See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/232418040

Intimacy in Conversational Style as a Function of the Degree of Closeness

Between Members of a Dyad

Article  in  Journal of Personality and Social Psychology · September 1985

DOI: 10.1037/0022-3514.49.3.671

CITATIONS READS

34 468

1 author:

Gail Hornstein
Mount Holyoke College
30 PUBLICATIONS   438 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Gail Hornstein on 27 February 2015.

The user has requested enhancement of the downloaded file.

Journal of Penonality and Social Psychology Copyright 1985 by the American Psychological Association, Inc.
1985, Vol. 49, No. 3, 671-681 0022-3514/85/$00.75

Intimacy in Conversational Style as a Function of the Degree

of Closeness Between Members of a Dyad
Gail A. Hornstein
Mount Holyoke College

Research on dyadic relationships has been based on the assumption that the closer
the relationship, the greater the amount of intimacy that is expressed within the
dyad. Most researchers have focused on how intimacy is conveyed in the content
of a conversation (i.e., which topics are discussed). The present study used a more
structural approach, focusing on how intimacy is conveyed by conversational style.
Fifty-eight naturally occurring telephone conversations between pairs of female
friends, acquaintances, and strangers were analyzed. In comparison with strangers,
friends used more implicit openings, raised more topics, were more responsive to
one another (e.g., by asking more questions), and used more complex forms of
closing. Acquaintances were generally more similar to strangers than they were to
friends. Implications of these findings for research on the development of intimacy
are discussed.

An assumption basic to research on dyadic turing their discussion. Underlying this struc-
relationships is that the closer the relationship, tural view is the assumption that intimacy is
the greater the amount of intimacy that is ex- conveyed by the style, and not just the content,
pressed within the dyad (Cozby, 1973). Al- of a conversation. Thus, for example, friends
though the precise nature of what constitutes would be expected to talk to one another in a
intimacy remains unclear (Sexton & Sexton, more intimate way than acquaintances would,
1982), disclosure of personal information, independent of the material being discussed.
feelings, and opinions has been identified by a Although the structural approach has re-
number of researchers as one key component ceived less attention than the more traditional
in the expression of intimacy (e.g., Gaebelein, content approach, it has generated some in-
1976; Jourard, 1971; Morton, 1978). teresting findings. The most important of these
Analyzing intimate exchange by examining is that intimate conversations seem to be more
the amount and type of disclosure that occurs jmplicit than nonintimate conversations along
within different kinds of dyads can be termed a number of different dimensions (Berger &
a content approach to the study of intimacy, Kellner, 1964; Garfinkel, 1967; Joos, 1962;
in that the focus is on the nature of the material O'Mara, 1973;Wemer&Kaplan, 1963). That
being discussed. An alternative approach to is, the shared experience of being in a close
this issue, which can be termed a structural relationship appears to allow the members of
approach, looks not at what the members of such dyads to talk together in ways that do not
a dyad talk about but how they go about struc- require full articulation of the assumptions on
which their exchange is based. The use of ab-
breviated expressions, rapid shifting from one
topic to another with little transition, and fre-
This article is based on a dissertation submitted to the quent ellipses are some of the features that
Department of Psychology, Clark University. Portions of characterize the implicit style of intimate con-
this study were presented at the meetings of the Association versation. These stylistic features are paralleled
for Women in Psychology, Seattle, March 1983 and the
Eastern Psychological Association, Philadelphia, April by such content characteristics as a lack of im-
1983. mediate reciprocity in self-disclosure (e.g.,
Grateful acknowledgement is made to Francine Deutsch, Won-Doomink, 1979).
Seymour Wapner, and an anonymous reviewer for their To specify more precisely how intimate
detailed and helpful comments on an earlier draft.
Requests for reprints should be sent to Gail A. Hornstein,
conversations are structured differently from
Department of Psychology and Education, Mount Holyoke nonintimate conversations, the present study
College, South Hadley, Massachusetts 01075-1462. built upon the above findings and on work by
671
672 GAIL A. HORNSTEIN

