Attribution of psychological characteristics from masked and unmasked conversations

Attribution of psychological characteristics from masked and unmasked conversations

JOURNAL OF COMMUNICATION DISORDERS 387 14 (1981). 387-397 ATTRIBUTION OF PSYCHOLOGICAL CHARACTERISTICS FROM MASKED AND UNMASKED CONVERSATIONS JOA...

672KB Sizes 0 Downloads 29 Views

JOURNAL

OF COMMUNICATION

DISORDERS

387

14 (1981). 387-397

ATTRIBUTION OF PSYCHOLOGICAL CHARACTERISTICS FROM MASKED AND UNMASKED CONVERSATIONS JOAN WELKOWITZ

and LLOYD

KAUFMAN

New York University

SUE SADD

Vera Institute ofJustice, New York, New York

A specific contribution of this study was the development of a procedure for separating the semantic and physical parameters of conversation. This procedure permits testing of the hypothesis that coverbal dimensions of conversation, that is, patterns of sounds and silence, convey information to independent listeners about the psychological states of speakers engaged in conversation. The masking procedure destroyed intelligibility of the conversation, but not the sequences of sounds and silences. Listeners rated speakers on three dimensions-warmth, anxiety, and dominance-under two conditions-masked and unmasked. Significant correlations were found between the ratings under both masked and unmasked conditions and between the ratings and a number of physical parameters of speech.

Introduction The quality of an interaction between two persons is conveyed not only by what they say to each other, that is, the semantic or lexical content of their speech, but also by how they converse. Thus, people are frequently judged by the ways in which they talk to others as well as by the content of their conversation. Research evidence (reviewed by Feldstein and Welkowitz, 1978) shows that the mean duration of the parameters of conversational rhythm, that is, pauses and vocalizations, as well as the mean level of vocal intensity (Welkowitz, Feldstein, Finkelstein, and Aylesworth, 1972), are stable characteristics of a person’s interaction style. However, it has also been demonstrated that the stability of the parameters can be altered as a function of certain experimentally manipulated psychological and situational conditions, such as increased stress (Cassotta, Feldstein, and Jaffe, 1967), loss of visual cues (Jaffe and Feldstein, 1970), personality characteristics of the speakers (Marcus, Welkowitz, Feldstein, and Jaffe, 1970), and degree of perceived interpersonal similarity (Welkowitz and Feldstein, 1969, 1970). Thus, it is becoming increasingly apparent that the coverbal dimensions of

Address correspondence to Joan Welkowitz, New York University, Washington Place, New York, New York 10003. Q Elsevier North Holland, Inc., 1981 52 Vanderbilt Ave. New York, NY 10017

Psychology

Building,

0021-9924/81/05387-l

6

1$2.50

388

J. WELKOWITZ

et al

conversation, that is, the patterns of sounds and silences, convey important information about the intentions and psychological states of two speakers engaged in conversation. The general purpose of this study is to examine judgments of psychological attributes made of conversational partners by independent observers. There is a fairly extensive literature relating personal perception to nonlexical parameters of conversational partners. Kramer (1963), in reviewing the early research, pointed out, however, that in order to avoid contamination of vocal cues by content, speakers were generally asked to read a uniform passage. More recent studies tended to concentrate on trait attribution of speakers engaged in natural conversation (e.g., Markel, 1965; Ramsey, 1966, 1968; Scherer, 1972; Welkowitz and Kuc, 1973; Welkowitz and Feldstein, 1973; Sadd, Welkowitz and Feldstein, 1978). The difficulty in all these studies of judgments of speaker characteristics has arisen from the fact that it has not been possible to separate the semantic content, or meaning of conversation, from the physical parameters. The specific contribution of this study was the separation of these semantic and physical parameters so that the influence of each on the apparent character of a conversation could be assessed. This was done without altering the rhythms of intensity patterns in the dialogue. The masking procedure, detailed in the Procedures section (vi& infra), produced sounds that, when played through a loudspeaker, sounded like speech in a foreign tongue, but the actual content of the conversation was unintelligible. It appeared to us that the nature of the conversation, the tones of friendliness or enmity, pleasure, or despair, could be detected in this processed speech. This procedure differs considerably from simply speeding up or slowing down speech, because this latter procedure alters the rhythmic patterns, as well as the frequency content. Method Subjects

