Vocal behavior in the dyadic interactions of preadolescent and early adolescent friends and acquaintances.

Feldstein, Stanley ; Field, Tiffany

An increasing number of studies are concerned with the extent to which the vocal behavior of pairs of individuals in conversational interactions with each other shows mutual vocal coordination or entrainment. Most of these studies have involved the interactions of adults and young adults (e.g., Crown, 1991; Feldstein & Welkowitz, 1987; Field et al., 1992) and, more recently, the interactions of infants with mothers and with strangers (e.g., Beebe, Alson, Jaffe, Feldstein, & Crown, 1988; Cohn & Tronick, 1988; Jaffe, Beebe, Feldstein, Crown, & Jasnow, 2001; Jasnow & Feldstein, 1986). The study reported here examined the interactions of preadolescent dyads. The study had three major aims. The first was to determine whether, and how much, entraimnent, or what we call coordinated interpersonal timing, occurs in the dialogues of preadolescent pairs. Coordinated interpersonal timing, or CIT, refers to changes in the temporal patterns of one person in a conversation as a function of changes in those of the other pers on. The second aim was to provide descriptive statistics of the states that comprise the vocal patterning of preadolescent interactions. The last, but not at all the least aim, was to compare, in terms of the state durations and frequencies and the coefficients of CIT, the interactions of friends and acquaintances in mixed- and same-gender dyads. Thus, the study is primarily an analysis of the temporal structure of preadolescent dialogues. This information was expected to extend our knowledge of what may be the basis for subsequent differences in the social interactions of friends and acquaintances. This expectation is based upon a "dyadic systems" position which holds that the two-person group is a basic psychological unit in which personality is originally formed (Sullivan, 1947) and in which the behavior of one of the individuals is determined by the behavior of both individuals (Jaffe, Beebe, Feldstein, Crown, & Jasnow, 2001). It has been shown (Jaffe et al., 2001) that the degree to which the temporal rh ythms of mothers and their infants are coordinated not only initiates, for four-month-old infants, the formation of an adult dialogue structure prelinguistically, but predicts the quality of the mother-infant relationship that will have developed by age 12 months.

The few investigations that have examined the chronography of children's interactions were concerned with whether the conversational time patterns of the children exhibited CIT, as well as the effects of age, gender, and ethnicity on the conversational time patterns of the children. One study (Welkowitz, Bond, & Feldstein, 1984a) of Hawaiian children found that the vocal time patterns are stable indices of children's conversational behavior, and that the patterns seem to vary as a function of the gender and ethnicity of the conversational pairs. Another study (Welkowitz, Bond, & Feldstein, 1984b) of Japanese-American children and adults in mixed- and same-gender pairs found gender effects for the adults but not for the children. Two earlier studies (Garvey & BenDebba, 1974; Welkowitz, Cariffe, & Feldstein, 1976) seemed to indicate that the development of CIT is positively related to age. However, none of these studies involved preadolescents, and the techniques for assessing CIT were relatively crude. In the present study, CIT was estimated by the use of time-series regression analyses.

METHOD

Participants

The 30 female and 26 male pre- and early adolescents who participated in the study were recruited from two sixth-grade classes at the West Laboratory Elementary School for a study by Field et al. (1992). The average age of the participants was 11.5 years, and the friends knew each other for an average of 4.2 years. The acquaintances knew each other for approximately five months. The selection of friends and acquaintances was made on the basis of a sociogram that presented cartoon faces with messages in balloon-like clouds coming from the cartoon faces such as "I know ______ the best" and "I know ______ the least." The choices of the students were validated by their teachers, who were asked to rank order, for each student, two of the student's closest friends (Field et al., 1992). For the present study, it is important simply to note that the boys and girls were assembled into same- and mixed-sex pairs, and that each pair participated in a 10-minute, face-to-face interaction across a small table, about any top ic or topics they desired, and that the interaction was audiotaped such that each voice was on a separate channel.

