Journal of Language and Social Psychology

http://jls.sagepub.com/

See you Online: Gender Issues in College Student Use of Instant Messaging
Naomi S. Baron
Journal of Language and Social Psychology 2004 23: 397
DOI: 10.1177/0261927X04269585

The online version of this article can be found at:

http://jls.sagepub.com/content/23/4/397

Published by:

http://www.sagepublications.com

Additional services and information for Journal of Language and Social Psychology can be found at:

Email Alerts: http://jls.sagepub.com/cgi/alerts

Subscriptions: http://jls.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://jls.sagepub.com/content/23/4/397.refs.html

>> Version of Record - Nov 10, 2004

What is This?

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

10.1177/0261927X04269585 ARTICLE
JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004
Baron / IM AND GENDER

SEE YOU ONLINE

Gender Issues in
College Student Use
of Instant Messaging

NAOMI S. BARON
American University

Instant Messaging (IM) is becoming a mainstay for online one-to-one communication.

Although IM is popularly described as a written version of informal speech, little empiri-
cal investigation of the linguistic nature of IM exists. Moreover, although gender issues are
being addressed for one-to-many forms of computer-mediated communication, we have no
comparable studies of IM. This article offers a linguistic profile of American college stu-
dent IM conversations. In addition to analyzing conversational scaffolding and lexical
issues, the article identifies gender divergences in IM usage. Some differences reflect com-
monly reported functional gender distinctions in face-to-face spoken conversation; other
differences indicate gender-based attitudes toward the importance of language standards
in speech and writing.

Keywords: instant messaging; IM; gender; computer-mediated communication

“See you online,” said one teenaged girl to another as they left school
for the day, each headed for her respective home—and connection to
the Internet. By “online,” the speaker meant Instant Messaging (IM),
through which a rapidly growing number of American teenagers com-
municate with one another (Lenhart, Rainie, & Lewis, 2001). Much has
been written in the popular press (e.g., Helderman, 2003; Lee, 2002)
about teenage use of IM. Although there are a handful of empirically
based studies of teenage IM (e.g., Boneva, Quinn, Kraut, Kiesler, &
Shklovski, in press; Grinter & Palen, 2002; Randall, 2002; Schiano
et al., 2002), there is almost no research on the linguistic characteris-
tics of these conversations (Jacobs, 2003, is an exception). Even less is
known about IM behavior as these users move into college and adult-
hood. Jones (2002) notes the popularity of IM on American college cam-
puses, though outside of Hård af Segerstad’s work (2002) on a form of
IM used in a specialized Swedish setting, we lack empirical studies of

AUTHOR’S NOTE: The author is grateful to Joe Walther for his valuable input to the final
version of this article.
JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY,
Vol. 23 No. 4, December 2004 397-423
DOI: 10.1177/0261927X04269585
 2004 Sage Publications
397

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

398 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

the linguistic nature of IM conversations constructed by perhaps one of

the greatest populations of its users, college students.
Instant messaging is one form of the larger online phenomenon of
computer-mediated communication (CMC). The term CMC refers to a
cluster of interpersonal communication systems used for conveying
written text, generally over the Internet. The two major parameters
across which types of CMC most significantly differ are first, whether
they are synchronous or asynchronous (i.e., whether or not transmis-
sion is essentially instantaneous and interlocutors are assumed to be
physically present to read messages and respond to them) and second,
whether the communication is one-to-one (i.e., between two people) or
one-to-many (i.e., one person’s message is broadcast to multiple po-
tential interlocutors). Schematically, the four classes of CMC are as
follows:

Synchronous Asynchronous

one-to-one IM e-mail, texting on mobile phones

1
one-to-many Chat, MUDs, MOOs, listservs, newsgroups
computer conferencing

In one-to-one forms of CMC (with the partial exception of e-mail), the

interlocutors generally know one another, whereas with one-to-many
forums, they often do not.
Linguists and other social scientists have been studying CMC for
well over a decade (e.g., Journal of Computer-Mediated Communica-
tion; Baron, 1984, 1998, 2003; Collot & Belmore, 1996; Crystal, 2001;
Ferrara, Brunner, & Whittemore, 1991; Herring, 1996, 2002; Maynor,
1994; Yates, 1996). However, we need to be aware that each type of
CMC has its own usage conditions and therefore, each needs to be ana-
lyzed in its own right. These usage conditions may, in turn, influence
the character of language produced in that medium (e.g., formal versus
informal, collaborative versus aggressive, verbose versus terse, edited
versus scattershot, informative versus whimsical).
One important question about the language we find in CMC is the
extent to which male and female users of the medium manifest differ-
ent linguistic patterns. Within the field of sociolinguistics, the topic of
gender differences in language has a long history (for reviews of the lit-
erature, see Cameron, 1998; Eckert, 1989; Holmes & Meyerhoff, 2003).
Most studies are based on spoken-language behavior, though a small
body of research has examined the possible effects of gender on written
style. Internet researchers have also been exploring gender-based cor-
relates of online behavior. Nearly all of the work to date has drawn
upon one-to-many data sources (e.g., Chat, listservs, computer con-
ferencing). We know very little about the kinds of language that male
versus female interlocutors create in one-to-one online venues such as

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 399

e-mail (though see Boneva & Kraut, 2002; Thomson, Murachver, &
Green, 2001) or IM.
The empirical goals of this article are twofold. The first is to de-
velop a linguistic profile of IM conversations used by contemporary
American undergraduate students. The second is to determine
whether the linguistic patterns in these conversations reveal gender-
based distinctions.
This study begins by outlining a linguistic framework for analyzing
IM as a form of CMC. The following section summarizes relevant liter-
ature on gender and speech, gender and writing, and gender and CMC.
The bulk of the article is devoted to an empirical analysis of an IM cor-
pus collected in spring 2003, including discussion of linguistic domains
that reveal gender differences. The article concludes by examining the
extent to which IM conversational behavior reflects or diverges from
gender-based language patterns previously reported in spoken and
traditional written venues.

INSTANT MESSAGING AS A FORM OF

COMPUTER-MEDIATED COMMUNICATION

Technologically, IM is a synchronous form of one-to-one CMC. Al-

though only two people are engaged in each conversation, users may
conduct multiple IM conversations simultaneously or IM while using
their computers for other functions (e.g., word processing, Web
searches, commercial transactions) at the same time, not to mention
engaging in simultaneous offline activities (e.g., talking, watching tele-
vision, eating). Therefore, it is common for interlocutors not to have one
another’s undivided attention, as they might better command in face-
to-face conversations. The current version (spring 2004) of America
Online Instant Messenger (AIM), which is the IM system used by the
majority of American college students, informs users when someone is
in the process of typing a message to them. However, the potential
recipient does not see any of the actual message until it has been sent.
As a form of conversation between two individuals, IM exchanges
can, in principle, be viewed in terms of the same kinds of linguistic vari-
ables as used in analyzing spoken face-to-face discourse between two
interlocutors (see Sacks, Schegloff, and Jefferson, 1974; Schegloff and
Sacks, 1973; Schiffrin, Tannen, and Hamilton, 2001, for discussion of
the analysis of spoken language discourse). For our purposes, we divide
such variables into four groups: individual turns, combining turns into
conversational sequences, openings and closings, and conversation
management.

Individual turns. The simplest way of looking at IM conversations is

at the level of individual turns. We can ask such quantifiable questions

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

400 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

as how long these turns are (in words), how many turns there are per
minute, and how often turns are made up of a single word rather than
multiple words. In examining the words (and word-like symbols) that
make up individual turns, we can consider such issues as how many
words are abbreviated or contracted, how many emoticons are used,
what kind of punctuation appears, or how frequently we find spelling
errors or self-corrections.

