 2011). There is no deployment of rules along constraining contexts of variation (Berdan, 1996) and no gradual diffusion over functionally increasingly grammaticalized meanings (Bernini,
2000). There is no setting or resetting of abstract knowledge about structure dependencies in syntactic hierarchies (Meisel, 1997) or at syntactic-semantic interfaces (Grter et al., 2010). Usage-based interlanguage researchers are left alone
with construction exemplars as physical instances of language which in themselves
(inseparable from the users and the uses) evolve over iterative meaningful use
events into evidence of qualitatively different knowledge. All experienced exemplars have a psycholinguistically similar status. They are all stored cognitive
routines, as much as grammar as a whole is the cognitive organization of ones

 Lourdes Ortega

experience with language (Bybee, 2010, p. 8). Eskildsen (2012) reminds readers
that any observable degree of specificity and schematicity is, analytically speaking,
equally interesting; all data, whether formulaic or creative, are equally relevant because whether memorized (un-analyzed) or schematized, all exemplars can
spark learning and must be traced longitudinally. In fact, the degree to which utterances are either exemplar-dependent and hence highly specific, or productive
and hence highly schematized is graded and can (must!) be measured in the data
by inspecting a learners attested inventory of constructions. In sum, in the
constructionist usage-based approach to interlanguage, there is no theoretically
tenable sense of teleology, that is, development proceeding to some desired or naturally posited end that researchers can expect (cf Larsen-Freeman, this volume).
A unique and unprecedented contribution of Eskildsen (2012) is the combination of usage-based constructionism with a conversation analytical inspection of
the very usage events in which negation occurs and of the socially distributed affordances that learners seize from their environment. For Eskildsen, the necessity
of investigating interactional contingencies of emergent L2 constructions (p. 363)
and thus the theoretical blend with conversation analysis falls naturally from heeding the tenets of usage-based linguistic development, because however abstract
constructional knowledge may ultimately become, it derives invariably from specific occasions of use (p. 367). In the conversation analytic part of his study,
Eskildsen is able to describe how all occasions for the use of negatives were afforded via interactions with the teacher or with other learners. It is rewarding to
see that interaction continues to be important in SLA and in the development of
interlanguage, after all these years. But of course, its importance has very different
epistemological content with conversation analysis (Kasper & Wagner, 2011).
The third and final usage-based study of negation to be reviewed in this section is Hauser (2013), who argues for the value of conversation analysis as a window into interlanguage development. As much if not most of conversation analysis
in SLA, the study takes a stance that is explicitly agnostic toward theory, reminiscent of Selinkers (1972) description before explanation dictum. Hauser studied a
Japanese adult who over a seven month period often used no(t) + X negation, and
less frequently X + no(t) negation, but both regularly and without any particular
noticeable changes over time. Upon closer conversation analytical inspection,
however, the data also offered two formulas, I dont know and I cant speak English,
only the first of which happened to undergo schematization, in the same sense as
Eskildsens (2012), from I dont know to I dont like and later from I dont to You
dont. Carefully documenting the interactional contexts in which these increasingly more abstract uses of the I dont know construction arose, Hauser showed
that development along this attested route unfolded in the moment by moment
opportunities to repeat negations offered by the interlocutor and often in the

Chapter 8. Trying out theories on interlanguage

structured, sequential actions of self-repair initiated by the interlocutor or by the

learner himself. Hausers conclusion was that Conversation Analysis, when applied to longitudinal data, can draw our attention to small areas of development
which may often be obscured against a background of stability (p. 31).
In a nutshell, the three usage-based studies of negation reviewed in this final
section exemplify a novel conceptualization of interlanguage development. Each
study represents a different usage-based approach and therefore encourages the
inspection of different evidence as theoretically relevant. System-internal selfadaptive variability is a main concern of the DST approach; cognitive internal developmental processes of schematization by abstraction are a central interest to
constructionism, and affordances of interactional social practices are at the heart
of conversation analysis for SLA. The foci are, however, not incompatible and can
instead complement one another, whether through theoretical blending (as in
Eskildsens 2012 program) or by synthetic incorporation of findings. The new usage-based habitus for doing interlanguage research shares broad ontological and
epistemological premises, as all three approaches assume language acquisition
happens in and through use, and thus the study of development cannot be separated from the local histories of practices and events in which language occurs and
which an individual learner experiences. Individual grammar emerges from actual, individual histories of usage, and histories of usage are by definition constituted by social and physical worlds at once.
Conclusion
This chapter began with two questions: What main theories have been tried on by
interlanguage researchers in four decades, since 1972? And what might one learn
about the virtues and limits of description prior to explanation with the hindsight
of 40 years of research history? The past 40 years of research into the L2 development of negation clearly show that interlanguage has been investigated from multiple theoretical perspectives, and the resulting polyphony of descriptions and
explanations of what in most cases can simply be said to be the same interlanguage
phenomenon is daunting. Interlanguage researchers have modeled the acquisition
of L2 negation by imagining the learners task quite differently as: traversing negation stages (Cancino et al., 1978); adding negation strategies in a sequence (Meisel
et al., 1981); continuously attuning usage sensitivities to probabilistic micro-
contextual influences (Berdan, 1996); learning (or not learning!) about structure
dependencies of NEG in syntactic hierarchies (Meisel, 1997) or in subtle syntacticsemantic interfaces (Grter et al., 2010); learning how to mean via expanding
information structuring resources and towards a widening of their functional

 Lourdes Ortega

range with increasingly more grammaticalized resources (Bernini, 2000); dynamically experiencing fluctuations, acceleration and deceleration, peaks and wells, as
they self-adapt resources for language use (Verspoor et al., 2008); building inventories of minimally and maximally schematized utterances experienced and
abstracted out of the contingencies of communication (Eskildsen, 2012); doing
interaction and doing learning as affordances constituted and structured by social
practices (Hauser, 2013).
Have we arrived, as Selinker (1972) urged us, to a clear idea of what is in [interlanguage], even if we cannot explain why it is there (p. 204)? For negation, at
least, there seems to be reasonable agreement in the accumulated descriptions,
and many of the facts stand scrutiny across theoretical lenses. For example, all
studies find that adult L2 learners of English start off by preposing the negator and
that this early preference may eventually fade out of their interlanguage system.
Studies looking simultaneously at other developmental milestones, whether they
highlight individual differences or shared patterns, suggest that certain grammatical phenomena lead the way to postposition, for example, some morphology at
least has begun to emerge (in the Basic Variety framework) or finiteness has set in
(in the formal linguistic framework). Studies that otherwise represent diverse theoretical lenses report some of the same finer descriptive findings. For example,
Hauser (2013) explicitly notes the incidence in his data for a use of the X + no(t)
construction for the focus-topic contrastive function reported by Bernini (2000),
and his data also support the first three of five factors found by Berdan (1996) as
favorable for the gradual diffusion of dont + V use (first person singular I, recent
dont V use, and more varied verb choice or less dependence on formulae). The
attention to description over 40 years is paying off.
How do different theoretical lenses affect description, given that each theory
probes the description of allegedly the same set of phenomena (i.e., negation) from
the vantage point of seeking explanations that are faithful to different interlanguage world views? By applying different theories, some findings appear to change
only in the details and yet they seem to bring different interlanguage truths to
the fore for consideration. This is the case of Berdans (1996) variationist re-interpretation of Albertos data, which overturned Schumanns (1978) conclusion about
fossilization by demonstrating small but statistically significant and contextually
systematic changes over time in the data. This is also the case of Verspoor et al.
(2008; van Dijk et al., 2011), whose disaggregated individual analyses uncovered
age effects and much less sharp boundaries along time changes than proclaimed
by traditional analyses. The application of a different theoretical lens also
occasionally leads to submitting a new descriptive fact for consideration. Both
Eskildsen (2012) and Hauser (2013) report not finding any evidence that the
modal can affects variable use of dont +V, contra what was reported initially by

Chapter 8. Trying out theories on interlanguage

Schumann (1978) and later by Berdan (1996) (cf. Table 1, stage 3). They both interpret this as evidence that, in consonance with what is postulated by usage-based
approaches to language development, different constructions and formulae entail
different affordances for different learners, and that this will depend on the contexts of usage and the personal and local histories of users. In sum, each new theory applied to the same interlanguage phenomena contributes theory-specific
analytical tools and theory-specific explanations that in turn can advance description by triangulation and falsification.
Looking forward
At the close of so many studies of interlanguage negation many more questions
open up as challenges for the future of interlanguage studies. I leave readers with
three pending issues worthy of future investment by interlanguage researchers.
The first issue concerns the study of variation in interlanguage, a construct
that has been and will continue to be a Trojan horse in interlanguage. While vibrant interest in interlanguage variation was witnessed in early years in the controversial notion of free or random variation in SLA (e.g., see R. Ellis, 1985; Tarone,
1979), empirical attention to questions of variation has not been sustained, as it
has depended on the theory driving each interlanguage agenda.
At the one extreme, variation has been eschewed in interlanguage approaches
that seek universal explanations of some kind, including general cognitive explanations (in stages and sequences work), functional explanations (in the Basic
Variety work), or formal linguistic explanations (in Universal Grammar work).
Somewhere in between, variation, albeit more in the sense of intra- and interlearner variability, rather than systematic variation, has been moderately accommodated in constructionist work by means of the notion of input variability, which
is seen as predicting patterns of developmental variability (N. Ellis et al., 2013;
see also Han, this volume). It is also indirectly accommodated in conversation
analysiss maximum to stay true to the local details of unfolding interactions
(Hauser, 2013; also Eskildsen, 2012).
At the opposite extreme, two approaches to interlanguage center around variation phenomena per se: the sociolinguistic variationist approach (e.g., Berdan,
1996) and the DST approach (e.g., Verspoor et al., 2008). Yet, there are key differences between them. The variationist reanalysis of English negation (Berdan,
1996) showed that any analysis that aspires to account for time variation alone
runs the risk of misguidedly pronouncing intractable, random looking variation
in the data, thereby missing evidence of learning. Context-related sources of systematic variation associated with linguistic and non-linguistic forces must therefore be accounted for when describing interlanguage. The study of this variation is

 Lourdes Ortega

linguistic or nonlinguistic but external to the system, quantitative, and often captured and modeled at the group level via statistical multivariate techniques such as
VARBRUL and newer Bayesian approaches (Gudmestad, House, & Geeslin, 2013).
The DST reanalysis of the same data (van Dijk et al., 2011; Verspoor et al., 2008)
showed that individual learning trajectories are obscured and precursors of development in them are missed if variability is ignored. The study of this variability is
system-internal, qualitative, and only observable at the individual learner level
over time. In the end, what has been clearly achieved with variationists and DST
researchers directly tackling theory-based explanations of the significance of variability/variation suggests the construct is all-important for explanations of
interlanguage. But there is great potential for a debilitating dispersion of efforts at
grappling with variation and variability in interlanguage as a whole, if the crosstheoretical explanatory tensions persist unexamined.
Furthermore, a theoretical-analytical element is still missing from the notions of variation examined in regards to studies of negation, namely, a broader
contextual sociolinguistic layer of the kind proposed by Tarone and Liu (1995),
one which would involve variation stemming from social interactions from across
a wide range of social roles, identity positionings, agency, and affective demands
for performativity. Ramptons work on language crossing (1995) and agentive
practices of stylization of self through language (2008) is also relevant here, as an
anonymous reviewer rightly noted. For example, in the 1995 study he documents
how his Pakistani youth participants in London playfully alternated canonical
English negation with preverbal negation (me no + verb) in interactions with
their Anglo teacher as a way to do identity work and resistance. The missing social
layer of language usage events and its attendant complexity is important in that
social forces can activate internal acquisition processes differentially because, as
Tarone and Liu showed, they also make differential linguistic demands (i.e., they
change usage).
In sum, in the future it may be advantageous for interlanguage researchers not
only to clarify theoretically relevant definitions of variation and variability, but
also to redress notions that have been underutilized in research and to acknowledge a certain complementarity in the available positions towards the issue. This
may inspire some to incorporate time-, context-, system-inherent, usage-attested,
and social identity variation in future interlanguage descriptions.
A second pending area is the surprising neglect in the study of crosslinguistic
influences vis--vis universal patterns in interlanguage. Selinker (1972) himself
was certainly interested in issues of L1 influence and illustrated them at length in
his seminal article. Nevertheless, most interlanguage studies have emphasized the
universal to the point that L1-related patterns are pushed to the background. This
is true even for studies that were carefully designed to elicit comparative evidence

Chapter 8. Trying out theories on interlanguage

across diverse L1 backgrounds, such as in the ingenious design devised by Klein

and Perdue (1997). Indeed, few studies have tackled this topic (but see Luk &
Shirai, 2009, and Odlin, this volume). Gauging the relative contributions and
weight of L1 knowledge vis--vis universal forces in development might have enjoyed among interlanguage researchers a lesser sense of descriptive urgency than
other areas, but its theoretical importance is clear, and this area is now ripe for
much more sustained attention in the future.
Third and last, in the future interlanguage studies will need to address a persistent ambivalence regarding the role of accuracy in development. The ambivalence is felt in the contemporary reading of Selinker (1972) and it lingers in the
studies reviewed as well as in the research of the past 40 years. On the one hand,
most interlanguage approaches make use of all data in their analyses, treating all
target-like and non-target-like data as useful windows into interlanguage and, in
many cases, as evidence for learning. On the other hand, each new stage, each new
sequence, even each new degree of schematicity or each new turn in which the
interaction supports the use of negation is proclaimed to be developmentally more
advanced or more propitious for learning only because each gets the learner
grammar closer to the right (i.e., nativelike) solution in the given target language: postverbal negator placement in the case of English. On the one hand,
Selinker (1972) told us, interlanguage must be studied as a system in its own right
via new constructs and new analytical units. And, as this chapter has shown, the
field has indeed generated a plethora of new constructs and analytical tools that
have enormously enriched the study of interlanguage. On the other hand, side by
side, researchers continue to judge ultimate adult language learning success by
unqualified reference to native speaker competence. On the one hand, mastery
and accuracy must be purposefully disentangled from development in interlanguage, which is said to constitute a natural linguistic system worthy of study in its
own right. On the other, mastery and accuracy, that is, teleological arriving to isomorphic conformity with idealized native speaker norms, is the only thinkable
way of defining (linguistic and developmental) success. These contradictions are
even more perplexing when the other central element in the equation is also considered: learners. Learners bring their own and varied goals, and their own and
diverse interpretations of success and failure to the learning task, reproducing, coconstructing, resisting, and/or subverting the goals and definitions from researchers, teachers, and other communities surrounding them. Larsen-Freeman (this
volume) offers her own important take on this ambivalence and advances some
possible solutions. I concur with her enthusiastically and have faith that complexity theory, as she proposes, together with other usage-based approaches, such as
the ones that have so fruitfully been applied to the study of negation, will help resolve this vexing ambivalence (Ortega, 2013). Whatever the solutions offered, it is

 Lourdes Ortega

clear that contemporary SLA must grapple with what is now, with the benefit of
historical hindsight, a contradiction in interlanguage studies as a project.
It is with a sense of excitement for the future of interlanguage studies that I
hope many others will re-read Selinker (1972) before or after reading this book.
They will, I am certain, re-live the exhilarating feeling and novelty of the concept
of interlanguage and the disciplinary invitation it initiated in SLA. The seminal
article urged the field to envision what adult learners do in building a new language as the building of a natural language a second time around. I have come to
see the process of interlanguage development as one of learning to become bilingual or multilingual later in life. This is indeed a process that offers hidden systematicities and mysterious successes to be unlocked by researchers. It deserves rigorous investigation and, indeed, its own thriving field of study following in
Selinkers footsteps.
References
Adamson, H.D., & Kovac, C. (1981). Variation theory and second language acquisition: An
analysis of Schumanns data. In D. Sankoff & H. Cedergren (Eds.), Variation Omnibus
(pp. 403410). Edmonton, Alberta: Linguistic Research.
Adamson, H.D., & Regan, V.M. (1991). The acquisition of community speech norms by Asian
immigrants learning English as a second language: A preliminary study. Studies in Second
Language Acquisition, 13, 122.
Atkinson, D. (Ed.). (2011). Alternative Approaches in Second Language Acquisition. New York,
NY: Routledge.
Bardovi-Harlig, K. (2000). Tense and Aspect in Second Language Acquisition: Form, Meaning,
and Use. Language Learning Monograph Series, 50(Suppl. 1). Malden, MA: Blackwell.
Bayley, R., & Tarone, E. (2013). Variationist approaches to SLA. In P. Robinson (Ed.), The Routledge Encyclopedia of Second Language Acquisition (pp. 679682). New York, NY: Routledge.
Beebe, L.M. (1988). Five sociolinguistic approaches to second language acquisition. In L.M. Beebe
(Ed.), Issues in Second Language Acquisition (pp. 4375). New York,NY: Newbury House.
Berdan, R. (1996). Disentangling language acquisition from language variation. In R. Bayley
& D.R. Preston (Eds.), Second Language Acquisition and Linguistic Variation (pp. 203243).
Amsterdam: John Benjamins.
Bernini, G. (2000). Negative items and negation strategies in nonnative Italian. Studies in Second
Language Acquisition, 22, 399440.
Bley-Vroman, R. (1983). The comparative fallacy in interlanguage studies: the case of systematicity. Language Learning, 33, 117.
Brown, R. (1973). A First language: The Early Stages. Cambridge, MA: Harvard University Press.
Bybee, J. (2010). Language, Usage and Cognition. Cambridge: Cambridge University Press.
Byrnes, H., Maxim, H.H., & Norris, J.M. (2010). Realizing Advanced Foreign Language Writing
Development in Collegiate Education: Curricular Design, Pedagogy, Assessment. Modern
Language Journal, 94(Supplement, issue 5).

Chapter 8. Trying out theories on interlanguage

Cancino, H., Rosansky, E. J., & Schumann, J. H. (1975). The acquisition of the English auxiliary
by native Spanish speakers. TESOL Quarterly, 9, 421430.
Cancino, H., Rosansky, E., & Schumann, J. H. (1978). The acquisition of English negatives and
interrogatives by native Spanish speakers In E. Hatch (Ed.), Second Language Acquisition: A
Book of Readings (pp. 207230). Rowley, MA: Newbury House.
Carroll, S. E. (2007). Autonomous induction theory. In B. VanPatten & J. Williams (Eds.),
Theories in Second Language Acquisition: An Introduction (pp. 155173). Mahwah, NJ:
Lawrence Erlbaum Associates.
Chomsky, N. (1959). [Reviews] Verbal behavior by B.F. Skinner. Language, 35(1), 2658.
Clahsen, H., & Muysken, P. (1986). The availability of universal grammar to adult and child
learners: A study of the acquisition of German word order. Second Language Research, 2,
93119.
Dulay, H., Burt, M., & Krashen, S. D. (1982). Language Two. Oxford: OUP.
Ellis, N.C. (1998). Emergentism, connectionism and language learning. Language Learning, 48,
631664.
Ellis, N.C., ODonnell, M.B., & Rmer, U. (2013). Usage-based language: Investigating the latent
structures that underpin acquisition. Language Learning, 63(S1), 2551.
Ellis, R. (1985). A variable competence model of second language acquisition. International
Review of Applied Linguistics, 23, 4759.
Ellis, R. (1989). Are classroom and naturalistic acquisition the same? A study of the classroom acquisition of German word order rules. Studies in Second Language Acquisition, 11, 305328.
Eskildsen, S. (2012). L2 negation constructions at work. Language Learning, 62, 335372.
Flynn, S. (1989). Spanish, Japanese and Chinese speakers acquisition of English relative clauses:
New evidence for the head-direction parameter. In K. Hyltenstam & L. Obler (Eds.),
Bilingualism across the Lifespan (pp. 116131). Cambridge: CUP.
Garca Mayo, M. del P., & Hawkins, R. (Eds.). (2009). Second Language Acquisition of Articles:
Empirical Findings and Theoretical Implications. Amsterdam: John Benjamins.
Gass, S. M. (1979). Language transfer and universal grammatical relations. Language Learning,
29, 327345.
Goldschneider, J., & DeKeyser, R.M. (2001). Explaining the natural order of L2 morpheme acquisition in English: A meta-analysis of multiple determinants. Language Learning, 51,150.
Grter, T., Lieberman, M., & Gualmini, A. (2010). Acquiring the scope of disjunction and negation in L2: A bidirectional study of learners of Japanese and English. Language Acquisition,
17, 127154.
Gudmestad, A. (2012). Acquiring a variable structure: An interlanguage analysis of second language mood use in Spanish. Language Learning, 62, 373402.
Gudmestad, A., House, L., & Geeslin, K.L. (2013). What a Bayesian analysis can do for SLA: New
tools for the sociolinguistic study of subject expression in L2 Spanish. Language Learning,
63, 371399.
Hansen Edwards, J.G. (2011). Deletion of /t, d/ and the acquisition of linguistic variation by
second language learners of English. Language Learning, 61, 12561301.
Hauser, E. (2013). Stability and change in one adults second language English negation. Language Learning, 63, 463498.
Hawkins, R. (2001). Second Language Syntax: A Generative Introduction. Malden, MA: Blackwell.
Hudson, T. (1993). Nothing does not equal zero: problems with applying developmental sequence findings to assessment and pedagogy. Studies in Second Language Acquisition 15,
461493.