Subjects, both speakers and listeners, were introductory psychology students, fulfilling course requirements by participating in experiments. Speakers were four men and two women. Nine women served as listeners. All subjects were strangers to each other. Procedures

The stimulus materials were three dialogues recorded from six speakers. Speakers were separated into same-sex pairs. Two speakers were seated at a table

ATTRIBUTION

OF PSYCHOLOGICAL

CHARACTERISTICS

389

and separated by a distance of about 2 m. Each speaker had a microphone clipped to his or her clothing. Speakers were instructed not to smoke, chew gum, or move about. They were told that the purpose of the experiment was to find out the ways in which strangers get acquainted and were requested to speak to their partners for 36 min. The voice of one speaker was recorded on one track and that of the other speaker on the other track of a Sony stereophonic taperecorder. Preliminary tests enabled us to set the gain controls of each channel of the recorder so as to minimize crosstalk. The resulting tapes were processed so that each session yielded two tapes. One of these was of the unprocessed (unmasked) conversation. The other processed (masked) tape provided the semantically meaningless sequences of sounds and silences. This was accomplished by applying the original recorded speech to a square law device and then to a low-pass filter. The operation of squaring has the effect of doubling all input frequencies. Thus, a continuous frequency spectrum quite different from that in normal speech was generated. The low-pass filter was set to attenuate all frequencies above 1 kHz at the rate of 6 dB per octave. As squaring produces frequency doubling, this filter operation resulted in a sound that resembied real speech in its frequency contour, but intelligibility was totally lost. Each of the nine listeners was told that we were interested in the ways in which people evaluated three personality dimensions from conversation: anxiety, dominance, and warmth. They were also told that some of the conversations were unintelligible, but they were to use whatever cues were available to make their evaluations. Thus, all listeners were exposed to both the processed (masked) and unprocessed (unmasked) versions of the same conversations, but the two versions were not presented contiguously. The listeners were unaware that they were, in fact, listening to the same conversations. The conversations were listened to during two sessions, 1 wk apart, in scrambled order, but always alternating masked with unmasked. Subjects listened to the dialogue for 3 min, at the end of which time an electronic buzzer sounded for 1 sec. Subjects were told to rate quickly each dialogic participant on each psychological dimension, that is, anxiety, dominance, and warmth, on a five-point scale from low to high. Subjects were not required to make ratings if they considered the stimulus material inadequate. In such cases, that particular rating was omitted. Approximately 95% of the ratings were completed. As listeners were rating each of two speakers every 3 min, and because each dialogue lasted 36 min, each tape yielded 12 x 2 = 24 ratings for each psychological dimension (12 for each speaker). At the end of 36 min, all listening subjects also gave a global rating, on each psychological dimension, to each speaker. At the end of the second scoring session, subjects were asked to comment on the masked tapes. They reported that, on occasion, some words were intelligible but that the conversation could not be followed. No subject claimed to recognize

390

.I. WELKOWITZ

that the three tapes containing the unmasked tapes.

