Vous êtes sur la page 1sur 22

Male and female backchannel strategies

in face-to-face conversation

C-level assignment for the Department of English, Uppsala University

by

E. Täljeblad-Steiner Supervisor: Pia Norell


Austrasse 27
CH-8134 Adliswil 2005

Key words: backchannel, form, function, men, women, gender, mixed, conversation,
support, difference, strategy, laughter
1

Table of Contents

1. INTRODUCTION 2
1.1. Informal conversation 2
1.2. Aim of study 3
1.3. Material and method 3

2. BACKGROUND 4
2.1. Turns and backchannels 4
2.2. Identifying backchannels 6
2.2.1 Back-backchannels 7
2.2.2 Unnoticed turns 8
2.2.3 Comment on the transcription marker (>) 8
2.3 Backchannel functions 9
2.4 Laughter 9

3. RESULTS 11
3.1 Backchannel forms 11
3.2 Backchannel functions 12
3.2.1 Defining a "level of interest" scale 13
3.3 Appropriate timing 17
3.4 Elicited backchannels 18

4. CONCLUDING REMARKS 18

5. REFERENCES AND NOTES 19

APPENDIX
2

1. INTRODUCTION

1.1 Informal conversation

In one of her TV programs, the journalist Stina Dabrowski had as her guest the prominent
politician Carl Bildt. The atmosphere was relaxed and Dabrowski picked the moment to
comment on the fact that she had never heard Bildt make one single mistake when he
spoke, regardless if he was prepared or not; no grammatical errors, no repetition of words,
not even a hesitation. Dabrowski said this in an unmistakingly disapproving tone, and Bildt
had to defend himself, asserting that he made as many errors as the next man when it
came to spontaneous speech. This was, however, expressed with such eloquence that the
audience burst out laughing and Dabrowski said something like "Well, you are incorrigible".
Bildt seemed a little embarrassed.

This example shows how important it is to know how spoken language functions and what
impact this knowledge or lack of knowledge has on our daily life. Crystal & Davy (1969:104)
explain the phenomenon as follows:
"Informal, spontaneous conversation is characterised by a very high proportion of 'errors',
compared with other spoken varieties, involving hesitation features of all kinds, slips of
the tongue (though these are by no means restricted to this variety), and a substantial
amount of overlapping or simultaneous speech."

Is there an interesting point to make in the fact that the above exchange took place
between participants of different sex? Thirty years ago, language and gender did not exist
as a research area in sociolinguistics (Coates, 2003). By tradition, research concentrated on
social class, ethnicity and age. Although gender has been said to be the primary category by
which the social world is organised, it was long avoided. Why? As Coates states, still after
the Second World War, the concept of man and person coincided. Man was the heart of the
society, holding all important positions, and this 'maleness' was not remarked on (2003:10).
A shift came with the Women's Movement and with well educated women, who focussed
their research on women's characteristics. From a deficit and dominance approach
(assuming that woman is powerless and subordinate, which is mirrored in her language),
modern research often takes the view that 'gender' – an alternative expression to the
biological 'sex' – is socially constructed, and that women and men are equal, but use
different communication strategies. This is the stance that I take.
3

1.2. Aim of study

In this paper, I am looking at the form and function of backchannels in daily conversation,
and more specifically, I want to look for any significant differences in the way men and
women use backchannels. My questions are: Do women and men use different forms of
backchannels? Do women use more backchannels than men? Is there evidence that women
give more support to show cooperation and sympathetic understanding? Is it true that men
interrupt in order to dominate and are they bad listeners?

1.3. Material and method

The primary source for my investigation is The London-Lund Corpus of English Conversation
(LLC), edited by Quirk and Svartvik in 1980. It includes 34 transcriptions of pieces of
informal conversation. Recordings made in the 70's were interpreted into orthographic
transcription with prosodic analysis, such as the beginning, end and nucleus of a tone unit,
stress, pauses, boosters, etc. In my examples, I will omit features that are irrelevant for my
discussion in order to facilitate their reading. Below you see an explanatory list of the
symbols used in my examples:

S.1.1 text identification

>Dave speaker continues where he/she left off

sylls syllables

+yes+ simultaneous talk

CAPITALS word is stressed

„ tone unit end

· brief pause of one light syllable

- unit pause

/m/ /sigh/ oral sounds

/- laugh/1 the length of the activity

Like Oreström (1983) and Tottie (1991), I have chosen to study only dyads, i. e. texts
representing conversation between two persons. These texts represent spontaneous
conversation between educated British speakers. The corpus contains all in all nine
4

transcriptions of face-to-face conversation: five are male-male, three male-female and one
female-female. I will examine three of these texts, one of each constellation:

- S.2.12, between a nurse aged 23 and a teacher aged 25, hereafter called Alice and Beth;
- S.1.6, between a female academic aged 45 and a male academic aged 28, called Claire
and Dave;
- S.2.1, between two male academics aged 43 and 34, called Eric and Frank.