 Lourdes Ortega
Hyltenstam, K. (1977). Implicational patterns in interlanguage syntax variation. Language
Learning, 27, 383411.
Hymes, D. (1972). Models of the interaction of language and social life: the ethnography of communication. In J. Gumperz & D. Hymes (Eds.), Directions in Sociolinguistics (pp. 3571).
New York: Holt, Rinehart & Winston.
Jansen, L. (2008). Acquisition of German word order in tutored learners: A cross-sectional study
in a wider theoretical context. Language Learning, 58, 185231.
Kasper, G., & Wagner, J. (2011). A conversation-analytic approach to second language acquisition. In D. Atkinson (Ed.), Theories of Second Language Acquisition: Alternative Approaches
and Orientations (pp. 117142). New York, NY: Routledge.
Klein, W., & Perdue, C. (1997). The basic variety (or: Couldnt natural languages be much simpler?). Second Language Research, 14, 301347.
Labov, W. (1966). The Social Stratification of English in New York City. Washington D.C.: Center for Applied Linguistics.
Labov, W. (1969). Contraction, deletion, and inherent variability of the English copula. Language, 45, 715762.
Larsen-Freeman, D. (1976). An explanation for the morpheme acquisition order of second language learners. Language Learning, 26, 125134.
Larsen-Freeman, D. (2011). A complexity theory approach to second language development/
acquisition. In D. Atkinson (Ed.), Alternative Approaches to Second Language Acquisition
(pp. 4872). New York, NY: Routledge.
Lenneberg, E.H. (1967). Biological Foundations of Language. New York, NY: Wiley.
Luk, Z. P., & Shirai, Y. (2009). Review article: Is the acquisition order of grammatical morphemes impervious to L1 knowledge? Evidence from the acquisition of plural -s, articles,
and possessives. Language Learning, 59, 721754.
Master, P. (1997). The English article system: Acquisition, function, and pedagogy. System, 25,
215232.
Meisel, J. (1997). The acquisition of the syntax of negation in French and German: Contrasting
first and second language development. Second Language Research, 13, 227263.
Meisel, J. (2011). First and Second Language Acquisition: Parallels and Differences. Cambridge:
CUP.
Meisel, J., Clahsen, H., & Pienemann, M. (1981). On determining developmental stages in natural second language acquisition. Studies in Second Language Acquisition, 3, 109135.
Mellow, J.D. (2006). The emergence of second language syntax: A case study of the acquisition
of relative clauses. Applied Linguistics, 27, 645670.
Norris, J. M., & Ortega, L. (2000). Effectiveness of L2 instruction: A research synthesis and
quantitative meta-analysis. Language Learning, 50, 417528.
OGrady, W. (2012). Three factors in the design and acquisition of language. Wiley Interdisciplinary Reviews: Cognitive Science, 3, 493499.
Ortega, L. (2009). Understanding second language acquisition. London: Hodder Arnold.
Ortega, L. (2013). Ways forward for a bi/multilingual turn in SLA. In S. May (Ed.), The Multilingual Turn: Implications for SLA, TESOL and Bilingual Education. New York, NY: Routledge.
Perales, S., Garcia Mayo, M. del P., & Liceras, J.M. (2009). The acquisition of L3 English negation
by bilingual (Spanish/Basque) learners in an institutional setting. International Journal of
Bilingualism, 13, 333.

Chapter 8. Trying out theories on interlanguage

Pienemann, M. (2011). Explaining developmental schedules. In M. Pienemann & J.-U. Keler

(Eds.), Studying Processability Theory: An Introductory Textbook (pp. 5063). Amsterdam:
John Benjamins.
Pienemann, M., Johnston, M., & Brindley, G. (1988). Constructing an acquisition-based procedure for second language assessment. Studies in Second Language Acquisition, 10, 217243.
Pienemann, M., & Keler, J.-U. (Eds.). (2011). Studying Processability Theory: An Introductory
Textbook Amsterdam: John Benjamins.
Preston, D., & Bayley, R. (2009). Variationist linguistics and second language acquisition.
In W.R.T. & T.K. Bhatia (Eds.), The New Handbook of Second Language Acquisition (pp.
89113). Bingley, UK: Emerald.
Rampton, B. (1995). Crossing: Language and Ethnicity among Adolescents. London: Longman.
Rampton, B. (2008). Style contrasts and the figuration of trajectory. Working Papers in Urban
Language and Literacies, 49, 120. <http://www.kcl.ac.uk/innovation/groups/ldc/publications/workingpapers/49.pdf>
Reder, S., Harris, K., & Setzler, K. (2003). A multimedia adult learner corpus. TESOL Quarterly,
37, 6578.
Schumann, J.H. (1976). Second language acquisition: The pidginization hypothesis. Language
Learning, 26, 391408.
Schumann, J.H. (1978). The Pidginization Process: A Model for Second Language Acquisition.
Rowley, MA: Newbury House.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10, 219231.
Selinker, L., & Han, Z.-H. (2001). Fossilization: Moving the concept into empirical longitudinal
study. In C. Elder, A. Brown, E. Grove, K. Hill, N. Iwashita, T. Lumley, T. McNamara &
K. OLoughlin (Eds.), Experimenting with Uncertainty: Essays in Honor of Alan Davies
(pp. 276291). Cambridge: CUP.
Slabakova, R., Montrul, S., & Prvost, P. (Eds.). (2006). Inquiries in Linguistic Development: In
Honor of Lydia White. Amsterdam: John Benjamins.
Stauble, A.E. (1978). The process of decreolization: A model for second language development.
Language Learning, 28, 2954.
Tarone, E. (1979). Interlanguage as chameleon. Language Learning, 29, 181191.
Tarone, E. (2007). Sociolinguistic approaches to second language acquisition research: 1997
2007. Modern Language Journal, 91, 837848.
Tarone, E., & Liu, G.Q. (1995). Situational context, variation, and second language aquisition
theory. In G. Cook & B. Seidhofer (Eds.), Principles and Practice in the Study of Language
(pp. 107124). Oxford: OUP.
Van Dijk, M., Verspoor, M., & Lowie, W. (2011). Variability and DST. In M.H. Verspoor, K. de
Bot & W. Lowie (Eds.), A Dynamic Approach to Second Language Development: Methods
and Techniques (pp. 5584). Amsterdam: John Benjamins.
VanPatten, B., & Williams, J. (Eds.). (2007). Theories in Second Language Acquisition: An Introduction. Mahwah, NJ: Lawrence Erlbaum Associates.
Verspoor, M., Lowie, W., & Van Dijk, M. (2008). Variability in second language development
from a Dynamic Systems perspective. Modern Language Journal, 92, 214231.
Zobl, H., & Liceras, J. (1994). Functional categories and acquisition orders. Language Learning,
44, 159180.

chapter 9

Another step to be taken Rethinking the

end point of the interlanguage continuum
Diane Larsen-Freeman
University of Michigan

Larry Selinkers Interlanguage article has had a formative role in shaping the
modern-day study of second language acquisition (SLA). This chapter begins by
singling out several of Selinkers contributions to the ontology of SLA. It goes
on to claim that at this point in the evolution of the study of SLA another step
needs to be taken, which is to reconsider the endpoint of the interlanguage
continuum. Using a biological analogy, it argues that there is no endpoint
for (inter-) language or its learning. Neither is extrinsically teleological. The
question then becomes how to reconcile the non-extrinsic teleology of language
and its learning with the normativity of teaching. The chapter concludes by
suggesting ways that a reconciliation might be achieved.

Introduction
I can still remember the excitement that the introduction of Larry Selinkers article
on interlanguage generated in 1972. I was a graduate student at the University of
Michigan at the time. My professor at Michigan, H. Douglas Brown, came to class
one day that year and announced that he had just read an article entitled Interlanguage, which he believed would prove very important to the field. Professor
Brown was right. The publication of Interlanguage (IL) was a watershed moment
in many ways. One significant way was Selinkers call to distinguish second language learning from second language teaching and to study the former apart from
the latter. At the time, this call was revolutionary, and it gave momentum to what
S. Pit Corders (1967) article had inspired 5 years earlier: the modern day study of
second language acquisition.
I begin this chapter by introducing five contributions of the Interlanguage article, which I later return to in this chapter. I have chosen to highlight these five
because I personally feel that they have made an enduring contribution. I list the
five in the order they appear in the article. Next, I explain why I believe that at this

 Diane Larsen-Freeman

juncture, another step needs to be taken. I conclude by offering some preliminary

thoughts on what implications this next step might augur for language teaching
and research.
Five enduring contributions of the interlanguage article
Selinker was remarkably prescient. The call for studying second language production was revolutionary when it was issued, and many of the claims in the article
have withstood the test of time. Here are five of them that I single out for their farreaching consequences.
The separation of teaching from learning
The first I have already mentioned is Selinkers call to separate teaching from learning. In the second paragraph of his article, Selinker writes:
It is also important to distinguish between a teaching perspective and a learning
one...In distinguishing between the two perspectives, claims about the internal
structures and processes of the learning organism take on a very secondary character in the teaching perspective...But such claims do provide the raison dtre for
viewing second-language learning from the learning perspective. This paper is
written from the learning perspective (Selinker, 1972, pp. 209210)

Selinkers distinguishing between teaching and learning and choosing to write

from a learning perspective were important moves in promoting the study of second language acquisition.
The definition of acceptable data
The second contribution that stands out for me is Selinkers definition of acceptable data.
In the learning perspective, what would constitute the psychologically-relevant data
of second-language learning? My own position is that such data would be those behavioral events which would lead to an understanding of the psycholinguistic structures and processes underlying attempted meaningful performance in a second
language. The term meaningful performance situation will be used here to refer to
the situation where an adult attempts to express meanings, which he may already
have, in a language which he is the process of learning. (Selinker, 1972, p. 210)

With this definition of data, Selinker deftly removes from contention as data student production in performance drills, the use of which was a common teaching

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

practice at the time. The focus he ascribed to meaning anticipated the huge shift
which ensued, where language came to be seen not only as a formal system, but
also as a system for making meaning, not only in SLA, but in linguistics and applied linguistics more generally.1
The existence of a separate linguistic system
...one would be completely justified in hypothesizing, perhaps even compelled to hypothesize, the existence of a separate linguistic system based on the observable output
which results from a learners attempted production of a TL [target language] norm.
This linguistic system we will call interlanguage (IL). (Selinker, 1972, p. 214)

Indeed, perhaps, the most enduring legacy of the article was its invitation to view
the learners attempts at producing the target language as a linguistic system, as a
language in its own right. This focused attention on the systematicity of learner
production, rather than seeing it as random performance.
Fossilization
As a fourth contribution, Selinker introduces the important, but contentious, idea
of fossilization to explain seemingly immutable forms in learners interlanguages.
Fossilizable linguistic phenomena are linguistic items, rules, and subsystems
which speakers of a particular NL will tend to keep in their IL, relative to a particular TL, no matter what the age of the learner or amount of explanation and
instruction he receives in the TL. (Selinker, 1972, p. 215)

Distinguishing linguistic units from psychological units

And, finally, in my select group, I put Selinkers distinction between linguistic units
and units of the psychology of second language learning. There is, after all, no
reason to expect learners language to be composed of some form of morphosyntactic structures.
...we should state that there is no necessary connection between relevant units
of linguistic theory and linguistically-relevant units of a psychology of secondlanguage learning. (Selinker, 1972, p. 225)

1. Of course, the importance of treating language as a meaning-making system did not originate with Selinker. Linguists such as Firth and Halliday long maintained this position.

 Diane Larsen-Freeman

Anyone would be very happy to leave a legacy as rich as this one. Perhaps the ultimate compliment, though Selinker may not see it as one, is that the term interlanguage is so ubiquitous in professional literature these days that it is often used
without attribution. One further point that I would like to make here is that
Selinker did not rest on his laurels. Over the years, modifications to the interlanguage concept have been made, such as adding the qualifying construct of discourse domains (Selinker & Douglas, 1985), which I take up at the end of this
chapter. Such updating is the very essence of good scholarship it seems to me: the
willingness to refine and even modify ones views as they mature. So there is a lot
to appreciate in the article Interlanguage and its author.
An enduring controversy, too
However, despite its enduring contributions, one significant question and answer
sequence in the article (which may stem from retaining a language teaching perspective) has sparked controversy:
...how does a second-language-learning novice become able to produce IL utterances whose surface constituents are correct, i.e., correct with respect to the TL
whose norm he is attempting to produce? (Selinker, 1972, p. 223)

A question followed immediately with his reply:

This question finally brings us face-to-face with the notion of success in absolute
terms: productive performance in the TL by the second-language-learner which is
identical to that produced by the native speaker of that TL. (Selinker, 1972, p. 223)

He quickly adds Of course, success need not be defined so absolutely. The teacher
or learner can be satisfied with the learners achieving what has been called communicative competence (p. 223). Nevertheless, despite his making allowances for
learners whose language production is not that of native speakers, Selinker persists in defining successful learning as being the reorganization of linguistic material from an IL to identity with a particular TL (Selinker, 1972, p. 224)
Another step to be taken
Now, the IL article was written at a time when our conception of language was very
much influenced by Chomsky and Lenneberg. Terms such as native speaker competence (p. 212) and surface constituents (p. 223) reflected the then current

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

theory and were the discourse of the time. It would be unfair to judge these expressions with todays sensibilities. They made sense in their day, and to many, they
still do. Nevertheless, at this time, I believe that there is a step that needs to be
taken and that is to challenge equating success with conformity to native speaker
norms. This position has already been challenged many times; here I will summarize the challenges on conceptual, ideological, and theoretical grounds.
As Ortega has put it
That is, wittingly or unwittingly, SLA researchers often portray development as a
transitional state that is (or should be) ever changing towards the target. Implied
in this construal is also an idealized monolingual native speaker, who is held to be
the ultimate yardstick of linguistic success. (Ortega, 2009, p. 140)

Embedded in Ortegas statement are 3 ideas worth expanding upon:

development as a transitional state ever changing towards the target
idealized monolingual native speaker
ultimate yardstick of linguistic success
I will address these three in turn.
Development as a transitional state changing towards the TL
Many others before me have pointed out the value of studying second language
development as its own system (e.g., Huebner, 1983) and of the misconception of
entertaining a view of IL in light of its development towards the TL. Perhaps most
famously, Bley-Vroman (1983) cautioned against the comparative fallacy, comparing L2 learners systems to the TL norms, a practice that may distort our understanding of the development of L2 speakers knowledge.
Vivian Cook (1996, p. 11) underscored this point by later adding
Bringing in the target L2 too soon may warp our analysis of the learners own
grammar towards the idiosyncrasies of the L2 rather than seeing it as a possible
human language that has to be discussed as a thing of its own.

In other words, introducing a target L2 perspective undermines the strongest argument for the interlanguage construct, i.e., that it is a language in its own right,
and needs to be examined as such.
As Year (2004) more recently observed
...the central issue of the comparative fallacy is that the IL data is not comparable
to the target language norm because this is ultimately an invalid comparison of L2
learners performance to native speakers hypothesized grammatical knowledge (or
competence). (Year, 2004, p. 2)

 Diane Larsen-Freeman

Besides the mismatch between learners performance and native speakers (Year
calls it hypothesized I prefer idealized) idealized competence, there is an additional problem: There is no definitive understanding of a native speaker. Han
(2004, p.166) comments that the construct of the native speaker figures prominently in SLA research and yet it is one of the least investigated and least understood concepts in the field. This observation is especially troubling, she adds,
because despite the lack of clarity concerning the concept, the native speaker has
been depicted, among other things, as a goal or a model for SLA or ... as a yardstick to measure second language knowledge. (p. 166)
Indeed, Davies (2003, p.180) as cited in Han (2004, p.167) claims
SLA research has always been more interested in the native speaker than in language proficiency. In particular it has compared native-speaker behavior and that
of various second language learners, asking the question: What does the second
language learner know and to what extent does this differ from what the native
speaker knows?

Han goes on (p. 167) to state that, Few would deny that Davies does herein capture a fundamental question of SLA researchers over the years, that is, whether or
not (adult) L2 learners can achieve linguistic competence that is indistinguishable
from that of a native speaker. Of little comfort to those who opt to answer the
question is the title of Hans article, suggested by Davies definition to be a native
speaker means not to be a nonnative speaker, which Han understandably says is
the only possible operational definition one can give of the native speaker concept
(Han, 2004, p.166). So, the point is that if we are to continue to compare nonnative
and native speakers, we are standing on shaky ground. But, the problem is not only
a conceptual one.
(Idealized) monolingual native speaker
Several researchers (Cook, 1991; Ortega, 2005; Seidlhofer, 2004) have rightly asserted that the monolingual native speaker is not a legitimate model for L2 learning. Yet, despite this assertion, it can still be said that most researchers
continue to apply monolingual norms, when conducting research on bi- and multilingualism, which means that, among other aspects, native-speaker language
proficiency is still used as the yardstick for all the languages of the multilingual
person and the multilingual subject and their languages can be investigated without taking all the languages in contact into consideration ... (Herdina & Jessner,
2013, p. 755)

Herdina and Jessners words ring even truer in light of the superdiversity (Vertovec,
2007) that characterizes mobility of populations, especially in Western Europe these

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

days. Although population flows are not a new phenomenon, we are more aware
that it is no longer possible to assume that language use is tied to particular nationalities or ethnicities. Even an appeal to hybridity to challenge this essentialism is
problematic because as Makoni and Makoni (2010) note, hybridity is predicated
upon and privileges the notion of languages as discrete entities. This implies that one
can deterermine where one language ends and the other begins. In contrast, Makoni
and Makonis term vague linguistique acknowledges that speakers have access to
diverse linguistic resources and use them in unpredictable ways. Their approach accords speakers agency in using bits and pieces of languages. Along similar lines,
but going even further, is Canagarajahs (2013) notion of translingual practice. While
in agreement that there are not separate competences for separately labeled languages, Canagarajah also includes other semiotic resources in translingual practice,
writing that communication transcends words (p. 7). Besides, many speakers have
no desire or intention of attempting to conform to the native speaker communityestablished standards of language practice (Preston, 1989). Even for those who do,
the fact is that the learners system and any idealized system will never converge.
In any case, a homogeneous native speaker speech community (or non-native
speaker, for that matter) does not exist (e.g., Firth & Wagner, 1997), and never has,
which brings us to the third factor.
The ultimate yardstick of linguistic success
There are several problems with this concept. First, the metaphor of a yardstick
suggests unidirectionality to the SLA process, which does not characterize L2
learning. Second, reprising the themes of the inappropriacy of using distinct L1
and L2 norms with multilinguals and of the impact of context on language use,
Herdina and Jessner (2013) continue
The reconstruction of the development of language systems based on the linear
projection from endpoint measures, as common in L1 and L2 acquisition models,
can be shown to be essentially misguided as within multilingual systems there
might be considerable fluctuation and variation within the respective language
systems as the respective languages wax and wane depending on environmental
demands. (Herdina & Jessner, 2013, p. 754)

A developmental yardstick also assumes that there is some agreed upon static target some common native speaker endpoint which we have already seen is a
vexing issue.
Finally, besides problems with the definition of native speaker endpoint, the
metaphor of a yardstick assumes that there is consensus on success. However, for
researchers, success is conceived of differently depending on ones theoretical

 Diane Larsen-Freeman

commitment. For instance, from a socio-cultural perspective, success would not

be determined by a learners individual performance, but instead by a learners
responsiveness to mediation. This means that what an individual is capable of
with assistance at one point in time, he or she will be able to do without assistance
at a future point in time (Lantolf & Thorne, 2007, p. 214). In short, for all these
reasons, conceptual, ideological, and theoretical, the definition of success as conformity to native speaker norms is problematic.
Byrnes (2013, p. 221) sums up the dilemma with regard to some of the points
I have been making:
In what has been called the bilingual turn in language studies, authors find fault
with (1) the undue weight being given to an accident of birth and a concomitant denial of the effects of history, culture and societal use; (2) the undisputed
authority and legitimacy in representing and arbitrating standards of form and
use enjoyed by native speakers; and (3) the troubling disregard of current social,
political and cultural realities of multilingualism and ever-changing forms of hybridity between multiple languages as learners adopt and adapt various identities
in diverse circumstances of life.

