Dorothea Hoffmann & Michael Ramsammy
Proceedings to the
18th International Postgraduate Linguistics Conference
Edited by Dorothea Hoffmann and Michael Ramsammy
6 May 2009 University of Manchester, UK
School of Languages, Linguistics and Cultures
University of Manchester
Oxford Road
Manchester, M13 9PL
United Kingdom
School of Languages, Linguistics and Cultures
Tel: +44 161 306 1240
SALC@manchester.ac.uk
http://www.alc.manchester.ac.uk
© 2013 The Authors
No part of this publication may be reproduced, stored in a retrieval system, or transmitted,
on any form or bay any means, electronic, mechanical, photocopying, recording or
otherwise, without the prior permission of the author(s) of that part of the publication,
except as permitted by UK copyright law.
ISBN: 978-0-9576682-3-2
This publication can be cited as:
Dorothea Hoffmann & Michael Ramsammy (eds). 2013. Proceedings to the 18th
International Linguistics Postgraduate Conference. Manchester: University of
Manchester
Contents
0. Introduction
DOROTHEA HOFFMANN & MICHAEL RAMSAMMY
The University of Manchester
iii
1. The semantics of three posture verbs gã ‘lie’, zi ‘sit’ and ze ‘stand’ in Gurenε:
A Cognitive Linguistics perspective
SAMUEL ATINTONO
The University of Manchester
1
2. Semantic Change – The evolution of lexical meaning in time and space: An
example from the Greek Language: The word µελαγχολία (Melancholy)
AMALIA KAZIANI
Aristotle University of Thessaloniki
14
3. Presently comparing pre-past: The expression of ‘past-perfectness’ in English and
Italian
SUSANNE SCHNEIDER
Freie Universität Berlin / University Ca’ Foscari of Venice
25
4. Are creoles tenseless languages?: A review of the creole tense and aspect system
ESTHER NÚÑEZ VILLANUEVA
The University of Manchester
47
5. How similar is a Belfast final rise to a Cambridge final fall?
JENNIFER SULLIVAN
University of Edinburgh
67
6. Against the “West Germanic Syntax” hypothesis: The V-2 constraint in Old
English and Old High German
ANNA CHICHOSZ
University of Łódź
82
7. Which-phrases do not move
BENJAMIN KRATZ
Johann Wolfgang Goethe-Universität Frankfurt
95
8. Adult Root Infinitives
NEVEN WENGER
Johann Wolfgang Goethe-Universität Frankfurt
115
~ ii ~
INTRODUCTION
PROCEEDINGS TO THE 18TH INTERNATIONAL POSTGRADUATE CONFERENCE IN
LINGUISTICS AT THE UNIVERSITY OF MANCHESTER
DOROTHEA HOFFMANN & MICHAEL RAMSAMMY
The University of Manchester
This collection of papers is a direct result of the 18th Postgraduate Conference in
Linguistics held at the University of Manchester on May 6th 2009.
In March 1992 the Department of Languages, Linguistics and Cultures at The University
of Manchester hosted its first Postgraduate Linguistics Conference (PLC). Since then this
annual event has grown, now drawing participants from across the world to exchange
ideas in a relaxed but stimulating environment. This makes PLC the longest continuously
running event of its kind in the UK.
The aim of the Conference was to bring together postgraduates from within the various
areas of the discipline of Linguistics and to allow them to present papers to their peers.
The plenary speakers for the event were Prof David Crystal from the University of
Bangor talking about ‘Language Death’ and Prof Eva Schultze-Berndt from Manchester
University on ‘Towards a Typology of overt verb classification’.
The following papers are a collection from speakers presenting at the conference covering
a wide range of topics within general linguistics, including phonetics, semantics, syntax,
and historical linguistics covering a number of different languages such as English,
Italian, German, Gurenε and Creole languages.
Samuel Atintono presents in chapter 1, “The Semantics of Three Posture Verbs gã ‘lie’, zi
‘sit’ and ze ‘stand’ in Gurenε” an analysis of posture verbs in a Gur (Niger-Congo)
language spoken in Northern Ghana. Within a cognitive linguistic framework and
drawing on a number of crosslinguistic typological studies, Atintono describes the
semantics and usage of the three posture verbs verbs gã ‘be lying’, zi ‘be sitting’, and ze’
‘be standing’. He places particular focus on the extension of meanings these basic posture
verbs have gone through in the language. They specifically encode spatial configurational
properties reflecting cultural and experiential realities of the speakers of the language. For
example the verb ze’ is associated with situations of unrest and effort. A person
experiencing any kind of troubles might be referred to as ‘standing’ indicating that their
mind might not be at ease and uncomfortable. He then concludes suggesting that in order
to describe the semantics of posture verbs in an adequate manner, one must consider not
only its most common basic uses, but also keep socio-cultural aspects in mind.
~ iii ~
In chapter 2 “Semantic Change – the evolution of lexical meaning in time and space”,
Amalia Kaziani reflects on the dynamic character of semantic change. She considers a
case study of the Greek word for ”melancholy” exploring its lexical changes across time
and space. The lexeme can be traced back to Ancient Greek and has extended its original
medical meaning to the mental health field and into today’s conventionalized usage.
Kaziani used extensive corpus data to prove that the original meaning of the word has
been lost in today’s usage.
Susanne Schneider analyses the expression of ‘past-perfectness’ in English and Italian in
chapter 3 “Presently comparing Pre-Past”. She investigates how the two languages code
past-perfectness. As a result of careful language-particular coding strategies, she was able
to compile a comprehensive record of similarities and differences between the languages’
use of past-perfectness. She concludes that the conceptual space of past-perfectness is
accessible by means of a dedicated Perfect-Past (PRF-PST) marking device that is used
differently in English and Italian.
In chapter 4, Esther Núñez Villanueva asks “Are creoles tenseless languages?” reviewing
the creole tense and aspect system. She analyses data from four different creole
languages, namely Guyanese Creole English, Haitian Creole French, Papiamentu Creole
Spanish, and Kituba and concludes that creoles are aspect-prominent, but not tenseless
languages. They tend to display a tripartite system of perfective and past and present
imperfective. Any differences between the languages can be accounted for by various
grammaticalization stages.
Jennifer Sullivan’s paper on “How similar is a Belfast final rise to a Cambridge final
fall?” in chapter 5 explores the possibility of the puzzling Belfast final ‘rise’ having
derived historically from final ‘falls’. Her results, however, show that the Belfast rise is
more similar to the timing of the Cambridge question ‘rise’ than a ‘fall’. As a tool of
analysis she uses an approach to quantification of similarity that has so far received little
attention.
In chapter 6, Anna Cichosz argues “Against the ‘West-Germanic syntax’ hypothesis”
based on an extensive corpus study of Old High German and Old English. She claims that
the syntax of both languages needs to be regarded as two independent systems starting to
develop before 1066 in the old Germanic period. She bases her corpus on a selection of
texts from both Old English and High German covering all of the main text types and thus
eliminating differences due to stylistic constraints. She concludes in stating that the V2constraint was present in both languages from an early stage, but that the degree of its
influence was different. It appears that the V-2 constraint was already a well-developed
phenomenon in Old High German. However, in Old English it never reached the status of
a rule and eventually disappeared from the syntactic system of English.
Benjamin Kratz in chapter 7 “Which-phrases do move” presents an approach where he
combines existing work on d-linked wh-phrases (DWH) with van Craenenbroek’s (2008)
original idea that the wh-word and are separate items. He presents work in progress some
puzzling aspects of DWH phrases such as theta-assignment, selection and reconstruction.
He presents shortcomings of van Craenenbroek’s analysis and provides empirical
examples challenging the approach. These are Doubly-filled COMP phenomena in
Frisian and dialectal Dutch, Swiping in English, Preposition Stranding and Free Relatives
in Dutch, and spading in dialectal Dutch. He then presents his own study of DWH phrases
~ iv ~
firstly by suggesting an analysis of DWH as topics and secondly by claiming a null—
head for wh-phrases.
The final chapter 8 by Neven Wenger is concerned with “Adult Root Infinites” (ARI), a
variety of infinitival structures occurring in root. They are different from other root
infinitives in their pragmatics which mark the speakers’ incredulity towards the
proposition of a previous utterance. Furthermore, they are available cross linguistically
showing morphosyntactic variation in this realm. Wenger analyses ARI within a
minimalist approach and provides a sketch of a syntax of (non)finiteness. The study
concludes in arguing for the possibility of analysing ARIs within a syntactic framework,
contrary to previous approaches. Furthermore, the complexity of ARIs qualifies them to
meet the requirements of nonfiniteness.
Overall these papers provide a fascinating insight into the work carried out by young
researchers at a postgraduate level at institutions in the UK, Germany, Poland, Italy, and
Greece. The variety and quality of topics present an enjoyable collection into many
aspects of linguistics.
~v~
CHAPTER 1
THE SEMANTICS OF THE THREE POSTURE VERBS GÃ ‘LIE’, ZI ‘SIT’ AND ZE
‘STAND’ IN GURENƐ
A COGNITIVE LINGUISTICS PERSPECTIVE
SAMUEL ATINTONO
The University of Manchester
This article discusses the meaning and use of three human posture verbs gã ‘be lying’, zi ‘be sitting’,
and ze’ ‘be standing’ in Gurenɛ, a Gur (Niger-Congo) language in Northern Ghana, West Africa.
The article specifically analyses the components of the meanings of these posture verbs from a
cognitive linguistics perspective. It is shown in this article that apart from their central meanings of
sitting, lying and standing these posture verbs encode spatial configurational properties that reflect
the experiential realities of the speakers of the language. The paper also discusses the extension of
the meanings of the three posture verbs to express the location of other inanimate figures such as
bottles, sticks, clothes and balls. It is argued in this paper that this extension of the basic meaning of
the posture verbs provides us with the conceptualization of the speakers in categorising other
entities. The data itself comes from the use of positional verb picture stimuli designed by experts at
the Max Planck Institute of Psycholinguistics (MPI) supplemented with real objects and spontaneous
posture tokens collected from natural speech contexts.
1 INTRODUCTION
The paper discusses the meaning and use of three human posture verbs gã ‘be lying’, zi
‘be sitting’, and ze’ ‘be standing’ in Gurenɛ, a Gur (Niger-Congo) language in Northern
Ghana, West Africa. I examine the components of the meanings of these posture verbs
from a cognitive linguistics perspective. This approach provides us with an opportunity to
appreciate the fact that the meanings of these verbs have spatial and cognitive
underpinnings. It is shown in this paper that apart from their central meanings of sitting,
lying and standing they encode spatial configurational properties that reflect the
experiential realities of the speakers of the language. The extension of the meanings of
the three posture verbs to express the location of other inanimate entities such as bottles,
sticks, clothes and balls is also discussed. Thus, the extension of the basic meaning of the
posture verbs provides us with an understanding of the conceptualization of the speakers
in categorising other entities. Posture expressions in human languages show a strong
lexical and semantic domain which requires an in depth investigation to properly
understand the semantics and the grammar (cf Schaefer & Egbohare 2008:215).
~1~
Samuel Atintono: The semantics of three posture verbs in Gurenɛ
Earlier studies on posture verbs from both typological and cognitive perspectives on
diverse languages (Newman 2002; Ameka and Levinson 2007) suggest that these posture
verbs encode concepts which are grounded in human experience. Two interesting
hypothesis emerged from this studies: First is the observation that posture verb properties
will vary greatly across languages, showing extreme variability and the second hypothesis
is that they will differ little, manifesting great similarity. The present analysis aims not to
test these claims but to provide evidence of the conceptualization patterns of the posture
verbs in Gurenɛ.
2 MOTIVATION FOR THE POSTURE VERBS IN GURENƐ
The posture verbs in Gurenɛ has never been explored to details to the best of my
knowledge except a short article by Atintono (2004) in which he discussed the semantics
and syntax of three posture verbs in Gurenɛ within Talmy’s model of motion events. This
was a good starting point but the paper itself is limited in scope and content. This present
study is of interest because its contribution has the potential of expanding our knowledge
on the semantics, and pragmatics of the Gur languages which are little studied. Newman
(2002), Ameka & Levinson (2007) tended to ignore the semantics of the posture verbs in
figurative contexts of which this present article provides a good account utilizing
cognitive linguistics concepts. The paper will have implications for cross-linguistics
typological claims on spatial cognition studies. As Ameka and Levinson (2007:850)
conceded the typological sample of the languages investigated so far for their postural
properties is limited and there is the need to expand the scope to ensure a wider analysis
to arrive at firm theoretical conclusions.
3 AN OVERVIEW
APPROACH
OF THE
ASSUMPTIONS
OF THE
COGNITIVE LINGUISTICS
In this section I provide an overview of the cognitive linguistics approach adopted for the
analysis of the three posture verbs in this article. It is anticipated that this will help us
appreciate the analysis of the data in the subsequent sections. Cognitive linguistics is
often considered as an approach and not a theory in its own right (Vyvyan & Green
2007:3, Geeraerts & Cuyckens 2007:4). The approach began in the late 1970s and early
1980s pioneered by Langacker (1987; 2000), Lakoff (1980; 1990) and further exemplified
by Gibbs (1996), Talmy (2000a, 2000b), Taylor (2000), Lee (2001), Croft and Cruse
(2004), Vyvyan and Green (2007). For lack of space I will not attempt to describe the
diverse range of complementary and overlapping theories of the cognitive linguistics
approach such as cognitive grammar, cognitive semantics, grammars of space and the like
that together constitute the cognitive linguistics approach. Instead I will provide the basic
principles and assumptions underpinning all these strands that collectively make up the
cognitive linguistics approach. The approach views language as a tool for organizing, and
conveying information conceptually (cf Geeraerts&Cuyckens 2007:3). In other words it is
concerned with the conceptual and experiential basis of language in the mind of the
~2~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
speaker. In this regard language is considered to be engrained in the cognitive capacity of
man. The central issue here as Casad (1996:1) observes is that language use is arguably
grounded in our daily experience of the real world. As it will be shown in the discussion
of the semantics and conceptualization patterns of the three Gurenɛ posture verbs their
application to the socio-cultural context reflects the conceptualization of the matrix of the
life of the speakers. The implication is that for us to fully understand the meaning of the
Gurenɛ posture verbs in this context we need to invoke some cognitive theoretical
constructs in order to provide a credible and satisfactory account of the phenomenon. In
what follows I now try to outline the key principles and assumptions of the cognitive
linguistics paradigm as conceived by the various cognitive linguists noted above.
One of the most probable important assumption or hypothesis that binds the diverse
cognitive linguistic theories and guides the approach is the orientation that natural
language can be adequately explained in terms of its semantics and function rather than
describing linguistic expressions in terms of formal rule system that is completely
independent of meaning. Thus the conception of meaning or language use is one of the
fundamental concerns of the cognitive linguistic approach. It is this view that sometimes
makes others to call it a usage-based approach. They reject the autonomous cognitive
faculty hypothesis proposed by generative grammarians especially in the Chomskyian
formalism. Instead cognitive linguists conceive knowledge of the world to be mediated
through language but generative linguists on the other hand concentrate on knowledge of
language and its mental representations (Geeraerts&Cuyckens 2007:7). As Croft and
Cruse (2004:2) notes the representation of linguistic knowledge in the human cognitive
faculty is essentially the same as the representation of other knowledge structures.
Language cannot therefore be said to constitute a separate cognitive faculty.
Another major assumption of the cognitive linguistics approach is that the grammar of a
language is conceptualization. This principle seeks to explain that a particular linguistic
structure or expression is linked to a particular way of conceptualizing a given situation
(Lee 2001:1). Language description is therefore motivated by the human conceptual
knowledge and experience of the world (cf Gibbs 1996:27). A key component of the
human faculty is therefore the conceptualization of our experiences and how this is
expressed through language. The linguistic knowledge that we possess in our minds is
therefore rooted in our conceptual system (see Croft and Cruse 2004:3, Langacker 2000).
The conceptualization principle further predicts that there is no isomorphic mapping of
elements of the external world onto linguistic form instead every situation can be
construed in different ways and that different ways of encoding a situation constitute a
different conceptualization. Linguistics structures therefore encode conceptualizations
which go beyond simple reference (Geeraerts&Cuyckens 2007:14). Rather
conceptualization is associated with both socio cultural and physiological base of human
experience. By this cognitive linguistics is thought to be concern with the full conception
of meaning and manifestation of some fundamental properties of the human mind as
viewed through the world. The posture verbs which are of interest in this article encode
spatial notions which are fundamentally rooted in the conceptual thought patterns of the
speakers and this explains the cultural variations of postural expressions across languages
~3~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
(see Levinson 2001, Levinson & Wilkinson 2006). We shall adopt these principles in the
analysis of the posture verbs in the rest of the paper.
4 THE SEMANTICS OF THE POSTURE VERBS; GÃ, ZI, ZE’
The three posture verbs have their basic meanings as well as their grammaticalized or
figurative extensions. In their basic usage they describe basic human postures with the
following central meanings; gã ‘to be in a lying position, zi ‘to be in a sitting position’,
and ze’ ‘to be in a standing position’. These three postures play an important role in our
every day activities and the verbs which encode them constitute a rich semantic class for
analysis. In examples (1) to (3) the posture verbs express the basic usage of these verbs.
In this sense they encode the distinct spatial configurations of the human body. So in (1)
the child’s body is in a horizontal contact with the mat while in (2) the woman’s lower
torso is resting on the chair. Example (3) shows that the man is in a vertical position and
is being supported by his feet. Notice that all the three verbs have stative interpretation
with the human entities. That is the human participants are in a state situation devoid of
any action or activity.
1.
Bia
la
gã
la
suŋɔ
Child DEF lie
FOC mat
‘The child is lying on the mat.’
puan
LOC
2.
Pɔka
la
zi
la
Woman
DEF sit
FOC
‘Woman is sitting on the chair.’
kuka zuo
chair head
3.
Budaa la
ze’
la
yiŋa
Man DEF stand FOC outside
‘The man is standing outside.’
The extension of the usage of these posture verbs in the figurative sense are expressed in
the following examples.
4.
Asɔ’ɔŋa
gã
la
na’am
zuo
Rabbit
lie
FOC chieftaincy
head
‘Mr. rabbit is enjoying wealth.’ (folktale text)
5.
A
zi
la
3SG sit
FOC
‘He stays at Accra.’
6.
Budaa la
ze’
la
Man DEF stand FOC
‘The man is in trouble.’
Ankara
Accra
yele
problem
~4~
puan
LOC
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
In examples (4) to (6) the meanings of the three posture verbs express the maintenance of
a pre-existing stative condition (see Scahefer & Egbohare 2008:217). They do not suggest
any relationship with a physical body positioning of the entities. For example in (4) it
does not suggest that the rabbit is physically lying on wealth. It only conceptualizes the
condition that he is experiencing. Similarly (5) expresses the fact that the person stays in a
place but does not imply that he is in a sitting posture. In (6) the man is said to be
literally standing in a problem. What this means is that when someone is in trouble that
person is naturally an unsettled person and this is conceptualized as been in a standing
posture. The fact is that a standing position does not provide rest and a person in trouble
is not at peace with himself. A much detail discussion is provided in the section on socio
cultural domain. This latter usage in Gurenɛ requires a cultural conceptual knowledge in
the semantics and pragmatics as well as the experiential world view of the speakers in
order to understand the meaning of these posture expressions.
4.1 Conceptualization patterns of the three posture verbs
In the cognitive linguistics approach of which this study is based the components of the
meanings of these verbs and the experiential realities associated with them encode
properties which together constitute a larger semantic frame which can conveniently be
grouped into four cognitive domains (see Newman 2002:2). These are; spatio-temporal
domain, force dynamic domain, active zone domain, and socio-cultural domain. We shall
look at each of them in turn.
4.1.1 The spatio-temporal domain
The spatio-temporal domain relates to the overall spatial configurations associated with
each posture and maintained through time. Thus zi ‘be sitting’ encodes a compact shape
of the entity while gã conceptualizes the horizontal elongation of the entity. So someone
who is in a lying posture obviously will align the body in a horizontal orientation.
Similarly ze’ ‘be standing’ designates an entity in an upright vertical position. These three
distinct spatio-temporal configuration patterns determine the spatial images in human
conceptualization and significantly play a role in the alternative categorization of the
position or location of other non human entities.
4.1.2 The force-dynamic domain
The force dynamic domain is a semantic category that characterizes the interaction of
force that manifest across a range of linguistic phenomenon that pertains to “the physical,
psychological, social, inferential, discourse, and mental-model domains of reference and
conception” (Talmy 2000:400). The notion of the force dynamic discussed in the context
of the posture verbs relates to the exercise of physical force through the sensorimotor
control of entities in an assumed posture. It suggests the notion of the ability of the entity
to exercise a rest state (Croft and Cruse 2004:66). The control of the force dynamic
patterns as Talmy (2000:413-414) points out is that it involves the steady-state opposition
~5~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
of two forces of which he calls Agonist (the entity that is located and exerts force on
another) and Antagonist (the place of location of an entity and which provides an
opposing force). In the posture verb paradigm it involves the Figure1 and the Ground2.
Cross-linguistically all the three verbs encode a maintained posture involving no
movement of the entity (human). Nonetheless there are differences regarding the
sensorimotor control required to maintain each posture. For example ze’ ‘be standing’
requires both the lower and upper parts of the human body to be in a sturdy and vertical
position. This requires some amount of sensorimotor control of the legs and the body to
maintain this posture. On the other hand, zi ‘be sitting’ demands that the lower torso be
rested on a solid support base with a supporting force providing the maintenance of the
sitting posture. The lying posture involving gã may not necessarily require any
sensorimotor muscular control of the body. The reason is that the body is completely at
rest requiring very little force to control it or none. So the degree of control as observed
by Newman (2002) varies with lying been the least while standing requires the most.
Children in their early years of development require the reverse of this that is, lying,
sitting, and standing.
4.1.3 Active zone domain
The term active zone is proposed by Langacker (1991:189-201) to refer to the specific
area or subpart of an entity that participates directly in a spatial relation. In other words
regarding the three posture verbs it is the part of the human body that is coincident with
the ground. The active zone should not be conceived of as a discrete or sharply bounded
region but rather the focal area of the relational interaction. For example, teeth constitute
the active zone of bite. Therefore for the three posture verbs; the active zone of zi will be
the buttocks and the upper part of the body that contribute to maintain the sitting position
while for ze’ the active zone suggests the legs. On the other hand the active zone of gã is
one side of the body that is in contact with the ground.
4.1.4 The socio-cultural domain
The socio-cultural domain refers to the world view of the speakers and how they
conceptualize the various posture states (cf Lemmens 2002:130). The various posture
states play different roles in the socio-cultural domains of the speakers. Sitting is
conceived as a comfortable position. This is manifested in the following Gurenɛ
figurative expressions. As all the examples suggest the cultural notions encoded by the
posture verbs in each of these sentences show that the subject entities are in control of
some state of affairs. The rich man in (7) is construed to be sitting freely conceptually.
However, its socio cultural interpretation is that he is without worry. Also (8) suggests in
the Gurenɛ world view that the rabbit has amassed wealth and is enjoying it comfortably.
1
Figure is the entity that is located with respect to another entity (Talmy 2000a:311-315). In the context of
my three posture verbs the human entities are figures while the place that they are at rest designates the
Ground.
2
Ground refers to the place where the Figure is located ( Talmy: ibid)
~6~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
Similarly anyone who is in a comfortable position is perceived to be experiencing a
chieftaincy condition as in (9). The reason is that chiefs in Gurnɛ culture do not have to
work but sit in their palaces and receive good treatment from their subjects.
(7)
Tata
la
zi
la
Rich.Person DEF sit
FOC
‘The rich man is without worry.’
fai
free
(8)
Asɔ’ɔŋa
la
zi
la
Rabbit
DEF sit
FOC
‘Mr. Rabbit is sitting on wealth.’
lɔgerɔ zuo
things head
(9)
A
zi
la
na’am
3SG sit
FOC chieftaincy
‘He is in a confortable position.’
zuo
head
The socio-cultural meanings of gã ‘lying’ are associated with rest, sleep, sickness, and
death as the following examples attest. In each of the examples below the semantic
character of the subjects and sometimes the complements in each expression provides a
good ground for the appropriate socio-cultural interpretation.
(10).
Kaara la
taregɛ
gã
mɛ
Farmer DEF be.tired
lie
AFF
‘The farmer is tired and is lying down.’
(11).
Bilia la
gã
mɛ
Baby DEF lie
AFF
‘The baby is sleeping.’
(12).
Asɔ’ɔŋa
pɔga gã
la
deo-n
Mr. Rabbit
wife lie
FOC room-LOC
‘Mr Rabbit’s wife is down with illness.’
(13)
Mam sira
gã
la
dɔgeta
My
husband
lie
FOC hospital
‘My husband is lying at the hospital.’
(14).
Na-katɛ
la
gã
la
Chief-big
DEF lie
FOC
‘The paramount chief is dead.’
la
with
bã’a
illness
deo-n
room-LOC
Examples (10) to (14) may encode the horizontal orientation of the Figure entity but the
socio cultural meanings do not necessarily imply this physical horizontal posture. For
example, (10) talks about a farmer who is tired and lying down but this suggests a resting
posture and not just lying down. In (11) the baby is said to be lying down but in the
~7~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
culture babies do not just lie down but are thought to always be sleeping. Sick persons are
also perceived in the culture to be weak and are conceptualized as lying down (cf Song
2002 on similar semantics of Korean posture verb of lying). Example (12) is an excerpt
from a folktale which seeks to portray rabbit’s wife as been sick and lying in the room.
Similarly (13) is a case of a woman who reports of her husband’s admission in the
hospital. It does not mean that her husband is in a lying posture at the hospital but that he
is in a state of illness. The death of prominent people such as chiefs in the society is not
normally announced as the chief is dead but rather expressed through euphemism such as
the chief is lying in the room as depicted in (14). There are a number of reasons for this.
First is the fact that they want to avoid a potentially conflict situation among the potential
successors to the throne from making an unhealthy contest before the burial of the
deceased chief. So by not publicly announcing or admitting that the chief is dead but only
lying down suggests he is only resting or taking a leave. This does not give the right to
any successor to contest for the throne that is not declared vacant. A second reason is that
a chief is perceived to be all powerful in the society and does not succumb to death. So
even when he is dead he is perceived to be alive in the spiritual world.
The experiential realities that are associated with ze’ are trouble, suffering and general
discomfort. The rich man in (15) is said to be literally standing in the sun. This means that
in the culture he is in difficult times. The fact is that the Gurenɛ community is in the
tropics and no one enjoys standing in the sun like people in the west do during the
summer. Instead people will prefer to stand in the shade. So any one who is standing in
the sun is said to be feeling the intense heat of the sun and therefore cannot be said to be
enjoying any comfort. A rich person who enjoys his wealth is conceptualized to be in the
shade but when his wealth is diminished he is said to be in the sun. In (16) the rabbit
features prominently in Gurenɛ stories as a wise character who always outwits his fellow
animals but occasionally comes into trouble. The rabbit is seen to be literally standing in a
problem. We see a link with the posture verb of standing because as explained in section
4.0 when one is in trouble you are psychologically unsettled.
~8~
Samuel Atintono: The semantics of three posture verbs in Gurenɛ
(15)
Tata
la
ze’
la
Rich.person DEF stand FOC
‘The rich person is suffereing.’
(16)
Asɔ’ɔŋa
ze’
la
Rabbit
stand FOC
‘Mr. Rabbit is in trouble.’
wuntɛɛŋa
sun
yele puan
trouble LOC
5 EXTENSION OF POSTURE TO LOCATIVE EXPRESSIONS
The posture verbs in Gurenɛ can be extended to help conceptualize the location or
position of other inanimate entities. Among the three posture verbs only two, gã and ze’
are used in this sense. Zi is restricted to human postures only as there is no instance of its
usage in the locative sense. A plausible explanation for its non occurrence in the locative
context is that no any other inanimate entity in the culture can assume a sitting posture
except humans. When dogs and cats sit on their rear while resting on their hind legs they
are described as squatting and not sitting. Some of the examples below were elicited
using the MPI positional verb picture series (numbers in brackets show the positional
picture numbers). It involves showing a picture of a posture scene to a consultants and
asking the question where is the entity X? The posture verb gã is used to describe the
location of the objects that are in a horizontal position on a surface. Example (17) is a
scene where a stick is put diagonally on top of a table while (18) depicts a posture scene
in which beans are spread on the floor. The example in (19) depicts a scene where a bottle
is placed on its side in a basket. The stick and the bottle are described as lying because
there is an extension of the conceptualization of human postures in a horizontal manner to
these entities. The description of the beans as lying might sound odd but again any entity
that lacks a base support (e.g. with legs) is construed as lying (also see Ameka 2007,
Kutscher & Schultze-Berndt 2007).
(17)
(18)
(19)
Dibega
la
pue
gã
la
teebule la
Stick
DEF cross lie
FOC table DEF
‘The stick is lying on top of the table.’ (PVPS 6)
zuo
head
Tɛala
yɛregɛ
gã
la
tiŋa.
Beans
DEF spread
lie
FOC ground
‘The beans are lying on the ground.’ (PVPS 11)
Tua la
gã
la
pi’ɔ la
puan
Bottle DEF lie
DEF basket DEF LOC
‘The bottle is lying in the basket.’ (PVPS 22)
Further ze’ is used to characterize the location of entities in a vertical position. Such
objects usually are conceived of having legs which are comparable to humans or they
~9~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
may have a base that projects them vertically. Objects such as tables, chairs, beds, cars,
trees, bottles, bowls, and houses can all be characterised as standing. The following
examples describe objects in a standing posture.
(20)
Tua la
ze’
la
tiŋa
Bottle DEF stand FOC land
‘The bottle is standing on the ground’. (PVPS 58)
(21)
Pi’ɔ la
ze’
la
bimbine
Basket DEF stand FOC platform
‘The basket is standing on the platform.’
(22)
Laa
la
ze’
la
tiŋa
Bowl DEF stand FOC land
‘The bowl is standing on the Ground.’
(23)
Naba yire la
ze’
la
kulega nuuren
Chief house DEF stand FOC river mouth
‘The chief’s palace is by the edge of the river.’
zuo
head
6 FORM CLASS OF VERBS
I discuss briefly in this section the form class of verbs that are used to code posture in
Gurenɛ. The three posture forms discussed above code stative situations. Talmy (2000a)
proposes that typologically positional verbs fall into three types namely stative “be in
position”, inchoative “get into a position”, and agentive “put into a position”. The three
posture verbs in Gurenɛ discussed in this article correspond to the ‘be in position type.’
The dynamic postural meanings or the inchoative type such as ‘to sit oneself down, to
move oneself into a ‘standing position’ and to lay oneself down are closely related
semantically to the corresponding stative meanings. At least for English this holds to
some extent. For example ‘sit’ can have dynamic stative interpretations as in ‘I sat on the
chair’ to mean ‘I sat myself on the chair’ or ‘I was sitting on the chair’. In Gurenɛ distinct
forms of the verb are used for the dynamic type as illustrated in (24) to (26).
(24)
A
ze’et-i
la
saazuo
3SG stand-IMPFV FOC upright
He is getting into a standing position
(25)
A
zi’ire la
kuka
3SG sit
FOC chair
‘He sat down on a chair’
zuo
head
~ 10 ~
Samuel Atintono: The semantics of three posture verbs in Gurenɛ
(26)
A
ga’at-i
la
tiŋa
3SG lie-IMPFV
FOC land
‘He is getting into a lying position on the ground.’
7 CONCLUSION
The article examines the meanings of three posture verbs zi ‘be sitting’, ze ‘be standing’
and gã ‘be lying’ from a cognitive linguistics perspective. It notes that the central
meanings of the three posture verbs code the spatial orientation of human beings but are
also extended to describe the location of inanimate objects. The verbs also manifest four
conceptualization patterns in their meanings. These are spatio-temporal, force-dynamic,
active zone, and socio-cultural domains. The spatio-temporal domain describes the
assumed posture of the entity described by each verb. Thus zi ‘sitting’ requires the lower
torso to be resting on a solid ground while the upper part of the body is compact. On the
other hand, ze’ characterizes the vertical position of the entity while gã spatio-temporal
feature is that the entity is in a horizontal position. The force-dynamic domain associated
with the posture verbs relates to the interaction of the sensorimotor control that is needed
to maintain each posture. Thus for that of ze’ it requires the legs and the body to be sturdy
in order to maintain the upright posture. Even inanimate entities such as tables must have
a rigid structure to be able to maintain a standing posture. Also zi requires only the
bottom part and the upper part of the body to maintain the sitting position. Similarly gã
demands the least effort to sustain because lying down does not require any much energy.
The active zone domain discusses the particular part of the human body that takes an
active part in the posture. The active part for ze’ is the feet and the legs that are directly
involved while that of sitting is the lower torso with the active zone of gã been any side of
the body that is in a horizontal contact.
The socio-cultural meanings of the posture verbs reflect some conceptualization of certain
experiences in the culture. In this regard the use of the posture verbs in this context do not
necessarily suggest that the entities are actually in such assumed postures. As shown in
the data the conception of rest is usually associated with gã ‘lying’ and also used to
describe the death of important people in the community such as rich people and chiefs.
In Gurenɛ culture the death of such people is not announced publicly because they have
much influence on the society. The idea is to prevent the emotional shock that people who
benefit from them can get and at the same time trying to preserve the social order. So by
describing them as been in a lying state psychologically relief the people of this shock
and seeks to convince them that they are only taking a rest but not gone into eternity.
Also people who are sick are often described with this posture verb to show that they are
in a state lacking good health. The posture verb zi in Gurenɛ worldview is associated with
conditions that are associated with comfort such as enjoying one’s wealth. It also
sometimes describes a powerless position of the subject described with such a posture
verb. The socio cultural meanings attributed to ze’ depict the subject entities experiencing
troubles. The point is that any person who encounters problems is construed to be
psychologically in an upright position because the mind is not at rest.
~ 11 ~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
The posture meanings of these verbs are also extended to include the locations of
inanimate entities such as pots, sticks, houses, bottles, and many other such objects. It is
observed in the data that zi ‘sit’ is the only posture verb with its meaning restricted to
only human posture. When the extension includes other entities their characteristic
features may sometimes have something similar to humans. For example it is observed
that entities that are described as standing either have leg-like support such as tables,
beds, cars or a base support, like houses, pots, television set, and baskets. The three
posture verbs lexicalise only stative postures with different forms encoding the dynamic.
REFERENCES
Ameka, K. Felix. 2007. The coding of topological relations in verbs: the case of Likpe
(S kp lé), Linguistics 45-5/6, 1065-1103.
Ameka, K. Felix 7 Levinson, C. Stephen. 2007. Introduction: The typology and semantics
of locative predicates: posturals, positionals, and other beasts, Linguistics 45-5/6,
847-871.
Atintono, A. Samuel. 2004. The syntax and semantics of posture verbs in Gurene, In
Dakubu, Kropp & Osam (eds), E.K., Studies in the Languages of the Volta Basin;
Proceedings of the Annual Colloquium of the Legon-Trondheim Linguistics Project,
Vol 2. 10-17.
Casad, H. Eugene. 1996. Cognitive Linguistics Research 6: Cognitive Linguistics in the
Redwoods: The expansion of a new paradigm in Linguistics. Berlin, New York:
Mouton de Gruyter
Croft, William & Cruse, B. Alan. 2004. Cognitive Linguistics. Cambridge: Cambridge
University Press
Dakubu, Kropp, Atintono, Winkene & Nsoh (eds). 2007. Gurenε-English Dictionary.
Legon: Linguistics Department
Evans. Vyvyan. 2007. A Glossary of Cognitive Linguistics. Edingburgh: Edingburgh
University Press.
Evans, Vyvian & Green, Melaine. 2007. Cognitive Linguistics: an introduction.
Edingburgh: Edingburgh University Press.
Frawley, William. 1992. Linguistic Semantics. Hillsdale, New Jersey, Hove & London:
Lawrence Erlbaum Associates, Publishers.
Geeraerts, Dirk & Cuyckens, Hubert (eds.). 2007. Introducing cognitive Linguistics. The
Oxford Handbook of cognitive Linguistics. Oxford: Oxford University Press.
Gibbs, W. Raymond. 1996. What’s cognitive about cognitive linguistics? In Casad, H.
Eugene (ed.). Cognitive Linguistics Research 6; Cognitive Linguistics in the
Redwoods; The expansion of a new paradigm in Linguistics, Berlin, New York:
Mouton de Gruyter.
Kutscher, Silvia & Schultze-Berndt, Eva. 2007. Why a folder lies in the basket although it
is not lying: the semantics and use of German positional verbs with inanimate
figures. In Ameka, K. Felix & Levinson, C. Stephen Linguistics 45-5/6, 983-1028.
Langacker, W. Ronald. 2000. (ed.).Cognitive Linguistics Research: Grammar and
Conceptualization. Berlin and New York: Mouton de Gruyter.
~ 12 ~
Samuel Atintonoː The semantics of three posture verbs in Gurenɛ
Langacker, W. Ronald. 1991. Concept, Image, and Symbol: The Cognitive Basis of
Grammar: In Dirven, Rene & Langacker, Ronald W. Cognitive Linguistics
Research: Grammar and Conceptualization. Berlin and New York: Mouton de
Gruyter
Langacker, W. Ronald. 1987. Foundations of Cognitive Grammar; theoretical
prerequisites. Vol 1. Stanford, California: Stanford University Press.
Lemmens, Maarten. 2002. The semantic network of Dutch posture verbs. In Newman,
John (ed.). The Linguistics of Sitting, Standing and Lying. Amsterdam/Philadelphia:
John Benjamins Publishing Company. 359-386.
Levinson, C. Stephen & Wilkins, David (eds.). 2006. Grammars of space: explorations in
Cognitive Diversity. Cambridge: Cambridge University Press.
Levinson, C. Stephen. 2003. (ed.). Space in Language and Cognition: explorations in
Cognitive Diversity. Cambridge: Cambridge University Press
Levinson, C. Stephen. 2001. The Linguistic Expression of space. IESBS.
Newman, John (ed.). 2002. The Linguistics of Sitting, Standing and Lying.
Amsterdam/Philadelphia: John Benjamins Publishing Company.
Schaefer, P. Ronald & Egbogare, O. Francis. 2008. A Preliminary Emai Posture verb
parameters. Journal of African Linguistics and Languages. www.JALL.com
Song, Jung Jae. 2002. The posture verbs in Korean: Basic and extended uses. In Newman,
John (ed.). The Linguistics of Sitting, Standing and Lying. Amsterdam/Philadelphia:
John Benjamins Publishing Company. 359-386.
Talmy, Leonard. 2000a. Toward a cognitive Semantics Volume I: Concept Structuring
Systems. Cambridge, Massachusetts, London, England: MT Press.
Talmy, Leonard. 2000b. Toward a cognitive Semantics Volume II: Typology and Process
in Concept. Cambridge, Massachusetts, London, England: MIT Press
~ 13 ~
CHAPTER 2
SEMANTIC CHANGE – THE EVOLUTION OF LEXICAL MEANING IN TIME AND SPACE
AN EXAMPLE FROM THE GREEK LANGUAGE: THE WORD ε αγχο ία (MELANCHOLY)3
AMALIA KAZIANI
Aristotle University of Thessaloniki
Semantic change is a phenomenon which reflects the dynamic character of lexical meaning. Word
forms and meanings stretch and evolve across time and space. The Greek word ε α χο ία is
considered to be a ‘living proof’ of the process of semantic change which a lexical item may
undergo. The original meaning of the word stretches back to ancient Greek. The word also
‘stretches’ in space; it is borrowed by more than 40 languages around the world. This paper shows
how the original medical meaning of the word has extended through metaphorical use to the
mental health field and resulted into its current everyday use through the process of
conventionalization of the metaphorical meaning. Sentences collected from a Greek corpus
illustrate the current use of the word ε α χο ία and show that, despite its obvious semantic
connection to the past, the everyday use of the word does not provide any clues about its original
meaning.
1 INTRODUCTION
Word forms and meanings stretch and evolve across time and space and the study of their
etymology sheds light on their history and the evolution of their meaning. The dynamic
character of lexical meaning is clearly depicted on the one hand in the variety of the
innovative ways in which writers express themselves in genres like poetry and science
fiction, and on the other hand in the examples of conventionalized semantic change,
recorded in a wide range of dictionary lemmas (Seuren 2000:423-426). The phenomenon
of semantic change across time and space might be viewed as a perpetual process of
interaction between linguistic, cognitive and cultural factors (Wardhaugh 1986:10). Kitis
(2007: 55) observed that language is “an open system that reaches out into the world”. I
consider the Greek word ε α χο ία to be the ‘living proof’ of this observation. The
original meaning of the word ε α χο ία stretches back to ancient Greece – around 400
B.C. The word also stretches in space; it is borrowed by more than 40 languages around
the world. Some examples are shown in the table below.
3
I would like to thank professor Eliza Kitis for her constructive comments and suggestions and professor
Dionysis Goutsos for having provided me with additional sources to supplement my study.
~ 14 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
(1): ε α χο ία in 25 languages4
Language
Albanian
Croatian
Czech
Danish
Dutch
English
Estonian
Finnish
French
German
Greek
Hungarian
Italian
Lithuanian
Norwegian
Polish
Portuguese
Romanian
Russian
Serbian
Slovak
Spanish
Swedish
Turkish
Ukrainian
Word
malinkonia
melanholija
melancholie
melankoli
melancholie
melancholy
melanhoolia
melankolia
mélancolie
melancholie
α χ α
melankólia
malinconia
melancholija
melankoli
melancholia
melancolia
melancolie
е а х ия
е а х ича
melanchólia
melancolía
melankoli
melankoli
Ме а х ія
2 THE PROCESS OF SEMANTIC CHANGE, AS REFLECTED IN THE EVOLUTION
OF THE MEANING OF THE WORD ε αγχο ία
With respect to the word’s ancient Greek origin, two key figures of medicine in ancient
Greece, Hippocrates and Galen, were involved in the formation of the lexical item’s
meaning; the former coined the term and the latter expanded its meaning (Millon & Davis
1996:36;
χα α
2003:126). The etymology of the compound word ε α χο ία
reveals its ancient Greek origin: Its constituents, “ α ” and “χ
” mean ‘black’ and
‘bile’ respectively ( α υ
1977). Thus, ε α χο ία could be thought of as a
representative sample of meaning’s journey into time and space. Before looking into the
evolution of the lexical item’s meaning, a clarification should be made with regard to its
corresponding term in English. More specifically, even though the Greek term has
retained its form for the last 2300 years, in the English language the semantic change of
4
Sources: www.logosdictionary.org, www.logos.it, www.wikiled.com
~ 15 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
the word seems to be reflected in the two slightly different terms: melancholia and
melancholy. According to the Longman Dictionary of Contemporary English
(www.ldoceonline.com), the former is an old-fashioned term denoting great sadness,
while the latter is a formal term referring to feelings of sadness. According to the Oxford
Advanced Learner’s Dictionary of Current English (Hornby, Wehmeier, McIntosh,
Turnbull & Ashby 2007:956), melancholia refers to a specific mental illness, while
melancholy simply refers to low spirits (table 2). The word melancholy was chosen as the
equivalent of ε α χο ία in the title of this paper. Melancholy was preferred to
melancholia because, as is shown later on in the examples from the Greek corpus, it
corresponds accurately to the contemporary meaning of the Modern Greek word
ε α χο ία.