masked speech used the same conversation

Definition

of Rhythmic

et al

as on

Parameters

The basic parameters related to onlookers’ perceptions were frequencies and mean durations of vocalizations, pauses, switching pauses, and simultaneous speech. 1. Vocalization. A continuous (uninterrupted) segment of speech (sound) uttered by the person who has the floor. 2. Pause. An interval of joint silence bounded by the vocalizations of the person who has the floor. 3. Switching pause. An interval of joint silence bounded by the vocalizations of different participants, that is, it follows a vocalization by the participant who has the floor and is terminated by the vocalization of another participant who thereby obtains the floor. Thus, the switching pause is a silence that marks a switch of speakers. Because it occurs during the turn of the person who relinquished the floor, it is credited to him or her. 4. Simultaneous speech. Speech uttered by a participant who does not have the floor. On the basis of its outcome, simultaneous speech may be divided into two types: noninterruptive and interruptive. a. Noninterruptive simultaneous speech. Begins and ends while the participant who has the floor is talking. b. Iriterruptive simultaneous speech. Part of a speech segment that begins while the person who has the floor is talking and ends after that person has stopped. Only that portion of the segment uttered while the other person is still talking is considereed interruptive simultaneous speech. The remaiing portion is considered a vocalization. Thus, interruptive speech culminnates in a change of which participant has the floor while noninterruptive simultaneous speech does not.

Computer

Analyses

of the Vocal Parameters

The recorded tapes were processed using programs developed for the PDP-12 computer (Martz and Welkowitz, 1977). The system uses an analog-to-digital converter with two channels. The programs detect presence and absence of sound from each speaker at a sampling rate specified by the user. The sampling interval used in the present study was 100 msec. Additional programs were used to count frequencies and determine durations and variabilities.

ATTRIBUTION

OF PSYCHOLOGICAL

391

CHARACTERISTICS

Results Reliability

The reliability analyses served three purposes: (1) to assess reliability of the raters for each condition (masked, unmasked) for the three psychological attributes, warmth, anxiety, and dominance; (2) to assess the consistency among raters for each speaker; and (3) to assess the reliability of the average duration and frequencies of the rhythmic parameters of the speakers. The focus of this study is on the relationship between ratings by listeners and the vocal parameters of speakers. However, the size of these reliability coefficients necessarily affects the size of the validity coefficients, that is, the correlation between ratings and the rhythmic measures. Intrarater reliability. The purpose of this analysis was to assess the consistency of the raters across 12 3-min speech segments for the six tapes (three masked, three unmasked). Ratings for each personality dimension were averaged over the first six time segments, or for the first 18 min (T,) of the tape and for the second 18 min (TJ, and split half reliability coefficients were obtained. The coefficients for dominance (r, = 0.68) and warmth (r, = 0.61) in the unmasked condition were considerably higher than in the masked condition (r, = 0.53 and r, = 0.21, respectively). The coefficient for anxiety in the masked condition (r, = 0.46) was higher than in the unmasked condition (rxr = 0.39). Interrater reliability. Consistency among raters across the 12 time segments was assessed for each speaker by obtaining a series of intraclass correlation coefficients (Table 1). The inctraclass coefficients represent the consistency

I?traclass

Coefficients

TABLE 1 for Nine Raters for Each Speaker (Interrater Personality

Speaker

Warmth Unmasked

1 2 3 4 5 6

0.58* 0.61* 0.52 0.34 0.31 0.74**

*p < 0.05. **p < 0.01.

Attribute

Anxiety Masked 0.21 0.13 0.47 0.77** 0.75** 0.37

Unmasked 0.22 0.77*+ 0.67* 0.60* 0.23 0.50

Reliability)

Dominance Masked 0.49 0.21 0.79** 0.30 0.43 0.63*

Unmasked 0.61* 0.25 0.72** 0.24 0.30 0.63*

Masked 0.28 0.38 0.87** 0.39 0.19 0.45

392

J. WELKOWITZ

et al

across the nine raters for each speaker. There is some tendency for raters to be more consistent with each other for particular speakers. There is also a tendency for greater rater consistency in the unmasked condition, but not for a particular psychological dimension. (An exception is for dominance in the masked condition where only one coefficient was significant. However, this coefficient was the highest for any condition .) Generally, the variance among the reliability coefficients does not provide evidence for systematic error that would affect a particular validity coefficient. Reliabilities offrequencies and durations of the rhythmic parameters: Speaker reliabilities. Speaker reliabilities were obtained by correlating first the fre-

quency and then the mean duration for the first 18 min of the dialogue with the last 18 min (Table 2). The reliabilities for average durations of pauses and switching pauses are similar to those reported by Feldstein and Welkowitz (1978). The lowest reliabilities are observed for frequency of switching pauses and mean duration of interruptive and noninterruptive simultaneous speech, that is, the speakers showed the least consistency across time in the use of these cues. Correlations Conditions