Garfinkel ( 1967), who outlined certain general of voice quality alone. Acquaintances were ex-
organizational features of conversation. Gar- pected to require more explicit identification
finkel argued that conversations are structured or to express uncertainty about one another's
in terms of a set of ends, which participants identity. Strangers were expected to provide
achieve by using a variety of means. Examples detailed and explicit identifying information
of such ends include initiating new topics, without the aid of prompting or questions.
maintaining the conversation once it is under- The second comparison concerned the ini-
way, and bringing the discussion to a close. tiation of topics. It was hypothesized that
Although Garfinkel claimed that there are friends would initiate the greatest number of
multiple means by which each of these ends topics; acquaintances, a moderate number; and
can be accomplished, he did not attempt to strangers, the least number of topics. This hy-
specify the factors that would lead a certain pothesis was based on the assumption that the
kind of dyad to use one means rather than more intimate the relationship, the wider the
another. However, the finding that intimate range of domains that can be drawn upon in
conversations are characterized by an implicit a given conversation (Altman & Taylor, 1973).
style suggests that the various means of ac- Furthermore, it was hypothesized that in con-
complishing conversational ends could be or- versations between strangers, there would be a
dered from more implicit to less implicit. On clear asymmetry between participants; callers
this view, dyads whose members have a close would initiate more topics than respondents
relationship would be expected to accomplish would. This asymmetry was expected to be less
a goal like initiating new topics in a more im- marked but also present in conversations be-
plicit way than dyads whose members are more tween acquaintances. In contrast, callers and
distant. Thus, although each dyad would be respondents were expected to initiate an equal
required to achieve the same set of conversa- number of topics in conversations between
tional ends, the means used to accomplish friends. These hypotheses were based on the
these tasks would be assumed to vary as a assumption that the differential role require-
function of the degree of closeness between ments of callers and respondents in telephone
members of the dyad. conversations would be highly salient to
In the present study, the conversational pat- strangers and somewhat salient to acquain-
terns of friends, acquaintances, and strangers tances but that the mutuality of friend rela-
were compared in order to test this general ex- tionships would override such role differences
pectation. Naturally occurring telephone con- in conversations between friends.
versations provided the source of data. 1 The The third comparison concerned the ways
overall hypothesis was that the three kinds of that dyads maintained the conversation once
dyads would differ with respect to how they it was underway. Fishman ( 1978) has pointed
accomplished the same set of conversational out that a conversation will simply die unless
ends, with friends using the most implicit, and the participants exert some effort to maintain
strangers the most explicit, means. Acquain- it; thus, she argues that their efforts at conver-
tances were expected to use a style that was sational maintenance can be termed interac-
moderately implicit. Specific hypotheses were tional work. Fishman suggests that because
formulated regarding how the three kinds of questions guarantee at least one response, they
dyads would achieve each of four conversa- serve as one of the most important means by
tional ends: identification/recognition, topic which to ensure that conversation continues.
initiation, conversational maintenance, and Another way to accomplish interactional work
closing. is to use what she terms minimal responses.
One unique feature of telephone conversa- By inserting a string of "hmms," "yeahs,"
tions is that participants must identify them-
selves to one another and indicate their rec-
1
ognition of each other's identity before the Although Schegloff (1979) has made the claim that
conversation can proceed (Schegloff, 1979). It "the talk people do on the telephone is not fundamentally
different from the other talk they do" (p. 25), telephone
was hypothesized that friends would accom- conversations do have certain unique organizational fea-
plish such recognition with little or no explicit tures (Ball, 1968; Schegloff, 1968) and thus cannot be con-
identification, in some instances on the basis sidered interchangeable with face-to-face conversations.
INTIMACY IN CONVERSATIONAL STYLE 673

"ohs," and so forth, into the stream of the required. This could take some time and would
speaker's talk, listeners can demonstrate their be likely to occur in a form that is uniquely
interest in what is being said and their con- appropriate for each particular dyad. These
tinuing participation in the conversation. issues were assumed to concern acquaintances
However, when such responses are made by to a lesser extent and strangers not at all. It
the listener at the end of the speaker's turn, was also hypothesized that callers would in-
they have the opposite effect in that they do variably initiate closings between strangers and
nothing to add to the topic and do not en- that they would typically but not invariably do
courage the speaker to continue. In conversa- so in conversations between acquaintances. In
tions between strangers, it was hypothesized contrast, in conversations between friends, the
that callers would bear most of the responsi- expectation was that the closing segment would
bility for conversational maintenance and be initiated equally often by callers and re-
would therefore ask more questions than re- · spondents, again reflecting the greater degree
spondents would. However, respondents were of mutuality in these dyads.
expected to use minimal response with high
frequency during speaking turns so as to dis- Method
courage further conversation. Callers in ac- Overview
quaintance conversations were also expected Naturally occurring telephone conversations between
to ask more questions than respondents, but women who were close friends, acquaintances, or strangers
respondents were expected to have a relatively provided the data for the study. Telephone conversations
high number of instances of minimal response have the advantage of rendering most nonverbal cues (other
than intonation) irrelevant, thereby making it possible to
while listening, which would serve the purpose use audiotape recordings for data analysis without sacri-
of encouraging the caller to continue talking. ficing any information. In addition, telephone conversations
Friends were expected to share interactional have been shown to have a more uniform organizational
work by asking one another frequent questions sequence than face-to-face conversations (Ball, 1968;
and by using minimal response with high fre- Scheglolf, 1979).
Data were collected by installing recording equipment
quency while listening so as to encourage the in the homes of 10 women, with their permission. Each of
other to continue speaking. A low rate of min- these 10 primary participants then recorded her telephone
imal response during speaking turns was ex- conversations with 4-6 secondary participants over a period
pected among friends, because the absence of of 2-4 weeks. From the large corpus of data that was gath-
ered, 58 conversations were randomly selected for analysis:
such a response pattern would serve to en- 20 between close friends, 20 between acquaintances, and
courage further conversation and friends were 18 between strangers.2 Conversations in each of the three
expected to want to do this. groups varied in length, but there were approximately equal
The fourth comparison examined how the numbers of short, medium, and long conversations in each
three kinds of dyads ended their conversations. group, and the differences between groups were not sig-
nificant, x 2(4, N = 58) = 0.38, p > .OS.
As Schegloff and Sacks ( 1973) have noted,
conversations do not simply end of their own Participants
accord; they must be deliberately brought to Sixty-one women participated, 10 as primary partici-
a close using one of a number of closing strat- pants and 5 I as secondary participants.3 Participants' ages
egies. It was hypothesized that strangers would
end their conversations by using very simple 2
It was difficult to locate strangers in order to secure
closing forms involving only a few ritualized permission for their data to be used in the study. This was
exchanges. Acquaintances were expected to use possible in only 18 cases; thus, rather than randomly se-
moderately complex closing forms that were lecting conversations between strangers, all available con-
more idiosyncratic. Friends were expected to versations were used. If equal sample sizes among groups
use complex forms of closing that would often had been necessary for data analysis, the friend and ac-
quaintance groups would have been reduced to 18 dyads
take some time to achieve and would be highly each to match the size of the stranger group. Because this
idiosyncratic. These hypotheses are based on was not necessary, it seemed desirable to use all 40 of the
the assumption that friends must find ways to friend and acquaintance cases for which permission had
end their conversations that do not threaten been secured, even though there were a few instances where
recording failure required dropping one or two of the friend
the closeness of their relationship; conse- dyads from a specific analysis (see footnotes 7 and 10).
quently, considerable implicit negotiation back 3
Some of the primary participants had the same friends
and forth about when to stop talking would be or acquaintances, so in a few instances a secondary par-
674 GAIL A. HORNSTEIN