(2)
Two definitions of melancholy
The definitions of melancholia and melancholy by the Longman Dictionary of
Contemporary English (www.ldoceonline.com):
melancholia [uncountable]
old-fashioned a feeling of great sadness and lack of energy
melancholy1 adjective very sad
melancholy2 noun [uncountable]
formal a feeling of sadness for no particular reason
The definitions of melancholia and melancholy by the Oxford Advanced
Learner’s Dictionary of Current English (Hornby, Wehmeier, McIntosh, Turnbull
& Ashby 2007:956):
melancholia /noun (old-fashioned) a mental illness in which the patient is
depressed and worried by unnecessary fears
melancholy / noun, adj.
noun [U] (formal) a deep feeling of sadness that lasts for a long time and
often cannot be explained
adj. very sad or making you feel sadness SYN: MOURNFUL, SOMBRE
The term ε α χο ία was first used by Hippocrates to describe a biological condition
characterized by excessive secretion of black bile in the human body. According to
Hippocrates’s Theory of the Four Humours (that is, fluids), which he put forward in the
4th century B.C., the human body is composed of four fluids: blood, phlegm, yellow bile
and black bile. He claimed that human behaviour is determined by biological and
physiological factors. Consequently, the imbalance of the fluids may result in various
types of physical illness as well as in mental disorders (Katsambas & Marketos 2007:860;
Millon & Davis 1996:36). Hence, the initial meaning of the term ε α χο ία was strictly
medical: it referred to a physical symptom – a biochemical disorder. The feelings of
sadness, fatigue or low spirits were only the secondary symptoms of the disease, which
often accompanied the clinical picture of some of the patients with ε α χο ία.
Cancer was one of the illnesses Hippocrates attributed to excessive amounts of black bile
(a condition which he also called “ α χυ α” – imbalance of the humours)
~ 16 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
( α α π π υ
2005:85). Different types of physical pain were also attributed to
increased levels of one of the four humours in the human body ( υ ω
υυ
et al 2009:125). Within the context of Hippocrates’s humoral doctrine, the excessive
secretion of yellow bile, black bile, blood and phlegm corresponded to four different
temperament types: the choleric, melancholic, sanguine and phlegmatic respectively
(Millon & Davis 1996:36). As Theodore Millon (1996:36) observes in the book titled
“Disorders of Personality. DSM-IV and Beyond”, “although the doctrine of humors has
been abandoned, giving way to scientific studies on topics such as neurormone chemistry,
its terminology and connotations still persist in such contemporary expressions as being
sanguine or good humored”.
It is worth mentioning that Hippocrates’s theory was radically innovative in that it refuted
the fixation with religion, which was prevalent at that time and introduced a systematic,
scientific method for the study of physiological problems and their causes. In other
words, the Theory of the Four Humours signified the end of a long era of preoccupation
with the metaphysical, which considered disease to be a “God-induced” punishment
(Katsambas & Marketos 2007:860; υ ω
υυ
et al 2009:127).
About 520 years later, Galen developed Hippocrates’s theory and tried to connect it with
human personality in a more systematic way (Nutton 2005:115). Galen looked more into
the correspondence of the four fluids – blood, phlegm, yellow bile and black bile – to the
four personality types: sanguine, phlegmatic, choleric and melancholic respectively and
he provided an analysis of these types (Millon & Davis 1996:36). Galen’s theory of the
four humours attached another meaning to the word ε α χο ία: that of a personality
disorder and the related behavioural patterns (Millon & Davis 1996:36). Apart from its
linguistic interest, the semantic change that the lexical item underwent also reflects the
radical change in the philosophy of medicine that Galen and Hippocrates brought about
with their theories. Both ancient Greek doctors introduced the holistic approach in
medicine, a philosophy that views human body and psyche as an unbreakable whole
(Orfanos 2007:854).
~ 17 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
(3)
ε α χο ία - The evolution of the meaning of
ε α χο ία
Original meaning
medical condition
excessive secretion of black
bile in the human body
Metaphorical use
personality disorder
a severe form of depression
Conventionalised meaning
emotion
sadness, low spirits
Returning to the linguistic aspect, Galen’s use of the term ε α χο ία could be considered
as the ‘bridge’ between the word’s original and its current meaning. More specifically,
while initially the term was used by Hippocrates to refer to a medical condition, the
Galenic theory enriched its meaning, by attaching behavioural aspects to it. The term was
then used metaphorically to denote a mental illness. This seems to be the case with the
word ε α χο ία, as it appears in a Greek psychiatric manual titled “ υ α
υχ α
” (Dynamic Psychiatry), which was published in 1971. In the aforementioned
manual the word is used as a synonym of depression (an affective disorder)
(Φ ππ π υ
1971:355). Finally, through its folklore use, the term was eventually
conventionalised: it lost its metaphorical dimension and became synonymous to sadness
or low spirits (Table 3).
Even though the connection between black bile and feelings of sadness that Galen
suggested was never confirmed, the current meaning of the word ε α χο ία – and of its
foreign counterparts – denotes ‘feelings of sadness’. This confirms Lakoff and Johnson’s
view that “truth is always relative to a conceptual system (…) any human conceptual
system is mostly metaphorical in nature (…) therefore, there is no fully objective,
unconditional or absolute truth” (Lakoff & Johnsonn 1980:185). If one adopted an
objectivistic approach of lexical meaning, once Galen’s theory was refuted, the word
ε α χο ία should cease to denote a mental state and be restricted to a physiological
symptom. On the contrary, the word is used in various creative ways in a wide range of
contexts and it is related to sadness, low spirits and bad mood among other, metaphoric
uses; these will be illustrated by instances retrieved from the Greek corpus.
As Seuren (2000:424-426) and Lakoff & Johnson (1980:186) observe, the
conventionalization of the metaphorical meaning of a word or construction, as a result of
its use in everyday language, often deprives the word’s meaning of its metaphorical
dimension. This seems to be the case with ε α χο ία. As will become clear in the
examples from the Greek corpus, the word ε α χο ία, with its current meaning, does not
refer either to the clinical condition of excessive secretion of black bile that the person
suffers from or to the mental symptoms that accompany it. However, its extensive use in
~ 18 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
everyday language throughout time has resulted in the word now literally meaning
‘sadness’, ‘bad mood’. Accordingly, the word’s original meaning underwent a semantic
change through its metaphorical use, which was followed by a conventionalization of its
metaphorical meaning.
With respect to the word’s use in formal mental health contexts, it should be stressed that,
as is indicated in the Greek versions of DSM-IV and ICD-10, two of the most widely
used diagnostic manuals of psychiatry worldwide, the word ε α χο ία is not included in
the formal definition of any affective disorder (Γ
α
1996:161-171;
φα ,
, αυ α 1993:135-163). More specifically, the Diagnostic and Statistical
Manual of Mental Disorders, 4th Edition (DSM-IV) is a manual published by the
American Psychiatric Association that includes detailed classifications and diagnostic
criteria for all the mental disorders. The Greek version of DSM-IV does not include the
word ε α χο ία as a classification of an Affective Disorder per se. It only refers to the
presence or absence of “melancholic features” (loss of pleasure, early morning
awakening, anorexia etc) as some of the criteria for the subclassification of a depressive
episode (Γ
α
1996:191). What is more, in the third edition of DSM the word
melancholia is characterized at the footnote as “a term from the past” which is only used
as a distinctive feature of a specific type of a depressive episode (American Psychiatric
Association 1980:205). Finally, in the Greek version of the Tenth revision of the
International Classification of Diseases (ICD-10) of the World Health Organization the
word ε α χο ία is also used only as a feature or symptom of a subcategory of a
depressive episode ( φα ,
& αυ α 1993:138).
3
CORPUS DATA
Before proceeding to the presentation of the corpus data, a short reference will be made to
the Hellenic National Corpus, from which I took the examples I used in my study
(http://hnc.ilsp.gr/). The Hellenic National Corpus has been developed by the Institute of
Language and Speech Processing (ILSP). It currently contains about 47,000,000 words
(May 2009), which correspond to over 2,000,000 sentences or 50,800 texts, while it is
constantly being updated. All texts have been selected, so as to present a realistic picture
of modern language use. It should be mentioned, however, that the Corpus contains
samples of written language exclusively. Most texts have been selected based on their
high readability (high circulation newspapers, best-selling books etc.). In order to include
different types of language, texts from several media, belonging to different genres and
dealing with various topics have been selected. The Corpus texts are classified according
to medium into the following categories: Books, Internet, Newspapers (daily or weekly),
Magazines (published every week, fortnight, month etc), and Various Other Sources (any
kind of text that does not belong to any of the above categories, such as: leaflets,
brochures, flyers, and all kinds of reports and documentation).
From the 100 sentences randomly selected from the Hellenic National Corpus (HNC) , 70
sentences were finally selected for the study. The remaining 30 were not analysed
because they were either too brief or didn’t provide the necessary contextual information
~ 19 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
needed in order to define the meaning of the key word in question (that is, ε α χο ία). In
47 out of the 70 sentences (i.e. 67.14%), ε α χο ία refers to an emotion, of mild
intensity and short duration, which often precedes or follows feelings of joy or
enthusiasm. In 28 sentences (that is, 40% of the sample) the word ε α χο ία appeared in
descriptions or reviews of various types of artistic work. In another 40% of the sample the
word was used metaphorically in various creative ways. Finally, in 17 of the 70 sentences
(24.28%) the word ε α χο ία appeared within sport – related contexts (see (4)).
(4)
The word ε α χο ία as it appeared in the Hellenic National Corpus
Number of sentences
(%)
ε α χο ία as a mental disorder
0
(0%)
ε α χο ία as an emotion
47
(67.14%)
ε α χο ία as a characteristic of
artistic work
28
(40%)
ε α χο ία used in metaphors
28
(40%)
ε α χο ία appearing in sportsrelated contexts
17
(24.28%)
It should be mentioned at this point that an attempt was made to translate the sentences
from the Hellenic National Corpus which are presented in the following examples, in
order to illustrate how the word ε α χο ία is used within the Greek language context. As
a consequence, some of the examples may seem semantically or syntactically awkward in
English. In addition, the word melancholy is being used as the English equivalent of
ε α χο ία in all the examples, even though it might not be the most suitable in every
case. The aforementioned word was chosen for homogeneity purposes.
4 ANALYSIS: CONCLUSION
As was mentioned earlier, in 67.14% of the sample the word ε α χο ία refers to an
emotion (table 5). More specifically, in the examples from the corpus ε α χο ία is
presented as an emotion which may be stirred up by a theatrical performance (example 3)
or which often characterizes songs (example 7), poems (example 8) or works of fiction
(example 9). As such, ε α χο ία is not an overwhelming or intense emotion. On the
contrary, the fact that it tends to appear in the same context with laughter (examples 3 and
5), humor (example 5) and sweetness (example 11) indicates that it is perceived to be a
mild emotion, which can easily fade and give way to joy (example 2). This observation
~ 20 ~
Amalia Kazianiː Sem
mantic change — the evolutiion of lexical meaning
m
in tim
me and space
migght be confiirmed by ex
xamples in tthe Greek corpus
c
in which
w
ε α χχο ία coexiists with
anniversaries (eexample 1),, enthusiasm
m (example 4) and cheeers (examplle 13).
(5)
ε α χοο ία as an em
motion
(11) “ π
υ α
π
π ω υ α
α χ α”
(11ª) “anniversaaries always haave a touch off melancholy””
(22) “ χ
α χ α
α
χα ”
(22ª) “yesterdayy’s melancholy
y turned into iirrepressible jo
oy”
(33) “ χ
α πα
α
α
, π
α,π
α π ,
α χ α”
(33ª) “he has creeated a uniquee performancee, full of laugh
hter and a littlee melancholy now and then
n”
(44) “
υ α
α
α χ α
α α
α αα υ α
α α, α
αα
υ
π υ
”
(44ª) “enthusiasm and melanccholy were thee two conflictting emotions which were m
mixed in the Council
C
of
M
Ministers”
(55) “π ω απ
χ
α
α α
α χ α υ
ωα”
(55ª) “behind huumor and laug
ghter hides thee melancholy of
o the central hero”
α χ α υ
(66) “
χ
φ
υ π
απ
χχα
α
χωφ”
(66ª) “His stylee isn’t reminisccent of Chekhhov’s smiling melancholy at all”
(77) “α α
α χ α α
χυ
α α
α α,
υ
ω Hopee of the Statess
α
α α π
”
(77ª) “even thoough melanch
holy is infuseed into our songs,
s
the music
m
of Hopee of the Stattes is not
ddepressive”
(88) “ α π
α
υ απ
α απ
α χ α”
(88ª) “melanchooly pervades his
h poetry”
(99) “
α υ χ
α
α φ
α χ α α
α
α
υ
”
(99ª) “his languaage has the po
ower to convey
ey melancholy
y and emotion in a lyrical toone”
Thee term ε α χο ία is alsso used to ddescribe varrious types of
o artistic w
work, as can
n be seen
in eexamples 3-9 in table 5. These eexamples confirm the observation
on that the word in
question has shhifted from the domainn of mediciine to other domains, llike sports (table
(
7)
and arts, amonng which po
oetry (exampple 8), music (examplee 7) and liteerature (exaamples 5
and 6).
Thee analysis off the corpuss sample alsso showed that
t
ε α χο
ο ία is usedd in various creative
wayys when refferring to em
motional staates. More specifically
y, in 28 exaamples (40%
% of the
sam
mple) ε α χο
χ ία had a more obviious metaph
horical funcction. It waas often perrsonified
(exaamples 5 and
a
6) or used
u
metapphorically to
t describe a person’ss facial exp
pression
(exaamples 12 and
a 15) or emotional
e
reeaction (exaamples 13, 14
1 and 16). The word was
w also
usedd creativelyy in descripttions of nonnhuman entiities, as show
wn in exam
mples 10 and
d 11:
~ 21 ~
Amalia Kaziianiː Semanticc change — thhe evolution off lexical mean
ning in time annd space
(6)
Metaphoriccal uses of ε α χο ία
α απ α
(10) “ α
φ
π α
, π υ χα
α
α α
α υπ
α χ α
α φω α α α
π α υ”
(10ª) ““The long winding road paassed throughh the forests which
w
had a vague and impposing melanccholy,
with ttheir clearingss and paths”
(11) “ φ
πω π υ α
υ
α χ α υ α
υπ
π
α υ”
(11ª) ““Autumn, whiich has such a sweet melanncholy and succh an emotive sweetness”
α
πα α
υ
(12) ““α α α
α, α
π
α χ α
αυ
…
α α π
α
α
”
(12ª) “she was obsserving this ou
utstanding dem
monstration off antiracism patiently, but w
with her eyes filled
with m
melancholy”
The sem
mantic contexts in which the wordd ε α χο ία
ί appeared more frequuently weree arts
(40% oof the sampple) and sports (24.288% of the sample).
s
Th
here were nno cases in
n the
corpus sample in which
w
the word
w
would refer to a mental
m
disorder. Hencee, following
g the
course of semanticc change, the
t meaningg of the wo
ord started from the ooriginal med
dical
field, ppassed throough the mental
m
heaalth field and
a
has en
nded up w
with its currrent
conventtionalized form.
f
The teerm, in its evveryday usee, does not seem
s
to havve any scien
ntific
connotaation; as cann be seen in
n examples 13-16 (table 7) the terrm is used tto refer, am
mong
other thhings, to thee feelings off the fans annd the playeers of the lossing team.
(7)
ε α χο ία in sport – related
r
conttexts:
(13) “ α
α
« υα
α
υ υ » πα υ
α «
»
α χ α
α
α φ ”
(13ª) ““the match ennded amidst “bluewhite” ccheers and “y
yellow” melan
ncholy in thee packed Ivan
nofeio
stadium
m”
(14) “ 40-48
α χ α”
36’
(14ª) ““40-48 at36 minutes
m
plungeed the neighboourhood into melancholy”
m
(15) “ π α
α χ α απ
α υπω
π
ωπ
υα
υ π
π υ ω χ ω ω
α
π
φα
υ π ”
(15ª) “seeing melaancholy engraaved on the face of the worthiest
w
beaarer of our nnational colou
urs in
European football”
α χ α
π α
απ
αυ α
υ
,
ω
ω α,
(16) “
”
(16ª) ””the melanchooly emanating
g from the em
mpty OAKA stadium,
s
due to
t the punishm
ment, was dou
ubled
becausse of Prelevic’s injury”
ning of the word ε α χο ία seem
ms to
The aboove evidencce indicatess that the prresent mean
derive ffrom its folkklore ratherr than its sciientific use.. The curren
nt use of thee word doess not
providee any clue about the item’s oriiginal mean
ning. Neveertheless, thhe study of its
etymoloogy revealss its obvious semantiic connection to the past. Finallly, it is worth
w
mentionning that evven though the
t term is not used in
n mental heaalth contextts, which means
m
that it iis not ‘officially’ conn
notative off a mental disorder,
d
in
n every dayy language it is
widely used to deescribe neg
gative moodd of varyin
ng intensity
y and duraation. It is also
extendeed to incluude a wide range of real-world entities (landscapes, songs, poems,
footballl matches) that
t may insspire feelinggs of melan
ncholy. This, interestinggly, seems to
t be
the casee not only with
w Greek, but also witth the otherr languages that have ad
adopted the term
t
~ 22 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
ε α χο ία. A future study of possible semantic analogies between languages might
provide enlightening information.
REFERENCES
English:
American Psychiatric Association. 1980. Diagnostic and Statistical Manual of Mental
Disorders (third edition). Washington, DC: American Psychiatric Association.
Freud, Sigmund. 1917. Mourning and Melancholia. The Standard Edition of the complete
works of Sigmund Freud. Trans: James Strachy. London, Hogarth Press: Institute of
Psycho-analysis, 1953-1974, 14, 243-258.
Katsambas, Andreas AND Marketos Spyros. 2007. Hippocratic Messages for Modern
Medicine (the Vindication of Hippocrates). Journal of the European Academy of
Dermatology and Venereology, 21, 859-861.
Kitis, Eliza. 2007. Semantics: The study of meaning in language. Department of English,
Aristotle University.
Lakoff, George & Johnson, Mark. 1980. Metaphors We Live By. Chicago: The
University of Chicago Press.
Millon, Theodore & Davis, Roger. 1996. Disorders of personality: DSM-IV and beyond.
New York: Wiley
Nutton, Vivian. 2005. The Fatal Embrace: Galen and the History of Ancient Medicine.
Science in Context 18(1), 111–121.
Hornby, Albert Sydney, Sally Wehmeier, Colin McIntosh, Joanna Turnbull and Michael
Ashby. 2007. Oxford Advanced Learner’s Dictionary of Current English. Oxford:
Oxford University Press.
Seuren, Pieter. 2000. Lexical eaning and etaphor, in Cognition in Language Use:
Selected Papers from the 7th International Pragmatics Conference. V. 1, 422-31.
Wardhaugh, Ronald. 1986. An Introduction to Sociolinguistics. New York: Blackwell
Greek:
Γ
, Ω
(
φ α , π
α). 1996. AMERICAN
PSYCHIATRIC ASSOCIATION: α ω
α DSM-IV.
α:
α
Γ
,Γ Ω Γ
. 2003. Έ
π χ
υ Γα
. In α
π
π ω υ
ω , 37 (3-4), 125-128
, . 1977. Ετυ ο ο ό
χα α
: Θ ω α,
.
Θ α
,Φ
Θ ω
Ω
,
,
, Ω
AND
,
. 2009.
α
ω απ
ππ
α
α
. χ α
α
2009, 26(1), 124–129
Φ
, Ω
,
, Ω
AND
,
~ 23 ~
Amalia Kazianiː Semantic change — the evolution of lexical meaning in time and space
. 1993. α
ICD-10 υχ
υ π φ
:
αφ
α
υ
α α α
πα υ
υχ
Φ
,Γ Ω Γ
. 1971. υ α
α α α
α α αχ
α
α:
α
υχ α
.
α
α α αχ
ω .
α:
α:
.
Electronic sources:
Hellenic National Corpus (HNC)™: The ILSP corpus. Institute for Language and Speech
Processing. Retrieved May 5, 2007 from http://hnc.ilsp.gr/
,
. 2005.
α
υ Γ
ω
χα ω
ω α
.
’
αφ
, 44, 85-87. Retrieved May 4, 2007, from
www.karaberopoulos.gr/karaberopoulos/ergasies/69.asp
Logos. Retrieved May 3, 2007 from: www.logos.it/pls/dictionary/
Logos Dictionary. Retrieved April 16, 2009 from: www.logosdictionary.org/
Longman English Dictionary Online. – LDOCE. Retrieved December 22, 2009 from:
www.ldoceonline.com
Wikiled Online Dictionary. Retrieved April 12, 2009 from: www.wikiled.com
~ 24 ~
CHAPTER 3
PRESENTLY COMPARING PRE-PAST
THE EXPRESSION OF ‘PAST-PERFECTNESS’ IN ENGLISH AND ITALIAN
SUSANNE SCHNEIDER
Free University of Berlin / University Ca’ Foscari of Venice
Tempo-aspectuality is a dimension of human conceptual space which has to be processed in all
natural languages. Consequently, every language must provide adequate means for the expression of
tempo-aspectual meaning. Starting from general observations about the linguistic codification of
time, this paper investigates how English and Italian code the complex notion of past perfectness. In
order to guarantee a methodologically sound comparison of the relevant language-particular coding
strategies, a descriptive apparatus, defined as ‘meta-category’, is first developed from cognitive and
typological insights and then applied to the languages in question. Since the ‘meta-category’ is an
empirical and illustratable tool, it allows for the principled inclusion and graphical representation of
both secondary and primary comparative data. The result is a comprehensive record of the
similarities and differences which characterise the expression of past perfectness in English and
Italian.
1 INTRODUCTION
Space and time are fundamental cognitive experiences which provide essential parameters
for a human being’s self-orientation in the world. Physically, an individual is always in
one specific place at one specific time. While everyone is free to some degree to
determine and influence the local component of their existence, the temporal component
is rigidly restricted to the present, i.e. the transitory moment of actual consciousness. It is
for this reason that during a telephone conversation a question like “Where are you?”,
which enquires about the spatial location of the absent interlocutor, seems rather normal
and relevant, while an equivalent enquiry about the person’s temporal location – “When
are you?” – strikes us as really irrelevant. This changes as soon as non-physical, i.e.
mental states are taken into account. In fact, the mind emancipates human beings from the
physical confines not only of space, but also of time. Thus, memories can take us back to
places that we saw at an earlier date, while visualisations of alternative and/or future
states of affairs allow for an imaginary projection of the self into hypothetical space and
time. As a consequence, even the temporal component appears to be no longer rigidly
fixed throughout, but becomes a somewhat more variable parameter, similar to space.
If human beings can indeed entertain thoughts about different places and times, then their
language must supply them with sufficient means to talk about these mental images.
Taking this cognitivist5 statement as a starting point, this paper focuses on the linguistic
5
The label ‘cognitive linguistics’ refers to a rather broad movement within (modern) linguistics which
holds that “language forms an integral part of human cognition” (Taylor 2002:4). For this reason,
~ 25 ~
Susanne Schneider: Presently comparing pre-past
codification of temporal notions. More precisely, it explores and compares how two
specific languages, namely English and Italian, cope with the task of representing
situations that the speaker perceives as belonging to the pre-past. This so-called pre-past
is a period which is anterior to another period which is itself located in the past with
respect to the speaker’s present, i.e. the moment in which s/he is making the utterance. A
comprehensive comparative overview of the relevant language-specific devices thus
constitutes the final aim of this paper and will be given in the form of a ‘meta-category’
in the Appendix.
The organisation of the paper is as follows: section 2 provides the theoretical outline and
section 3 treats actual linguistic data. More precisely, section 3.1 focuses on typological
data, while language-particular secondary and primary sources are analysed in sections
3.2.1 and 3.2.2, respectively. Finally, the conclusions presented in the last section of this
paper briefly summarise the main findings of the preceding sections and highlight the
adequacy of the ‘meta-category’ for the description of the similar and/or different
codification strategies that English and Italian speakers adopt when referring back to prepast situations.
2 THEORETICAL INTRODUCTION
2.1 The linguistic expression of time and essential premises of its
comparative description
In his monograph on Hopi, an Amerindian language, which famously (but falsely) had
been reckoned by Whorf (1940 [1998:93]) to be a ‘timeless language’, Malotki
(1983:630) argues that time forms a “fundamental experience conceptualised by every
human mind and processed linguistically by all languages to some degree or other.”
While even Hopi has now repeatedly been shown to constitute no exception to this
general rule and, consequently, to possess a number of linguistic devices for the
codification of time in utterances, the reasons for Whorf’s wrong conclusion may become
clearer if one considers the fact that though all languages are indeed capable of encoding
temporal reference, they do so in rather different ways, assigning different functional
weight to lexical and/or grammatical resources. Grouping the relevant linguistic strategies
according to their structural properties, Comrie (1985:8) distinguishes three major classes
of temporal expressions, namely lexically composite expressions, lexical items and
grammatical categories. The lack or infrequent use of expressions belonging to one or two
of these classes does not imply that a language fails to encode temporal reference.6 It
cognitivist scholars typically ground their analyses of linguistic phenomena in what is generally known
about human cognition.
6
Comrie accuses Whorf of having fallen prey to exactly this fallacy, since his analysis of Hopi as timeless
was “based simply on the fact that the language in question has no grammatical device for expressing
location in time, i.e. has no tense” (1985:4). More recently, Dahl (2001) made a case for considering
Maybrat (a Bird’s Head language of New Guinea) a language without tense and aspect. However, in the
conclusive remarks of his paper, he admits that even in this language “some compensatory mechanisms are
found” (ibid, 172).
~ 26 ~
Susanne Schneider: Presently comparing pre-past
merely means that the language in question employs different means for achieving an
overall similar effect.
While a natural language may or may not rely on certain specific strategies for the
codification of temporal notions, some general tendencies can be observed. On the one
hand, relevant lexical devices are typically members of rather large sets of forms from
which the speaker may choose if the specific temporal information that the item expresses
is necessary and/or relevant to the message he intends to communicate. On the other
hand, the principal grammatical categories traditionally regarded to deal with time are
tense and aspect, i.e. two categories only, each of which comprises a finite set of
constructions from which the speaker has to choose the most adequate form for every
utterance he makes. The grammatical categories of tense and aspect thus appear to play a
very important role in the linguistic codification of time and we will discuss them later in
the paper without, however, forgetting that lexical means may contribute to the
codification of temporal meanings.
Tense and aspect thus constitute the grammatical expression of time in language, where
tense is concerned with “location in time” (Comrie 1985:9), and aspect with “different
ways of viewing the internal temporal constituency of a situation” (Comrie 1976:3).
Being categories of the verb, tense and aspect may be expressed either synthetically, i.e.
by inflectional morphology, or periphrastically, i.e. involving several forms of words,
such as auxiliaries, in addition to the main verb, yet without ceasing to count as one
constructional unit. While there are predicates in which the tense and aspect markers can
rather easily be recognised as distinct morphemes, it is not an infrequent phenomenon to
find temporal and aspectual notions fused in a single grammatical form. For example,
while English I was singing visibly aligns a past tense marker (carried by the auxiliary)
and a periphrastic progressive marker (be V-ing) both past tense and imperfective aspect
are interwoven in the Italian synthetic form cantavo (‘I was singing’ / ‘I used to sing’).
There are thus not always neatly delineated categorical distinctions between temporal and
aspectual markers and even simple tense forms, such as the English Simple Present7 in I
sing, appear to carry an inherent aspectual meaning, which in the example given is most
likely to be interpreted as habitual like in the utterance She sings in a choir.
The fact that languages divide up the space of tempo-aspectuality in rather different ways,
distributing sections of tempo-aspectual meaning among their particular inventories of
forms, may cause initial methodological problems for a comparative approach. For
example, the specific languages selected for comparison may differ greatly as to the
number of the tempo-aspectual forms they possess because distinctions that are made in
one language may be neutralised in the others and therefore sometimes lack
morphological marking. While such a situation would render the choice of comparable
forms difficult at first, cases of forms that are identically or similarly named across the
selected languages might make such a choice all too obvious and tempt us into forgetting
7
We will follow the widely accepted practice introduced by Comrie (1976) and write designations of
language-specific categories with initial capitals. Single quotes around lower case denominations are used
for the dimensions of conceptual space.
~ 27 ~
Susanne Schneider: Presently comparing pre-past
that traditional language-specific denominations should not be taken as descriptive of the
actual forms’ functions in present-day languages (cf. Coseriu 1972:47). In order to avoid
these and similar pitfalls, linguists who intend to embark on comparative studies should
bear in mind that, as emphasised by Johanson (2000:45), “linguistic values determined
within differently structured systems cannot be compared with each other in a direct
way.”
What, then, enables us to compare the various linguistic phenomena related to the
expression of time that we encounter across languages? After having determined exactly
what it is that we want to compare, i.e. after having decided on the language-specific
tempo-aspectual forms as our units of comparison (at this point our choice to concentrate
on grammatical markers helps us to ensure that our units retain manageable proportions),
and after having ensured that these units share a general common foundation, namely
their reference to particular sections of the conceptual space of tempo-aspectuality, we
need to establish a set of stable parameters in terms of which we can describe the
languages we are interested in. Naturally, such parameters cannot be taken from any of
the object languages themselves since this would lead us to impose the categories and
terms of one language onto the others, which would result in a depersonalisation of the
latter ones. On the contrary, such parameters must really be a solid tertium comparationis,
which is constant, free from the idiosyncrasies of any individual linguistic system and
thus equally applicable across the entire set of languages that we want to compare.
For a linguistic comparison within the field of tempo-aspectuality, such a stable tertium
must necessarily reflect the structures of general conceptual space because, as mentioned
earlier, all languages have to cope with the uniform task of ‘coding’ these notional
structures into linguistic form (cf. Langacker 1976). The adoption of such purely semantic
parameters alone, however, can “be inadequate and misleading” and must therefore be
“constrained formally” (original emphasis), as rightly argued by Krzeszowski (1984:302).
In fact, only through the principled implementation of a combined semantic and formal
tertium comparationis can we (1) confirm the essential functional equivalence of the
compared forms and (2) guarantee a strict correlation between the relevant semantic
components on the one hand and grammatical structure on the other. In the following
section, we will show how such a complex tertium comparationis can be generated from
certain cross-linguistic tendencies revealed in typological studies and how this step can
help us to create a valid methodological tool for the comparison of tempo-aspectual
reference.
2.2 A typologically based comparative tool: the ‘meta-category’
Croft (2008:5) formulates one of the fundamental assumptions of linguistic typology
when he states that “each language expresses the same meanings or functions,” but “the
encoding can vary in ways that are not predictable from how meaning is encoded in one
particular language.” The emphasis here lies on ‘one particular language’, because what
typologists are concerned with is the detection of “language universals via cross-linguistic
generalizations” (Croft 2001:7). These can only be achieved through the study of large
~ 28 ~
Susanne Schneider: Presently comparing pre-past
samples of languages which are representative of a wide range of language families and
geographical regions. Bybee (1985), Dahl (1985) and Bybee et al. (1994) are perhaps the
most prominent typological works dedicated to the grammatical codification of tempoaspectual notions. Whereas Bybee (ibid) examined secondary sources (mainly reference
grammars) for a sample of 50 maximally unrelated languages, extending the sample to
comprise 76 languages in the co-authored publication, Dahl (ibid) analysed primary data
gathered with the help of a translation questionnaire for a total of 64 languages. The
striking overall similarity of their discoveries led these scholars to integrate their work
and to develop what has come to be referred to as the ‘Bybee & Dahl approach’ (Dahl
2000), the groundwork for which they laid in a joint paper published in the end of the
1980s (Bybee and Dahl 1989). In this article, the authors show that tempo-aspectual
notions not only generally tend to find expression in grammatical rather than in lexical
means, but they also tend to be encoded by similar verb-relating grammatical morphemes
(‘grams’) across various languages. In fact, the grouping of language-specific markers
according to the meanings they express results in a restricted set of cross-linguistic types
(‘gram-types’), each of which appears to manifest a clear predilection for certain regular
means of expression which have developed along similar pathways from very similar
lexical sources.
In the multidimensional space of tempo-aspectuality, gram-types thus constitute focal
areas which are very likely to be grammatically encoded in the languages of the world. To
the extent that these gram-types are indeed idealised cross-linguistic constructs, they can
be introduced into our theory of linguistic comparison, where they can function as a valid
tertium comparationis. In fact, the adoption of these gram-types effectively helps us to
meet two important methodological requirements. Firstly, it provides us with the
necessary stable parameters through which we can evaluate and compare the languagespecific categories that we are interested in. Secondly, it supplies a full-fledged
metalanguage, i.e. an inventory of cross-linguistically valid labels, in which we can
phrase our statements about the selected languages without running the risk of using
incommensurable terms.
Cross-linguistic gram-types thus enable and facilitate comparison. They allow us to
determine which language-specific forms can be compared meaningfully and list a
number of potential semantic and formal properties against which we can check these
forms. While any large-scale typological investigation would probably come to a halt at
this rather superficial level of analysis, due to the elevated number of languages under
scrutiny, it is a great advantage of more narrowly circumscribed comparative studies,
such as this one, that greater attention can be paid to even the subtler peculiarities of the
relevant language-particular markers. On the one hand, such an in-depth analysis allows
for the determination and detailed description of the extended semantic space that a
language-specific item covers and of which the cross-linguistic focus typically constitutes
but the central part. On the other hand, it can help identify alternative strategies which a
particular language may employ in order to reveal certain tempo-aspectual notions and
which, consequently, are in constant competition with the canonical marker under
investigation. In addition, the extended meaning area encompassed by a certain particular
tempo-aspectual marker in one language can then be projected onto the other language(s),
~ 29 ~
Susanne Schneider: Presently comparing pre-past
whereby parallel semantic extensions and/or functional overlaps might be uncovered. The
detection of such multi-dimensional correspondences can shed light onto interlinguistically recurring connections between conceptual categories and linguistic
structures. While these connections might perhaps not qualify as linguistic universals,
they can nevertheless help create new and challenging cognitive perspectives.
3 THE EXPRESSION OF ‘PAST-PERFECTNESS’ IN ENGLISH AND ITALIAN
3.1 The gram-type PRF-PST
Cross-linguistic investigations into the expression of tempo-aspectual reference have
shown that in addition to the six major gram-types, namely PFV (perfective), IPFV
(imperfective), PROG (progressive), FUT (future), PST (past) and PRF ([present]
perfect),8 many languages possess a dedicated marker for the codification of the notion of
pre-past. Formally, these constructions appear to be analysable as straightforward
combinations of the gram-types PRF and PST. While scholars such as Thieroff (1994)
and Bybee et al. (1994) treat language-specific grams of this kind in exactly this
simplified way, Dahl and Velupillai (2005:271) point out that, semantically, such markers
have a “relatively strong tendency to develop noncompositional readings; that is they
become semantically independent of pasts and perfects.” In this paper, we will recognise
the afore-mentioned formal parallels to other gram-types by choosing the composite PRFPST (past perfect) label rather than the abbreviation PLPFCT (pluperfect) that was
introduced by Dahl (1985). In our semantic analysis, however, we will not assume
compositionality throughout, but include ‘noncompositional meanings’, too.9
In general typological terms, the gram-type PRF-PST typically indicates that “there is a
reference point in the past, and that the situation in question is located prior to that
reference point” (Comrie 1985:65). This relation of pure anteriority may, in certain
contexts, imply that the situation is also relevant to the past reference time. The fact that
this relevance can be of different kinds is reflected by the various specific uses to which
the language-particular markers belonging to the gram-type PRF-PST can regularly be
put. The experiential use, for instance, indicates that, due to pre-past experiences, certain
qualities can be ascribed to the agent at a past reference point. And while the inclusive
past use represents a pre-past situation as leading up to a past reference point and maybe
beyond it, the resultative use points to the persistence of the result of a pre-past situation
at such a past reference time. If all these contexts, which share relevance as a semantic
8
The upper case denominations serve as labels for cross-linguistic gram-types. While typologists often
make up their own abbreviations, these labels have been adopted from the ‘List of Standard Abbreviations’
provided in Comrie et al. (2004). The use of square brackets in the unabbreviated denomination of the
gram-type PRF highlights the fact that though this gram-type is simply called PERFECT, what it is usually
taken to designate is the cross-linguistic category of present perfect.
9
Incidentally, this seems to be the strategy also followed by Johanson (2000), who chooses to use a
composite label, i.e. his very own ‘+PAST (+POST)’, for the cross-linguistic gram-type in question, yet
explicitly states that “so-called Pluperfects do not have the same semantic structure as so-called Perfects”
(ibid, 107).
~ 30 ~
Susanne Schneider: Presently comparing pre-past
feature, are perfectal and thus attributable to the common meaning component PRF, there
are a number of additional, non-focal uses which are not characteristic of PRF itself. On
the one hand, markers of PRF-PST often tend to be employed in reference to situations
which are located in the distant past (cf. Binnick 2001:561). On the other hand, they
frequently co-occur with definite time adverbials, thereby allowing for the pre-past
situation to be explicitly anchored in the pre-past period or, as Johanson (2000:108) terms
it, to be referred to ‘historically’ rather than ‘diagnostically’.
The extended semantics which typological studies assign to the gram-type PRF-PST are
often explained on the grounds of the cross-linguistically typical morphosyntactic features
as well as the systemic values by which the various language-specific manifestations of
this gram-type are characterised. In general, PRF-PST grams tend to be combinations of
perfect and past. In fact, analogous to markers of PRF, instances of PRF-PST are rather
consistently expressed periphrastically, involving an auxiliary in combination with a past
participle (or similar form) of the lexical verb. The additional past meaning component
essential to the gram-type PRF-PST is carried by the auxiliary of the construction, which
thus may itself be an instance of the gram-types PST, PFV or IPFV. While a PST gram is
the only option for languages that have no PFV:IPFV distinction, all languages that do
make this distinction could in theory form two different markers of PRF-PST. However,
typological studies have revealed a strong tendency for the regular use of only one past
perfect construction, at least with respect to European languages (Johanson 2000:120).
This implies that markers of PRF-PST, unlike those of PRF, frequently lack competing
grams and that they are thus likely to constitute functional parallels of both the perfect
and the past in the pre-past stratum.
Typological literature also supplies more detailed information about the
grammaticalisation of PRF(-PST) grams. Diachronically, (past) perfects appear to be
derived from at least three different sources, namely from (1) resultative constructions
involving a past participle with or without a copula, (2) transitive possessive
constructions and (3) constructions comprising words such as ‘finish’, ‘come from’,
‘throw away’ or ‘already’ (cf. Dahl and Velupillai 2000). While these lexical sources
represent the first stages of a common grammaticalisation path leading towards perfect
grams, interesting observations have been made with respect to the most probable
continuation of this path. There is, in fact, a tendency for PRF and PRF-PST grams to
develop into markers of recent and remote past, respectively.
We can now systematise these typological insights by means of the comparative tool
which we developed in section 2.2. Around the typological meaning label PRF-PST, we
can thus build the cross-linguistically typical core section of the ‘meta-category of pastperfectness’ (figure 1) by associating the regularly recurring meanings (or uses) as well as
the underlying conceptual space to one side of the model and by supplying empty slots
towards the other side into which any language-particular denominations and
constructions can be inserted.
~ 31 ~
Susanne Schneider: Presently comparing pre-past
Conceptual
Space
Meanings
(focal, extended)
Typological
Standard(s)
Form Term
Morphological TempoForm
aspectual
Reference
p
a
s
t
p
e
r
f
e
c
t
n
e
s
s
anteriority/relevance to past R
PRF-PST
___________
___________
experiential past perfect
PRF-PST
___________
___________
inclusive past
PRF-PST
___________
___________
resultative past perfect
PRF-PST
___________
___________
remote past
PRF-PST
___________
___________
‘historical’ past reference
PRF-PST
___________
___________
PAST
PERFECTAL
PRE-PAST
PERFECTIVE
Figure 1: The cross-linguistic standard PRF-PST as characterised in typological research
As can be seen, the above figure constitutes a simple graphical template which
summarises the general tendencies that characterise the codification of the basic cognitive
category of past perfectness in a large number of the world’s languages. We can now
apply this template in parallel fashion to our selected languages in order to show the
similarities and differences regarding the particular restructurings and extensions of this
essentially universal area of conceptual space in English and in Italian. The following
subsections of this paper reflect the fact that such a practical application can be based on
two different types of data, namely (1) secondary data gained from published language
descriptions and (2) primary data obtained by working with native speaker informants. In
what follows we will not only present the relevant information that these sources reveal,
but we will also discuss the relative appropriateness of both sources.
3.2 The ‘meta-category of PAST PERFECTNESS (PRF-PST)’
3.2.1 Secondary data
English and Italian undoubtedly belong to those languages whose tempo-aspectual
systems have been described extensively. Nevertheless, if this were to lead us to expect to
find a vast pool of pre-processed data relevant to our comparative study, we would soon
be proven wrong. In fact, Salkie’s (1989:1) statement about the “relatively little attention”
that scholars had paid to the English marker of past perfectness by the time he was
writing his article appears to be valid even today. What is more, it seems to apply equally
well to the literature available on Italian. While this marked scarcity of in-depth studies
constitutes a potential source of descriptive gaps,10 we can fruitfully integrate the few
available texts with the condensed accounts given in English and Italian reference
grammars as well as with the brief remarks that typological scholars make about these
10
Notable exceptions are the already mentioned Salkie (1989) as well as Declerck (2006) for English,
Bertinetto (1986) for Italian and Squartini (1999) for both languages.
~ 32 ~
Susanne Schneider: Presently comparing pre-past
two languages. Such a combination of different secondary sources will allow us to
compile a detailed state-of-the-art report. If the systematisation of the available secondary
data is carried out according to the conventions of our comparative tool, all observations
can be phrased in a uniform meta-language. Moreover, the resulting descriptive account
remains flexible enough to incorporate any improvement and extension that a subsequent
study of primary data is likely to suggest.
A first projection of the cross-linguistic standard PRF-PST (cf. figure 1) onto the
grammatical system of English clearly identifies the morphologically composite
construction had V-en as the relevant language-particular manifestation form. On closer
analysis, this verbal complex, which freely combines with verbs of all actionality classes,
consists of the Simple Past form of the auxiliary have followed by the past participle of
the lexical verb. The English Past Perfect thus turns out to conform closely to the crosslinguistic tendencies regarding the formal make-up of PRF-PST markers. The specific
choice of have as the only possible auxiliary in present-day English11 points towards the
form’s diachronic origin in a transitive possessive construction and identifies it as an
instance of the so-called ‘have’ (past) perfects (cf. Lindstedt 2000:367).
While had V-en thus constitutes the only dedicated marker of PRF-PST in English, the
Italian tense and aspect system shows a greater variety of forms. The application of the
cross-linguistic standard to Italian results in the identification of two relevant
constructions. The composite aveva V-ato, on the one hand, constitutes another instance
of a ‘have’ (past) perfect, this time with the auxiliary avere (‘have’) conjugated in the
Imperfetto (a past imperfective tense).12 The construction era V-ato, on the other hand,
has the auxiliary essere (‘be’), again marked for the Imperfetto, precede the past
participle of the lexical verb, which exhibits gender and number agreement with its
subject.13 It can thus be regarded as a ‘be’ (past) perfect (cf. Lindstedt 2000:367), whose
diachronic source can be traced back to a resultative construction involving a copula and
a past participle. The co-existence of these two composite markers of PRF-PST does,
however, not usually result in a direct competition of these constructions, since their
distribution is strictly dependent on the lexical verb and, therefore, complementary.14
Both constructions, in fact, are traditionally considered to be instantiations of the Italian
Trapassato Prossimo (or Piuccheperfetto), a tempo-aspectual form which freely combines
11
A detailed account of the history of the have V-en construction is given in Bybee et al. (1994), where
some mention is made of a parallel be V-en construction which, however, no longer exists as a regular
formation.