Together, the texts amount to 15,000 words. I am aware that it is not enough material to
be able to present any statistically significant data, but I can look for patterns and
tendencies.

As a rule, the recordings were made without the prior knowledge of the participants, so-
called surreptitious recording. However, Beth and Eric were aware of being recorded.

I have had no access to the material in other forms, such as recordings or computer disks.

In studying the texts, I listed the different forms of backchannels and counted the instances
of each form. Also, I identified the most important functions on an interactional level. I
compared the results in search of differences between male and female use of backchannels.
Moreover, I compared the timing and number of elicited responses.

The first step, to define a backchannel, turned out to be more difficult than I anticipated.
Therefore, I will next describe the background to my set of definitions as clearly as possible,
and comment on the transcription.

2. BACKGROUND

2.1 Turns and backchannels

In 1970, Yngve (1970:568) brought forward a characteristic feature of spoken language


while discussing turntaking in conversation:
"The passing of the turn from one party to another is nearly the most obvious aspect of
conversation (---) the distinction between having the turn or not is not the same as the
traditional distinction between speaker and listener, for it is possible to speak out of turn
and it is even reasonably frequent that a conversationalist speaks out of turn. (---) This is
because of the existence of what I call the backchannel, over which the person who has
5

the turn receives short messages such as 'yes' and 'uh-huh' without relinquishing the
turn."

This kind of listener responses function as a direct feedback to speakers, signalling that their
messages have been received, understood, agreed to and/or caused a certain effect
(Oreström 1983:24). There are various labels for this phenomenon; it seems that every
linguist coins his/her own term. Those I have come across are: response token, signal of
attention, accompaniment signal, acknowledger, support, feedback signal, reassurer,
continuer, a go-o, minimal response. I have chosen to use Yngve's term backchannel, which
has come to designate a backchannel message, which can contain one or several
backchannel items.

The difficulties connected with defining a backchannel have been recognised, if not given
any systematic study or satisfactory solutions by the authors I have consulted for this
paper. Oreström (1983:23) states that there are two types of utterances: speaking turns
and backchannel items. These should be kept apart because of their different functions.
With reference to Henne and Watzlawick among others, he defines a speaking turn as a
sequence that conveys new information and expands the topic, whereas a backchannel item
has a low value on the content level but a relatively high value on the relationship level of
communication. It is a question of different roles played by the interlocutors, where one
party is the active speaker, advancing the topic and delivering turns, while the other party is
the active listener, staying in the background and showing interest by emitting
backchannels.

Example (1) shows a typical and indisputable backchannel (marked in italics by me). As
speaker Claire pauses to take a breath at the end of a tone unit, Dave conveniently puts in
an m, which does not interrupt A's flood of words in any way:

(1) Claire: ...and I went back to my old JOB in the civil SERVICE and I found it SO DULL ·
Dave: /m/
>Claire: that I got this LECTURING job in a teacher's TRAINING COLLEGE -
(S.1.6:143)

Unfortunately, not all backchannels are as easily identified. The form of a turn and a
backchannel overlap. To identify a backchannel, you need to take into account the context
as well as timing, intonation and pause lengths.
6

2.2 Identifying backchannels

When I first started scrutinising the selected texts, I meant to follow Tottie's (1991:260-
261) criteria for determining backchannels. She states that:
"The most important principle is that backchannel status can be determined only on
the basis of the following utterance. Thus in (1), A's utterance ^oh G\od' yes ^that's
the :w\orst 'one might have been a backchannel, but as it provoked a response from
B, y\eah# ((it's)) ^really !\awful, it came to serve as a turn.

(1) B: we ^came 'back 'by . from ^Windlebury \East# ^that's the one
that :st\ops at# . *^every !st\ation#*
A: *^oh ^G\od 'yes* **^that's the :w\orst 'one#**
B: **y\eah# ((it's)) ^really** !\awful# (S.5.9:1221-7)2"

At first, it seemed to be an ingenious guiding line, but as I studied my texts and kept filling
my margins with question marks, I realised that I did not agree. It turned out that about
20% of the utterances that I felt were intended as backchannels were noticed and answered
to by the turnholder. Were they to be labelled turns, although there was no claim of the turn
from the active listener? In a way, Tottie (1991:255) answers my question in her
introduction:
"Backchannels are the sounds (and gestures) made in conversation by the current non-
speaker, which grease the wheels of conversation but constitute no claim to take over the
3
turn. "

On the other hand, I noticed that an utterance that starts off as a backchannel may change
into a turn: "yeah / ə:m/ well this was a scheme which…". This happens when the other
party yields the turn.