Combining turns into conversational sequences. The next step up in

discourse complexity is to consider how turns are combined into
sequences and conversations. For example, we need to look at
sequences of turns used by a sender (i.e., without interruption from his
or her interlocutor), at how users break up single sentences into multi-
ple turns, and at how the larger conversation is structured (e.g., how
long are IM conversations—with respect to both total number of turns
and clock time).

Openings and closings. An important subset of discourse construc-

tion is how interlocutors open and close their conversations. Are there
special formulae used within a dyad or a larger discourse group? How
many turns (and/or how much time on the clock) does it take to initiate
or terminate an IM exchange?

Conversation management. The most complex level of IM discourse

analysis is that of conversation management. How do the interlocutors
structure the body of their conversation? Some of the variables at issue
are the same as in spoken conversation: How do you take and hold the
floor? What constitutes a felicitous reply to an interlocutor’s utter-
ance? When may you interrupt a chain of transmissions (or a chain of
thought)? In addition, IM technology introduces the possibility of mes-
sage transmissions overlapping, with the result that a new conversa-
tional thread may be introduced before an earlier topic is concluded.
How do interlocutors sort out multiple simultaneous conversation
threads?
Beyond these four discourse domains lie various meta-issues. One of
these is the degree of attention to the conversation at hand. Many peo-
ple sometimes use IM asynchronously, wandering away from the com-
puter for minutes or even hours at a time (i.e., while still logged on),
conducting multiple simultaneous conversations, and even editing an
IM turn they have typed before actually sending it (Nardi, Whittaker,
& Bradner, 2000; Gloria Jacobs, personal communication, October
2004). Other meta-issues affecting IM discourse originate in variables
going beyond the scope of the conversation itself. Among these vari-
ables are age of the interlocutors, context of usage (e.g., personal ver-
sus business), conditions of usage (e.g., private messaging sent from an
office computer; teenagers with parents looking over their shoulders),

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 401

local or national culture, extent of experience with the medium, and, of

course, gender.

GENDER AND LANGUAGE

Linguists have long recognized that a person’s gender may affect his
or her linguistic productions. At the most obvious level, some lan-
guages of the world restrict particular lexical, phonological, or gram-
matical usage patterns to males or females. In Australia, for example,
there is a group of aboriginal women who use a sign language that
males are forbidden to learn (Kendon, 1980). Other gender differences
are the result of more subtle acculturation. For example, it is com-
monly reported that females tend to use more politeness indicators
than males (e.g., Coates, 1993), whereas men more frequently inter-
rupt women than vice versa (e.g., Tannen, 1994). Some of these differ-
ences appear not just in the United States or the West more generally,
but cross-culturally (Chambers, 1992; Holmes, 1993). If we hope to
understand potential gender-based differences in IM conversations,
it behooves us to understand how gender distinctions work in offline
language.
Gender analysis of IM is further complicated by the fact that IM
(like other forms of CMC) is arguably based on both spoken and writ-
ten premises. That is, although the messages themselves are physi-
cally written and bear some characteristics of written language, they
also have important resemblances to speech (Baron, 2000; Crystal,
2001). We therefore need to look not only at the literature on gender
and speech but also at studies on gender and writing. In addition, the
growing body of research on gender issues in one-to-many CMC plat-
forms may help illuminate gender analysis of a one-to-one CMC plat-
form such as IM.

SPEECH AND GENDER

The majority of studies concerning language and gender have been

based upon the analysis of spoken language, drawing upon direct
observation, interviews, or transcriptions appearing in large-scale cor-
pora. Two of the themes most commonly addressed in these studies
have been first, the use of social versus more neutrally informative
speech, and second, the extent to which speech adheres to normative
standards.

Social versus informative. Sociolinguists (e.g., Cameron, 1998;

Coates, 1993; Eckert & McConnell-Ginet, 2003; Holmes, 1993;
Romaine, 2003; Tannen, 1994) have commented on the tendency of
women (largely, though not exclusively in the West) to use conversa-

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

402 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

tion predominantly as a tool for facilitating social interaction, whereas

their male counterparts are more prone to use conversation for convey-
ing information. In Holmes’ (1995, p. 2) words, whereas women “use
language to establish, nurture and develop personal relationships,”
men’s use of conversation is more typically “a means to an end.” These
two social constructs (social versus informative) derive from observa-
tions of concrete linguistic features in which the speech of at least
many women diverges from that of at least many men. For example,
women tend to use more affective markers (e.g., “I know how you feel”),
more diminutives (e.g., “little bitty insect”), more hedge words (e.g.,
perhaps, sort of), more politeness markers (e.g., “I hate to bother you”),
and more tag questions (e.g., “We’re leaving at 8:00 pm, aren’t we?”)
than do men. Men, on the other hand, are likely to use more referential
language (e.g., “The stock market took a nosedive today”), more profan-
ity, and fewer first-person pronouns than are women.
Another way of looking at the social function of language is to ask
how much talking takes place. Do women (as commonly assumed) talk
more than men? Multiple findings reported by Holmes (1993) suggest
that males rather than females tend to dominate public conversations
among mixed-sex participants. Although there is now a growing litera-
ture on same-sex dyadic conversation between friends (e.g., Coates,
1998), there seems to be a paucity of studies specifically comparing
same-sex male and female verbosity, particularly among adolescents
or young adults.

Standard versus substandard usage. A second common finding in

the gender and language literature has been that on average, women’s
speech reflects standard phonological, lexical, and grammatical pat-
terns more than men’s does (e.g., Chambers, 1992; Holmes, 1993;
James, 1996; Labov, 1991). More detailed analysis has revealed this
pattern to be especially strong among lower middle-class females ver-
sus males (see Labov [1991] for a summary of the evidence). A variety
of explanations have been offered for the gender discrepancies: that
women are socialized to speak more “correctly”; that women are inter-
ested in social mobility; that at least among children and young adults,
there is a biological basis for women’s higher verbal abilities (Cham-
bers, 1992); or that because women tend to have less social power than
men, language is one of the few domains in which women can exercise
social superiority. Without entering into this larger debate, we can note
that especially in the modern West, women’s social roles in the family
and in the public sphere have made acquisition and use of standard
language patterns personally advantageous. On one hand, because
women do the majority of the child rearing, they can model standard
language usage for their progeny. On the other hand, in much of the
20th century, when women’s professional choices were largely circum-
scribed, the positions that were broadly open to women (e.g., teacher,

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 403

secretary, airline stewardess) required their incumbents to be well-

spoken.

WRITING AND GENDER

There are a small number of studies pertaining to gender differ-

ences in written language. Some of the data analyzed are historical in
nature (relying heavily upon personal letters), whereas other data
derive from large-scale written corpora or experimental essay compo-
sition tasks. In most empirical analyses, the same two themes we
looked at regarding gender and speech have been judged to be signifi-
cant, namely the extent to which texts function socially or referentially,
and the extent to which writers adhere to normative standards. How-
ever, rather than organize our synopsis of the literature in terms of
these two categories, we will first look at findings by genre (specifically,
personal letters) and research agenda (identifying gender via textual
analysis), and then consider national measurements of children’s writ-
ing skills in the United States. We will then summarize the findings in
terms of the social versus informational and standard versus non-
standard dimensions.