12
The form aveva V-ato represents the third person singular. While the past participle is the same for all
persons, the auxiliary is conjugated as follows: avevo, avevi, aveva, avevamo, avevate, avevano.
13
The cited form of the past participle, V-ato, represents the form used with a masculine singular subject.
The remaining forms are V-ata (feminine, singular), V-ati (masculine, plural) and V-ate (feminine, plural).
The conjugational paradigm of the auxiliary essere in the Imperfetto is the following: ero, eri, era,
eravamo, eravate, erano.
14
In general terms, the copula-based construction era V-ato is used with all reflexive and pronominal verbs,
while aveva V-ato regularly occurs with all transitive verbs. While most intransitive verbs typically select
either the one or the other construction, verbs expressing “meteorological or atmospheric conditions… may
take either auxiliary” (Maiden and Robustelli 2000:267).
~ 33 ~
Susanne Schneider: Presently comparing pre-past
with verbs of all actionality classes and has the defining characteristic of featuring an
auxiliary marked for the imperfective rather than perfective past. Italian thus clearly
exhibits a basic PFV:IPFV distinction and, in accordance to what we observed in section
3.1, a formally analogous construction involving avere or essere in the Passato Remoto (a
perfective past tense) is indeed a possible formation. However, there are good reasons
against counting these two complementary Trapassato Remoto constructions as
manifestations of the cross-linguistic gram-type PRF-PST. On the one hand, unlike
aveva/era V-ato, the forms ebbe/fu V-ato are extremely rare in Modern Italian and
virtually absent from everyday language (cf. Bertinetto 1986:467). On the other hand,
even the few and distinctly archaic instances of these forms which might occur in high
register literary texts are subject to severe actionality restrictions and strict syntactic rules.
Due to the very specific relational meaning expressed by the Trapassato Remoto, namely
that of a completed past event that is adjacent to a subsequent past event, the two relevant
constructions only select telic (or contextually telicised) verbs in temporal subordinate
clauses which are introduced by a subordinating conjunction suggesting immediate
anteriority with respect to a situation encoded by the Passato Remoto in the main clause.
A brief check against the cross-linguistic standard as described and illustrated in section
3.1 makes clear that any form with such a limited distribution and usage cannot qualify as
a regular manifestation of the gram-type PRF-PST. We will therefore ignore this form in
what follows and code possible occurrences of it as <PRF-PST, i.e. as a construction
which is similar, but not close enough to the cross-linguistic standard PRF-PST. Not only
do these observations on Italian confirm the typological tendency towards the regular use
of only one marker of PRF-PST per language, but they also suggest that the Italian
Trapassato Prossimo is indeed a general past perfect despite the aspectually
circumscribed imperfective morphology exhibited by its auxiliaries. Focusing on the
semantic analyses found in secondary literature, the following sections will clarify this
point.
As we have seen in the preceding paragraphs, both English and Italian feature one
language-specific tempo-aspectual category which, according to morphological
parameters, closely corresponds to the formal tendencies discerned for the cross-linguistic
standard PRF-PST. However, as is commonly emphasised by typologists and argued for
in this paper with regard to the Passato Remoto, morphological criteria alone do not
suffice in order for an item to qualify as a marker of a certain gram-type. On the contrary,
such formal features rather “appear to be a regular correlative of the specific semantics of
these forms” (Schneider 2009:288). Hence, it is essential that the two identified
constructions closely comply with the focal semantic features ascribed to the crosslinguistic standard. In reference grammars of English and Italian, we find very similar
descriptions of the central meaning expressed by the Past Perfect and the Trapassato
Prossimo, respectively. Both categories are, in fact, understood to represent situations
which are anterior to a past reference time, i.e. a reference time which is anterior to the
time of utterance (cf. Dardano and Trifone 1995:356 for Italian, Huddleston and Pullum
2002:140 for English). This univocal definition, formulated against the background of
two different language-particular tense and aspect systems, repeats almost literally the
cross-linguistic definition cited in section 3.1. Paying closer attention to the contexts in
which the Past Perfect and the Trapassato Prossimo are shown to occur, it becomes clear
~ 34 ~
Susanne Schneider: Presently comparing pre-past
that both forms cover all of the focal uses listed for the gram-type PRF-PST (figure 1) –
i.e. they typically encode the relational meanings of anteriority and relevance and are
regularly found in past experiential, past inclusive and past resultative contexts. Due to
their common relational value, both language-particular markers of PRF-PST are very
likely to appear in temporal subordinate clauses. In this specific syntactic environment,
they represent situations as anterior and relevant to the situation expressed by the
predicate in the main clause, which may be marked for any of the language-specific past
tenses. It is interesting to note that in cases where the temporal conjunction itself
explicitly encodes the anteriority relation, the Past Perfect and the Trapassato Prossimo
are sometimes substituted by the Simple Past (PST) and by the Passato Prossimo (<PRFPRS), respectively (cf. Leech 1971:43).
Despite their predilection for syntactically subordinate positions, the English and the
Italian PRF-PST markers are also frequently found in independent sentences where they
either rely on sentence-internal temporal adverbials or on the linguistic co-text for their
past reference times. A closer look at the specific functions of these adverbials reveals
that while it is true that they are a common means for providing the relevant past
reference times from which anteriority relationships are computed, in some contexts they
may also specify the specific times at which the pre-past situations represented by the
PRF-PST constructions had obtained. Both language-particular forms can thus refer
diagnostically as well as historically to pre-past situations. Squartini (1999), relying like
many other scholars on the terminology introduced by Comrie (1976:56), adopts these
two general uses as the so-called ‘perfect-in-the-past’ and ‘past-in-the-past’ functions into
his analysis of the English Past Perfect and the Italian Trapassato Prossimo.15 He also
adds a third semantic value, namely the establishment of ‘past temporal frames’, which he
shows to be a rather common feature of the Italian PRF-PST marker. What is especially
interesting is the fact that in this specific usage the Trapassato Prossimo no longer appears
to express the relational meaning of anteriority, but rather to locate a situation directly in
what, strictly speaking, should be a pre-past period. Since such contexts, however,
typically lack an intermediate past reference time, the Trapassato Prossimo is interpreted
as underlining the relative remoteness of the time at which the speaker perceives the
encoded situation to have obtained.16 Interpreting the term ‘remoteness’ as a measure of
psychological rather than exclusively temporal distance, the ‘remote past’ meaning
clearly features among the extended semantics encompassed by the Italian Trapassato
Prossimo. While, in this respect, Squartini remains rather vague in his remarks on
English, Comrie (1985:68/69) provides the relevant information about the Past Perfect,
15
In his paper, Squartini (1999) actually uses the term ‘Pluperfect’ in order to refer to the cross-linguistic
gram-type as well as to the relevant language-specific categories, thereby conflating theoretical concepts
belonging to two different descriptive levels in one single term. Since such broadly defined technical terms
may easily be the source of misinterpretations, we will stick to our strategy of distinguishing carefully
between these levels, not only from the conceptual, but also from the terminological point of view.
16
Since this particular usage of the Trapassato Prossimo is not very frequently described, the following
examples are adopted from Squartini (1999:58) for sake of clarity: Quel disegno lo avevo fatto io il primo
giorno che lavoravo all’istituto. (‘I made (lit., had made) that drawing on the first day I worked at the
institute.’); Chi avevi conosciuto, quando eri stata a Pisa? (‘Who did you meet (lit., had met), when you
were (lit., had been) in Pisa?’).
~ 35 ~
Susanne Schneider: Presently comparing pre-past
stating that its use “to indicate temporal remoteness when there is no intervening
reference point available from the context will simply disorient the English speaker.”
The English, unlike the Italian, PRF-PST marker thus fundamentally depends on the
existence of a past reference time, which must be recoverable from the surrounding
context. This, however, does not mean that the relevant reference time must be mentioned
explicitly in the linguistic co-text. One usage in particular adequately demonstrates that
the implicit presence of a past reference time in the general discourse situation may in
some cases suffice. Relying on Declerck’s (2006:508) analysis of English, we can
describe this specific use as referring to “a durative situation [which] has never actualized
in a period up to t0 but is actualizing at t0” (‘t0’ being the time of utterance). Squartini
(1999:57), attesting referring to instances of the Trapassato Prossimo in very similar
contexts, gives a more condensed description of this extended usage, observing that it
expresses ‘reversed results’.17 Neither of the two authors, however, claims an obligatory
use of the respective PRF-PST marker in these particular contexts. On the contrary, the
Present Perfect (PRF-PRS) and the Passato Prossimo (<PRF-PRS) are cited as possible
(even if not strictly synonymous) language-particular alternatives.
In addition to these temporal meanings, the English and Italian markers of PRF-PST also
cover a number of modal notions. The specific areas of modal meaning they encompass
and the syntactic environments in which these occur, however, differ to a rather large
extent. In fact, while the Past Perfect constitutes the canonical means for the expression of
hypothetical situations in the protases of counterfactual sentences, the Trapassato
Prossimo may occur in a similar position only in distinctly informal discourse. In this
case, it is often also employed in the accompanying apodoses. Complement clauses of
desiderative verbs constitute yet another syntactic environment in which the English
marker of PRF-PST may be used to represent hypothetical situations which, if actualised,
would have obtained before the moment of utterance. While the Trapassato Prossimo
never occurs in such contexts, it has free access to another modal notion which is only
rarely encoded by the English Past Perfect. Italian speakers, in fact, may employ the PRFPST marker in order to represent their current hopes, intentions or desires in a less direct,
less pressing way. This so-called attenuative use of the Trapassato Prossimo (and the Past
Perfect) thus allows the speaker to formulate polite indirect requests. A similar courteous
overtone may, however, also be achieved by the Imperfetto (IPFV) and the Simple Past
(PST), respectively.
17
As before, here are two examples adopted from Declerck (2006:508) and Squartini (1999:57) that
illustrate this somewhat rarely described usage of the Past Perfect and the Trapassato Prossimo,
respectively: I had never dreamed of meeting these people before.; Me lo aveva promesso, ma adesso fa
finta di non ricordarsene. ‘He did promise (lit., had promised), but now he pretends not to remember it.’
Note that while the English sentence, in accordance with Declerck’s definition, contains a durative
predicate, the Italian example features a non-durative verb in the Trapassato Prossimo, suggesting the
existence of subtle inter-linguistic differences with respect to this usage. Our analysis of primary data in
section 3.2.2 will help us to throw some light on this problem.
~ 36 ~
Susanne Schneider: Presently comparing pre-past
If the English and the Italian manifestations of PRF-PST have been shown to differ
slightly with respect to the semantic space they cover as well as to the syntactic
environments they may occur in, a further difference can be noted on the morphosyntactic
level. Focusing on the general compatibility of these forms, we can observe that only the
English Past Perfect may co-occur with an additional tempo-aspectual marker, namely the
Progressive (PROG). This complex construction is typically observed in past inclusive
contexts, where it suggests that a situation which had begun in the pre-past continued on
up to the past reference time and beyond it.
3.2.2 Primary data
The study of secondary sources in the previous section has helped us to identify the
language-particular forms that are most relevant to our investigation into the linguistic
expression of past perfectness in English and Italian. What is more, it has resulted in a
general outline of the major similarities and differences between these constructions,
showing that while there is a basic congruence between the two at the level of their focal
meanings (this congruence having been the reason for identifying these markers as the
relevant language-specific manifestations of PRF-PST in the first place), important
differences pertain in the extended, more marginal meaning areas they cover. Apart from
the few uncertainties which were shown to be a direct result both of the relative scarcity
of secondary material as well as of the vagueness of some descriptions, we now have
quite clear a picture of what the Past Perfect and the Trapassato Prossimo are capable of
expressing. But do native speakers actually use these constructions or are they just
optional devices which can and usually are replaced by alternative language-specific
forms in natural discourse? Coseriu (1972:56) rightly points out that in order to draw a
comprehensive comparison between two or more languages, it is not sufficient that we
know what a speaker might be able to say, but we must also know what s/he is most
likely to say in a particular context.18 Not only can primary data help us to acquire exactly
this kind of knowledge, but it can also cross-check, illustrate, refine and complete the
global theoretical information provided by secondary sources.
In order to obtain detailed primary data regarding the codification of tempo-aspectual
notions in English and Italian, an elicitation questionnaire was constructed and then
translated into both languages. Modelled on the cross-linguistic survey conducted by Dahl
in 1985 and on the various EUROTYP questionnaires published in a volume edited by
Dahl in 2000, the ATAM-questionnaire19 used here consisted of 230 sentences and short
texts which were accompanied by brief indicators of context and whose predicates were
supplied in the infinitive (printed in capitals). In a field study, the English and Italian
versions of the questionnaire were circulated at random to university students in the
18
The original statement on which this translation is based runs as follows: “Es genügt also nicht zu wissen,
was man in einer Sprache sagen könnte, man muß auch wissen, was normalerweise in bestimmten
Situationen gesagt wird” (Coseriu 1976:56).
19
This denomination employs the acronym ATAM introduced by Bertinetto and Nocetti (2006), which aptly
summarises the four dimensions that together form the semantic domain of verbal time, namely
A(ctionality), T(emporal Reference), A(spectuality) and M(odality).
~ 37 ~
Susanne Schneider: Presently comparing pre-past
United Kingdom, the United States of America, Canada and Italy, who were asked to
complete each of the sentences with the form of the verb that seemed most natural to
them. The following discussion of primary data is based on a random sample of ten of
these questionnaires per language.20
The analysis of the English and Italian questionnaire samples has revealed significant
inter- as well as intra-linguistic variation in the frequency of occurrence of the languageparticular PRF-PST markers. English informants employed had V-en 16 times on average
(with individual totals ranging from a minimum of 9 to a maximum of 24 occurrences).
By contrast, Italian speakers used aveva/era V-ato between 4 and 13 times, thus
averaging only 10 instances per speaker. While the overall low number of occurrences of
the PRF-PST markers can be partly explained by the conceptual complexity of the prepast and by the fact that human beings have a naturally stronger preoccupation for all
situations that make up, or directly influence, their present moment of experience (both
factors, of course, directly influenced the questionnaire design), the gap between the
respective average for the two groups suggests that there may be indeed no absolute
functional equivalence between the English and Italian markers of PRF-PST. In addition,
the fact that intra-linguistic variation is shown to more than double (in English) or even
more than triple (in Italian) the minimum counted representatives of PRF-PST suggests
that individual and/or regional preferences may also influence the selection of PRF-PST
devices.
Analysing in detail the questionnaire sentences in which language-specific markers of
PRF-PST occur, it becomes apparent that, quite predictably, most of the focal meanings
are rather consistently expressed by the Past Perfect and the Trapassato Prossimo,
respectively. The first example given below, which elicited PRF-PST markers
throughout, features a situation which is not only anterior and relevant to another past
situation, but at the same time part of a past habitual scenario, illustrating that both
language-specific forms are easily compatible with habitual meaning:21
(1)
[The speaker used to meet his friend once a week. Nowadays he never sees him:]
a. Every time I met him in those years, he would tell me about the film he had
just seen. [10/10]
b. Ogni volta che lo incontravo in quegli anni mi parlava del film che aveva
appena visto. [10/10]
20
The English sample comprises one informant from each of the following cities: Cardiff (GB), Dublin
(IRE), Edinburgh (GB), Edmonton (CA), Leeds (GB), London (GB), New York (USA), Nottingham (GB),
Reading (GB) and West Bromwich (GB). As regards the Italian informants, there are two speakers from
Udine, and one speaker from each of the following Italian cities: Brindisi, Genoa, Lecce, Pisa, Rome,
Salerno, Venice and Vicenza.
21
Though directly adopted from the ATAM-questionnaire, some of the sentences reproduced in this paper
are likely also to have appeared in the previously published surveys on which this questionnaire has been
modelled. For reasons of space, the contexts accompanying the questionnaire sentences are supplied here in
English only, while answers are cited verbatim in both languages. Square brackets will provide the number
of occurrences of the PRF-PST marker.
~ 38 ~
Susanne Schneider: Presently comparing pre-past
The counted occurrences of the English and Italian PRF-PST markers, however, turn out
to be distinctly lower in cases where the anteriority relation is explicitly spelled out by
some other form in the linguistic context. Sentence (2) is an example of a relevant
anterior situation encoded in a subordinate clause which is introduced by the temporal
conjunctions after/dopo che. As the informants’ answers show, the conceptual
redundancy of an additional marker of anteriority rather frequently results in a
substitution of the Simple Past (PST) or Passato Prossimo (<PRF-PRS) for the respective
language-particular PRF-PST constructions:
(2)
a. (Yesterday,) Mary came home after John had arrived / arrived. [05/10]
b. (Ieri,) Maria è tornata a casa dopo che Giovanni era arrivato / è arrivato.
[07/10]
As we noted earlier, the notion of relevance at a past reference time can be of different
kinds, resulting in a number of distinguishable uses of PRF-PST. The following sentence
constitutes an example of an experiential context. While both the English and the Italian
groups display a decided preference for the respective language-specific PRF-PST
marker, two of the English informants alternatively resort to the Simple Past (PST) and
one of the Italian informants to the Passato Prossimo (<PRF-PRS). By doing so, these
speakers seem to treat the encoded pre-past situation as mere fact rather than to explicitly
assert its relevance at the given past reference time:
(3)
[Q: When you came to this place a year ago, did you know my brother? A:]
a. (Yes,) I had met / met him at least once before I came here. [08/10]
b. (Sì,) l’avevo incontrato / l’ho incontrato almeno una volta prima di venire
qua. [09/10]
A very interesting phenomenon can be observed with respect to another of the
fundamentally perfectal focal meanings, namely past inclusivity. As we noted earlier,
secondary sources on English typically list such contexts as paradigmatic environments of
the complex Past Perfect Progressive (PRF-PST-PROG). The questionnaire sentence
given in (3), however, illustrates that some speakers prefer to use the non-progressive
Past Perfect construction in these contexts:
(4)
a.
b.
When I found her (yesterday), she had already knocked / had already been
knocking at our neighbour’s door for half an hour. [02+08/10]
Quando l’ho trovata (ieri), aveva bussato / stava bussando alla porta del
nostro vicino già da mezz’ora. [02/10]
The juxtaposed Italian data is even more remarkable. While the Perifrasi Progressiva
stava V-ando (PROG-PST), i.e. the Italian periphrastic marker of progressivity, here
marked for the Imperfect, was employed by most informants, two instances of the
Trapassato Prossimo could be recorded in this distinctly inclusive context. Asking those
informants who had not used the PRF-PST marker to judge the relative acceptability of
this alternative choice, they usually pointed out that the temporal adverbial introduced by
da (‘for’, indicating an interval which is essentially open with respect to its terminal
~ 39 ~
Susanne Schneider: Presently comparing pre-past
point) made the use of the PRF-PST construction highly unlikely, since this verbal form
suggested an interruption of the encoded situation at the past reference point and would
thus rather have to combine with a temporal adverbial introduced by per (‘for’, indicating
a closed interval that measures the duration of the situation from its initial to its final
point). For these speakers, the Trapassato Prossimo thus turned out to be blocked in cases
where the evaluation period was demarcated as extending beyond the reference time.
Back-checking with the two informants who did employ the PRF-PST marker, their
answers invariably contained the Trapassato Prossimo, suggesting the existence of
speaker-related semantic extensions.22 As the following example illustrates, instances of
the Italian PRF-PST marker can be observed to multiply in those contexts where the
situation is understood as leading up to but not beyond the past reference point (past-upto-then). The English sentence shows that such a contextual change may elicit some
instances of the Simple Past (PST):
(5)
a.
b.
When John retired at the age of 70, he was tired because he had worked / had
been working / worked hard all his life. [07+01/10]
Quando Aldo è andato in pensione all’età di 70 anni era stanco perché aveva
faticato / ha faticato tutta la vita. [07/10]
The resultative use is yet another of the focal meanings typically encompassed by PRFPST markers. The questionnaire shows that such contexts consistently elicited the use of
the had V-en and aveva/era V-ato constructions if the past reference time was given in the
immediate linguistic co-text, as illustrated in (6). A delegation of the relevant past
reference time to the wider linguistic context significantly reduced the number of counted
PRF-PST forms and elicited instances of the Simple Past (PST) and the Passato Prossimo
(<PRF-PRS), as shown in (7):
(6)
a.
b.
(7)
When I came home yesterday, he had written two letters [finished before I
arrived]. [10/10]
Quando sono arrivato a casa ieri, aveva scritto due lettere [finite prima del
mio arrivo]. [10/10]
[Q: Did you find your brother at home? A:]
a. (No, we did not, we were unlucky.) He had left / left (just before we came).
[07/10]
b. (No, non l’abbiamo trovato. Siamo stati sfortunati.) Era partito / È partito
(appena prima che noi arrivassimo). [07/10]
22
The fact that the two informants who allow for the PRF-PST marker in (3) are from the Italian cities of
Lecce and Salerno might suggest that this semantic extension is a phenomenon more typical of the Southern
than of the Central or Northern Italian varieties. However, since the primary data presented in this paper is
based on a mere ten questionnaires per language, it is difficult to make any reliable statements about
regional variation. What is more, the Southern Italian speaker from Brindisi does not employ the PRF-PST
marker and thus runs contrary to the observed tendency.
~ 40 ~
Susanne Schneider: Presently comparing pre-past
An even more pronounced avoidance of the PRF-PST marker can be observed among the
ten English informants in what secondary sources refer to as the ‘reversed result’ use. As
noted in section 3.2.1, these specific contexts lack explicit mentioning of an appropriate
past reference time and rather rely on an implicit reference time provided by the general
discourse situation. Sentence (8) gives an example of this particular use. Analysed in
detail, it presents a durative situation which has never actualised before but is actualising
at the time of utterance. Judging from the counted occurrences of the language-particular
PRF-PST markers, the Past Perfect, unlike the Trapassato Prossimo, proves to appear
only very sporadically in these contexts:
(8)
[A and B are in a room of B’s house. Leaving the room, B asks A:]
a. Had you been / Were you / Have you been in this room before? [01/10]
b. Eri stato / Sei stato in questa stanza prima? [07/10]
Interestingly, the number of English PRF-PST markers can be observed to increase
slightly in (9), where a telic rather than an atelic situation is referred to. What is especially
interesting with respect to this particular sentence is that the past reference time appears
to be established by the help of a physical image of the past, i.e. the picture that the
interlocutors are looking at. The use of the PRF-PST marker in this specific context
explicitly locates the situation it refers to prior to the time which is immortalised in the
picture and thus seems to insert it into a kind of temporal frame situated in a relative
remote past. Quite predictably, the implicitness of the intermediate past reference time in
this as well as in the preceding example allows for other tempo-aspectual forms to occur,
such as the Simple Past (PST) and the Present Perfect (PRF-PRS) in English and the
Passato Prossimo (<PRF-PRS) as well as the rarer Passato Remoto (PFV) in Italian:
(9)
[Looking at a picture of a house which has been torn down:]
a. Who had built / built this house? [02/10]
b. Chi aveva costruito / ha costruito / costruì questa casa? [06/10]
The questionnaire also elicited several occurrences of the English PRF-PST marker
encoding distinctly modal meanings. In an example such as (10), the Past Perfect is
employed to encode a counterfactual condition. While some informants avoided the PRFPST construction in this sentence and employed the Simple Past (PST) instead, the vain
wish expressed in (11) elicited no alternative tempo-aspectual forms among the English
speakers. The ten Italian informants, by contrast, consistently resorted to an analytical
marker of the subjunctive mood, namely the Congiuntivo Trapassato (SUBJ-PRF-PST),
in order to encode these modal notions:
(10)
[The speaker knows that the boy was expecting money and that he did not get it:]
a. If the boy had got / got the money (yesterday), he would have bought a
present for the girl. [07/10]
b. Se il ragazzo avesse ricevuto i soldi (ieri), avrebbe comprato un regalo per la
ragazza. [00/10]
~ 41 ~
Susanne Schneider: Presently comparing pre-past
(11)
[Yesterday, A went shopping with her boyfriend. They saw a beautiful but much
too expensive dress. Today, A tells her best friend about the dress and exclaims:]
a. (Yesterday in the shop,) if only he had had enough money on him! [10/10]
b. (Ieri nel negozio,) se solo avesse avuto abbastanza soldi con sé! [00/10]
No instance of a Trapassato Prossimo expressing a modal notion could be elicited by the
230 sentences that make up the questionnaire. Of course, this does not imply that the
Italian PRF-PST marker is never actually used to encode modal meaning. It rather makes
us aware of a potential shortcoming of questionnaires used as a methodological tool for
gathering primary data. No questionnaire, in fact, will ever be long enough to include all
the phenomena that an in-depth study of tempo-aspectuality should be able to account for.
Notwithstanding this weakness, the questionnaire method has been shown to give
valuable insight into the use of the PRF-PST constructions in English and Italian. Not
only has the highly comparable data yielded by the ATAM-questionnaire confirmed the
observations made in the secondary literature, but it has also helped us to refine and
extend this set of data.
4 CONCLUSION
Coseriu (1972:41), whom we have already cited several times in this paper, makes the
interesting observation that the fact that ‘a language B can express the same contexts as a
language A does not imply that language B needs to encode these meanings with similar
means or needs to encode them at all.’ Language B, in fact, may either not possess
devices that are similar to those encountered in language A, or it may possess yet not
employ them in the same way, due to the existence of alternative and/or semantically
more general codification strategies. Applied to the topic we have treated in this paper,
these words translate into the following central statement: while both English and Italian
are able to express the notion of ‘past perfectness’, they might not encode it by similar
means nor encode it explicitly on all occasions. At this point, i.e. after the detailed
application of the ‘meta-category’ as a methodologically sound apparatus of linguistic
comparison, created through the strict implementation of a typologically guided
comparative framework, we are able to confirm and corroborate this statement.
As this paper has shown, the conceptual space of ‘past perfectness’ is accessible by
means of a dedicated PRF-PST marking device in both English (Past Perfect) and Italian
(Trapassato Prossimo). While English speakers quite consistently employ this marker to
encode all of the focal past perfectal meanings, the respective Italian marker only occurs
very sporadically in truly inclusive contexts, where the continuance of the situation after
the past reference point favours the choice of an explicitly progressive construction
(PROG-PST), a tendency which is also notable in English, where the PRF-PST marker
combines with progressive morphology. Even in the expression of the remaining focal
meanings, and even in the case of English, the PRF-PST construction rarely constitutes
the only possible marking strategy, but is more or less frequently replaced by the
language-specific present perfect (PRF-PRS, <PRF-PRS) or past tense markers (PST,
PFV). The language-particular manifestations of PRF-PST thus lose some of their
~ 42 ~
Susanne Schneider: Presently comparing pre-past
potential occurrences within the core meaning areas to other tempo-aspectual forms. At
the same time, however, they gain functional load outside the semantic focus. In fact,
while the Trapassato Prossimo, significantly more frequently than the Past Perfect,
registers extended uses expressing ‘relatively remote’ past perfective reference, the Past
Perfect canonically extends to encode counterfactual meaning. Italian, unlike English,
possesses dedicated subjunctive markers, which are typically used in such modal
contexts. The Trapassato Prossimo does thus not usually occur in such modal contexts,
though informal and colloquial speech may provide instances of the Italian PRF-PST
construction encoding counterfactual meaning.
The practical application of the ‘meta-category’ has thus resulted in a comprehensive
parallel description of the grammatical constructions used in two specific languages,
namely English and Italian, to encode ‘past perfectness’ and (closely) related notions. The
fact that the ‘meta-category’ is an illustratable tool represents an additional advantage. As
shown in the Appendix, the insertion of the comparative data (both secondary and
primary) into the graphical template offered by the ‘meta-category’ results in a very
easily consultable record of (1) the grammatical devices which encode the notion of ‘past
perfectness’ in English and in Italian, and (2) the ways in which the language-specific
markers identified as manifestations of the cross-linguistic standard PRF-PST extend
beyond this focal meaning. What is more, the graphical representation of the ‘metacategory’ proves to be an essentially open structure. Not only can it accommodate any
additional variations or changes that may develop in language use, but it may also be
extended to fit the peculiarities of any language-particular system. It is this global
adaptability that renders the ‘meta-category’ a perfect tool for linguists to better
understand the regularities of both cross-linguistic as well as language-particular
structures.
REFERENCES
Bertinetto, Pier Marco. 1986. Tempo, Aspetto e Azione nel Verbo Italiano. Il
Sistema dell’Indicativo. Firenze: L’Accademia della Crusca.
Bertinetto, Pier Marco & Noccetti, Sabrina. 2006. Prolegomena to ATAM Acquisition.
Theoretical Premises and Corpus Labeling. Quaderni del Laboratorio di
Linguistica della Scuola Normale Superiore di Pisa 6.
Binnik, Robert I. 2001. Temporality and Aspectuality. In Haspelmath, Martin, Ekkehard
König, Wulf Oesterreicher & Wolfgang Raible (eds.) Language Typology and
Language Universals: An International Handbook — Volume 1. Berlin: Walter de
Gruyter, 557-567.
Bybee, Joan L. 1985. Morphology: A Study of the Relation between Meaning and Form.
Amsterdam: John Benjamins.
Bybee, Joan L. & Dahl, Östen. 1989. The Creation of Tense and Aspect Systems in the
Languages of the World. Studies in Language 13-1, 51-103.
~ 43 ~
Susanne Schneider: Presently comparing pre-past
Bybee, Joan L., Revere Perkins & William Pagliuca. 1994. The Evolution of Grammar:
Tense, Aspect, and Modality in the Languages of the World. Chicago:
Chicago
University Press.
Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.
Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.
Comrie, Bernard, Martin Haspelmath & Balthasar Bickel. 2004. The Leipzig Glossing
Rules: Conventions for Interlinear Morpheme-by-Morpheme Glosses.
Coseriu, Eugenio. 1972. Über Leistung und Grenzen der kontrastiven Grammatik. In
Gerhard Nickel (ed.) Reader zur kontrastiven Linguistik. Frankfurt am Main:
Athenäum Fischer Taschenbuch Verlag, 39-58.
Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological
Perspective. Oxford: Oxford University Press.
Croft, William. 2008. Methods for Finding Language Universals in Syntax.
Dahl, Östen. 1985. Tense and Aspect Systems. Oxford: Basil Blackwell.
Dahl, Östen (ed.) 2000. Tense and Aspect in the Languages of Europe. Berlin: Mouton de
Gruyter.
Dahl, Östen. 2001. Languages without Tense and Aspect. In Ebert, Karen H. and Zúñiga,
Fernando (eds.) Aktionsart and Aspectotemporality in Non-European Languages.
Zürich: ASAS-Verlag, 159-173.
Dahl, Östen and Velupillai, Viveka. 2005. Tense and Aspect. In Haspelmath, Martin,
Matthew S. Dryer, David Gil & Bernard Comrie (eds.) The World Atlas of
Linguistic Structures. Oxford: Oxford University Press, 266-281.
Dardano, Maurizio & Trifone, Pietro. 1995. Grammatica Italiana con Nozioni di
Linguistica. Bologna: Zanichelli.
Declerck, Renaat. 2006. The Grammar of the English Verb Phrase. Vol. 1: The
Grammar of the English Tense System. Berlin: Mouton de Gruyter.
Huddleson, Rodney & Pullum, Geoffrey K. 2002. The Cambridge Grammar of the
English Language. Cambridge: Cambridge University Press.
Johanson, Lars. 2000. Viewpoint Operators in European Languages. In Dahl, Osten (ed.)
Tense and Aspect in the Languages of Europe. Berlin: Mouton de Gruyter, 27-187.
Krzeszowski, Tomasz P. 1984. Tertium Comparationis. In Fisiak, Jacek (ed.) Contrastive
Linguistics: Prospects and Problems. Berlin, New York and Amsterdam: Mouton,
301-312.
Langacker, Ronald W. 1976. Semantic Representations and the Linguistic Relativity
Hypothesis. Foundations of Language 14, 307-357.
Leech, Geoffrey N. 1971. Meaning and the English Verb. London: Longman.
Lindstedt, Jouko. 2000. The Perfect – Aspectual, Temporal and Evidential. In Dahl, Östen
(ed.) Tense and Aspect in the Languages of Europe. Berlin: Mouton de Gruyter,
365-383.
Maiden, Martin & Robustelli, Cecilia. 2000. A Reference Grammar of Modern Italian.
London: Arnold.
Malotki, Ekkehart. 1983. Hopi Time. Berlin, New York, Amsterdam: Mouton.
Salkie, Raphael. 1989. Perfect and Pluperfect: What Is the Relationship? Journal of
Linguistics 25, 1-34.
Schneider, Susanne. 2009. ‘Progressivity’ in English and Italian: A Typologically Guided
Comparative Study. New Trends and Methodologies in Applied English Language
~ 44 ~
Susanne Schneider: Presently comparing pre-past
Research. In Prado-Alonso, Carlos, Lidia Gómez-García, Iria Pastor-Gómez and
David Tizón-Couto (eds.) Diachronic, Diatopic and Contrastive Studies. Bern,
Berlin, Bruxelles, Frankfurt am Main, New York, Oxford, Wien: Peter Lang, 279307.
Squartini, Mario. 1999. On the Semantics of the Pluperfect: Evidence from Germanic
and Romance. Linguistic Typology 3, 51-89.
Taylor, John R. 2002. Cognitive Grammar. Oxford: Oxford University Press.
Thieroff, Rolf. 1994. Inherent Verb Categories and Categorizations in European
Languages. In Tieroff, Rolf & Ballweg, Joachim (eds.) Tense Systems in
European Languages. Tübingen: Niemeyer, 3-45.
Whorf, Benjamin Lee. 1940. Science and Linguistics. In Bennett, Milton J. (ed.) Basic
Concepts of Intercultural Communication: Selected Readings. 1998. Yarmouth,
Maine: Intercultural Press, 85-96.
ABBREVIATIONS
COND-PRF
FUT
IPFV
PFV
PRF
PRF-PRS
PRF-PST
PRF-PST-PROG
PROG
PROG-PST
PST
SUBJ-PRF-PST
conditional perfect
future
imperfective
perfective
perfect
present perfect
past perfect
past perfect progressive
progressive
past progressive
past
past perfect subjunctive
~ 45 ~
Susanne Schneider: Presently comparing pre-past
APPENDIX
The ‘meta-category of PAST PERFECTNESS (PRF-PST)’ in English
and Italian
CS
p
a
s
t
p
e
r
f
e
c
t
n
e
s
s
m
o
d
a
l
i
t
y
Meanings
Cross-linguistic Form Term
(focal, extended) Category-type(s) in English
in English
Cross-linguistic Form Term
Category-type(s) in Italian
in Italian
anteriority/
relevance
to past-R
PRF-PST
PST
Past Perfect
Simple Past
PRF-PST
<PRF-PRS
{<PRF-PST}
Trapassato Prossimo
Passato Prossimo
Trapassato Remoto
experiential
past perfect
PRF-PST
PST
Past Perfect
Simple Past
PRF-PST
<PRF-PRS
Trapassato Prossimo
Passato Prossimo
inclusive past
PRF-PST
Past Perfect
{PRF-PST}
PRF-PST-PROG Past Perf. Prog. PROG-PST
past up-to-then PRF-PST
Past Perfect
PRF-PST
PRF-PST-PROG Past Perf. Prog. <PRF-PRS
PST
Simple Past
Trapassato Prossimo
Perifrasi Progressiva
all’Imperfetto
Tempoaspectual
Reference
PAST
PERFECTAL
Trassato Prossimo
Passato Prossimo
resultative
past perfect
PRF-PST
PST
Past Perfect
Simple Past
PRF-PST
<PRF-PRS
Trapassato Prossimo
Passato Prossimo
reversed result
PRF-PST
PRF-PRS
PST
Past Perfect
Present Perfect
Simple Past
PRF-PST
<PRF-PRS
Trapassato Prossimo
Passato Prossimo
past in the past PRF-PST
(‘historical’
past reference)
Past Perfect
PRF-PST
Trapassato Prossimo
(remote) past
PRF-PST
temporal frame PST
Past Perfect
Simple Past
PRF-PST
<PRF-PRS
PFV
counterfactual PRF-PST
conditioning sit. PST
Past Perfect
Simple Past
{PRF-PST}
Trapassato Prossimo
SUBJ-PRF-PST Congiuntivo Trapassato
counterfactual COND-PRF
conditioned sit.
Cond. Perfect
{PRF-PST}
COND-PRF
counterfactual
desires
PRF-PST
Past Perfect
SUBJ-PRF-PST Congiuntivo Trapassato
attenuative
statements
PRF-PST
PST
Past Perfect
Simple Past
PRF-PST
IPFV
PRE-PAST
PERFECTIVE
Trapassato Prossimo
Passato Prossimo
Passato Remoto
Trapassato Prossimo
Condizionale Passato
HYPOTHETICAL
PERFECTAL/
PERFECTIVE
Trapassato Prossimo
Imperfetto
[PRF-PST (PAST PERFECT); P(a)ST; PRF-PRS (PRESENT PERFECT); PRF-PST-PROG (PAST PERFECT PROGRESSIVE);
PROG-PST (PAST PROGRESSIVE); P(er)F(ecti)V(e); I(m)P(er)F(ecti)V(e); SUBJ-PRF-PST (PAST PERFECT SUBJUNCTIVE)]
[‘<’ close, but not good enough instance of typological standard; ‘{}’ rarely attested instances; ‘CS’ Conceptual Space]
~ 46 ~
CHAPTER 4
ARE CREOLES TENSELESS LANGUAGES?
A REVIEW OF THE CREOLE TENSE AND ASPECT SYSTEM
ESTHER NÚÑEZ VILLANUEVA
University of Manchester
Following the seminal work of Bickerton (1975; 1981), the Tense and Aspect (TA) system of creoles
has been invariably described as consisting of three preverbal markers expressing relative tense,
mood and aspect. However, more recently, it has also been claimed that creoles are tenseless
languages which mark only aspectual and mood distinctions in the verb phrase (Binnick 1991). The
analyses of data from four different creoles, Guyanese Creole English (Guyanese CE), Haitian
Creole French (Haitian CF), Papiamentu Creole Spanish (Papiamentu CS) and Kituba, suggest that
creoles are aspect-prominent languages, tending to display a tripartite system of perfective, past
imperfective and present imperfective. Following the views of Sankoff (1990) and Sidnell (2002),
the differences among the four creoles analysed could be explained by the different stages of
grammaticalisation of their markers.
1 INTRODUCTION
Despite its relatively short life, creolistics has been a quite divided field. As Winford
(1996:97) expressed, ‘creolists sometimes seem to take more than their fair share of
delight in disagreement, and appear almost proud of the fact that no two creolists appear
to agree on anything’. One of the major bones of contention in the field is the definition
of creole itself and the classification of different contact varieties under such a label.
On one hand, authors such as Mufwene (1996, 1997) have argued that ‘creole’ represents
a sociohistorical label, bonding together languages that were born from similar extreme
situations of multilingual contact. On the other hand, there is the view that creoles are
definable in terms of specific structural features, due to some universal linguistic
principles involved in the genesis of the creole, a view held by Bickerton (1975, 1981)
and McWhorter (1998) who in his paper defined ‘creole’ as a typological class
synchronically distinguishable from other languages.
~ 47 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Bickerton’s influential theories relied heavily on the common patterns that unrelated
creoles show in their Tense and Aspect (TA)23 systems. Bickerton claimed that creoles
shared a system of three preverbal markers expressing:24
(i) Relative past tense, also termed anterior tense, which expresses past tense with
statives and past-before-past with nonstatives.
(ii) Non-punctual aspect, later re-named imperfective aspect, which is restricted in use
to nonstatives and the range of meanings that it covers is described by Bickerton
as ‘progressive-durative plus habitual-iterative’ (Bickerton 1981:58.).
(ii) Irrealis mood, which expresses future and unrealised conditions.
This system is illustrated below by examples provided by Bickerton (1975, 1981) from
Guyanese Creole English (Guyanese CE):
(i) Example of the relative past tense marker bin. In the first instance, bin accompanies
the stative verb gatu ‘to have to’ and has a past reference. In the second instance, it
accompanies the nonstative hapn ‘happen’ and adds the meaning of past-before-past:
(1)
(Example 2.51 in Bickerton 1975:36)
Dem bin gatu get we an kom dis said, lef di plees an get we, bikaz terabl
ting bin hapn wid dem chiren25
‘They had to get away and come over here, leave the place and get away,
because terrible things had happened to their children.’
(ii) Examples of the non-punctual marker a. Example (2) has a progressive reading
while example (3) has an habitual reading:
(2)
(Example 2.3 in Bickerton 1975:29)
mi na no wai dem a du dis ting
‘I don’t know why they are doing this.’
(3)
(Example 2.45 in Bickerton, 1975:34)
evri de mi a ron a raisfiil
‘Every day I hurry to the ricefield.’
(iii) Example of the Irrealis marker go with a future reading:
(4)
(Example 2.80 in Bickerton 1975:42)
Fraidi awi go mek
‘We’ll make [some] on Friday.’
23
TA will be used throughout the paper as an acronym for Tense and Aspect.
Bickerton uses auxiliary, marker and particle interchangeably. I have decided to use marker throughout
this paper since it has become the most popular term in the creole literature.
25
When no interlinear gloss has been provided by the authors, I have highlighted the TA markers for easier
reference.
24
~ 48 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Moreover, he established that the meaning of the unmarked verb stem depended on the
inherent lexical aspect of the predicate. In example (2) above, no is a stative verb and it is
unmarked, therefore it is interpreted as making reference to a present state.
The prototypical creole TA system proposed by Bickerton is summarised in Table 4.1.
The Irrealis marker will not be analysed along tense and aspect in this paper and it has not
been included in the following table.
Stative verb
Progressive and
habitual
Past reference
Nonstative verb
Ø
Non-punctual marker
Relative past tense marker
Ø
Relative past tense marker: pastbefore-past
Table 4.1: The prototypical creole TA system according to Bickerton (1975)
Bickerton’s theories spurned a wealth of publications on creole TA systems in the early
and mid-eighties and more and more creoles were cited as fitting Bickerton’s prototype.
These subsequent analyses mostly rejected the innate explanation for the creole TA
system proposed by Bickerton, the so-called Language Bioprogram Hypothesis, but
adopted without further questioning both the terminology and the morphosyntactic
analysis of Bickerton. Muysken (1981), for example, identifies the three preverbal
markers in at least thirteen creoles, comprising both Atlantic (i.e. Jamaican) and Pacific
creoles (i.e. Indo-Portuguese).
Importantly, Bickerton’s analysis would distinguish creoles from non-creoles. In the latter
cases, a marker of relative past tense will rarely be found. When compared with crosslinguistic studies of TA systems such as the ones conducted by Bybee et al. (1994) and
Dahl (1985), it is evident that ‘relative past’ (or ‘anterior’) tense does not constitute a
cross-linguistic category found in other languages. Is that proof that creoles display a
tense distinction not found in other languages? Two scholars came to a different
conclusion in subsequent years; namely that Bickerton’s relative past tense marker is
indeed another aspectual marker, concretely a perfective marker. Andersen (1990)
reached this very conclusion in his analysis of Papiamentu, a Spanish-based creole, while
Binnick (1991) takes the analysis further when classifying creoles as tenseless languages.