ranged from early 20s to mid-50s. Primary participants tances, and strangers. The categorization of secondary
were recruited from a college community, with the first 10 participants as friends, acquaintances, or strangers was
volunteers being chosen. 4 made by the primary participant using standardized cri-
teria. Because the literature on interpersonal relationships
does not provide any one universally agreed-upon set of
Procedure criteria to distinguish one kind of relationship from another,
I integrated various viewpoints (Becker, 1973; Bell, 1975;
Tape recording. A meeting was held with each primary Jourard, 1971; Kurth, 1970; Levinger, 1978; Levinger &
participant to explain the procedure, describe the purposes Snoek, 1972; Lewin, 1948; Orlinsky, 1979; Rose, 1979;
of the study, and most important, review the ethical safe- Sadler, 1970; Suttles, 1970; Walker & Wright, 1976; Wright,
guards that had been built into the procedure. Primary 1974) to arrive at a conservative set of criteria that are in
participants signed a consent form at this meeting indi- accordance with most existing views. The criteria were as
cating that they agreed to have recording equipment in- follows:
stalled in their homes and that they would provide the 1. Friend relationships were defined as those involving
names of those women whose conversations were recorded reciprocal self-disclosure and a high degree of mutual de-
so that they could be contacted for permission to use their pendence, involvement, and obligation between partici-
data in the study. The consent form also guaranteed the pants. A friend relationship is developed with a specific,
participant the right to tum off the equipment at any time unique individual who is chosen from some larger group
for any reason. on the basis of her unique constellation of characteristics
During the interval between this meeting and the in- and qualities, not on the basis of her ability to meet a
stallation of the recording apparatus, primary participants specific need or provide a service. Individuals involved in
discussed the project with those women with whom they a friend relationship see or talk with one another frequently,
had regular telephone conversations. These women were and often do so just to interact, rather than for some other
told that if one of their conversations happened to be re- purpose.
corded, they would be sent a letter describing the study 2. Acquaintance relationships were defined as those in-
and the procedures used to maintain confidentiality and volving little self-disclosure and a low degree of mutual
that they could grant or refuse permission for the actual dependence, involvement, and obligation between partic-
use of the conversation in the study. This procedure ensured ipants. An acquaintance relationship is typically restricted
that many of the secondary participants were aware of the to only one setting (e.g., school, work, softball team), and
recording before it took place but prevented discussions there is no warrant in the structure of the relationship to
about the study and the procedure from taking place during interact in other settings. Individuals involved in an ac-
the actual recording. quaintance relationship see or talk with one another on an
The equipment consisted of two parts: a telephone re- irregular basis, and typically do so for a specific purpose
cording control (Radio Shack No. 43-236) and a portable rather than solely for the purpose of interacting. Following
cassette tape recorder. The control device connects at one Kurth ( 1970), acquaintance relationships were seen as
end to the tape recorder and at the other end to a modular friendly but not as intimate enough to be friendships.
plug attached to the telephone wire. This device can be set 3. Stranger relationships were defined as those involving
to operate automatically in such a way that when a tele- no prior contact between participants. They therefore have
phone receiver is lifted anywhere in the house, the tape none of the characteristics (e.g., dependence, obligation,
recorder is activated and recording begins. Thus, sensitivity involvement, disclosure) that are present in the other two
to the recording procedure is reduced because participants categories. Contacts between strangers are almost always
do not have to tum on the apparatus manually at the be- focused around the delivery of a service or the presentation
ginning of each conversation. However, the control device of information of a specific nature, and there is typically
can easily be switched off, and participants were encouraged no further interaction once this exchange has occurred.
to tum it off as often and for as long as they liked.' Re- Although only the primary participant was involved in
cording took place over a period of several weeks, with deciding whether a given relationship fit the friend, ac-
each primary participant providing at least four 90-min quaintance, or stranger category, the above criteria were
cassette tapes of data. All participants reported that once
they became accustomed to the apparatus, they were only
4
minimally aware of the recording. Given the ethically sensitive nature of telephone re-
Data preparation. Tapes were transcribed using a cording, only those individuals who had no reservations
modified form of the transcription coding system devised about volunteering were accepted as primary participants.
by Jefferson (as presented in Sacks, Schegloff, & Jefferson, The fact that this meant including some roommates and
1974). This system provides a standardized method for friends among the primary participant group was seen as
indicating such information as pauses, simultaneous less problematic than the alternative choice of allowing
speech, and general features of intonation. some reluctant individuals to participate.
Criteria for distinguishing between friends, acq11ain- ' Primary participants reported that they turned off the
recording apparatus during conversations (a) with men,
(b) with family members, and (c) with those women with
ticipant conversed on one occasion with a friend and then whom they were especially intimate. They had been in-
on another occasion with an acquaintance. Further, some structed not to record conversations of the first two types.
of the primary participants were roommates of one another, The fact that they chose not to record conversations of the
so in a few instances a secondary participant conversed on third type means that the intimate conversations in this
one occasion with a stranger (the roommate) and then on study are not representative of the entire range of intimate
another occasion with her friend or acquaintance. exchange that occurs between women.
INTIMACY IN CONVERSATIONAL STYLE 675