Given these serious and mounting criticisms, one can justifiably wonder why
then the native speaker target still affords a benchmark for second language development at all. It may be useful at this point to review what Wolfgang Klein (1998) has
written about the matter. Klein offers three answers to this question: First, a normative perspective is retained in teaching, with teachers helping students move as close
as possible to some arbitrarily imposed norm. Second, this normative point of view
is reinforced by all of us who have learned a language through formal instruction,
where the language is well-defined in textbooks, grammars and dictionaries and
correspondingly assessed (though see VanPatten, this volume). Third, according to
Klein, researchers are implicated as well. Having a TL perspective from which we
can measure IL deviance provides a simple clear design for empirical work.
Klein has therefore broadened culpability to include members of the research
community as offenders. If we accept that a TL perspective should not be entertained, at least exclusively, then what can be done about it? My answer is that we
need to take another step. We need to reconceive language and therefore interlanguage. SLA is still suffering from an externally teleological view of (inter-)language
(Ortega, 2009, this volume; see also Ortega & Byrnes, 2008, p. 287). But, this view
is flawed. As I wrote some years ago, language has no end, and it has no state
(Larsen-Freeman, 2006a).
A teleological view of language
Teleology means completion or end-directedness, goal, purpose. The genitive
form of its Greek root, telos-, is also germane. It forms the prefix of English words

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

such as telescope and telepathy, both of which imply a relationship to something occurring at a distance, despite physical discontinuity. In other words,
Deacon (2012, p. 24) writes We recognize teleological phenomena by their development toward something they are not, but which they are implicitly determined
with respect to...It is the end for the sake of which they exist... (p. 24). I believe that
this has been the assumption that the field of SLA has been operating under: That
IL is a teleological phenomenon (implicitly) determined with respect to the TL. As
we have already seen, this assumption is faulty for a number of reasons. I will use
Deacons example, contrasting a discussion of artifacts crafted by humans with
natural phenomena that develop on their own, to make this point another way.
Think about a human craft, such as pottery. I imagine that the potter has a
mental image of a finished ceramic bowl when she begins. However, it takes shape
and is somewhat remodeled as the clay is worked on the wheel. The bowl is being
constructed for some purpose, and whatever that is, it is likely that the purpose
guides the selection and modification of its physical characteristics, e.g., the bowl
should be shaped with a flat bottom to keep its contents within. The point is that
what guides the creation and use of the bowl is located extrinsically, so a bowl
derives its end-directed features parasitically, from the teleology of the designer
or user. It is not intrinsic (Deacon, 2012, p. 25). In contrast, a body organ is not
dependent on any extrinsic teleology. Of course, it, too, has a function but a biological function lacks an explicit representation of the end with respect to which it
operates (Deacon, 2012, p. 25). It could be said, though, to have an intrinsic or
immanent teleology (Weber & Varela, 2002).
How organisms work is just not the way artifacts work: the latter always point to an
external purpose they are made or used for, the former are purposes with the goal of
keeping existent by organizing themselves. (Weber & Varela, 2002, pp. 106107)

In other words, a biological organism is autopoietic. [A]utopoiesis entails the

production and maintenance of a dynamic entity in the face of material change
(Thompson, 2007, p. 146). Its identity is not fixed, as it is with a ceramic bowl,
because the cells it is composed of are constantly being renewed. Further, an autopoietic organism needs to be adaptive, to change in relation to changing conditions (Thompson, 2007).2 Importantly, though, this understanding of adaptivity is
not as it is in Darwins theory of evolution. For neo-Darwinians, evolution involves the optimization of adaptation through natural selection. From an autopoietic perspective, however, adaptation is an invariant background condition of all
2. In organisms, there is a twofold purpose of identity (self-production) and sense making
(adaptivity and cognition), based on autopoiesis (Thompson, 2007, p. 153). Sensemaking is
behaviour or conduct in relation to environmental significance and valence, which the organism
itself enacts or brings forth on the basis of its autonomy (Thompson & Stapleton, 2008, p. 3).

 Diane Larsen-Freeman

life (Maturana and Varela 1987, pp. 94117)... (as cited in Thompson, 2007,
p. 159). Furthermore, interaction with the environment can alter its development
(Gilbert & Sarkar, 2000)3 and its size.
Thus, in these ways (independence from an extrinsic teleology and its continual adaptation to changing conditions), I think language is more like a biological organ than an artifact. There are several other key differences between the bowl
and a body organ. The potter begins with the clay for the pot from the start. From
these, she shapes her bowl. On the other hand, an organ, like the heart, grows from
one cell by dividing, multiplying, and differentiating. Even when the heart reaches
full size, it does not stop changing. As I have just noted, its cells are constantly being replaced (Wade, 2009). Finally, the life of an organism is not resident in its
parts. It is whole from the start, embodied in the global organization of the living
processes (Deacon, 2012, p. 135).
I do not want to carry this analogy too far. Clearly, human language is not a
biological organ. It is functionally different. We are not just the carriers, but its
authors. Its construction takes place within the context of a collective speech
community. Furthermore, language in use is semiotic.4 On the other hand, language in use does have an autopoietic nature, maintaining its identity all the
while it is changing (Larsen-Freeman, 2011a). In addition, we should remember
that, just like a biological organ5, a language in use grows and changes with very
little conscious intervention of humans (Keller, 1985) (cf. VanPatten, this volume). Moreover, seeing language in use more as an organism than an artifact
also does away with the need to posit preformationism (the assumption that in
order to build a complex structure you need to begin with a detailed plan or
template) (Deacon, 2012, p. 50). Indeed, from a complex systems point of view,
language complexity is not due to the unfolding of some prearranged plan
(Tucker and Hirsh-Pasek, 1993, p. 364) because all that is required to account for
3. Gilbert and Sarkar give some fascinating examples, such as that a female ant larva can become a queen or a worker, depending on the food she has been given, a turtle can develop into
male or a female, depending on the temperature at a critical point in its incubation, and a wrasse
fish can become a male or a female depending on whether there is a male already resident in the
reef. These life history strategies make up a large part of contemporary ecology (Gilbert &
Sarkar, 2000, p. 7).
4. I note, though, that there is a field of biosemiotics, which introduces a profound change of
perspective implied when life is considered not just from the perspectives of molecules and chemistry, but as signs conveyed and interpreted by other living signs in a variety of ways... (Wikipedia
entry on biosemiotics). Biosemiotics seeks to link the two kingdoms of mind and matter, in
order to give humanity its place in nature (Hoffmeyer, 1996, p. 94, as cited in Kull, 1999).
5. Rutherford (1987), too, has argued for a more organic metaphor for grammar, in contrast
to a machine metaphor.

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

complexity is a sensitive dependence on initial conditions and a language-using

context within which the system can adapt and change.
Even when we speak not of the language of a speech community, but rather
those language resources of the individual learner, the analogy is useful. The language resources of an individual may be few, but they constitute a wholly functional system, one that grows under propitious conditions. One way it grows is, as
with cells, through differentiation (cf. Andersens (1984) one to one principle),
following from meaningful interaction, be it in conversation, in private speech, in
interaction with a computer or a written text. New forms are not mere additions to
the system; they change the system itself (Feldman, 2006). The system develops
from experience (Ellis & Larsen-Freeman, 2006), afforded by the environment.
The ambient language does, therefore, have a role in its shape. But the point is that
it does not determine it, nor does it define the learning trajectory. If it did, there
would be no way to account for the individual developmental paths that learners
take. Larsen-Freeman and Cameron (2008, p. 158) put it this way:
Embodied learners soft assemble their language resources interacting with a changing environment. As they do so, their language resources change. Learning is not
the taking in of linguistic forms by learners, but the constant adaptation [creation,]
and enactment of language-using patterns in the service of meaning-making in
response to the affordances that emerge in a dynamic communicative situation.

This view of language as a complex adaptive system (Ellis & Larsen-Freeman,

2009) counters the tendency to portray learner language as being an incomplete
and deficient version of native speaker language. Indeed, implicit in this understanding of language as a self-modifying, emerging system is that the developmental change process is never complete and neither is its learning. It also, to my mind,
offers an interpretation of fossilization from a probabilistic variationist approach,
which attributes an apparent halt in development to an entrenched item-based
pattern which simply remains available for use for the time being (Eskildsen 2012,
p. 366), in other words a probabilistic pattern, not an absolute one (Berdan, 1996).
Furthermore, this point of view supports Selinkers observation that learners
language production units are not linguists constructs. Instead, they are languageusing patterns, variegated in form and length, which are psycholinguistically identical (Croft & Cruse, 2004). It [language] is the way it is because of the way it has
been used, its emergent stabilities arising out of interaction (Larsen-Freeman &
Cameron, 2008, p.115), not because of an extrinsic teleology.
In sum, the view I am putting forth here (as I have for some time, e.g., LarsenFreeman, 1997) is that language as realized in a speech community is an open
system, always changing, never fixed and that the language resources of its speakers are a dynamic network of language-using patterns: emergent, mutable and

 Diane Larsen-Freeman

self-organizing. Their development within language learners, then, is not as an act

of conformity, but rather is extended from continuing dynamic adaptedness to a
specific present and ever-changing context (Larsen-Freeman, 2006b, 2011b).
Thibault (2011, p. 211) put it this way
Advances in the theory of dynamical systems, grounded in our understandings
of neurobiology ... and ecosocial systems, have opened up the possibility of a different approach. The new approach stresses the centrality of co-acting agents who
extend their worlds and their own agency through embodied, embedded processes of languaging, rather than uses of an abstract language system...

Reconciling the non-teleological nature of language with the normativity

of language teaching
It is important to acknowledge that the view of language learning that I have been
promulgating in this chapter is that of natural language development. While it is
essential for teachers to understand the natural process, they are responsible for
aiding learners to at least approach the norms of the community in which the
learners seek membership. To do so, teachers need to be concerned not only with
change from below, i.e., the change initiated by language learners, but also with
change from above (Tarone, 2007), i.e., change prompted by instruction, often itself influenced by standardized examinations. The question then becomes how to
help learners extend their linguistic worlds, all the while making possible their
membership in the discourse communities to which they desire admission. The
following are some possible moves to reconcile the two:
1. Set the overall goal of language teaching as developing capacity (Widdowson,
1983).
By capacity, Widdowson means the ability to create meaning with language.
Capacity is that which enables learners to move beyond speech formulas in order
to innovate. It is what accounts for the fact that language changes all the time, and
that it does so due to the cumulative innovations that language users make at the
local level as they adapt their language resources to new communicative contexts
(see Montrul this volume). Capacity is an active force for continuing creativity
(1983, p. 27).
2. Within this overall goal, identify particular contexts of use, contexts in which
norms for local success can be established.
As Schleppegrell (2006) proposes, such contexts, e.g., tasks, genres, assignments,
and situations, establish the expectations for language use. In other words, we

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

should not be thinking of teaching language, writ large, but rather teaching the
lexicogrammatical resources present in particular oral or written texts (Byrnes,
forthcoming). We need to think of language as locally contingent and situated
(Eskildsen, 2012, p. 353).
On a related note, Selinker and Douglas (1985) also point out that local contingency is true for individual learners. [Interlanguage] processes do not take
place primarily across ILs, but in fact take place in internally created discourse
domains that are important to individual learners: domains that the learner greatly needs or wishes to interact in... (1985, p. 199).
3. Engage learners in activities that are rich in affordances.
The concept of affordances recognizes that learners perceive their own learning
opportunities, despite what the teachers goal is for a particular pedagogical activity. Rather than thinking of providing students with input, then, teachers should
think of activities which allow students access to the language through multiple
entry points. If the language learner is active and engaged, she will perceive linguistic affordances and use them for linguistic action (van Lier, 2000, p. 252).
Indeed, the gap between the teachers intentions for a given activity and how a
learner engages in the same activity can be a key resource for understanding
the processes of learning as processes of formation of agency (Engestrm &
Sannino, 2012, p. 46).
4. Create activities that encourage transfer appropriate processing (Lightbown,
2008; 2013; Larsen-Freeman, 2013a).
Transfer appropriate processing takes place when the conditions of learning and
the conditions of use are aligned. It is not the same as using language in authentic
situations. What is important is that the conditions are semiotically aligned, where
speakers have a psychologically genuine need to communicate.
5. Design activities where language-using patterns as identified by contexts of use
(in keeping with the learners goals) are iterated (Larsen-Freeman, 2012a).
Students need to encounter and use the patterns present in the contexts of use iteratively. It is navigating the tension between convention and innovation that is
central to learning. As Larsen-Freeman and Cameron (2008, p. 190) have observed
When we make use of genres in speaking or writing, we use the stabilized patterns but exploit the variability around them to create what is uniquely needed for
that particular literacy or discourse event. The challenge for the teacher is to provide for iteration in a way that does not undermine the meaningful use of language. This can be achieved by designing learning activities that are inherently
repetitive (Gatbonton & Segalowitz, 2005), but offer affordances for creativity.

 Diane Larsen-Freeman

6. Teach reasons for form-based rules (Larsen-Freeman, 2003). (For a different

view of rule-based teaching, see VanPatten, this volume.)
Help students develop awareness of the choice they have in how they use patterns
within a context of use (Larsen-Freeman, 2002). Teach students to understand that
language rules are not as arbitrary and fixed as they seem. For example, a rule of
English grammar that says that you cannot use the present participle with a stative
verb is not inviolate, if you use the stative verb meaningfully in an active sense
(He is being humble), or use it to emphasize change (We are loving our new
schedule), or a span of time (I have been wanting a new car). Helping students
see the reasons equips them better to see language as a meaning-making system,
not one filled with arbitrary rules.
7. Teach adaptation.
What we should be teaching is not only language, but also the process of adaptation: Teaching students to take their present system and mold it to a new context
for a present purpose (Larsen-Freeman, 2013b). One means of doing so is to build
on Stevicks idea of technemes: When teachers change the conditions for completing a task successfully from one time to the next, however slightly, a new challenge
is presented. For instance, if the students have to complete the same task a second
time, but do so taking less time, students have an opportunity to learn to adapt
their language resources.
8. Develop students capacity for semiotic agility.
Priors concept of semiotic agility is the capacity for shifting rapidly and fluently
between and among semiotic worlds (Prior, 2010, p. 233). To do this, teachers would
guide students to move easily among the many different modes of communication
in the many different contexts students encounter. As Thorne (2011) has put it This
calls for an ontological shift in how cognition/action and development are related
From the dead hand of competence (Geertz, 1973: 88) to semiotic agility.
9. Help students learn to negotiate in translingual contexts (Canagarajah, 2013).
Helping students to negotiate effectively for intelligibility and communicative success in global contact zones calls for them to develop performance competence
(Canagarajah, 2013). As there is no homogeneous norm to call upon in such zones,
performance competence requires suitable translingual negotiation strategies beyond teaching students to negotiate meaning. They include rhetorical and social
considerations as well as cultivating cooperative dispositions.
10. As a counterbalance to the use of formal assessment for measuring what learners have acquired, find ways of assessing progress in terms of development.

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum

This entails assessing learning in a self-referential way: Looking at what learners

are doing over time, expanding their repertoire of language resources, for instance,
and defining progress in terms of where a learner wants to go, not looking at what
the learner is not doing in light of some idealized target. As Clarke (2013, p. 295)
writes, we have to recognize that measures of success cannot be universally applied to all ... students. This injunction applies to both teachers and researchers,
no doubt a different and perhaps difficult way of operating, but one that improves
on the assumption that progress is unidirectional and achieved when it is identical
with native speaker production.
Conclusion
In this chapter, I have suggested that the concept of IL provided an indispensable
initial step in establishing the legitimacy of looking at language learning, independent of the normative perspective of language teaching. It has stimulated much
research and has had a formative role in setting research agendas. Furthermore, it
did later itself move in a non-teleological direction through the construct of discourse domain, as internally defined by the learner.
We must now take another step to distance our focus from a teleological view.
By continuing to equate identity with idealized native speaker production as a
definition of success, it is difficult to avoid seeing the learners IL as anything but
deficient. Instead, I have called for reconceiving language as a complex adaptive
dynamic system, one locally constituted without an extrinsic teleology. This reconception means that language learning success need not be determined by a common distal endpoint. I believe that enacting an instructional approach, such as
called for here, would facilitate membership in the learners preferred discourse
community as well as support their capacity for extending the system. We can then
look to the learners ever-expanding capacity to use their palette of lexicogrammatical resources (Byrnes, forthcoming) to engage in meaningful languaging in a
multitude of socially created and internally defined contexts in an agile fashion. In
actuality, learners actively transform their linguistic world; they do not merely
conform to it (Larsen-Freeman, 2012b).
References
Andersen, R. (1984). The one-to-one principle of interlanguage construction. Language Learning, 34, 7795.
Berdan, R. (1996). Disentangling language acquisition from language variation. In R. Bayley &
D.R. Preston (Eds.), Second Language Acquisition and Linguistic Variation (pp. 203243).
Amsterdam: John Benjamins.