Coding Vocal Interactions

The coding of the vocal behavior of a dialogue is accomplished by means of the direct input of the two audio signals, one for each person, into a specialized computer system known as the Automated Vocal Transaction Analyzer (AVTA; Jaffe & Feldstein, 1970). AVTA first performs an analogue-to-digital conversion. Specifically, the two channels of incoming audio signals are sampled synchronously every 250 msec to determine whether the signal in each channel is on or off, without regard to the frequency or intonational characteristics of the sounds. The sole requirement is that the sounds be within the range of human hearing. These observed time series are stored digitally in the computer in the form of a sequence of four number codes that index the four observable dyadic states: when one signal is on and the other is off, the code is "1"; if vice versa, the code is "2"; if both signals are on, the code is "3"; or if both are off, the code number is "0." In the present study, the AVTA software transforms these dec imal numbers into a set of dialogic vocal states (defined next) and averages their durations for a fixed time unit. Although the observable states 1 and 2 distinguish between the speakers, the coactive states 3 and 0 do not. To have them do so requires a superordinate characteristic of dialogue that allows for unequivocal identification of the speaker. The speaking turn is the characteristic that provides such identification. It begins the instant either participant vocalizes alone, and it is held by that participant until the other vocalizes alone, at which point the other participant takes the turn. By virtue of the turn, the AVTA system is able to compute five vocal states (Feldstein & Welkowitz, 1987; Jaffe & Feldstein, 1970) that are subsumed by the turn: vocalizations (V), pauses (P), switching pauses (SP), and interruptive and noninterruptive simultaneous speech (ISS and NSS, respectively). Figure 1 diagrams an interactional sequence. The numbered line at the bottom represents 250-msec units. The arrow s that point down denote the end of speaker 1's turns; those that turn up denote the end of speaker 2's turns.

A vocalization is a continuous utterance containing no silence as long as 250 msec (silences less than 250 msec are attributable to stop consonants in speech and are bridged by our A/D converter). Joint silence is classified in terms of its outcome. A pause is a joint silence greater than 250 msec bounded by the vocalizations of the speaker who holds the turn. A switching pause is a joint silence greater than 250 msec initiated by the speaker who holds the turn, but terminated by the other speaker, who thereby gains the turn. In relatively unconstrained dialogues, it is assigned, for theoretical and empirical reasons, to the speaker whose turn it terminates.

Similarly, simultaneous (joint) speech is that uttered by the person who does not hold the turn during a vocalization of the person who does and it, too, is classified in terms of its outcome. Noninterruptive simultaneous speech (NSS) is a segment of speech that begins and ends while the person who holds the turn vocalizes continuously. Interruptive simultaneous speech is part of a speech segment that begins while the person who holds the turn is vocalizing but continues when he or she stops. Only that part of the segment uttered coactively is considered interruptive simultaneous speech. The trailing portion, inasmuch as it is then a unilateral utterance, marks the beginning of the turn for the person who initiated the interruption and is, therefore, considered his or her vocalization.

The AVTA system computes, for turns and each of the vocal states, the average duration for the entire interaction session but also the average duration for every five seconds of the interaction. The latter computation is for the purpose of subjecting the data to time-series analyses.

RESULTS

Frequencies and Durations

The descriptive statistics of the states for the interactions studied are presented in Tables 1 and 2. Separate factorial ANOVAs were used to compare speaking-turn frequencies and durations, the vocal-state frequencies and durations of the pairs of friends and acquaintances, and their gender and dyad gender (whether same-gender or cross-gender). The results provide no evidence (with regard to frequencies and durations) that gender or dyad gender (same or mixed), as a main effect, differentiated the two groups for any of the vocal states. Frequency of turns, however, yielded au F ratio that, although only bordering on significance, F(1, 108) = 3.53, p = .06, [euro] = .15, tended to suggest that the girls used more turns than did the boys.

The participants' use of interruptive and noninterruptive simultaneous speech yielded significant differences. The boys engaged in more frequent interruptive and noninterruptive simultaneous speech (ISS and NSS, respectively) when they were with acquaintances than when they were with friends whereas the girls engaged in more interruptive and noninterruptive simultaneous speech when they were with friends than when they were with acquaintances, F(1), 108) = 17.05, p = .00, [euro] = .36, and F(1, 108) = 14.31, p = .00, [euro] = .33, respectively. Figure 2 graphs the influence of gender by dyad type on ISS and Figure 3 graphs the same type of interaction for NSS.