Tottie (1991:261) also recognises that probably "no two researchers will arrive at exactly
the same solutions", and "the best one can do is to state one's criteria for inclusion of back
channels as explicitly as possible". Unfortunately, she has not been very explicit herself.
Apart from the quotation above, she mentions that "most problems arose with longer
backchannels". Admittedly, it would be too lengthy to describe all criteria for identifying a
backchannel, but I will show the two most recurring difficulties that I experienced while
examining the transcription.
7

2.2.1 Back-backchannels

In Example (2), Eric's first are they is intended as an encouragement to Frank to go on, and
not a proper question. Frank hears this and breaks off to give a short answer before he goes
on with his originally intended line. Again, he receives the same supporting signal, but this
time, he pays no attention. Seen in context, Frank keeps the turn over a longer period
before and after these exchanges. Although Eric elicited a reaction from Frank, it does not
change the direction of the conversation long enough to be seen as a turn. It stays a
backchannel as intended, and Frank's short reaction can be called a back-backchannel, a
term borrowed from Oreström (1983).

(2) Frank: although Rank XEROX are doing their best to make sure that it IS
Eric: are they
Frank: oh they are DEVILS you know · they're buying EVERYBODY out
Eric: are they
Frank: and they then they they've not only put up their PRICES but the've tacked on
all kinds of HANDLING charges and CUTTING charges and – PACKAGING and…
(S.2.1:941)

Another common back-backchannel is when the turnholder reacts to laughing, either by


laughing back or making a short comment, as in Example (3). The giggles convey no other
information than that Dave finds Claire's utterance amusing. Furthermore, it does not make
Claire replan or change direction of the subject, although she breaks off for an instant. Dave
has no intention of taking the floor. Even as Claire chooses to react to the giggles, they
remain a backchannel.

(3) Claire: + I was an UNDER+ GRADUATE here of very RIPE YEARS until last JULY

/e/ last JUNE

Dave: / - giggles/
Claire: YEAH - - and I went back to my old JOB in the Civil SERVICE ...
(S.1.6:137)

I have not included the back-backchannels in my count, since I regard them as tightly
connected with the active speaker's turn.
8

2.2.2. Unnoticed turns

The notion that utterance status is determined on the basis of the next speaker's utterance
might also easily lead to the wrong idea that any utterance that is answered is a turn, and
vice versa, that any utterance not answered constitutes a backchannel. As is noticed in
Example (4), this is not the case. Here, Eric's utterance next term is a turn in its own right,
adding a clarification of the first question /ə/ when are you submitting it. Frank continues
without a pause, which is also signalled by the symbol (>). Alternatively, this exchange can
be seen as the two men being turnholders almost simultaneously.

(4) Eric: /ə/ when are you submitting it


Frank: /ə:h/ - well +it+
Eric: +next+ term
>Frank: it WOULD have been this - AUTUMN - but / ə:/ - I had to go to WORK – THIS
WINTER – and that really (3 sylls)…
(S.2.1:9)
2.2.3 Comment on the transcription marker (>)

The marker (>) is occasionally found before the letter which indicates the speaker. It marks
that the 'speaker continues where he left off' (Quirk and Svartvik, 1980:21). It puzzled me,
because there are more instances, where the speaker has no (>) marker, although he or
she speaks without interruption. Does the marker keep an otherwise broken tone unit
together? No, there were lots of exceptions to that rule. As no further explanation was
given, I figured that it might mark that a single phrase is broken in two by a supporting
signal (Crystal and Davy 1974:109), and thus would be a certain indicator of the presence
of a backchannel. This idea did not prove to be watertight either; however, out of 71
instances of (>), only five were exceptions to the 'backchannel indicator rule', including the
one shown in Example (4) above.

With the above examples, I have pointed at some of the difficulties that arise with picking
out backchannels from a piece of transcribed spoken discourse. This means that my results
are far from exact; however, they show a tendency of what types of, and to what extent,
backchannels occur.
9

2.3 Backchannel functions

The primary function of the backchannel is entirely interactional. It is a device to convey the
act of listening. Backchannels come in a variety of forms and functions that do not
necessarily agree. Thus, "m" may convey acknowledgement, agreement, surprise, or
boredom, depending on intonation (Oreström, 1983).