However, only a brief section of Binnick’s general volume on tense and aspect is devoted
to creoles and it is unclear how Binnick arrived at such a conclusion.
The aim of this paper is to investigate the claim that creoles constitute tenseless languages
which mark only aspectual and mood distinctions in the verb phrase, a claim found in
Binnick (1991). Such a claim could constitute an important typological claim regarding
creoles but it is not found in the creole literature. In addition, it could solve the main
inadequacies of Bickerton’s theories. In the first place, it is important to note that this
claim does not imply that all creoles mark the same aspectual distinctions or by the same
~ 49 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
strategies, allowing room for variation among creoles provided that such variation is still
within the category of aspect. Secondly, it proposes a valid parameter for language
comparison, since all languages have some marking of either tense or aspect or both, with
the exception of restricted pidgins.
For this purpose, data from four different creoles Guyanese Creole English (Guyanese
CE), Haitian Creole French (Haitian CF), Papiamentu Creole Spanish (Papiamentu CS)
and Kituba will be analysed.26 Section two gives more details on the sample chosen.
Section three contains a discussion of the data within a neo-Reichenbachian theory of
tense and aspect (Binnick 1991:115f.) and the final conclusions are presented in section
four.
It is important to note that only the markers previously identified as fitting Bickerton’s
model are analysed in this paper. The four creoles studied have further verbal markers,
such as completive markers or infinitive markers. Interested readers can consult the
excellent volume Comparative Creole Syntax, edited by John Holm and Peter Patrick
(2007), in which the grammars of 18 creoles are outlined. Please note that the creole TA
system has been analysed following Bickerton’s theories in this volume.
2 SAMPLE
The data used in my analysis comes from published works. The four creoles analysed
were chosen on the basis of availability of scholarly research and learners’ grammars,
while trying to include as much variety as possible with regards to typological
classification of superstrate and substrate languages, different degrees of contact with the
superstrate or other prestigious languages, and differences in the sociolinguistic status of
the languages. The remainder of this section includes a brief description of the language
contact situation that gave birth to the creoles and of the synchronic sociolinguistic
situation of each creole. Bibliographic details of the data sources have also been included.
2.1 Guyanese CE
Guyanese CE, or Creolese as the Guyanese themselves call it, is widely spoken in the
former colony of British Guyana, in which it co-exists with endangered Amerindian
languages (Holbrook and Holbrook 2001). Ethnologue (Gordon 2005) reports a figure of
650,000 speakers for Guyanese CE.
English remains the national language of Guyana. All teaching is in English27 and
Creolese is not used, for the most part, in television or radio broadcasting, although a
26
I have followed Holm (1988) in the denomination of these four creoles. With the exception of Kituba, the
name of each creole is followed by a reference to the lexifier language. The acronyms used are CE: Creole
English, CF: Creole French, and CS: Creole Spanish.
27
Data from the Barometer of Human and Trade Union Rights in Education, elaborated by Education
International.
~ 50 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
large percentage of the population does not have adequate competency in Standard
English (Holbrook and Holbrook 2001). From all the creoles analysed, Guyanese is the
one with a lesser prestige.
Guyana only became a British colony at the end of the 18th century and, in quite a short
period of time, a large influx of slaves was brought to the colony from West Africa and
from other Caribbean colonies, such as Barbados, Antigua and Jamaica. The creole was
probably imported by these slaves from elsewhere in the region (Holm 1989).
Data sources: Most of the data is taken from Bickerton (1975) although a more recent
revision of the expression of habituality and imperfectivity by Sidnell (2002) has also
been reviewed. Bickerton provides an account of the three layers of variation in the creole
(basilect, mesolect and acrolect), but only data from the basilect has been taken into
account.
2.2 Haitian CF
Kreyòl, the official name of Haitian CF, is spoken by more than seven million people in
the former Caribbean French colony of Haiti. Most Haitians are monolingual in Kreyòl,
with only about one-fifth of the population proficient in French. French is used almost
exclusively as the language of writing in all domains, from education to politics, although
the creole is taking on new functions in school (Holm 1989, DeGraff 2007).
Lefebvre (1998) calculates that the period of creation of Haitian CF lies between 1680
and 1740, a period along which Haiti became a classic plantation colony and
consequently the number of slaves increased dramatically, reaching 92% of the
population in 1791. The slaves spoke languages from the Niger-Congo group, particularly
from the Kwa (Gbe speakers are estimated to have formed half of the slave population)
and Bantu language families.
Haitian CF coexisted with African languages for about a hundred years, while the
ongoing stream of newly arrived slaves continued (Lefebvre 1998). Bickerton (1981)
considers Haitian CF as a characteristic radical creole, while McWhorter (1998) has
equally described it as the Caribbean French Creole least affected by French.
Data sources: Spears (1990) deals with the TA system of Haitian CF, while Lefebvre
(1998) constitutes a comprehensive analysis of this creole in the light of the substrate
theory of creole genesis. No reference is made by any of these scholars regarding the
level of formality of the language variety portrayed, although both studies are based on
data collected in the field. The more recent contribution of DeGraff (2007) to a collected
volume on creole syntax has also been consulted.
2.3 Papiamentu CS
Papiamentu is spoken in Curaçao, Aruba and Bonaire, three islands belonging to the
Netherlands Antilles. The Antillean islands enjoy a high degree of multilingualism in
~ 51 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Dutch, Spanish, English and Papiamentu. This creole has been classified as a semi-creole
by many creolists due to its extensive contact with European languages (McWhorter
1998).
Perl (1999) estimates that 90% of the communication in the three islands is in
Papiamentu. The creole was also learned and used by the white settlers, which explains its
high prestige nowadays. It is spoken by all social classes and used in the media and
education.28
The genesis of Papiamentu remains a mystery. It was not born in a plantation, as Curaçao
was mainly used as a holding camp for slaves who were later sent to other Caribbean
destinations (McWhorter 2000). Holm (1989) and McWhorter (2000) consider
Papiamentu a Spanish-based creole, while Maurer (1998) describes it as a creole of a
mixed lexical base (both Spanish and Portuguese). Holm’s analysis and terminology have
been followed in this paper.
Data sources: Andersen (1990) analyses the TA system of the basilectal variety of
Papiamentu. The more recent contribution of Kouwenberg & Ramos-Michel (2007) to a
collected volume on creole syntax has also been consulted.
2.4 Kituba
Kituba is a widely spoken language in the area at the mouth of the Congo River. Kituba is
mostly spoken in the southern provinces of the Republic of Congo and is one of the three
national languages of that country, along with Lingala and French. It is also spoken in the
north-west provinces of the Democratic Republic of Congo (formerly Zaire), although it
is not an official language in that country (Ethnologue, Gordon 2005).
Ethnologue (Gordon 2005) reports that Kituba has more than four million speakers and
800,000 second language users. It was used by French and Belgian colonisers, which led
to its identification with the language of the administration (Swift & Zola 1963:x). As a
national language in the Republic of Congo, it is used in radio and television, but not in
education (Woods 1994). Most of its speakers are therefore multilingual (Woods 1994).
Holm (1989) reports that Kituba has the prestige of a big-city language associated with
modern life and Woods (1994) reports that it is increasingly being used in more social
domains, unlike French.
Also known as Kikongo-Kituba, Kituba seems to have arisen out of a need for
intercommunication between speakers of Lingala and speakers of Kikongo, which is
considered its lexifier (Swift & Zola 1963:x). Lingala and Kikongo are typologically
similar languages, agglutinating Bantu languages (Mufwene 1990:97).
28
Data from the Barometer of Human and Trade Union Rights in Education, elaborated by Education
International.
~ 52 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Data sources: Mufwene (1990) has described the TA system of the eastern dialect of
Kituba, which is not in contact with Kikongo. The manual of Swift & Zola (1963)
represents the western dialect and it probably portrays the most standard variety of the
language, as its aim is to serve as a self-taught manual to learn the language by foreigners.
Three of these languages (Guyanese CE, Haitian CF and Papiamentu CS) comply with
the sociohistorical profile defined by McWhorter (1998:791), that is, they constitute
‘spoken languages that were created via rapid adoption as a lingua franca by slave
populations five hundred years ago or less’ and have also traditionally been classified as
Atlantic creoles (Holm & Patrick 2007:v). The superstrate of these three languages
constitute three different European languages.
The sociohistorical circumstances that led to the birth of Kituba are very different from
those of the above Atlantic creoles, as it arose as a regional lingua franca (Swift & Zola
1963:x). Kituba has remained on the edge of creolistics and is considered by some
scholars a koiné (Mufwene 1997).
3 DISCUSSION
3.1 Modern Reichenbachian approaches to tense and aspect
Following modern Reichenbachian approaches to tense and aspect as described by
Binnick (1991:115f.), we can define tense and aspect in terms of three independent points
on a timeline:
x S, the time of speech or writing;
x E, the time when the event or state takes place; and
x R, the more abstract reference point from which E is seen, the temporal ‘point of
view’.
As illustrated by Figure 1, tense represents the relationship between speech time (S) and
reference point (R) while aspect encodes the relationship between event (E) and reference
point (R) (Binnick 1991:115).
Tense
S
R
Aspect
E
Figure 4.1: Neo-Reichenbachian approach to tense and aspect (Binnick 1991:115)
In many languages, the expression of aspect is not independent of tense. To represent
adequately these TA systems, it has been proposed to use two pairwise orderings of E and
R (aspect) and R and S (tense) (Binnick 1991:115).
~ 53 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Simple or absolute tenses represent the relationship between the times of events or states
(E) and the time of speech (S), they are therefore ‘regarded purely from the point of view
of the present (the moment of speech or writing)’ (Binnick 1991:40). To represent simple
tenses, R and E always coincide and both either precede S (simple past), coincide with S
(simple present) or follow S (simple future) (Binnick 1991:111). The sentence
(5) I had pizza for lunch
could therefore be represented as R,E:R-S. The reference point (R) coincides with the
event (E, ‘eating the pizza’) and both precede the speech time (S, now).
Aspect is defined by Chung & Timberlake (1985:213) as ‘the relationship of a predicate
to the time interval over which it occurs’, that is, aspect describes how an event (E)
develops in relation to a reference point (R), independently of the speech time (S). For
example, in the sentence
(6) John was reading all night
the event (E, ‘reading’) is ongoing at R (‘all night’) and that reference point precedes the
speech time. Example (6) is represented as R-E:R-S.
Aside from simple tenses and aspectual distinctions, it is also possible to distinguish
relative tenses. Dahl (1985:25) and Comrie (1976:5) define a relative tense as a tense
always regarded from the point of view of another tense, in itself established in relation to
speech time. Relative tenses are restricted to subordinate clauses and non-finite clauses.
See Dahl (1985:25) and Comrie (1976:5) for further reference.
3.2 Perfective/imperfective
The aspectual distinction between perfective and imperfective plays an important role in
many verb systems. In English, such distinction is not marked explicitly (Binnick
1991:296), while in languages such as Rendille, they are (Dahl & Velupillai 2008b). See
the following examples:
(7)
(Example 1 in Dahl & Velupillai 2008b)
a. khadaabbe chiirta.
letter.pl
write. IPFV
‘He writes/is writing/wrote/was writing/will write letters.’
b. khadaabbe chiirte.
letter.pl
write. PFV
‘He wrote letters.’
Traditionally, the imperfective is described as looking at the action as a development over
time, as ongoing (Binnick 1991:212). In Reichenbachian terms, the imperfective
~ 54 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
represents the situation in which an event partly overlaps with the reference time but its
termination follows the reference time (R-E).
In the following representation of aspect, the arrows represent possible reference times,
while event time is illustrated by the brackets:
Perfective aspect
------------------------------(
Event time
)-------------------------------
Imperfective
Figure 4.2: The representation of aspect (adapted from Binnick 1991: 210)
The imperfective views the event as incomplete or in-progress while the perfective
presents the event as an unanalysable whole. In example (7a) above, the event of writing
is ongoing at the reference time, whether the reference time is the time of speaking, the
past or the future.
The perfective aspect denotes that an event is contained in its totality within the reference
time (E,R) (Binnick 1991:210). According to Cruse (2004:288), the perfective is
described as viewing an event holistically, without paying attention to the event itself;
that is to whether it is a punctual or a durative event. Example (7b) presents the event as
completed at the reference time.
Since the perfective describes events as complete or bounded, it is natural that such
events are past events (Dahl & Vellupillai 2008a). Bybee et al. (1994:51) acknowledge
that since pasts and perfectives both develop from the same sources and in much the same
way, they have very similar semantic content. What is conveyed in one language by the
perfective aspect, is conveyed in another by the past tense (Cruse 2004). Example (7b)
above is translated into English as the past tense sentence He wrote letters.
As hinted above, a way of differentiating the perfective aspect from the past tense is how
they are integrated in a linguistic system. The perfective is the contrast partner of the
imperfective (see example 7 above), while the past can co-occur with the imperfective
(Bybee et al. 1994:84). The sentence He was writing letters is an example of a past
imperfective in English.
~ 55 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
3.3 Bickerton’s anterior marker is a perfective marker
Perfective
marker
Guyanese CE
Haitian CF
Papiamentu CS
Kituba
bin
te
a
- á(k)a (suffix)29
Table 4.2: Summary of markers discussed in this section
There is enough evidence from the four creoles analysed to support the view that these
creoles mark the perfective/imperfective distinction. In the four creoles analysed, the
perfective marker, which in Kituba is a suffix, incorporates a perfect and a simple past
reading depending on the context. The function of such marker in the four creoles is to
denote completion, whether in relation to S or to R, as it is shown by the following
examples.
(8)
Example from Haitian CF (example 4 in DeGraff 2007:103)
Bouki te konn repons lan
B PFV know answer DEF 30
‘Bouki knew the answer.’
(9)
Example from Haitian CF (example 5 in DeGraff 2007:103)
Bouki te ale (anvan Boukinèt vini)
B PFV go before B
come
‘Bouki had left (before Boukinèt came).’
Comparison of examples (8) and (9) reveals that the same marker te indicates a past event
in relation with the time of speaking (example 8) and a past event in relation with another
reference point (example 9). Similar readings of the marker termed Anterior tense by
Bickerton are found in the other three creoles, as becomes clear from the following
examples. Compare examples (10) and (11) for the two readings of the marker a in
Papiamentu and the two instances of the suffix -áka in example (12):
(10)
Example from Papiamentu CS (example 24 in Kouwenberg & Ramos-Michel
2007:312)
Mi a lubida ariba dje
1s PFV forget on 3s
‘I forgot about it.’
29
Segments into brackets are written but not pronounced.
I have used the abbreviations IPFV (imperfective), PFV (perfective) and IPFV_PAST (past imperfective)
in the interlinear gloss for the Tense and Aspect markers. In all other respects, the gloss provided by the
authors have been faithfully reproduced.
30
~ 56 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
(11)
Example from Papiamentu CS (example 25 in Kouwenberg & Ramos-Michel
2007:312)
Mi a
kome bonchi kaba
1s PFV eat bean already
‘I’ve already eaten beans.’
(12)
Example from Kituba (example 2d in Mufwene 1990:99)
Na ki+ntéte
ngé
tub+áka nde María kwend+á(k)a na
LOC 9+Monday you-sg say+PFV COMP Mary go+PFV
LOC
ki+sálu mazóno.
9+work yesterday31
‘On Monday you said that Mary had gone to work the day before [Sunday].’
Example 1 above, reproduced below for easier reference, includes two instances of
Guyanese CE bin. This marker can similarly have a past reading in relation with the time
of speaking (Dem bin gatu get we) and a past-before-past reading (bikaz terabl ting bin
hapn):
(1)
Example from Guyanese CE (example 2.51 in Bickerton 1975:36)
Dem bin gatu get we an kom dis said, lef di plees an get we, bikaz terabl ting bin
hapn wid dem chiren.
‘They had to get away and come over here, leave the place and get away, because
terrible things had happened to their children.’
It seems that the simple past reading is preferred with statives, but this might be due to the
fact that the perfect is more frequent with nonstatives (Bybee et al 1994:69).
The only difference among the creoles is that in two of the creoles analysed the perfective
marker is not always obligatory. In Kituba and Papiamentu CS, the markers are
obligatory and the verb stem is unacceptable in most environments, aside from a group of
statives. On the contrary, in Haitian CF and Guyanese CE, the perfective marker appears
in variation with the unmarked verb stem. Especially with activity-type predicates, as in
examples (13) from Guyanese CE and (14) from Haitian CF, the interpretation of the
unmarked verb stem could entail a certain degree of difficulty, as it could have both a
completed event reading or a present habitual meaning. Cf. example (13) with example
(1) and example (14) with example (8).
(13)
31
Example from Guyanese CE (example 19 in Sidnell 2002:163)
I se shi dadii Ø taak nais wid am.
‘He said that her father talks/talked nicely to him.’
The digit before the gloss refers to the noun class.
~ 57 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
(14)
Example from Haitian CF (example 57 in Lefebvre 1998:134)
Mari prepare pat.
Mary prepare dough
‘Mary (generally) prepares/ prepared dough.’
In Haitian CF and Guyanese CE, the perfective marker is only obligatory to convey a
perfect reading, such as in example (9) above.
3.4 The creole imperfective marker
Guyanese CE Haitian CF
Papiamentu CS
Kituba
Tenseless imperfective
marker
a
-
ta
-
Present imperfective marker
-
ap
-
ke
Past imperfective marker
-
ta
-
vandá(k)a
Perfect progressive/past
imperfective
bina
-
tabata
-
Table 4.3: Summary of markers discussed in this section:
As stated above, the perfective is the contrasting partner of the imperfective. While the
perfective marker denotes completion, the four creoles analysed display a range of
markers denoting an ongoing event at the reference point.
In Guyanese CE and Papiamentu CS there is an imperfective marker covering both the
progressive and the habitual meanings. Cf. examples (15) and (16) below for Guyanese
CE:
~ 58 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
(15)
(16)
Example from Guyanese CE (example 2.59 in Bickerton 1975:37)
Mi tel am wa mi a du
‘I told him what I was doing.’
Example from Guyanese CE (example 2.47 in Bickerton, 1975:34)
Evribadi bin gatu wach aut an evribadi a de aal abaut a rood, striit,
dam.
‘Everybody had to be on the watch and everyone used to be all over the place, on
roads, streets, dams.’
In example (15), the progressive expressed by a entails that the event (du) fills the
reference time (the time when Mi tel am) but it may terminate after the reference time,
that is, the activity expressed by a du ‘doing’ may carry on after the action of Mi tel am ‘I
told him’. In neo-Reichenbachian terms, it is represented as R-E:R-S. In example (16),
the marker a qualifies a stative verb. a de ‘used to be’ makes reference to a past habitual
state. Once again, the event (a de) is ongoing at the reference time, which precedes the
speech time, that is R-E:R-S32. Guyanese CE a groups the meanings of progressive and
habitual and it is tenseless. In the examples above, the tense reference of the clause
containing the marker a is established by the previous clause.
Papiamentu CS marker ta has a present reading in example (17) and a past reading in
example (18), both sharing the feature of on-going action at the reference time. As
Guyanese CE a, the Papiamentu marker is also tenseless. It is important to note that in
Papiamentu ta incorporates the readings of progressive, habitual and even future. The
Guyanese imperfective mark appears in variation with the unmarked verb stem to express
habituality (see example (13) above).
(17)
32
Example from Papiamentu CS (example 17 in Kouwenberg & Ramos-Michel
2007:310)
Wan ta kanta / awor-akí / tur dia / otro luna
W IPFV sing/ now-Dem/ every day/ next month
‘Wan sings [Generic]/ is singing right now [Prog] / sings every day [Hab] /will
sing next month [Fut]’.
According to Binnick (1991), the habitual implies that the event happens at some time during the
reference time, but with gaps while the progressive implies that the action is ongoing without gaps. See
the following examples (from Binnick 1991: 459):
(a) Mr Blandings was building his dream house
(b) Mr Blandings builds his dream house (on Wednesdays)
In (b), the event of building is punctuated by the Thursdays to Tuesdays gaps.
~ 59 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
(18)
Example from Papiamentu CS (example 4 in Andersen 1990:70)
Antó el a weta un hòmber yongotá ei bou, ta
saka
then he PFV see a man
kneeling there below, IPFV remove
awa ku un makutu basha den un bari sin
bom.
water with a bucket empty in a barrel without bottom
‘Then he saw a man kneeling down below, taking out water with a bucket (and)
pouring (it) in a bottomless barrel.’
In addition, Guyanese CE and Papiamentu CS have developed a perfect progressive
marker, which in both languages is spreading to mark past state. Guyanese CE displays a
perfect progressive marker, bina, probably derived from the combination of bin + a, that
is the perfective marker bin and the imperfective marker a. Such combinations are found
in other creoles and are further discussed below. The following example (19) contrasts
with example (15) above, reproduced below for easier reference.
(19)
Example from Guyanese CE (example 2.60 in Bickerton 1975:37)
Mi tel am wa mi bina
du
‘I told him what I had been doing.’
(15)
Example from Guyanese CE (example 2.59 in Bickerton 1975:37)
Mi tel am wa mi a du
‘I told him what I was doing.
Guyanese CE bina appears in variation with the tenseless imperfective marker a to
express past habitual. Compare example (20) below with example (16) above, reproduced
here for easier reference. The perfect progressive marker has extended its meaning in
Guyanese CE to include a past habitual meaning, that is, an imperfective meaning
restricted to past reference.
(20)
(16)
Example from Guyanese CE (Example 2.57 in Bickerton 1975: 37)
wan blakman an i waif bina liv abak
‘A negro and his wife used to live inland. ‘[and they live there no longer]
Example from Guyanese CE (example 2.47 in Bickerton, 1975:34)
Evribadi bin gatu wach aut an evribadi a de aal abaut a rood, striit,
dam.
‘Everybody had to be on the watch and everyone used to be all over the place, on
roads, streets, dams.’
Similarly, in Papiamentu, we can also find a past imperfective marker, which is
compulsory for the expression of perfect progressive (as in example (21) below). Tabata
is also the default option to mark statives with past reference (as in example (22) below):
~ 60 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
(21)
Example from Papiamentu CS (example 3 in Andersen 1990:69)
El a
haña un hòmber sintá den port’e kamber.
he PFV find a man seated in door_of room
e hòmber ku ta’ata
yena awa ku makutu
the man
who IPFV_PST fill water with bucket
‘He found a man sitting in the doorway. The man who had been filling a bucket
with water.’
(22)
Example from Papiamentu CS (example 15 in Kouwenberg & Ramos-Michel
2007:310)
Mi ta
/ tabata
malu
1s IPFV / IPFV_Past sick
‘I am/was sick.’
In Kituba and Haitian CF, on the other hand, there is an imperfective marker restricted in
its tense reference to the present progressive and they have also developed a past
progressive marker. In example (23) below, the Haitian CF marker ap has a progressive
reading, but it is restricted to present reference.
(23)
Example from Haitian CF (example 4c in Spears 1990:121)
M ap
pale ak Mari.
1sg IPFV talk with Marie
‘I’m talking to Marie.’
To mark past progressive, the perfective marker is combined with the progressive marker
(t a or t ap from te + ap), as in example (24), in a similar way to Guyanese CE bina. This
could lead us to classify the Haitian CF perfective marker as a past marker instead, since
it co-occurs with the progressive marker. However, the fact that te also has a perfect
reading makes it difficult to classify it as a past marker (cf. example (9)). The
combination of perfective and imperfective marker is further discussed below.
(24) Example from Haitian CF (example 24 in Spears 1990:134)
M t
a mande kòman li
t
ap ... kòman l apr
1sg IPFV_PAST ask how 3sg IPFV_PAST how 3sg IPFV
ale la
a.
go there DET
‘I was asking him... how he was going to move.’
Kituba also displays a present imperfective marker ke, which combines reference to
present state for statives and progressive for nonstatives, including future reference, as in
example (25). However, Kituba also has another marker, vand+á(k)a (spelled wandaka
by Swift & Zola 1963), with exclusive reference to past progressive, as in example (26).
~ 61 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
(25)
Example from Kituba (example 7a in Mufwene 1990:104)
Yándi ké
kwísa mbási.
He IPFV come tomorrow
‘He comes/is coming tomorrow.’
(26)
Example from Kituba (example in Swift & Zola 1963:193)
Yandi wandaka
fimpa
nzutu na munu yonso.
He/she IPFV_Past examine body LOC me all
‘He was examining my whole body.’
To sum up, the four creoles analysed share a perfective marker with the same range of
meanings. However, the spectrum of the imperfective is divided differently by each
creole. More importantly, there are aspectual markers restricted in their tense reference,
such as the Kituba past progressive or the Haitian CF present progressive. Table 4.4
below exemplifies the range of meanings of the imperfective markers in the four creoles.
Guyanese CE
Haitian CF
Imperfective Tenseless
Present
marker
progressive
+ Progressive
tenseless habitual + future
future
Past
Perfect progressive.
imperfective Past habitual
marker
Papiamentu CS
Tenseless
+ progressive
present state
Past progressive Perfect
progressive
past state
Kituba
Present
+ progressive
present state
Past progressive
+
Table 4.4: Imperfective markers
Creole imperfective markers may incorporate the meanings of progressive and habitual
(Papiamentu CS and Guyanese CE), it may be restricted to progressive meaning (Haitian
CF) or it may be compulsory for present reference, aside from a restricted group of
statives (Papiamentu CS and Kituba).
4 CONCLUSIONS
4.1 Are creoles tenseless languages?
To conclude, we can generally define the creole TA system as aspect-prominent, since it
incorporates the imperfective/perfective distinctions. However, the four creoles analysed
are not tenseless since tense restrictions are commonly found in the spectrum of the
imperfective. Binnick’s claim that creoles are tenseless languages is not substantiated by
~ 62 ~
+
Esther Núñez Villanueva: Are creoles tenseless languanges?
the data. The prediction, found in Bybee et al. (1998:83), that tense distinctions are only
relevant in the imperfective is born out in the data of these creoles.
Taking this into account, the Creole TA system shows important similarities to the
tripartite system represented in Table 4.5.
Perfective
Imperfective
Present imperfective
Past imperfective
Table 4.5: Representation of the major TA markers in creole
However, the above TA system is not unique to creoles. It is found in many IndoEuropean languages and Semitic language (Dahl 1985:189). Nonetheless, if supported by
further analyses of other creoles, the TA system could be one piece of evidence to defend
the theory that creoles constitute a typological class, the main differences among creoles
lying in the obligatory nature of the markers and the range of meanings of the
imperfective marker.
4.2 Differences among the creoles analysed
In Haitian CF and Guyanese CE the verb stem is acceptable in affirmative, simple
sentences but its interpretation relies heavily on context. According to Sankoff’s view
(1990), the use of the bare verb stem can constitute an historical residue of the pidgin
stage, in which the creole markers had not yet evolved. The evidence here suggests that in
Guyanese CE and Haitian CF the progressive and the perfect are the only environments in
which the TA markers are fully grammaticised, while in other environments the TA
markers are in variation with the unmarked verb stem. In Haitian CF and Guyanese CE,
the unmarked verb stem of nonstatives can include all of the following meanings: simple
past, present perfect, present habitual or past habitual, whilst the unmarked verb stem of
statives can have both a present habitual and a present state reading. As a consequence,
the imperfective/perfective markers are not obligatory to express such meanings.
It would be reasonable to consider then that Kituba and Papiamentu CS’s markers
correspond to a further stage in the grammaticalisation of the TA markers since in both
creoles the markers are obligatory and the verb stem is unacceptable in most
environments, aside from a group of statives. The different level of grammaticalisation of
their markers would explain some of the differences found in the TA system of the above
creoles.
The suggestion that these creoles’ TA markers present a different stage of
grammaticalisation begs the question of whether the TA markers are following equivalent
paths of grammaticalisation in each language. We can tentatively point to some data that
may shed some light on the issue, although this is by no means a thorough claim, but
more an indication of future research.
~ 63 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Sidnell (2002:53) reports that in Guyanese CE the imperfective marker emerged through
the grammaticalisation of a locative expression consisting of the locative copula de
(derived from English ‘there’) and the preposition a (derived from English at) into a
tenseless progressive (marker a). The tenseless progressive marker could have later been
adopted to express tenseless habitual, as in example (16).
Sidnell’s description for Guyanese CE constitutes a well attested path of
grammaticalisation and Bybee et al. (1994:129) claim that the majority of progressive
forms in their database derive from expressions involving locative elements, such as a
verbal auxiliary or an adposition. Haitian CF, Papiamentu CS and Kituba could have
followed a similar pattern, since the three of them present locative auxiliaries as markers
of imperfective aspect. In Kituba and Papiamentu CS, the imperfective marker could have
further developed into a present marker.
The perfect has also been identified as the source of pasts and perfectives (Bybee et al.
(1994:56). The perfect is, along with the progressive, the other environment in which the
TA markers are fully grammaticised in Guyanese CE and Haitian CF. This could be
evidence that the perfect marker is in the process of incorporating a perfective reading in
these two creoles and such a process could have taken place also in Kituba and
Papiamentu CS.
The original meanings of progressive and perfect is clear from the combination of both
markers, which rendered bina in Guyanese CE (bin + a), compulsory to express perfect
progressive (Cf. example (19). Such combination must have taken place before the
progressive became an imperfective (combining the meanings of progressive and
habitual) and before the perfect became a perfective (combining the meanings of perfect
and past). The perfect progressive seems to be spreading to mark past habitual in
Guyanese CE (see example (20).
The perfect progressive could havebeen the source of the past imperfective marker in
Haitian CF. In this creole, the marker t a (or t ap) is derived from the perfective marker te
and the imperfective marker ap. It is not clear whether t a incorporates a perfect
progressive reading, since there are no relevant examples in the data sources consulted.
The only other evidence for such process is in the Papiamentu CS tabata, which
incorporates the meanings of perfect progressive and past state (example (22)).
This scenario seems to imply that eventually the unmarked verb stems in creoles will be
used in less contexts and would explain the differences among the range of meanings of
the perfective and imperfective markers in the four creoles analysed. However, more data
is needed regarding the changes happening in creoles over time to substantiate such a
claim. The truth is that it leaves many questions unanswered, such as why the progressive
and the perfect markers are the first to be grammaticised and why creoles follow the same
path of grammaticalisation. Similarly, it may not be causal that the two radical creoles
analysed, Guyanese CE and Haitian CF, differ from the other two creoles in the level of
grammaticalisation of the TA markers.
~ 64 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
REFERENCES
Andersen, Roger W. 1990. Papiamentu Tense-Aspect, with special attention to discourse.
In John Victor Singler (ed.), 59-96.
Bickerton, Derek. 1975. Dynamics of a creole system. Cambridge: Cambridge University
Press.
Bickerton, Derek. 1981. Roots of language. Ann Arbor, MI: Karoma.
Binnick, Robert I. 1991. Time and the verb: A guide to tense and aspect. Oxford: Oxford
University Press.
Bybee, Joan, Revere Perkins & William Pagliuca. 1994. The evolution of grammar:
Tense, aspect, and modality in the languages of the world. Chicago & London: The
University of Chicago Press.
Chung, Sandra & Alan Timberlake. 1985. Tense, aspect and mood. In Timothy Shopen
(ed.), vol. 3, 202-258.
Comrie, Bernard. 1976. Aspect (Cambridge Textbooks in Linguistics). Cambridge:
Cambridge University Press.
Cruse, Alan 2004: Meaning in Language: An Introduction to Semantics and Pragmatics.
(Second edition.) Oxford: Oxford University Press.
Dahl, Östen. 1985. Tense and aspect systems. Oxford: Blackwell.
Dahl, Östen & Viveka Velupillai. 2008a. Tense and aspect. In Matthew S. Dryer et al.
(eds.),
introduction
to
chapter
65-68.
Available
online
at
http://wals.info/supplement/intro65-68. Accessed on 06/08/08.
Dahl, Östen & Viveka Vellupillai. 2008b. Perfective/imperfective aspect. In Matthew S.
Dryer,
et
al.
(eds.),
chapter
65.
Available
online
at
http://wals.info/feature/description/65. Accessed on 06/08/08.
DeGraff, Michael. 2007. Kreyòl Ayisyen, or Haitian Creole (Creole French). In John
Holm & Peter L. Patrick (eds.), 101-126.
Dryer, Matthew S., Martin Haspelmath, David Gil & Bernard Comrie (eds.). 2008. The
world atlas of language structures online. Munich: Max Planck Digital Library.
Education International. 2008. Barometer of Human & Trade Union Rights in Education.
Available online at http://www.ei-ie.org/barometer/en/index.php. Accessed on
22/08/08.
Gordon, Raymond G., Jr. (ed.). 2005. Ethnologue: Languages of the world.15th edn.
Dallas,
Tex.:
SIL
International
Online.
Available
online
at
http://www.ethnologue.com. Accessed on 11/02/2010.
Holbrook, David J. & Holly A. Holbrook. 2001. Guyanese Creole survey report.
Available online at http://www.ethnologue.com/show_work.asp?id=41939.
Accessed on 22/08/08.
Holm, John. 1988. Pidgins and creoles 1: Theory and structure. Cambridge: Cambridge
University Press.
Holm, John. 1989. Pidgins and creoles 2: Reference survey. Cambridge: Cambridge
University Press.
Holm, John & Peter L. Patrick (eds.). 2007. Comparative creole syntax: Parallel outlines
of 18 creole grammars (Westminster Creolistics Series). London: Battlebridge
Kouwenberg, Silvia & Abigail Ramos-Michel. 2007. Papiamentu (Creole
Spanish/Portuguese). In John Holm & Peter L. Patrick (eds.), 307-332.
~ 65 ~
Esther Núñez Villanueva: Are creoles tenseless languanges?
Lefebvre, Claire. 1998. Creole genesis and the acquisition of grammar: The case of
Haitian Creole (Cambridge Studies in Linguistics). Cambridge: Cambridge
University Press.
Maurer, P. 1998. El papiamentu de Curazao. In Matthias Perl & Armin Schwegler (eds.),
139-217.
McWhorter, John H. 2000. The missing Spanish creoles: recovering the birth of
plantation contact languages. California: University of California Press.
McWhorter, John H. 1998. Identifying the Creole Prototype: Vindicating a typological
class. Language 74 (4). 788-818.
Mufwene, Salikoko S. 1990. Time Reference in Kikongo-Kituba. In John Victor Singler
(ed.), 97-118.
Mufwene, Salikoko S. 1996. The founder principle in creole genesis. Diachronica 13. 83134.
Mufwene, Salikoko S. 1997. Jargons, pidgins, creoles and koines: what are they? In
Arthur K. Spears & Donald Winford (eds.), 35-70.
Muysken, Pieter. 1981. Creole tense/mood/aspect systems: The unmarked case. In Pieter
Muysken (ed.), Generative studies on creole languages, 181-199. Dordrecht: Foris.
Muysken, Pieter C. (ed.). 1981. Generative studies on creole languages. Dordrecht: Foris.
Perl, Matthias.1999. Problemas actuales de la estandarización del papiamentu. In
Zimmermann (ed.), 251 - 262.
Perl, Matthias & Armin Schwegler (eds.). 1998. América negra: panorámica actual de
los estudios lingüísticos sobre variedades hispanas, portuguesas y criollas.
Frankfurt am Main: Vervuert.
Sankoff, Gillian. 1990. The grammaticalization of tense and aspect in Tok Pisin and
Sranan. Language Variation and Change 2 (3). 295-312.
Shopen, Timothy (ed.). 1985. Language typology and syntactic description, vol. 3:
Grammatical categories and the lexicon. Cambridge: Cambridge University Press.
Sidnell, Jack. 2002. Habitual and imperfective in Guyanese Creole. Journal of Pidgin and
Creole Languages 17 (2). 151-189.
Singler, John Victor (ed.). 1990. Pidgin and creole tense-mood-aspect systems (Creole
Language Library 6). Amsterdam: Benjamins.
Spears, Arthur K. 1990. Tense, mood and aspect in the Haitian Creole preverbal marker
system. In John Victor Singler (ed.), 119 - 142.
Spears, Arthur K. & Donald Winford (eds.). 1997. The structure and status of pidgins and
creoles (Creole Language Library 19). Amsterdam: Benjamins.
Swift, Lloyd B. & Emile W. A. Zola. 1963. Kituba basic course (Foreign Service Institute
Basic Course Series). Washington, D.C.: Foreign Service Institute.
Winford, Donald. 1996. Common ground and creole TMA. Journal of Pidgin and Creole
Languages 11 (1). 71-84.
Woods, David R. 1994. Changing patterns of language utilization in the Republic of
Congo. African Languages and Cultures 7 (1).19- 35.
Zimmermann, Klaus (ed.). 1999. Lenguas criollas de base lexical española y portuguesa.
Frankfurt am Main/Madrid: Vervuert/Iberoamericana.
~ 66 ~
CHAPTER 5
HOW SIMILAR IS A BELFAST FINAL RISE TO A
CAMBRIDGE FINAL FALL?
JENNIFER SULLIVAN
University of Edinburgh
The phenomenon of statements with final (nuclear) intonational rises in Belfast and other Northern
British/Irish regions has continually puzzled researchers. Are these ‘rises’ really rises at all
(Cruttenden 1997)? I propose to address this with the following hypothesis: nuclear statement ‘rises’
in Belfast are actually more similar to nuclear statement ‘falls’ in Cambridge English than to
Cambridge question ‘rises’. This is based on the possibility that the Belfast ‘rises’ may have derived
historically from ‘falls’. However, results show that the timing of the Belfast ‘rise’ is actually much
more similar to the timing of the Cambridge question ‘rise’. The relative height of the L and H
points of the Belfast ‘rise’ is different to both the Cambridge ‘falls’ and the Cambridge ‘rises’.
Thus, my original hypothesis is not strongly supported. However, the approach to quantification of
similarity in intonation which we outline is something which has until now received very little
attention.
1 INTRODUCTION
It is well-known that speakers often use intonation as one tool to distinguish questions
from statements. Statements in Standard British English varieties (e.g. Cambridge
English) are generally expected to end with a final pitch fall, whereas questions are
typically associated with final rising pitch (Grabe et al 2000). However, in Belfast
English, statements often end with a final rise. Are Belfast statements more similar to
Standard English questions simply because both contain final rises? Could a rise
sometimes be more similar to fall? If so, which elements would they share? Trying to
start answering these challenging questions is the goal of this paper, but it is also where
we reach a big gap in previous research. Recent years have seen a noticeable increase in
objective measurements of phonetic similarity, particularly among vowels and consonants
(e.g. McMahon et al 2007). However, intonation has usually been left to the side in these
measurements (exceptions include Connolly (1997) and Gussenhoven & Rietveld
(1991)). In addition, the components of a measure of segmental similarity would seem
quite inappropriate for measuring intonation, especially as segmental measures tend to
make heavy use of phonetic feature systems, and such features are not a standard part of
the most prominent current intonational theories. I define intonation here along the lines
of Ladd (2008), as the use primarily of fundamental frequency (f0) (roughly
representative of pitch), but also aspects of intensity and duration to express meaning at a
sentence/utterance level. Distinguishing a question from a statement is one example of a
sentence/utterance level meaning contrast. This is different from lexical tone/accent, in
which pitch is used to distinguish meanings at the word level (e.g. ‘anden’ means ‘the
duck’ with Accent I in Swedish, but ‘the ghost’ with Accent II (Bruce 1977: 15).
~ 67 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
2 HYPOTHESIS
My initial hypothesis is that Belfast statement final (nuclear) rises would actually be more
similar to nuclear statement falls than to nuclear question rises. The primary reason I
propose this hypothesis is a plausible scenario about the historical development of these
Belfast statement rises. This is that they may not always have been realised as rises, but
rather were originally final falling contours which changed over time into rises. The
suggestion that rises of the kind associated with Belfast might be kinds of falls was
explored in Cruttenden (1997, with reference to Knowles’ work on Liverpool English).
Of course, similarity between falls and rises is not necessarily linked with a shared
background but it is a useful place to start.
2.1 Framework
A change from a fall to a rise can be accounted for very elegantly by AutosegmentalMetrical (AM) theory (see Bruce (1977), Ladd (2008), and Pierrehumbert (1980)),
elements of which I use in my analysis. AM theory does not treat rises and falls as units,
but rather decomposes the intonation contour down simply into high (H) and low (L)
target points around prominent stressed syllables (known as pitch-accented syllables) and
phrase edges in the utterance. This theory analyses the phonetic realisation of these H
and L points along two key parameters, which form the main components of my
intonational similarity measurements to date. These parameters are Alignment and
Scaling. Alignment refers to the precise timing of the H and L points with respect to their
associated syllable. Where exactly in the syllable does the f0 peak (phonetic realisation of
H) occur, for example? Is it timed shortly after the vowel onset or at the end of the
vowel? Scaling refers to the relative height of the H and L points, with respect to the
speaker’s pitch range at that part of the utterance.
2.2 Alignment and Intonational Change
It is the Alignment parameter that can easily account for a change from a ‘fall’ into a
‘rise’. Imagine that the f0 peak (H) is located near the onset of the vowel in the most
prominent stressed syllable in the utterance, which is usually also the final stressed
syllable (known as the nuclear syllable). The pitch must first rise up to reach the peak.
After this peak, the pitch falls off over any following unstressed syllables. So we hear a
final fall. However, if the peak moves gradually rightwards, eventually it will occur
beyond the stressed syllable, such that low pitch may now occur on the stressed syllable
and rise up to the peak in the following unstressed syllable. The final fall part of the
contour may then be truncated (see also Grabe et al 2000), as there may no longer be
enough room to produce or perceive a final fall. So both L and H points of the rise before
the main fall in the original contour now form what is perceived as a final rise. Though I
have not analysed historical data at present (cf. Kim 2006), I illustrate this possible
change from a fall to a rise with contemporary data from Cambridge and Belfast English
below (data from Grabe et al 2001).33
33
Figures (1)-(3) were produced using Praat (Boersma & Weenink 2009) with a script adapted from the
Praat scripts of Pauline Welby (http://www.ling.ohio-state.edu/~welby/praat.html).
~ 68 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
Fundamental frequency (Hz)
400
350
300
250
200
150
100
0
0.5
Time (s)
ɪ
s
n
d
ə
ɹ
ɛ
l
ɐ
Figure 5.1: Extract of an utterance from a female speaker of Cambridge English of the nuclear accented
word ‘Cinderella’. Notice the rise to reach the peak in stressed vowel, followed by the fall.