Although not formally significant, there was some tendency of friends to generally use longer durations of ISS in their interactions than did acquaintances, F(1, 108) = 3.77, p = .06, [euro] = .14, but the analysis also yielded three significant two-way interactions of gender by dyad type, F(1, 108) = 22.22, p = .00, [euro] = .40, gender by dyad gender, F(1, 108) 6.60, p = .01, [euro] = .22, and dyad type by dyad gender, F(1, 108) = 5.88, p .02, [euro] = .20. The boys used longer durations of ISS than did girls when they were with acquaintances, whereas the girls used longer ISS durations than did boys when they were with friends (Figure 4). However, the girls used somewhat shorter segments of ISS than did the boys when they were in mixed (cross-gender) dyads, but longer segments of ISS than the boys when they were in same-gender dyads (Figure 5). Finally, friends used longer segments of ISS in same-gender dyads than in cross-gender dyads, but acquaintances did the opposite (Figure 6).

Coordinated Interpersonal Timing

Time-series regression. In order to assess CIT, the turn durations and the average state durations of each pair were subjected to bivariate time-series regression analyses that examined the coordination of each partner's vocal behavior with the prior vocal behavior of the other partner. The ARIMA modeling procedures of SPSS were used to "pre-whiten" the data, and the ACF subprogram of SPSS Trends was used to allow for visual and statistical tests of the best-fitting model. After the model was selected, the AREG subprogram computed the equations. The analyses yield squared semi-partial regression coefficients that are the proportions of variance of one partner's vocal behavior accounted for by the other partner's vocal behavior (Table 3). The square roots of these coefficients index the strength of the coordination between the partners.

After completing the time-series regression analyses, a special meta-analytic procedure was used to determine, treating each dyad as a separate experiment, whether the groups of friends and acquaintances each exhibited, in general, significant degrees of coordination with the prior temporal behaviors of their partners. The meta-analytic approach suggested by R. Rosenthal (personal communication, September, 1988) was employed and involved three simple steps for a particular group of CIT values: (a) transform the probability values associated with the coefficients to normal standard deviate scores (z scores); (b) square the z scores to yield chi-squares, each with a degree of freedom; and (c) sum the chi-squares for each group to provide a chi-square with degrees of freedom equal to the number of chi-squares in the sum. The summed chi-square value indicates whether the coefficients (the square roots of the CIT values) are significantly different from zero. The reason for this approach is that the usual meta-ana lysis tests a directional hypothesis, whereas in this study the coefficients support the hypothesis that coordinated interpersonal timing has occurred whether they are positive or negative. The only value that does not support the hypothesis is zero. A separate meta-analysis was computed for each of the vocal measures for each of the groups. The results (Table 4) of the analyses reveal that, in general, there was a considerable degree of temporal coordination across groups and states. An exception was that neither friends nor acquaintances coordinated the durations of their vocalizations with the prior vocalizations of their partners.

Comparisons of Gender and Groups

Multiple regression analyses were performed to determine whether CIT was influenced by gender, dyad type, and dyad gender. The dependent variable in each of the hierarchical equations was comprised of the squared semi-partial coefficients ([r.sup.2.sub.sp]) indexing the CIT of one of the vocal states. The independent variables were gender, dyad type, dyad gender, and the products of the three variables that carry the interaction effects.

The results yielded an interaction of gender by dyad type, indicating that although the boys' turn, F(1, 108) = 5.187, p = .02, [r.sub.sp] = -.21 (Figure 7), and switching-pause, F(1, 108) = 4.611, p = .03, [r.sub.sp] = -.20 (Figure 8), durations showed greater coordination with the prior turns and switching pauses of their partners when they were friends than when they were acquaintances, the girls' turn and switching-pause durations showed greater coordination with those of their partners when they were acquaintances than when they were friends. In short, they boys coordinated the timing of their vocal behavior to a greater extent when they were friends than when they were acquaintances, and the girls did the opposite.