reported to be so clear that there was no difficulty in making if an acquaintance uses an opening form that
these categorizations. No primary participant reported that is too explicit, she will end up sounding like a
she thought it was possible that the secondary participant
would have categorized the relationship in a different way. stranger, thereby violating the assumption of
moderate familiarity that underlies her rela-
Results and Discussion tionship to the partner. What seems to happen,
therefore, is that acquaintances use a whole
Opening Identification/Recognition Sequence variety of different opening forms in an at-
To score the way that participants identified tempt to resolve the recognition problem in a
themselves to one another, the opening iden- way that will seem appropriate to the partner
tification/recognition sequence of each con- in that specific conversation. It may be that
versation was assigned to one of three groups: acquaintances who are on the way to becoming
high implicitness, moderate to low implicit- friends use forms of opening that are more
ness, or high explicitness. 6 These categoriza- implicit, whereas those who are likely to
tions were made independently by two judges maintain an acquaintance relationship use less
with 94% agreement. An example of a highly implicit openings.
implicit opening would be self-identification
in a form recognizable without a name (e.g.,
Topic Initiation
respondent: "Hello"; caller: "Hi. It's me"). An
example of a moderate to low implicit opening To examine differences in the pattern of
would be a question regarding the identity of topic initiation used by the three kinds of
the answerer (e.g., respondent: "Hello"; caller: dyads, it was necessary to have a way of mark-
"Uh, hi, who's this?"). An example of an ex- ing the starting and ending point of each topic
plicit opening would be self-identification with in a given conversation. Following Fishman
one's full name (e.g., respondent: "Hello"; (1979, p. 12), an utterance was considered to
caller: "Good day. This is Gertrude McElroy"). initiate a topic if "it addressed itself to a dif-
Results indicated that 79% of friend dyads ferent subject than the utterance preceding it."
used opening sequences scored as highly im- This criterion is also similar to those used by
plicit, 7 whereas stranger dyads showed just the Martlew ( 1980) and Weiner and Goodenough
opposite pattern, with 89% using explicit se- ( 1977) in scoring conversational moves. There
quences. Acquaintance dyads showed an in- was a high degree of reliability between scorers
termediate pattern, with 65% using moderately in judging topic initiations and conclusions;
implicit opening sequences, 20% using explicit two independent judges reached a 93% rate of
sequences, and 15% using highly implicit se- agreement.
quences. These findings fit the predictions, and
the differences between groups are highly sig-
nificant (kappa= .66, p < .001). 8 6
The specific scoring categories were modified from the
The variability among the responses of ac- system developed by Schegloff ( 1979). In Schegloff's sys-
quaintance dyads indicates that identification/ tem, the caller's first utterance is assigned to one of nine
recognition may be more problematic for ac- specific opening types. However, in the present study, all
quaintances than for either of the other two utterances made by both the caller and respondent during
groups. Such a finding makes sense in that un- the opening identification/recognition sequence were con-
sidered in categorizing the opening. The entire category
like friends, whose frequent contact makes system, with examples of each scoring type, is available
recognition unproblematic, and strangers, from the author.
7
whose lack of prior contact requires the use of Only 19 of the 20 friend dyads were included in analyses
explicit forms of identification as a matter of of opening types because the beginning segment of one
friend conversation was lost owing to recording failure.
course, acquaintances cannot identify them- 8 In this and some subsequent analyses, differences be-
selves to one another in one of a small number tween dyadic groups were assessed by means of kappa
of standard ways. Rather, acquaintances are in (Cohen, J960), a statistic designed to test for ordered dif-
a somewhat awkward position: If they rely on ferences between nominal categories. Dyads, not individ-
forms of opening that are too implicit, they uals, were used as the unit of analysis. Although some in-
dividuals participated in more than one dyad, no dyad was
run the risk of putting their partner in the un- identical to any other dyad, thus posing no challenge to
comfortable situation of struggling to figure the assumption of independence on which the kappa and
out to whom she is talking. On the other hand, other nonparametric statistics are based.
676 GAIL A. HORNSTEIN

The numbers of topics initiated by the caller Table I

and by the respondent were computed, and Mean Number of Topics Initiated by the Three
these scores were analyzed by means of a two- Types of Dyads
way analysis of variance. The variables were Type of dyad
type of dyad (friend, acquaintance, stranger) Participant
and participant role (caller, respondent). The role Friends Acquaintances Strangers
main effect for type of dyad was significant,
F(2, 110) = 26.88, p < .001; however, neither Caller 7.75 2.00 1.72
Respondent 6.55 1.50 1.17
the main effect for participant role nor the in-
teraction between the two variables was sig- Note. For 15 comparisons (df 110), mean difference re-
nificant. quired for significance at p < .01 = 3.66.
To determine specifically which groups dif-
fered from one another, the Bonferroni t sta-
tistic was used (Wike, 1971 ). The criteria for (questions initiating a topic, questions occur-
significance provided by Bailey (1977) indicate ring during discussion of a topic). This third
that friends initiated significantly more topics variable was a repeated measure. There was a
than either acquaintances or strangers, who did significant main effect for type of dyad, F(2,
not differ significantly from one another (see 110) = 21.18, p < .00 l; a significant main ef-
Table 1). fect for type of question, F(l, 110) = 60.80,
These findings do not provide support for p < .001; and a significant interaction between
the prediction that there would be a difference type of dyad and type of question, F(2, 110) =
in the number of topics initiated by callers ver- 18.72,p < .001. Post-hoc comparison of means
sus respondents. Inspection of the means in using the Bonferroni t statistic indicated that
Table 1 indicates that for each group, the these effects were due to the extremely high
numbers of topics initiated by callers and by number of questions asked by both callers
respondents were almost equal, and there was and respondents in friend dyads as they dis-
no tendency for the difference to be greatest cussed a topic. The means for these two cate-
in the stranger group, as had been predicted. gories were significantly higher than all of
However, this may be due to the fact that over- the others; none of the other dyad groupings
all, strangers and acquaintances initiated very differed significantly from one another (see
few topics. It had been hypothesized that Table 2).
friends would initiate more topics than either These findings provide partial support for
acquaintances or strangers; what was not ex- the hypotheses. The high incidence of ques-
pected was that strangers and acquaintances tions among friends was expected, because the
would initiate so few topics that any difference mutuality of friend relationships was assumed
between callers and respondents would be so to generate a significant degree of interest in
small as to be nonsignificant. the details of the other's life, which could best
be elicited through the use of frequent ques-
Maintenance of Conversation tions. However, the differences that had been
expected between callers and respondents in
Two types of measures were used to score stranger and acquaintance conversations were
conversational maintenance: (a) the number not found. Inspection of the means in Table 2
of questions asked to initiate a topic and during indicates that in each instance, the mean for
discussion of a topic and (b) the number of callers was slightly higher than that for re-
minimal responses (e.g., "yeah," "uh-huh," spondents, but these differences were not sig-
etc.) made while listening and during a speak- nificant, and the expectation that the difference
ing tum. Both of these measures were com- between callers and respondents would be
puted separately for callers and respondents. greatest in stranger dyads was not supported.
To examine differences in the frequency of Thus, respondents appear to work as hard as
questions, a three-way analysis of variance was callers do, with both participants doing the
performed, with the variables type of dyad most work in conversations between friends.
(friend, acquaintance, stranger); participant To determine whether there were differences
role (caller, respondent); and type of question in the use of minimal response by callers and
INTIMACY IN CONVERSATIONAL STYLE 677