Fundamental frequency (Hz)
400
350
300
250
200
150
100
0
0.5
Time (s)
s
ɪ
n
d
ə
ɹ
ɛ
l
ɘ
Figure 5.2: Similar to the above, but this time spoken by a speaker of Belfast English. Notice the peak is
further to the right, now timed with the segment /l/ at the boundary between stressed and unstressed syllable
in ‘Cinderella’
An alignment difference like this in relation to the H has been at the heart of an account
of the lexical accent distinction in Stockholm Swedish (Bruce 1977). A difference in the
alignment of H can account for the presence of a rise on the stressed syllable in Accent I
and the fall in Accent II. A similar alignment difference was also invoked in relation to
intonational differences in the varieties of Orkney and Shetland English (van Leyden &
van Heuven 2006). Crucially, an alignment difference was also briefly explored in trying
to account for dialect differences in the Irish language (Dalton & Ní Chasaide 2005,
Dalton 2007). A very similar phenomenon to Belfast English occurs in the Northern
dialect of the Irish language where we find nuclear statement rises, in contrast to nuclear
falls in the Southern dialects. Dalton & Ní Chasaide’s comparison of the Northern
statement rises against Southern statement falls from one dialect led to their argument that
~ 69 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
they were not phonetically similar enough to each other to support the hypothesis that
there had been a ‘Re-alignment’ of the H, turning the fall into the rise. However, I
wished to examine how this hypothesis would hold up in Belfast English data, and I also
wished to expand greatly on the theoretical treatment begun by Dalton & Ní Chasaide.
Fundamental frequency (Hz)
400
350
300
250
200
150
100
0
0.5
Time (s)
s
ɪ
n
d
ə
ɹ
ɛ
l
ɘ
Figure 5.3: An extract from a different Belfast English speaker, showing the peak now in the final
unstressed vowel, with hardly any final fall afterwards
3 DATA
To examine intonational similarity and potential Alignment change, I focused on the
phenomenon of nuclear statement rises in Belfast English and compared them to nuclear
statement (rise)-falls34 in Cambridge English, and to nuclear question rises in both Belfast
and Cambridge English. I used data from the Intonational Variation in English (IViE)
corpus (Grabe et al 2001). This corpus consists of recordings of teenage speakers of 9
varieties of British and Irish English in five different speaking styles. I refer in this paper
to two of these styles: the Read Sentences, in which participants read a list of declarative
and interrogative (including coordination, declarative questions, y/n and wh- questions)
sentences; and the Read Passage, in which participants read a version of the fairytale
Cinderella. For consistency, I only examined nuclear stressed syllables in which there
was just one following unstressed syllable. I measured the alignment and scaling of the L
and H points on the nuclear syllable and surrounding syllables.
34
I henceforth refer to the statement contours in Cambridge English as (rise)-falls as an important reminder
of the rise up to reach the peak H in these contours.
~ 70 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
3.1 Methodological Details
The numbers of utterances (Intonational phrases) examined in each category are listed
below:
Sentences List
Belfast statement rises
74 Intonational Phrases
(IPs) (6 male, 6 female
speakers)
Cambridge statement (rise)- 0
falls
Belfast question rises
92 IPs (6 male, 6 female)
Cambridge question rises
48 IPs (6 male, 6 female)
Cinderella Passage
65 IPs (6 female, 2 male)
67 IPs (6 female, 5 male)
0
0
Table 5.1: Numbers of sentences analysed
Though statements in the Sentences list task were recorded by Cambridge speakers, it was
inappropriate to analyse most of them for the present paper, as the majority contain
‘downstep’ i.e. “the stepwise lowering of pitch (or of pitch range) at specific pitch
accents” (Ladd 2008: 76). The phonetic differences between ‘downstepped’ and ‘nondownstepped’ accented syllables would have added confounds to the Alignment and
Scaling measurements I intended to make. The Cinderella Passage had very few
questions, which is why the question rises come from the Sentences list only. Further, the
Cambridge speakers produced some of their questions with different intonational patterns
than plain rises, which explains why the number of IPs analysed here is notably less than
for the Belfast questions. I emphasise that this is still very much work in progress, one
reason why the number of IPs analysed is unequal between categories, between males and
females, and within each speaker.
F0 contours were smoothed to reduce the effect of consonantal effects on f0. The main
stressed (nuclear) syllable as well as the preceding and following unstressed syllables
(prenuclear and postnuclear respectively) syllables were all labelled, as were the segments
within them, mainly according to the segmentation criteria laid out in Turk et al (2006).
When these critiera could not be applied, we chose midpoints of formant transitions in
marking segment boundaries between sonorant segments. However, I acknowledge that
several cases of segmentation were very difficult and accept that this has implications for
the Alignment results below. This is because Alignment is measured in milliseconds (ms)
from the beginning or end of specific segments e.g. the onset of nuclear stressed vowel.
All Alignment and Scaling measurements were calculated and extracted using Praat
scripts (modified versions of scripts by Pauline Welby http://www.ling.ohiostate.edu/~welby/praat.html).
Before giving details of the results, I recall the initial hypothesis: that Belfast statement
rises would be more similar to the Cambridge statement (rise)-falls than to question rises,
following from the possibility that the Belfast statement rises had developed from nuclear
falls (see section 2 above).
~ 71 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
4 RESULTS
I now present the results of the Alignment and Scaling measurements for each of the
categories. The statistical analysis carried out so far involved One-Way ANOVAs on the
Sentence/Variety types on the four dependent variables of Alignment of H, Alignment of
L, Scaling of H, and Scaling of H. Where appropriate these were followed by post-hoc ttests with the Bonferroni correction for multiple comparisons applied. Non-parametric
versions of these tests gave very similar results. All statistical procedures were done with
R (R Core Design Team 2009).
4.1 Alignment of H
First, I deal with the Alignment of H. H was deemed to be the f0 maximum, except
where there was a clear ‘elbow’ in the contour marking the end of the main rise, even if
the pitch continued to rise very slightly after this or level out. In the Cambridge statement
(rise)-falls, the H was aligned within the nuclear stressed syllable, specifically shortly
after the onset of the nuclear vowel (mean 30 ms after onset, standard deviation (s.d.) 70).
In the Belfast statement rises, the H was aligned much later, this time in the following
unstressed syllable, well beyond the onset of the following postnuclear unstressed vowel
(Sentences List: mean 110 ms after onset, s.d. 60; Cinderella Passage: mean 101 ms,
s.d.53). In the Cambridge question rises, the H was in fact aligned very similarly to the
Belfast statement rises (mean 94 ms after postnuclear vowel onset, s.d. 118). In the
Belfast question rises, the H was also aligned in a very similar location (mean 100 ms,
s.d. 50).
There was no significant difference in the Alignment of H between Belfast statements,
Cambridge questions, and Belfast questions: F (3, 275) = 0.5831, n.s. I did not test the
significance of the Alignment of the H in the Cambridge statements against the
Alignment of H in the other categories. This was because I had measured it against a
different segmental landmark and in any case, it was obviously so much further away
from the Alignment of H in the Belfast statements and in the question rises. I expected the
alignment of H to be later in the Belfast statements than in the Cambridge statements,
directly following from the scenario of potential Alignment change between the Belfast
and Cambridge statements. However, I had not expected the H in the Belfast statements
to be aligned as closely to the H in the question rises as it turned out to be. I had thought
the H would be earlier in the Belfast statements. This was due to previous descriptions
of Belfast statement rises as ‘rise-plateaux’ or ‘rise-plateaux-slumps’ (Cruttenden 1997,
Grabe 2002, Ladd 2008), where the main rise is followed by a levelling off or slight fall,
in contrast to question rises which are expected to keep rising. When there is just one
unstressed syllable after the accented syllable, this proposed difference does not actually
appear to be systematic, though further work is needed.
4.2 Alignment of L
The alignment of the L point beginning the rise was primarily measured with a special
line-fitting script to mark the point of greatest acceleration of the rise (for further details,
~ 72 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
see the accompanying notes on Welby’s ‘Elbow’ scripts at http://www.ling.ohiostate.edu/~welby/praat.html and references therein). We did not focus on the true f0
minimum as often in level stretches of low pitch this corresponds merely to a random f0
value. However, if errors were made with the line-fitting script and if the true f0
minimum was in an appropriate location, then that was taken as representative of L. In the
Cambridge statement (rise)-falls, the L refers to the beginning of the rise leading up to the
H (NOT the final low point at the end of the fall). It was aligned in the unstressed
prenuclear syllable prior to the nuclear stressed syllable (mean 38 ms after onset of
prenuclear vowel, s.d. 72). In the Belfast statement rises, the L was again aligned much
later, around the end of the nuclear vowel (which often corresponded to end of the nuclear
syllable) (Sentences List: mean 10 ms after the vowel offset, s.d. 48; Cinderella Passage:
mean 11ms, s.d. 71). The L in the Cambridge question rises (mean 6ms before the vowel
offset, s.d. 87) and in the Belfast question rises (mean 19ms after the vowel offset, s.d.
40) was timed very similarly.
There was no significant difference between the alignment of L in these three categories:
F (3, 252) 35 = 1.586, n.s. We did not test the significance of the alignment of the L in the
Cambridge statement (rise)-falls against the other categories for the same reason in
relation to alignment of H above. Again, although we had expected the L in the Belfast
statements to be aligned later than in the Cambridge statements, we had not expected it to
be as similarly timed as the question rises. So in terms of the Alignment parameter, the
Belfast statements rises are very like question rises in relation to both L and H. This goes
against our hypothesis that the Belfast statement rises would be more similar to the
Cambridge statement rises and therefore makes it hard to link the Belfast statement rise
back easily to an original falling contour. The similarity in Alignment of the Belfast
statements and Belfast questions is also different to Makarova’s (2007) work on Russian,
where an alignment difference between questions and statements was found.
4.3 Scaling
Now turning to the Scaling parameter, which refers to the height of the L and H points.
First I measured these f0 points in the standard Hertz (Hz) scale. Then I decided to
convert them to the ERB scale, which is argued to be a better approximation of perception
and for the inclusion of the different ranges of male and female voices together (Arvaniti
et al 2006, Glasberg & Moore 1990, Ladd 2008). In each IP, Praat scripts extracted the
Hz and ERB value of the L and H points and also of the speaker’s mean pitch value in
that utterance. Obviously, different speakers have different pitch ranges so I normalised
the L and H by dividing each by the speaker’s mean ERB value (see Ladd 2008 for this
and other methods of normalising the scaling). This meant that the L and H values from
different speakers could be more appropriately compared with each other.
35
The reason the degrees of freedom are not the same in the Alignment of L as in the Alignment of H is as
follows: in some IPs, there were errors in the extracted measurements for the Alignment of L but not for the
Alignment of H and vice versa. A similar situation occurred in relation to the Scaling measurements, which
is why the degrees of freedom again will be different there.
~ 73 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
4.4 Scaling of H
In terms of the Scaling of the H, it was highest in the Cambridge questions (mean 1.19,
s.d. 0.1 normalised ERB), followed by the Cambridge statements (mean 1.14, s.d. 0.12),
the Belfast questions (mean 1.1, s.d. 0.08) and the Belfast statements (Sentences List
mean 1.07, s.d. 0.06; Cinderella Passage: mean 1.03, s.d. 0.07). Among these categories,
there was a significant difference in the Scaling of H: F (4, 321) = 28.87, p<0.001. Posthoc t tests with the Bonferroni correction revealed that the Cambridge questions were
significantly higher in H than the Belfast questions (t 5.9052, p<0.001) and Belfast
statements (Sentences List: t 7.5845, p<0.001; Passage t -9.8262, p<0.001), but not
significantly higher than the Cambridge statements (t -2.5602, n.s.). The Cambridge
statements were significantly higher in H than the Belfast statements (Sentences List: t
3.7038, p<0.01; Passage: t 5.9251, p<0.001) but not significantly higher than the Belfast
questions (t 2.2744, n.s.). The Belfast questions were only significantly higher in H than
Belfast statements from the Passage (t -5.6124, p<0.001), not from the Belfast statements
from the Sentences List (t 2.2368, n.s.). Finally, among the two sets of Belfast
statements, the Sentences List statements were significantly higher in scaling of H than
the Cinderella Passage statements (t -3.73, p<0.01).
One of the main points that we glean from these results is that again the expected
phonetic similarity of the Belfast statement to the Cambridge statement must be called
into question, contrary to our original hypothesis. The Cambridge statements have H
higher with respect to the speaker’s mean pitch than the Belfast statements. Though the
Belfast statements are different to the Cambridge questions in this way, they are more
similar to Belfast questions than to either of the Cambridge sentence types. The trend
here for higher scaling of H in the questions within each variety reflects a pattern
observed in many languages (e.g. Yuan et al 2002).
4.5 Scaling of L
In the Scaling of the L, the highest L was in the Cambridge statements (mean 0.99
normalised ERB, s.d. 0.11), followed by the Cambridge questions (mean 0.96, s.d. 0.07),
the Belfast questions (mean 0.92, s.d. 0.07) and the Belfast statements (Sentences List:
mean 0.92, s.d. 0.06; Passage: mean 0.91, s.d. 0.07). Among these categories, there was a
significant difference in the Scaling of L: F (4, 314)=12.19, p <0.001. Post-hoc tests
revealed that there were significant differences between the varieties of Cambridge and
Belfast, but not within these varieties. The Scaling of L was significantly higher in the
Cambridge statements than in Belfast statements (Sentences List: t 4.5594, p <0.001;
Passage t 4.8184, p <0.001) and questions (t 4.2929, p <0.001), but not higher than the
Cambridge questions (t 1.3429, n.s.). Likewise, the Cambridge questions were also
significantly higher in the Scaling of L than the Belfast statements (Sentences List: t
3.5971, p<0.01; Passage: t -3.9175, p <0.01) and questions (t 3.381, p<0.01). As
indicated though, there was no significant difference in the Scaling of L between the
Belfast statements and questions (Sentences List: t -0.5972, n.s.; Passage: t 0.0513, n.s.).
Nor was there any difference between the two groups of Belfast statements (t -0.6761,
n.s.).
~ 74 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
These results reflect my own impressionistic observations of the data, such that the pitch
at the beginning of the Belfast statements and questions dips down noticeably on the
accented syllable, whereas in the Cambridge statements and questions, the pitch remains
quite high in the speaker’s range before rising further. So again we see an important
difference between the Belfast statements and the Cambridge statements, which goes
against the initial hypothesis. The L in the Belfast statements is lower with respect to the
speaker’s mean pitch than the L in the Cambridge statements. Further, the Belfast
statements are extremely similar to the Belfast questions in the Scaling of L, though they
are different to the Cambridge questions in this regard.
4.6 Assessment of the Results
Figure 5.4: Rough stylised diagram indicating the Alignment and Scaling of the statements and questions
from Belfast and Cambridge. The two black rings are intended to show how similar the alignment of the L
and H are in the Belfast statements, Belfast questions and Cambridge questions.
In the overall assessment of similarity between Belfast statements and Cambridge
statements, none of the parameters of Alignment and Scaling have pointed to a
particularly close connection between them. This is contrast to the almost entirely
overlapping measurements between the Belfast statements and Belfast questions. So this
analysis of contemporary corpus data, rather than supporting a view of the Belfast
statements rises as having formed from statement (rise)-falls, actually shows the Belfast
statement rises as very similar to question rises in many respects. I have also begun to
incorporate the individual Alignment and Scaling measurements into an overall
composite score of similarity. Very tentative early results from this show Belfast
~ 75 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
statements closest to Belfast questions, then to Cambridge questions and farthest from
Cambridge statements.
5 THEORETICAL PROPOSALS
We conclude from this analysis of the IViE data that the concept of similarity and
indications of potential intonational change need to be taken in a new direction.
Therefore, our next main step was to take the Alignment and Scaling parameters and
develop more theoretical proposals as to how they behave with regard to similarity and
potential change from a (rise)-fall to a rise. There are three main parts to my proposals at
their present stage of development.
5.1 H Drift
The first part is to do with the direction of change. Many studies of languages with a
lexical tonal contrast report that H tones have a tendency to spread rightwards (e.g.
Hyman 2007, Kim 2005). There is a parallel in intonational studies which have found
that H target points have a tendency to drift rightwards alongside an increase in the
number of unstressed syllables after the main stress (e.g. Dalton & Ní Chasaide 2005,
Gussenhoven 2007, Silverman & Pierrehumbert 1990). However, I admit that it is unclear
how this context could lead to change over time in intonation. Anyway, if the H peak
drifts too much away from the stressed syllable, it will no longer be perceived as
associated to that syllable. Therefore, some phonological reorganisation would be needed,
as the nuclear stressed syllable being a tone bearing unit (TBU), requires a tone. At this
point, the low pitch from the rise to the peak would become phonologised as L tone on
the stressed syllable. This putative phonological change and perceptual motivations
behind it may be somewhat akin to metathesis in segmental phonology (e.g. Blevins &
Garrett 1998). There is some evidence to suggest that L tones need to be realised as flat
stretches or broad dips down on the stressed syllable in order to be perceived (Dilley
2005). So the resulting contour would have quite a different shape to the original
statement (rise)-fall. This would link with lower scaling of the L in the Belfast
statements, though admittedly this also occurred in the Belfast questions.
5.2 Favoured Alignment points
The second part is my argument that there are favoured places within the syllable with
which L and H target points like to align themselves. Linking with the first part of my
proposals above, H target points are often aligned slightly beyond the main stressed
syllable in a number of languages (e.g. Arvaniti et al 2000 on Greek). It may be easier
both to produce and to perceive high pitch on the stressed syllable if H is located here
(e.g. Rossi 1971, Silverman 1997, Hyman 2007). There are also arguments that for H
targets that the accented vowel may be broken up into two or three domains for
categorical phonological distinctions or pragmatic differences (Ladd 2008 and references
therein). There is increasing recent evidence to suggest that H targets at any rate may be
coordinated very closely with articulatory gestures (e.g. Mücke et al 2009). In relation to
~ 76 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
L, a low turning point beginning the rise located at the end of the nuclear stressed syllable
(similar to the Belfast statements and also the question contours) may also be a favoured
alignment location from an articulatory point of view. It is possible that the L point is
coordinated with the peak velocity of the maximum closure of the consonant onset of the
following unstressed syllable (Mücke, p.c.). The H point in Catalan nuclear rises and
Italian statements is also coordinated with this point (Mücke et al 2009, Prieto et al 2007).
Overall, I extrapolate from this that if the alignment of an L or H point undergoes change
that it may not just move gradually but may undergo more abrupt shifts between favoured
locations (cf. Stevens (1989) on quantal relationships between articulation and acoustics
in segmental phonology). The main implication of this for similarity is simple linear
Alignment measurements in milliseconds may not capture perceptual similarity or the
nature of change itself. By building up an inventory of these favoured locations, we
would be able to express in terms of number of steps how far away the Cambridge
statement H is from the Belfast statement H etc.
5.3 Tonal Crowding Effects
The third part posits a further rightward movement of the H. If the L is located around the
edge of the nuclear stressed syllable and the H is just a bit beyond it in the following
unstressed syllable, we may have the phenomenon of tonal crowding. This phenomenon
is well-known in both languages with lexical tone and those with intonation only. When
two tones are too close to each other, one or both of them may move apart. In this
instance, I suggest that the H tone may move further to the right, away from the L tone.
This pattern of the presence of the L tone resulting in later alignment of the following H
tone has been reported by Arvaniti et al (2006) for Greek intonation, and by Kristoffersen
(2007) and Peters (2007) for the lexical accent in Norwegian and Hasselt Flemish
respectively. This could account for the major alignment differences between the
Cambridge statements and the Belfast statements. We have tested this prediction by
seeing if the later that L was aligned in the Belfast statements (Passage data), the later the
H was also aligned. However, there was no significant correlation between the alignment
of L and the alignment of H (Pearson product-moment correlation: t 1.1858, d.f. 59, n.s.).
6 COULD BELFAST RISES HAVE COME FROM QUESTION RISES?
What is clear at present is that whichever way we assess alignment (by counting steps or
by taking absolute measurements), the (majority of the) Belfast statement rises remain
much more similar to the question rises from either Belfast itself or Cambridge than to the
Cambridge statement (rise)-falls. We had originally expected that the Belfast statement
rises might have the L and H alignment more intermediate between statement (rise)-falls
and question rises. However, such cases only occurred in a small portion of the data and
mainly from a single speaker (extremely interesting though such cases in themselves are).
Without historical data, therefore, I believe that it is not possible to claim that my results
nor my theoretical proposals support an idea of an alignment change from a statement
(rise-)fall to a statement rise. This is because it would be so easy to ask why the Belfast
~ 77 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
statements could not be modifications of questions since their alignment appears so
similar.
So if the Belfast statements are phonetic modifications of question rises (see Bolinger
1978 mentioned in Ladd 2008)36, what kind of changes would have to be made to turn a
question rise into a statement rise? The main change here would be in Scaling. The
Cambridge questions in particular were significantly higher in their H scaling than the
Belfast statements, though my results show a tendency for Belfast questions also to have
higher scaling than Belfast statements. Higher scaling of H is a well-known
distinguishing feature of questions from statements (e.g. Yuan et al 2002). Alignment
change through rightward movement of H is well-attested in languages with lexical tone
and is increasingly invoked in relation to intonation too (Dalton 2007 Gussenhoven 2007,
Hyman 2007, Silverman 1997). However, are there any reports that Scaling changes can
happen in languages with lexical tone or in intonation? Arvaniti & Ladd (2009) argue
that alignment of H in Greek appears to be more variable than the scaling of H in a
specific synchronic context (tonal crowding). So this might be a sign that height changes
are less likely than alignment changes. However, in Chinese languages reductions of the
height of lexical tones in terms of change over time have been attested (Chen 2000). So
although much more work needs to be done on establishing how likely alignment and
scaling are to change over time, we must acknowledge for the present that scaling
changes are possible. This at first makes it even harder to uphold the hypothesis that this
data shows that Belfast statement rises have ultimately come from statement (rise)-falls
than from question rises (again without having historical data available). To turn a
Cambridge question into a Belfast statement, we would have to lower the scaling of the L
and the H, but could leave the alignment pretty much unchanged. To turn the Cambridge
statement into a Belfast statement, we would have to lower the scaling of the L and the H,
and also make extensive alignment changes.
7 CONCLUSIONS AND PROPOSALS FOR THE FUTURE
The implications of the work this paper has uncovered go contrary to the initial
hypothesis, that Belfast statements would be more similar phonetically to Cambridge
statements reflecting a possible shared origin. In fact, these Belfast statements are very
similar to question rises, in relation to the Alignment of L and H. In scaling, they are also
very similar overall to Belfast questions, but have lower scaling of L and H than the
Cambridge data (both questions and statements). Of course, we should clarify that
similarity and change are not always linked. There are well-attested examples from
segmental phonology where two very divergent pronunciations can be shown by
principled methods to have a shared background and others where two very similar forms
have different backgrounds. So the big challenges that remain for this work include trying
to establish what natural sound changes in intonation actually are. We now have ample
data on patterns of synchronic intonational variation in different languages/varieties (e.g.
36
Belfast statements do not function like questions though.
~ 78 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
Grabe 2002) and through a wide range of prosodic factors (e.g. Silverman &
Pierrehumbert 1990). How do these patterns link with diachronic change in intonation?
For future work, we also wish to consider other possibilities about the origin of the
Belfast statement rise. Instead of being derived from a statement (rise)-fall or from a
question rise, we wish to examine whether it could be a form of continuation rise, but
with its function extended. We also wish to compare Belfast English with Glasgow
English, and see if we would fit the potential differences between them into a potential
trajectory of intonational change. This is further work. Overall, we assess that these
phonetic parameters of Alignment and Scaling provide a very useful way to understand
the similarities between different sentence types in different varieties. They also provide
a useful starting point for exploring hypotheses about intonational change. However, in
the context of Belfast statement rises and Cambridge statement (rise)-falls, it does not
support this hypothesis on its own. We now need further parameters (e.g. including
amplitude as well as f0 measurements), historical data, and/or much stronger theoretical
arguments for showing that the changes from a (rise)-fall to a rise would indeed be more
plausible than the lesser changes from questions to statement rises. Alternatively, these
parameters of similarity should motivate us to accept that we do not have clear phonetic
connections between Cambridge statements and Belfast statements, and we now need to
look elsewhere for an explanation of why Belfast statements rise at the end. Either way,
the phenomenon of Belfast statement rises does not have an uncontroversial explanation
yet.
REFERENCES
Arvaniti, Amalia & Ladd, D. Robert. 2009. Greek wh-questions and the phonology of
intonation. Phonology 26. 43-74.
Arvaniti, Amalai, Ladd, D. Robert, & Mennen, Ineke. 2000. What is a starred tone?
Evidence from Greek. In Broe, Michael & Pierrehumbert, Janet B. (eds.) Papers in
Laboratory Phonology V. Cambridge: CUP. 119-31
Arvaniti, Amalia, Ladd, D. Robert, & Mennen, Ineke. 2006. Phonetic effects of focus and
‘tonal crowding’ in intonation: Evidence from Greek polar questions. Speech
Communication 48. 667-96.
Blevins, Juliette AND Garret, Andrew. 1998. The origins of consonant-vowel metathesis.
Language 74.3. 508-56.
Boersma, Paul AND Weenink, David. 2009. Praat: Doing phonetics by computer.
(Version 5.0.47). http://www.praat.org (Jan 2010).
Bolinger, Dwight. 1978. Intonation across languages. In Greenberg, Joseph (ed.)
Universals of Human Language, Vol. II: Phonology. 471-524. Palo Alto: Stanford
University Press.
Bruce, Gösta. 1977. Swedish word accents in sentence perspective. Lund: Gleerup.
Chen, Matthew. 2000. Tone Sandhi. Cambridge: CUP.
Connolly, John H. 1997. Quantifying target-realization differences. Clinical Linguistics
and Phonetics 11. 267-98.
~ 79 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
Cruttenden, Alan. 1997. Intonation, 2nd Edn. Cambridge: CUP.
Dalton, Martha. 2007. The phonetics and phonology of the intonation of Irish dialects.
Dublin: Trinity College PhD Thesis.
Dalton, Martha & Ni Chasaide, Ailbhe. 2005. Tonal alignment in Irish dialects.
Language and Speech 48.4. 441-64.
Dilley, Laura Christine. 2005. The phonetics and phonology of tonal systems. Cambridge,
MA: MIT PhD thesis.
Grabe, Esther. 2002. Variation adds to prosodic typology. Proceedings of the Speech
Prosody Conference. 127-32. Aix-en-Provence, April 2002.
Grabe, Esther, Brechtje Post & Francis Nolan. 2001. The IViE Corpus. Department of
Linguistics, University of Cambridge.
Grabe, Esther, Brechtje Post, Nolan, Francis & Farrar, Kimberley. 2000. Pitch accent
realization in four varieties of British English. Journal of Phonetics 28. 161-85.
Gussenhoven, Carlos. 2007. The phonology of intonation. In de Lacy, Paul (ed.) The
Cambridge Handbook of Phonology. Cambridge: CUP. 253-80
Gussenhoven, Carlos & A. C. M. Rietveld. 1991. An experimental evaluation of two
nuclear-tone taxonomies. Linguistics 29. 423-49.
Hyman, Larry M. Universals of tone rules: 30 years later. In Riad, Tomas and
Gussenhoven, Carlos (eds.) Tones and tunes. Vol. 1: Typological studies in word
and sentence prosody. 1-34. Berlin: Mouton de Gruyter.
Kim, Yuni. 2005. Some observations on the intonational phonology of Kuki Thaadow, a
Tibeto-Burman tone language. Ms., UC Berkeley.
Kim, Yuni. 2006. Variation and Finnish influence in Finland-Swedish dialect intonation.
Proceedings from Fonetik 2006. Working Papers from the Department of
Linguistics and Phonetics 52. Lund University.
Krifsotffersen, Gjert. 2007. Dialect variation in East Norwegian tone. In Riad, Tomas and
Gussenhoven, Carlos (eds.) Tones and tunes. Vol. 1: Typological studies in word
and sentence prosody. 99-111. Berlin: Mouton de Gruyter.
Ladd, D. Robert. 2008. Intonational Phonology, 2nd Edn. Cambridge: CUP.
Ladd, D. Robert, Mennen, Ineke and Schepman, Astrid. 2000. Phonological conditioning
of peak alignment in rising pitch accents in Dutch. Journal of the Acoustical Society
of America 107.5. 2685-96.
McMahon, April, Heggarty, Paul, McManon, Robert and Maguire, Warren. 2007. The
sound patterns of Englishes: Representing phonetic similarity. English Language
and Linguistics 11.1. 113-42.
Mücke, Doris, Grice, Martine, Becker, Johannes and Hermes, Anne. 2009. Sources of
variation in tonal alignment: Evidence from acoustic and kinematic data. Journal of
Phonetics 37.3. 321-338.
Peters, Jörg. 2007. Bitonal lexical pitch accents in the Limburgian dialect of Borgloon.
Tones and tunes. Vol. 1: Typological studies in word and sentence prosody, ed. by
Tomas Riad and Carlos Gussenhoven, 167-98. Berlin: Mouton de Gruyter.
Pierrehumbert, Janet B. 1980. The phonetics and phonology of English intonation.
Cambridge, MA: MIT PhD thesis.
Prieto, Pilar, Mücke, Doris, Becker, Johannes, and Grice, Martine. 2007. Coordination
patterns between pitch movements and oral gestures in Catalan. Proceedings of the
16th International Congress of Phonetic Sciences (ICPhS). 989-92. Saarbrücken.
~ 80 ~
Jennifer Sullivan: How similar is a Belfast final rise to a Cambridge final fall?
R Development Core Team. 2009. R: A Language and Environment for Statistical
Computing. R Foundation for Statistical Computing, Vienna, Austria. {ISBN} 3900051-07-0. http://www.R-project.org (Jan 2010).
Rossi, M. 1971. Le seuil de glissando ou seuil de perception des variations tonales pour
les sons de la parole. Phonetica 23. 1-33.
Silverman, Daniel. 1997. Tone sandhi in Comaltepec Chinantec. Language 73. 473-92.
Silverman, Kim and Pierrehumbert, Janet. 1990. The timing of prenuclear high accents in
English. Papers in Laboratory Phonology I: Between the grammar and physics of
speech, ed. by John Kingston and Mary Beckman, 72-106. Cambridge: CUP.
Stevens, Kenneth N. 1989. On the quantal nature of speech. Journal of Phonetics 17. 345.
Turk, Alice, Nakai, Satsuki and Sugahara, Mariko. 2006. Acoustic segment durations in
prosodic research: A practical guide. In Sudhoff, Stefan, Denisa Lenertová, Roland
Meyer, Sandra Pappert, Petra Augurzky, Ina Mleinek, Nicole Richter and Johannes
Schlie er Methods in empirical prosody research. 1-28. Berlin/New York: Mouton
de Gruyter.
Van Leyden, Klaske and Van Heuven, Vincent. 2006. On the prosody of Orkney and
Shetland dialects. Phonetica 63. 149-74.
Yuan, Jianhong, Shih, Chilin and Kochanski Greg P. 2002. Comparison of declarative
and interrogative intonation in Chinese. Proceedings of the Speech Prosody
Conference. 711-4. Aix-en-Provence, April 2002.
~ 81 ~
CHAPTER 6
AGAINST THE “WEST GERMANIC SYNTAX” HYPOTHESIS
THE V-2 CONSTRAINT IN OLD ENGLISH AND OLD HIGH GERMAN
ANNA CICHOSZ
University of Łódź
Even though both English and German belong to the same West-Germanic group of Indo-European
languages, their present word order seems to have very little in common. While practically all
English clauses follow the SVO order, Modern German differentiates between V-final (in most
subordinate clauses) and V-2 pattern (characteristic of main declarative clauses). This paper will
focus on the so-called V-2 constraint. In the Old Germanic period, both languages tended to place
their finite verb on the second position of the main declarative clause, and this – among other
similarities – leads some scholars to assume that the two syntactic systems were practically identical
(cf. Davis and Bernhardt 2002). In the present paper I would like to claim that the differentiation
started to take place before 1066, in the Old Germanic period, and thus it is necessary to regard the
syntax of Old English and Old High German as two independent systems.
1 INTRODUCTION: THE V-2 CONSTRAINT
The V-2 constraint is a phenomenon well-known to all scholars of Modern German. This
language is characterised by a very specific word-order rule which says that in main
declarative clauses the finite verb always takes the second position. Therefore, all
grammatically correct German clauses represent the following pattern:
(1) Ich bin nach London gefahren.
[I am to London gone.]
I went to London.
(2) Gestern bin ich nach London gefahren.
[Yesterday am I to London gone.]
Yesterday I went to London.
~ 82 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
As illustrated in example (2), when the clause is introduced by an element other than the
subject, the subject and the finite verb get inverted so that the finite verb could keep its
position in the clause. This is the phenomenon known as the V-2 constraint, V-2 rule,
verb-seconding or Zweitstellung, and it is present in other Germanic languages as well
(e.g. Dutch). The translations which accompany each example show, however, that
English deviates from this pattern and consistently demonstrates the Subject – Verb order,
no matter which constituent stands at the beginning. Nonetheless, there are some traces of
the V-2 constraint in Modern English, though they are not so easy to trace:
(3) Never have I seen such a wonderful place.
Example 3 illustrates a fossilised English structure with Subject-Verb inversion caused by
the negative adverb “never”, an existing though rather unproductive pattern, which is in
fact a remnant of the V-2 constraint that used to be present in the English system in the
Old Germanic period, that is to say before the 11th century, when English was still quite
similar to its Germanic cousins.
Yet, the fact that the V-2 structure is characteristic of Germanic languages does not mean
that it is a rule that they have all inherited and consistently used ever since, while English
was the only one that lost it almost completely. In the Old Germanic period the V-2
pattern was not yet a rule but rather a very strong tendency, and clauses illustrating it are
very widespread in both Old English and Old High German corpus:
(4) Eft clipode se engel Abraham (OE, Genesis)37
[Again called the angel Abraham]
(5) Hier begin ih einna reda umbe diu tier (OHG, Physiologus)
[Here start I my tale about the animal]
The examples show two clauses with Subject-Verb inversion caused by the initial
adverbial, which clearly indicates that some sort of the V-2 constraint operated in both
languages. However, the corpus also includes numerous counterexamples:
(6) be þam man mihte oncnawan (OE, The Battle of Maldon)
[by that one could know]
(7) So ir selbo quhad dhurah zachariam (OHG, Isidor)
[so he himself said through Zacharias]
Here the situation is exactly the same, both clauses are introduced by an adverbial, but the
subject and the finite verb are not inverted. This leads to two preliminary conclusions: the
V-2 pattern is not an unbreakable rule in Old English and Old High German and; the
37
All the Old English and Old High German examples quoted in the present paper come from the ENHIG
database which is available on the Internet at http://ia.uni.lodz.pl/cichosz/enhig. The abbreviations OE and
OHG, accompanying each example, stand for Old English and Old High German respectively.
~ 83 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
situation in both languages seems similar, which has to be checked in a representative
corpus of texts.
2 THE “WEST/OLD GERMANIC SYNTAX” HYPOTHESIS
In the only existing comparative study of Old English and Old High German syntax,
Davis and Bernhardt (2002:1) claim that:
It is meaningful to speak of a common syntax of West Germanic, which must have existed
in Proto-West-Germanic and which is evidenced in the major languages in the group.
Thus, as a consequence,
The study of the syntax of one Old West Germanic language is an indicator of the syntax of
others; in particular the syntax of the better-recorded Old English acts as a guide to the
syntax of Old High German (Davis and Bernhardt 2002:1).
Davis, in his latest book on comparative syntax of Old English and Old Icelandic, has
modified this theory to include languages from both the North and East Germanic branch,
stating that:
The patterns of word order exhibited by all the Old Germanic languages may be regarded
as identical. This is tantamount to saying that the syntax of these languages is the same in
all its important points (Davis 2006: 53).
The present paper aims to question this hypothesis, showing that the two languages did
demonstrate significant word order differences that need to be taken into consideration.
3 STRUCTURE OF THE STUDY CORPUS AND METHODOLOGY
Davis and Bernhardt (2002) based their hypothesis on the analysis of two long prose
works: homilies by Ælfric (late 10th century) for Old English and the so-called Tatian
Gospel Translation (early 9th century) for Old High German. The problem of sources is a
very important aspect of the analysis; since there are no ideally comparable texts written
in the two languages and a large portion of available texts comprises poetry and
translations, it is relatively difficult to create a good study corpus. As many scholars
working in this field admit, it is especially problematic in the case of Old High German:
Therefore, we have to conclude that none of the Old High German sources of a
considerable size may be viewed as a good example of original prose representative for the
system of the dialects spoken at the period of time (Petrova and Solf in press: 2)
To investigate the details of Old High German word order presents some problems. The
Old High German documents are numerous, but it is a problem to find corpora which are
unquestionably representative of the language in terms of word order. On the one hand, the
prose material is composed almost completely of translations of earlier Latin religious
~ 84 ~
Anna Cicchosz: Againstt the “West-G
Germanic synta
ax” hypothesis
is
writings and
a follows the order off the originaal with only minor exceeptions. On the
t other
hand, som
me of the earlliest poetic w
works are preeserved only in fragmentts and some represent
r
artistic exxperiments which
w
may haave brought into play im
mportant variaations from expected
word ordeer. Of coursee, a study succh as this mu
ust work witth what is attttested. I hav
ve chosen
to take a very
v
wide saampling of ssentences fro
om seven diffferent sourcees in the hop
pe that if
significannt idiosyncrasies do appeear in any on
ne corpus, th
hey may be brought to light and
identified by the data from
f
the othher corpora an
nd evaluated
d accordinglyy (Smith 1971: 43).
t strategy
y has been tto include samples
s
from all mainn text types (poetry,
In tthis study, the
origginal prose and translaated prose) and thus, by numero
ous comparrisons, to eliminate
diffferences that are duee to stylisstic constraaints and isolate reaal similaritties and
disccrepancies between
b
thee languagess (which is in accordan
nce with thhe attitude taken
t
by
Smiith, though his corpuss is unfortuunately very
y small). The
T structurre of the corpus is
pressented below
w:
d
aree:
Thee texts incluuded in the database
x for Old English: Beowulf (exccerpts), Caedmon’s Hym
mn, The Baattle of Mald
don, The
anderer, Wiidsith, Ælfrric’s homily
y Alia Visiio, Laws off Alfred
Seafareer, The Wa
(excerppts), The Anglo-Saxonn Chroniclee (excerpts)), Wulfstann’s Sermo Lupi
L
ad
Anglos,, Genesis trranslation (eexcerpts), West
W Saxon Gospels (exxcerpts);
x for Old
O
High German:: Hildebrrandslied, Ludwigsliied, Mersseburger
Zauberssprüche, Muspilli,
M
O
Otfrid’s Evangelienb
E
buch (exceerpts), Petruslied,
Wessobbrunner Geebet, Old High Gerrman Physsiologus, N
Notker’s Prrologue,
Wessobbrunner Prredigt, Wieener Hund
desegen, Isidor
I
trannslation (ex
xcerpts),
Stra buurger Eide, Tatian Gosppel Translattion (excerp
pts)
Thee database with
w all the texts
t
used iin the study, as well as a search toool, are avaiilable on
the Internet at: http://ia.un
ni.lodz.pl/cicchosz/enhig
g.38
5 THE FREQ
QUENCY OF
O V-2 DE CLARATIV
VE CLAUSE
ES
If O
Old Englishh and Old High
H
Germaan had the same syntaactic system
ms, then wee should
expect similar frequenciess of variouss word ordeer patterns in
i the sampples. This, however,
h
38
Thhe author wouuld like to thaank Piotr Peziik (University
y of Lodz), wh
ho helped to ccreate the database and
the oonline interfacce (corpus design based on P
Pezik, Levin and
a Uzar 2006
6).
~ 85 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
is simply not the case. Table 6.1 shows the distribution of V-2 main non-conjoined
declarative clauses (conjoined clauses must be analysed separately since in Old Germanic
languages coordinating conjunctions could trigger different word order patterns).
text type
poetry
original prose
translated prose
Old English
%39
33
136
38
46
64
90
Old High German
%
57
128
81
116
41
78
40
Table 6.1: V-2 non-conjoined declarative clauses (ambiguous patterns excluded).
It is apparent that the two languages do behave in a different way. The first and most
important conclusion to be drawn from this table is that in two native samples, i.e. poetry
and original prose, Old High German demonstrates a visibly higher frequency of the
analysed pattern (57 vs. 33% and 81 vs. 38%). In translations, though, the tendency is
reverse (41 vs. 64%). In order to investigate the reasons for this puzzling discrepancy, it is
necessary to consider the function of the V-2 order.
6 FUNCTION OF V-2
The V-2 pattern seems to have functioned as a neutral, unmarked order of main clauses in
all Old Germanic languages. According to Smith (1971), the V-2 pattern is most common
in independent statements both in Old English and Old High German. Basing his
conclusions on the results from a few Old Germanic languages, he claims that:
After 600 A.D. the verb-second order seems to have attained the status of non-marked
order in the Germanic dialects generally (Smith 1971: 138).
Smith analyses the three basic positions of the finite verb and comes to the following
conclusions (1971: 291):
a) verb-final was the primary Germanic unmarked order, inherited from Proto-IndoEuropean, used mainly in subordinate clauses;
b) verb-initial was the primary Germanic marked order, also inherited from ProtoIndo-European, used in commands, conjoined clauses and dramatic sentences;
c) these two orders which Germanic inherited from Proto-Indo-European were
finally supplemented by a third pattern – verb-second – which came as a strong
innovation.
39
The corpus contains 1328 clauses of the analysed type.
Ambiguous patterns are short, structurally ambiguous clauses consisting of the finite verb only (which
makes it impossible to decide whether they should be classified as V-1 or V-final) or a constituent followed
by a finite verb (which makes it impossible to decide whether they should be classified as V-2 or V-final).
40
~ 86 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
Nonetheless, this basic distinction does not explain the exceptional behaviour of
translations. What can be useful here is the study of discourse relations in Old High
German translated prose, conducted by a team at the Humboldt University of Berlin. The
scholars analysed the relation of V-1 and V-2 clauses and discovered that:
… verb-initial structures establish coordinative discourse relations whereas verb-second
clauses signal subordinating linkage to the previous discourse part. In this sense, a verbinitial occurrence within the text, even involving an already established discourse referent,
may be perceived as a signal that the utterance quits a previous passage of subordination
and returns to the main line of the discourse (Hinterhölzl and Petrova 2005: 3).
This observation is very significant, as it shows that the relative frequency of V-1 and V-2
main clauses depends on the structure of the narrative, not on the language itself. Thus, it
is the sum of V-1 and V-2 clauses that should be considered as the basis for further
analyses, and the statistics are presented in Table 6.2.
V-1
V-2
Total
V-1
V-2
Total
V-1
V-2
Total
OE poetry
%
25
104
33
136
58
240
OE original prose
%
12
15
38
46
50
61
OE translated prose
%
5
7
64
90
69
97
OHG poetry
%
22
50
57
128
79
178
OHG original prose
%
1
2
81
116
82
118
OHG translated prose
%
29
54
41
78
70
132
Table 6.2: The interdependence of the V-1 and V-2 pattern in main non-conjoined declarative clauses
(ambiguous patterns excluded).
It is interesting to observe that after summing up the frequency of V-1 and V-2 clauses,
the two translation samples demonstrate practically identical proportions (69 vs. 70%).
Therefore, it is only logical to assume that the apparent difference in the frequency of V-2
clauses is directly related to discourse relations and various narration techniques, with
Old High German translations employing the V-1 pattern relatively often (29%) and Old
~ 87 ~
A
Anna
Cichoszz: Against the “West-Germa
anic syntax” hypothesis
h
Englishh translationns clearly avoiding iit (5%).41 In
I the otheer samples,, however, the
summedd frequenciies are stilll higher forr Old High
h German, which
w
sugggests a stronger
influencce of the V--2 constrain
nt on this lannguage.
