DURING THE YEAR 1954-1955 my wife and I

were engaged in a study of child-rearing in a

semi-rural New England village. In the course
of the study I had occasion to record two or
more interviews on Audograph discs or tapes,
with each of the 24 children of our sample.
Previously certain inconsistencies in the chil-
dren's speech had attracted my attention,
especially the variation between -in and -ing
for the present participle ending. (The variation
in this dialect between -in and -ing in the
participle ending does not extend to words with
a final -m in an unstressed syllable in standard
speech. This variation is therefore probably
best viewed as a case of free alternation of two
allomorphs which happen to differ in respect
to one phoneme, rather than as a case of phono-
logical free variation.) Accordingly, in transcrib-
ing the discs and tapes, I decided to note the
choice of these two variants, and this paper is
intended to summarize and discuss this infor-
mation.
To begin with, all of the 24 children, except
three, used both forms to some extent at least.
The three exceptions used only the -ing form,
and since they were less loquacious than most
of the other children, it is possible that a larger
sample of their speech would have revealed the
use of the other variant as well. This may then
be regarded as a case of so-called free variation
of two linguistic forms within a local speech
community, and within the speech of most
individual members of our sample community.
In general, the choice of one or the other of
the variants would not affect the denotation of
acts, states, or events by the word.
"Free variation" is of course a label, not an
explanation. It does not tell us where the
variants came from nor why the speakers use
JOHN L FISCHER
them in differing proportions, but is rather a
way of excluding such questions from the scope
of immediate inquiry. Historically, I presume
that one could investigate the spread of one of
these variants into the territory of another
through contact and migration, and this would
constitute one useful sort of explanation.
However, another sort of explanation is possible
in terms of current factors which lead a given
child in given circumstances to produce one of
the variants rather than another, and it is this
which I wish to discuss here.
Before discussing the determinants of selection
of the variants it will be helpful to understand a
little of the general background of the data.
The 24 children in our sample consisted of an
equal number of boys and girls, both divided
into two equal age groups, ages 3-6 and 7-10.
By the time the recordings were made my
wife and I had been observing the children
periodically for eight to ten months and most
of the children were fairly well acquainted
with us. Most of the children were interviewed
in an office in our house, which was located
in the middle of the village. Most of the children
had visited our house before, some a number of
times. Four younger children who had not were
interviewed in their own homes. Three general
types of text were obtained:
1. Protocols for all children for a verbal
thematic apperception test (TAT) in which
the children were asked to make up stories
starting out from short sentences given by
the investigator.
2. For older children only, answers to a formal
questionnaire.
3. For a few of the older children, informal
interviews asking them to recount their
recent activities.
483
Social Influences on the Choice of 5
a Linguistic Variant

484 SOCIAL INFLUENCES
I shall present first some counts of variants in
the TAT protocols, since this test was adminis-
tered to all the children. As is shown in Table 1,
a markedly greater number of girls used -ing
more frequently, while more boys used more -
in.
The first boy was regarded by his teacher and
others as a "model" boy. He did his school
work well , was popular among his peers,
reputed to be thoughtful and considerate. The
second boy was generally regarded as a "typical"
boyphysically strong, dominating, full of
mischief, but disarmingly frank about his
transgressions. The "model" boy used almost
exclusively the -ing ending here, while the
"typical" boy used the -in ending more than
half the time, as shown above.
In Table 3 one may also note a slight tendency
for the -ing variant to be associated with higher
socio-economic status, although this is not
statistically significant with a sample of this
size. The community studied is fairly small and
does not have strong class lines, vhich is prob-
ably why more marked results did not appear.
(Most previous studies of sociological factors
connected with linguistic variants have been
concerned with linguistic indices of class, cast
or occupational groups. Group boundaries
have been regarded, implicity or explicitly, as
barriers to communication analogous to political
boundaries, geographical distance, etc. The
emphasis in this paper is rather on variations
within a face-to-face community whose members
are in frequent free communication: variations
between social categories of speakers and between
individual speakers, and situational variations in
the speech of individual speakers, as noted
below.)
Besides asking who uses which variant and
how much, we may also ask whether there are
situational differences in when a single speaker
uses these variants. One variant in the situation
may be descri bed as degree of formal i ty:
in the children's terms I would think of this as
degree of similarity to a formal classroom
recitation. The best child to examine for this
variable is the "model" boy of Table 2 since
he was interviewed in all three situations
mentioned above and was obligingly talkative
in each. As Table 4 shows, the frequency of
choice of variants changed from an almost
exclusive use of -ing in the TAT situation to a
predominance of-tn in the informal interviews.