Personal letters. Biber and his colleagues (Biber, 1988; Biber,

Conrad, & Reppen, 1998; Biber & Finegan, 1997) have studied speech
and writing by using corpus-based data drawn from a variety of ven-
ues, enabling them to analyze register variation across language-use
contexts. The five major dimensions of analysis Biber identified were
(a) involved versus informational production, (b) narrative versus
nonnarrative discourse, (c) elaborated versus situation-dependent ref-
erence, (d) overt expression of argumentation, and (e) impersonal ver-
sus nonpersonal style. Each dimension represents a cluster of linguis-
tic features. For example, “involved versus informational production”
measures such variables as use of present-tense verbs, appearance of
first- and second-person pronouns, use of contractions, and use of “pri-
vate” verbs such as think or feel. Note that the “involved versus infor-
mational” dimension roughly parallels the “social versus informative”
dichotomy that has commonly been used to distinguish (at least statis-
tically) between female versus male speech patterns.
Biber et al. (1998) applied this linguistic framework to a corpus of
personal letters written by both men and women over the course of four
centuries (from the 17th through the 20th), particularly with regard to
the dimension “involved versus informational.” Included in their sam-
ple were letters written by women to women (FF), by women to men
(FM), by men to women (MF), and by men to men (MM). Although the
number of letters analyzed was small (especially in the FF and FM
samples), Biber et al. report some interesting gender differences, as
well as changes in gender patterns over time. In both the 17th and the

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

404 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

20th centuries, personal letters written by women (i.e., FF and FM)

showed a higher index of “involved” language than did letters written
by men (i.e., MF and MM).

In a related study, Palander-Collin (1999) examined personal letters

written by men and women in the 17th century. Focusing on the phrase
“I think,” which combines two elements of Biber’s (1995) dimension of
involvement (i.e., the first person pronoun I and the private verb
think), Palander-Collin confirmed findings by Biber et al. (1998) that
personal letters written by women showed higher levels of interper-
sonal involvement than did letters written by men.

Identifying gender via textual analysis. Personal letters closely

approximate the conditions of face-to-face speech. Both linguistic
transactions take place between two interlocutors who not only know
one another but typically share personal experiences. Moreover, at
least paradigmatically, both personal letters and conversation tend to
be more informal than formal. We therefore expect (and find) a signifi-
cant amount of “involved” behavior in both instances (Biber, 1995),
although female levels tends to be higher than those of males.
Does gender distinction carry over into other written venues? Mulac
and Lundell (1994) studied impromptu essays that college students
were asked to write describing landscape scenes that were projected
onto a large screen. Drawing upon earlier work in the language and
gender literature, researchers coded the essays with respect to 17 lin-
guistic features, including “male language variables” (e.g., references
to quantity, judgmental adjectives, elliptical sentences, locatives, and
sentence-initial conjunctions or filler words) and “female language
variables” (e.g., intensive adverbs, references to emotion, dependent
clauses, sentence-initial adverbials, uncertainty verbs, hedges, and
long mean-length sentences). The authors found that the analysis of
the essays with respect to gender-coded language variables correctly
identified the essay-writer’s gender 72.5% of the time.
More recently, Argamon, Koppel, Fine, and Shimoni (2003) created
an algorithm for identifying a writer’s gender by using computers to
analyze written texts. The investigators compiled lists of linguistic fea-
tures they judged to be indicative of female writers (e.g., use of pro-
nouns) or of male writers (e.g., use of determiners, quantifiers, and car-
dinal numbers). The authors note that these feature sets align with
Biber’s (1995) distinction between “involved” versus “informational”
styles. Applying their algorithm to new samples, the algorithm cor-
rectly identifies the writer’s gender approximately 80% of the time
(Koppel, Argamon, & Shimoni, 2002).

Standardized tests of written language skills. A very different venue

for assessing gender differences in written language is standardized

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 405

achievement tests. In the United States, the best-known yardstick of

precollege academic achievement is The Nation’s Report Card, com-
piled by the National Assessment of Educational Progress (NAEP),
under the auspices of the National Center for Educational Statistics
(2002). Over the years, girls have consistently outpaced boys on the
writing component of the testing rubric. The 2002 study reports that
for students tested in grades 4, 8, and 12, females continued to outscore
their male counterparts, with the score gap between females and
males being greatest in 12th grade. Evaluation was done using a 6-
point scale (ranging from unsatisfactory to excellent) that looked at
each writing sample as a whole, rather than giving separate scores on
individual dimensions (e.g., level of vocabulary, development of an
argument, length, or writing mechanics).

Summary of writing and gender findings. We can look at our find-

ings regarding writing and gender in terms of the two dichotomies we
identified for analyzing speech and gender: namely, social versus infor-
mative and standard versus nonstandard. All of the corpus-based
studies and the essay study support the speech-based finding that in
general, female language is more social (“involved”) in character than
is male. None of the studies we reviewed directly addressed the issue of
standard versus nonstandard language usage. However, the NAEP
reports can be seen as providing supporting evidence for the general
claim that the written language of females shows greater command of
the “standard” than does the written language of males.

GENDER AND CMC

The field of CMC has sparked interest from sociolinguists concerned

with the extent to which traditional gender dichotomies in face-to-face
communication would play out in online behavior. Initial optimism
that the relative anonymity of the medium would support an equaliza-
tion of gender roles (e.g., Danet, 1998) soon gave way to the realization
that online dynamics often replicated offline gender distinctions.
Herring (2003) offers a thorough analysis of language and gender
issues in one-to-many CMC forums such as listservs and newsgroups
(both of which involve asynchronous communication) and Chat,
MUDs, and MOOs (all of which involve synchronous communication).
In both venues, Herring reports gender asymmetries. On asynchron-
ous discussion lists and newsgroups, “males are more likely to post lon-
ger messages, begin and close discussions in mixed-sex groups, assert
opinions strongly as ‘facts,’ use crude language (including insults and
profanity), and in general manifest an adversarial orientation toward
their interlocutors,” whereas females “tend to post relatively short
messages, and are more likely to qualify and justify their assertions,
apologize, express support of others, and in general, manifest an

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

406 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

‘aligned’ orientation toward their interlocutors” (Herring, 2003, p.

207).
In one-to-many synchronous CMC forums, gender roles are in some
ways more balanced (e.g., more equal participation in Chat environ-
ments as measured by number of messages and message length; see
Herring, 1999). However, Herring (2003) goes on to note that gen-
der differences (and often inequalities) still pervade Chat and social
MUDS or MOOs. For example, males use more aggressive and insult-
ing speech acts, whereas females type three times as many represen-
tations of smiles or laughter. Male discourse is oppositional and ad-
versarial, whereas female discourse style is aligned and supportive
(Herring, 2003).
In considering the literature on one-to-many CMC gender issues, we
need to be mindful that the social conditions for one-to-one CMC are
quite different. Most of the one-to-many forums that have been studied
are open to large numbers of users, the majority of whom a given inter-
locutor does not personally know. Moreover, in many such forums,
users can anonymize their postings, representing themselves as being
of an age, gender, and so on that does not correspond to who they are in
real life. In one-to-one venues such as IM, e-mail, or texting on mobile
phones, conversation typically takes place between interlocutors who
know one another. IM conversations generally involve dyads who have
ongoing face-to-face relationships that are supplemented by online
discourse. Thus, although some of the gender findings from the body of
one-to-many CMC literature may prove relevant for a gender analysis
of IM language (e.g., use of emoticons representing laughing or smil-
ing), other variables (e.g., dominance of a conversation) may be less
applicable.
Given this background on the nature of IM as a form of CMC and on
previous studies of language and gender, we now turn to the IM corpus
upon which our own analysis is based.