 Diane Larsen-Freeman
Bley-Vroman, R. (1983). The comparative fallacy in interlanguage studies: The case of systematicity. Language Learning, 33, 117.
Byrnes, H. (2013). Renting language in the ownership society: Reflections on language and
learning in a multilingual world. In J. Arnold & T. Murphey (Eds.), Meaningful Action: Earl
Stevicks Influence on Language Teaching (pp. 216234). Cambridge: CUP.
Byrnes, H. (forthcoming). Theorizing writing development at the intersection of task and L2
writing: Reconsidering complexity. In H. Byrnes and R. Manchn (Eds.), Task-based Language Learning: Insights from and for L2 Writing. Amsterdam: John Benjamins.
Canagarajah, S. (2013). Translingual Practice. New York, NY: Routledge.
Clarke, M. A. (2013). Individual and organizational learning: Exploring the psycho-social dynamics of change. In J. Arnold & T. Murphey (Eds.), Meaningful Action: Earl Stevicks Influence on Language Teaching (pp. 281297). Cambridge: CUP.
Cook, V. J. (1991). The poverty-of-the-stimulus argument and multi-competence. Second
Language Research, 7(2), 103117.
Cook, V. J. (1996). Some relationships between linguistics and second language acquisition. In
K. Sajavaara & C. Fairweather (Eds.), Approaches to Second Language Acquisition, Jyvskyl
Cross-Language Studies, 17, 713.
Corder, S. P. (1967). The significance of learners errors. International Review of Applied
Linguistics, 5, 161170.
Croft, W., & Cruse, A. (2004). Cognitive Linguistics. Cambridge: CUP.
Davies, A. (2003). The Native Speaker: Myth and Reality. Clevedon: Multilingual Matters.
Deacon, T. (2012). Incomplete Nature. New York, NY: W.W. Norton & Company.
de Angelis, J., & Jessner, U. (2012). Writing across languages in a bilingual context: A dynamic
systems theory approach. In R. Manchn (Ed.), L2 writing development: Multiple perspectives (pp. 4768). Berlin: de Gruyter.
Ellis, N. C., & Larsen-Freeman, D. (Eds.). Language as a complex adaptive system. Language
Learning, 59(Suppl. 1), 90112.
Engestrm, Y., & Sannino, A. (2012). Whatever happened to process theories of learning?
Language, Culture and Social Interaction, 1, 4556.
Eskildsen, S. (2012). L2 negation constructions at work. Language Learning, 62(2), 335372.
Feldman, J. (2006). From Molecule to Metaphor. A Neural Theory of Language. Cambridge, MA:
The MIT Press.
Firth, A., & Wagner, J. (1997). On discourse, communication, and (some) fundamental concepts
in SLA research. The Modern Language Journal, 81(3), 285300.
Gatbonton, E., & Segalowitz, N. (2005). Rethinking communicative language teaching. The
Canadian Modern Language Review, 61(3), 325353.
Geertz, C. (1973). The Interpretation of Cultures. New York: Basic Books, Inc.
Gilbert, S. F., & Sarkar S. (2000). Embracing complexity: Organicism for the 21st Century.
Developmental Dynamics, 219(1), 19.
Han, Z.-H. (2004). To be a native speaker means not to be a nonnative speaker. Second Language
Research, 20(2), 166187.
Herdina, P., & Jessner, U. (2013). Epilogue: The implications of language attrition for dynamic
systems theory: Next steps and consequences. International Journal of Bilingualism, 17(6),
752756.
Hoffmeyer, J. (1996). Signs of Meaning in the Universe. (translated by B.J. Haveland). Bloomington,
IN: Indiana University Press.
Huebner, T. (1983). A Longitudinal Analysis of the Acquisition of English. Ann Arbor, MI: Karoma.

 Chapter 9. Another step to be taken Rethinking the end point of the interlanguage continuum
Kull, K. (1999). Biosemiotics in the twentieth century: A view from biology. Semiotica, 1271/4,
385414.
Klein, W. (1998). The contribution of second language acquisition research. Language Learning,
48(4), 527550.
Lantolf J., & Thorne, S. (2007). Sociocultual theory and second language learning. In B. VanPatten & J. Williams (Eds.), Theories in Second Language Acquisition (pp. 201224). New York,
NY: Routledge.
Larsen-Freeman, D. (1997). Chaos/complexity science and second language acquisition. Applied
Linguistics, 18, 14165.
Larsen-Freeman, D. (2002). The grammar of choice. In E. Hinkel & S. Fotos (Eds.), New Perspectives on Grammar Teaching (pp. 103118). Mahwah, NJ: Lawrence Erlbaum Associates.
Larsen-Freeman, D. (2003). Teaching Language: From Grammar to Grammaring. Boston: Heinle/
Cengage.
Larsen-Freeman, D. (2006a). Second language acquisition and the issue of fossilization: There is
no end, and there is no state. In Z-H. Han & T. Odlin (Eds.), Studies of Fossilization in
Second Language Acquisition (pp. 189200). Clevedon: Multilingual Matters.
Larsen-Freeman, D. (2006b). On the need for a new understanding of language and its development. Journal of Applied Linguistics, 3l(3), 281304.
Larsen-Freeman, D. (2011a). A complexity theory approach to second language development/
acquisition. In D. Atkinson (Ed.), Alternative Approaches to Second Language Acquisition
(pp. 4872). Oxford: OUP.
Larsen-Freeman, D. (2011b). Saying what we mean: Making a case for second language acquisition to become second language development. A plenary address delivered at the 16th
World Congress of Applied Linguistics (AILA), August 28, 2011.
Larsen-Freeman, D. (2012a). On the roles of repetition in language teaching and learning.
Applied Linguistics Review, 3(2), 195210.
Larsen-Freeman, D. (2012b). The emancipation of the language learner. Studies in Second
Language Learning and Teaching, 2(3), 297309.
Larsen-Freeman, D. (2013a). Transfer of learning transformed. Language Learning, Supplement
1, 107129.
Larsen-Freeman, D. (2013b). Complex systems and technemes: Learning as iterative adaptations. In J. Arnold & T. Murphey (Eds.), Meaningful Action: Earl Stevicks Influence on
Language Teaching (pp. 184195). Cambridge: CUP.
Larsen-Freeman, D., & Cameron, L. (2008). Complex Systems and Applied Linguistics. Oxford:
OUP.
Lightbown, P. (2008). Transfer appropriate processing as a model for classroom second language
acquisition. In Z.-H. Han (Ed.), Understanding Second Language Process (pp. 2744).
Clevedon: Multilingual Matters.
Lightbown, P.M. (2013). Transfer appropriate processing. In P. Robinson (Ed.), Encyclopedia of
Second Language Acquisition (pp. 652655). London: Routledge.
Makoni, B., & Makoni, S. (2010). Multilingual discourses on wheels and public English in
Africa. A case for vague linguistique. In J. Maybin & J. Swann (Eds.), The Routledge Companion to English Language Studies (pp. 258 270). Abingdon: Routledge.
Maturana, H. & Varela, F. (1989). The Tree of Knowledge: The Biological Roots of Human Understanding. Boston: Shambala Press.
Ortega, L. (2005). For what and for whom is our research? The ethical as transformative lens in
instructed SLA. The Modern Language Journal, 89, 427443.

 Diane Larsen-Freeman
Ortega, L. (2009). Understanding Second Language Acquisition. London: Hodder Education.
Ortega, L., & Byrnes, H. (Eds). (2008). The Longitudinal Study of Advanced L2 Capacities. New
York, NY: Routledge.
Preston, D. (1989). Sociolinguistics and Second Language Acquisition. Oxford: Blackwell.
Prior, P. (2010). Remaking IO: Semiotic remediation in the design process. In P. Prior & J. Hengst
(Eds.), Exploring Semiotic Remediation as Discourse Process (pp. 206235). Houndmills:
Palgrave Macmillan.
Rutherford, W. (1987). Second Language Grammar: Learning and Teaching. London: Longman.
Schleppegrell, M. (2006). The linguistic features of advanced language use: The grammar of exposition. In H. Byrnes (Ed.), Advanced Language Learning: The Contribution of Halliday
and Vygotsky (pp. 134146). London: Continuum.
Seidlhofer, B. (2004). Research perspectives on teaching English as a lingua franca. Annual
Review of Applied Linguistics, 24, 209239.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics, 10(3), 209231.
Selinker, L., & Douglas, D. (1985). Wrestling with context in interlanguage theory. Applied Linguistics, 6, 190204.
Tarone, E. (2007). Sociolinguistic approaches to second language acquisition research19972007.
The Modern Language Journal, 91, 837848.
Thibault, P. (2011). First-order languaging dynamics and second-order language: The distributed language view. Ecological Psychology, 23, 210245.
Thorne, S. (2011). Languaging phenomenologies of time, place, and space. Insights into applied
linguistics: Languaging, agency and ecologies. Plenary at the 30th Summer School of Applied Language Studies. June 7, 2012. Jyvskyl, Finland.
Thompson, E. (2007). Mind in Life. Cambridge, MA: Harvard University Press.
Thompson, E., & Stapleton, M. (2008). Making sense of sense-making: Reflections on enactive
and extended mind theories. Topoi, 28(1): 2330.
Tucker, M., & Hirsch-Pasek, K. (1993). Systems and language: Implications for acquisition. In L.
Smith & E. Thelen (Eds.), A Dynamic Systems Approach to Development: Applications
(pp. 359384). Cambridge, MA: The MIT Press.
Van Lier, L. (2000). From input to affordance: Social-interactive learning from an ecological
perspective. In J. Lantolf (Ed.), Sociocultural Theory and Second Language Learning (pp.
243259). Oxford: OUP.
Vertovec, S. (2007). Super-diversity and its implications. Ethnic and Racial Studies, 9(6),
10241054.
Wade, N. (2009). Heart muscle renewed over lifetime. New York Times, April 3, 2009, pg. A12 of
the New York edition.
Weber, A., & Varela, F. (2002). Life after Kant: Natural purposes and the autopoietic foundations
of biological individuality. Phenomenology and the Cognitive Sciences, 1, 97125.
Widdowson, H. G. (1983). Competence and capacity in language learning. In M. Clarke & J.
Handscombe (Eds.), On TESOL 82: Pacific Perspectives on Language Learning and Teaching
(pp. 97106). Washington, DC.
Year, J. E. (2004). Instances of the comparative fallacy. Teachers College, Columbia University
working papers in TESOL & Applied Linguistics, 4(1), 17.

chapter 10

Interlanguage 40 years on
Three themes from here
Larry Selinker

New York University and Research Production Associates

This chapter pulls together 40 years of interlanguage study in terms of three
interlinked themes: (1) historic roots, emphasizing ongoing questions/issues
that needed interlanguage to be asked; (2) interlanguage as a system in its own
right, centering on evidence for the systematic nature of interlanguage; (3) a
suggested future centering on creation of a deep interlanguage semantics guided
by not only learning from, but contributing to, an exponentially-changing
computational world. The chapter concludes with a peroration addressed to
current students and younger colleagues stressing development of doubt as a
crucial skill not being too quick to discard possible answers to core concepts
and questioning whether we are training students adequately in the right skillsets
for productive research.

Introduction
This concluding chapter is organized around three linked themes that have been
calling for unified treatment for some time. What links these three themes and
serves as background motif to much of the volume, is a concern with organization
of information in this field, a current take on interlanguage principles, and, following from this, inferred doubt about some current beliefs and practices.
The first major theme is historical understanding. We organize this theme by
focusing on two linked observations:
a. pre-interlanguage, there were important issues and questions that have not
been resolved, and we should understand their history, rejoice at progress
made, but ponder why they still hang in the air. Conversely and important to
this theme of historical understanding, we also wish to emphasize that
b. post-interlanguage, there are robust issues and resulting questions that could
NOT have been proposed before we had the concept of interlanguage, or
something like it.

 Larry Selinker

Examination and organization of older and newer interlanguage principles provides a basis for revealing how they are understood by current students and young
colleagues, and this raises doubt as to the preparation of new researchers, and
outcomes of their work.
The second major theme is interlanguage as a system in its own right. In order to understand principles underlying the original concept of interlanguage
(Selinker, 1972), the analyst should avoid imposing externally-derived, preconceived categories and units borrowed from other fields upon interlanguage data.
There are underlying methodological issues raised here and I have particular doubt
about current practices (my own included). The problem is that not imposing externally-derived categories on primary interlanguage data has proven very hard to
do. We always seem to have to resort to some standard. In raising doubt about
current analytical practices and their theoretical ramifications, it is proposed that
progress can be made through development of the third major theme.
The third major theme is the creation of a deep interlanguage semantics. Interlanguage analysts do not know exactly what the represented meanings and concepts are when we see and hear target-language-like words and novel forms used
in interlanguage. We are only beginning the task of working out interlanguage
meanings, and this is where I feel we must expend energy. In this section, I delineate a deep interlanguage semantics, the search for universal and idiosyncratic
interlanguage logical propositions, and contextual-particular ones. I briefly outline some promising developments in an exponentially-changing technical world
environment, where we can both learn and contribute.
Finally, I share some thoughts of what to tell young students and new colleagues who want to know and to make sense of the bedrock categories of this
branch of human knowledge.

Three major themes

Historical understanding
As a result of the Interlanguage Hypothesis, perspectives on learner talk and
writing were profoundly changed from traditional concentration on the chaos of
errors a concern going back centuries to viewing such talk and writing as
being highly systematic and, thus being emblematic of complex acquisitional and
developmental processes.
Realizing that from the beginning, Interlanguage Studies has never been a
monolithic enterprise, I will use this working definition of interlanguage:

Chapter 10. Interlanguage 40 years on

Interlanguage is that linguistic/cognitive space that exists between the native

language and the language that one is learning. Interlanguages are non-native
languages which are created and spoken whenever there is language contact.

People create interlanguage when attempting to express meanings in a second language. Interlanguages are highly structured, containing new/novel forms. They
result from attempted production of a target language and are never perfect when
measured in terms of the target language, but deviate in structured ways. Unlike
other areas of human behavior, very interestingly, here practice does not make
perfect in ultimate success terms1 (Selinker, 2012).
Additionally, with interlanguage, or as Ortega (this volume) succinctly puts it:
What adult learners do in building a new language, there has been a concurrent
shift away from over-reliance on the forms of the target language that have been
acquired (except maybe in the testing/assessment industry) to focus on knowledge
that learners have actually acquired and constructed as idiosyncratic interlanguage
and how interlanguage knowledge is used. Historically, we realize there are two
different sets of issues and questions:
a. pre-interlanguage issues and questions, many still unanswered
b.post-interlanguage issues and questions that could NOT have been asked, even
contemplated, before we had interlanguage as a recognized and established
category.

We can begin to organize interlanguage information around these two sets of issues and questions. Set (a) involves primarily issues of language transfer and, secondarily, issues about age and memory. Set (b) initially concerned influences that
shape/re-shape developing and stabilized parts of interlanguage system(s), as well
as questions of individual variation. But much more has been revealed.
Concerning gathering and organizing interlanguage information, there is still
the unresolved issue of what constitute the psychologically-relevant data of second-language learning (Selinker, 1972). In linguistic circles at the time, there was
much discussion of the newly proposed use of grammaticality judgments by
generativists. It seemed to me then, and still does, that using these tests gets the
analyst information about a system (especially instructed information about rules
[see VanPatten, this volume]) that is different from the knowledge system controlling interlanguage use. But gaining psychologically-relevant data of second-language learning has proven more difficult than was originally envisioned. Gass &
Polio (this volume) show the complexities of this issue in an historical context and
provide an up-to-date view on what appropriate data for analysis might be. I am
1. As Larsen-Freeman (this volume) emphasizes, the notion of success in a second language,
as portrayed in the Interlanguage paper, has created great controversy.

 Larry Selinker

in sympathy with their postulate that a multiplicity of data types are needed, that
appropriate data should not be judged in a vacuum but should be relatable to
particular research questions asked.
Pre-interlanguage issues and questions that are still unanswered
Central to the contrastive analysis endeavor was the classical concept of language
transfer (von Humboldt, 1836, Whitney, 1881, Fries, 1945; Weinreich, 1953, Lado,
1957). Odlin notes that von Humboldt:
...sees second language acquisition as the only way an individual may escape from
the conceptual world of the native language. Yet the escape is never, in the view of
von Humboldt, completely successful. (Odlin, 2005)

But for traditional contrastive analysis there was only a dichotomous choice, all or
nothing, transfer/no transfer. This dichotomy was a powerful orthodoxy and even
a tyranny (more later). Important here is that though much has been learned in 40
years about language transfer, from the beginning it was seen and is still seen, as a
central interlanguage process. Weinreich (1953) described learners seeking to establish interlingual identifications or making the same what cannot be the
same. Language transfer effects are now regarded as more subtle (cf. Odlin, this
volume). We now see more different types of transfer, such as reverse transfer
from a second language to a first (Cook, 2003), interlanguage transfer from one
interlanguage to the next (Schmidt & Frota, 1986; Selinker & Baumgartner-Cohen,
1995; Amaro et al, 2012; Odlin, this volume). Even with extensive work on language transfer, we should doubt our analyses, since we still lack serious predictive
principles, as:
What are constraints and conditions under which interlanguage transfer occurs
from interlanguage1 to interlanguage2 versus when interlanguage transfer is
blocked by the native language, and native language transfer occurs, versus when
no transfer occurs?

Questions about memory always seem to have been there in the background. For
example, Lado (1957) had a strong hypothesis that a central part of learning a
second language involves expanding memory span, still a good hypothesis, though
I know of no work on it. There now exist a range of questions about such matters
as working memory (Williams, 2012) and its relationship to long-term memory
storage; on differential abilities in controlling information so that learners can
carry out further processing; on what is stored temporarily, and therefore available
later; and on the still unresolved relationship between perception and production
now study-able in terms of the creation of interlanguage systems (Gass & Mackey
2012 and Robinson, 2012).

Chapter 10. Interlanguage 40 years on

Questions comparing learning a first language with learning a second are

very old (cf. especially Lenneberg, 1967; Selinker, 1972; Ortega, this volume). A
post-interlanguage obsession (cf. White, 2003) is whether universal grammar is
reactivated in learning a second language, but solid conclusions are just not
there, in my view (but see Meisel, 2011). One solid post-interlanguage result is that
first language acquisition leads to convergent grammars and second language acquisition leads to divergent grammars (Bley-Vroman, 1983, 1989; Sorace, 2011;
passim, this volume).
In terms of this first set of questions, then, there has been real progress with
more abundant, more refined questions and more clearly organized data. Interestingly, some vexing older problems are not problems anymore. One such involves
the putative dichotomous phonetic/phonemic distinction in chasing contrastive
error predictions.2 Analysts now do not have to search for dichotomous bounded
categories if none exist.3
We move to questions that need the construct of interlanguage.

Issues and questions that could NOT have been raised before interlanguage
We start with a basic fact about interlanguage that is profound in its implications,
and seems to be supported in every study:
Learners in their interlanguages create new/novel forms, i.e. forms not in their
native or target languages.

Often, these new interlanguage forms have unclear meanings (examples below).
Historically, we have to admit that some contrastive analysts noticed such new/
novel forms in their data. They were baffled and admittedly had no idea how to
handle them. They just honestly listed these unaccounted-for forms in a section of
the contrastive study they called lacunae or residue (Kleinjans, 1958; Alatis,
1966). Increasing pervasiveness of these new/novel forms turned out to be crucial
small facts (Goffman, 1974, 1981) in overturning the tyranny of the strong contrastive analysis hypothesis and have become central to the interlanguage endeavor (see Selinker, 1992). Mike Long (personal communication) points out that we
2. We here owe a debt to Chomsky, wiping out clear theoretical dividing lines between phonetics/phonemics, especially the clever work on the Russian devoicing rule (Chomsky & Halle,
1968) where there occur general phonetic rules applying across both putative independent subsystems, a situation regularly found in interlanguage, but little researched. Similar results regarding Hebrew appear in Selinker (1967).
3. None of this is to say that the classical categorical phonetic/phonemic distinction may not
be of value. Adrian Palmer (personal communication) points out that the distinction, while
probably leaky, is still worth keeping. Language teachers would probably find themselves without an important organizing principle if phonemic categories were completely discarded.