As the pposition of the finite veerb is not ennough to drraw any deffinite concluusions, it is now
necessaary to analysse another important
i
phhenomenon
n related to the V-2 connstraint, nam
mely
subject inversion.
7 SU
UBJECT INV
VERSION
As it w
was shown inn example 2,
2 in a langguage with the
t V-2 con
nstraint the subject and
d the
finite verb are inveerted when the clause is introduced by anoth
her constituuent (most often
o
an adveerbial). Thiss results in a relatively high numb
ber of V-S clauses.
c
Figuure 6.1 pressents
the relaative frequenncies of S-V
V and V-S ppatterns acro
oss the samp
ples.
Figure 6.1:
6 The rate of V-S inversioon in main non
n-conjoined declarative
d
claauses.
omogenous , the S-V order
o
It is vissible that alll the three Old Englissh samples are very ho
clearly dominates, though theere is a subbstantial nu
umber of V-S clauses ppresent. In Old
High G
German, how
wever, the frequenciess of V-S clauses
c
are much highher and the two
patternss seem to be rather ballanced, whiich is in acccordance wiith the earliier observattions
pointingg to the stroonger position of the V
V-2 rule in Old High German. Taable 6.3, which
w
presentss the frequeency of subjjects that avvoid inversion, in yet an
nother conffirmation off this
preliminnary concluusion.
41
The diffference betw
ween the frequeency of V-1 cclauses in Old English and Old
O High Germ
rman translatio
ons is
not the toopic of this paaper, but a detaailed explanattion can be fou
und in Cichossz 2009. Sufficce it to say that the
discrepanncy is most probably relateed to a diachrronic development (the OH
HG translationns analysed fo
or the
purpose oof this study come
c
from thee 9th century, tthe OE translaations were created later, att the end of the 10th
century).
~ 88 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
text type
poetry
original prose
translated prose
Old English
%
42
91
42
29
28
27
Old High German
%
17
14
13
9
30
22
Table 6.3: S-Z clauses introduced by an element other than the subject.
When another constituent is placed at the beginning of the main declarative clause, the
subject still precedes the finite verb in a substantial portion of the Old English sample
(42% in poetry and original prose), whereas in Old High German such a situation is rather
rare (only 17% in poetry and 13% in original prose). Translations behave in a very
consistent way, which suggests a similar influence of Latin on the samples.
Examples 8-10 illustrate the lack of inversion:
(8) Æfter ðisum ic wearð gebroht (OE, Alia Visio)
[After that I was brought]
(9) be þam man mihte oncnawan (OE, The Battle of Maldon)
[by that one could know]
(10) Þær ic ne gehyrde butan hlimman sæ (OE, The Seafarer)
[there I could hear nothing but the roar of the sea]
It is necessary to point out here that all the three subjects in the clauses presented above
are light; the issue of weight will be discussed in detail in section 7.
As mentioned before, the inversion of the subject and the finite verb is usually caused by
an adverbial. According to the “West Germanic Syntax” hypothesis, it is even an
obligatory phenomenon:
Where there is an initial adverbial word the word-order pattern is altered; the subject
follows the verb, and object and complement words follow the subject (Davis and
Bernhardt 2002: 63).
Figure 6.2 shows how often the initial adverbial triggers inversion in the study corpus
used in the present analysis.
~ 89 ~
A
Anna
Cichoszz: Against the “West-Germa
anic syntax” hypothesis
h
Figu
ure 6.2: The rate
r of V-S invversion in maiin non-conjoin
ned declarativve clauses intrroduced by an
n
aadverbial.
It is cleear from tabble 2 that th
he rate of innversion is highest in the
t case of two native Old
High G
German sam
mples: poeetry (82%) and origin
nal prose (92%). Whhereas the two
translation samplees once again demonnstrate simillar proportions, the tw
two native Old
Englishh samples innvert definittely less ofteen (44% in poetry and 62% in origginal prose)).
On the whole, it apppears that the mechannism of inveersion is rellatively stroong in Old High
H
Germann, with bothh heavy (lex
xical) and liight (pronominal) phraases behavinng accordin
ng to
the sam
me rule, as inn the examp
ples presenteed below:
(11)) Hier beginn ih einna reeda umbe ddiu tier (AHD
D Physiolog
gus)
[Here start I my story of
o an animaal]
(12)) mit geru scal
s man geb
ba infahan, ort widar orte
o (Hildebrrandslied)
[with weappon should one
o gifts acccept, sword against swo
ord]
(13)) Thanne spprah hluduig
g (Ludwigsllied)
[Then spokke Ludwig]
At this point, the difference
d
between
b
Oldd English an
nd Old High
h German iis quite evid
dent.
Davis aand Bernharrdt (2002: 55),
5 howeveer, claim th
hat in clausees with an iinitial adverrbial
71% off OE and 75%
7
of OH
HG main nnon-conjoineed declaratives demonnstrate the V-S
order. T
The results obtained du
uring the prresent study
y are compleetely inconssistent with
h this
observaation, with frequenciess ranging ffrom 44% (OE poetry
y) to 92% (OHG orig
ginal
prose). Thus, it seeems that th
he distributiion of this structure iss very muchh dependen
nt on
dy corpus made the results totally
text typpe and a different structure oof the stud
incompparable. Thhe present analysis ssuggests th
hat, apart from
f
the w
word order in
nt operated m
more efficieently in the Old High G
German sam
mple.
translations, the V--2 constrain
~ 90 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
8 THE INFLUENCE OF WEIGHT
The last factor that needs to be included in the analysis is the influence of subject weight.
Numerous studies of Old English indicate that subjects and objects tend to appear in
different positions depending on their weight, with light pronominal phrases located more
towards the beginning of the clause and heavier phrases “postponed”:
Main clauses introduced by a constituent other than the subject show an interesting quirk:
while inversion takes place with all types of finite verb in a large majority of cases when
the subject is nominal, a personal pronoun subject remains in preverbal position (Fischer at
al. 2000: 49).
In the case of inversion, this means that light subjects should be inverted less often than
heavier subjects. Table 6.4 shows the rate of inversion of light subjects across all the
samples.
text type
poetry
original prose
translated prose
Old English
inverted
non-inverted
%
%
30
21
70
50
42
16
58
22
44
19
56
24
Old High German
inverted
non-inverted
%
%
84
41
16
8
85
41
15
7
50
7
50
7
Table 6.4: The behaviour of light subjects in non-conjoined declaratives
It is apparent that in all Old English samples most light subjects were not inverted, though
the phenomenon is most evident in poetry (70% stay on the second position). On the other
hand, in Old High German light phrases were inverted in the vast majority of cases (84%
in poetry and 85% in original prose). The behaviour of Old High German translations
must be ignored since only 14 light subjects were discovered in the sample42, and this
number is not enough to draw any reliable conclusions. Yet, when we compare all the
other samples, it becomes clear that in Old High German the V-2 constraint was strong
enough to exert an influence on all subjects regardless of their weight, whereas in Old
English light phrases were less liable to conform with the rule.
9 CONCLUSION
The present analysis has proven that the V-2 constraint was present in both languages
under investigation, but the degree of its influence was different. In Old High German the
phenomenon of verb-seconding was visibly stronger than in Old English, which is
illustrated by the higher incidence of V-2 declaratives in general, the greater rate of
subject inversion and a substantially weaker impact of weight on the word order of main
42
This is related to the very high incidence of subjectless clauses in Old High German translations
(pronominal subjects were usually unexpressed and therefore it is impossible to investigate their typical
position in the clause).
~ 91 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
clauses. It may be safely assumed that the V-2 constraint was already a well-developed
phenomenon in Old High German, whereas in Old English it never reached the status of a
rule and after a period of constituting a visible tendency it started to lose strength and
disappear from the syntactic system of English.
This observation is in accordance with the analysis of Old High German conducted by
Katrin Axel:
... I will argue extensively in the present study that some crucial steps toward the verbsecond grammar, such as the generalisation of verb movement in root clauses, has already
been taken in OHG (Axel 2007: 1)
and Marian Bean who claims that Old High German:
(...) looks remarkably like the V-2 language of modern German: the verb is in second
position in main clauses preceded by any single element and in final position in subordinate
clauses (Bean 1983: 52)
Bean even claims that English was never really on its way to acquire the V-2 rule since
according to her theory: “X’VS is a narrative device rather than a sign of a developing
verb-second constraint” (Bean 1983: 137).
The differences between languages discussed above can also be used to refute the “West
Germanic Syntax” hypothesis. The discrepancies between Old English and Old High
German word order exist and they cannot be seen as occasional deviations from
“tendencies approaching the status of rules” (Davis and Bernhardt 2002). Of course, the
syntax of Old English and Old High German exhibited many similarities; they were two
closely related languages from the same group that were bound to resemble each other
after only a few centuries of isolation. Even though the two languages were still quite
similar, significant differences started to appear they should not be disregarded; the
discrepancies are present in the corpus and clearly show that the “West Germanic Syntax”
is a theoretical construct which does not reflect the textual, and – as a consequence linguistic, reality of Old English and Old High German.
Another important conclusion that may be drawn from the present analysis is the fact that
Old English and Old High German started to differentiate before the end of the Old
Germanic period, i.e. before the 11th century. This means that the changes which took
place in the English system have to be attributed to some mechanisms which started to
operate already in the Old Germanic period. What was the real reason that triggered the
change is a question that goes beyond the scope of the present study. One can only
speculate as to other foreign influences (Old Norse spoken in the Danelaw being the most
obvious candidate) or some internal mechanisms which started the chain reaction
described by Bean (1983) as “phonological-process-leading-to-morphological-lossleading-to-syntactic-change”. Yet, the view that the loss of case markings was the source
of all the syntactic changes has recently been questioned, with the new theory suggesting
a more parallel mechanism:
~ 92 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
The two case studies examined here do not support the view that the fixing of word order in
English was driven solely by the loss of case inflections or the assumption that at every
stage of English, there was a simple correlation between less inflection and more fixed
word order. This is not to say that deflexion did not play an important role in the
disappearance of some previously possible word orders. It is reasonable to suggest that
although it may have been pragmatic considerations which gave the initial impetus to
making certain word orders more dominant than others, deflexion played a role in making
these orders increasingly dominant. It seems likely that the two developments worked hand
in hand; more fixed word order allowed for less overt case marking, which in turn
increased the reliance on word order (Allen 2009: 220).
Whatever the reason for all the changes that affected English, they started to take place
before the 11th century and thus differences between the two languages under
investigation started to appear already in the Old Germanic period, which points out that
any overgeneralisations like “the Old Germanic syntax” need to be treated with caution
and checked in a varied sample of texts before being widely accepted.
REFERENCES
Allen, Cynthia. 2009. Case Syncretism and Word Order Change. In van Kemenade, Ans
and Los, Bettelou (eds.). The Handbook of The History of English, Oxford: WileyBlackwell, 201-223.
Axel, Katrin. 2007. Studies on Old High German Syntax: Left Sentence Periphery, Verb
Placement and Verb Second. Amsterdam/Philadelphia: John Benjamins.
Bean, Marian. 1983. The Development of Word Order Patterns in Old English. London:
Croom Helm.
Cishosz, Anna. 2009. The influence of text type on the word order of Old Germanic
languages: a corpus-based contrastive study of Old English and Old High German.
Unpublished Ph.D. Dissertation, University of Lodz.
Cichosz, Anna & Pezik, Piotr. 2008. ENHIG: Old English and Old High German
database <http://ia.uni.lodz.pl/cichosz.enhig>.
Davis, Graeme & Bernhardt, Karl A. 2002. Syntax of West Germanic: The Syntax of Old
English and Old High German. Göppingen: Kümmerle Verlag.
Davis, Graeme. 2006. Comparative Syntax of Old English and Old Icelandic: Linguistic,
Literary and Historical Implications. Oxford: Peter Lang.
Fischer, Olga, van Kemenade, Ans, Koopman, Willem & van der Wurff, Willem. 2000.
The Syntax of Early English. Cambridge: Cambridge University Press.
Hinyerhölzl, Roland & Petrova Svetlana. Rhetorical Relations and Verb Placement in
Early Germanic Languages. Evidence from the Old High German Tatian translation
(9th century). In Stede, M. et al. (eds). Salience in Discourse. Multidisciplinary
Approaches to Discourse. Münster: Stichting/Nodus, 71-79.
Petrova, Svetlana & Solf, Michael (in press). On the Methods of Information-Structural
Analysis of Texts from Historical Corpora. A Case Study on the OHG Tatian. In
Hinterlhölzl, R. & Petrova, S. (eds). New Approaches to Word Order Variation and
Word Order Change, Berlin: Mouton de Gruyter.
~ 93 ~
Anna Cichosz: Against the “West-Germanic syntax” hypothesis
Pezik, Piotr, Levin, Eric & Uzar, Rafael. 2006. Zastosowania baz danych w
językoznawstwie. [Using Databases in Linguistics]. In Lewandowska-Tomaszczyk .
Barbara (ed.). Podstawy językoznawstwa korpusowego. Łódź: Wydawnictwo
Uniwersytetu Łódzkiego.
Smith, Jesse. 1971. Word Order in the Older Germanic Dialects. Unpublished Ph.D.
Dissertation, University of Illinois.
~ 94 ~
CHAPTER 7
WHICH-PHRASES DO MOVE
BENJAMIN KRATZ
Graduiertenkolleg “Satzarten: Variation und Interpretation“
Johann Wolfgang Goethe-Universität Frankfurt
Since Pesetsky (1987) it is widely accepted that d-linked wh-phrases (DWH) lack operator status.
Van Craenenbroeck (2008) is an attempt to frame this insight into a syntactic analysis separating the
element bearing wh-morphology from the operator-function. A number of shortcomings make it
appropriate to refine this analysis. First, there are problems with respect to theta-assignment,
selection, and reconstruction since van Craenenbroeck takes the DWH itself to be base-generated in
the left-periphery. Second, the analysis does not take into account any other recurrent claims on the
nature of which-phrases (e.g. that DWH are topics). It is shown why the proposal in van
Craenenbroeck (2008) is problematic and an alternative analysis is sketched which does not run into
these problems. This alternative analysis tries to incorporate the existing work on DWH and the
original idea of van Craenenbroeck that the wh-word and the syntactic operator are separate items.
1 INTRODUCTION
It is a well known fact that which-phrases show a number of properties that set them apart
from other wh-phrases.43 One of these properties concerns the operator-status of whichphrases. Although it is widely accepted that all wh-questions involve an operator-variable
dependency, which-phrases are claimed not to be syntactic operators (cf. Pesetsky 1987).
Van Craenenbroeck (2008) - henceforth VC – is an attempt to make sense of this
contradiction. To do so, he presents seven sets of data from Germanic (some of them new
to the literature on which-phrases). His basic claim is that which-phrases are basegenerated in the C-domain (and at the same time there is an operator first-merged in the
argument position which is later moved to a position in the CP to check of its operatorfeature), whereas other (simplex) wh-phrases are base-generated in the VP.
43
A note on terminology: van Craenenbroeck (2008) uses the term ‘complex wh-phrases’ but all of his
examples (except one) involve which-phrases. Since these are generally taken to be inherently discourselinked wh-phrases (DWH), I will use the terms which-phrases and DWH in alternation throughout this
chapter. For ease of exposition, the term ‘simple wh-phrases’ will be used to refer to wh-words like who
and what.
I refrain from classifying all non-monomorphematic wh-phrases as belonging to a single group.
This would include such diverse items as which N, whose N, how many Ns etc. Although I believe the
differences are encoded in the morphosyntax of these expressions, it does not seem reasonable to extend the
claim about the separation of the lexical wh-item and the operator (see below) to all of them.
~ 95 ~
Benjamin Kratz: Which-phrases do move
The aim of this chapter is threefold: First, I want to show that most of the data discussed
in van Craenenbroeck (2008) do not force an analysis based on the idea that whichphrases are base-generated in the C-domain. Some of the data simply show that whichphrases end up in a position higher in the C-layer than other wh-phrases do. Others can
easily be explained without claiming that which-phrases are base-generated in the Cdomain. Second, despite this, I want to maintain what I think is the most important insight
in VC (besides the very interesting new data). Namely that separating the operator and the
wh-element is the right strategy to account for some of the special properties of whichphrases. Third, I want to sketch an analysis which incorporates this idea and at the same
time is able to account for other properties of DWH not touched on by van
Craenenbroeck (2008). These include the licensing of resumptive pronouns by DWH and
the lack of superiority-effects with DWH.
I start with a brief summary of VC’s analysis of which-phrases in section two. Section
three lists a number of conceptual arguments against this analysis as well as criticism
concerning the analyses of particular data-sets VC uses to justify his claims. In the forth
section, I lay out the basics of my alternative analysis and present empirical and
conceptual arguments supporting it. Section five concludes this chapter.
2 BASE-GENERATION
OF
WHICH-PHRASES
CRAENENBROECK’S (2008) ANALYSIS
IN
CP:
VAN
In his 2008 paper “Complex wh-phrases don’t move: On the interaction between the split
CP-hypothesis and the syntax of wh-movement”, van Craenenbroeck proposes that CP is
split into (at least) two projections:
(1)
a. Wh-features are checked in CP1
b. Operator-features are checked in CP244
Additionally, it is claimed that which-phrases behave differently because of (2):
(2)
With DWH the operator is not a genuine part of the wh-phrase
Based on these assumptions, the following derivations for simple wh-phrases and DWH
are proposed (I will refer back to (4) as analysis A1):
44
What the general properties of such an operator-feature should be is left open in VC (and by many other
authors who assume it). It is not a trivial move to decide whether the property of being an operator can be
encoded by a feature under any conception of this notion or not.
~ 96 ~
Benjamin K
Kratz: Which-p
phrases do mo
ove
(3)
simple wh-phrases
w
s
(4)
which-pphrases (A1
1)
8) himself aadmits that only
o
two off the data-seets he preseents lead
Vann Craenenbrroeck (2008
one to assumee an analysiis like A1 – especially
y the claim
m that whichh-phrases are
a firstmerrged in SpeecCP1. He offers
o
an altternative in
n which the operator iss base-geneerated in
the CP-domainn and the wh
hich-phrase base-generrated inside VP (I will refer back to
t (5) as
45
anallysis A2):
45
Noote that this iss in a sense a recast
r
of Peseetsky’s (1987)) original analy
ysis of DWH in terms of un
nselective
bindding of the whiich-phrase by a Q morphem
me in the C-lay
yer.
~ 97 ~
Beenjamin Kratzz: Which-phra
ases do move
(5)
which-phraases (A2)
According to van Craenenbro
C
eck, the twoo empirical facts that are
a problem
matic for succh an
approacch are: (i) thhe wh-copyiing construcction in Gerrman; and (ii)
( prepositiion strandin
ng in
Dutch. Since the main aim of this chhapter is to show thatt which-phrrases are baseb
c
with this cllaim,
generated in VP likke any other argument--wh-phrase and A2 is compatible
ows concernns analysis A1
A (exceptiions are marrked).
most off the criticissm that follo
3 PR
ROBLEMAT
TIC ASPEC
CTS OF AN
NALYSIS OF
O A1
There aare a numbeer of probleematic aspe cts for an analysis
a
likee A1. I list them here with
only briief explanattions as to why
w I thinkk they are reeal shortcom
mings. In thhe next sectiion I
elaboratte my criticcism on som
me of them and try to show
s
how my
m own anaalysis of wh
hichphrasess offers posssibilities to overcome thhese probleems.
ual Problems
3.1 Conceptu
erb
3.1.1 Selection by the ve
The (reestriction off the) which-phrase is aan argumen
nt of the verrb (i.e. it is selected by
y the
umptions shhould be baase-generateed inside thhe VP to recceive
verb), aand under sttandard assu
its thetaa-role (or case;
c
see beelow). Onee could opp
pose this crriticism of aanalysis A1 by
claiminng that (for example) Aoun
A
and L
Li (2003) maake a similar claim abbout certain whconstructions in Lebanese
L
Arrabic. Supeerficially, th
his seems to
t be true, but there is
i an
importaant differencce between the data disscussed in th
hese two works. In conntrast to thee fact
that in all examplles in van Craenenbro
C
oeck (2008)) the DWH is only diisplaced claauseinternall, the exam
mples in Aoun
A
and L
Li (2003) (for which the authoors claim baseb
generation of the which-phraase in the C
C-domain) involve deependenciess across claauseboundarries (and isslands in paarticular) annd the wh-p
phrase is ‘ttaken-up’ bby a resump
ptive
pronounn in the arrgument-position in thhe lower cllause. In otther words,, Aoun & Li’s
analysiss is rooted in widespreead assumpptions aboutt general prroperties off movementt and
its connnection to isslands and (one particul
ular type of) resumption
n.
~ 98 ~
Benjamin Kratz: Which-phrases do move
Another related problem with base-generation of DWH inside the C-domain is caseassignment. How does a which-phrase receive case - either inside VP or in IP - if it never
occupies a position this low? The following data from German show that the nominal
restriction as well as the lexical element which can bear case-morphology:46
(6)
a. [Welches
Mantels]GEN hat sich der Typ entledigt?
Which-GEN coat-GEN has REF the guy get-rid-of
‘Which coat did the guy carry out?’
b. [Welchen
Mädels]DAT hat Peter eine Uhr gegeben?
Which-DAT girls-DAT has Peter a
watch given
‘Which girl did Peter give a watch to?’
c. [Welches
Buch]ACC hat Jens gelesen?
Which-AKK book-AKK has Jens read
‘Which book did Jens read?’
3.1.2 Reconstruction
If DWH never occupy a position in the IP (and VP), it is not possible to interpret it there
at LF under standard assumptions of reconstruction (as an activation of a lower copy/trace
at LF).47 Even a simply example like the following is a problem for an analysis of this
kind:
(7) [Which pictures of himselfj]i did Johnj like ti?
If the which-phrase is base-generated in its surface position (as in analysis A1), then there
is no possibility for John to bind the anaphor himself inside the wh-phrase (under standard
Minimalist assumptions about binding of anaphors; see e.g. Radford 2004: 93, 197).
The selection- and the reconstruction-problem for analysis A1 just described are
essentially the same Boeckx (2003) determined for the analysis of wh-in-situ and whresumption constructions in Lebanese Arabic in Aoun & Li (2003). Boeckx writes:”[…]
such an approach would lead to the conclusion that the very same elements can be
licensed by being bound (in situ) or by binding (resumption). I know of no other element
that can be both a binder and a bindee in identical configurations. In addition, a nonmovement approach would have to posit two different First-Merge mechanisms for the
very same elements (either [who] or [which X] are base-generated in their theta-position
or they are base-generated in SpecCP). Such a theory would then lose any hope of
regularizing First-Merge operations (“base structures” in a pre-theoretic sense) (23)”.
46
Thanks to Andreas Blümel (p.c.) for encouraging me to include these data.
Van Craenenbroeck (2008: 17) admits that reconstruction is a problem for analysis A1 and hints at a
semantic reconstruction mechanism to overcome it but leaves open as to how such a mechanism works.
47
~ 99 ~
Benjamin Kratz: Which-phrases do move
3.1.3 No reference to other work on which-phrases
There is a huge amount of literature on d-linking in which a number of additional
properties of DWH are discussed. For example, Rizzi (2000), Polinsky (2001) and
Grewendorf (2008) claim that which-phrases are syntactic topics. Taking these findings
into consideration, some of the data used in VC to justify analysis A1 can receive a
different analysis and therefore are no longer corroborating A1. The differences between
which-phrases and other wh-phrases most often reported in the literature are the following
three phenomena:
(a)
DWH can escape weak islands
(b)
DWH are able to/must licence resumptive pronouns
(c)
DWH are immune to superiority
Admittedly, if which-phrases are base-generated in the CP it seems to follow naturally
that they can obviate superiority. A problem for such reasoning is the fact that the ability
of DWH to appear before a ‘superior’ wh-word is in most cases only optional. The whichphrase can also follow the ‘superior’ wh-element in the left-periphery (or even stay insitu).
At one point of his paper, van Craenenbroeck claims that which-phrases are not
necessarily base-generated in the C-layer: “[…] while the complex wh-phrase is merged
in SpecCP1 in [(4)], in a multiple wh-question (where it is not required to type the clause)
it can just as easily be merged in an argument position (2008: 6)”. This implies a ‘lookahead-mechanism’, because at the point of the derivation the argument of the verb is firstmerged, the computational system needs to “know” if the sentence is a multiple-whquestion or not (among other things). It is also not obvious how the which-phrase and the
operator can be merged together in the argument position under the assumptions on which
analysis A1 is based.
3.2
Problems with the Analyses of Particular Empirical Phenomena
Besides these more general problems, I also identified a number of problems with van
Craenenbroeck’s (2008) particular analyses of the data-patterns he presents. These are
listed below:
3.2.1 Doubly filled COMP phenomena in Frisian and dialectal Dutch
The first two sets of data are from Frisian and dialectal Dutch. The patterns are virtually
identical. Therefore, I will illustrate my criticism only by using the Dutch data. The
relevant examples come from Strijen Dutch:
~ 100 ~
Benjamin Kratz: Which-phrases do move
(8)
a. Ik weet nie <of> met wie <of> Jan oan et proate was.
I know not if with who if John on it talkINF was
‘I don’t know who John was talking to.’
b. Ik weet nie of met wie dat Jan oan et proate was.
I know not if with who that John on it talkINF was
‘I don’t know who John was talking to.’
c. Ik vroag me af <*of> welke jonge <of> die maisjes gistere gezien hebbe.
I ask me PRT if which boy if the girls yesterday seen have
‘I wonder which boy the girls saw yesterday.’
d. * Ik vroag me af of welke jonge dat die maisjes gistere
gezien hebbe.
I ask me PRT if which boy that the girls
yesterday seen have
INTENDED: ‘I wonder which boy the girls saw yesterday.’
The data in (8) do not show any connection to the operator-status of the wh-elements
involved (and therefore cannot be used to argue for an analysis like A1). They only
illustrates that which-phrases end up in a position rather high up in the C-domain (they
can only precede of), whereas other wh-phrases can occupy this position or a lower one
(preceding dat and following of). Furthermore, (8a/c) and (8b/d) do not form minimal
pairs: (8a) and (8b) involve wh-PPs and (8c) and (8d) do not. Without more data, one
could not decide whether the differences in grammaticality are due to this additional
factor.
3.2.2 Swiping in English
Swiping is an acronym for “Sluiced Wh-word Inversion In Northern Germanic”. Sluicing
in general is the deletion of the complement of a wh-item in embedded wh-sentences.
Swiping is exceptional because the wh-word is followed by an inversed preposition. It is
restricted to simple wh-phrases, i.e. which-phrases are excluded, as Merchant (2002)
originally observes:48
(9)
a. Ed gave a lecture, but I don’t know what about.
b. *Ed gave a lecture, but I don’t know which topic about.
VC takes this as a proof for the claim that which-phrases occupy a smaller number of
positions in the C-domain than other wh-phrases. He claims that the preposition in (9a)
gets stranded in SpecCP2 with the wh-element moving higher up to SpecCP1. Since DWH
never occupy SpecCP2 in analysis A1, it looks as if it can easily explain why (9b) is
ungrammatical. But again, nothing hinges on the fact that DWH are base-generated in the
C-layer. Analysis A2 – in which SpecCP2 is occupied by an empty operator – can explain
the differences in (9) without appealing to base-generation of the which-phrase in
SpecCP1.
48
Looking at the list of wh-elements that can partake in swiping in Merchant (2002), the term ‘complex
wh-phrases’ seems to be justified in this case. But as the short summary of Merchant’s proposal in the text
below should make clear, this has nothing to do with the operator-status of the wh-phrases involved.
~ 101 ~
Benjamin Kratz: Which-phrases do move
Looking at Merchant’s (2002) original analysis for these facts, it seems reasonable to
explain the difference in (9) by reference to the different internal structures of
monomorphematic wh-elements and which-phrases. Merchant’s analysis is based on the
claim that swiping involves head-movement and incorporation of a wh-head to the Phead. Since which-phrases are always phrases (they obligatorily take a noun-complement;
with bare which this can be phonologically empty) they cannot raise to the P-head and
incorporate into it.49
And finally, proposal A1 predicts that which-phrases can never strand a preposition in the
IP/VP-domain, since they never occupy a position that low in the structure. This
prediction is not borne out as the following data from Radford (2004: 192) show:
(10)
a. IKEA only actually has ten stores [from which to sell from]
b. The hearing mechanism is a peripheral, passive system over which we have no
control over.
3.2.3 Wh-Copying in German
Wh-copying as illustrated by (11a) is generally taken to result from the multiple spell-out
of copies of a wh-phrase (which is usually taken to be base-generated in the subordinate
clause). To account for the difference in (11), VC proposes that which-phrases “do not
undergo movement at all throughout the derivation (2008: 10)”. Therefore, they cannot
leave a copy in the lower clause which could be spelled-out.
(11)
a. Wen glaubt Hans wen Jakob gesehen hat?
WhoACC thinks Hans whoACC Jacob seen
has
‘Who does Hans think that Jacob saw?’
b. * Welches BuchACC glaubst du welches BuchACC Hans liest?
Which book think you which book John reads
‘Which book do you think Hans reads?’
The problem that arises for analysis A1 is essentially the same as mentioned in section
3.1: how does the computational system come to know that in these cases, the whichphrase is not first-merged in the argument position?
49
Merchant (2002) takes this head-movement to occur after spell-out. But in his footnote 12, he sketches
some possibilities to analyze it as taking place in syntax proper. Since this issue is orthogonal to the topic of
this section, I will not discuss the pros and cons of a “head-movement in syntax” approach to swiping.
~ 102 ~
Benjamin Kratz: Which-phrases do move
Besides this, the wh-copying paradigm in German is more complex than the data in van
Craenenbroeck (2008) suggests. There are even simple wh-phrases that cannot participate
in this construction:50
(12)
a. *?Wem glaubst du wem
Hans das Buch gegeben hat?
WhoDAT think you whoDAT Hans the book gave
‘Who do you think Hans gave the book to?’
b. *?Wie glaubst du wie Hans geschlafen hat?
How think you how John slept
has
‘How do you think Hans has slept?’
So it seems as if there are external factors governing the appearance of wh-items in this
construction. It does not suffice to recur to the operator-status of the wh-elements
involved or the first-merge site of the elements under examination. This leads me to the
conclusion that at this point wh-copying cannot be used to decide whether an analysis of
which-phrases is more appropriate than another.
As a last point it should be mentioned that verb-second (in Germanic) is usually taken to
be a reflex of (in the relevant cases) wh-movement. Take the data in (13):
(13)
a. Du glaubst, Hans liest die Zeit.
You believe John reads the Times
‘You believe that John is reading the Times.’
b. Welche Zeitung
glaubst du liest Hans?
Which newspaper believes you reads John
‘Which newspaper do you believe John reads?’
c. *Welche Zeitung glaubst du Hans liest?
Which newspaper believes you John reads
‘Which newspaper do you believe John reads?’
If the which-phrase is base-generated in the C-domain, why does the finite verb in the
embedded clause obligatorily surface clause-initially in (13b) (the position it occupies in
V2-sentences) and not clause-finally as in (13c)?51
50
Note that the degree of grammaticality of wh-copying also varies among speakers of German. (i) is
judged as grammatical in Van Craenenbroeck (2008: 10) and this is taken as evidence for the claim that the
difference between e.g. wer (who) and welch- (which) does not depend on the head-phrase-distinction:
(i)
*Mit wem glaubst du mit wem Hans spricht?
‘Who do you think Hans is talking with?’
For me, being a native speaker of German, this sentence is clearly ungrammatical (or at least highly
marked) and this fact is supportive of the idea that some kind of the head/phrase distinction is at work here.
51
Again, thanks to Andreas Blümel (p.c.) for reminding me of this fact.
~ 103 ~
Benjamin Kratz: Which-phrases do move
3.2.4 Preposition-stranding in Dutch
Van Riemsdijk (1978) observes that Dutch is a partial preposition-stranding language,
inasmuch as only R-pronouns and empty operators can strand a preposition. VC claims
that which-phrases can also strand a preposition in Dutch, whereas other wh-phrases
cannot:
(14)
a. * Wie wil je niet mee samenwerken?
Who want you not with cooperate
INTENDED: ‘Who don’t you want to cooperate with?’
b. ? Welke jongen wil je niet mee samenwerken?
Which boy
want you not with cooperate
‘Which boy don’t you want to cooperate with?’ (Dutch)
In an proposal like van Craenenbroeck’s, which takes an empty operator to be involved in
the derivation of which-phrases, van Riemsdijk’s generalization can be retained: The
preposition is stranded by the empty operator, not the element which.
I follow VC in taking this observation as evidence for the fact that with DWH, the
operator is not part of the genuine wh-word. But as with the first two data-sets, nothing
hinges on the assumption that which-phrases are first-merged in SpecCP1. It would
suffice to show that the operator(-feature) is not part of the lexical element which, but
rather occupies its own position inside the DWH.
3.2.5 Free relatives in Dutch
In a number of languages (e.g. Dutch, German, English) free relatives cannot be
introduced by which-phrases. The following examples are from German:
(15)
a. Wer das nicht versteht, ist ein Idiot.
Who this not understands is an idiot
b. *Welcher Mensch das nicht versteht,
ist ein Idiot.
Which human this not understands is an idiot
In order to enable his analyses (both A1 and A2) to derive these facts, van Craenenbroeck
proposes that with free relatives, “we are dealing with a truncated C-domain, in which
CP2 is present, but CP1 is not (2008: 12)”. Since DWH never occupy CP2, they are not
able to occur in free relatives.52
The assumption that CP1 is missing with free relatives gives rise to another problem: If
VC is right and CP1 is the locus of wh-feature-checking, how is the wh-feature on a
52
Van Craenenbroeck (2008: 12) gives an example from Dutch to support his claim. Remember that in
Dutch, dat occupies CP2 and of occupies CP1. Now, only dat can occur with free relatives in Dutch, of is not
possible. This argument relies on the idea that dat in free relatives is the same element as the dat in the
embedded sentences in Dutch or occupies the same position. If not, the example mentioned does not
support the claim about the truncated CP with free relatives.
~ 104 ~
Benjamin Kratz: Which-phrases do move
simple wh-phrase in SpecCP2 checked in free relatives in Dutch? This seems to be a
nontrivial problem to me, if one assumes that the wh-movement in relatives is also due to
the checking requirement of the wh-feature/morphology (cf. den Dikken 2003 and Šimík
2007).
3.2.6 Spading in dialectal Dutch
The last construction discussed in VC is spading (Sluicing Plus a Demonstrative In Noninsular Germanic) in dialectal Dutch. Again, which-phrases are excluded from spading
(the examples are from Wambeek Dutch):
(16)
a. Jef eid iemand gezien, mo ik weet nie wou dat.
Jeff has someone seen but I know not who thatDEM.
‘Jeff saw someone, but I don’t know who.’
b. *Jef eid ne student gezien, mo ik weet nie welke student dat.
Jeff has a student seen but I know not which student thatDEM.
‘Jeff saw a student, but I don’t know which student.’
VC analyzes spading as follows: The demonstrative originates in SpecIP and focusmoves to SpecCP2. Then, the wh-element raises out of the VP, “tucks in” under the
demonstrative in SpecCP2 (cf. Richards 2001) and moves further to SpecCP1 in the course
of the derivation. To explain why DWH are excluded from spading, it is argued that with
which-phrases, the ellipsis site under sluicing is CP2 and not IP. This will delete the
demonstrative and the contrast in (16) is derived. This assumption is grounded in
Merchant’s (2001) observation that sluicing deletes the complement of the C-head which
hosts the wh-phrase. But an important caveat is in order here: Since both simple whphrases and DWH end up in SpecCP1 in analysis A1, it is not clear why the demonstrative
is not deleted with simple wh-phrases either.
4 COMPONENTS OF AN ALTERNATIVE ANALYSIS
4.1
Preliminaries
Summing up the previous section, van Craenenbroeck’s (2008) analyses (no matter if one
looks at A1 or A2) seem to be able to account for the data-sets he presents. On closer
inspection, however, this is only achieved by a range of assumptions which seem to be
only motivated by the analysis. In addition, some of the data-sets can easily be explained
without the claim that which-phrases are base-generated in the C-layer (as in analysis
A1).
Before I continue, I want to highlight that I agree with van Craenenbroeck on three
things: (i) With which-phrases, the lexical wh-element is not the (wh-)operator; (ii) The
differences between DWH and other wh-phrases are structural (VC’s footnote 1); and (iii)
A split-CP offers a way of accounting for these differences (page 4 of VC 2008). In
addition to that, van Craenenbroeck’s (2008) paper offers a range of very interesting new
data which every analysis of which-phrases has to explain.
~ 105 ~
Benjamin Kratz: Which-phrases do move
4.2
Basic Claims
I want to begin to lay out the basics of my own proposal by noting that I believe the two
positions van Craenenbroeck calls CP1 and CP2 are of a different nature than he claims. In
my view, it is appropriate to say that CP2 is FocP and CP1 is TopP.
There are several reasons for this assumption. First, it is standard to assume that whereas
FocP has operator-properties, TopP does not (cf. Grewendorf 2004; Benincà & Poletto
2004). So the idea that the (wh-)operator targets a position low in the C-domain can be
maintained. Second, identifying CP2 as FocP brings the additional benefit of being a
manifestation of the idea that focus and wh are (at least in a large array of languages)
closely tied (following a long tradition based on Horvath 1986). So what VC calls an
operator-feature can be analyzed as a focus-feature.5354 Third, TopP is generally taken to
sit higher in the structure than FocP (see Benincà & Poletto for arguments against a topicposition below FocP). Fourth, it seems to be the case that there are different positions
targeted by wh-elements – not only in different languages but also in a single language
itself, depending on the kind of wh-movement examined. Nevertheless, the focus-position
seems to be the locus of wh-feature checking in the majority of cases (or at least it is
involved in the derivation of most wh-questions). This opens up the possibility to analyse
CP1 as the projection of another formal-feature, the topic-feature, and not of the whfeature.
The idea that the derivation of which-phrases involves a topic-feature (i.e. that DWH are
syntactic topics) can be found in Comorovski (1996), Rizzi (2000), Polinsky (2001),
Richards (2001) and Grewendorf (2008). Following Erteshik-Shir (1997) and Rizzi
(2000), I take the topic-feature to be located on the lexical restriction of the which-phrase.
The restriction is contextually-given like topics (i.e. it is presupposed/specific).55
Together with the idea that CP1 is really TopP, and that TopP is located higher in the
structure than FocP, this topic-feature can account for the fact that DWH can move up to
a higher position in the C-domain than other wh-items.
The second main point of my analysis is that DWH are headed by a null D-head. The idea
of a null D-head with which-phrases goes back to the semantic analysis by Rullmann &
Beck (1998). They observed that which-phrases project presuppositions the same way a
non-wh-definite would. Boeckx & Grohmann (2004) expand on this and claim that this
empty D-head can get stranded in the base-position and can be spelled-out as a
resumptive pronoun. Through this move, they are able to account for the obligatory
53
VC himself writes: “[…] the demonstrative focus-moves to specCP2” (2008: 16).
This move also circumvents the conceptual problems that the assumption of an operator-feature faces.
Maybe it is more appropriate to assume that certain positions in the C-layer are marked for being operatorpositions (cf. Boeckx & Grohmann 2004) and to justified the features that drive movement to this positions
on independent grounds (like the focus-feature).
55
Cinque (1990: 53) notes that which-phrases and topicalized elements are very similar in their
referentiality-status (the semantic notion he assumes to be the basis of d-linking).
54
~ 106 ~
Benjamin K
Kratz: Which-p
phrases do mo
ove
pressence of reesumptive pronouns
p
w
with which-p
phrases in languages llike Roman
nian (cf.
Dobbrovie-Sorinn 1993; see also Doronn 1982 and Sharvit
S
1999 for Hebreew).
Conncerning thee relation off the lexicall item which
h and the op
perator, forr the purpose of this
Another po
chappter I follow
w the claim
m that the opperator is base-genera
b
ted in CP. A
ossibility
wouuld be to foollow Hagsttrom (19988) and Bošk
koviΕ (1998
8) in assumi
ming a Q-mo
orpheme
thatt can be gennerated together with diifferent wh--elements in
n a multiplee-wh-questio
on. Note
thatt both alternnatives are compatible with the viiew that thee lexical iteem which iss not the
bearrer of what could be caalled operatoor-propertiees.
I suuggest the foollowing intternal structture for whiich-phrases: On top off the structure, there
is ann empty D--head. This D-head takkes the wh-eelement whiich (bearingg the wh-feaature) as
its ccomplementt. At the botttom of the structure we
w find the (n
nominal) reestriction off the whelem
ment whichh bears a topic-feature.. In cases of
o movemen
nt of the w
wh-phrase, the WhP
raises out of the DP via Sp
pecDP. The correspond
ding tree loo
oks like (177):
(17))
Internall structure of
o which-phhrases
t structures for whicch-phrases proposed
p
in
n Boeckx & Grohmann
n (2004)
I deepart from the
and Boeckx (22003) and assume
a
an aadditional projection,
p
called
c
WhPP here, betw
ween the
DP and the NP
P headed by which. W
With this I follow,
f
amo
ong others, Radford (2
2004) in
assuuming that which
w
headss its own prrojection.56
18). The
Thee derivationn of which--phrases acccording to my analyssis is illusttrated in (1
operrator is firsst-merged in
n Foc0 - thiis can vary
y depending
g on the lannguage/consstruction
undder examinaation. From this positio n it is able to bind the variable thaat is represeented by
the whole DP
P in the baase-position (recall thaat the D-heead can geet a spell-o
out as a
resuumptive whhich also neeeds to be boound by an operator in
n A’-positioon). I assum
me clause
typiing to be able
a
to occu
ur via Agreee, i.e. the wh-phrase does not hhave to rise to the
speccifier of a designated
d
position
p
to check its wh-feature.
w
Independen
I
nt of this isssue, it is
mosst plausible to assume that
t the wh--feature is lo
ocated in th
he wh-word itself.
56
Thhe exact label is of no impo
ortance for thee present chaptter.
~ 107 ~
Beenjamin Kratzz: Which-phra
ases do move
(18)
Derivation of DWH:
To sum
mmarise the main
m pointss of the propposed analy
ysis so far:
(aa) DWH haave a comp
plex DP-strructure. Th
hey are heaaded by an empty D-h
head
w
which can bee stranded and
a spelled- out as a RP
P.
(bb) The restriiction of a DWH
D
is enddowed with a topic-feature.
(cc) The operaator with DW
WH is not iidentical wiith the lexical element w
which.
In the nnext two subbsections, I present a rrange of em
mpirical argu
uments that corroboratee the
first two claims. Inn subsection
n 4.5. I willl go throug
gh the data-sets discusssed in section 3
and shoow how an analysis allong the linnes sketched
d above cou
uld be ablee to account for
them).
4.3
Supporting Evidence I: DW
WH as top
pics
Since H
Horvath (19886), it has been
b
claimedd that all no
on-echo-wh
h-phrases aree endowed with
a focus-feature and therefore, one may ask if it is tenable to assume a w
which-phrasse to
bear a topic featuure. In a sense,
s
focuus and topiic seem to exclude eeach other. But
indepenndent of DW
WH, it hass been claim
med that th
he notion of
o topic is iinvolved in
n the
derivatiion of som
me instancess of wh-queestions (cf. Grohmann
n 2006 for German). This
possibillity for a whh-phrase to act as a toppic is illustraated by the Chinese exaample from
m Wu
(1996) in (19b) wiith a topicaalized wh-phhrase.57 Chiinese is a wh-in-situ
w
laanguage, so
o the
57
Not suurprisingly, thee sentence can
n only be usedd in a situatio
on that exactly
y fits the definnition of d-lin
nking:
It can oonly be uttered if the possible
p
answ
wers can be drawn from
m a set preeestablished in
n the
context/ddiscourse.