The results also indicate that the girl's pause durations show greater coordination with the prior pauses of their partners than do those of the boys, regardless of whether the pairs were friends or acquaintances or of same or mixed gender. F(1,108) = 5.78, p = .02, [r.sub.sp] = .22. Unlike switching pauses, which comprise an interpersonal variable in the sense of marking the boundary between the speakers' turns and the fact that their occurrences and durations are jointly determined, pausing is an intrapersonal behavior. That the girls coordinated the durations of their pauses to a greater extent than did the boys may indicate that they were more attentive and involved in their interactions than were the boys.

Finally, there was an interaction between dyad type and dyad gender, F(1, 108) = 5.l68,p = .02, [r.sup.sp] = -.21 (Figure 9), such that friends coordinated the durations of their ISS with those of their partners to a significantly greater extent in same-gender dyads than in mixed-gender dyads, whereas acquaintances did the opposite. These results are similar to those found previously for differences in the durations of ISS rather than in their coordination, and suggest that interruptive behavior is the one state that is sensitive to the gender composition of a dyad, but only when mediated by the degree of relationship between the partners.

DISCUSSION

Interruptive behavior is important not only because it has been tagged as a significant predictor of coronary artery disease (Siegman, Feldstein, Tomasso, Ringel, & Lating, 1987), but also because it is thought to reflect dominance in interpersonal interactions. Most, if not all of the relevant literature indicates that men interrupt more than do women, that women are interrupted more than are men, and that, in mixed-gender dyads, men interrupt women more than women interrupt men (Deaux, 1976; Dindia, 1987; Eakins & Eakins, 1978; Henley, 1977; Zimmerman & West, 1975). Dindia (1987), who published a relatively brief but useful review of the literature, argued that these conclusions are based upon both inadequate designs and faulty statistical analyses. Her own study, which attempted to correct such problems, found that men did not interrupt more than did women, and women did not get interrupted more than did men. She also found that mixedgender pairs produced more interruptions than same-gender pairs and that interruptions were differentially distributed in mixed- and same-gender pairs, but not as a function of gender. Even more persuasive is an experiment by Crown and Cummins (1998) which demonstrated that, despite the fact that the conversations they heard were chosen because the men and women in them had equal numbers of objectively measure interruptive behaviors, judges perceived women in the study to use more interruptive behavior than did the men.

Our results suggest that the frequency, duration, and even the interpersonal coordination of interruptive behavior depend upon the social-relationship context within which it is used. In particular, whether the pairs of boys or girls in this study engaged in more ISS depended primarily upon whether they were friends or acquaintances. Even the interruptive behavior in mixed- or same-gender dyads appeared to differ as a function of whether the partners were friends or acquaintances. That there were no differences in interruptive behavior simply as a function of gender accords with the findings of Dindia (1987) and Crown and Cummins (1998), and the findings with regard to dyad gender are not at variance with those of Dindia (1987) or Natale, Entin, and Jaffe (1979). That there were no simple differences between the interruptive behaviors of boys and girls argues against the prevalent notion that interruptive behavior reflects interpersonal dominance, which is a second-order inference based upon men's alleged gre ater use of such behavior. It is also in line with the adult findings of Welkowitz, Bond, and Feldstein (1984b).

It may be more profitable to consider the interruptive behavior of the pre- and early adolescents in somewhat broader terms. Deaux (1977) proposed that the strategy of men in interactive contexts tends to be competitive whereas that of women tends to be affiliative. Although the constructs of competitiveness and dominance share certain attributes, to be competitive is not necessarily to be domineering. Might it be that the boys' more frequent use, and the girls' less frequent use of ISS and NSS when they were acquaintances reflect such strategies? It may well be that having to interact with acquaintances elicits competitive behaviors on the part of boys because of the inherent uncertainties about each other and the situation. But because the situation is an interpersonal one, the same uncertainties may elicit affiliative behavior on the part of girls. On the other hand, it may not be as necessary for boys to be as competitive, or for girls to be as affiliative when they are interacting with friends.