JOHN L.FISCHER 485
Of course, these three situations should not be
regarded as exhaustive of the frequency range
of these variants in this boy's speech. In the
interviews I myself used the -ing variant consist-
tently and this probably influenced the inform-
ant's speech somewhat. Probably in casual
conversation with his peers the -in!-ing ratio
is even higher than in the informal interview.
Another measure similar in implication to the
frequency of variants by type of interview
would be differences in frequency between the
beginning and later parts of a single interview.
Especially in the TAT protocols, which are
the most formal text, I noticed for a number of
children that the -i ng frequency was higher
in the beginning of the interview and later
dropped off, presumably as the child became
more relaxed and accustomed to the situation.
In only one child was the reverse trend noted,
and there are reasons to believe that this
particular child may have become more tense
during the administration of the test.
A linguist might ask whether there is any
association between the suffix variants and
specific verbs. The corpus is not large enough
to establish stable frequency indices for the
suffixes of individual words, but there is
certainly a trend for markedly "formal" verbs
to have the -ing suffix and markedly "informal"
verbs to have the -in suffix. The first boy in
Table 2 above, for instance, used -ing in
criticizing,correcting, reading, visiting, interesting,
and used -in in punehin, flubbin, ncimmin,
chetcin, hittin. For some common verbs, how-
ever, such as play, go, and do he used both
alternatively. Probably only a few verbs are
formal or informal enough in their connotations
so that the same variant would always be used
with them. Of course, the choice of verb
vocabulary is itself related to personality and
situational factors.
In brief, then, the choice between the -ing and
the -in variants appear to be related to sex,
class, personality (aggressive/cooperative), and
mood (tense/relaxed) of the speakers (and
doubtless of the person spoken to, although
this was not investigated), to the formality of
the conversation, and to the specific verb
spoken. While these are "free variants" in the
standard type of description of languages in
which only grammatical facts and differences
in none but "denotative" meaning are taken
into account, if we widen our scope of study to
include the meaning of these variants to the
conversants we might call them "socially
conditioned variants," or "socio-symbolic var-
iants," on the grounds that they serve to
symbolize things about the relative status of
the conversants and their attitudes toward
each other, rather than denoting any difference
in the universe of primary discourse (the "outer
world"). (Uriel Weinrich has suggested to me
the term "symptomatic signs," after Karl
Biihler, as an alternative for "socio-symbolic
variant" which already has a basis in estab-
lished usage. However, it seems to me that
"symptomatic signs" might be in one sense
too broad and i n another too narrow: too
broad i n the sense that it might be i nter-
preted to refer to "non-linguistic" features of
speech such as general pitch, loudness, timbre,
rate, etc., and too narrow in the sense that
Biihler appears to regard the symptomatic
function as, ideally, purely expressive of the
speaker, while I am looking for a broader term
which would cover this function but also include
expression of the dyadic relationship between
the conversants. This cannot simply be taken
care of by adding in Buhler's "signal" function
which deals with the "appeal" to the listener,
since at least some aspects of the relationship do
not exist primarily either in speaker or listener
but rather between them, e.g., relative age,
relative rank. See Buhler [1934], especially
p. 28. Whether I should here introduce a term
incorporating "symbol" is a further question
which I acknowledge but do not discuss here,
as it is complex and is not directly relevant to
the main argument of the paper.) What are the
wider implications for linguistics of such an
analysis of social factors influencing choice of
linguistic variants? For one thing, many
linguists have recognized that "free" variation
is a logically necessary stage in most of all
linguistic change. (I find in checking over the
literature that this statement seems to be based
more on my impressions of conversations with
linguists than on published statements. One
clear statement of this principle, however, is to
be found on p. 367 of Vogt [1954]. A more
general statement applying to any type of
cultural element, and by implication linguistic
elements can be found i n Li nton [1936, p.
280].) Less widely appreciated but also recogn-
ized by some is another fact: Although the
mechanisms of psychic economy are becoming
486 SOCIAL I NFL VENCES
better understood in diachronic phonemics,
they are not always sufficient to explain fully
the progressive adaption of variant forms, and
that people adopt a variant primarily not
because it is easier to pronounce (which it
most frequently is, but not always), or because
it facilitates some important distinction in
denotational meaning, but because it expresses
how they feel about their relative status versus
other conversants.