EMPIRICAL STUDY OF IM:

CORPUS, TERMINOLOGY, VARIABLES

An IM corpus was collected in April 2003 from 22 college-aged stu-

dents who were attending school or had graduated the semester before
the study was undertaken. This pilot study was part of a larger, ongo-
ing project comparing online and face-to-face communication among
American college students. The discussion that follows is restricted to
a large subset of linguistic features for which data analysis has been
completed.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 407

THE IM CORPUS

Using America Online’s freely downloadable program known as

AIM (AOL Instant Messenger), IM conversations were initiated by a
cohort of current (or recently graduated) students at American Univer-
sity in Washington, D.C. American University is an academically com-
petitive, largely residential institution with approximately 5,500
undergraduates. These conversational initiators (“student experi-
menters”) were asked to IM specified numbers of male and female
peers who were on the initiator’s AIM Buddy List. (A Buddy List is a set
of IM screen names that a user chooses to associate with his or her
account. When one of these users logs on to AIM, he or she can see
which people on the Buddy List are currently online.)
At the beginning of each conversation, initiators requested permis-
sion of their interlocutor to save the IM conversation that was to follow
so that it could be forwarded to the author for research purposes. For-
mal consent forms were distributed electronically to all parties (stu-
dent experimenters and their conversational partners) at the end of
the IM conversation. Both members of each conversation were given
the opportunity to edit out any words or turns they wished to delete (an
option rarely taken), and user screen names were anonymized. Stu-
dent experimenters then electronically forwarded the IM conversation
files (and consent forms) to a site established for project analysis. Ini-
tial discussion in the IM conversations regarding the research project
was eliminated from actual data analysis, thus precluding an analysis
of conversational openings.
The corpus consisted of 23 distinct IM conversations. A total of 9 con-
versations took place between females (FF) and 9 between males
(MM). An additional 5 conversations involved female/male dyads
(FM). In a number of the FF and FM conversations, a single student
experimenter had conversations with several people on his or her
Buddy List. Because several student experimenters withdrew from
the project and could not be replaced, most of the MM conversations
were between the same two interlocutors.
Taken collectively, the 23 IM conversations contained a total of
2,185 conversational turns, made up of 11,718 words. Some of the anal-
yses discussed below were performed on the entire corpus whereas oth-
ers were restricted to comparison of the 9 FF and 9 MM conversations
(together totaling 1,861 conversational turns).

TERMINOLOGY

As yet, there is no generally accepted linguistic terminology for ana-

lyzing IM data. Therefore, we established the following definitions:

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

408 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

Turn: composition (i.e., by typing) and transmission of an instant message.

Max: hey man
Utterance: rough equivalent of a sentence in IM.
Susan: Somebody shoot me!
Sequence: a number of IM turns in a row (from 1 to N) sent seriatim by the
same interlocutor.
Max: hey man
Max: whassup [sequence = 2 turns]
Utterance chunking: breaking a single IM utterance into two or more turns.
Joan: that must be nice
to be in love
in the spring
Closing: series of turns (between interlocutors) at the end of an IM conver-
sation, beginning with one party initiating a closure of the conversation
and ending with termination of the IM connection.
Sam: Hey, I gotta go [first indication that will terminate conversation]
[subsequent conversational turns]
Sam: I’m outta here [final turn in conversation]

LINGUISTIC VARIABLES ANALYZED

A range of linguistic variables were analyzed. The first three cate-

gories involved conversational scaffolding (essentially, how conver-
sations were constructed). The first category was turns, analyzed in
terms of average turn length (in words), longest turns (in words), per-
centage of one-word turns (out of total turns), and turns per minute.
Second, sequences were analyzed as longest sequence per conver-
sation, average number of turns per sequence, and percentage of
sequences including >1 turn. Third, conversation length was analyzed
for entire conversations (turns per conversation; time), and closings
(number of turns; time).
We also considered a number of lexical issues that constitute short-
enings of words or phrases (abbreviations, acronyms, and contractions)
as well as looking at emoticons (i.e., graphic images that are intended
to serve as sentence modifiers, expressing the sender’s attitude toward
the text preceding it). Tabulations of all linguistic variables were per-
formed both for the entire corpus and by gender.

GENERAL LINGUISTIC PROFILE OF IM

The entire corpus was coded for each of the relevant variables, with
instances of each variable then tallied and analyzed by hand. Despite
some design and implementation limitations (including the overall
size of the corpus and the skewed male sampling), the linguistic profile
of IM that emerged clearly suggests a number of distinctive patterns in
American college student IM conversations, as well as suggestive dif-
ferences between male and female use of the medium. In this section,

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 409

Table 1
Turns Summary

Average turn length (in words) 5.4

Longest turn (in words) 44.0
Percentage of one-word turns (out of total turns) 21.8%
Average turns per minute 4.0

Note. Average turn length and turns per minute were calculated over the entire database
of 2,185 turns. However, longest turns and percentage of one-word turns were calculated
over just the female to female (FF) and just the male to male (MM) conversations (total-
ing 1,864 turns).

we will review the findings for the overall corpus with respect to con-
versational scaffolding and lexical issues. In the following section, we
consider these same domains with regard to gender.

CONVERSATIONAL SCAFFOLDING

Turns. Table 1 summarizes the structural nature of individual

turns. The aggregated data suggest that average IM turns are fairly
short (5.4 words per turn). However, averages obscure some of the
important characteristics of contemporary college student IM conver-
sations. One of these characteristics is that a number of IM turns are
quite long indeed (the longest in the corpus being 44 words). A second
characteristic is the high proportion of one-word turns (21.8% of the FF
and MM combined corpus).
Because all of the turns in the IM conversations were time-stamped
(a feature available directly through AIM), it was possible to calculate
not only how long each conversation lasted but also how many turns
took place per minute. The average—barely 4 turns per minute—is
surprisingly low, considering how few seconds it takes to type an aver-
age of 5.4 words and transmit the message.

Sequences. For counting purposes, a sequence was defined as one or

more consecutive terms from the same interlocutor. If a sender’s multi-
ple turn sequence was interrupted by a message from his or her inter-
locutor but the sender did not attend to the interruption, all consecu-
tive turns dealing with the same theme were considered to be part of a
single sequence. Table 2 summarizes the findings regarding
sequences.
The first step in the analysis was to cluster all the data into
sequences. There were 1,292 sequences, built out of a total of 2,185
turns. (Recall that sequence is here defined as one or more consecu-
tive turns, so single turns are included in this statistic.) Sequences
ranged in length from 1 to 18 turns. The average number of turns per
sequence was 1.7. However, 42% of the IM sequences contained more

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

410 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

Table 2
Sequences Summary

Number of sequences in total database 1,292

Longest IM sequence per conversation
(i.e., from same interlocutor) 18 sequential turns
Average number of turns per IM sequence 1.7
Percentage of IM sequences including >1 turn 42.0%

Table 3
Conversation and Closing Length

Entire Conversations Mean

Average turns per conversation 93.8 turns

Average time per conversation 23.7 minutes
Closings Mean

Average number of turns to close 7.0 turns

Average time to close 31.9 seconds

than one turn (543 multiturn sequences out of 1,292 total sequences).
If we eliminate the one-word turns (which constitute 21.8% of the total
conversational turns but are rarely part of multiturn sequences),
multiturn sequences constitute nearly half of the IM discourse.