 Larry Selinker

do not know the range of new/novel forms, where they come from, and that we
gave up interlanguage analysis too soon for hypothesis testing. There are many issues here regarding such forms, including what classes of interlanguage grammar
and semantics they belong to and how they might relate to possible universals of
language and interlanguage. To the best of my knowledge, this stream of research
has never been pursued.
Fossilization is the quintessential interlanguage phenomenon. Here, too, we
must acknowledge that, before interlanguage, this phenomenon of getting stuck in
a second language was noticed by some scholars, for example Weinreich (1953)
and Nemser (1961). But, it must be emphasized that before its labeling as fossilization, there was no organized empirical work on it.
Fossilization may be what singles out second language acquisition as a unique
phenomenon.4 In Selinker (1992) I claimed that without fossilization, there would
be no field of Second Language Acquisition (discussed in Long, 2003; Han &
Selinker 2005; Selinker 2011; Han, 2013). Without this non-convergent property
of getting stuck in a second language, leading to divergent grammars (Bley-
Vroman, 1989, 2009 and Han, this volume), acquisition would be acquisition,
and why distinguish? But we do distinguish. All the work that has gone into understanding the idiosyncratic nature of stabilization/fossilization phenomena
would not have been possible without the interlanguage construct, or something
like it, to spur that work on. Many things are still not clear most especially, the
exact basic unit of fossilization. The best bet, in my view, is still interlingual units
(though now, more tied to meaning, discourse and information structure) that are
latent in a distinct structure, activated only when learners attempt to express
meaning in a second language. This conceptualization fits the criteria which tie
together the neuro-cognitive mechanism, the realizational unit, and the sociopsychological condition of fossilization (Han, 2013).
Fossilization is a concept that is now used outside the second language acquisition community (Han, 2013, this volume). Fossilization research and its practical
implications are major growth industries. There are factors that interact with each
other that control and are controlled by the phenomenon. That is, when the five
central processes of interlanguage (Selinker, 1972) interact with each other and
with other processes, such as phonetic similarity (Haugen, 1950, 1953) or structural symmetry (Kellerman, 1989), their effect will be stronger on fossilization
projection. Most important is the interactive link between fossilization and language transfer. This interconnected link has been explored in terms of the Multiple
Effects Principle (Selinker & Lakshmanan, 1992; Lardiere, 2003; Schwartz, 2013),
4. Montrul (this volume) argues that fossilization and the other 4 central interlanguage processes are normal psycholinguistics processes. But see interlanguage-type processes below.

Chapter 10. Interlanguage 40 years on

with fossilization being selective and local and not a global phenomenon as
suggested, wrongly, in the Interlanguage paper (Han, this volume).
Interlanguage over the years has been shown to be anything but monolithic.
Variation in interlanguage may be the most empirically studied of all interlanguage phenomena; its study in differential inter- and intra-learner development
continues to be another growth industry. A seminal early paper is Tarone (1979;
see also 1988); much of the work since then is summarized in Bayley and Tarone
(2012), organized around key issues in an historical context. All the interactional and variationist research about the effect of recasts, positive and negative
feedback, error corrections related to input of many and various types, and developing and stabilized interlanguages could not have been done without the
interlanguage construct.
Since Adjmian (1976), we have spent a lot of time on a series of related questions of a universalist nature, either about structure or development, for example:
How do interlanguage data fit into concerns of universal grammar? How does
universal grammar fit into concerns of interlanguage data? The list of studies is
long and well-explored (Schwartz & Sprouse 1996; White, 2003; Lardiere 2007;
Cook 2009; Montrul, this volume). Adjmians crucial insight that interlanguage is
different from other natural languages in that it is permeable has yet to enter the
debate in a lasting way.
Processing strategies were initially set up in Selinker (1972) as communication versus learning strategies. This dichotomy has proven problematic. A number of sources document processing strategies and strategy learning (Cohen,
2011; Pienemann & Kessler, 2012). This research has much practical positive
fallout. The intersection of processing instruction (VanPatten, 2011; VanPatten
et al, 2013) with interlanguage development/fossilization creates a natural and
ongoing research area (VanPatten, this volume). Odlin (this volume) relates understanding of transfer to concerns of language-specific vs. possible universals
of language processing.
Many questions about the process of noticing the gap between ones current
state of interlanguage knowledge and a target, have arisen since Schmidt and Frota
(1986; cf. Schmidt, 2010). This gap is important to interlanguage development
given our definition of interlanguage as that linguistic/cognitive space that exists
between the native language and the language one is learning. This idea of cognitive space between two languages is developed in terms of intercultural
awareness(Kramsch, 1998). Much cognitive work and issues on spatial processing
and representation (e.g. Gnzel, 2007) are of value in understanding interlanguage
from the perspective of this metaphor, especially in light of current (over)use of
social media across cultures, often through interlanguage. Issues arise:

 Larry Selinker

Where does private interlanguage space begin and end?

What is it that makes this private space more public?
What exactly has to be noticed to create that move?
Questions abound about accuracy and what it can mask in terms of proficiency
level. To me, accuracy is about using grammar precisely and the grammaticalization
of interlanguage (Heine & Kuteva, 2005; Kuteva & Heine, 2010; Larsen-Freeman,
this volume; Ortega, this volume). Linguistic precision, and its (inevitable) targetlanguage focus, 5 can run into the paradox of second language learning. It has been
known for a long time (e.g. Saporta, 1966; Lawler & Selinker, 1971) that attention
to grammar while speaking gets in the way of successful linguistic performance6
and perhaps interlanguage growth. The place of conscious knowledge and processing in second language acquisition is still unresolved (Robinson et al, 2012).
Grammaticalizing an interlanguage can mean the beginning of creation of a basic
variety (cf. Klein & Perdue, 1992, 1997; Perdue, 1984, 1993) with restructuring of
this type of interlanguage being meaning-driven (whether target-like or not) by
expanding grammatical resources.
The way information is organized has been seriously studied in the area of
linguistics called cognitive grammar (Langacker, 1991, 2008), and applied to second language learning and teaching (Tyler, 2012). Is information coded and organized differently in interlanguage? Is language transfer a factor? What is stored
temporarily, and therefore available later can now be studied in terms of the creation of interlanguage systems (Gass & Mackey, 2012; Robinson, 2012).
The latent psychological structure, a genetically determined mechanism in
the brain (at least partially distinct from the latent language structure postulated
by Lenneberg, 1967) has distinguishing characteristics: no genetic timetable; may
not be activated or realized as a natural language; may overlap with other cognitive
structures. Whether there is one language acquisition device to create both L1 and
interlanguage, or two language acquisition devices (or none) has been
roundly debated (e.g. Ritchie & Bhatia, 2009). Selinker (1972) postulated a latent
5. Ortega (this volume) sees a persistent and vexing ambivalence between a) interlanguage as a system in its own right and b) the need to refer to a target language, a contradiction
in interlanguage studies as a project. Given that an interlanguage is created in the context of
study or use of another language, I see no recourse but to refer to that target for conditions of
accuracy. This intellectual issue has to be disentangled from a political one: the privileged status of native speakers, but target standards could be non-native ones, like English as a Lingua
Franca, depending on the context.
6. Saporta (1966: 49) states this paradox of second language learning most eloquently:
... the ability or inclination to formulate the rules apparently interferes with the performance
which is supposed to lead to making the application of the rules automatic.

Chapter 10. Interlanguage 40 years on

psychological structure for interlanguage and I have seen no reason to abandon

this position (cf. Odlin, this volume; Han, 2013; Han, this volume).
Questions of memory extend to interlanguage in a permeable way (Adjmian,
1976) in terms of re-learning and the retrieval of interlanguage items one thought
one knew (Morgan-Short & Ullman, 2012), an issue important to heritage language
learning (Kagan & Dillon, 2012). I have seen no integration of this common memory problem with its permeable dimension into linguistic concerns of interlanguage.
There are puzzling questions about different types of interlanguage, with
too many equating learner language with interlanguage. Non-native speakers
(NNS) with complex grammaticalized syntax cannot seriously be called language learners anymore, but they still have divergent phonetic, syntactic, semantic/pragmatic systems. Researchers studying Interlanguage Pragmatics insist
that it can exist apart from SLA research (Kasper & Blum-Kulka, 1993; BardoviHarlig, 2012), purposefully disassociating their work from second language acquisition, concentrating on second language use (Rose, 2012). One wonders if
there is a discernible cut-off point between learner language and these other
types of interlanguage.
How does interlanguage fit into a larger scheme of things? Many believe that
interlanguage processes such as the creation of new/novel forms, and even fossilization, are NOT unique to interlanguage (cf. Ortega, Montrul, this volume). Since
the 1980s such new/novel forms have been shown to exist in interdialect, i.e forms
not in the local nor in the national standard dialects (Trudgill, 1986; Chambers &
Trudgill, 2004). The interdialect approach differs from earlier dialect approaches
in its attention and sensitivity to stages in learning a second dialect. The context
here is that dialect speakers find themselves confronting, in meaningful performance situations, a standard dialect; interlanguage-type processes seem to be
involved7. Effects go broader than that: interlanguage processes seem to be differentially widespread among other cognitive intersystems interculture, interliteracies and maybe all cognitive intersystems.
Before one conflates all these intersystems into one, e.g. claiming (falsely,
I believe) that dialect acquisition is a subarea of second language acquisition
(Siegel, 2010), note that both propositions may be true:
Second language acquisition is a unique kind of learning.
Interlanguage-type processes are widespread, though unevenly distributed,
among all the cognitive intersystems.

7. The cognitive and social connection between language/dialect acquisition and use has been
discussed well beyond second language/dialect acquisition and technical linguistics, often in the
context of its racial and political implications (cf. Alim & Smitherman, 2012).

 Larry Selinker

The hypothesis proposed (Selinker, 2007, 2012) is that:

Interlanguage, interdialect, interculture, interliteracy... are all types of intersystems.

Since attested new/novel forms are also found in interculture (called C3, Kramsch,
1998) and may also occur in interliteracy, it follows that:
Interlanguages are subsets of intersystems (Selinker, 2007) as are interdialects and
intercultures since there exist attested new/novel forms in all three. This would
apply to interliteracies if such forms are found. (Selinker, 2012)

It is important to emphasize that these are separable sub-systems of the phenomenon we call intersystem, and thus, they are not all the same cognitive entity.
Processes of learning or stagnation in one should not necessarily be superimposed
on another. Each has to be argued for independently on system-internal
grounds.
We are back to an original interlanguage question (Selinker, 1972) but raised
to a higher level: What are the basic units of intersystem? In this case, what are the
basic units of interlanguage?
Interlanguage as a system in its own right (Selinker 1972)
Theme two concerns the principle that interlanguages must be described in their
own terms, that units/meanings/syntactic bracketing of interlanguage must be decided on system-internal grounds, and that one must put description before explanation, preferably in meaningful performance situations (Bardovi-Harlig, this
volume). As a corollary, it means that one must avoid the infamous comparative
fallacy (Bley-Vroman, 1983, 1989). This fallacy involves deciding units/meanings/syntactic bracketing of interlanguage by referring either to native or target
language. Committing this fallacy does not allow us to get to the true units/meanings of interlanguage; it gives us something else, knowledge of another system or
knowledge of something unclear.
Bearing in mind the principle of interlanguage as a system in its own right,
we have doubt about much current analytical practice, which continues to impose
externally-derived, preconceived categories upon interlanguage data, claiming
they are the basic units of interlanguage knowledge. We reject imposing such categories whether they arise from standard languages or from some form of universal grammar. Slapping target-language or universal grammatical categories on
target-language-like words that compose a majority of interlanguage data does not
work because such target-language-like forms (Spolsky, 1973) often have unclear, indeterminate, ambiguous meanings.

Chapter 10. Interlanguage 40 years on

This lack of clear meanings for interlanguage words has been a problem since
our earliest days, when we knew that we couldnt take at face value what interlanguage speakers mean when they use target words (Bayley & Tarone, 2012). Odlin
(this volume) states:
SLA researchers must exercise greater circumspection in their inferences about
what an (interlanguage) production actually means.

This problem of inherent ambiguity of interlanguage data, has been known since
Corders (1967) extensive discussion of what he called covert errors. One example
still resonates, the German speaker in Edinburgh who said:
You must not take off your hat.

for the target:

You dont have to take off your hat.

We here have perfectly grammaticalized target language-sounding words in their

proper syntactic form, but with clearly divergent interlanguage meaning in the
semantic area of modality.
It is crucial to agree that interlanguage analysts have no immediate access to
meanings of the interlanguage being described. Before interlanguage, Weinreich
(1953) noticed that NNSs can perceive target language lexical items with different
sets of semantic features than native speakers, being at a loss for exact selectional/
collocational restrictions attached to lexical items. Since interlanguage intuition
data is suspect (though not prohibited), we revert to dependence on an original
interlanguage principle (Selinker, 1972):
In determining interlanguage meanings, we have only target language lexical items
and their ordering (and, perhaps contextual information) as our primary data. Interlanguage intuitions are secondary data and must be regarded as suspect.

If this principle is true, then the following corollary applies:

Interlanguage analysts cannot easily tap intentions and meanings of learners and
other, more mature interlanguage users.

This emphasizes that interlanguage units and idiosyncratic meanings are suspect
and must be explicitly justified internally with interlanguage-particular phonetic,
semantic and contextual data. Without such explicit argumentation, we have reason
to doubt the validity of analyses presented. We must know what interlanguage forms
mean to interlanguage speakers who produce them. What do interlanguage speakers intend in particular meaningful performance situations by the use of specific
words, when we seriously consider interlanguage as a meaning making system?

 Larry Selinker

Consider phrasal verbs. We cannot code in target-language terms interlanguage lexicalization of target phrasal verbs. An example is the production of look
for both target language look over (meaning search/review) and also for look
meaning look. Interlanguage-only synonyms are created where there are none in
the target language, with interlanguage look meaning both look and look over
to the interlanguage speaker.
How does one code interlanguage when a speaker deletes (in target language
terms) a particle or preposition, or sometimes, purely in interlanguage terms,
chooses and uses a favorite particle? In a case study. Bulgarian-English speakers
produced off as a default-particle substitution, knowing that something has to be
there, producing very strange sentences, like pick off the children (Selinker, 1996;
Kaplan & Selinker, 1997). We need to code the interlanguage semantics of such
phrases in a logical way. In terms of interlanguage analysis, these are looked at not
as errors, but as interlanguage-particular semantic codings of interlanguage units,
involving native-language transferred grammaticalization processes of lexicalization of meaning from phrasal unit onto the main verb.
More complex semantic/discourse confusion obtains in interlanguage coding
of verbs like claim which presents a case of absence without presence (Selinker
& Kinahan, 2000) due to transfer-of-training. Even very advanced NNS I have
worked with in technical and academic writing use claim as a synonym for
show or even prove and do not know that claim, in certain constructions,
presupposes that you are about to deny, or at the least question, the proposition of
the person you are discussing. A classical example is the sentence:
Chomsky claims that syntax must be autonomous of semantics.

with the entailment I doubt it, without directly saying so. But such subtle semantic learning may be blocked by the central interlanguage process of transfer-oftraining (Selinker1972) where grammar books, even enlightened ones such as
Biber and Conrad (2009), miss this entailment and lump claim together with
other verbs such as argue, imply, postulate, indicate, propose, contend, maintain,
suggest, hypothesize under the vague heading of: reporting verbs with a
certainty level of less certain. Most essentially, in target-language terms, several
of these verbs do not have this potential presupposition of denial; others are neutral to it.
Another intriguing type of construction involves the all-important structuredependent island constraints (Ross, 1969) used by NNSs variably where interlanguage emphasis is involved. Note interlanguage sentences such as:

You put MYSELF in this terrible situation.

or:

I wonder if my colleagues will defend MYSELF against X.

Chapter 10. Interlanguage 40 years on

Here the interlanguage is inherently ambiguous. You have to know that this particular NNS uses native-like island constraints in the standard manner in some
domains and not others, not violating the standard syntactic rule but only, when
intending emphasis substituting myself for me, using the reflexive pronoun
emphatically, allowing the pragmatic use of reflexive pronoun to take precedence
over the syntactic rule forbidding reflexive pronouns in that slot in that construction. (Lawler, personal communication)
There are oddities here without end. It would be nice to have an interactive
web site where we could contemplate and add relevant information. I recommend that we look carefully at reviving in a more current fashion, an old linguistic tradition, creating books of interlanguage texts like those vast texts gathered
in Amerindian studies by linguistic anthropologists, beginning over 100 years
ago, texts that were contextually rich and full of investigator notes about context
and intention (e.g. Jacobs, 1926, 1933; Givn, 2013). Such books of interlanguage
texts would go beyond current databases. They would be richer in interlanguage
semantics, glossed properly and presented with detailed contexts and organized
information for semantic scenario data mining, such texts organized to be marriable with current and future technologies, creating a common interlanguage data
infrastructure.
How do we design an interlanguage database that accounts for idiosyncratic and variable projections of interlanguage intentions and propositions
onto target-language-like words? Here, we have an ultimate design problem:
Target-language-like words plus novel forms with idiosyncratic meanings must
be made to fit exactly into unique and possibly idiosyncratic semantic propositions and paths. We must design such large databases so that interlanguage
forms using target-language-like words have semantic coding that allows for
both convergent and divergent semantics, as well as novel semantics. This is
way beyond what we have now. But, one thing is clear: coding surface errors
alone wont do; raw data with detailed background information is a step forward (Landolfi, 2012).
From the above, we get the important interlanguage principle:
Assume that interlanguage forms, whether target-language-like or not, have idiosyncratic meanings and deep semantic implications, until shown otherwise.

It is thus claimed that to understand interlanguage, progress can only be made in

avoiding the comparative fallacy through a serious development of the third major
theme, the creation of an interlanguage semantics, where we now turn.

 Larry Selinker

The creation of a deep interlanguage semantics

We start with an earlier point:
Interlanguage analysts do not know exactly what the represented meanings and

concepts are when we hear and see target-language-like words and novel
forms used in interlanguage speech and writing.

We cannot create accurate interlanguage descriptions (and subsequent explanations) without a deep interlanguage semantics, one that goes beyond word and
sentence level. This undeveloped research area seems to me naturally connected
with computational life. We need systematic descriptions of interlanguage semantic systems in evolution, and where these systems get stuck. In this section I outline what I mean by a deep interlanguage semantics and relate that to some current
computational work.
Methodologically, in interlanguage semantics, there are two interlanguage
maxims (cf. Corder, 1977; Selinker, 1972, 1992) that should be kept in mind.
1. Any interlanguage data should be considered idiosyncratic until shown to be
otherwise;
2. No matter how advanced the interlanguage speaker, there will exist both similarities and differences between interlanguage form/meaning combinations
and those of the target language.
It appears that, with the advanced learner, while target language-like words may
work in a particular context, native speakers may say they wouldnt put it that
way (Byrnes, 2012). This issue involves the expression of and the organization of
information, and other semantic subtleties.
To my surprise, I find few deep interlanguage semantic descriptions in the literature. There is a strong, robust lexical/vocabulary acquisition literature that we
can utilize (e.g., Fitzpatrick, Al-Qarni & Meara, 2008). There is the large aspects
hypothesis literature (Shirai & Andersen, 1995; Andersen & Shirai, 1996). BardoviHarlig (this volume) shows the range of interlanguage linguistic devices that speakers use to express the concept of temporality. These descriptions are important, but
we need to link them with work on interlanguage intentions, entailments, presuppositions, construals8 and the like, and in a computational frame.
This theme flows from our failure to deal with ambiguous interlanguage meanings and tendency to impose externally-derived, preconceived categories on interlanguage data with their unclear and often ambiguous meanings. An important
8. The concept of construals, how we perceive the actions of others, is essential but difficult
to work through in interlanguage semantics. I have depended on, in native-language linguistics,
Langacker (1991, 2008, to appear).

Chapter 10. Interlanguage 40 years on

corollary to this theme is our over-reliance on hypothesis testing9 in second language acquisition, as opposed to a hopefully clear interlanguage analysis which
sheds light on the range of idiosyncratic phenomena we are dealing with. In my
view, creation of a deep interlanguage semantics involves matching intentions of
interlanguage speakers with logical propositions and then matching these with
target language-like words and novel forms used in idiosyncratic ways, organized
to be compatible with the world of current, and in principle, future computational
machines.
A deep semantics can start with logical propositions getting things done with
words. Take the proposition: complaining about the neighbors barking dogs.
You can imagine this logical proposition being highly similar across languages and
cultures, but the semantics of the particular interlanguage forms chosen to express
that proposition, their selectional details, their semantic frame connections, entailments, etc. often are idiosyncratic in interlanguage and effected by conceptual
transfer (Odlin, this volume). We will surely become involved in at least some of
the following semantic concepts of interlanguage: polysemy, synonymy and contextual paraphrase relations, interlanguage-created idiomaticity, metaphors, indexicality, modality, topicalization and emphasis, finiteness, case, tense/aspect,
construals, creation of interlanguage-particular frames, and most tricky but essential, ambiguity in the interlanguage, not in the target language.
Interlanguage speakers in meaningful performance situations get tripped up
here all the time, wrongly assuming that their interlanguage target-like words and
phrases have the same meanings in the target system as those the interlanguage
speaker has coded in the interlanguage system. One basic question is:
How do non-native speakers construct idiosyncratic meanings, structure-dependent bracketing, entailments, implicatures, perceived presuppositions, polysemy, synonymy and contextual paraphrase relations, idiomaticity, indexicality,
peculiar semantic frames, construals...?
Stakes are high with such questions; resolving some would shed light on the
depth of semantic concepts which interlanguage speakers regularly create (even at
the highest levels of proficiency) that native speakers would never project from
their target lexical items and sentences.
One tricky type of example is where interlanguage speakers have divergent or
incomplete entailments, leading to misunderstandings or no understandings.
This example comes from an interlanguage speaker who knows English very well,
who said,
They are broken,
9. The fact of our over-reliance on hypothesis testing in second language acquisition is argued
convincingly by Norris and Ortega (2012), with credible alternatives presented.