~ 108 ~
Benjamin Kratz: Which-phrases do move
fronting of a wh-element is exceptional, and one way to account for this is assuming a
topic-layer which the dislocated wh-item can target:
(19)
a. Zhangsan mai-le shenme?
Zhangsan buy-ASP what?
‚What did Zhangsan buy?‘
b. Shenme Zhangsan mai-le?
Strong syntactic evidence for the assumption that DWH are syntactic topics can be found
in Tsez (Polinsky 2001). This language exhibits a construction (long-distanceagreement/LDA) which forces a matrix verb to agree with the absolutive argument of the
embedded clause. The absolutive argument must also bear a topic-marker:
(20)
a. Enir
[uza magalu
-gon bac’ruli] b-iy-xo.
Mother boy bread.ABS.III-TOP ate
III-know-PRS(present tense)
‘The mother knows that the bread, the boy ate’
b. *Enir
[uza magalu
-gon bac’ruli] r-iy-xo.
Mother boy bread.ABS.III-TOP ate
IV-know-PRS
c. *Enir
[uza magalu
-kin bac’ruli] b-iy-xo.
Mother boy bread.ABS.III-FOC ate
III-know-PRS
An embedded nasi N (which N) must trigger LDA. This contrasts with NDWH which
may not trigger LDA:
(21)
a. Dar
[nasi kec’
nesir
b-ati-ru-li]
b-iy-x-anu
Me.DAT which song.III.ABS him.DAT III-like-past part-NL III-know-PRSNEG
‘I don’t know which song he liked
b. *Dar
[nasi kec’
nesir
b-ati-ru-li]
r-iy-x-anu
Me.DAT which song.III.ABS him.DAT III-like-past part-NL IV-know-PRSNEG
Grewendorf (2008) observes that in German the degree of grammaticality of extraction
from wh-islands is the same for movement of DWH as for movement of topics:
(22)
a. ?[Welches Buch]i weißt du nicht, [wem du ti geben sollst]?
Which bookacc know you not
whodat you give should
b. ?[Radios]i kann ich mich nicht erinnern, [wie man ti repariert].
Radiosacc can I refl. not remember how one repairs
c. *Wasi weißt du nicht, [wem du ti geben sollst]?
Whatacc know you not
whodat you give should
Furthermore, Grohmann (2006) claims that all instances of wh-fronting in German are
instances of topic-movement. Interestingly, German does not display superiority-effects
in general. Grohmann’s claim can be compared to BoškoviΕ’s (2002) analysis of
~ 109 ~
Benjamin Kratz: Which-phrases do move
multiple-wh-fronting in Slavic, due to which only the fronting of the superior wh-phrase
is an instance of wh-fronting, all other wh-phrases moving to check their focus-features
(cf. Sabel 1998; Grewendorf 2008: 11). See Reglero (2003) for an analysis of superiority
in Basque which heavily relies on the contrast between wh-foci and wh-topics.
So the absence of superiority with DWH seems to depend on the fact that the topicpositions in the left-periphery are located higher than the focus-checking position (in
general) and that DWH can target this position(s), whereas other wh-phrases cannot.58
4.4
Supporting Evidence II: The null D-head
As has already been mentioned, the idea that which-phrases are headed by a null D-head
goes back to the semantic analysis of DWH by Rullmann & Beck (1998). Besides their
semantic reasoning, there are a number of languages that realize the D-head with DWH
overtly. The following Albanian example is taken from Kalluli (1999):
(23)
Cil-et
libra (i)
solli Ana?
Which-the books them bought Ana.
‘Which books did Ana buy?’
In Portuguese, the appearance of the overt definite determiner o in front of quê (what) is
obligatory if this receives a d-linked interpretation (as it does in the in-situ example in
(24) from Boeckx 2003):
(24)
A Maria viu *quê/o quê.
The Maria saw what/the what
‘Which thing did Maria see?’
In addition, Boeckx (2003) points to the fact that forms like Archaic Dutch hetwelk (thewhich) and Bavarian an waichan (the which-one) can as well be taken as instances of the
overt spell-out of the D-head. So, there is empirical syntactic evidence for the possibility
of the structure Rullmann & Beck (1998) proposed on semantic grounds.
And finally, I take the proposed D-head to explain the specific nature of DWH as they
have been identified by cf. Kiss (1993). In this context it is worthwhile to point to the
fact that in Enç’s (1991) influential (semantic) theory of specificity, the definition of
specificity is exactly the one Pesetsky (1987) gave for d-linking (as Enç himself observes
in her footnote 8). This connection of d-linking and specificity gains more plausibility in
light of the following facts: (i) the element pe in Romanian (which is obligatory with
DWH) is only preceding elements supporting a definite/specific reading (Boeckx 2003:
36); and (ii) resumptive pronouns (which are obligatory with DWH in some languages)
appear to trigger a specific reading on the antecedent (Boeckx 2003: 19, 32).
58
BoškoviΕ (2002: 360) and Boeckx & Grohmann (2004) observe a number of similarities between DWH
and (long-distance) scrambled phrases (e.g. lack of superiority-effects, insensitivity to weak islands,
obligatory reconstruction, triggering of clitic-doubling). If Grewendorf (2005) is right and scrambling is not
an instance of optional (stylistic) reordering of constituents but due to the checking of discourse-related
features like topic and focus, this correlation is expected.
~ 110 ~
Benjamin Kratz: Which-phrases do move
4.5
How my analysis can account for the data in
Craenenbroeck 2008
4.5.1 Doubly filled COMP phenomena in Frisian and dialectal Dutch
Van
Which-phrases have to move to TopP and cannot stay in FocP. Therefore, they can only
licence of (being the complementizer in the head-position of TopP). The fact that wie is
also able to move to TopP can easily explained by the possibility of wh-elements like who
to receive a d-linked interpretation, i.e. to act as syntactic topics (cf. Grewendorf 2008,
among others).
4.5.2 Swiping in English
As has already been mentioned in section 3.2, the exclusion of which-phrases in swiping
seems to depend on the more complex structure of which compared to bare wh-items like
who. The internal structure of which renders this element an XP and prevents it from
head-moving to P0 (and incorporating into the preposition). This ban can be accounted for
by something like the ‘structure-preservation principle’: an XP cannot move to an X0
position (cf. Radford 2004).
4.5.3 Wh-copying in German
Although I do not have an account for this construction (at least none which is relevant
for the topic of this chapter), I argued in section 3.2 that wh-copying cannot be used to
decide whether an analysis of which-phrases is more appropriate than another. Because of
this, I do not take the lack of an analysis on my part as a possible counter-argument
against my view on the derivation of which-phrases (especially that they are first-merged
inside the VP).
4.5.4 Preposition-stranding in Dutch
If one is not reluctant to expand van Riemsdijk’s (1978) generalization about the elements
that can strand a preposition in Dutch to which-phrases, there seems to be no problem for
an analysis along the lines sketched in the previous section. I do not think the inclusion of
DWH complicates the generalization (as VC does). At the moment, I do not have
anything to say about why simple wh-phrases are not able to strand a preposition in
Dutch.
4.5.5 Free relatives in Dutch
On the one hand, I have to admit that my analysis has nothing to say about the fact that
which (and its cognates) cannot appear in free relatives in English, German and Dutch.
But on the other hand, it does not run into the problems for VC’s analysis I mentioned.
Even if one accepts VC’s claim that CP1 is missing with free relatives, the problem how
to check the wh-feature does not arise in my analysis: Since I take CP1 to be TopP,
truncating it would not interfere with wh-feature-checking.
~ 111 ~
Benjamin Kratz: Which-phrases do move
Research on the distribution of wh-items as relative-pronouns in German and English
may shed light on this problem: In German, wer (who) cannot introduce a relative clause,
but welch- (which) can. Even the use of was (what), being a genuine wh-operator
(Grewendorf 2008) is restricted in colloquial German and not allowed in more formal
registers. In English, who can appear in headed relative clauses but what cannot.
Explaining these differences seems to me to be the first step to explaining why which
cannot appear in free relatives.
4.5.6 Spading in dialectal Dutch
As far as I can see, there is a simpler way to account for the contrast in (16). If the
demonstrative is base-generated together with the wh-element, one could exclude the
demonstrative with DWH by claiming that the internal structure of which-phrases does
not include a position for the demonstrative: the lexical element which obligatorily takes a
noun complement (in case of a bare which, this receives a null spell-out) that cannot take
the demonstrative as a complement itself. Therefore, the demonstrative in spading is
ungrammatical with DWH.
5
Conclusion
The analysis of which-phrases I proposed in the preceding section does not face the
conceptual problems analysis A1 faces. In particular, my analysis: (i) is compatible with
standard assumptions about argument selection; (ii) has no problem with reconstruction;
(iii) can explain the data discussed in section 3 in accordance with standard assumptions
on wh-movement (i.e. it does not rely on assumptions like ‘which-phrases never move’).
By connecting van Craenenbroeck’s (2008) claim (2) with the insights gained in
the already existing work on which-phrases (e.g. the topic character of DWH and the
specific interpretation of which-phrases), the analysis supported in section 4 is potentially
able to account for other phenomena associated with DWH: (i) the lack of superiority
effects (wh vs. focus vs. topic); (ii) the licensing of resumptive pronouns (D-head); and
(iii) possibly the insensitivity to weak islands (topic-status and D-head).
Since this is work in progress, there are naturally some issues to be more
thoroughly investigated. Besides the problems for my analysis I have identified in section
4.4., questions to be asked (and hopefully answered) in my future research include:
(a) Can [TOP] and [FOC] appear on the same element/in the same phrase?
Although I have argued for a positive answer to this question on syntactic grounds,
semantic and information-structural considerations have to be evaluated, too.59
(b) What is the role of the operator in wh-questions and (how) is it encoded syntactically?
59
See also Krifka & Féry (2009) for additional arguments supporting the claim that focus and topic are not
mutually exclusive.
~ 112 ~
Benjamin Kratz: Which-phrases do move
The discussion so far should have made clear that the biggest puzzle concerns the
operator-status of which-phrases and how to implement differences to other wh-phrases in
a theory. This is tightly connected to the question whether the bearer of operatorproperties are lexical elements or syntactic positions. Maybe both is correct, but then the
question discussed in note 13 (is there an operator-feature?) gains even more prominence.
(c) What is the connection between the focus-feature and the wh-feature?
As has been said already, it is often assumed that the wh-feature is often accompanied by
a focus-feature. On the other hand, they are definitely different items. If DWH are
endowed with a topic-feature, do they also really bear a focus-feature and what are the
details of the interaction of both (or all three)? I just roughly sketched the way my
analysis can account for the obviation of superiority with DWH, but surely, for my
account to be gaining explanatory power, the intricate interplay of the features under
discussion in different languages has to be explored in more detail.
REFERENCES
Aoun, Joseph & Yen-Hui Audrey Li. 2003. Essays on the representational and
derivational nature of grammar: the diversity of wh-constructions. Cambridge:
MIT-Press.
Benincà, Paola & Cecilia Poletto. 2004. “Topic, Focus and V2: Defining the CP
Sublayers”. In: Rizzi, Luigi (ed.). The structure of CP and IP. Oxford: OUP. 52-75.
Boeckx, Cedric (2003). Islands and Chains - Resumption and Stranding. Amsterdam:
John Benjamins.
Boeckx, Cedric. & Kleanthes K. Grohmann. 2004. “SubMove: Towards A Unified
Account of Scrambling and D-Linking”. In: Adger et al. (eds.), Peripheries.
Dordrecht: Kluwer. 241-257.
BoškoviΕ, Željko. 1998. On the Interpretation of Multiple Questions. In: Fodor et al.
(eds.). A Celebration: Essays for Noam Chomsky’s 70th Birthday. Cambridge.
Mass.: MIT Press. 1-9.
BoškoviΕ, Željko. 2002. On Multiple Wh-Fronting. Linguistic Inquiry 33, 351-383.
Cinque, Guglielmo. 1990. Types of A’- Dependencies. Cambridge, Mass.: MIT Press.
Comorovski, Ileana. 1996. Interrogative Phrases and the Syntax-Semantics Interface.
Dordrecht: Kluwer.
Den Dikken, Marcel. 2003. “On the morphosyntax of wh-movement”. In: Boeckx, Cedric
& K. K. Grohmann (eds.). Multiple Wh-fronting. Amsterdam: Benjamins. 77-98.
Dobrovie-Sorin, Carmen. 1993. The Syntax of Romanian. Berlin: Mouton de Gruyter.
Doron, Edit. 1982. On the Syntax and Semantics of Resumptive Pronouns. Texas
Linguistic Forum 19. 1-48.
Enç, Mürvet. 1991. The Semantics of Specificity. Linguistic Inquiry 22. 1-25.
Erteschik-Shir, Nomi. 1997. The Dynamics of Focus Structure. Cambridge: CUP.
Grewendorf, Günther. 2002. Minimalistische Syntax. Tübingen: Francke-Verlag (UTB).
~ 113 ~
Benjamin Kratz: Which-phrases do move
Grewendorf, Günther. 2005. The discourse configurationality of scrambling. In: Sabel,
Joachim & M. Saito (eds.). The Free Word Order Phenomenon: Its Syntactic
Sources and Diversity. Berlin: Mouton de Gruyter. 75-135.
Grewendorf, Günther. 2008. Wh-movement as topic movement. Mskr. Universität
Frankfurt a.M.
Grohmann, Kleanthes K. 2005. Top Issues in Questions: Topics-TopicalizationTopicalizability. In: Cheng, Lisa & Norbert Corver (eds.). Wh-Movement Moving
on. Cambridge: MIT Press. 249-288.
Hagstrom, Paul. 1998. Decomposing Questions. Doctoral dissertation. Massachusetts
Institute of Technology. Cambridge.
Horvath, Julia. 1986. Focus in the Theory of Grammar and the Syntax of Hungarian.
Dordrecht: Foris.
Kalluli, Dalina. 1999. The comparative syntax of Albanian: on the distribution of
syntactic types to propositional interpretation. Doctoral dissertation. University of
Durham.
Kiss, Katalin. 1993. Wh-movement and Specificity. Natural language and Linguistic
Theory. 11. 85-120.
Krifka, Manfred & Caroline Féry. 2009. Information Structure: Notional Distinctions,
Ways of Expression. Proceedings of the 18. International Conference of
Linguistics. Seoul.
Merchant, Jason. 2001. The syntax of silence. Sluicing, islands and the theory of ellipsis.
Oxford: OUP.
Merchant, Jason. 2002. Swiping in Germanic. In: Zwart, C. J.-W & W. Abraham (eds.).
Studies in Comparative Germanic Syntax. Proceedings of the 15th Workshop on
Comparative Germanic Syntax. Amsterdam: John Benjamins. 289-315.
Pesetsky, David. 1987. Wh-in-situ: Movement and Unselective Binding. In: Reuland &
ter Meulen (eds.). The Representation of (In)definiteness. Cambridge, Mass.: CUT.
98-129.
Polinsky, Maria 2001. Information Structure and Syntax: Topic, Discourse-linking, and
Agreement. Mskr. University of California.
Radford, Andrew. 2004. Minimalist Syntax. Exploring the structure of English.
Cambridge: CUP.
Reglero, Lara. 2003. Non-wh-fronting in Basque. In: Boeckx, Cedric & Kleanthes K.
Grohmann (eds.). Multiple wh-fronting. Amsterdam: John Benjamins. 187-227.
Richards, Norvin. 2001. Movement in Language. Oxford: OUP.
Rizzi, Luigi. 2000. Reconstruction, Weak Island Sensitivity, and Agreement, Mskr.
University of Siena.
Rullmann, Hotze & Sigrid Beck. 1998. Presupposition Projection and the Interpretation of
which-Questions. Proceedings from Semantics and Linguistic Theory 8. Ithaca:
Cornell University. 215-232.
Sabel, Joachim. 1998. Principles and Parameters of Wh-Movement. Habilitationsschrift,
Universität Frankfurt a.M.
Šimík, Radek. 2007. “The source of wh-morphology in questions and relative clauses”.
In: Blaho, S. et al. (eds.). Proceedings of ConSOLE XV. 1-22.
~ 114 ~
Benjamin Kratz: Which-phrases do move
Sharvit, Yael. 1999. “Resumptive Pronouns in Relative Clauses”. Natural Language and
Linguistic Theory 17. 587-612.
Van Craenenbroeck, Jeroen. 2008. Complex wh-phrases don’t move. Mskr.
CatholicUniversity of Brussels.
Van Riemsdijk, Henk. 1978. On the diagnosis of wh-movement. In: S. J. Keyser (ed.).
Transformational studies in European languages. Cambridge: MIT-Press. 189-206.
Wu, Jianxin. 1996. Wh-topics, Wh-focus and Wh-in-situ. In: University of Maryland
Working Papers in Linguistics 4. 173-192.
~ 115 ~
CHAPTER 8
ADULT ROOT INFINITIVES
NEVEN WENGER
University of Frankfurt
Adult Root Infinitives (ARIs) are a variety of infinitival structures occurring in root (i.e. main
clause/independent) contexts. A central feature that distinguishes ARIs from other root infinitives
(RIs, e.g. infinitival wh-questions) is their pragmatic meaning: a speaker’s incredulity towards the
proposition of a previous utterance. Other notable aspects are the cross-linguistic availability of
ARIs (an Indo-European phenomenon), and the morphosyntactic variation across these languages
(e.g. variable Subject Case, or an optional coordinator). As an apparently impoverished and
idiosyncratic syntagm, ARIs have often been put forth to argue against generativist-derivational
theories of syntax (e.g. Fillmore et al. 1988, Lambrecht 1990). Taking sides with Etxepare &
Grohmann (2002 et seq.), I argue that the ARI can well be subsumed under a generalised
derivational principle like those of Minimalist Syntax (cf. Chomsky 1995 et seq.). The more
general discussion of various aspects of the ARI is followed by a sketch of a syntax of
(non)finiteness formulated within a phase-based minimalist framework.
1 INTRODUCTION
This paper deals with a syntactic construction referred to as Adult Root Infinitive (ARI; cf.
Etxepare & Grohmann 2002 et seq.). Section 2 is a survey of the grammatical properties
of the ARI, mainly a condensed version of Wenger (2008). It touches on the discoursepragmatics and the prosodic characteristics of the ARI, as wells as on its relation to child
language (2.1.). In addition, three morphosyntactic properties are explored: the restriction
to embedded domains, cross-linguistic variation/distribution, and Subject Case (2.2.). The
subsection on Subject Case (2.2.3) anticipates the syntactic analysis in section 3
(formulated in the theoretical framework of phase-based minimalism; cf. Chomsky 2001
et seq.), where it will be shown that the key to understanding the syntax of ARIs lies in
the concept of nonfiniteness and its epiphenomena (Subject Case60, verb inflection,
syntactic dependence). It concludes that ARIs most likely constitute Tense phrases
(TP/IP).
60
Technical terms in caps (e.g. ‘Case’) refer to the generative definition of a given notion. ‘Case’, e.g., is
structural Case, i.e. case assigned to a nominal by another element in its syntactic environment (e.g.,
Nominative by T(ense)).
~ 116 ~
Neven Wenger: Adult root infinitives
2 GRAMMATICAL PROPERTIES
2.1 Discourse-pragmatics,
variation
prosody,
semantics
&
ontogenetic
As shown in the following mini-discourse (speaker A – speaker B), the illocutionary force
of the ARI is to express incredulity (orthographically marked by ‘!?’) towards a
previously uttered proposition (within a reference turn; cf. Bücker 2008; Incredulity
Response Construction in Lambrecht 1990). Typically, it is preceded by another, simplex
interjection of incredulity (What!?), and/or followed by a dissentive expression
confirming the incredulity raised by the preceding ARI.
(1) A: I heard Quagmire’s preparing a paper for the forthcoming MLC18
proceedings…– B: What!? Him prepare a paper!? No way, dude! He’s got other
things on his mind…
The incredulity expressed by the ARI also correlates with a distinct prosodic structure:
the Subject obligatorily bears focal stress (represented by caps), and the intonation
contour of ARIs is final-rising (a global rise, typical of open interrogatives; represented
by ‘[↑]’)61.
(2)
HIM prepare a paper [↑]!?
Another property of the ARI, reflected in its attribute adult, is that it is restricted to adult
registers. This is an important qualification since root infinitives (RIs) are far from
uncommon in child language – indeed, at least during one stage of language acquisition,
Child Root Infinitives (CRIs) make up the majority of utterances (cf. Rizzi 1993; (3) from
Radford 1990).
(3)
a. Baby eat cookies.
b. The baby eat cookies!?
– CRI
– ARI
However, despite the formal resemblance of CRIs to ARIs, they differ in one crucial
respect: CRIs may in principle have one of a whole range of aspectual, temporal, and
modal, as well as illocutionary meanings – a degree of polysemy (or, grammatical
underspecification) that presumably needs (post-syntactic) disambiguation based on
contextual information (cf. Avrutin’s 1999 approach to RIs). Accordingly, (3) may be
variably translated into adult English as ‘The baby eats/is eating/ate/should eat/…
cookies’. ARIs, on the other hand, are rather specialised in that they can only carry what
61
There seems to exist a prosodic variant with falling intonation (orthographically marked by ‘...’ instead of
‘!?’) and unfocussed Subject: Yeah, yeah… Him like books...Whatever you say.... In this case, however, the
ARI expresses non-genuine incredulity. The syntactic (pragmatics notwithstanding) analysis presented in
sections 3 should carry over to it.
~ 117 ~
Neven Wenger: Adult root infinitives
might be dubbed incredulitive force (as discussed above). Interestingly, the functional
domains of CRIs and ARIs seem to be mutually exclusive: CRIs can express almost any
of the aforementioned meanings, with the exception of incredulity, to which, in turn,
ARIs are restricted. This opposition is also reflected in the relative frequency of ARIs vs.
CRIs in natural discourse: While CRIs – as pointed out above – are an omnipresent
phenomenon of child language, ARIs seem to occur considerably less frequently in adult
speech62. Finally, in addition to its special incredulitive force, there is another feature that
distinguishes ARIs from CRIs, which is the irrealis semantics typical of infinitives: they
denote situations/propositions evaluated w.r.t. a non-actual (‘unreal’, hence irrealis)
world. CRIs, on the other hand, may be realis in that they can be used to assert63.
With this rough outline of the prosodic and discourse-pragmatic properties of ARIs,
including a brief comparison of ARIs with CRIs, I now turn to the morphosyntactic
characterisation of ARIs.
2.2 Morphosyntax
2.2.1 Restriction to root domain
The attribute root of the ARI refers to the fact that ARIs occur as independent main
clauses only (clauses forming the root (domain) of an upside-down syntactic tree graph).
This is noteworthy insofar as nonfinite structures (infinitives, gerunds, participles, etc.)
appear to be restricted to dependent/subordinate contexts, as the following examples of
nonfinite complements show (which yield ungrammatical sentences standing alone, e.g.
My thesis finished; cf. (d)):
(4)
a. I’ll make [him like books].
b. I want [him to like books].
c. I saw [him reading].
d. I consider [my thesis finished].
– CAUSATIVE (BARE INFINITIVE)
– VOLITIONAL (TO-INFINITIVE)
-- PERCEPTIVE (PRESENT PARTICIPLE)
– ‘SMALL CLAUSE’ (PAST PARTICIPLE)
ARIs, on the other hand, not only are restricted to root contexts, but moreover, they
cannot be embedded at all (under equivalent – i.e. incredulitive, or dubitative –
predicates). They are a root phenomenon64.
(5)
*I doubt/wonder/don’t believe [him like books].
62
ARIs seem to be restricted to oral speech (cf. Bücker 2007, 2008 for a number of attested German ARIs
from internet newsgroups).
63
There is a heated debate on the semantics of CRS in terms of their modality. Some authors maintain that
CRIs do exclusively express irrealis modality (cf. Hoekstra & Hyams 1998).
64
There exist yet other types of RIs, such as infinitival wh-interrogatives (cf. Reis 2003 for German):
Warum nur Linguistik studieren? ‘Why (only) study linguistics?’. Ultimately, these (and other) RIs should
share a derivational commonality, the additional domain of variance having to do with illocutionary
meaning.
~ 118 ~
Neven Wenger: Adult root infinitives
While the restriction of ARIs to root domains is already exceptional within a languagespecific system (like English), even more interestingly, this oddity seems to be available
to numerous Indo-European languages, as sketched in the following typological section65.
2.2.2 Cross-linguistic distribution
As shown by Etxepare and Grohmann (2002 et seq.) and Bücker (2008:7), the ARI is no
phenomenon restricted to central European languages66. For the following languages the
availability of the ARI is attested: Germanic (German, Dutch, English, Swedish,
Norwegian), Romance (French, Spanish, Portuguese, Galician, Catalan, Italian), Slavic
(Polish, Russian, Latvian), Uralic (Hungarian)67.
Although the ARIs of different languages seem to share a set of core characteristics, they
do differ along specific dimensions. As Grohmann and Etxepare (2003) and Etxepare and
Grohmann (2007) show, for instance, variation exists in the compatibility with
quantificational Subjects, the permissibility of discourse-fronted material (topics, etc.), or
the availability of temporal-aspectual modification (e.g. deictic adverbials, perfect
auxiliaries).
Instead of elaborating on the typological dimension (cf. Wenger 2008 for more
discussion), I would like to take one step back in what follows, and examine the ARI
from a more basic angle, discussing whether the ARI constitutes a unified, integrated
syntagm that can be subjected to syntactic analysis at all.
2.2.3 Subject Case
The common characterisation of the morphosyntax of infinitives (and of nonfinite
structures in general) includes the absence of Subject-verb agreement (SVA), of verbal
tense inflection, and of overt nominative (NOM) Subjects. Evidently, the latter criterion is
only partially met in ARIs, which do have overt Subjects, which in English, however,
surface in the accusative (ACC) form (visible if pronominal)68:
(6)
Him[ACC]/*He[NOM] prepare a paper!?
As in generative theory it is required for overt Subjects (all DPs more generally) to be
65
Potts & Roeper (2006) offer a pragmatic/semantic explanation for the unembeddability of ARIs.
A more general problem in this respect is posed by the imprecise semantic and morphosyntactic
definition of the notion of finiteness (cf. Nikolaeva 2007 for an overview and different views; cf. Wenger
2009 for my own point of view).
67
My own empirical research has produced two other, Slavic languages that possess ARIs: Bulgarian and
Croatian. Thanks to my informants Iwo Iwanov and Iva Riekert-Wenger.
68
This seems to be also true of French: Lui[ACC]/?*Il[NOM] aimer les livres!? ‘Him like books!?’.
66
~ 119 ~
Neven Wenger: Adult root infinitives
Case-marked (this intuition goes back to at least Vergnaud's (1977 [2008]) Case Filter),
the central question with regard to the ARI Subject is where its Case stems from.
Structural Cases, i.e. nominative and accusative, are considered to be assigned in
particular configurations, by particular categories: Thus, while ACC is typically assigned
to direct Objects by transitive verbs, NOM (a.k.a. Subject Case) has typically been
correlated with ‘finite’ I (1980s), later on ‘finite’ T (1990s) – ultimately with a [+Tense]
category (cf. Chomsky 1981:19). A Subject agreement feature AgrS (i.e. SVA), either
contained within I (Chomsky 1981), or projecting independently (AgrSP > TP; cf. Pollock
1989; Belletti 1990), has always been considered a licensor for [NOM] as well (as opposed
to [+Tense]), at least since George & Kornfilt (1981). The correlation between AgrS and
NOM re-entered the latest revision of generative theory – minimalist syntax – as [NOM]
being a post-syntactic reflex of SVA (now complete [uφ] on T instead of AgrS; cf.
Chomsky 2001:6, 16). Whatever the licensor of [NOM], it has rarely been conceived of as
primitive, but always as contingent on another morphosyntactic property.
Provided such a theory of Case, ARIs raise the question of what licenses their [ACC]
Subjects – it cannot be a (higher) predicate (as in ECM-clauses; cf. (4)), nor, obviously, a
[+finite]-feature (be it [+Tense], or SVA). Besides structural Case (and inherent Case, i.e.
dative/oblique), there exists yet another ‘type of Case’, which seems to be operative in
ARIs: default Case (cf. Schütze 1997). In certain syntactic environments, where no
structural Case-assigner is available (V[+trans], I[+fin]), some sort of default
mechanism69 kicks in, which provides an unvalued Case feature with a default value, thus
‘rescuing’ a derivation (cf. the Case Filter mentioned above: ‘overtness requires Case’).
This default Case appears to be identical with the unmarked nominative Case in most of
the relevant languages (e.g. German); English, however, produces [ACC] Subject Case in
a number of environments where an explanation based on default Case seems reasonable,
given that there is no other possible Case-assigner (finite T, prepositional C(OMP),
matrix transitive predicate):
(7)
a. Him/*He/*His sleeping, I went out alone.
b. A: Who is it? – B: It’s me/*I/*my.
c. Him/*He/*His, he digs cheese cake.
– PARTICIPLE CLAUSE
– FOCUS POSITION
– LEFT DISLOCATION
What these nonfinite domains share is the lack of a Case-assigner, which is why the
default mechanism kicks in, providing the Subjects with the English default Case [ACC]
(and not, say, [GEN])70.
To conclude this subsection, I would like to briefly compare two alternative approaches to
the Subject Case of ARIs with the superior (viz. more economical) default strategy. The
prosodic nature of the ARI Subject, i.e. their being obligatorily focussed, is one obvious
69
This mechanism, which provides nominals with a default Case (and operates more generally in terms of a
default grammar; e.g. infinitival morphology might be another case at hand), is construed as a postsyntactic, morphological process, as implemented in Distributed Morphology (cf. Schütze 1997 for details).
70
Visser (1963:237ff.) identifies [NOM] as the default Case for older stages of English.
~ 120 ~
Neven Wenger: Adult root infinitives
candidate for the source of the [ACC] Case, the idea being that focus yields some kind of a
strong pronominal form, on the assumption that the English [NOM] pronouns are too
weak, in terms of their lexical-phonological complexity, to bear focal stress. Whatever the
pronominal status of English pronouns (strong, weak, clitic; cf. Cardinaletti & Starke
1999 for an overview), it is rather easy to demonstrate that focussing a pronoun does not
necessarily yield [ACC] Case (indeed, excludes it), which would be unexpected given the
obligatoriness of [ACC] in ARIs:
(8)
a. HE is the double agent (, not Herbert)!
b. ?*HIM is the double agent (, not Herbert)!
Here, the Subject pronoun bears a contrastive focus (as indicated by the bracketed
Herbert, one member of the contrast set), but receives [NOM] rather than [ACC] Case,
which actually yields at least a deviant sentence (marked ‘?*’), if not ungrammatical.
Although the focus of ARI Subjects seems to be of a different nature than contrastive
focus (incredulity, rather than contrast), it is reasonable to dissociate focus assignment
from Subject Case licensing, given the heterogeneous distribution of both categories (also
cf. (7) above).
The second alternative explanation is related to one class of nonfinite structures licensing
[ACC] Subject Case, where in some cases an overt, non-verbal Case-assigning element is
available (prepositional COMPs; (a, b)), in some cases none at all (c, d):
(9)
a. I want [for him to read less Chomsky].
b. [With him gone to bed], the party started.
c. [Him distracting the cat], I was able to grab its tail.
d. [Him kissing the goldfish] is a disturbing image.
– FOR…TO-INFINITVAL
– ‘SMALL CLAUSE’
– PARTICIPIAL CLAUSE
– CLAUSAL GERUND
These clauses all have in common that there is no finite predicate assigning the [ACC]
Case to their Subjects, as in ‘ECM’-infinitivals (e.g. I want him to read less Chomsky).
However, the nonfinite clauses in (a, b) are headed by prepositional COMPs (for and
with), which might act as Case-assigners71. Now, one methodological cornerstone of
generative theory being maximal reduction of various phenomena to common sources
(i.e. generalisation), one might envision an analogical extension of (a, b) to (c,d) qua null
COMPs (something that is done e.g. in Radford (2009), possibly for textbook didactics).
Thus, while as for (a, b) it is the nonfinite COMPs for/with in C that assign [ACC] to the
Subject ‘under government’, this is achieved by the null COMP (‘’) in (c,d) (provided
the tripartition of the clausal spine into CP > IP > vP (v = transitive V); cf. Carnie 2008:
ch. 11 for a concise but illuminating overview):
(10) a. I want [CP [C° for][ACC] [IP him[ACC] [I° to] [vP read less Chomsky] ] ].
71
Indeed, it is uncontroversial that their properly prepositional counterparts (and possibly their diachronic
sources) do assign [ACC].
~ 121 ~
Neven Wenger: Adult root infinitives
b. [CP [C° ][ACC] [IP Him[ACC] I° [vP kissing the goldfish] ] ] is a disturbing
image.
One domain of inquiry central to generative theory are infinitival structures such as
Control, raising, and ‘ECM’-infinitivals. What is essential here is that these are all
dependent on (= selected or licensed by) a higher predicate. Accordingly, the source of
the [ACC] Case of Subjects in nonfinite argument clauses (= Complements) has been
ascribed to the dependency between matrix predicate and embedded Subject, something
that is evidently not an option with the relevant examples above, which are Adjunct (=
‘non-selected/-licensed’, thus omissible) clauses. Importantly, the study of nonfinite
Complementation (cf. BoškoviΕ 1997 for an elaborate survey) has produced an
asymmetry w.r.t. the structural complexity of Control and COMP-infinitivals on the one
hand, and raising and ‘ECM’-infinitvals on the other: While the former are considered
CPs (cf. Davies & Dubinsky 2004 for Control infinitivals), the latter are considered
truncated (= reduced) to IP, essentially for the lack of any evidence to the contrary. This
structural asymmetry is reflected in the different mobility of e.g. Control vs. ‘ECM’infinitvals:
(11) a. [CP PROk To read less Chomsky]i is what Ik promise ti.
b. *[IP Him to read less Chomsky]i is what I want ti.
To sum up, only CPs can be moved (pseudo-clefted, topicalised), passivised, phonetically
isolated, etc., but IPs cannot. While these contrasts may prove valid, it is far from clear
that they are reducible to the contrast in structural complexity – the CP vs. IP opposition
retains an ad hoc flavour. What is worse is that these differences (or at least some of
them) do not seem to extend easily to the Adjunct clause given in (9c, d) above. While
their similarity to ‘ECM’-Complements (no overt COMP, [ACC] Subject Case) suggests a
treatment as IPs, they are not subject to the same mobility constraints72:
(12) a. It is [?P him kissing the goldfish] that is a disturbing image. – CLEFT
b. What is a disturbing image is [?P him kissing the goldfish]. –
PSEUDOCLEFT
Thus, the categorical status (CP vs. IP) of the clausal gerund in (12) remains unclear (?P).
In any case, attractive as a generalisation of superficially different nonfinite structures to a
uniform structural representation might be, I think it misses a central desideratum of
minimalist theory: economy. All that e.g. a null COMP theory of the nonfinite Adjuncts
discussed above does is extend the explanation of one phenomenon (i.e. Case-assigning
prepositional COMPs) to another one (i.e. COMP-less nonfinite clauses with [ACC]
Subjects), while there is an equally adequate, but, importantly, more economical solution
readily available, not appealing to an ad hoc stipulation like null COMPs: default Case.
72
Granted, copular sentences, and clefts, and pseudo-clefts in particular, are a difficult case in their own
right. The asymmetry in grammaticality w.r.t. mobility between the Adjunct clause in (12), and the ‘ECM’clause in (11b), however, remains a fact.
~ 122 ~
Neven W
Wenger: Adult root
r
infinitives
Theerefore, insttead of assu
uming nulll Case assig
gners wherrever a nonn-canonicallly Casemarrked Subjecct occurs in a nonfinitee structure, I maintain the
t more ecconomical approach
a
baseed on defauult Case assiignment (prrepositional COMPS no
otwithstandding).
Agaainst this baackground, I would likee to discuss in the follo
owing sectioon evidencee for and
agaiinst the struuctural com
mplexity of ARIs. Morre concretelly, based onn the tripartition of
the clause into CP > IP > vP (where eeach phrasee is likely to
o stand for a more fine-grained
clauuse structuree along the lines of Caartographic approachess to phrase--structure; cf.
c Rizzi
19997, among many
m
others)), boils dow
wn to the qu
uestion: Are ARIs CPs, IPs, or vPs?
2.3
3
The phrase-strructural c
complexitty of ARIs
s
2.3
3.1 Minima
alist synta
ax
In thhis section, I give a co
oncise accouunt of the essentials off the most reecent frameework of
generative synttax.
(
aaround Cho
omsky’s
Synntactic theoory in the (mainstreaam) generattive vein (evolving
worrks), which in its latest incarnationn (going bacck to Chomsky 1993) m
might be refferred to
as m
minimalist syntax, asssumes a phhrase-structtural skeletton for eacch clause (reduced
(
clauuses being excluded
e
forr now) conssisting of th
hree domain
ns nested wiithin each other like
in a Matryoshkka doll: CP
P > IP > vP
P. The loweest, verbal domain
d
vP is where arrgument
struucture is licensed in a uniform faashion, as th
hematic rolles (AGENT, PATIENT, etc.)
e
are
concerned (thhe so-calleed Universsal Theta-A
Assignment Hypothessis (UTAH
H), first
form
mulated in Baker
B
1988
8). Thus, AG
GENT nomin
nals are alw
ways licenseed in the Sp
pec(ifier)
of vvP as an extternal argum
ment, and PA
ATIENTs (a..k.a. THEME
Es in generaative vocabu
ulary) as
the sister (i.e. thhe Complem
ment) of vP
P as internal arguments..
(13)
What is particuularly intereesting in thiis respect iss a contrast in thematicc structure deriving
from
m the split of intransitive verbs into two classes: veerbs that haave a PATIE
ENT-like
Subbject (unaccusatives, e.g
g. fall), andd those that have an AG
GENT-like Su
ubject (unerrgatives,
e.g. dance). Provided thee UTAH iss valid, unaaccusatives, lacking eexternal arg
guments,
wouuld evidentlly fail to yield
y
the acctually obseerved surfacce word ord
rder of intraansitiveunaccusative sentences
s
iff their PATTIENT argum
ment were spelt out iin situ, i.e. as the
Com
mplement of
o v. Ratherr, they wouuld producee the ungraammatical **Fell the guillotine
instead of the correct
c
The guillotine ffell. This lin
ne of argum
mentation is the base off the VPinteernal Subjecct Hypotheesis (VPISH
H), which captures
c
on
ne aspect a principle like the
~ 123 ~
Neven
N
Wengeer: Adult root infinitives
UTAH entails – thhat the notio
on of Subjecct is a deriv
vative one. Thus,
T
it is aassumed thaat the
Subjectt-to-be arguument (be itt AGENT, PA
ATIENT, etc..), i.e. the highest
h
nom
minal within
n the
vP, mooves from its
i vP-interrnal positioon to (the Spec of) a higher, vPP-external one,
becominng the gram
mmatical Su
ubject. The Subject po
osition is geenerally asssumed to bee the
Spec off IP (or off another I-rrelated heaad – or of an
a even hig
gher one, uunder particcular
circumsstances, as proposed
p
in
n more recennt works; cff. e.g. Rizzi & Shlonskyy 2006, 200
07).
(114)
The infl
flectional heead I of the clausal dom
main superseding vP no
ot only licennses Subjeccts in
its Specc (implemennted as a feeature [EPP] scanning the
t domain in its scopee for a nom
minal
and atttracting it),, but also hosts Tennse, Aspectt, and Modality/Moood features (all
ultimateely surfacinng as verbaal inflectionn), as well as an Agreeement (moore preciselly, a
SVA, oor, in more recent
r
generative vocab
abulary, a φ--feature com
mprising PEERSON, NUMBER,
and GEN
NDER), and a feature re
elated to Suubject Casee, i.e. [NOM]73. The rouugh idea behind
this feaature accum
mulation aro
ound I goess as follow
ws: If finiteness can bbe equated with
morphoological Tennse (but cf.. the discusssion below
w), which iss licensed bby I, then SVA
S
must bee licensed around
a
I as well
w since iit only surfaaces in finitte environm
ments. The same
s
holds foor [NOM], which
w
is thee typical Suubject Case in finite contexts (defa
fault [ACC] Case
C
being thhe nonfinitee counterparrt; cf. abovee). If I contains no Ten
nse feature, however, i.e. if
it is nonfinite, it still
s seems to
t be able tto attract a Subject to its Spec poosition – ass the
precedinng discussion of e.g. ECM-infinit
E
tivals has sh
hown –, thee differencee with regarrd to
finite I bbeing the laack of Tensee, SVA, andd [NOM].
With thhis rough ouutline of the principless behind the derivation
n of a canoonical indicaative
sentencce (considerred the unm
marked casse, underly
ying other, more com
mploex sentence
types), I am now inn the positio
on to discusss the structtural compllexity of AR
RIs. The secction
wing will – as announcced above – take the discussion
d
oof the syntaax of
immediiately follow
ARIs one step back,
b
and address
a
wheether ARIs qualify as ‘clauses’, i.e. as unified,
73
In whhat follows, these
t
aspects of the I-dom
main are heaavily simplifieed for presenntational purp
poses.
Actually,, the relation between
b
Tensse, SVA, [NOM
M], and finiten
ness have been undergoingg constant reviisions
at least siince Chomskyy (1981).
~ 124 ~
Neven Wenger: Adult root infinitives
integrated syntactic structures, at all. This will be followed by a discussion of ARIs as
structures of different, increasing complexity, ranging from vP to CP.
2.3.2 ARIs as clauses
While, intuitively, the ARI might come across as a clause proper, on closer inspection,
from a narrowly syntactic perspective, things are not as straightforward. A number of its
characteristics, raise the question of whether ARIs are really mono-clausal, i.e. one
syntactically unified phrase (vP, IP, CP), or rather bi-phrasal, i.e. two syntactically
distinct phrasal chunks.
The first controversial aspect is the intonational (prosodic) structure of ARIs. Speaker
judgements differ as to whether ARIs form one intonation phrase (IntP) (a), or rather two
distinct ones (b):
(15) a. [IntP HIM like books]!?
b. [IntP HIM]!? [IntP Like books]!?
– UNIFIED INTONATION CONTOUR
– COMMA INTONATION
Thus, a prosodic structure like (a), representing a unified intonation contour, would
support a mono-clausal analysis of ARIs, while the comma intonation in (b) might favour
a bi-phrasal one (with connectivity possibly established post-syntactically). Solely on
prosodic grounds, the question cannot be easily settled. Thus, the most careful
observation to be made would be that prosodic variation exists with ARIs, with a unified
intonation contour and an interrupted one (comma intonation) coexisting. However, this
entails the question of whether we are dealing with two distinct syntactic phenomena, or
with only one, which is prosodically variable. While this might be hard to answer, there is
evidence that at least the prosodic subset represented in (a) also constitutes a unified
syntactic domain.