A further function exists on the semantic level in the manner that the backchannel gives the
current talker a clue to whether the points introduced are new or already known to the
listener - and whether the talker was right in presuming the one or the other in the first
place. Terasaki (2004:171) describes how it is widely observed, that utterances can be
divided into two main groups: those that refer to a previously introduced item, so-called
"given" information, and utterances that introduce a piece of information as "new".

Gardner (2001) surveyed the eight most common mono- and bisyllabic response tokens in
the British language (mm, mm hm, uh huh, yeah, oh, really, okay and alright) in the light
of their functions as continuers, acknowledgement tokens, newsmarkers and change-of-
activity tokens and found that these tokens are rather a complex bunch of utterances that
does very varied work. He complains that many researchers tend to lump a whole range of
sounds and utterances together as backchannels, and treat them superficially as a group.

My conclusion is similar; in many studies of conversation, the backchannel functions are


mentioned, but shallowly treated in favour of other properties of conversation. Heritage
(1989:30), however, sums up some work that has been made in this area. He has shown
that 'oh' is distinctive in being used to indicate some change of state of current awareness.
Schegloff has shown the importance of the precise placement of utterances, such as 'mhm'
by reference to the boundaries of turn-constructional units within a segment of talk.
Jefferson has shown that while 'mm' and 'hm' is a token of passive recipiency, 'yes' may
function as topic-shifting or topic-curtailing activity. Jefferson has further shown that
'really', 'did you', 'you did', etc., promote the telling of news.

Finally, Heritage complains that the fast development of advanced technical devices has led
to a strongly empirical approach to conversation, especially in the field of social psychology.
Experiments are prearranged and carried out in laboratories, and the computer data findings
are believed - until proved otherwise - to bring important facts about the participants of
10

conversation. In this context, Heritage (1989:21-47) accuses researchers of oversimplifying


the function of backchannels and he would like to see more theory constructions.
In view of this last statement, and because I was not able to find a suitable method or
model in the consulted literature, I decided to create my own model, built on
considerations that I will explain in the Results section.

2.4. Laughter

I have decided to devote a whole sub-chapter to laughter, as it was completely omitted in


some research reports (Oreström, Stenström, Tottie), albeit it is a very prominent form of
backchannel. I speculate that it could have something to do with the way it is marked in the
LLC transcriptions. For some reason, Quirk and Svartvik (1980:24) have chosen to include
the sound of laughs - not to vocalisations like /m/, as one would expect - but to a group of
features they call 'contextual comments', which indicate "non-linguistic activity, such as
laughs, coughs, enters, telephone rings or technical mishaps, e.g. gap in recording". Other
oral sounds, such as coughs, sighs and groans are included here to.

This gives the wrong impression, as enters, telephone rings and mishaps on one hand are
mechanical happenings, whereas laughter, coughs, sighs and groans are. Although perhaps
mostly unplanned, the latter can be used to cause attention and/or show emotions. This
causes obvious problems working with computerised data, which Oreström (1983) did, and
this might explain why he omitted laughter.

American researchers have included laughs as a grammatical element though. Schegloff


(1996:102), for example:
"…such as laughter (---) This last is a systematically produced acoustic component of the
"speech stream", which surely contributes to the "meaning" and "import" and
"understanding" of the speech production of which it forms a part."

Heritage (1989:30) refers among others to Jefferson, who points out that not only is
laughter never reported "verbatim", it is also rarely transcribed. Heritage: "Moreover
laughter is apt to be regarded as an out-of-control activity and not as a phenomenon which
is strongly structured in its occurrence, organisation and tasks."

I agree that laughter represents a spontaneous and unplanned feature of communication.


This makes it even more interesting and I see no reason why it should be ignored as an
important form of backchannel in conversation.
11

3. RESULTS

Before discussing the results, I need to comment on the conversations that were being
examined for this paper. The first one takes place between two women, who know each
other well and it is very informal and lively. It is dominated by Alice, who in detail recounts
episodes from her life, so-called "storytelling", and she keeps the floor practically through
the whole conversation and offers only two backchannels. Her friend, Beth, knows about the
recording, which may be the reason why she stays in the background by means of regular
backchanneling. She also asks questions to start a new topic when a previous one dies out.
The conversation is dominated by laughing. Alice keeps laughing while telling, and Beth
laughs with her, producing an outstanding total of 50 laughter backchannels.