Conversation length. Table 3 presents findings on number of turns

and chronological length of both entire conversations and closing
sequences.
Aggregated data suggest that IM conversations are fairly lengthy:
more than 93 turns apiece on average and nearly 24 minutes long. In
reality, IM conversations show enormous variety, ranging from quick
three- or four-turn volleys to conversations stretching over more than
200 turns and well over an hour. Moreover, although the communica-
tion channel may technically be open for an extended period of time,
the two interlocutors are not necessarily directly engaging with one
another over the entire IM conversational span. For example, in one
lengthy FF conversation (142 turns, 88 minutes), there was a 15-
minute gap when no turns occurred.
The data on IM conversational closings suggest that much as in
real-life encounters, interlocutors who know one another reasonably
well often take awhile before actually terminating a conversation.
Here is an example of a typical multiturn IM conversational closing
between female interlocutors. The sequence took 19 seconds:

Gale: hey I gotta run

Sally: Okay.
Sally: I’ll ttyl?
Gale: gotta do errands.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 411

Gale: yep!!
Sally: Okay.
Sally: :)
Gale: talk to you soon
Sally: Alrighty.

The median closing sequence took seven turns and averaged nearly 32
seconds.

LEXICAL ISSUES

We turn now to the four lexical issues examined in this study: abbre-
viations, acronyms, contractions, and emoticons. As we will see, the
data argue for a somewhat different profile of IM (at least as used by
American college students) than the image presented in the popular
press of a medium filled with lexical short-cuts and emoticons, not to
mention chaotic punctuation and devil-may-care spelling.

Abbreviations. Table 4 summarizes the use of what we might call

CMC abbreviations (i.e., abbreviations that appear to be distinctive to
CMC communication). Excluded from this tabulation are abbrevia-
tions that although appearing in CMC messages, are also part of com-
mon offline written usage (e.g., hrs = hours) or are direct representa-
tions of spoken usage (e.g., cuz = because). Admittedly, the line between
common offline and CMC-specific usage is sometimes difficult to draw
(e.g., b/c for because was included in the tally of CMC abbreviations,
whereas prob for problem or convo for conversation was not).
Considering that the IM corpus contained nearly 12,000 words (i.e.,
tokens), the fact that only 31 CMC abbreviations appeared (less than
.3% of the corpus) is noteworthy.

Acronyms. Table 5 summarizes the use of what we might call CMC

acronyms (i.e., acronyms that appear to be distinctive to CMC commu-
nication). Excluded from this tabulation are acronyms that, although
appearing in CMC messages, are also part of common offline written
usage (e.g., US = United States or TA = teaching assistant).
As in the case of CMC abbreviations, the limited number (and vari-
ety) of CMC acronyms is small (90 tokens, which represents less than
.8% of the total words in the corpus). The most frequent acronym used
was lol (laughing out loud), an acronym that has even found its way
into spoken usage among some college students. Interestingly, how-
ever, the IM (or even spoken) term is not always used to indicate the
humorous response suggested by the words “laughing out loud.”
Rather, both lol and heehee (or haha) are commonly used as phatic fill-
ers for the equivalent of OK, cool, or yeah. Thus, we find a contrast
between “literal” CMC meaning, for example,

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

412 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

Table 4
Computer-Mediated Communication (CMC) Abbreviations

Total CMC abbreviations 31 (out of 11,718 words)

Type/token distribution (arranged alphabetically)
bc (also b/c) = because 5
bf = boyfriend 2
cya = see you 7
k = OK 16
y? = why 1

Table 5
Computer-Mediated Communication (CMC) Acronyms

Total acronyms 90 (out of 11,718 words)

Type/token distribution (arranged alphabetically)
brb = be right back 3
btw = by the way 2
g/g (also g2g) = got to go 2
LMAO = laughing my _____ off 1
lol (also LOL) = laughing out loud 76
OMG = Oh my God 1
ttyl = talk to you later 5

Laura: What’s your favorite dish to make?

Cynthia: hehehe
Cynthia: chocolate mousse
Laura: freedom fries?
Cynthia: LOL

and phatic meaning, for example,

Mark: i’ve got this thing that logs all convos [=conversations] now
Jim: really?
Jim: why’s that
Mark: i have ever [=every] conversation i’ve had with anybody since
the 16th
Mark: i got a mod [=module] for aim [=AIM], and it just does it
Mark: i’m not sure why
Jim: lol
Jim: cool

Contractions. Much as popular conceptions of IM lead us to antici-

pate a high frequency of textual shortening through use of abbrevia-
tions and acronyms, we might expect writers to make use of contrac-
tions wherever the language permits (e.g., I’m instead of I am or he’s
rather than he is). However, just as the college IM corpus yielded sur-
prisingly few CMC abbreviations or acronyms, an analysis of tokens
representing potential contractions suggests that again, many users

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 413

Table 6
Use of Contracted Versus Uncontracted Lexical Pairs

Contracted 498
Uncontracted 265
Total 763
% Contracted (of total) 65.3%

Note. This analysis is based upon the 9 female to female (FF) and 9 male to male (MM)
conversations.

Table 7
Emoticon Summary

Total emoticons 49
(out of 11,718 words)
Type/token distribution
(by descending frequency)
:-) = smiley 31
:-( = frowny 5
O:-) = angel 4
:-P = sticking out tongue, with nose 3
;-) = winking 2
:-\ = undecided 1
:-[ = embarrassed 1
:P = sticking out tongue, without nose 1
:- = [probably a typographical error] 1

are not taking advantage of possible lexical shortcuts. Table 6 summa-

rizes the occurrences of contracted versus uncontracted lexical pairs
occurring in the corpus.
Contracted forms constitute the majority of usages of contraction
pair choices. In 65.3% of situations, the contracted option was chosen.
However, the finding that in nearly 35% of the linguistic situations
offering an option, users chose the uncontracted “long form” (e.g., I am,
he is, do not, they are) was unexpected.

Emoticons. Table 7 summarizes all types and tokens of emoticons

appearing in the full corpus of nearly 12,000 words.
Given the amount of publicity that emoticons continue to receive in
the popular press (e.g., Lorenzi, 2002), it was surprising to see how few
emoticons appeared in the entire text. Of the 49 emoticons used, 31 of
them (nearly two thirds) were a “smiley.” Moreover, a small number of
participants were responsible for using the majority of the 49
emoticons. Out of 22 interlocutors, just 3 people accounted for 33 of the
emoticons in the corpus. This pattern is consistent with Walther and
D’Addario’s (2001) finding in an e-mail experiment that the emotional
meaning emoticons convey is overshadowed by the verbiage
accompanying them.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

414 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

Table 8
Turns Summary by Gender Pairs

Gender Pairs
FF MM

Average turn length (in words) 5.3 5.2

Longest turns (in words) averaged over 9 conversations 27.6 19.6
Absolute longest turn 44 34
One-word turns as percentage of total 20.0% 24.4%
Average turns per minute 3.9 4.5

Note. Whereas Table 8 tabulates only the 9 female to female (FF) and the 9 male to male
(MM) conversations, the earlier turns summary in Table 1 calculates average turn
length and turns per minute over the entire database, but calculates longest turns and
percentage of one-word turns just over the FF and MM conversations.

GENDER ISSUES IN IM

The IM corpus was next examined with respect to gender. For con-
versational scaffolding, comparisons were made for turn structure and
conversation length. (Sequence data were not tabulated data by gen-
der.) For lexical issues, gender comparisons were done for abbrevia-
tions, acronyms, contractions, and emoticons. No gender-based pat-
terns emerged for abbreviations or acronyms, and therefore neither
issue will be discussed in this section. However, in the case of both con-
tractions and emoticons, gender appears to be a relevant variable.