The finding that the girls engaged in more coordinated vocal behavior when they were acquaintances than when they were friends while the boys did the opposite presents an interesting comparison with the findings of a number of previous studies. Field and her colleagues (Field et al., 1992), using spectral analysis, found that there was greater coherence in the vocal behavior of pairs of acquaintances than of pairs of friends without regard to gender, which supports the present findings to some extent. Crown (1991) investigated the coordination of vocal timing in the conversations of college students who liked each other, disliked each other, and were unacquainted with each other. She found that the most coordination occurred in the conversations of the pairs who were unacquainted and the pairs who disliked each other. But all of her students were women. The results of the current study raise the question of whether the temporal coordination of men might be greater in their conversations with friends than with acquaintances.

The rather complex pattern of results also provides a salutary caution that things are often more complicated than they might otherwise appear. Specifically, these results point to the importance of considering the context of social interactions as a prime determinant of the function of any particular behavior. It is difficult, if not impossible to interpret, in the sense of main effects, the meaning and function of a particular behavior. We found that understanding the meaning of CIT in the conversations of these pre- and early adolescents required examination of the interactions rather than main effects. Clearly, CIT plays an important role in structuring these dialogues. However, one cannot interpret the role of CIT without considering the gender of each participant, the gender composition of the dyad, and the nature of the relationship (friend or acquaintance). The necessity of considering higher-order interactions replicates the findings of Jaffe and colleagues (2001) regarding the role of CIT in the dev elopment of adult dialogue. They found that the predictive utility of CIT varied significantly as a function of infant gender, whether the infant was interacting with mother or a stranger, the site of the interaction (laboratory or home), and the dependent variable under consideration (attachment or cognitive development). The degree of coordination between infant and adult assumed a very different meaning depending upon the context examined in any particular set of interactions. While the scope of the present study is more circumscribed than that of Jaffe et al., it too replicates the essential fact that the meaning of CIT can best be understood through the sophisticated analysis of higher-order interrelations. The subtleties of social interaction are not well captured solely by main effects.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

Table 1

Means and Standard Deviations of the Speaking-Turn and Vocal-State
Frequencies that Comprise Vocal Behavior

 T V P SP NSS ISS

MALE
SAME GENDER

 Friends
 M 92.11 178.94 86.83 45.39 26.22 21.50
 SD 15.68 56.06 54.38 14.92 14.53 11.09
 Acquaintances
 M 103.45 180.41 76.95 44.14 52.86 36.09
 SD 30.20 51.89 36.97 19.53 54.78 22.03

MIXED GENDER

 Friends
 M 107.75 169.13 61.38 44.63 40.38 37.63
 SD 27.37 29.17 34.06 8.37 21.56 23.13
 Acquaintances
 M 88.75 145.25 56.50 46.13 29.50 24.75
 SD 20.84 53.60 42.50 18.16 31.18 19.48

FEMALE
SAME GENDER

 Friends
 M 89.33 174.42 84.70 34.08 74.29 32.46
 SD 43.28 77.85 54.30 23.89 77.16 16.99
 Acquaintances
 M 77.50 165.61 88.11 36.56 25.56 19.78
 SD 26.29 55.07 54.49 10.69 21.54 17.56

MIXED GENDER

 Friends
 M 107.75 193.25 85.50 49.13 39.88 31.88
 SD 27.47 39.83 26.84 12.37 23.81 18.73
 Acquaintances
 M 88.63 188.63 100.00 44.38 31.25 23.63
 SD 20.79 56.81 54.87 19.17 32.84 20.85

Note. T = Speaking Turns

V = Vocalizations

P = Pauses

SP = Switching Pauses

ISS = Interruptive Simultaneous Speech

NSS = Noninterruptive Simultaneous Speech.