A classic diagnostic for clausal domains is binding theory (cf. Chomsky 1981), in
particular the coreferential binding of reflexives like himself, etc. (anaphors in generative
terminology; coreferentiality indicated by co-indices), which is captured by the binding
condition A (here, a simplified version by Büring 2005:55; for the original formulation,
cf. Chomsky 1981:188): “A reflexive must be bound within the smallest category [α]
containing it, its case assigner, and a Subject”. Put even more simply, a reflexive and its
antecedent must be clause-mates (clause = domain/smallest category α). Otherwise, a
clause containing a reflexive would not be well-formed, as shown by the following
example (16), where the reflexive himself does not find an appropriate antecedent within
α (requiring the personal pronoun him):
(16) Courtneyi and Kurtk have just married. But hek knew [α that shei didn’t love
himk/*himselfk].
Interestingly, reflexives do not pose any problem in ARIs – of course provided their
~ 125 ~
Neven Wenger: Adult root infinitives
antecedent is within the same domain:
(17) I’ve just heard that Kurt committed suicide… – What!? [α Kurti kill
himselfi/*himi]!? This can’t be true!
If the ARI in (17) were represented as a bi-phrasal structure, the grammaticality with a
reflexive (himself) contained in it would be unexpected: As shown in (16), a reflexive
should not be able to occur in a ‘clause’ (α) on its own, without an antecedent. Thus, what
is deviant in the following representation is the notation indicating the syntactically
licensed coreferentiality between a reflexive (himself) and its antecedent (Kurt), which is
at odds (marked by a starred index) with the insights gained from Binding Theory (viz.
that both must share a domain). This would be a wrong binding-theoretic prediction since
the ARI per se is of course well-formed containing a reflexive:
(18) [α Kurti]!? [α Kill himself*i]!?
While the preceding argument from binding-theoretic considerations relies on a
negative/indirect line of argumentation, another, more clearly morphosyntactic one comes
from the phenomenon of inflected infinitives. As laid out above, ARIs are available in a
whole range of languages (language families), among them Portuguese. Now,
interestingly, varieties of the latter possess an agreeing infinitives, i.e. infinitives
morphologically marked for SVA (the following is an example from Brazilian Portuguese
(BP)74).
(19)
Eles
saírem
cedo de casa!? Impossível!
PRN.3PL get.out-INF-3PL early of home impossible
‘Them leave home early!? Impossible!’
Whatever the specific implementation, most theories assume that agreement (SVA being
a specific subcase) is a local dependency between two elements, i.e. one holding within
certain syntactic domains (of variable complexity), but never across sentences75.
Trivially, in order for the BP infinitive to surface inflected for φ-features (i.e. saírem
instead of saír), then, the Subject pronoun eles must share with the verb saír a unified,
local domain – one phrase (= a clause), not two distinct ones.
A third, strong piece of evidence for the mono-clausal analysis is provided by an
investigation of the left-peripheral activity in ARIs (Do they license topics, etc.?). As this
point will be taken up in 2.3.4. in more detail, I here provide only one example from
74
Thanks to Marcello Modesto for the Brazilian Portuguese data. It should be added that the agreeing
infinitive in BP might be restricted to particular registers/varieties.
75
Of course, while agreement dependencies are also formed transsententially in discourse (e.g. any personal
pronoun without a referring expression in the same clause), this would not qualify as syntactic agreement in
the narrow sense, which – contrary to discourse agreement – is subject to domain-specific locality
constraints.
~ 126 ~
Neven Wenger: Adult root infinitives
Spanish, which allows for clitic left dislocation (CLLD) in ARIs, whereby a constituent is
fronted to the left, leaving behind a resumptive clitic pronoun (example from Etxepare &
Grohmann 2005:130):
(20)
elecciones]i [ganar-[las]i ]k Schröder tk!?
the elections
win-them.CL Schröder
‘Schröder win the elections!?’ (‘#The elections, Schröder win!?’)
[Las
Given that the unmarked, assertive-futurate ‘base form’ of the sentence is Schröder
ganará las elecciones ‘Schröder will win the elections’, i.e. SVO word order, the
information-structurally modified (20) not only has its direct object las elleciones ‘the
elections’ moved to a left-peripheral topic position, but also the infinitive ganar ‘win’
moves higher than the Subject Schröder (be it in SpecvP or SpecIP). If the Spanish ARI
in (20) would really be constituted by two phrases, movement operations like CLLD –
from one phrase into another one – would clearly be unexpected, even prohibited under
the general ban on transsentential movement.
A final, somewhat conceptual, argument derives from the minimalist theory of thematic
structure (UTAH) discussed above: In short, if any v is subcategorised for a specific
number of arguments, these must be exhaustively satisfied in the course of the syntactic
derivation of any vP. If the ARI were taken to consist of two phrases – one containing the
predicate, one the Subject –, then the predicate phrase (vP) would clearly violate any
conditions formulated within such a theory given that no external argument would be
licensed (not to mention the unaccusative–unergative divide).
I take the evidence from connectivity and movement (binding, SVA, CLLD) to conclude
that at least a (prosodic?) subset of ARIs must constitute a unified clausal structure. It is
this very subset that is of interest here. What remains to be examined, then, is the
structural complexity, i.e. the categorial status, of ARIs (roughly: vPs, IPs, or CPs).
2.3.3 ARIs as vPs
Once the syntactically unified nature of ARIs has been established, the most minimal way
of analysing them would be as bare vPs. The notion of ‘bare vP’ is closely related to
theories of Small Clauses (SCs; cf. Williams 1975), which are verbless predicational
structures like the following:
(21) I consider [?P Fritz a moron]. (‘I think Fritz is a moron.’)
In principle, there are three options to deal with SCs syntactically in order to determine
their categorial status: (i) as exocentric structures labelled SC; (ii) as endocentric
~ 127 ~
Neven Wenger: Adult root infinitives
structures headed by the predicate (A, P, D76, or v/V); or (iii), as vPs headed by a covert
predicator similar to be (vBE, but not A, P, or D).
(22) a. I consider [SC [NP Fritz] [DP a moron] ].
b. I consider [DP [NP Fritz] [D° a] [NP moron] ].
c. I consider [vP [NP Fritz] v°BE [DP a moron] ].
Interestingly, a corpus-based, empirical perspective shows that most tokens of the ARI –
as rarely as it may occur compared to other syntactic ‘constructions’ – instantiate a
verbless, SC-like structure, most often an adjectival one (the following German one is
attested):
(23)
schwanger!?
PRN.1SG pregnant
‘Me pregnant!?’
Ich
Disregarding option (i), one might attempt to subsume both verbless and verbal ARIs
under a SC-analysis, be it along the lines of (ii), or (iii). Indeed, this is assumed by
Progovac (2006) in her survey of various nonsentential structures, among them the ARI,
which she refers to as verbal Root Small Clause (VRSC), i.e. a main clause headed by
v/V (= vP). However, beyond these basic observations, there exists evidence for a more
complex syntactic structure for ARIs that cannot be discarded.
2.3.4 ARIs as IPs
No matter how one would syntactically represent an adjectival SC as that in (23) (as SC,
AP, or vP), the asymmetry between the thematic structure and the surface syntax of
unaccusative verbs ([vP fall he] vs. He fell) strongly suggests a structure even more
complex than vP, according to the thematic theory outlined above. While this might be an
argument resting on theory-internal grounds, modification of ARIs by aspectual adverbs
(like often) shows that there must be more structure than a thematically complete vP since
semantically these adverbs need to take scope/quantify over a whole eventuality, which is
syntactically represented by vP (‘It is ASPECToften [that AGENT v PATIENT]’). In conjunction
with the VPISH, then, adverbial modification yields a schematic structure along the lines
of the following representation, where the highest vP-internal nominal must have moved
across the adverb often (be it in SpecAspP, in the outer SpecvP, or maximally adjoined to
vP):
76
In generative theory, what had been known as nominal phrases (NPs) have been treated as Determiner
phrases (DPs) at least since Abney (1987).
~ 128 ~
Neven W
Wenger: Adult root
r
infinitives
(24)
Inteerestingly, ARIs
A
are no
ot only com
mpatible witth adverb ty
ypes that m
modify even
ntualities
(i.e.. vP’s) as a whole, but also with adverbs th
hat are licen
nsed still hiigher in thee clausal
struucture, within the I-domain. Thosse adverbs comprise among
a
otheers root (≈ deontic)
moddality adverrbs, as oppo
osed to advverbs licenseed still high
her within tthe C-domaain (incl.
episstemic modaality, eviden
ntial, etc.; ccf. next sectiion).
(25) Frittz necessa
arilyROOT diigs girls since
s
he’s a man… – What!!? Fritz
neccessarilyROOTT dig girls!?? Why!? As I see it, he might well be into men
n…
In aaddition, AR
RIs are com
mpatible wiith temporaal adverb(iaal)s, but onlly with non
n-deictic
ones, deictic onnes being prrohibited byy the nonfin
niteness chaaracteristic oof ARIs (to which I
willl return in seection 3).
(26) Frittz be a theo
oretical synttactician som
me day[-deictiic]/*tomorroow[+deictic]!? You got
it completely
c
wrong!
w
He’lll be a syntaactic theoretician!
On the assumpption, then, that temporral adverbiaals are generally licens ed by the I--head T,
whiich is also where tensse values arre introduceed, it is safe to conceeive of the ARI as
projjecting at leeast to IP (ass given in (224)).
2.3
3.5 ARIs as
a CPs
Thee common diagnosticss to determ
mine the prresence/com
mplexity off the C-dom
main of
clauuses are (i) modificatio
on by speakker-oriented
d adverbs (ccf. above w..r.t. IP and vP); (ii)
operrations targgeting the left peripherry (topicalisation, etc.)). To cut thhings short,, for the
majority of lannguages exaamined (inccl. English),, there is no
o indicationn of a left periphery
(exccept for a grroup of Western Romaance languages; cf. (20), hence no reason to assume
a
a
C-ddomain:
~ 129 ~
Neven Wenger: Adult root infinitives
(27) Him (*unfortunately) prepare a paper!? (cf. He’s unfortunately preparing a
paper.)
(28) *[A paper]i, him prepare ti!? (cf. A paper, he’s preparing (but no thesis).)
One feature of the ARI, which seems to have been neglected in the syntactic literature, is
that it exhibits a rather systematic behaviour when it comes to the mobility of its vP: the
vP may undergo what looks like topicalisation, leaving behind the Subject (the TopP here
is without any particular theoretical commitment, generic):
(29) a. Prepare a paper, him!?
b. [TopP [vP Prepare a paper]i Top° [IP him I° ti ] ]. – vP-TOPICALISATION
Rather straightforwardly, this raises the question whether the derived front position of the
predicate in (29) is a syntactic fronting operation at all, or really just an instance of
conjunct reversibility, supporting the analysis of ARIs as biphrasal coordination
structures (Prepare a paper!? Him!?; cf. (29). While one might be inclined to intuitively
take the latter view, German ARIs, which may (optionally?) occur with an overt
coordinator und ‘and’ linking Subject and predicate, suggest otherwise77. In coordinative
ARIs, vP-fronting is barred, and so is &P-fronting (fronting of the phrase headed by the
coordinator):
(30)
(31)
a. Der
ein Paper schreiben!?
that.one a paper write!?
‘Him write a paper!?’
b. [vP Ein Paper schreiben]i, der ti!?
und ein Paper schreiben!? – COORDINATIVE ARI
that.one and a paper write!?
‘Him (and) write a paper!?’
b. *[vP Ein Paper schreiben]i, der und ti!?
c. *[&P Und ein Paper schreiben]i, der ti!?
a. Der
It is all but clear what kind of functional element the coordinator in ARIs really is: It is no
Boolean coordinator (cf. Potts & Roeper 2006:198), neither of the symmetrical kind,
denoting intersection (e.g. [N John] and [N Paul]…), nor of the asymmetrical kind,
denoting logical spatio-temporal or modal relations (e.g. He tripped, and fell). The latter
both require identical categorial and semantic types for the conjuncts (where equicategoriality may be obstructed by ellipsis). Rather, the coordinator in ARIs appears to be
a predicator, akin to the as coocurring with the verb regard (I regard [PredP him [Pred° as]
77
The coordinative ARI is also available to English, though to a limited degree/non-productively (cf.
Bücker 2008:65ff.).
~ 130 ~
Neven Wenger: Adult root infinitives
[an idiot]]). Leaving the discussion of this aspect of the ARI to future investigation, what
is relevant here is that ARIs do seem to license movement operations, as indicated
negatively by intervening restrictions (coordinator).
A final question that remains to be addressed is the target of the movement operation: in
general, at least some nonfinite structures are taken to be without a C-domain (i.e.
truncated to IP). However, it is the C-domain where topics are assumed to be licensed,
according to varying theoretical implementations (adjunction theories excluded, all
approaches syntactic in character): (i) classically, in SpecCP; (ii) in the SpecTopP/FocP
of an articulated C-domain: Force > Top > Foc > Fin (cf. Rizzi 1997, and others); (iii) in
the Specs of the non-information-structural heads of a Rizzian C-system (i.e. in
SpecForceP and/or SpecFinP; cf. López 2009). Although interesting (in particular w.r.t. to
a group of Western Romance languages, which do allow for left-peripheral operations in
ARIs, though restrictively; cf. (30) and Wenger 2008:52), for reasons of space I cannot
pursue investigations into the information-structural (IS) syntax of ARIs. Nonetheless, a
theory of the syntax of finiteness (as to be rudimentarily sketched in section 3) should
shed light on these matters as well since it is exactly here, around the C–I interface, where
the derivation of both left-peripheral IS-operations and finiteness is computed.
2.3.6 A short note on the transsentential character of ARIs
It should be pointed out that not only the internal structure of ARIs is of controversial
interest, but also the external, i.e. transsentential, context, in which they occur within
discourse. As already laid out in the introductory section, ARIs are likely to be
accompanied by a follow-up expression – a Coda –, which spells out (= confirms) their
incredulitive force lexically:
(32) What!? Him prepare a paper!? No way, dude!/Never!/I can’t believe it!/…
This feature of the ARI raises the question of what the linguistic nature of this
dependency is – discursive or syntactic? While the former solution is relatively plausible
(as e.g. assumed by Lambrecht 1990), the latter one, as it is maintained by Etxepare &
Grohmann (2000 et seq.), is much less obvious. They argue for a quantificational
tripartite syntactic representation of ARIs (ultimately following Heim 1982, Diesing
1992), headed by the quantificational operator R (as proposed by Zanuttini & Portner
2003 for exclamative clauses), contrasting an abnormal/unexpected event expressed by
the ARI with a presupposed set of normal/expected situations (= the function of
widening). R is taken to take as its restriction its external argument (the ARI), mapping it
into its nuclear scope (the Coda) (cf. Wenger 2008:55ff. for a more detailed exposition
and discussion).
~ 131 ~
Neven
N
Wengeer: Adult root infinitives
(333)
As emppirical evideence, Etxep
pare and Grrohmann (e.g. Etxepare & Grohm
mann 2005) rely
on connnectivity eff
ffects that seeem to holdd between ARI
A and Co
oda. One suuch effect iss the
licensinng of Negatiive Polarity
y Items (NP
PIs) like anyy-X (in the non-generic
n
c reading). NPIs
N
must bbe licensed by ‘emphatic’ elemeents (negative, interro
ogative,…) c-comman
nding
them:
(334) a. I donn’tNEG want to have anyyNPI/*some more coffee.
b. I waant to have *any
* NPI/som
me more cofffee.
As it tturns out, NPIs contaained withiin ARIs seeem to need a (negaative) licen
nsing
expresssion in their Coda (exam
mple taken form Etxep
pare & Groh
hmann 20055:134):
hop!? Neverr!/No way!/*
*Of course!!/*Okay!
(335) Me buyy anythingNPI
N in that sh
Howeveer, this piece of eviidence facees several problems, one set cconcerning the
grammaatical naturee of the Coda itself, thhe other the explanatory
y base for tthe connectivity
effects. First, it is not
n evident that the Cooda is an oblligatory folllow-up to A
ARIs at all since
s
may well be uttered on their
t
own, m
maybe acco
ompanied by
y an extralinnguistic (faacial)
ARIs m
expresssion of dissbelief – which,
w
how
wever, doess not consttitute a naarrow synta
actic
(gramm
matical) property. Besiides that, thhe Coda ex
xpression ittself is possitionally raather
variablee (it can weell occur prreceding ann ARI: No way!
w
Him read books! ?), and a raather
heteroggeneous set with regard
d to its cate gorial statu
us (QP, DP, CP, ellipticcal expressiions,
etc.) – characterisstics that in my view
w favour a discourse--based expllanation off the
connecttivity, whicch is more loose thann syntactic connectivitty. In addittion, as forr the
licensinng of NPIss, a more straightforrward acco
ount is avaailable: Thhe interrogaative
dimensiion of AR
RIs (ARIs clearly invvolve a fo
orce type consisting
c
of at leastt an
exclamaative and an interro
ogative valuue, hence the ‘!?’). Interrogattivity, how
wever
represennted syntacttically, licen
nses NPIs, aas shown in
n the following examplle:
(336) a. Wouuld Steven like any morre coffee?
b. Stevven like any more coffeee!?
– QU
UESTION
– AR
RI
In whatt follows, I will
w treat th
he ARI as a mono-claussal structuree, leaving thhe discussio
on of
its highher order struucture (synttactic vs. diiscursive) to
o future reseearch.
~ 132 ~
Neven Wenger: Adult root infinitives
3 FINITENESS
In this section, I will sketch a syntax of finiteness, which essentially reduces to an
examination of the syntax-internal C–Infl interface. I believe the key to understanding the
syntactic dimension of finiteness lies in the syntax of tense (more abstractly, referential
anchoring). Before laying out a syntactic account of finiteness, specifically w.r.t. the
nonfiniteness in ARIs, I would first like to sketch the ontology of finiteness within the
generative paradigm.
Traditionally, finiteness itself has been a theory-peripheral, often merely descriptive,
notion associated with (at least) four phenomena: (i) Tense (TNS); (ii) NOM Case; (iii)
AgrS/SVA; and (iv), syntactic dependence. While initially (i.e. at least since the
grammarians of the antiquity) the focus was on the morphological dimension of the verb
(w.r.t. tense and φ-agreement, particularly PERSON), it has shifted to a broader, clausal
context in modern theories of grammar: finite clauses license NOM Subjects; finite clauses
can be syntactically independent (i.e. root clauses), nonfinite clauses are dependents, and
do not generally license Subjects, etc.. Within generative theory the features (i)-(iii) have
been described in terms of implicational licensing correlations of the following kind, at
least since Chomsky (1981), which is how the notion of finiteness found its way into
generative syntax, though initially as a rather descriptive label [±Fin]78:
(37) a. I[+Fin]: [+Tns] o [+Agr] o [NOM]
b. I[–Fin]: [–Tns] o [–Agr] o ([/NULL]79)
An assumption within the formulation of this correlation has always been that structural
Case – the last link in the chain – is contingent on other phenomena, alternately on TNS or
Agr. In current phase-based minimalism à la Chomsky (2000 et seq.), the licensing
correlation is derivationally instantiated in T (= I), which contains [TNS] and [φ] (φ is the
former Agr, and consists of the feature bundle [PERSON, NUMBER, GENDER]), NOM being
considered a post-syntactic PF-reflex licensed by a complete φ-feature on T. Accordingly,
nonfinite contexts are considered to be headed by a φ-defective Tdef, which does not
license Case. While the licensing chain as formulated in (37) does not reflect minimalist
syntax, where there is no interdependency between [TNS] and [φ], it is nonetheless
assumed to be φ-completeness/defectiveness that differentiates nonfinite from finite
domains. Hence, the capacity of agreement is equated with finiteness (according to option
(iii) above).
At first sight, neither SVA, nor any of the other four options listed above prove sufficient
78
Although occasional occurrences in several GBT-era works suggest otherwise, e.g. FP within Pollock's
(1989:372, 394) split-IP hypothesis, the vagueness and subsequent re-disappearance confirms the notion’s
(formerly) descriptive character.
79
Default Case aside (which is ACC in English), English has two other nonfinite Subject Cases not licensed
by I: ACC in so-called ECM-infinitivals (classically, i.e. AcI, e.g. I want him to leave), where ACC is
licensed by the matrix verb, and GEN, which in nominal gerunds (His visualising of my ideas is very
creative), is licensed DP-internally.
~ 133 ~
Neven Wenger: Adult root infinitives
as candidates for finiteness: cross-linguistically there are nonfinite structures that (i) do
show SVA (contra the conception just sketched; cf. (19)); that (ii) can occur as
independent root structures (e.g. the ARI); that (iii) can license NOM Subjects (e.g.
Hungarian; cf. Szabolcsi 2005:619); and that (iii) can manifest overt (= morphological)
Tense, as the following example from Classic Latin shows (from Cecchetto & Oniga
2001:15):
(38)
Vellem
hoc scripisse
would-like-1SG that write-PST-INF
‘I would like to have written that.’
On the plausible assumptions that NOM Case is a secondary property, and that the
dependence of nonfinite structures can be overridden rather easily (with syntax-external
help, ‘pragmatic-illocutionary enrichment’), I would like to sketch a schematic derivation
of ARIs with regard to nonfiniteness that relies on properties of tense rather than
agreement80, 81.
On closer scrutiny, examples of so-called tensed infinitives like the Latin one in (38) turn
out not to involve fully referential, i.e. deictic, tense, but rather dependent anaphoric
tense, which cannot referentially anchor to the speech event (the NOW of the origo), but
derives its temporal interpretation contextually, either by entering into a dependency
(binding, valuation, checking, etc.) with a higher matrix predicate, or – in case of root
infinitives – by appealing to non-syntactic means. For the above example from Latin this
means that the Tense morpheme -is(s)- of the infinitive scripisse is, strictly speaking, not
a real PAST morpheme (viz. PST). Past tense orders a structure corresponding to an
eventuality (or to an extended situation/interval), including an indication of event time
(ET) and Reference time (RT), anterior to the Speech Time (ST) (cf. Reichenbach 1947
for these notions) (i.e. RT < ST). Crucially, -is(s)- orders the eventuality anterior to the
ET/RT of the matrix predicate velle (i.e. anaphorically: RTscribere < ETvelle), which itself is
ordered contemporaneously with ST (PRESENT: ST,RT,ETvelle).
Arguably, the anaphoric nature of the temporal reference of the inflected Latin infinitive
is just what would be expected for non-inflected counterparts in e.g. English. While this
has been a major point of controversy ever since Stowell (1982), according to whom
some linguists have maintained that (some) infinitivals have a future-irrealis tense
semantics (viz. Control infinitivals), many others view infinitivals as transparent domains
whose temporal interpretation is context-dependent, i.e. anaphorically determined by the
80
I am not discussing agreeing infinitives as those found in Brazilian Portuguese here (cf. (19)).
To my knowledge, yet other candidates for the identification of finiteness have been verbal Mood (e.g.
Aygen 2002), assertability (e.g. Klein 2006), or none at all (or, in other words, a composite notion), as in
Landau’s (2004) scale of finiteness. Particularly interesting is the (not new) idea of categorising
(non)finiteness as a type of verbal Mood, nonfinite verb forms being in complementary distribution with
other verbal Moods such as the Indicative, or the Subjunctive. For reasons of space, though, I will not
discuss this option here.
81
~ 134 ~
Neven Wenger: Adult root infinitives
main clause tense (cf. e.g. Hornstein 1990:146ff.). Based on Neo-Reichenbachian
approaches to the syntax of tense, the three time points/intervals are represented in the
tripartite backbone of the clause, where finiteness is a function of the referential
anchoring of the extended situation (i.e. IP ⊃ RT–ET) to C (⊃ ST) (cf. Enç 1987;
Bianchi 2003):
(39) [CP C°[ST] [IP I[RT] [vP v°[ET] ] ] ]
Just like some of the works just mentioned, I would like to capitalise on the notion of
defectiveness (a privative approach) – but rather as phrase-structural underspecification
(i.e. scalar truncation82) than as featural underspecification (e.g. a missing φ-probe on T,
or a lacking Tense-feature). The dichotomy of phrase-structural and featural
underspecification of structures mirrors the tension between mainstream (cf. e.g.
Chomsky 2000) and Cartographic generative theories of phrase-structure (cf. e.g. Rizzi
1997), the former assuming the clausal skeleton to consist of macro-categories (C > T > v
> V), the latter of fine-grained sequences of varying complexity (e.g. C = Force > Top >
Foc > Fin > IP; cf. Rizzi 1997). On the truncation view, then, nonfiniteness results from
the absence of structure, specifically, C (⊃ ST): the IP remains unanchored within its
own domain, either linking up to a higher clause, or to a discourse context, as in the case
of ARIs.
Going back to the licensing of agreement, Tense, and Case, then, the Neo-Reichenbachian
syntax of tense just outlined in a simplistic fashion might be integrated with current
reasoning as follows: On the assumption that it is C (or Fin/Cmin = the lowest head of a
split C-domain) that licenses [NOM] and [uφ: ], and not T (I) (cf. the similar idea of C–T
feature inheritance Chomsky 2007, 2008), the absence of SVA and NOM follows if an
infinitival projects only to TP (or Imax), to the very exclusion of C[uφ: , NOM]. Subject
raising can still be assured if one conceives of the [EPP] as a leftness condition (i.e. Spec
condition) on domains (which may be overridden subsequently), having a nominal raise
to the left edge of the I-domain under Attract Closest (cf. Chomsky 1995:297).
Finally, what remains to be sketched in more detail are the tense properties of infinitival
structures. A more recent trend is the assumption that (mono-clausal) syntactic structures
involve more than just one tense category, namely one that is lower in the clausal
structure, topping off the vP, and a higher one, the classic T(P) (cf. e.g. Pesetsky &
Torrego 2004). In the present approach, the higher Tense (proper, i.e. referential tense) is
reanalysed as C/Fin (or as contained therein), while the lower tense is identified as the
classic T, but redefined as (containing) a time variable [TNS:val] (on possible values cf.
below) that denotes the temporal orientation of an extended event, which may be
interacting with the event argument on v (ET) as well as aspectual categories to produce
82
Given a functional sequence a > b > c (an fseq, i.e. a fixed hierarchy of functional projections; cf. Starke
2001), the lack of b entails the lack of a, i.e. only scalar truncation of the fseq is permitted. Allowing for
selective truncation, e.g. ?a > c, is not restrictive enough (indeed, not restrictive at all) in capturing phrasestructurally a configurational property of a syntactic structure.
~ 135 ~
Neven
N
Wengeer: Adult root infinitives
distinctions of view
wpoint and
d inner aspeect. The tem
mporal depeendency beetween C an
nd T
may be construed in
i terms of Agree-baseed feature valuation:
v
Besides [uφ: ] and [NOM
M], C
contains a (defaullt?) time vaalue denotinng the NOW
W (referential index), ssay [TNS:NO
OW],
while T carries a feature
f
deno
oting the rellative temporal orientaation of the extended event
e
it headds (i.e. relaative tense)), which caan take on
ne of threee values [A
ANT(ERIORIT
TY)],
83
[POST(EERIORITY)], or [COIN(CIDENCE)] . Furthermo
ore, if the anaphoric
a
reelation is to be
conceivved of in terrms of valu
uation (insteead of, say, binding), T also carriees an unvallued,
uninterppretable ‘llinking’ feature.
f
Ovverall, this treatmen
nt of Teense yields a
decomppositional-dderivational syntactic appproach to Tense
T
(e.g. simple pastt = C[TNS:NOW
N
]
> T[TNSS:ANT] > Assp[TNS:COIN
N]).
(400)
C must carry an acctive probe [uTNS: ], w
which search
hes for a matching valu
lue [TNS:vall], in
order foor Agree(C,,T) to be esttablished. T
Thus, in the present exaample, C[uTTNS: ] probes for
and finnds T[TNS:ANT
A ], which
h heads ann extended situation typed
t
‘anteerior’ (temp
poral
orientattion) (and, irrelevant here, ‘perrfective’, or
o simple, aspect). A
At the syn
ntax–
morphoology interfface, an according Teense morph
heme – herre, in Engliish, {-ED} – is
insertedd into [uTNSS:ANT], witth affix hoppping (cf. Chomsky
C
19
957 [2002])) producing
g the
84
surface order (low,, i.e. v, in En
nglish) .
about nonfinnite structurres? Evidenntly, with C,
C and thus the ‘properr’ Tense feaature
What ab
[uTNS] as well as the referen
ntial anchorr [TNS:NOW
W], being ab
bsent, an A
ARI is correectly
predicteed not to diisplay deictiic Tense, buut to still bee capable of producingg distinction
ns of
relativee tense (teemporal oriientation). Indeed, iff languagess possess free or bo
ound
morphoology that iss able to lex
xicalise a nnonfinite (i.ee. C-less) T,
T this is posssible underr the
proper contextual circumstan
nces (contrrary to wh
hat is claim
med, e.g. iin Etxepare &
Grohmaann 2007). As
A for anterriority, this is typical of past perfect contexts::
83
This trripartition cann be derived frrom a feature hierarchy of temporal
t
coin
ncidence (cf. R
Ritter & Wiltsschko
2005) com
mposed of [±ccoincidence] and
a [±anteriorrity/posteriority] (or similarrly, e.g. privattively).
84
Recall that the higheest [uTNS] feaature would bee (contained in)
i the finiteness category FFin in a more fineb dominated by (at least) another functtional
grained pperspective onn phrase struccture, and wouuld thus still be
projection, viz. Force.
~ 136 ~
Neven Wenger: Adult root infinitives
(41)
A: When I came home yesterday, Axel had already left… – B: What!? Him
have left (already)!? Not true! I spotted him behind the sofa when I came home!
In nonfinite contexts, the perfect auxiliary, which is commonly associated with tempoaspectual concepts like ‘present relevance’ or ‘current orientation’ (an eventuality anterior
to ST extends into the NOW), is reduced to an expression of anteriority, as predicted by the
lack of anchoring85. The fact that ARIs – like most nonfinites – are interpreted as irrealis
is likely to be due to extra-syntactic default interpretation, assigning an unactualised (=
irrealis) semantics to unanchored situations.
4 CONCLUSION AND OUTLOOK
A careful survey of the linguistic properties of ARIs has shown two things: (i) that it can
be subjected to syntactic analysis, contrary to arguments that see it as a ‘defective’
construction consisting of two loosely linked phrases; and (ii) that ARIs have an abstract
syntactic structure that is more complex than meets the eye (vP/SC), but just as complex
to match the requirements imposed by nonfiniteness (TP, but not CP).
The analysis of ARIs as TPs (syntactically) and extended situations (semantically) still
lacks one component of meaning – pragmatics, especially, illocutionary meaning.
Uncontroversially, ARIs possess a kind of force (incredulitive), which is rather
specialised and cannot be easily overridden (as is the case e.g. with declaratives as
questions). However, provided that illocutionary force is somehow represented in syntax
(as sentence mood, sentential force, or, put more simply, as a ‘clause-typing’ operator),
ARIs (and reduced nonfinite structures more generally) prove problematic given that the
locus of force is generally assumed to be very high in the clausal spine, in C (or Force), as
suggested by COMPs indicating clause-type (that o declarative, if o interrogative).
W.r.t. ARIs then, I can only speculate that their illocutionary force is derived
compositionally extra-syntactically, rather than being represented structurally, e.g. by
operators. In any case, the role of illocutionary meaning in nonfinite root structures, as
well as the effect of prosody on illocutionary meaning need to be taken into account to be
able to tackle these questions – a topic for another paper.
REFERENCES
Abney, Steven P. 1987. The English Noun Phrase in Its Sentential Aspect. Ph.D. thesis,
MIT.
Avrutin, Sergey. 1999. Development of the Syntax-Discourse Interface. Dordrecht:
Kluwer.
Aygen, Nigar Gulsat. 2002. Finiteness, Case and Clausal Architecture. Ph.D. thesis,
Graduate School of Arts and Sciences, Harvard University.
Baker, Mark C. 1988. Incorporation: A Theory of Grammatical Function Changing.
85
Also cf. the discussion around the Latin example (38) and p. 19. Latin realises the anteriority realised by
have synthetically, by an inflectional affix.
~ 137 ~
Neven Wenger: Adult root infinitives
Chicago, IL: University of Chicago Press.
Belletti, Adriana 1990. Generalized Verb Movement: Aspects of Verb Syntax. Turin:
Rosenberg & Sellier.
Bianchi, Valentina. 2003. On finiteness as logophoric anchoring. In: Jacqueline Guéron &
Liliane Tasmovski. Temps et Points de Vue/Tense and Point of View. Nanterre:
Université Paris X, 213-246.
Boeckx, Cedric. 2008. Bare Syntax. New York, NY: OUP.
BoškoviΕ, Željko. 1997. The Syntax of Nonfinite Complementation: An Economy
Approach. Cambridge, MA: MIT Press.
Brandt, Patrick. 2006. Receiving and Perceiving Datives (Cipients): A View from
German. In: Daniel Hole et al. Datives and Other Cases: Between Argument
Structure and Event Structure. Amsterdam, Netherlands: Benjamins, 103-139.
Bücker, Jörg. 2007 ‘Du und tolerant?’ – Zur Syntax, Semantik und Pragmatik deutscher
‘Incredulity Response Constructions’ (IRCs). ‘Grammatik in der Interaktion’
(GIDI) Working Papers 5.
Bücker, Jörg. 2008. Polythetic Classes and ‘Constructional Chains’: ‘Incredulity
Response Constructions’ (IRCs) and Related Constructions from a Cross-Linguistic
and Diachronic Point of View. ‘Grammatik in der Interaktion’ (GIDI) Working
Papers 13.
Büring, Daniel. 2005. Binding Theory. New York, NY: CUP.
Butler, Jonny. 2004. Phase structure, Phrase structure, and Quantification. Ph.D. thesis,
Department of Language & Linguistic Science, The University of York.
Cardinaletti, Anna & Michal Starke. 1999.The typology of structural deficiency: A case
study of the three classes of pronouns. In: Henk van Riemsdijk et al. (1999). Clitics
in the Languages of Europe. Berlin: Mouton de Gruyter, 145-233.
Carnie, Andrew.2008. Constituent Structure. New York, NY: OUP.
Cecchetto, Carlo & Renato Oniga. 2001. Consequences of the Analysis of Latin Infinitival
Clauses for the Theory of Case and Control. University of Milano-Bicocca &
University of Udine.
Chomsky, Noam. 1957 [2002]. Syntactic Structures. The Hague: Mouton.
Chomsky, Noam. 1981. Lectures on Government and Binding: The Pisa Lectures.
Dordrecht: Foris.
Chomsky, Noam. 1993: A minimalist program for linguistic theory. In: Kenneth Hale &
Samuel Jay Keyser. The View from Building 20: Essays in Linguistics in Honor of
Sylvain Bromberger. Cambridge, MA: MIT Press, 1-52.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Chomsky, Noam.2000. Minimalist inquiries: The framework. In: Roger Martin et al.
(2000). Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik.
Cambridge, MA: MIT Press, 89-156.
Chomsky, Noam. 2001. Derivation by phase. In: Michael Kenstowicz et al. Ken Hale: A
Life in Language. Cambridge, MA: MIT Press, 1-52.
Chomsky, Noam. 2007: Approaching UG from below. In: Uli Sauerland & Hans-Martin
Gärtner. Interfaces + Recursion = Language? Chomsky's Minimalism and the View
from Syntax-Semantics. Berlin, Germany: Mouton de Gruyter, 1-29.
Chomsky, Noam. 2008. On phases. In: Robert Freidin et al. Foundational Issues in
Linguistic Theory. Cambridge, MA: MIT, 133-166.
~ 138 ~
Neven Wenger: Adult root infinitives
Cinque, Guglielmo.1999). Adverbs and Functional Heads: A Cross-linguistic
Perspective. New York, NY: OUP.
Davies, William D. & Stanley Dubinsky. 2004. The Grammar of Raising and Control.
Malden, MA, et al.: Blackwell Publishing.
den Dikken, Marcel. 2006. Relators and Linkers: The Syntax of Predication, Predicate
Inversion, and Copulas. Cambridge, MA: MIT Press.
Diesing, Molly. 1992. Indefinites. Cambridge, MA: MIT Press.
Drubig, H. Bernhard. 2007. Phases and the typology of focus constructions. In: Kerstin
Schwabe & Susanne Winkler. On Information Structure, Meaning and Form:
Generalizations across Languages. Amsterdam, Netherlands: Benjamins, 33-66.
Enç, Mürvet. 1987. Anchoring Conditions for Tense. Linguistic Inquiry 18:4, 633-659.
Etxepare, Ricardo & Kleanthes K. Grohmann. 2002. Towards a syntax of Adult Root
Infinitives. In: Teresa Satterfield et al. Current Issues in Romance Languages.
Amsterdam, Netherlands: Benjamins, 65-79.
Etxepare, Ricardo & Kleanthes K. Grohmann.2005. Towards a grammar of Adult Root
Infinitives. In: John Alderete et al. Proceedings of the 24th West Coast Conference
on Formal Linguistics. Somerville, MA, USA: Cascadilla Proceedings Project, 129137.
Etxepare, Ricardo & Kleanthes K. Grohmann. 2007. Temporal and aspectual variation in
Adult Root Infinitives. In: Louis de Saussure et al. Recent Advances in the Syntax
and Semantics of Tense, Aspect, and Modality. Berlin: Mouton de Gruyter, 147168.
Fillmore, Charles J., Paul Kay et al. 1988. Regularity and idiomaticity in grammatical
constructions: The case of 'Let Alone'. Language: Journal of the Linguistic Society
of America 64:3, 501-538.
George, Leland M. & Jaklin Kornfilt. 1981. Finiteness and boundedness in Turkish. In:
Frank Heny. Binding and Filtering. Cambridge: MITPress, 105-127.
Grohmann, Kleanthes K. & Ricardo Etxepare (2003): "Root infinitives: A comparative
view". Probus: International Journal of Latin and Romance Linguistics 15:2, 201236.
Heim, Irene. 1982. The Semantics of Definite and Indefinite Noun Phrases. thesis,
Department of Linguistics, University of Massachusetts Amherst.
Hoekstra, Teun & Nina Hyams. 1998. Aspects of root infinitives. Lingua: International
Review of General Linguistics 106:1-4, 81-112.
Hornstein, Norbert. 1990. As Time Goes By: Tense and Universal Grammar. Cambridge:
MIT Press.
Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge,
MA: MIT Press.
Klein, Wolfgang. 2006. On finiteness. In: Veerle van Geenhoven. Semantics in
Acquisition. Dordrecht: Springer, 245–272.
Lambrecht, Knud. 1990. 'What, Me Worry?': Mad Magazine Sentences revisited. In: Kira
Hall et al.. Proceedings of the Sixteenth Annual Meeting of the Berkeley Linguistics
Society. Berkeley: Berkeley Linguistics Society, 215-228.
López, Luis. 2009. A Derivational Syntax for Information Structure. New York, NY:
OUP.
Nikolaeva, Irina, Ed. 2007. Finiteness: Theoretical and Empirical Foundations. New
~ 139 ~
Neven Wenger: Adult root infinitives
York, NY: OUP.
Pesetsky, David & Esther Torrego. 2001. T-to-C movement: Causes and consequences.
In: Michael Kenstowicz et al. Ken Hale: A Life in Language. Cambridge, MA: MIT
Press, 355-426.
Pesetsky, David & Esther Torrego. 2004. Tense, Case, and the nature of syntactic
categories. In: Jacqueline Guéron & Jacqueline Lecarme. The Syntax of Time.
Cambridge, MA: MIT, 495-537.
Pesetsky, David & Esther Torrego. 2007. The syntax of valuation and the interpretability
of features. In: Simin Karimi et al. Phrasal and Clausal Architecture: Syntactic
Derivation and Interpretation. Amsterdam, Netherlands: Benjamins, 262-294.
Pollock, Jean-Yves. 1989. Verb movement, Universal Grammar, and the structure of IP.
Linguistic Inquiry 20:3, 365-424.
Potts, Christopher & Tom Roeper. 2006. The narrowing acquisition path: From
Expressive Small Clauses to declaratives. In: Ljiljana Progovac et al.. The Syntax of
Nonsententials: Multidisciplinary Perspectives. Amsterdam, Netherlands:
Benjamins, 183-201.
Progovac, Ljiljana. 2006. The syntax of nonsententials: Small Clauses and phrases at the
root. In: Ljiljana Progovac et al.. The Syntax of Nonsententials: Multidisciplinary
Perspectives. Amsterdam, Netherlands: Benjamins, 33-71.
Radford, Andrew. 1990. Syntactic Theory and the Acquisition of English Syntax: The
Nature of Early Child Grammars of English. Oxford: Blackwell.
Radford, Andrew. 2009. An Introduction to English Sentence Structure. Cambridge:
CUP.
Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York, NY: McMillan.
Ritter, Elizabeth & Martina Wiltschko. 2005. Anchoring events to utterances without
tense. In: John Alderete et al. Proceedings of the 24th West Coast Conference on
Formal Linguistics. Somerville, MA, USA: Cascadilla Proceedings Project, 343351.
Rizzi, Luigi. 1993. Some notes on linguistic theory and language development: The case
of root infinitives. Language Acquisition: A Journal of Developmental Linguistics
3:4, 371-393.
Rizzi, Luigi. 1997. The fine structure of the Left Periphery. In: Liliane Haegeman.
Elements of Grammar: Handbook of Generative Syntax. Dordrecht: Kluwer, 281337.
Rizzi, Luigi & Ur Shlonsky. 2006. Satisfying the Subject Criterion by a non-Subject:
English Locative Inversion and Heavy NP-Shift. In: Mara Frascarelli. Phases of
Interpretation. Berlin, Germany: Mouton de Gruyter, 341-361.
Rizzi, Luigi & Ur Shlonsky. 2007. Strategies of Subject extraction. In: Uli Sauerland &
Hans-Martin Gärtner. Interfaces + Recursion = Language? Chomsky's Minimalism
and the View from Syntax-Semantics. Berlin, Germany: Mouton de Gruyter, 115160.
Schütze, Carson Theodore Robert. 1997. INFL in Child and Adult Language: Agreement,
Case and Licensing, Massachusetts Institute of Technology.
Starke, Michal. 2001. Move Dissolves into Merge: A Theory of Locality. Ph.D. thesis,
University of Geneva,
Stowell, Tim. 1982. The Tense of Infinitives. Linguistic Inquiry 13:3, 561-570.
~ 140 ~
Neven Wenger: Adult root infinitives
Vergnaud, Jean-Roger. 2008. Letter to Noam Chomsky and Howard Lasnik on 'Filters
and Control,' April 17, 1977. In: Robert Freidin et al.. Foundational Issues in
Linguistic Theory. Cambridge, MA: MIT, 3-15.
Visser, Frederikus Theodorus. 1963. An Historical Syntax of the English Language.
Leiden: Brill Academic Publishers.
Wenger, Neven. 2008. "Me a sentence type?!”: English Adult Root Infinitives
Reconsidered in a Minimalist Syntax Framework. M.A. thesis, Anglistisches
Seminar, Heidelberg University.
Wenger, Neven. 2009. The Syntax of Finiteness. Frankfurt a. M.
Williams, Edwin S. 1975. Small Clauses in English. Syntax and Semantics 4, 249-273.
Zanuttini, Raffaella & Paul Portner. 2003. Exclamative Clauses: At the Syntax-Semantics
Interface. Language: Journal of the Linguistic Society of America 79:1, 39-81.
~ 141 ~