The second and the third texts are different from the first one but similar to each other in
that the parties have a working relationship, and do not know each other very well. In the
second text, the atmosphere is friendly and non-competitive between Claire and Dave. They
take turns in holding the floor. In the third text, the exchanges are short and many between
Eric and Frank, and you get the impression of fast turns with hardly any pauses.

3.1. Backchannel forms

As shown in Table 1, the single item /m/ (33%) is the most frequent backchannel item,
followed by /laughs/(19%), yes (7%), mhm (6%), yeah (5%), /hm/ (2%) and no (1%).
These figures show about the same trend as the results of Oreström (1983:121), whose
most frequent single word support types were /m/ (50%) and yes (34%), and Tottie's result
of yes (44%), followed by /m/ (36%). The higher number of yes can be explained with the
fact that she included the complex backchannels, which often started or ended with yes.

Table 1. Female and male occurrences of backchannels

Women % Men % Total


/m/ 53 13% 85 20%
/laughs/ 60 14% 18 5%
yes 16 4% 14 3%
/mhm/ 3 1% 21 5%
yeah 7 2% 14 3%
no 3 1% 3 1%
/hm/ 0 0% 7 2%
complex 32 8% 37 9%
other 20 5% 27 6%
Total 194 46% 226 54% 420
12

All in all, men offered more backchannels than women: 226 (54%) to 194 (46%). The use
of the different forms are evenly spread, with exceptions for /laugh/, which women used
three times more than men, and /m/ which men used 1,5 times more than women. Also, all
7 instances of /hm/ were made by men. These represent 1% of all backchannels. Oreström
(1983) does not mention /hm/ at all (did he not find any?), but Tottie's (1991) number of
/hm/ also constituted 1%, however with no mention of gender.

According to some of Oreström's results (1983:123), the females used mhm and yeah much
less often than the males. This is also the case in my study, where men used /mhm/ 7 times
more and yeah twice as often. The complex backchannels make up 13%. They are either a
combination of single ones (yes yes, /hm/ /hm/, I see yes yes) or unique utterances: you
know this is what's so awful about academics isn't this is the worst side of them (text
S.1.6:1130). "Other" backchannels make up 15%, and include (1) questions (really, have
you, you think so, oh), (2) restatements (kept in Circencester yes), (3) sentence
completions (viceroy /m), (4) exclamations or emotion signals (crikey, amazing, ugh, wow,
oh lord /m/, /sigh/, /gasp/, /groan/).

3.2. Backchannel functions

Is there in my results any evidence for the general notions that exist about male and female
conversation strategies and what are these notions?

Coates (2003:87-93), in summing up various research work, states that women are said to
be more polite, more cooperative and make use of more backchannels in conversation than
men. Women are also believed to employ more questions as a conversational strategy. Men,
on the other hand, are said to follow strategies of non-cooperation. They interrupt and take
hold of the floor without regard to timing, and they are more unwilling to give support in the
way of backchannels. She goes on: "Research on the use of minimal responses is
unanimous in showing that women use them more than men, and at appropriate moments,
that is, at points in conversation which indicate the listener's support for the current
speaker. --- when men do use minimal responses, these are often delayed, a tactic which
undermines the current speaker and reinforces male dominance."

There are a lot of studies which focus on interruptions as a disruptive behaviour, and
although these usually exclude backchannels, there is no clear boundary between the two.
In their review of studies on interruptive behaviour, James and Clarke (1993:258) sum up
that:
13

"not only are instances of interruption not necessarily disruptive in nature, but they can
function to indicate support, collaboration, and solidarity. There is considerable evidence,
however, that women tend to perform more positive socioemotional behavior of this kind
in interactions than do men. For example, many studies have found women to do more
agreeing and showing of support, in both same- and mixed-sex interaction (---) the
majority of studies which have examined the use of back channel responses by listeners
have found women to use more."

In order to compare my results with the above cited generalisations, I wanted a model
that would include all backchannels on one hand and mirror emotional behaviour on
the other. I decided my starting point to be that:
- not only is the primary function of the backchannel to show interest in way of
agreement, understanding, question, surprise, etc., but to signal the extent of this
interest.

It was not possible to undertake a systematic study of the prosodic features, such as
intonation and loudness, for two reasons. I did not have access to computerised
transcriptions, which would have meant overwhelming work to keep track of all
features on paper, and two of the six speakers were aware of the recording. As a
consequence, the prosodic features of their speech were not transcribed.

As I did not find a model in consulted literature that suited my purpose, I decided to
organise the backchannels on a level-of-interest scale, starting with the least
emotional and ending with the most expressive.