CONVERSATIONAL SCAFFOLDING

Turns. Table 8 summarizes the gender comparison of the 9 FF ver-

sus the 9 MM IM conversations with regard to turn scaffolding.
On most measures involving turns (average turn length in words,
percentage of one-word turns, and average turns per minute), there
appears to be little difference between female and male usage pat-
terns. The only domain in which genders showed obvious variation is
longest turns. For the FF and MM IM conversations analyzed, if we
consider just the longest turn in each conversation (i.e., the longest
turn in each of the 9 FF conversations; the longest turn in each of the 9
MM conversations), we find that female “longest turns” (27.6 words)
were an average of 8 words lengthier than male “longest turns” (19.6
words). Moreover, the longest single turn in all of the FF conversations
was 44 words, whereas the longest single turn in all of the MM conver-
sations was 34 words.

Conversation length. Table 9 presents the gender breakdown re-

garding conversation length issues.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 415

Table 9
Conversation and Closing Length by Gender Pairs

Gender Pairs
FF MM

Entire conversations
Average turns per conversation 121.9 85.2
Average minutes per conversation 31.3 18.9

Closings
Average number of turns to close 9.8 4.3
Average seconds to close 41.0 16.3

Gender seems to make a difference with regard both to overall con-

versation length (in number of turns and clock time) and to the length
of conversational closings. With regard to overall conversation length,
FF conversations were roughly a third longer (in turns and time) than
were MM conversations, although the considerable range in turns per
conversation makes direct gender comparison difficult. When it came
to closing IM conversations, however, females clearly used a greater
number of turns, and lengthier time spans, than did males.2

LEXICAL ISSUES

Contractions. Table 10 summarizes the use of contractions in the FF

and MM corpus, by gender.
Both males and females used higher percentages of contracted than
uncontracted forms in their IM conversations (e.g., using don’t rather
than do not). However, males used a much greater proportion of con-
tracted forms than did females.3 Although males chose contracted
forms more than three quarters of the time (77.1%), females were more
evenly balanced in their choices, using contracted forms only 57% of
the time.

Emoticons. Table 11 summarizes the use of emoticons by gender for

the entire IM corpus.
A greater proportion of females used emoticons than did males.4 Of
the 16 female participants in the study, three quarters used one or
more emoticons in their IM conversations. Of the 6 male participants,
only 1 used emoticons. As for total emoticon usage by gender, 34 (of the
total 49) emotions were used by females and 15 by males. However, all
of the 15 male usages are attributable to a single male interlocutor, and
all 15 occurred when that male was in IM conversations with female
interlocutors.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

416 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

Table 10
Use of Contracted Versus Uncontracted Lexical Pairs by Gender

Gender
FF MM

Contracted 256 242

Uncontracted 193 72
Total 449 314

% Contracted (of total) 57.0% 77.1%

Table 11
Emoticon Usage by Gender

Females (n = 16) Males (n = 6)

Number of interlocutors using emoticons 12 1

Total emoticons used 34 15

IM AND GENDER:
DISCUSSION AND FUTURE RESEARCH

GENERAL LINGUISTIC PROFILE AND INTERPRETATION

Our analysis of a corpus of IM conversations constructed by Ameri-

can college students offers an initial linguistic profile of IM practices
by a subset of contemporary users. For the database as a whole, we
found that with respect to conversational scaffolding, turns vary in
length and frequency, but average 5.4 words per turn and 4 turns per
minute. The use of multiturn sequences by interlocutors is common,
constituting 42% of the corpus. Conversations vary widely in length
and typically take multiple turns to close. Looking at the entire corpus
with respect to lexical issues, we saw that the use of CMC abbrevia-
tions and CMC acronyms was minimal, with lol by far the most com-
mon acronym. In lexical situations where either a contacted or uncon-
tracted form could be used, interlocutors used contracted forms only
two thirds of the time. The use of emoticons was minimal, with the
smiley by far the most common.
The overall linguistic analysis provides evidence that IM conversa-
tions, at least as constructed by the present American college student
sample, represent a blend of both spoken and written language con-
ventions. As for speech, the conversations make relatively frequent use
of the acronym lol as a phatic filler, roughly comparable to OK, really, or
yeah in spoken discourse. Similarly, interlocutors took multiple turns
to close IM conversations, a feature familiar in face-to-face discourse.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 417

As for writing, interlocutors did not regularly avail themselves of

the common “shorthand” techniques popularly associated with IM. Not
only were CMC abbreviations and acronyms infrequent, but even con-
tractions were not as pervasive as might be expected in a medium
touted as favoring brevity. As was reported elsewhere (Baron & Ling,
2003), this IM corpus revealed other signs of a more formal written
style, including a paucity of spelling errors and presence of a number of
attempts to correct one’s own typing errors made in an immediately
preceding turn.
Equally striking was the preponderance of multiturn sequences.
Although motivations for breaking single utterances into multiple
transmissions still need to be formally investigated, anecdotal evi-
dence from students suggests three plausible explanations. The first
is technological. By transmitting pieces of a message seriatim, senders
provide some text for their interlocutors to begin reading (while the
sender is typing more of the message). This piecemeal transmission
style keeps the interlocutors from waiting and, equally important,
maintains the interlocutors’ attention so they do not shift to another
task (e.g., participating in an IM conversation with someone else, surf-
ing the Web, making a phone call). The second motivation may have to
do with layout of text on the page. It is easier (and faster) to read short
lines of text than to read long lines spanning the entire page horizon-
tally. The final motivation may be aesthetic. Several IM users have
reported that when they divide IM utterances into chunks across mul-
tiple turns, they are consciously attempting to make the results vis-
ually resemble a poem.
Perhaps the most unexpected finding was how few average turns (4)
were exchanged per minute. With turns averaging 5.4 words, an aver-
age of 21.6 words were sent per minute, which probably represents
considerably less than half the typing speed of most college students.
Anecdotal evidence indicates that the majority of college students en-
gage in multitasking while doing IM. Given the time gaps between
many turns, we can only conclude that in many instances, this “syn-
chronous” form of CMC is being used asynchronously.

GENDER FINDINGS AND INTERPRETATION

When we reexamined the corpus looking for gender-based patterns

of linguistic usage, we found several aspects of conversational scaffold-
ing in which male and female patterns differed. Whereas average turn
length was similar across genders, the longest single turns in FF con-
versations were longer than those in MM conversations. The average
FF conversation was longer (both in number of turns and in time) than
the average MM conversation. Females took roughly twice as long (in
number of turns and in time) to close conversations as did males. With
regard to lexical issues, females used more uncontracted lexical forms

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

418 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

than did males, and females were far more likely to use emoticons than
were males.
We can think about these gender findings in terms of the two broad
categories we earlier identified for looking at language and gender
issues, namely social/involved versus informative, and standard ver-
sus nonstandard.

Social/involved versus informative. All of the IM conversational

scaffolding findings indicate that female interlocutors were more
“talkative” than their male counterparts. Females took longer turns,
had longer overall conversations, and took longer to say goodbye. In
our earlier review of the sociolinguistic literature, we found no grounds
for arguing that even in same-sex dyadic face-to-face conversation,
women are more talkative than men. However, in their study of infor-
mal essay-writing, Mulac and Lundell (1994) reported that females
used longer sentences than did males. Thus, it is possible that our IM
findings reflect a female writing style rather than a female speech
style.
It is instructive to compare the IM gender findings with those
reported for one-to-many CMC communication. In the lexical domain,
our finding that females were far more likely to use emoticons than
were males is consonant with Herring’s report (2003) that women were
three times as likely to use representations of smiles or laughter than
were males in one-to-many synchronous communication. However, our
IM findings with regard to verbosity were not paralleled in one-to-
many CMC, for which Herring (1999, 2003) reported men used longer
messages in asynchronous communication and more balanced num-
bers and length of messages in synchronous CMC. Herring’s findings
more closely parallel those summarized by Holmes (1993) for public
spoken conversations among mixed-sex participants.