Table 2 spondents in friend dyads made more respon-

Mean Number of Questions Asked by the Three ses while listening than any other group (see
Types of Dyads Table 3).
Type of dyad
The high incidence among friends of min-
imal response while listening was expected,
Acquain- because the use of this strategy serves to en-
Type of question Friends tances Strangers courage the other to continue speaking, and
friends were considered most likely to want to
Caller
During discus- do this. However, the asymmetry between call-
sion of topic 17.05 4.90 3.44 ers and respondents in friend conversations was
Initiating not expected. At least with respect to minimal
topic 2.70 0.80 0.78 response, it appears that respondents in friend
Respondent
During discus-
dyads do more of the work of maintaining the
sion of topic 16.65 4.15 2.72 conversation than callers do. Also contrary to
Initiating expectation, respondents in stranger conver-
topic 2.25 0.65 0.44 sations did not use minimal response with high
frequency during speaking turns so as to dis-
Note. For 66 comparisons (df = 110), mean difference re-
quired for significance at p < .01 = 8.29. courage further conversation.
To summarize, the findings indicate that re-
spondents in friend conversations relied heavily
on minimal response while listening as a tech-
respondents in the three kinds of dyads, a nique to encourage the caller to continue
three-way analysis of variance was performed. speaking. Callers in friend conversations also
Again, the first two variables were type of dyad relied on this technique, although to a lesser
(friend, acquaintance, stranger) and partici- degree than respondents did. Acquaintances
pant role (caller, respondent). The third vari- and strangers used minimal response less often
able, a repeated measure, was type of minimal than friends did and did not show the same
response (responses made while listening, re- asymmetry between callers and respondents.
sponses made during a speaking turn). The All dyads made fewer responses during speak-
main effect for type of dyad was significant, ing turns than they did while listening.
F(2, 110) = 18.53, p < .001, as was the main Taken together, the findings regarding con-
effect for type of minimal response, F( 1, versational maintenance demonstrate that
110) = 31. 76, p < .001. The interaction be- friends use specific techniques to keep a con-
tween these two factors was also significant, versation going, whereas strangers and ac-
F(2, 110) = 10.69, p < .001. The main effect quaintances do so to a considerably lesser ex-
for participant role was not significant, but the tent. Because the use of such techniques as
interaction between type of dyad and partici-
pant role reached borderline significance, F(2,
110) = 2.89, p = .06. The other interactions Table 3
were not significant. Mean Number of Minimal Responses Made
To understand these findings, it is helpful by the Three Types of Dyads
to examine the group means, which are pre- Type of dyad
sented in Table 3. Post-hoc comparisons in-
dicated that the main effect for type of dyad Acquain-
was due to the large number ofresponses that Type of response Friends tances Strangers
both callers and respondents in friend dyads Caller
made while listening. Both of the significant While listening 19.10 9.50 3.50
interactions can be attributed to the extremely While speaking 5.95 2.75 1.56
high number of responses that respondents Respondent
in friend dyads made while listening. Thus, While listening 35.65 5.25 6.10
While speaking 8.65 2.60 1.50
these analyses indicate that friends used mini-
mal responses more frequently while listening Note. For 66 comparisons (df = 110), mean difference re-
than acquaintances or strangers did, and re- quired for significance at p < .01 = 14.98.
678 GAIL A. HORNSTEIN