3.2.1 Defining a "level-o- interest" scale

Starting with the most neutral, I organised the backchannel realisations in the following
groups:
1. the single forms m, yes, yeah mhm, hm, no
2. complex (quite quite quite quite)4
3. questions (really, you think so, oh)
4. restatements (they relax m)
5. sentence completions (fifteen thousand)
6. exclamations (oh lord no, christ)
7. laughs

To arrive at this model, I looked at the length of the backchannel, if it seemed to carry
several functions and/or emotional weight, and made the following considerations:
14

- a short form shows less evidence of active interest and is more void of emotion than a
longer one; you can "m" and "hm" along without really listening. Exceptions are:
- oh, which adds the signal of surprise at the made utterance and therefore shows
evidence of active listening;
- /laugh/ (see below); and
- yes and yeah, which may show evidence of high interest and belong to the group of
exclamations. Where available, I looked at the transcripted intonation and timing to
decide on the level of interest. This means, that in my scheme, form and function
generally agree, but not always.

- Complex forms, such as yeah yeah yeah show more interest than a single one, but are
monotone, not creative, and showing less interest than an exclamatory question, such
as really.

- Restatement and sentence completion require thought as well as creativity and


therefore belong to the high interest end, as do exclamations, whose intensity of
shown emotion bring them further to the high interest end.

- Finally, laughter, although perhaps not creative in its physical realisation, is a


spontaneous, i.e. strongly emotional, reaction following a creative thought as part of a
necessary active listener participation. It is a backchannel form that you can hardly
fake or control by will, and the combination of these features place laughter at the
furthermost high interest end of this scale.

Perhaps not totally surprisingly, the "unruly" cases of backchannels that I - in want for
better ideas while sampling – grouped as "other" backchannels turned out to correspond to
the higher interest end of the scale; they were questions, restatements, sentence
completions and exclamations.

Chart 1 shows the result of the backchannels offered by men and women on the "level of
interest" scale. Single and complex backchannels as a group were offered 114 times by
women and substantially more, 181 times, by men. Questions, restatements and sentence
completions as a group were represented three times more by men than women. Finally, at
the high interest end, women offered three times more exclamations and laughs. Looking at
the illustration, you can see how men are strongest at the low interest end, how women and
men follow each other closely towards the middle - men always slightly ahead - and how
they break up towards the end and differ in choice of form and intensity of backchanneling.
15

Chart 1.Distribution of backchannels offered by men and women on an interest level scale.

160
140
120
100
Men
80
Women
60
40
20
0

ns
ts

ns

hs
le

ex

ns
ng

en

io
io

ug
pl

tio

at
m

et
em
si

es

la
co

m
pl
qu

at

la
st

ex
co
re

low interest high interest

Chart 2 shows that the single gender conversations contained a slightly larger amount of
backchannels than the mixed gender one. The all female conversation sticks out from the
other two in that Beth does practically all the backchanneling: 133 instances of the total
135. As already mentioned, Beth was aware of the recording, and it is possible that her
reaction to that was to encourage Alice to keep talking. Therefore, this result may not be
representative. On the other hand, Eric also knew about the recording, and participated in
the conversation as both active talker and active listener. Tannen comments on the dilemma
that as long as participants are aware of the presence of the tape recorder, their talk is not
natural. However, the participants also tend to soon forget the tape recorder (1984:33-34).
The mixed conversation contained the lowest amount of backchannels, 122 instances, fairly
evenly divided between Claire (59) and Dave (63). The largest amount of backchannels was
found in the all male conversation with 163 instances.

A seemingly constant observation from research on single sex conversations is that while
men tend to disagree with or ignore each other's utterances, women tend to acknowledge
and build on them. This is done by women in a way that Coates (2004) calls "jam session"
and means that the floor is open to all at the same time, and that speakers co-construct
utterances (sentence completion). She continues: "In contrast, male speakers prefer a one-
at-a-time model of turn-taking. Overlapping talk is rare in all-male talk and is interpreted as
an illegitimate attempt to grab the floor. This means that, in mixed conversation, women
and men may come into conflict over overlapping talk."
16

Chart 2. Distribution of backchannels in the same gender and mixed conversations

140

120

100
single
complex
80
questions
restatements
60 completions
exlamations
40 /laughs/

20

0
Alice Beth Claire Dave Eric Frank

This manner of conversation was confirmed by Alice and Beth, producing the vast majority
of backchannels simultaneously (section 3.3). Similarly, the all-male conversation was
dominated with backchannels produced between talk. In the mixed gender conversation,
however, I could not detect any signs of conflict or misunderstanding.