Standard versus nonstandard. Our finding that females used fewer

contracted forms than did males suggests that females have a greater
tendency toward treating IM as a written medium (because in the aca-
demic domain, students are still taught that contractions should not be
used in formal writing). The number of CMC abbreviations and acro-
nyms in the IM corpus was too small for drawing any gender-based dis-
tinctions. Preliminary (though incomplete) analysis of punctuation
and capitalization in the IM conversations suggests that females are
more normative than are males in that females employed more stan-
dard punctuation and capitalization than did their male counterparts.
Other studies of one-to-one CMC such as Ling’s (in press) work on Nor-
wegian short-text messaging report that female teenagers and young
adults used more standard punctuation and capitalization in their text
messaging on mobile phones than did males.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 419

FUTURE RESEARCH

The results reported in this article represent a first step toward

understanding the issue of gender in IM conversations. However, addi-
tional data collection and analysis remain to be done before we can be
more definitive about the linguistic nature of contemporary IM. First,
IM data need to be collected from a broader range of cohorts (including
college students from other types of academic settings, teenagers,
adults in the work force, etc.). Second, IM data need to be coded for a
variety of additional variables. With respect to the social versus infor-
mative dimension of language use, we need to look at such issues as
content analysis (cf. Boneva & Kraut, 2002, on content analysis of e-
mail), pronoun use (cf. reports by Argamon et al., 2003, and Biber et al.,
1998, that females use more pronouns than males), and use of apolo-
gies (cf. Herring, 2003, regarding women’s heavier use of apologies in
one-to-many CMC). With respect to the standards issue, we need to
look closely at punctuation and capitalization, as well as at spelling
errors and error correction. That is, we need to study the IM data more
extensively to determine the extent to which gender distinctions previ-
ously reported in spoken and written language are reflected in IM
conversations.
Third, we need to study the nature of multitasking behavior while
engaging in IM conversations. We need to look for possible gender dis-
tinctions here, because anecdotal evidence indicates that males may be
heavier multitaskers than females. Data on multitasking will provide
a better understanding of the extent to which IM should be thought of
as a form of synchronous conversation as opposed to being a potentially
less engaged form of communication. And fourth, to better understand
the issue of comparative female verbosity in IM conversations, it will
be important to collect both dyadic speech data and offline written
compositions from comparable cohorts so that we can judge with better
certainty the extent to which male and female IM conversations
approximate spoken or written discourse. Existing gender studies of
speech or writing may not reflect the language patterns of contempo-
rary college students and therefore may not make for appropriate com-
parisons with college student IM conversations.

CONCLUSIONS

Although the IM conversations of American college students reveal

considerable variation across users, as a whole, they evidence a low
proportion of the characteristics assumed in the popular press. Analy-
sis of average message length and average turns per minute suggests
that IM is frequently used asynchronously, in tandem with multi-

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

420 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

tasking behaviors, though the precise character of such multitasking

remains to be studied.
Unlike one-to-many forms of CMC, IM conversations contain gen-
der patterns that in some ways reflect gender distinctions in spoken
and (offline) written language. Our findings that females use longer
IM turns and fewer contractions than males suggest that female IM
conversations reflect more a female writing style than a female speech
style, though comparative samples of face-to-face conversation and of
writing by college-aged students will need to be collected and analyzed
before this hypothesis can be confirmed.

NOTES

1. MUDs and MOOs are acronyms for Multiple User Dungeons (or Dimensions or
Discussions—the term has evolved over the years) and MUDs Object Oriented, respec-
tively. The term Object-Oriented refers to a programming style.
2. Large variances associated with the means on turns per conversations (female to
female [FF] M = 121.89, SD = 82.68; male to male [MM] M = 85.22, SD = 88.96) and on
minutes per conversation (FF M = 31.33, SD = 25.57; MM M = 18.89, SD = 17.27) ren-
dered no statistically significant differences between gender pairs. For instant message
(IM) closings, however, FF/MM differences were significant both for number of turns per
closing (FF M = 9.78, SD = 5.12; MM M = 4.29, SD = 1.13), t (14) = 2.77, p = .015; as well as
seconds taken to close (FF M = 41.00, SD = 20.12; MM M = 16.29, SD = 15.48), t (14) =
2.68, p = .018.
3. Fisher’s exact test p = .022.
2
4. χ (1) = 50.44, p < .001.

REFERENCES

Argamon, S., Koppel, M., Fine, J., & Shimoni, A. R. (2003). Gender, genre, and writing
style in formal written texts. Text, 23, 321-346.
Baron, N. S. (1984). Computer mediated communication as a force in language change.
Visible Language, 18, 118-141.
Baron, N. S. (1998). Letters by phone or speech by other means: The linguistics of email.
Language and Communication, 18, 133-170.
Baron, N. S. (2000). Alphabet to email: How written English evolved and where it’s head-
ing. London and New York: Routledge.
Baron, N. S. (2003). Language and the Internet. In A. Farghali (Ed.), The Stanford hand-
book for language engineers (pp. 59-127). Stanford, CA: CSLI.
Baron, N. S., & Ling, R. (2003, October). IM and SMS: A linguistic comparison. Paper pre-
sented at the Fourth International Conference of the Association of Internet Re-
searchers, Toronto, Ontario, Canada.
Biber, D. (1988). Variation across speech and writing. Cambridge: Cambridge University
Press.
Biber, D. (1995). Dimensions of Register Variation. Cambridge: Cambridge University
Press.
Biber, D., Conrad, S., & Reppen, R. (1998). Corpus linguistics: Investigating language
structure and use. Cambridge: Cambridge University Press.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 421