 Larry Selinker

referring to some neighbors. She was not understood by a mixed group of native
and NNS. This phrase was an interlanguage substitute for standard:
They are broke,

meaning that they have no money. One must refer to a standard here for meaning. In Selinker (2011), I showed how to do a detailed interlanguage semantic
analysis on this sort of item, using the machinery of entailment, where only one
form (broke) of the noun/verb lemma, break, which the advanced and proficient
NNS did not know, has the idiomatized meaning of they have no money.
Such cases, and there are many, lead to an interlanguage principle:
Non-native speakers tend to remember main lexical items of the idiom but not the
grammar owned by native speakers for that idiomatic meaning; as a result they
often lose the idiomatic meaning.

We do not have organized results for questions such as:

What is the interlanguage range for particular core and peripheral semantic propositions? What are the idiosyncratic variations, if interlanguage norms can

be found?

To add to the complexity, take another core interlanguage semantic area, interlanguage synonymy, an understudied issue. Weinreich (1980) gave the example
of the interlanguage speaker who says wise men for wise guys but knows
wise guys when he hears it. Semantically, wise men is not the same as wise
guys, though interlanguage users may make the two phrases interlanguage-only
synonyms. Such synonymy is also created by phonetic processes, as when an
interlanguage exists with few final consonants due to language transfer, unexpected homonyms appear; in one case, the phonetic interlanguage sequence
[da] stood for both jar and job, the phonetic interlanguage sequence [mai]
in [mai] weather,not my weather but mild weather; [bri], later more traffic; much later context, the proposition intended turned out to be more traffic
around (GW) bridge. One thing is clear: Intuition that we as analysts regularly
use, appealing to native speaker semantics especially, will not divulge interlanguage semantics. More sophisticated methods are called for. We need to explore
the range of interlanguage synonymy in an area of great idiosyncratic (re-)structuring and variability.
The role of interlanguage intentions in the making of interlanguage meaning
is crucial since, if such intentions can be reliably tapped, they become the empirical clues to a deep interlanguage semantics. Such intentions can be discovered
with various methodologies: for example, translation with on the go revelations
in discourse with contextual clues such as intonation helping us to understand the

Chapter 10. Interlanguage 40 years on

making of meaning on the fly in context-dependent and flexibly-construed

situations (Gee & Hayes, 2011; Gee, 2013).
In computational use fields, progress is being made discovering how to get at
intentions, often within a closed computer environment. For example, making
sense of intention is a problem that screen reading software has dealt with (Lane,
2013). If, through close analysis, we can build reliable and valid closed contexts,
even with artificial languages IF they have meaning and structure dependency
(Gass &Polio, this volume), developing digital tools for interlanguage analysis, we
could then move to more open contexts and contribute to this digital world.
This theme of the development of a deep interlanguage semantics is important
in and of itself, but more important, projecting from the world we live in, it can be
a source of jobs, in that we can then exploit and contribute to computational linguistics, especially in the areas of NNS speech recognition, talk to text and text to
talk, the interlinguistics of automatic translation and even encryption. We have
barely touched the surface here. The reader is referred to Selinker (2012) for a
strong connection with interlanguage in computational life, where, in terms of
automatic translation, there now exist interlanguage-wiki programs, where propositions are stored in a data bank at the loading time of the source article and
tapped when a translation is demanded through various portals.10
Students and younger colleagues seem to be open to conversation about interlanguage semantics, about technological use in interlanguage analysis and reframing of durable historical issues, about contributing to computational life, but do not
seem to know how to go about connecting the technological and historical dots, let
alone knowledgeable in current computational linguistics. What stops them from
connecting these dots? Are we preparing students and mentoring younger colleagues with proper future use skills and understandings? These concerns lead me
to a perorative concluding statement on how to conduct our future work.
Peroration: Developing doubt as a critical skill
I wish to address these final remarks to current students11 and younger colleagues.
It is important to regularly contemplate bedrock categories of this branch of
human knowledge. Focus on doubt, conceived of not in negative terms, but on
10. Interlanguage web semantic details: http://de.wikipedia.org/wiki/Datei:Interlanguage_Semantics.svg. Interlanguage wiki links are listed at: http://en.wikipedia.org/wiki/Help: Interlanguage_links.
11. It is interesting to recall that it is students who have consistently contributed to our topic
here traced way back, certainly back to the influence of Hakutas senior thesis at Harvard (1974,
1975, 1976)

 Larry Selinker

developing disciplined doubt as a skill.12 You can start, first and foremost, by being aware of a subtle but potentially pernicious phenomenon:
Be wary of a false consensus manifesting itself as dogma.

As an Applied Linguistics student at Georgetown (196263), I kept hearing a particular assumption stated as fact, which I know I doubted from the very beginning. There was what I now conceive as a false consensus with at least 80 years of
empirical work behind it (Whitney, 1881). Contrastive analysts, for all their
strengths, were blinded by an ideology as to what actually and factually happens
with learners. I was a student in courses in contrastive analysis, language testing,
and the like, taught among others by writers of two of my founding texts, Robert
Lado and Bob DiPietro. 13 After Lado had said repeatedly,
The learner transfers his forms, meanings, and distributions from the native lan-

guage to the target language. (Lado, 1957)

I jumped up and seemingly shouted:

Tell me: How can i transfer my native language English patterns to someone elses
French? There must be another system involved.
Lado, to his everlasting credit, said in essence: prove it! And I tried to do just that with
the education Georgetown linguistics gave me. I did fieldwork in Tel Aviv (Selinker,
1966, 1969) applying tools learned in extensive lab settings working on big texts, from
extensive training in detailed analytical tools of linguistic field methods and contrastive analysis, and I am forever grateful. Most of all I learned at Georgetown:
Doubt, do not fear censure, learn to take time and see the evidence.

Data are not out there waiting to be picked; phenomena are out there. Data have
to be constructed, organized and nursed by disciplined watching, until you actually see evidence. To be sure, one cannot pick up such skills alone; one needs long
tutored experience with detail of solid analytical tools, being shown how to look
and gain awareness of philosophical pitfalls. Science proceeds through error and
refining questions, as well as close attention to multifold details of data. In developing skills of disciplined doubt, you must always question the obvious:
12. The concept of disciplining doubt can be traced to Kathryn Schulz, Being Wrong (Schulz,
2010). She is not responsible for my applications here.
13. Bob DiPietro was one of the best teachers I have ever known and an unsung hero of both
contrastive analysis and bilingualism studies (e.g., DiPietro,1961,1964). Historically, his 1965
treatise (Agard & DiPietro, 1965) shows the detail of using the robust machinery of contrastive
analysis as a technology.

Chapter 10. Interlanguage 40 years on

How do you know what you know?

Why do you believe what you believe?

Nurture your founding texts14 by asking colleagues to state assumptions fully,

watching for hidden assumptions; be ready to challenge. As you do, also
Be wary of false equivalencies.

A false equivalency might be second language acquisition made equivalent to

FIRST language acquisition. Parallel naming leads one astray. It has been pointed
out forcefully that it is hard to know what one can learn about second language
acquisition from generalizing from first.15
Young scholars should make basic to their doubt training practice in critically
analyzing dichotomies that academics are forever creating (and believing). Thus,
Be wary of all dichotomies.

Making a claim that two entities are sharply distinct implies a clear statement of
boundary criteria that apply in every empirical case. When we set up categories,
other logic is involved: how they are set up, how the category relates to other possible categories, boundaries of categories, and internal structure of the category.
Proliferation of dichotomous categories persists; unfortunately, we have many:

second language acquisition versus first language acquisition

native speaker versus non-native speaker
competence versus performance
learning versus acquisition
semantics versus pragmatics
explicit learning versus implicit learning.

Regarding this last dichotomy, for instance, even if two distinct types of knowledge, explicit and implicit knowledge, can be shown to exist and they are disassociated and distinct systems (Robinson et al, 2012, 251/252) represented in two
distinct parts of the brain, it does not automatically follow that there exist
two distinct types of learning, coincidentally with the same names (cf. Han &
Finneran, 2013). This is a muddled area. Reductive conflation of categories is not
a good idea. Assume that things are always more complicated. Look for alternative
more subtle explanations.
Read widely, use abductive reasoning, seek alternative possible explanations.
14. Ones founding texts are personalized. I provide detail about keeping track and nurturing
ones founding texts in Selinker (1979, 1992).
15. Foster-Cohen (2001) shows the detail of how in every instance, the equivalency (first/second etc.) is false, with in one case, at least six possible situation combinations.

 Larry Selinker

None can be accepted per se; each must be argued for within the domain of enquiry. In general,
Be wary of flawed data.

It is hard to establish what is FACT: a statement about the world that most everyone accepts as true, or something that is actually the case. Ideology enters in. Even
something that has occurred often can be contentious since each side is certain
that their position is factual. Verifiability is tough. Scientific facts are verified by
repeatable experiments, but in our domain of enquiry it is hard to get exact replication. Try to work out ways to show that you have reliable data, doubting imposition of externally-defined categories and creating metrics to decide confounding
data. Doubt as fact interlanguage results that are far from meaningful performance
situations. Follow the principles:
Observe interlanguage speakers in meaningful performance situations, and Inte-

grate interlanguage semantics in elicitation devices if possible.

The interlanguage maxim of relevance here is:

Misinterpretation of input by learners is common, maybe the norm.

Doubt even the most basic maxims, e.g. that interlanguage is a linguistic system
in its own right. Again, seek out alternatives. Doubt the putative claims about
mind/brain which are set as assumptions; e.g. doubt the reality of already formulated arrangements or architectures, such as latent language structure (Lenneberg,
1967), latent psychological structure (Selinker, 1972), or even, maybe especially,
universal grammar. Evidence for such big questions will always be indirect and
often vague, and all you can really say is: it is possible.
Finally, keep track of all arguments, even where they appear to contradict. As
a Rabbi friend once put it: Keep track of all arguments...who knows what will be
true in 50 years? This is done in Talmudic thought developed over more than 700
years and is consistent with original interlanguage work. It focuses on questions,
on developing possible answers, often in implicit chaotic ways based on history of
opinion behind each commentary rather than on proposing definite answers,
keeping in mind the strong possibility that there may be none16.
16. One could write a treatise on the belief that the original interlanguage work is consistent
with Talmudic thought and logic. The idea of Talmudic commentaries is to bring centuries of
thought and precedent to bear on new questions which occur with new situations (introduction
to Selinker, 1992). We proposed a Talmudic approach to grammar (Selinker & Naiditch,2004).
Interestingly, something like this happens in Chomsky (1965). A specific model of grammar is
impressively developed over several hundred pages, but then Chomsky says, basically, but it
could also be that and in a few pages, proposes a different perspective from which develop
several strands leading (chaotically) to current minimalism.

Chapter 10. Interlanguage 40 years on

Conclusion
So I have to ask: Are we training students in crucial skills of developing doubt,
historical understanding, choosing and honoring founding texts, keeping track of
all arguments, abductively reading widely and paying serious attention to semantic and computational linguistics developments? OR, are we neglecting the passing
on of crucial skills and understandings for the world ahead?
A final thought, for students and young scholars: if you do not slip; if you have
patience and do not move into any orthodoxy; if you keep honing your critical
skills, especially paying attention to the history of questions asked in this field,
learning as much computer science and computational linguistics as you can, then
the next 40 years will be one long, bright day for interlanguage and interlanguage
studies. Knowledge about interlanguage, interdialect, interculture, interliteracy
and other intersystems, and the processes that create them, will always be needed.
Forty years ago, there was no field of second language acquisition. It is clear that
the interlanguage hypothesis and the concepts associated with it, helped bring this
vibrant field into existence. This field is now progressing way beyond the wildest
dreams of its founders.
Acknowledgements
I want to thank Buzz Palmer, Terry Odlin and the Editors who read and commented on previous versions. I greatly profited from their intelligence and well
thought-through perspectives. It is clear from this volume that whatever the lasting value in the Interlanguage Hypothesis, it has involved the work of many.
References
Adjmian, C. (1976). On the nature of interlanguage systems. Language Learning, 26(2), 297 320.
Agard, F., & DiPietro, R. (1965). The sounds of English and Italian. Chicago, IL: University of
Chicago Press.
Alatis, J.E. (1966). The American English Pronunciation of Greek Immigrants: A Study in Language Contact with Pedagogical Implications. Unpublished doctoral dissertation. Ohio State
University.
Alim, H.S., & Smitherman, G. (2012). Articulate While Black: Barack Obama, Language, and
Race in the US. Oxford: OUP.
Amaro, J., Flynn, S., & Rothman, J. (Eds.). (2012). Third Language Acquisition in Adulthood.
Amsterdam: John Benjamins.

 Larry Selinker
Andersen, R., & Shirai, Y. (1996). The primacy of aspect in first and second language acquisition:
The pidgin-creole connection. In W.C. Ritchie & T. K. Bathia (Eds.), Handbook of Second
Language Acquisition (pp.527570). San Diego, CA: Academic Press.
Bardovi-Harlig, K. (2012). Pragmatics in second language acquisition. In S.M. Gass & A. Mackey
(Eds.), The Routledge Handbook of Second Language Acquisition (pp. 147162). New York,
NY: Routledge.
Bayley, R., & Tarone, E. (2012). Variationist perspectives. In S.M. Gass & A. Mackey (Eds.), The
Routledge Handbook of Second Language Acquisition (pp. 4156). New York, NY: Routledge.
Biber, D., & Conrad, S. (2009). Register, Genre, and Style. Cambridge: CUP.
Bley-Vroman, R. (1983). The comparative fallacy in interlanguage studies: The case of systematicity. Language Learning, 33, 117.
Bley-Vroman, R. (1989). The logical problem of second language learning. In S. M. Gass & J.
Schachter (Eds.), Linguistic Perspectives on Second Language (pp.1 68). Cambridge: CUP.
Bley-Vroman, R. (2009). The evolving context of the fundamental difference hypothesis. Studies
in Second Language Acquisition, 31, 175198.
Byrnes, H. (2012). Advanced learner proficiency. In S.M. Gass & A. Mackey (Eds.), The Routledge Handbook of Second Language Acquisition (pp. 506522). New York, NY: Routledge.
Chambers, J., & Trudgill, P. (2004). Dialectology (2nd ed.). Cambridge: CUP.
Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: The MIT Press.
Chomsky, N., & Halle, M. (1968). The Sound Pattern of English. Cambridge, MA: The MIT Press.
Cohen, A.D. (2011). L2 learner strategies. In E. Hinkel (Ed.), Handbook of Research in Second
Language Teaching and Learning: Volume 2 (pp. 681698). New York, NY: Routledge.
Cook, V. (Ed.). (2003). Effects of the second language on the first. Clevedon: Multilingual Matters.
Cook, V. (2009). Multilingual Universal Grammar as the norm. In I. Leung (Ed.), Third Language Acquisition and Universal Grammar (pp 5570), Bristol: Multilingual Matters.
Corder, S. P. (1967). The significance of learners errors. IRAL, 5(4), 161170.
Corder, S. P. (1977). Language continua and the interlanguage hypothesis. In S.P. Corder & E.
Roulet (Eds.), The Notions of Simplification, Interlanguages, and Pidgins and their Relation
to Second Language Pedagogy (pp 5172). Geneva: Droz.
DiPietro, R. (1961). Borrowing: Its effect as a mechanism of linguistic change in American
Sicilian. General Linguistics, 5, 3036.
Di Pietro, R. (1964). Learning problems involving Italian [s], [z] and English /s/, /z/. Proceedings
of the Ninth International Congress of Linguists, 1962. The Hague: Mouton.
Fitzpatrick, T., Al-Qarni I., & Meara, P. (2008). Intensive vocabulary learning: A case study.
Retrieved July 20, 2013, from the Swansea Vocabulary Acquisition Research Group Web
site: <http://www.lognostics.co.uk/vlibrary/index.htm>
Foster-Cohen, S. (2001). First language acquisition...second language acquisition: Whats
Hecuba to him or he to Hecuba? Second Language Research, 17(4), 329344.
Fries, C.C. (1945). Teaching and Learning English as a Foreign Language. Ann Arbor, MI:
University of Michigan Press.
Gass, S.M., & Mackey, A. (Eds.). (2012). The Routledge Handbook of Second Language Acquisition. New York, NY: Routledge.
Gee, J. (2013). The anti-education Era: Creating Smarter Students Through Digital Learning. New
York, NY: Palgrave, MacMillan.
Gee, J., & Hayes, E. (2011). Language and Learning in the Digital Age. New York, NY: Routledge.
Givn, T. (Ed.). (2013). Ute Texts. Amsterdam: John Benjamins.

Chapter 10. Interlanguage 40 years on

Goffman, E. (1974). Frame Analysis: An Essay on the Organization of Experience. Cambridge,
MA: Harvard University Press.
Goffman, E. (1981). Forms of Talk. Philadelphia, PA: University of Pennsylvania Press.
Gnzel, S. (2007). Topologie: Zur Raumbeschreibung in den Kultur- und Medienwissenschaften.
Kultur- und Medientheorie. Bielefeld: Transkript Verlag.
Hakuta, K. (1974). A preliminary report on the development of grammatical morphemes in a
Japanese girl learning English as a second language. Working Papers on Bilingualism, 3,
1843.
Hakuta, K. (1975). Prefabricated patterns and the emergence of structure in second language
acquisition. Language Learning, 24, 28797.
Hakuta, K. (1976). A case study of a Japanese child learning English as a second language.
Language Learning, 26, 32151.
Han, Z.-H. (2013). Forty years later: Updating the fossilization hypothesis. Language Teaching,
46(2), 133171.
Han, Z.-H., & Finneran, R. (2013). Re-engaging the interface debate in SLA: Strong, weak, none
or all? The International Journal of Applied Linguistics. doi: 10.1111/ijal.12034.
Han, Z.-H., & Selinker, L. (2005). Fossilization in L2 learners. In E. Hinkel (Ed.), Handbook of
Research in Second Language Teaching and Research (pp. 455483). Mahwah, NJ: Lawrence
Erlbaum Associates.
Haugen, E. (1950). The analysis of linguistic borrowing. Language, 26, 210231.
Haugen, E. (1953). The Norwegian Language in America. Philadelphia, PA: University of
Pennsylvania Press.
Heine, B., & Kuteva, T. (2005). Language Contact and Grammatical Change. Cambridge: CUP.
Jacobs, M. (1926, 1933). Special Collections, University of Washington Libraries [19261971
(Accession No. 1693001) including Microfilm of Jacobss Coos field notes, 19331934
(Accession No. 1693002)]
Kagan, O., & Dillon, K. (2012). Heritage languages and L2 learning. In S.M. Gass & A. Mackey
(Eds.), The Routledge Handbook of Second Language Acquisition (pp.491505). New York,
NY: Routledge.
Kaplan, T., & Selinker, L. (1997). Empty, null, deleted, missing, omitted, absent... items in interlanguage. Review article of Klein 1993. Second Language Research, 13(2), 170186.
Kasper, G., & Blum-Kulka, S. (Eds.). (1993). Interlanguage Pragmatics: An Introduction. Oxford:
OUP.
Kellerman, E. (1989). The imperfect conditional. In K. Hyltenstam & L.K. Obler (Eds.), Bilingualism across the Lifespan: Aspects of Acquisition, Maturity and Loss (pp. 87115). Cambridge: CUP.
Klein, W., & Perdue, C. (1997). The basic variety (or: Couldnt natural languages be much simpler?). Second Language Research, 13, 301347.
Klein, W., & Perdue, C. (1992). Utterance Structure: Developing Grammars Again. Amsterdam:
Benjamins.
Kleinjans, E. (1958). A Descriptive-comparative Study Predicting Interference for Japanese in
Learning English Noun-head Modification Patterns. Unpublished doctoral dissertation, University of Michigan.
Kramsch, C. (1998). Language and Culture. Oxford: OUP.
Kuteva, T., & Heine, B. (2010). Contact and grammaticalization. In R. Hickey (Ed.), The
Handbook of Language Contact (pp. 128). Malden, MA: Wiley-Blackwell.
Lado, R. (1957). Linguistics across Cultures. Ann Arbor, MI: University of Michigan Press.