In their reviews of studies, James and Drakich (1993:289) conclude that "higher-status"
individuals, e.g. by rank, race or occupation, are more willing to contribute to the interaction
than will "lower-status", and are viewed by themselves and others as more intellectually
competent, and likely to perform better. Lower-status participants are content to make more
support signals.

This theory clearly do not apply to "Almighty Alice", who is 23 and a nurse, and
"Backchanneling Beth", who is 25 and a teacher. There is plainly something else than
profession, age or gender rank that is reflected in this conversation. Looking at the two
academics Claire and Dave, however, it fits the picture that Dave is younger and also the
one offering most backchannels, if only marginally. Interestingly, Dave's second most
offered backchannel was /laugh/, a typically female, cooperative feature (16 instances to
Claire's 10). Similarly, Claire's most presented backchannel was /m/, a typical male non-
committed form. Is it possible that both parties more or less unconsciously know of the
other's preference of strategy and copies it to show cooperation and avoid any possible
conflicts? Finally, it also fits that Frank in the all-male talk is younger, seeks approval from
17

Eric and is eager to show interest by using a number of sentence completions. This makes
Frank the lower-status talker and as expected he also gives (marginally) more
backchannels: 86 compared to 77.

3.3 Appropriate timing

Chart 3 shows that about 40% of the female backchannels were simultaneous speech, and
20% of the male. It is worth noticing that with both genders, most simultaneous
backchannels were uttered at a tone unit end (predicting a pause?), and turned into
simultaneous talk only because the turnholder just got started again, see Example 5.

(5) Dave: which he ADMITTED„ it was


Claire: +YES+
Dave: +EXACTLY+ what he'd DONE

(S.1.6: 326)

I looked for especially long turnholder pauses – one or more units – where the listening
party seemed slow to give a backchannel, but I could not find any evidence that men delay
their supporting signals as a conversational strategy. Assuming that men indeed prefer
talking one-at-a-time, my interpretation is that they do not want to seem to interrupt – as a
cooperative strategy – and take care to backchannel when there is a pause, whereas women
do not take offence at simultaneous talk, and therefore overlapping speech takes place more
often.

Chart 3. Female and male timing of backchannels

250

200

150
in between
simultaneous
100

50

0
females males
18

3.4 Elicited backchannels

Another cooperative device is elicited backchannels, i.e. a reaction specifically asked for by
the turnholder. The eliciting mechanism from the turnholder was laughter and 'you know',
tag questions and a first person remark such as 'I think', 'I don't know'. I found 42 elicited
backchannels, twice as often offered by women than by men. These were laughs, single and
complex items, as well as questions, in that order of frequency (Chart 4).

Chart 4. Male and female elicited backchannel responses

18
16
14
12
10
male
8 female

6
4
2
0
single complex question laugh

The conversations go like waves, with many short exchanges for a while, and then longer
exchanges, with hardly any backchannels at all. There is clearly a rhythm of speech, where
the listener is quiet for a longer stretch of time. However, instead there are probably gazes,
nodding, smiling, and such body signals, to show active listening.

I speculate that it might be tiring for the turnholder – in the middle of storytelling – if the
active listener keeps making noises ALL the time. There is a fine line between supportive
and interruptive behaviour. In this context, it is worth noticing that the elicits and elicited
responses did not appear after long stretches of talk, but in a wave of a number of
backchannel realisations. It could as well be that the 'eliciting device' is in fact NOT the
turnholder's way of asking for support, but quite the reverse. It could mean saying thank
you for the already given backchannels by including the other party in the speech.

4. CONCLUDING REMARKS

In this study, I could show some tendencies of gender difference in backchannel strategies.
I looked at the form and frequency of backchannels and found that men, contrary to many
other studies gave a greater amount of backchannels. The forms used were the same except
19

for /hm/ that was offered only by men. Other forms that stood out in this context were /m/,
which men used 1.5 times as much, and /laugh/, which women used three times as often.

I looked at functions, employing a "level-of-interest" scale, and found that men used
sentence completion as a strategy of cooperation, however, thereby also dominating the
floor for a short while; in other words, showing interest as well as creativity in a competitive
way. Women were not found to be competitive, but solely cooperative, and more so than
men. This was confirmed in the way they showed more interest in the way of more emotion.
They were creative, using a great variety of exclamations. However, contrary to some other
studies exemplified by Coates (2004), women were not more cooperative than men in the
manner of using exclamatory questions, such as oh really, did you.