Table 2

Means and Standard Deviations of the Speaking-Turn and Vocal-State
Durations that Comprise Vocal Behavior

 T V P SP NSS ISS

MALE
SAME GENDER

 Friends
 M 3.176 1.057 .640 .819 .661 .385
 SD 1.274 .472 .218 .329 1.078 .081
 Acquaintances
 M 3.090 1.264 .653 .693 .470 .468
 SD 1.142 .412 .184 .171 .155 .214

MIXED GENDER

 Friends
 M 2.682 1.117 .682 .673 .404 .377
 SD 1.391 .207 .230 .263 .102 .067
 Acquaintances
 M 2.484 1.007 .706 .774 .469 .456
 SD .722 .445 .233 .249 .123 .226

FEMALE
SAME GENDER

 Friends
 M 4.887 1.601 .610 .784 .571 .651
 SD 5.585 1.772 .294 .586 .352 .405
 Acquaintances
 M 4.221 1.185 .607 .749 .353 .336
 SD 3.343 .429 .157 .252 .084 .079

MIXED GENDER

 Friends
 M 3.094 1.174 .630 .721 .404 .413
 SD .938 .326 .206 .205 .073 .088
 Acquaintances
 M 3.740 1.177 .659 .749 .384 .406
 SD 1.374 .406 .139 .167 .143 .202

Note. The average durations are in seconds. T = Speaking Turns; V =
Vocalizations; P = Pauses; SP = Switching Pauses; ISS = Interruptive
Simultaneous Speech; NSS = Noninterruptive Simultaneous Speech.

Table 3

Means and Standard Deviations of the Rs that Index the Coordination of
Each Partner's Temporal Behavior with the Prior Temporal Behavior of the
Other Partner

 T V P SP NSS ISS

MALE
SAME GENDER

 Friends
 M .216 .117 .150 .150 .142 .188
 SD .138 .047 .085 .057 .087 .109
 Acquaintances
 M .167 .109 .108 .105 .133 .123
 SD .142 .059 .066 .048 .082 .068

MIXED GENDER

 Friends
 M .220 .144 .094 .170 .138 .118
 SD .172 .059 .035 .064 .037 .037
 Acquaintances
 M .264 .113 .092 .128 .140 .158
 SD .261 .066 .031 .058 .081 .059

FEMALE
SAME GENDER

 Friends
 M .146 .108 .150 .122 .173 .142
 SD .138 .047 .106 .100 .109 .117
 Acquaintances
 M .240 .130 .168 .142 .133 .108
 SD .145 .054 .129 .071 .087 .047

MIXED GENDER

 Friends
 M .124 .119 .146 .118 .105 .123
 SD .123 .034 .058 .060 .053 .052
 Acquaintances
 M .296 .139 .160 .126 .126 .140
 SD .241 .049 .082 .102 .081 .056

Note. See Table 1 for definitions of the state symbols. The n for the
current coefficients is 56.

Table 4

Summary of Meta-Analyses of the Coefficients of Coordination for Friends
and Acquaintances

 V P SP ISS NSS

Friends
 Partner A
 R .34 .37 .34 .38 .39
 [chi square] 31.61 95.11 * 43.73 * 82.83 * 76.85 *
 Partner B
 R .33 .37 .38 .38 .36
 [chi square] 34.88 52.57 96.87 * 79.84 * 82.92 *

Acquaintances
 Partner A
 R .34 .36 .34 .34 .35
 [chi square] 30.28 66.05 * 27.75 29.41 55.47 *
 Partner B
 R .34 .35 .35 .36 .36
 [chi square] 32.43 77.53 * 49.28 * 53.79 * 59.70 *

Note. The chi-squares index the significance of the CIT associated with
the groups of friends and acquaintances. The Rs index the strength of
the coordination.

* P < .05

REFERENCES

Beebe, B., Alson, D., Jaffe, J., Feldstein, S., & Crown, C. L. (1988). Vocal congruence in mother-infant play. Journal of Psycholinguistic Research, 17, 245-259.

Cohn, J., & Tronick, E. (1988). Mother-infant face-to-face interaction: Influence is bidirectional and unrelated to periodic cycles in either partner's behavior. Developmental Psychology, 24, 386-392.

Crown, C. L. (1991). Coordinated interpersonal timing of vision and voice as a function of interpersonal attraction. Journal of Language and Social Psychology, 10, 29-46.