 Larry Selinker
Landolfi, L. (2012). P..C.E.: An Italian-English Corpus Based on EFL Students. Napoli: Liguori
Editore.
Lane, D. (2013). Screen reader text. <http://dalelane.co.uk/blog/?p=2641> Technical specifications: <http://www-10.lotus.com/ldd/lfwiki.nsf/xpDocViewer.xsp?lookupName=Forms+
Designer+8+documentation#action=openDocument&res_title=Adding_screen_reader_
text_FD_8&content=pdcontent>; blog: <http://dalelane.co.uk/blog/>
Langacker, R. (1991). Foundations of Cognitive Grammar. Stanford CA: Stanford University Press.
Langacker, R. (2008). Cognitive Grammar: A Basic Introduction. Oxford: OUP.
Langacker, R. (To appear). Construal. entry #159, <http://idiom.ucsd.edu/~rwl/rwl_publ.pdf>
Lardiere. D. (2003). The comparative fallacy revisited: A reply to Lakshmanan & Selinker. Second
Language Research, 19, 129143.
Lardiere, D. (2007). Ultimate Attainment in Second Language Acquisition: A Case Study. Mahwah,
NJ: Lawrence Erlbaum Associates.
Lawler, J., & Selinker, L. (1971). On paradoxes, rules, and research in second-language learning.
Language Learning, 21, 2742.
Lenneberg, E.H. (1967). Biological Foundations of Language. New York, NY: John Wiley and
Sons.
Long, M. (2003). Stabilization and fossilization in interlanguage development. In C. Doughty &
M. Long (Eds.), The Handbook of Second Language Acquisition. Oxford: Blackwell.
Meisel, J. (2011). First and Second Language Acquisition: Parallels and Differences. Cambridge:
CUP.
Morgan-Short, K., & Ullman, M. (2012). The neurocognition of second language. In S.M. Gass
& A. Mackey (Eds.), The Routledge Handbook of Second Language Acquisition (pp. 282
300). New York, NY: Routledge.
Nemser, W. (1961). An Experimental Study of Phonological Interference in the English of
Hungarians. Unpublished doctoral dissertation, Columbia University.
Norris, J., & Ortega, L. (2012). Assessing learner knowledge. In S.M. Gass & A. Mackey (Eds.),
The Routledge Handbook of Second Language Acquisition (pp. 573589). New York, NY:
Routledge.
Odlin, T. (2005). Crosslinguistic influence and conceptual transfer: What are the concepts? Annual Review of Applied Linguistics, 25, 325.
Perdue, C. (Ed.). (1984). Second Language Acquisition by Adult Immigrants: A Field Manual.
Nijmegen: Max Planck Institute for Psycholinguistics.
Perdue, C. (1993). Adult Language Acquisition: Cross-linguistic Perspectives. Cambridge: CUP.
Pienemann, M., & Kessler, J.-U. (2012), Processability theory. In S.M. Gass & A. Mackey (Eds.),
The Routledge Handbook of Second Language Acquisition (pp. 506522). New York, NY:
Routledge.
Ritchie, W. & Bhatia, T.K. (Eds.) (2009). The New Handbook of Second Language Acquisition.
Bingley, UK: Emerald.
Robinson, P. (Ed.). (2012). The Routledge Encyclopedia of Second Language Acquisition. New
York, NY: Routledge.
Robinson, P., Mackey, A., Gass, S., & Schmidt, R. (2012). Attention and awareness in second
language acquisition. In S.M. Gass & A. Mackey (Eds.), The Routledge Handbook of Second
Language Acquisition (pp. 247267). New York, NY: Routledge.
Rose, K. (2012). Pragmatics, In P. Robinson (Ed.), The Routledge Encyclopedia of Second Language Acquisition (pp. 501504). New York, NY: Routledge.

Chapter 10. Interlanguage 40 years on

Ross, J.R. (1969). Adjectives as nominal phrases. In D.A. Reibel & S.A. Schane, (Eds.), Modern
Studies in English: Readings in Transformational Grammar. Englewood Cliffs, NJ: Prentice
Hall.
Saporta, S. (1966). Applied linguistics and generative grammar. In A. Valdman (Ed.), Trends in
Language Teaching. New York, NY: McGraw-Hill.
Schmidt, R. (2010). Attention, awareness, and individual differences in language learning. In W.
M. Chan, S. Chi, K.N. Cin, J. Istanto, M. Nagami, J.W. Sew, T. Suthiwan, & I. Walker, Proceedings of CLaSIC 2010, Singapore, December 24 (pp. 721737). Singapore: National
University of Singapore, Centre for Language Studies.
Schmidt, R., & Frota, S. (1986). Developing basic conversational ability in a second language: A
case study of an adult learner of Portuguese. In R.R. Day (Ed.), Talking to Learn: Conversation in Second Language Acquisition (pp. 237326). Rowley, MA: Newbury House.
Schulz, K. (2010). Being Wrong: Adventures in the Margin of Error. New York, NY: Ecco Press.
Schwartz, B. (2013, April). At what level, transfer? Paper presented at the meeting of the
Generative Approaches to Second Language Acquisition, Gainesville, FL.
Schwartz, B., & Sprouse, R. (1996). L2 cognitive states and the Full Transfer/Full Access model.
Second Language Research, 12, 4072.
Selinker, L. (1966). A Psycholinguistic Study of LanguageTtransfer. Unpublished PhD dissertation, Georgetown University.
Selinker, L. (1967). Language contact: Resultant semantic categories in Israeli Hebrew. In D.G.
Stuart (Ed.), Linguistic Studies in Memory of Richard Slade Harrell. Washington, DC:
Georgetown University Press.
Selinker, L. (1969). Language transfer. General Linguistics, 9(2), 6792.
Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics,10, 209231.
Selinker, L. (1979). On the use of informants in discourse analysis and language for specific
purposes. International Review of Applied Linguistics, 17, 189215.
Selinker, L. (1992). Rediscovering Interlanguage. London: Longman.
Selinker, L. (1996). On the notion interlanguage competence in early SLA research: An aid to
understanding some baffling current issues. In G. Brown, K. Malmkjaer, & J. Williams
(Eds.), Performance and Competence in Second Language Acquisition (pp. 89113).
Cambridge: CUP.
Selinker, L. (2007). The Interlanguage Hypothesis: Context, structure and process, where do
we go from here? Special issue of NOA (Norsk som andresprk): Proceedings, Den 7.
konferansen om Nordens sprk som andresprk (Norwegian as a second language), Oslo:
Novus forlag. 931.
Selinker, L. (2011, May). Some unresolved issues in an ELT new media age: Towards building an
interlanguage semantics. Paper presented at the Seminar in Advanced SLA, Teachers
College, Columbia University, NY.
Selinker, L. (2012). Interlanguage In P. Robinson (Ed.), The Routledge Encyclopedia of Second
Language Acquisition (pp. 520523). New York, NY: Routledge.
Selinker, L., & Baumgartner-Cohen, B. (1995). Multiple language acquisition: Damn it, why
cant I keep these two languages apart?. Language Culture and Curriculum, 8(2), 115121.
Selinker, L., & Kinahan, C. (2000). Online interlanguage analysis: Pilgrimage from Santiago. In
I. Plag & K. P. Schneider (Eds). Language Use, Acquisition and Language History: (Mostly)
Empirical Studies in Honor of Ruediger Zimmerman (pp 8389). Trier: Wissenshaftlicher
Verlag Trier.

 Larry Selinker
Selinker, L., & Lakshmanan, U. (1992). Language transfer and fossilization: The Multiple Effects
Principle. In S.M. Gass & L. Selinker, (Eds.), Language Transfer in Language Learning (pp.
197216). Amsterdam: John Benjamins.
Selinker, L., & Naiditch, F. (2004). Reframing grammar and grammar teaching: A Talmudic approach. Rivista di Psicolinguistica Applicata, 4(1), 1339.
Shirai, Y., & Andersen, R. (1995). The acquisition of tense-aspect morphology: A prototype account. Language, 71, 743762.
Siegel, J. (2010). Second Dialect Acquisition. Cambridge: CUP.
Sorace, A. (2011). Pinning down the concept of interface in bilingualism. Linguistic Approaches
to Bilingualism, 1, 133.
Spolsky, B. (1973). What does it mean to know a language? Or, how do you get someone to perform his competence? In J.W. Oller & J. Richards (Eds.), Focus on the Learner: Pragmatic
Perspectives for the Language Teacher (pp. 164176). Rowley, MA: Newbury House.
Tarone, E. (1979). Interlanguage as chameleon. Language Learning, 29, 18191.
Tarone, E. (1988). Variation in Interlanguage. London: Edward Arnold.
Trudgill, P. (1986). Dialects in Contact. Oxford: Blackwell.
Tyler, A. (2012). Cognitive Linguistics and Second Language Acquisition: Theoretical Basics and
Experimental Evidence. New York, NY: Routledge.
VanPatten, B. (2011). Stubborn syntax: How it resists explicit teaching and learning. In C. Sanz
& R. Leow (Eds.), Implicit and Explicit Language Learning: Conditions, Processing, and
Knowledge (pp. 921). Washington, DC: Georgetown University Press.
VanPatten, B., Borst, S., Collopy, E., Qualin, A., & Price, J. (2013). Explicit information, grammatical sensitivity, and the First-noun Principle: A cross-linguistic study in processing instruction. The Modern Language Journal, 97, 504525.
von Humboldt, W. (1836/1960). ber die Verschiedenheit des menschlichen Sprachbaues und
ihren Einfluss auf die geistige Entwickelung des Menschengeschlechts. [On the diversity of human language construction and its influence on human development]. Bonn: Dmmler.
Weinreich, U. (1953). Languages in Contact. The Hague: Mouton.
Weinreich, U. (1980). Problems in the analysis of idioms. In W. Labov & B. Weinreich (Eds.), On
Semantics. Philadelphia, PA: University of Pennsylvania Press.
White, L. (2003). Second Language Acquisition and Universal Grammar. Cambridge: CUP.
Whitney, D. (1881). On mixing in language. Transactions of the American Philological Association, 12, 116.
Williams, J.N. (2012). Working memory and SLA. In S.M. Gass & A. Mackey (Eds.), The Routledge Handbook of Second Language Acquisition (pp. 427221). New York, NY: Routledge.

Name index
A
Abrahamsson 89, 90, 93, 100,
157159, 169
Adamson 181, 183, 198
Adjmian 9, 11, 15, 24, 56, 69,
227, 229, 241
Agard 238, 241
Alario 44
Alatis 225, 241
Alhawary 33, 43
Alim 229, 241
Alonso-Ovalle 108, 124
Al-Qarni 234, 242
Alvarez Torres 170
Amaro 224, 241
Andersen 61, 69, 137, 145, 213,
217, 234, 242, 246
Atkinson 108, 124, 146, 174, 198,
200, 219
B
Baird 162, 164, 169
Bardovi-Harlig 3, 127, 128,
130132, 134136, 138141, 143,
145, 146, 177, 187, 198, 229, 230,
234, 242
Barlow 36, 44
Bates 94, 101, 102
Baumgartner-Cohen 224, 245
Bayley 17, 18, 2426, 183, 198,
201, 217, 227, 231, 242
Beebe 17, 24, 181, 198
Bell 11, 17, 24, 133
Belletti 56, 57, 70
Benmamoun 94, 100
Bennati 70
Berdan 181183, 186, 188, 191,
193195, 198, 213, 217
Berko 162, 163, 169
Bernini 187, 188, 191, 194, 198
Bhatia 25, 70, 74, 95, 102, 201,
228, 244
Bhatt 91, 95, 100, 102
Biber 232, 242
Bigelow 19, 20, 24, 26

Bird 163, 164, 169

Birdsong 11, 24, 47, 66, 68, 70,
155, 156, 169
Bleakney 163, 171
Bley-Vroman vii, 9, 14, 24, 42,
43, 47, 48, 50, 51, 66, 67, 70, 78,
82, 88, 89, 93, 100, 180, 198,
207, 218, 225, 226, 230, 242
Block 60, 108, 124
Blum-Kulka 229, 243
Boas 35, 36, 43
Bodman 141, 145
Boeckx 112, 124
Borst 126, 246
Bohnacker 57, 70
Boustagui 71
Bowles 149, 169
Brire 147, 169
Brindley 180, 201
Brown 24, 72, 94, 176, 180, 198,
201, 203, 245
Burt 75, 101, 141, 146, 175, 181,
199
Bybee 192, 198
Byrnes 176, 198, 210, 215, 217,
218, 220, 234, 242
C
Cadierno 44, 71, 123, 126
Cameron 4, 28, 44, 213, 215, 219
Canagarajah 209, 216, 218
Cancino 178, 179, 181, 189, 191,
193, 199
Caramazza 37, 44
Carroll 61, 70, 178, 199
Castaeda-Jimnez 34, 44
Chambers 229, 242
Chan 55, 71, 89, 93, 101, 245
Chesterman 36, 42, 44
Chomsky 8, 10, 24, 77, 100, 174,
181, 199, 206, 225, 232, 240, 242
Clahsen 77, 82, 89, 100, 176, 179,
180, 199, 200
Clarke 217, 218, 220
Clifton 124

Cohen 145, 227, 242

Collopy 126, 246
Conrad 232, 242
Cook, G. 26, 201
Cook, V.J. 14, 24, 67, 70, 71, 207,
208, 218, 224, 227, 242
Cooreman 129, 145
Coppieters 49, 68, 70, 155, 156,
169
Corbett 36, 44
Corder 3, 8, 11, 24, 75, 76, 81,
100, 101, 152155, 169, 203, 218,
231, 234, 242
Cornish 166, 167, 169
Costa 44, 179
Coulter 84, 100
Craig 38, 44
Croft 213, 218
Crossley 34, 44
Crothers 147, 169
Cruse 213, 218
D
Dahl 130, 145
Davies 146, 201, 208, 218
Deacon 211, 212, 218
de Angelis 218
de Graaff 123, 124
DeKeyser 89, 93, 101, 176, 199
Desmarais 49, 73
Dickerson 17, 24
Dietrich 127, 145, 146
Diffley 154, 171
Dillon 229, 243
Dimitrakopoulou 93, 104
DiPIetro 238, 241, 242
Donaldson 55, 58, 59, 70
Drnyei 149, 169
Doughty 18, 24, 45, 72, 101, 121,
122, 124, 171, 244
Douglas 19, 25, 65, 73, 203, 206,
215, 220
Duff 149, 169
Dulay 75, 101, 141, 146, 175, 181,
199

 Interlanguage Forty Years Later

Dumas 12, 19, 26, 52, 73, 99, 103
E
Earles 165, 167, 170
Eckman 11, 24
Eisenstein 6, 141, 145
Elder 169171, 201
Ellis, N 12, 24, 25, 45, 47, 60, 70,
78, 79, 101, 106, 176, 190, 195,
199, 213, 218
Ellis, R vii, 15, 24, 25, 28, 44,
109, 110, 118, 124, 157, 159,
169171, 176, 195, 199
El Tigi 71
Engestrm 215, 218
Epstein 82, 101
Erlam 169171
Eskildsen 190195, 199, 213,
215, 218
Eubank 11, 24, 82, 101
F
Fasold 17, 25
Fein 72
Feldman 213, 218
Felix 89, 100
Ferguson 36, 44, 145
Fernndez 121, 126
Fernndez-Solera 124
Filiaci 55, 56, 59, 68, 73, 74
Finneran 67, 69, 71, 239, 243
Firth 205, 209, 218
Fitzpatrick 234, 242
Flynn 77, 81, 82, 101, 177, 199, 241
Foster-Cohen 239, 242
Fotos 18, 25, 219
Frazier 124
Freeman 22, 25
Fries 27, 51, 70, 224, 242
Frota 224, 227, 245
G
Garcia Mayo 200
Gass 3, 6, 11, 13, 16, 20, 22,
2426, 66, 6874, 100, 101,
125, 147, 149, 155, 158, 170, 171,
177, 199, 223, 224, 228, 237,
242244, 246
Gatbonton 17, 25, 215, 218
Gee 237, 242
Geertz 216, 218
Geeslin 196, 199
Gilbert 212, 218
Gilmaz 101

Givn 31, 36, 44, 233, 242

Glass 80, 102, 112, 125
Goffman 225, 243
Goldschneider 176, 199
Goo 157, 158, 170
Granena 93, 101
Grosjean 98, 101
Grter 185, 191, 193, 199
Gualmini 185, 199
Gudmestad 183, 196, 199
Gullberg 24, 157, 171
Gnzel 227, 243
Grel 97, 101
Gutirrez 157, 170
H
Haegeman 138, 146
Hakuta 11, 25, 237, 243
Halle 225, 242
Han vii, 2, 13, 25, 3739, 41, 44,
51, 53, 54, 5660, 6372, 76,
8688, 90, 92, 99, 101, 124, 125,
178, 183, 201, 208, 218, 219,
226, 227, 229, 239, 243
Hansen 20, 24, 26, 183, 199
Hansen Edwards 183, 199
Harley 107, 124
Harris 201
Hatch 152, 171, 199
Haugen 226, 243
Hauser 192195, 199
Hawkins 48, 54, 55, 71, 73, 89, 93,
101, 177, 184, 185, 199
Hayes 237, 242
Heine 228, 243
Helmick 163, 171
Helms-Park 29, 33, 44
Hendricks 71
Hendriks 56, 57, 71
Herdina 208, 209, 218
Hernndez 94, 102
Herschensohn 87, 101, 103
Heycock 74
Hickmann 56, 71
Hill 51, 71, 201, 245
Hirsch-Pasek 220
Hoffmeyer 212, 218
Hopp 58, 68, 71
House 196, 199
Huebner 145, 152, 170, 207, 218
Hulstijn vii, 6, 117, 124
Hyltenstam 89, 90, 93, 100,
146, 154, 155, 170, 178, 199,
200, 243

Hymes 181, 200

I
Indefey 171
Ionin 8486, 94, 95, 101, 102
Ioup 47, 49, 71, 89, 100
Ivanov 56, 73
J
Jackson 28, 45
Jacobs 233, 243
Jansen 176, 200
Jarvis vii, 27, 29, 30, 3437,
39, 44
Jegerski 108, 124
Jespersen 36, 44
Jessner 208, 209, 218
Jiang 149, 170
Johnson, J 51, 68, 72, 93, 101, 155,
156, 170
Johnson, K 22, 25
Johnston 180, 201
Jones 165, 167, 170
K
Kagan 229, 243
Kamimoto 28, 44
Kanno 80, 101, 112, 124
Kaplan 232, 243
Kaschak, M 165, 167, 170
Kasper 138, 146, 192, 200, 229,
243
Kaufman 92, 103
Keating 108, 115, 116, 121, 124,
126
Keijzer 97, 101, 102
Keller 56, 73, 74, 212
Kellerman 28, 44, 48, 61, 72, 74,
226, 243
Kersten 165, 167, 170
Kessler 110, 122, 124, 227, 244
Kilborn 129, 145
Kim 96, 101
Kinahan 232, 245
Kirby 166, 169
Klein, E 80, 101 243
Klein, W 127130, 134, 145, 146,
186, 187, 197, 200, 210, 219,
228, 243
Kleinjans 225, 243
Ko 84, 101
Kohnert 94, 101, 102
Kpke 97
Kovac 198