asking questions and making minimal respon- as tightly clustered in the high complexity
ses allows a topic to be discussed in more group as had been predicted. This variability
depth, it might be said that conversations be- among the closing types used in friend con-
tween friends are structurally more complex versations indicates that the task of ending the
(i.e., more topics are discussed in greater conversation may be more problematic for
depth). Fishman's notion ofinteractional work friends than for either of the other two groups.
is quite useful here, because it highlights the On reflection this is not surprising; if the clos-
fact that discussion of this kind does not just ing is not handled carefully (i.e., by being ex-
happen; it requires effort or work on the part ecuted too abruptly or in a form not appro-
of both participants. It makes sense that friends priate for that specific partner), the mutuality
would be more motivated than acquaintances of the friend relationship could be jeopardized.
or strangers to perform such interactional The use of distinct closing forms by friends,
work. · acquaintances, and strangers may also reflect
The recent work on conversational "open- the different purposes for which conversations
ers" (Miller, Berg, & Archer, 1983; Purvis, were undertaken by each group. 11 To evaluate
Dabbs, & Hopper, 1984) parallels these find- this possibility, each conversation was assigned
ings on conversational maintenance. Openers, to l of 14 categories that described, from the
defined as "those individuals able to promote caller's perspective, the primary reason for the
conversation and self-disclosure in their part- conversation. Prior to scoring, these categories
ners" (Purvis et al., 1984, p. 65), apparently were grouped in terms of whether they were
use specific behaviors such as gaze and simul- indicative of high intimacy, moderate to low
taneous speech to encourage their partners' intimacy, or no intimacy. Agreement on these
participation. The research on openers has not judgments was 86%. The number of friend,
addressed the question of whether intimate acquaintance, and stranger dyads assigned to
dyads are more likely than nonintimate dyads each of the three intimacy groupings was com-
to include an opener, but such a hypothesis puted, and the differences between groups were
would seem reasonable on the basis of the highly significant (kappa = .90, p < .00 l ). Re-
present findings regarding conversational sults indicated that strangers typically called
maintenance among friends. one another in order to ask for or receive spe-
cific information. Acquaintance calls typically
Closing Segment involved making arrangements of some sort.
To score the ways that participants ended In contrast, friends often called each other
their conversations, a procedure similar to that simply for the pleasure of talking together.
used to score opening sequences was used. The These findings serve to clarify the results on
closing segment of each conversation was as- conversational closing. Because strangers and
signed to one of three groups: high complexity, acquaintances called one another for specific
moderate to low complexity, or very low com- reasons, they could close their conversations
plexity. 9 These categorizations were made in- once the information had been transmitted or
dependently by two judges with 87% agree- the arrangements made. However, because
ment.
Results indicated that 94% of stranger dyads
used closings that were scored as having very 9
The specific scoring categories were modified from the
low complexity, and 95% of acquaintance system developed by Schegloff and Sacks ( 1973). The
dyads used closings in the moderate to low modified system used 23 subcategories, each of which de-
complexity group. Friends showed a more scribed a particular sequence of events occurring during
mixed pattern: 61 % used highly complex clos- the closing segment (e.g., each participant says "OK" fol-
ings, 39% used moderate to low complexity lowed by a pause, they discuss arrangements for future
contact, and then engage in a terminal exchange such as
closings, and no dyads used very low com- "Bye. See you"). The complete scoring system, with ex-
plexity closings. 10 These findings generally fit amples of each subcategory, is available from the author.
10
the predictions, and the differences between Only 18 of the 20 friend dyads were included in anal-
groups are highly significant (kappa = .76, yses of closing types because the closing segments of two
friend conversations were lost owing to recording failure.
p < .001). 11
Analyses of conversational content, including the rea-
The only unexpected aspect of these results son for which the call was initiated, are summarized here;
is that closings among friend dyads were not more detailed information is available from the author.
INTIMACY IN CONVERSATIONAL STYLE 679

friend conversations were motivated by a more form of opening and closing), they clearly dif-
general desire to talk together, their conversa- fered from both friends and strangers. How-
tions could not end until each participant had ever, in other respects (e.g., number of topics
said everything she wished to say or until one initiated, use of techniques for maintaining
participant indicated that she would rather do conversation), acquaintances were similar to
something other than talk. In either of these strangers. The clear differences between groups
cases, closing the conversation becomes a suggest that a structural approach to the study
complex task of negotiating between the needs of conversational intimacy can add a useful
of each individual and the more general need perspective.
to maintain closeness in the relationship. That Advocates of the more traditional content
friends showed more variability in closing approach might argue that these differences in
types than the other two kinds of dyads prob- conversational style simply reflect the different
ably reflects the greater complexity of their purposes for which dyads initiated their con-
task. versations. In other words, perhaps it is the
An additional analysis of conversational purpose per se, rather than the degree of in-
closing examined which participant initiated timacy of the dyad, that is associated with one
the closing segment. As predicted, callers ini- conversational style rather than another. This
tiated almost all closings in stranger conver- alternative hypothesis would only carry any
sations (94%) and many of the closings in ac- weight if conversational purposes and dyad in-
quaintance conversations (60%). In addition, timacy were orthogonal variables. However, the
as expected, callers and respondents initiated results of this study demonstrate, not surpris-
an equal number of closings in friend conver- ingly, that this is clearly not the case. The pur-
sations (50% each). A chi-square analysis in- poses for which friends, acquaintances, and
dicated that these differences between groups strangers initiate conversations are intimately
were clearly significant, x2(2, N = 56) = 9.03, tied to the needs of each of these kinds of re-
p .01. These results illustrate one way in lationship and cannot be considered as some-
which stranger and acquaintance conversations how separate from such needs. In other words,
are more asymmetrical than conversations be- intrinsic to the nature of friend relationships
tween friends. To some extent, this may be a is a desire to talk with the friend simply for
specific feature of telephone conversations, the pleasure of doing so; such a purpose would
where the roles of caller and respondent are make no sense in the context of a stranger re-
highly differentiated. However, it is possible lationship. Thus, to say that conversational
that this asymmetry may also hold for face-to- style is partly shaped by the purpose of the
face conversations, in that the person who ini- conversation is merely a more specific way of
tiates the conversation may be seen as having saying that the degree of intimacy between
the responsibility for ending it as well. Because members of a dyad affects the way in which
friends are presumed to welcome contact with they converse.
one another irrespective of who initiates it, the A conceptual issue raised by the results of
task of ending the conversation may, in general, the present study has to do with the ambiguity
be more equally shared. of the acquaintance relationship and the need
to consider in more detail how the degree of
Summary and Conclusions closeness of a relationship is differentiated in
experience. Some researchers (e.g., Altman &
The findings from the present study support Taylor, 1973; Hatfield, Utne, & Traupmann,
the notion that the closer a dyad's relationship, 1979; Huesmann & Levinger, 1976; Waister,
the more likely the dyad is to use a highly im- Berscheid, & Waister, 1976) have taken the
plicit conversational style. In comparison with view that relationships vary along one or more
strangers, friends used more implicit openings, continuous dimensions (e.g., equitable-ineq-
raised more topics, were more responsive to uitable, high exchange-low exchange). On this
one another (in terms of asking more questions view, friends would be seen as being at one
and making more minimal responses while pole of a dimension like closeness, strangers
listening), and used more complex forms of would be seen as being at the other pole, and
closing. The pattern of findings for acquain- acquaintances would be seen as being at the
tances was more mixed: In some respects (e.g., midpoint. However, other researchers (e.g.,
680 GAIL A. HORNSTEIN