The ratings attribute were and unmasked warmth, r = dominance. It

Reliability

between Ratings for Masked und Unmasked

of the speakers across the 12 segments for each personality averaged and Pearson correlations obtained between the masked conditions. The resulting coefficients were r = -0.10 (NS) for 0.81 @ < 0.05) for anxiety, and r = 0.92 (p < 0.01) for can thus be inferred that when ratings are obtained for anxiety and

Coefficients

for Frequency

TABLE 2 and Mean Duration

of Nonlexical

Reliability Nonlexical

Parameter

Vocalizations Pauses Switching pauses Noninterruptive simultaneous speech Interruptive simultaneous speech n As these coefficients reflect six tape recordings, Therefore, the alpha criterion level is 0.10. *p < 0.10.

**p < 0.05. ***p < 0.01.

Frequency

Parameters”

Coefficients Mean Duration

0.92*** 0.69 0.24 0.81** 0.98*** the power of the significance

0.70 0.72* 0.75s 0.44 -0.08 test is low.

ATTRIBUTION

dominance used. Comparison

OF PSYCHOLOGICAL

under masked and unmasked

CHARACTERISTICS

conditions,

similar

393

speaker cues are

of Segment and Global Ratings

The method of using a series of ratings at the end of small time segments was used in the hope of yielding more reliable data than those produced by global ratings, that is, data yielding higher validity coefficients when correlated with speech parameters. Global ratings may, in fact, depend more on the effect speakers have had on the listener during the last few minutes than across the entire 36 min period. For exploratory purposes, global ratings were compared with the average segment ratings by correlating the two sets (Pearson r). For the masked condition, the correlations between segment and global ratings were as follows: warmth (r = 0.29, NS), anxiety (r = 0.26, NS), and dominance (r = 0.95, p < 0.01). For the unmasked condition, the correlations between segment and global ratings were the following: warmth (r = 0.71, NS), anxiety (r = 0.90, p < 0.05), and dominance (r = 0.91, p < 0.05). In the unmasked condition, there is a definite trend for the global ratings to represent the average rating across time segments. The trend is much less clear in the masked condition, except for dominance. Validity CoefSicients-Relationship Nonlexical Cues

of Personality

Attributions

to

These coefficients were obtained by correlating the frequencies and average durations of the speech parameters for each conversational partner with mean ratings for the psychological dimensions warmth, anxiety, and dominance. These correlations were obtained separately for each condition, masked and unmasked, and for two time segments (1 and 2). Results are given in Table 3. The correlations show considerable fluctuation across time segments for anxiety-for both masked and unmasked conditions-with a definite trend toward increasing strength for time 2.This may represent, for this variable, a greater stabilization over time in the cues attended to by listeners in rating speakers. The coefficients for ratings of warmth are essentially zero under both conditions for both time segments, and therefore none reaches conventional levels of significance. Ratings of anxiety under masked and unmasked conditions are positively related to mean durations of pauses and switching pauses. Under both masked and unmasked conditions, mean duration of interruptive simultaneous speech and frequency of noninterruptive simultaneous speech are negatively related to ratings of anxiety. (In fact, when correlation coefficients are significant in both masked and unmasked conditions, they are always in the same direction.) In the

(MD)

Vocalizations

OF, Frequency; *p < 0.05. **p < 0.01.