Also contrary to expectation, men were good listeners, predicting the next pause and timing
their backchannel to this pause. It seems to instead confirm the theory that men prefer a
one-at-a-time strategy. Men were not found to wait too long before giving support.

The same gender conversations seemed to confirm the belief that women talk with each
other simultaneously, whereas men prefer one-at-a-time. In the mixed gender conversation,
I thought I could detect a willingness from the parties to "overuse" a backchannel form that
was typical of the other; Claire used /m/ and Dave used /laugh/.

Both genders made room for longer stretches of speech from the turnholder, also allowing
for longer stretches of pauses without breaking the silence. There seemed to be a rhythmic
alternation between short and long exchanges.

5. REFERENCES AND NOTES

5.1. Primary Source

Quirk, Randolph and Svartvik, Jan (eds). 1980. The London-Lund Corpus of Spoken English,
152–172, 372–398, 686–706. Lund: Lund University Press.

5.2. Secondary Sources

Coates, Jennifer. 2003 (3rd edition). Women, Men and Language. A Sociolinguistic Account
of Gender Difference in Language. Harlow: Pearson Education Limited.
Crystal, David and Davy, Derek. 1969. Investigating English Style. London: Longman.
20

Eckert, Penelope. 1993. Cooperative Competition in Adolescent "Girl Talk". In: Deborah
Tannen (ed). Gender and Conversational Interaction. 32-61. New York & Oxford:
Oxford University Press.
Gardner, Rod. 2001. When Listeners Talk. Response tokens and listener stance. Amsterdam
& Philadelphia: John Benjamins Publishing Company.
Heritage, John. 1989. Current developments in conversation analysis. In: Derek Roger and
Peter Bull (eds). Conversation: An Interdisciplinary Perspective. 21-47. Clevedon &
Philadelphia: Multilingual Matters Ltd.
James, Deborah and Clarke, Sandra. 1993. Women, Men and Interruptions: A Critical
Review. In: Deborah Tannen (ed). Gender and Conversational Interaction. 231-280.
New York & Oxford: Oxford University Press.
James, Deborah and Drakich, Janice. 1993. Understanding Gender Differences in Amount of
Talk: A Critical Review of Research. 281-312. In: Deborah Tannen (ed). Gender and
Conversational Interaction. 231-280. New York & Oxford: Oxford University Press.
Oreström, Bengt. 1983. Turn-taking in English Conversation. Lund: Liber Förlag.
Stenström, Anna-Brita. 1984. Questions and Responses In English Conversation. Lund: Liber
Förlag.
Tannen, Deborah. 1984. Conversational Style. Analyzing Talk Among Friends. Norwood, New
Jersey: Ablex Publishing Corporation.
Terasaki, Alene Kiku. 2004. Pre-announcement sequences in conversation. In: Gene H.
Lerner (ed). Conversation Analysis. Studies from the first generation. 171–223.
Amsterdam & Philadelphia: John Benjamins Publishing Company.
Tottie, G. 1991. Conversational style in British and American English: The Case of
Backchannels. In: Karin Aijmer & Bengt Altenberg (eds). English Corpus Linguistics.
254-271. London: Longman.
Yngve, V.H. 1970. On Getting a Word in Edgewise. In: Papers from the Sixth Regional
Meeting of the Chicago Linguistic Society, April 16-18, 1970. 567-578. Chicago:
University of Chicago, Department of Linguistics.

1
I put laughter in between slashes together with vocalisations and phonetic sounds and do not follow
Quirk and Svartvik's transcription, which puts laughter within brackets, together with non-
conversational sounds like when a door opens.
2
this example is taken from Corpus of Spoken American English, which differs somewhat from the
London-Lund Corpus of Spoken English (Tottie 1991:260)
3
my italics
4
Oreström (1983:124) argues, however, that a chain of backchannel items also has another function;
it can indicate the listener's raised interest in taking over the turn, similar to raising a hand in the
classroom.
APPENDIX

Basic Data: Occurrence of backchannels in the texts S.2.12, S.1.6 and S.2.1.

Alice Beth Claire Dave Eric Frank Total %

/m/ 0 32 21 38 27 20 138 33%


/laughs/ 0 50 10 16 1 1 78 19%
yes 1 4 11 0 8 6 30 7%
/mhm/ 0 2 1 3 5 13 24 6%
yeah 0 7 0 0 4 10 21 5%
no 1 0 2 0 2 1 6 2%
/hm/ 0 0 0 0 6 1 7 1%
complex 0 19 13 3 15 19 69 16%
other 0 19 1 3 9 15 47 11%

Total 2 133 59 63 77 86 420 100%

Vous aimerez peut-être aussi