Clark & Mills, 1979; Rubin, 197 4) have argued they do on park benches or at bus stops. It is
that certain kinds of relationships are quali- difficult to know how a broader range of
tatively different from others (e.g., exchange stranger conversations could be collected
versus communal relationships, liking versus without violating the participants' right to pri-
loving relationships). On this view, friends vacy, but this issue deserves further thought.
would be seen as having intimate relationships, In addition to the directions for future re-
whereas strangers and acquaintances would be search noted earlier, an important next step
seen as having nonintimate relationships. involves examining the microgenesis of close
The findings of the present study seem to relationships and the changes in conversational
support this second view, in that in a number style that accompany it. By following a group
of respects, acquaintances appeared to be more of dyads over a period of months from the time
similar to strangers than they were to friends. their members first met, the development of
Such a view would imply that when a rela- an increasingly intimate conversational style
tionship becomes a friendship, it undergoes a could be traced in a detailed and systematic
qualitative and not simply a quantitative way. Studies of this sort would serve to extend
transformation from a nonintimate to an in- the findings of cross-sectional research such as
timate relation. The best way to determine that reported here. An initial step toward this
whether this is in fact the case would be to goal has been taken by Hornstein and Truesdell
study relationships as they were developing so (1984), who studied female acquaintances over
that the participants' experience of closeness a 3-month period as they moved toward be-
at different points in time could be repeatedly coming friends. The results of this study pro-
assessed to see whether there is a different sort vide some support for the developmental hy-
of change involved in moving from acquain- pothesis, in that by the end of the third month,
tance to friend than there is in moving from acquaintances had begun to show an increase
stranger to acquaintance. in conversational intimacy and in other re-
A central task for such research would be spects did not differ from friends. However,
to determine whether the qualitative difference this initial study also made clear that a longer
in closeness hypothesized here as characteristic data collection period may be necessary to fully
of friendship is specific to friendships between examine the dynamics of intimacy as they un-
women or whether it is also true for men. fold within an ongoing relationship.
Women's friendships have been shown to be
highly intimate (Armstrong, 1968; Weiss & References
Lowenthal, 1973), in contrast to men's friend- Altman, I., & Taylor, D. A. ( 1973). Social penetration: The
ships, which are somewhat superficial (Lewis, development of interpersonal relationships. New York:
Holt, Rinehart & Winston.
1978; Powers & Bultena, 1976). Furthermore, Aries, E. ( 1976). Interaction patterns and themes of male,
female friends report talking about more per- female, and mixed groups. Small Group Behavior. 7, 7-
sonal and intimate topics and in more depth 18.
than male friends do (Aries, 1976; Johnson & Armstrong, J.C. (1968). Perceived intimate friendship as
a quasitherapeutic agent. Journal of Counseling Psy-
Aries, 1980). Taken together, these findings chology. 15, 137-141.
suggest that men may experience less of a dif- Bailey, B. J. R. (1977). Tables of the Bonferroni t statistic.
ference than women between acquaintance Journal ofthe American Statistical Association, 72. 469-
and friend relationships, but this issue has not 478.
been directly investigated and thus remains an Ball, D. T. ( 1968). Toward a sociology of telephones and
telephoners. In M. Truzzi (Ed.), Sociology and everyday
empirical question. life (pp. 59-75). Englewood Cliffs, NJ: Prentice-Hall.
One methodological feature of the present Becker, C. S. (1973). A phenomenological explication of
study that needs to be addressed in future work friendship: As exemplified by most important college
concerns the development of procedures that women friends (Doctoral dissertation, Duquesne Uni-
would allow for the collection of a wider range versity, 1973). Dissertation Abstracts International, 34,
1739B.
of stranger conversations. In the present study, Bell, J. (1975). The samaritan concept of befriending.
the restricted range of stranger conversations British Journal of Social Work. 5. 413-422.
was partly the result of using telephone con- Berger, P., & Kellner, H. (1964). Marriage and the con-
struction of reality. Diogenes, 46, 1-24.
versations as the source of data; it is simply Clark, M. S., & Mills, J . (1979). Interpersonal attraction
not the case that strangers chat with one an- in exchange and communal relationships. Journal of
other on the telephone in the same ways that Personality and Social Psychology. 37, 12-24.
INTIMACY IN CONVERSATIONAL STYLE 681

Cozby, P. C. (1973). Self-disclosure: A literature review. A developmental interpersonal perspective. In H. Otto