The clearest and most comprehensive state-
ment of social factors in linguistic change which
I have encountered is found in an article by
Martin Joos (19S2) dealing with medieval
sibilants. (Others have separately recognized
the importance of fashion in linguistic change,
especially in the spread of standard dialects,
and to a lesser degree have recognized the
complementary process of using distinctive
linguistic features to emphasize social exclusive-
ness. J. O. Hertzler (1953) gives a bibliography
including studies of both sorts. Joos's statement
however appears to me to be unique in his
recognition that the two processes combine to
constitute a self-perpetuating cycle. Since Joos
is noted for his rigorous definition of the scope
of linguistics proper it is perhaps all the more
interesting that he should throw in this "socio-
logical" aside.) He speaks of "the phonetic
drift, which was kept going in the usual way:
that is, the dialects and idiolects of higher
prestige were more advanced in this direction,
and their speakers carried the drift further
along so as to maintain the prestige-marking
difference against their pursuers. The vanity
factor is needed to explain why phonetic drifts
tend to continue in the same direction; the
'inertia sometimes invoked is a label and not
an argument.'" This protracted pursuit of an
elite by an envious mass and consequent
"flight" of the elite is in my opinion the most
important mechanism in linguistic drift, not
only in the phonetic drift which Joos discusses,
but in syntactic and lexical drift as well.
(Incidentally, this flight-pursuit mechanism
might be regarded as an explanation of the
constant rate of decay of basic "non-cultural"
vocabulary postulated by Morris Swadesh's
theory of glottochronology. To make it suffice
one would also need to assume that all societies
possess some form of elite groupif only the
"ideal conformist" in some societiesand that
mass envy of the elite and ambition to join them
are everywhere the same. These assumptions
may seem radical and against common sense,
but they are not as easy to refute as one might
think. Needless to say, one would not assume
that the elite is always a property or authority
elite. In politically and economically undifler-
entiated societies, the most important criterion
might be technical skill and productivity in
consumer goods, admired personality traits, etc.)
The study of social factors in linguistic drift is in
the field of the sociology of language rather
than linguistics proper. However, this study
can not reach ultimate fruition without certain
linguistic studies by competent linguists. I refer
here to studies of individual variations in
linguistic forms in small, face-to-face speech
communities, and of variations in these forms
in the speech of single individuals in a range of
social situations. Studies of this sort constitute
tasks of respectable magnitude which have, in
the main, been neglected. (The classic study in
this field is Gauchat [1905]. Other references
are cited by von Wartburg [1946, p. 33]. Modern
techniques, of course, open entirely new
perspectives for research.) A student of
social factors in the choice of linguistic
variants would wish to know for a fairly
large stratified sample of a speech
community how often members of a given
sub-group used a sizable sample of series of
socially significant variants, and for at least some
of the sub-groups one would want to know
how these frequencies of choice of variants
changed under different situations and in the
presence of conversants of different social status
and personal relationships. A linguist as such
would not wish to analyze these social factors
in great detail. But it would be well within the
scope of linguistics to identify individual
informants in a unitary speech community by
name or code number and group them accord-
ing to their similarity or dissimilarity in the
use of variants in some standard situation, say,
in conversation with the linguist. The psychol-
ogist and sociologist could then take these
groups and see what sense they made in their
terms. In practice, of course, such a rigorous
separation between linguistics and the more
general social sciences is not required since
linguists and other laymen are presumably
capable of making a number of distinctions of
considerable sociological interest, such as male
versus female, etc.
A word about the relation of the proposed
study to dialectology is appropriate here. It has
generally been the aim of dialectologists to
describe linguistic variations between groups
which are separated by some communications
barrier, especially geography or social class.
What I am advocating here is the study of
linguistic variations within small groups where
there is free and relatively intense communica-
tion, so that as far as possible the lack of con-
tact between speakers is not a reason for failure
to use the same forms. Of course in a large
society such as ours, small closed groups are
rare, and some of the variation among the
individuals of any group picked for study will
be due to the fact that they have different con-
tacts outside the group. But this empirical
fact does not reduce the importance of studying
variation within the face-to-face community,
although it suggests that the best place to study
such variation would be on a remote Pacific
atoll with a small, long-established population.
What I am proposing might be called compar-
ative idiolectology rather than dialectology.
Ideally, a thorough description of a single
dialect would be based on the study of a
sizable sample of the idiolects in a local speech
community, in the same way that a through
description of a language would be based on
the study of a sizable sample of its dialects. In
comparative idiolectology one might, as a
device of field work, still concentrate on a
single informant, but one would want to follow
him around with a portable recording machine
and note changes in his speech in different
settings and situations and with different
conversants. Moreover, since phenomenolog-
ically language is as much listening as speaking
one would be led to analyze what was said
comprehensibly to him by others as well as
what he said himself.