Crown, C. L., & Cummins, D. A. (1998). Objective vs. perceived interruptions in the dialogues of unacquainted pairs, friends, and couples. Journal of Language and Social Psychology, 17(3), 372-389.

Deaux, K. (1976). The behavior of women and men. Monterey, CA: Brooks/Cole.

Deaux, K. (1977). Sex differences. In T. Blass (Ed.), Personality variables in social behavior (pp. 357-377). Hillsdale, NJ: Erlbaum Associates.

Dindia, K. (1987). The effects of sex of subject and sex of partner on interruptions. Human Communication Research, 13, 345-371.

Eakins, B. W., & Eakins, R. G. (1978). Sex differences in human communication. Boston: Houghton Mifflin.

Feldstein, S., & Welkowitz, J. (1987). A chronography of conversation: In defense of an objective approach. In A. W. Siegman & S. Feldstein (Eds.), Nonverbal behavior and communication (2nd ed., pp. 435-499). Hillsdale, N.J.: Erlbaum Associates.

Field, T., Greenwald, P., Morrow, C., Foster, T., Guthertz, M., Healy, B., & Frost, P. (1992). Behavior state matching during interactions of preadolescent friends versus acquaintances. Developmental Psychology, 28, 242-250.

Garvey, C., & BenDebba, M. (1974). Effects of age, sex, and partner on children's dyadic speech. Child Development, 45, 1159-1161.

Gottman, J. M., Markman, H., & Notarius, C. (1977). Topography of marital conflict: A sequential analysis of verbal and nonverbal behavior. Journal of Marriage and the Family, 39, 461-577.

Henley, N. (1977). Body politics. Englewood Cliffs, NJ: Prentice-Hall.

Jaffe, J., & Feldstein, S. (1970). Rhythms of dialogue. New York: Academic Press.

Jaffe, J., Beebe, B., Feldstein, S., Crown, C. L., & Jasnow, M. D. (2001). Rhythms of dialogue in infancy. Monographs of the Society for Research in Child Development, 66(Serial No. 265), 1-131.

Jasnow, M. D., & Feldstein, S. (1986). Adult-like temporal characteristics of mother-infant vocal interactions. Child Development, 57, 754-761.

Natale, M., Entin, E., & Jaffe, J. (1979). Vocal interruption in dyadic communication as a function of speech and social anxiety. Journal of Personality and Social Psychology, 37, 865-878.

Siegman, A. W., Feldstein, S., Tomasso, C. T., Ringel, N., & Lating, J. (1987). Expressive behavior and the severity of coronary artery disease. Psychosomatic Medicine, 49, 545-561.

Sullivan, H. S. (1947). Conceptions of modern psychiatry. Washington, DC: William Alanson White Foundation.

Welkowitz, J., Bond, R. N., & Feldstein, S. (1984a). Conversational time patterns of Hawaiian children as a function of ethnicity and gender. Language and Speech, 27, 173-191.

Welkowitz, J., Bond, R. N., & Feldstein, S. (1984b). Gender and conversational time patterns of Japanese-American adults and children in same- and mixed-gender dyads. Journal of Language and Social Psychology, 3, 127-138.

Welkowitz, J., Cariffe, G., & Feldstein, S. (1976). Conversational congruence as a criterion of socialization in children. Child Development, 47, 269-272.

Zimmerman, D. H., & West, C. (1975). Sex roles, interruptions and silences in conversation. In B. Thorn & H. Henley (Eds.), Language and sex: Differences and dominance (pp. 105-129). Rowley, MA: Newbury House.

The authors are indebted to the Academic Computing Center of the University of Maryland, Baltimore County, for making their facilities and time readily available. They are also indebted to Dr. Michael Jasnow for his invaluable comments. Finally, the authors are grateful for the participation and cooperation of the teachers, parents, and children at the West Laboratory Elementary School in Miami, Florida.

Tiffany Field, Touch Research Institutes, University of Miami School of Medicine, Florida.

Reprint requests to Stanley Feldstein, Department of Psychology, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, Maryland 21250. Electronic mail may be sent to feldstei@)umhc.edu.