Psychological Bulletin, 79, 73-91. (Ed.), Love today: A new exploration (pp. 135-150). New
Cohen, J. (1960). A coefficient of agreement for nominal York: Association Press.
scales. Educational and Psychological Measurement, 20, Powers, E., & Bultena, G. (1976). Sex differences in inti-
37-46. mate friendships of old age. Journal of Marriage and
Fishman, P. M. ( 1978). Interaction: The work women do. the Family. 38, 739-747.
Social Problems, 25, 397-406. Purvis, J. A., Dabbs, J.M., & Hopper, C.H. (1984). The
Fishman, P. M. ( 1979). What do couples talk about when "opener": Skilled user of facial expression and speech
they're alone? In D. Butturff & E. L. Epstein (Eds.), pattern. Personality and Social Psychology Bulletin, JO,
Women's language and style (pp. 11-22). Akron: Uni- 61-66.
versity of Ohio Press. Rose, S. ( 1979, April). Maintenance and functions of
Gaebelein, J. (1976). Self-disclosure among friends, ac- friendship in early adulthood. Paper presented at the
quaintances, and strangers. Psychological Reports, 38, meeting of the Eastern Psychological Association, Phil-
967-970. adelphia.
Garfinkel, H. (1967). Studies in ethnomethodology. Engle- Rubin, Z. (1974). Lovers and other strangers: The devel-
wood Cliffs, NJ: Prentice-Hall. opment of intimacy in encounters and relationships.
Hatfield, E., Utne, M. K., & Traupmann, J. ( 1979). Equity American Scientist, 62, 182-190.
theory and intimate relationships. In R. Burgess & T. L. Sacks, H., Schegloff, E., &Jefferson, G. (1974). A simplest
Huston (Eds.), Social exchange in developing relation- systematics for the organization of tum-taking in con-
ships (pp. 99-133). New York: Academic Press. versation. Language, 50, 696-735.
Hornstein, G. A., & Truesdell, S. E. (1984). The develop- Sadler, W. A. ( 1970). The experience of friendship. Hu-
ment of intimate conversation in close relationships. manitas, 6, 177-209.
Unpublished manuscript, Mount Holyoke College. Schegloff, E. ( 1968). Sequencing in conversational openings.
Huesmann, L. R., & Levinger, G. ( 1976). Incremental ex- American Anthropologist, 70, 1075-1095.
change theory: A formal model for progression in dyadic Schegloff, E. ( 1979). Identification and recognition in tele-
social interaction. In E. Waister & L. Berkowitz (Eds.), phone conversation openings. In G. Psathas (Ed.), Ev-
Equity theory: Toward a general theory of social inter- eryday language: Studies in ethnomethodology (pp. 23-
action (pp. 191-229). New York: Academic Press. 78). New York: Irvington.
Johnson, F. L., & Aries, E. ( 1980). Close friendship in Schegloff, E. A., & Sacks, H. ( 1973). Opening up closings.
adulthood: Conversational content between same-sex Semiotica, 8, 289-327.
friends. Unpublished manuscript, University of Massa- Sexton, R. E., & Sexton, V. S. ( 1982). Intimacy: A historical
chusetts/Amherst College. perspective. In M. Fisher & G. Stricker (Eds.), Intimacy
Joos, M. ( 1962). The five clocks (Publication No. 22). (pp. 1-20). New York: Plenum Press.
Bloomington: Indiana University Research Center in Suttles, G. D. ( 1970). Friendship as a social institution. In
Anthropology, Folklore, and Linguistics. G. J. McCall, M. M. McCall, N. K. Denzin, G. D. Sut-
Jourard, S. M. ( 1971 ). Self-disclosure: An experimental tles, & S. B. Kurth (Eds.), Social relationships (pp. 95-
analysis of the transparent self. New York: Wiley. 135). Chicago: Aldine.
Kurth, S. (1970). Friendship and friendly relations. Jn G. J. Walker, L. S., & Wright, P. M. ( 1976). Self-disclosure in
McCall, M. M. McCall, N. K. Denzin, G. D. Suttles, & friendship. Perceptual and Motor Skills, 42, 735-742.
S. B. Kurth (Eds.), Social relationships (pp. 136-170). Waister, E., Berscheid, E., & Waister, G. W. (1976). New
Chicago: Aldine. directions in equity research. In E. Waister & L. Ber-
Levinger, G. ( 1978, August). Models ofclose relationships: kowitz (Eds.), Equity theory: Toward a general theory of
Some new directions. Paper presented at the meeting of social interaction (pp. 1-42). New York: Academic Press.
the American Psychological Association, Toronto, Can- Weiner, S. L., & Goodenough, D. R. ( 1977). A move toward
ada. a psychology of conversation. In R. 0. Freedle (Ed.),
Levinger, G., & Snoek, J. D. (1972). Attraction in rela- Discourse processing and comprehension (pp. 213-225).
tionship: A new look at interpersonal attraction. Mor- Norwood, NJ: Ablex.
ristown, NJ: General Leaming Press. Weiss, L., & Lowenthal, M. F. ( 1973, August). Perceptions
Lewin, K. (1948). Resolving social conflicts. New York: and complexities offriendship in four stages ofthe adult
Harper & Row. life cycle. Paper presented at the meeting of the American
Lewis, R. A. ( 1978). Emotional intimacy among men. Psychological Association, Montreal, Canada.
Journal of Social Issues, 34, 108-121. Werner, H., & Kaplan, B. (1963). Symbol formation . New
Martlew, M. ( 1980). Mothers' control strategies in dyadic York: Wiley.
mother/child conversations. Journal ofPsycholinguistic Wike, E. L. ( 1971 ). Data analysis. Chicago: Aldine-Ath-
Research, 9, 327-347. erton.
Miller, L. C., Berg, J. H., & Archer, R. L. (1983). Openers: Won-Doomink, M. J. (1979). On getting to know you:
Individuals who elicit intimate self-disclosure. Journal The association between the stage of relationship and
of Personality and Social Psychology. 44, 1234-1244. reciprocity of self-disclosure. Journal of Experimental
Morton, T. L. (1978). Intimacy and reciprocity of exchange: Social Psychology. 15, 229-241.
A comparison of spouses and strangers. Journal ofPer- Wright, P.H. (1974). The delineation and measurement
sonality and Social Psychology. 36, 72-81. of some key variables in the study of friendship. Rep-
O'Mara, J. J. (1973). Characteristics of dyadic conversation: resentative Research in Social Psychology. 5, 93-96.
An ecological approach (Doctoral dissertation, Du-
quesne University, 1973). Dissertation Abstracts Inter-
national, 34, 6220B. Received July 23, 1984
Orlinsky, D. E. ( 1979). Love relationships in the life cycle: Revision received November 6, 1984 •
View publication stats