Name index
Kovacs 163, 164, 170
Kramsch 25, 227, 230, 243
Krashen 15, 18, 25, 75, 101, 106,
122, 124, 130, 141, 146, 175,
181, 199
Kull 212, 219
Kuteva 228, 243
Kwak 102
L
Labov 16, 17, 25, 65, 181, 182,
200, 246
Lado 8, 27, 28, 34, 44, 51, 72, 81,
102, 125, 224, 238, 243
Lakshmanan 11, 13, 19, 26, 59,
73, 226, 244, 246
Lamendella 65, 73
Landolfi 233, 244
Lane 237, 244
Langacker 228, 234, 244
Lantolf 210, 219, 220
Lardiere 51, 52, 54, 55, 63, 67,
68, 72, 76, 87, 92, 102, 226,
227, 244
Larsen-Freeman 4, 14, 28, 44,
51, 58, 64, 67, 72, 75, 90, 93,
98, 148, 176, 190, 192, 197, 200,
203, 210, 212219, 223, 228
Larson-Hall 149, 170
Lasky 161, 164, 170
Lawler 147, 170, 228, 233, 244
Lee, M 102
Lee, O S 102
Leech 131, 146
Leeser 115, 116, 121, 126
Lehmann 36, 44
Lenneberg 49, 72, 77, 78, 88, 89,
102, 174, 181, 200, 206, 225,
228, 240, 244
Levelt 65, 72
Lew vii, 65, 71
Li, C 33, 44, 56, 57, 72
Li, M-C 87, 102
Liceras 77, 81, 102, 176, 184,
200, 201
Lieberman 185, 199
Liebner 110, 122, 124
Lightbown vii, 109, 124, 215,
219
Liu 17, 18, 26, 196, 201
Loewen 157, 169171
Long 18, 25, 45, 62, 72, 89, 93,
101, 102, 124, 129, 146, 171, 244
Lowie 189, 190, 201

Lozano 80, 102

Lucy 38, 39, 44
Luk 33, 44, 197, 200
M
Mackey 3, 24, 25, 68, 70, 71, 125,
148, 149, 170, 171, 224, 228,
242244, 246
Makoni, B 209, 219
Makoni, S 209, 219
Mansouri 110, 122, 124
Martohardjono 82, 101
Master 49, 177, 200
Matlin 161, 164, 171
Maturana 212, 219
Maxim 176, 198, 240
McDonough 149, 171
McReynolds 161, 164, 171
McWhorter 90, 92, 100, 102
Meara 234, 242
Medin 72
Meisel 82, 89, 102, 130, 131, 146,
176, 179, 180, 184, 186, 191, 193,
200, 225, 244
Mellow 177, 200
Migge 33, 44
Miozzo 44
Montalbetti 80, 102, 108, 124
Montrul 2, 11, 53, 56, 57, 72, 75,
85, 86, 90, 9397, 100102,
178, 186, 201, 214, 226, 227,
229
Morgan-Short 166, 167, 171,
229, 244
Moselle 71
Muysken 77, 82, 89, 100, 176,
179, 199
N
Naiditch 240, 246
Nassaji 18, 25
Nemser 147, 171, 226, 244
Newport 68, 72, 93, 101, 155, 156,
165, 170, 171
Nielsen 34, 44
Norris 121, 125, 148, 149, 171, 176,
198, 200, 235, 244
Novoa 49, 72
Noyau 127, 145, 146
Noyer 107, 124
O
Obler 49, 72, 146, 199, 243

Odlin 2, 11, 13, 25, 27, 28, 33, 34,

40, 45, 53, 54, 59, 71, 72, 76, 86,
197, 219, 224, 227, 229, 231, 235,
241, 244
ODonnell 199
OGrady 93, 95, 102, 178, 200
Orr 29, 31, 32, 34, 36, 45
Ortega 4, 61, 67, 75, 86, 90, 93,
98, 109, 121, 125, 148, 149, 171,
173, 176, 179, 197, 200, 207,
208, 210, 219, 220, 223, 225,
228, 229, 235, 244
P
Paik 66, 68, 70
Paradis 15, 25, 89, 102
Pavlenko 37, 44
Pawley 144, 146
Perales 184, 200
Perdue 186, 187, 197, 200, 228,
243, 244
Prez-Leroux 80, 102, 112, 125
Perpian 81, 97, 102, 103
Prvost 55, 72, 178, 201
Philp 169171
Pienemann 124, 176, 179181,
186, 191, 200, 201, 227, 244
Polinsky 9396, 100, 103
Polio 3, 13, 16, 20, 66, 147, 223,
237
Pomerantz 136, 146
Preston 17, 18, 21, 2426, 72, 183,
198, 201, 209, 217, 220
Price 85, 126, 246
Prior 216, 220
Py 98, 101
Q
Qualin 126, 246
R
Racsmany 163, 164, 170
Rampton 17, 25, 196, 201
Read 162, 164, 171
Rebuschat 159, 171
Reder 190, 201
Regan 183, 198
Reinders 169171
Ringbom 32, 35, 39, 45
Ritchie 25, 70, 74, 228, 242, 244
Roberts 157, 158, 171
Robinson vii, 12, 25, 45, 70, 72,
118, 123, 125, 198, 219, 224, 228,
239, 244, 245

 Interlanguage Forty Years Later

Roland 71
Rmer 199
Rosansky 178, 199
Rose 146, 229, 244
Rosn 57, 70
Ross 232, 245
Rothman 88, 103, 106, 109, 118,
122, 125, 126, 241
Roussou 89, 103
Rutherford 56, 73, 212, 220
S
Sabourin 29, 36, 45
Saffran 170
Salsbury 136, 145, 146
Snchez-Walker 96, 97, 102
Sannino 215, 218
Sanz 125, 166, 171, 246
Saporta 228, 245
Sarkar 212, 218
Sato 129, 146
Schachter 11, 25, 28, 34, 44, 45,
56, 67, 70, 7274, 89, 90, 100,
103, 154, 171, 242
Schiller 44
Schleppegrell 214, 220
Schmid 97, 101103
Schmidt 15, 25, 47, 50, 73, 125,
141, 146, 224, 227, 244, 245
Schneiderman 49, 73
Schulz 157, 159, 171, 238, 245
Schumann vii, 47, 50, 73, 178,
179, 182, 183, 186, 194, 195, 198,
199, 201
Schtze 153, 171
Schwartz 76, 82, 103, 106, 111,
125, 226, 227, 245
Segalowitz 215, 218
Seidlhofer 26, 208, 220
Selinker vii, 116, 19, 20, 22,
23, 2527, 35, 39, 42, 43, 45,
4749, 52, 53, 58, 59, 6569,
72, 73, 7579, 81, 8492, 94,
98, 99, 101, 103, 105, 106, 110,
112, 114, 117, 122125, 127,
128, 144, 146154, 159, 160,
162, 164, 167171, 173175,
180, 181, 183, 186, 192, 194,
196198, 201, 203206, 213,
215, 220228, 230232, 234,
236240, 243246
Silva-Corvaln 90, 93, 103

Serratrice 56, 59, 68, 73

Setzler 201
Sharma 91, 92, 103
Sharwood Smith 74, 106, 122,
125
Sheen 64, 73
Shenkman 72
Shimura 28, 44
Shirai 33, 44, 197, 200, 234, 242,
246
Siegel 229, 246
Skehan 59, 73, 123, 125
Slabakova 56, 72, 73, 83, 84, 103,
178, 201
Slobin 35, 36, 45, 64, 65, 73
Smith, H. 51, 74
Smith, L. 220
Smith, M C 112, 114, 116, 121,
124, 125
Smitherman 229, 241
Solomon 162, 164, 171
Sorace 5456, 59, 68, 70, 73, 74,
89, 97, 103, 225, 246
Spada vii, 15, 109, 121, 124, 125
Spolsky 70, 230, 246
Sproat 107, 125
Sprouse 76, 82, 103, 145, 227, 245
Stang 161, 164, 171
Stapleton 211, 220
Stauble 178, 179, 201
Steinhauer 166, 171
Strapp 163, 164, 171
Suppes 147, 169
Swain 12, 19, 26, 52, 73, 99, 103
Swierzbin 1, 20, 22, 26, 146
Syder 144, 146
T
Tabakowska 43, 45
Tamariz 166, 169
Tarone vii, 1, 6, 7, 1620, 22, 24,
26, 37, 61, 65, 73, 76, 86, 124,
146, 148, 152, 154, 155, 171, 183,
195, 196, 198, 201, 214, 220, 227,
231, 242, 246
Thibault 214, 220
Thomas 26, 75, 103
Thomason 90, 92, 103
Thompson 33, 44, 56, 57, 72, 211,
212, 220
Thorne 210, 216, 219, 220
Tomita 121, 125

Tonkovich 163, 171

Toribio 111, 125
Towell 48, 73
Trager 51, 74
Trahey 64, 74
Trvise 56, 74
Trvisiol 57, 74
Trofimovich 149, 171
Trudgill 92, 103, 229, 242, 246
Truscott 11, 26, 106, 121, 122, 125
Tsimpli 56, 74, 89, 93, 103, 104
Tucker 212, 220
Tyler 228, 246
Tyson 154, 171
U
Ullman 15, 26, 89, 104, 166, 171,
229, 244
V
Vainnika 104
VanBuren 50, 51, 74
Van Dijk 189, 190, 194, 196, 201
Van Lier 215, 220
VanPatten 3, 15, 18, 23, 26, 48,
70, 74, 78, 101, 105, 106, 108,
109, 112, 115, 116, 118, 120126,
145, 148, 174, 178, 185, 191, 199,
201, 210, 212, 216, 219, 223,
227, 246
Varela 211, 212, 219, 220
Verspoor 189, 190, 194196, 201
Vertovec 208, 220
von Humboldt 224, 246
von Stutterheim 128, 129, 146
W
Wade 212, 220
Wagner 152, 171, 192, 200, 209,
218
Wagner-Gough 152
Wang 51, 74
Wardhaugh 28, 45
Webb 163, 164, 171
Weber 211, 220
Weinreich 90, 104, 224, 226, 231,
236, 246
Wexler 84, 101
White 11, 26, 51, 55, 56, 64, 68,
72, 74, 7678, 8082, 88, 104,
111, 126, 201, 225, 227, 246
Whitman 28, 45

Name index
Whitney 224, 238, 246
Widdowson 26, 214, 220
Williams, J. 18, 23, 24, 26, 48,
70, 72, 74, 101, 122, 126, 145,
163, 164, 169, 174, 199, 201, 219,
224, 245, 246
Williams, F 18, 23, 24, 26, 48,
70, 72, 74, 101, 122, 126, 145,
163, 164, 169, 174, 199, 201, 219,
224, 245, 246

Wilson 56, 74
Y
Year 207, 208, 220
Yip 56, 74
Yoon 96, 101
Yorio 141, 146
Young 17, 26
Young-Scholten 20, 103, 104

Z
Zobl 176, 201
Zuengler 17, 24

Subject index
A
accuracy 4, 14, 16, 23, 35, 51,
87, 88, 173, 175, 176, 179, 181,
197, 228
adaptation 211213, 216
adverb, adverbial 63, 88, 115, 116,
129134, 139, 143, 158, 167
affordances 173, 192195, 213, 215
articles 27, 2931, 3338, 4043,
51, 63, 8486, 91, 92, 94, 95,
143, 167, 177
artificial language 4, 147, 160,
165168
aspect 3, 83, 84, 128, 130,
132134, 177, 234, 235
attention 14, 1619, 40, 53, 59,
60, 65, 161, 228
attrition 55, 56, 97
autopoietic 211, 212, 220
B
basic variety 186188, 194, 195,
228
bilingual 14, 17, 35, 55, 56, 7577,
90, 93, 9699, 184, 198, 210
bilingualism 2, 3, 71, 75, 76, 89,
93, 95, 96, 98100, 238
C
children 9, 10, 12, 47, 48, 57,
7780, 87, 93, 96, 97, 99, 114,
156, 161, 162, 165, 176, 178, 184,
185, 187, 189, 232
chronological order 130135, 144
comparative fallacy 180, 207,
230, 233
competence 3, 4, 8, 13, 14, 4750,
53, 66, 77, 78, 83, 86-90, 92, 93,
99, 105107, 110, 111, 117, 120,
122, 123, 154, 156, 159, 184, 186,
197, 206-209, 216, 239
see also knowledge system
multi-competence

complex adaptive system 4,

166, 213
computational linguistics 237,
241
concept-oriented approach 3,
128, 129
conventional expression 127,
128, 140-144
copula 91, 109, 115, 121, 180, 188
D
determiner (definite,
indefinite) 32, 37, 40, 41, 88,
91, 178,
developmental sequence/
stage 110, 135, 144, 176,
179181
dialect 7, 14, 86, 90, 111, 229
dialectal 91, 9699
see also interdialect
differential failure 54
differential object marking
(DOM) 96, 98, 99
differential success 4751, 60,
61, 63, 68
E
early bilinguals/
bilingualism 94-96
endpoint 4, 83, 97, 203, 209,
217
endstate 51, 63, 6668
error 8, 19, 22, 28, 29, 40, 67, 75,
80, 81, 84, 86, 8892, 97, 99,
135, 141, 153, 162, 222, 225, 227,
232, 233
covert 231
explicit
instruction 15, 18, 66, 80, 110,
111, 115, 120, 121, 123
learning 99, 111, 117, 120,
122, 239

knowledge 15, 16, 18, 24, 66,

118, 124, 159, 169171, 239
memory 99, 163, 164
rule 109, 121
F
first language (L1) 3, 8, 10, 11,
34, 47, 52, 53, 55, 60, 77, 81, 85,
87, 9799, 108, 128, 149, 176,
225, 239
formal instruction 15, 18, 20, 108,
110, 114, 115, 117, 118, 122, 210
see also explicit instruction
formal rules 18, 20
see also explicit rules
formal style 1618
formulas 140, 141, 180, 183, 190,
192, 194, 195, 214
formulaic 127, 137, 141, 180,
192
fossilize, fossilization 7, 9, 1214,
23, 39, 4769
selective 47, 5456, 60, 63,
66, 69
frequency 34, 50, 58, 6063, 108,
142, 161, 163, 164, 176, 187189
Full Transfer/Full Access
Hypothesis 82, 103
functional/ functionalist 81, 159,
176, 184, 188, 193, 195
functional approach 3, 107, 128,
129, 186, 187
Fundamental Difference
Hypothesis (FDH) 47, 78,
88, 89, 93
G
gender 2934, 36, 37, 95, 96, 109,
111, 118, 121, 158, 166,
grammaticality judgment 9,
11, 1316, 20, 49, 58, 120, 147,
150, 151, 152160, 164, 166,
168, 223

 Interlanguage Forty Years Later

H
heritage
language 9395, 229
speaker 3, 77, 9398
I
idiosyncrasy 2, 27, 39, 40, 41,
54, 59
illiterate 17, 20, 23, 92
implicit
instruction 121
knowledge/system 15, 16, 18,
77, 106, 118, 159, 170, 239
learning 99, 239
memory 99, 163
incomplete acquisition 72, 76,
90, 92, 93, 95, 97, 99, 102, 103
individual difference 62
inflectional morphology
(agreement, tense)
informal
context 18
style/use 18, 58
input processing 3, 21, 105, 117,
120, 123
input robustness 61, 62, 64
instructed second language
acquisition (instructed
SLA) 105, 114, 117, 121, 122,
123, 224
instruction 2, 3, 5, 15, 18, 20, 52,
68, 69, 78, 80, 86, 88, 105123,
205, 210, 214, 217, 227
interculture 229, 230, 241
interdialect 229, 230, 241
interlingual identification 2, 9,
14, 37, 90, 175, 224
interliteracy 229, 230, 241
intersystem 229, 230, 241
interrogative 50, 178
intuition 153156, 158, 159, 168,
231, 236

language norm 2, 14, 17, 18, 20,

21, 47, 58, 67, 69, 78, 180, 197,
205210, 236, 240
language transfer 2, 8, 11, 2743,
52, 53, 56, 57, 59, 60, 64, 65,
75, 76, 8192, 94100, 176,
223, 224, 226228, 232, 235,
236, 238
languaging 214, 217
latent language structure 77, 78,
88, 89, 94, 228, 240
latent psychological structure
(LPS) 2, 11, 27, 35, 37, 39, 43,
5254, 77, 78, 89, 228, 240
lexicalization 232
literacy 2, 5, 7, 19, 20, 215
see also interliteracy
literate 19, 20, 23, 52
M
markedness 62, 63, 64
memory 39, 62, 148, 169, 170,
223, 224, 229
associative 49
long-term 224
short-term 163, 164
working 20, 115, 158, 224
see also explicit memory,
implicit memory
mental representation 3, 10, 89,
105108, 117, 120, 122
Missing Surface Inflection
Hypothesis 55
Morphological Misreading
Hypothesis 55
multi-competence 14

K
knowledge system 1416, 18, 223

N
negation 109, 115, 121, 178194
non-native varieties 9092, 99
nonsense syllables and
words 14, 147, 150152,
160164, 167, 168
noticing 15, 17, 21, 22, 120, 227,
228

L
language change 3, 18, 21, 76, 90,
98, 99, 166, 214
language contact 33, 34, 89, 90,
100, 223

O
Observers Paradox 16
order of acquisition 18, 141
see also developmental
sequence/stage

P
performance 1114, 16, 33, 37, 52,
53, 58, 66, 7678, 85, 87, 89, 127,
144, 149, 150, 153, 156, 159, 160,
162, 164, 165, 186, 204208,
210, 216, 228231, 235, 239, 240
permeable/permeability 11, 15,
52, 56, 227, 229
poverty of the stimulus
(POS) 108-109, 111114, 185
pragmatics 35, 66, 94, 127129,
135138, 141, 144, 229, 239
see also syntax-pragmatics
Processability Theory 21, 48,
107, 176, 179181, 186
R
relative clause 28, 32, 51, 80, 81,
88, 131133, 155, 158, 177
reverse-order report 130135
S
semantics, 35, 49, 65, 85, 129,
221, 222, 232237
social context 16, 17, 18, 65, 135,
142, 181, 182, 196
speech act 136, 144
speech community 18, 20, 21,
98, 99, 183, 209, 212, 213
staged development 109111,
114, 115, 121
see also developmental
sequence/stage
style-shifting 1618
see also variation,
sociolinguistic
syntax-discourse 55, 57, 58, 64
syntax-pragmatics 55, 57, 58
T
Talmud 10, 23, 240
teleology, teleological 14, 181,
192, 197, 203, 210214, 217
telic, telicity 83, 84
temporal, temporality 127, 128,
130132, 135, 144, 234
tense, verb 87, 88, 91, 92, 110, 115,
118, 119, 128, 130135, 177, 235
present 115, 117
past tense 87, 88, 91, 92, 110,
115, 163

Subject index
transfer, language 8, 11, 2743,
52, 53, 56, 57, 59, 60, 64, 65,
75, 76, 8192, 94100, 176,
223, 224, 226228, 232, 235,
236, 238
transfer of training 27, 52, 232
transitional competence 8

U
UG (Universal Grammar) 10
12, 55, 68, 75-82, 8890, 99,
118120, 156, 167, 178, 184186,
225, 227, 230, 240
ultimate attainment 51, 60,
6769, 83, 8688, 93, 99, 155,
158
usage-based approach 78, 79,
173, 176, 189193, 195, 197

V
variation
individual 20, 37, 40
sociolinguistic 1618, 155, 183,
195, 196, 227
see also style-shifting
variationist research 1618,
181183, 187, 188, 194196