The untrained listener will not, of course,
generally be able to reproduce or identify the
differences in the speech of others whom he
encounters, unless he is an accomplished mimic.
But he does react to these differences by
making interpretations about the social situation
on the basis of them and will be able to tell
when a speaker is talking like a woman, like
an upper class person, like a relaxed person,
etc., even though he cannot specify all the
variant forms on which he bases his judgment.
(The "tape experiment" described by Putnam
JOHN L.FISCHER 487
and O'Hern [1955] investigates language and
social status in this manner, although the
speakers were not members of a single
face-to-face community, so the complication
of barriers to communication is introduced.)
(This is not to deny the presence or importance
of other "non-linguistic" features of speech as
well as things entirely unconnected with speech
such as dress, physical appearance, gestures,
etc., which also serve as cues for judgments of
the conversational situation.) In analyzing socio-
symbolic variants there will obviously be a
certain amount of association between variant
series. In many of the series at least one variant
could be distinguished as "formal," and
another as "informal." But it is a question for
empirical investigation whether this distinction
applies to all variant series, and, if so, with how
much force. I have suggested above a number
of factors which influence the -in j-ing
distinction. Conceivably they all bear on
formality, that is, compliance, tenseness,
femaleness, and high class all make for formal
behavior. But even if this is true for these
factors in American culture, are they a unitary
complex in all cultures, and may there not be
other social factors affecting socio-symbolic
variants which are independent of the formality
complex? Are variants associated with being
female always associated as well with formality ?
In three languages with which I am acquainted,
English, Japanese, and Ponapean, I can think
of a number of instances where this link is
found, but there also appear to be exceptions.
In Ponapean, for instance, a minority of women
have an unusual allophone for the r phoneme,
but this seems to have no relation to the degree
of formality. Lisping in English is regarded as
feminine, but would indicate little about degree
of formality.
Even where the same factor determines the
choice of alternants in several series of variants,
the breaking point for each series will probably
be different. For instance, in the TAT texts
discussed above, three of the children used the
pronunciation ty for the indefinite article a.
This pronunciation can be regarded as formal
to the point of being artificial and is much more
restricted for speakers in this community
than the -ing variant of the present participle
ending, yet the direction of social symbolism
is the same, though not the intensity. In other
words, ty in itself is more a sign of formality
488 SOCIAL I NFL UENCES
than -ing though both are signs of formality.
The "formality" index of a given text would be
determined by the variant chosen in several
series of socio-symbolic variants, each of which
would have a different socio-symbolic level with
respect to formality. Presumably these series
could be ordered in terms of i ncreasingly
greater thresholds of formality required to
bring about the shift from the informal to the
formal form.
I have been stressing here the synchronic
implications of socio-symbolic variants. The
diachronic implications are at least equally
interesti ng. Obviously the threshol d for a
given variant does not necessarily remain the
same, generation after generation. If a particular
variant has for whatever reason greater prestige,
it will gradually be adopted in more situations
by more people: its threshold will be lowered.
But as its threshold is lowered and approaches
universality in the speech community, its socio-
symbolic load is reduced and eventually
vanishes. One could hardly convey much of an
air of informality, for example, by saying a
for the indefinite article, though saying ey would
be quite stilted. But presumably new series of
variants keep arising to replace those which
achieve uniformity in this way.
Now what is meant by "variants of greater
prestige" ? One could determine which of a
pair of variants had the greater prestige by
noting which tended to "spread" when two
conversants who in other situations differed in
their choice came together. But the grounds of
prestige clearly vary according to individuals
and societies. A variant which one man uses
because he wants to seem dignified another
man would reject because he did not want to
seem stiff. Societies likewise have characteristic
average value preferences. Using the variable of
formality, it is quite possible that one society
would show a tendency, at least in some
situations, to show a preference for adoption
of formal forms of speech, and another in anal-
ogous situations show a preference for informal
forms. These preferences could in turn be
related by persons so inclined to social structure.
One would end up with a statement not simply
of the direction of linguistic drift, but what this
drift meant psychologically and what social
changes might check it. It would be very
interesting, for instance, to find and examine
cognate variants from some related societies
with differing descent practices, and see whether
the current drift is in the direction of feminiza-
tion or masculinization. Such data would not
only illuminate the mechanism of linguistic
drift, but would provide students of social
structure with extremely valuable indices of
the distribution of envy and cross-segmental
identification in the communities speaking the
language studied.