Biber, D., & Finegan, E. (1997). Diachronic relations among speech-based and written
registers in English. In T. Nevalainen & L. Kahlas-Tarkka (Eds.), To explain the pres-
ent: Studies in the changing English language in honour of Matti Rissanen (pp. 253-
275). Helsinki: Modern Language Society.
Boneva, B., & Kraut, R. (2002). Email, gender, and personal relations. In B. Wellman &
C. Haythornthwaite (Eds.), The Internet in everyday life (pp. 372-403). Oxford:
Blackwell.
Boneva, B., Quinn, A., Kraut, R., Kiesler, S., & Shklovski, I. (in press). Teenage communi-
cation in the instant messaging era. In R. Kraut, M. Brynin, & S. Kiesler (Eds.), Infor-
mation technology at home. Oxford: Oxford University Press.
Cameron, D. (1998). Gender, language, and discourse: A review essay. Signs: Journal of
Women in Culture and Society, 23, 995-973.
Chambers, J. K. (1992). Linguistic correlates of gender and sex. English World-Wide, 13,
173-218.
Collot, M., & Belmore, N. (1996). Electronic language: A new variety of English. In S. Her-
ring (Ed.), Computer-mediated communication: Linguistic, social, and cross-cultural
perspectives (pp. 13-28). Amsterdam: John Benjamins.
Coates, J. (1993). Women, men, and language (2nd ed.). London and New York: Longman.
Coates, J. (Ed.) (1998). Language and gender: A reader. Oxford: Blackwell.
Crystal, D. (2001). Language and the Internet. Cambridge, UK: Cambridge University
Press.
Danet, B. (1998). Text as mask: Gender and identity on the Internet. In S. Jones (Ed.),
Cybersociety 2.0 (pp. 129-158). Thousand Oaks, CA: Sage.
Eckert, P. (1989). Jocks and burnouts: Social categories and identities in high school. New
York: Teachers College Press.
Eckert, P., & McConnell-Ginet, S. (2003). Language and gender. New York: Cambridge
University Press.
Ferrara, K., Brunner, H., & Whittemore, G. (1991). Interactive written discourse as an
emergent register. Written Communication, 8, 8-34.
Grinter, R. E., & Palen, L. (2002). Instant messaging in teen life. Proceedings of the ACM
Conference on Computer Supported Cooperative Work (CSCW ’02), New Orleans, LA,
November 16-20. New York: ACM Press.
Hård af Segerstad, Y. (2002). Use and adaptation of written language to the conditions of
computer-mediated communication. Doctoral Dissertation, Department of Linguis-
tics, Göteborg University, Göteborg, Sweden.
Helderman, R. (2003, May 20). Click by click, teens polish writing. Washington Post, p. B1.
Herring, S. (Ed.) (1996). Computer-mediated communication: Linguistic, social, and
cross-cultural perspectives. Amsterdam: John Benjamins.
Herring, S. (1999). The rhetorical dynamics of gender harassment online. The Informa-
tion Society, 15, 151-167.
Herring, S. (2002). Computer-mediated communication and the Internet. In B. Cronin
(Ed.), Annual review of information science and technology 36 (pp. 109-168). Medford,
NJ: American Society for Information Science and Technology.
Herring, S. (2003). Gender and power in on-line communication. In J. Holmes &
M. Meyerhoff (Eds.), The handbook of language and gender (pp. 202-228). Malden,
MA: Blackwell.
Holmes, J. (1993). Women’s talk: The question of sociolinguistic universals. Australian
Journal of Communication, 20, 125-149.
Holmes, J. (1995). Women, men, and politeness. New York: Longman.
Holmes, J., & Meyerhoff, M. (Eds.) (2003). The handbook of language and gender. Malden,
MA: Blackwell.
Jacobs, G. (2003, April). Breaking down virtual walls: Understanding the real space/
cyberspace connections of language and literacy in adolescents’ use of instant messag-
ing. Paper presented at the annual meeting of the American Educational Research
Association, Chicago, IL.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

422 JOURNAL OF LANGUAGE AND SOCIAL PSYCHOLOGY / December 2004

James, D. (1996). Women, men, and prestige speech forms: A critical review. In V. L.
Bergvall, J. M. Bing, & A. F. Freed (Eds.), Rethinking language and gender research
(pp. 98-125). London: Longman.
Jones, S. (2002). The Internet goes to college: How students are living in the future with
today’s technology. Pew Internet & American Life Project. Retrieved July 25, 2004,
from http://www.pewinternet.org/reports/toc.asp?Report=71
Kendon, A. (1980). The sign language of the women of Yuendumu: A preliminary report
on the structure of Warlpiri sign language. Sign Language Studies, 27, 101-112.
Koppel, M., Argamon, S., & Shimoni, A. R. (2002). Automatically categorizing written
texts by author gender. Literary and Linguistic Computing, 17, 401-412.
Labov, W. (1991). The intersection of sex and social class in the course of linguistic
change. Language Variation and Change, 2, 205-254.
Lee, J. (2002, Sept. 19). I think, therefore IM. New York Times, Circuits Section, p. G1.
Lenhart, A., Rainie, L., & Lewis, O. (2001). Teenage life online: The rise of the instant-
message generation and the Internet’s impact on friendships and family relation-
ships. Pew Internet & American Life Project. Retrieved July 25, 2004, from http://
www.pewinternet.org/pdfs/PIP_Teens_Report.pdf
Ling, R. (in press). The sociolinguistics of SMS: An analysis of SMS use by a random sam-
ple of Norwegians. In R. Ling & P. Pedersen (Eds.), Front-stage/back-stage: Mobile
communication and the renegotiation of the social. London: Springer-Verlag.
Lorenzi, R. (2002). ‘Smiley’ emoticon turns 20. Retrieved July 25, 2004, from http://
dsc.discovery.com/news/briefs/20020916/emoticon.html
Maynor, N. (1994). The language of electronic mail: Written speech? In G. D. Little &
M. Montgomery (Eds.), Centennial usage studies (pp. 48-54). Tuscaloosa: University
of Alabama Press.
Mulac, A., & Lundell, T. L. (1994). Effects of gender-linked language differences in adults’
written discourse: Multivariate tests of language effects. Language and Communica-
tion, 14, 299-309.
Nardi, B., Whittaker, S., & Bradner, E. (2000). Interaction and outeraction: Instant
messaging in action. Proceedings of the ACM Conference on Computer Supported
Cooperative Work (CSCW ’00) (pp. 79-88). New York: ACM Press.
National Center for Educational Statistics (2002). The nation’s report card: Writing
2002. National Assessment of Educational Progress. Washington, DC: U.S. Depart-
ment of Education, Institute of Educational Sciences.
Palander-Collin, M. (1999). Male and female styles in seventeenth century correspon-
dence. Language Variation and Change, 11, 123-141.
Randall, N. (2002). Lingo online: A report on the language of the keyboard generation.
Department of English, University of Waterloo. Retrieved July 25, 2004, from http://
www.arts.uwaterloo.ca/nrandall/LingoOnline-finalreport.pdf
Romaine, S. (2003). Variation in language and gender. In J. Holmes & M. Meyerhoff
(Eds.), The handbook of language and gender (pp. 98-118). Malden, MA: Blackwell.
Sacks, H., Schegloff, E., & Jefferson, G. (1974). A simplest systematics for the organiza-
tion of turn-taking in conversation. Language, 50, 696-735.
Schegloff, E., & Sacks, H. (1973). Opening up closings. Semiotica, 7, 289-327.
Schiffrin, D., Tannen, D., & Hamilton, H. E. (Eds.) (2001). The handbook of discourse
analysis. Malden, MA: Blackwell.
Schiano, D., Chen, C., Ginsberg, J., Gretarsdottir, U., Huddleston, M., & Isaacs, E. (2002).
Teen use of messaging media. Proceedings of the ACM Conference on Human Factors
in Computing Systems (CHI ’02), Minneapolis, MN (pp. 594-595). New York: ACM
Press.
Tannen, D. (1994). Gender and discourse. New York: Oxford University Press.
Thomson, R., Murachver, T., & Green, J. (2001). Where is the gender in gendered lan-
guage? Psychological Science, 12, 171-175.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014

Baron / IM AND GENDER 423

Walther, J. B., & D’Addario, K. P. (2001). The impacts of emoticons on message interpreta-
tion in computer-mediated communication. Social Science Computer Review, 19, 323-
345.
Yates, S. (1996). Oral and written linguistic aspects of computer conferencing. In
S. Herring (Ed.), Computer-mediated communication: Linguistic, social, and cross-
cultural perspectives (pp. 29-46). Amsterdam: John Benjamins.

Naomi S. Baron (Ph.D., Stanford University) is a professor of linguistics at Ameri-

can University in Washington, DC. Her research interests include computer-
mediated communication, the influence of technology on spoken and written lan-
guage, child language acquisition, and the history and structure of English. Her
current research focuses on relationships between instant messaging, face-to-face
speech, and offline writing.

Downloaded from jls.sagepub.com at NORTHEASTERN UNIV LIBRARY on April 5, 2014