-0.03

MD, mean duration; M, masked; U, unmasked

-0.12

~0.12

-0.28*

0.03

0.09

Interruptive (MD) simultaneous speech

-0.14

0.00

-0.15

-0.07

0.05

-0.18

-0.08

-0.05

PO.09

-0.18

-0.03

-0.10

0.17

-0.07

1

-0.29*

0.29*

-0.17

0.44**

-0.20

-0.37**

-0.21

2

Anxiety u 2

-0.28*

0.52**

-0.04

PO.07

-0.05 -0.23

-0.28*

-0.53**

-0.34*

-0.27*

0.27* -0.11

0.21 0.46**

0.34**

-0.17

-0.23

-0.20

2

0.48**

M

-0.39**

0.38**

-0.40**

0.33*

-0.22

-0.46** 0.28*

-0.53**

-0.15

1

-0.19

-0.48**

Attribute

0.14

0.11

0.23

-0.07

-0.09

-0.26

0.01

1

M

0.01

-0.11

0.11

Interruptive (F) simultaneous speech

-0.14

PO. 10 PO.05

-0.12

-0.05

0.08

-0.03

0.15

0.07

-0.03

0.15

0.00

-0.05

-0.01

2

-0.06

Nonintenuptive (MD) simultaneous speech

-0.17

-0.23

0.22

-0.01

-0.06

-0.09

1

IJ

-0.16

0.12

-0.04

0.14

0.23

-0.23

Noninterruptive (F) simultaneous speech

Switching Pauses (MD)

0.15

0.11

-0.19

2

0.15

-0.04

0.17

1

M

Warmth

Personality

TABLE 3 Coefficients between the Frequency and Mean Duration (MD) of the Nonlexical Parameters and Mean Ratings of Personality Conditions-Masked and Unmasked and for Time Segments 1 and 2 ”

Switching Pauses (F)

Pauses (MD)

Pauses (F)

(F)

Vocalizations

Parameter

Pearson Correlation

-0.36**

-0.10

-0.11

-0.13

-0.19

-0.38**

0.22

-0.29*

0.32*

-0.23

0.44**

-0.43**

-0.34*

0.41**

0.21

0.38**

-0.26

-0.23

-0.29* -0.28*

2

for Two

1

U

Attributions

ATTRIBUTION

OF PSYCHOLOGICAL

CHARACTERISTICS

395

masked condition, mean duration of vocalizations is negatively related to anxiety, that is, speakers with very short vocalizations are perceived as more anxious. In the unmasked condition, frequency of pauses, switching pauses, and vocalizations as well as mean duration of noninterruptive simultaneous speech are all negatively correlated with anxiety. In the masked and unmasked conditions for dominance, mean durations of vocalizations and noninterrruptive simultaneous speech are negatively related to ratings of dominance, whereas mean durations of pauses and switching pauses ale positively related to these ratings. Frequency of switching pauses and mean duration of interruptive simultaneous speech are positively related to ratings of dominance in the masked condition. In the unmasked condition, frequencies of vocalizations, pauses, and noninterruptive simultaneous speech are all negatively correlated with dominance

Discussion This study represents an attempt to identify those features in patterns of sound, rather than in the content of conversation, that are related to independent evaluations of psychological attributes. To do this it was necessary to separate the semantic content of conversation from the physical parameters of the and interruptions-without destroying the speech-pauses, vocalizations, rhythmic patterns of the dialogue. Ratings of the attributes of warmth, anxiety, and dominance were selected because of the recognition in the psychological literature of their importance in the development of a taxonomic basis for personality research. In an attempt to develop an observation language, Norman ( 1963) used 20 peer nomination rating scales drawn from the results of several analyses presented by Tupes and Christal (1958). Analyses yielded five orthogonal personality factors, labeled extroversion, agreeableness, conscientiousness, emotional stability, and culture. Digman and Takemoto (1978), in reanalyzing the correlations of seven studies using personality rating scales indicated that, indeed, the domain may be adequately described by the five factors described by Norman (1963). Three of the traits described by Norman+xtroversion, agreeableness, and emotional stabilityclosely resemble those selected for the present study, namely, dominance, warmth, and anxiety. The results of the present study indicate that the more anxious speaker, under both masked and unmasked conditions, is characterized as having longer interand intrapersonal pauses. (Pauses between speakers, that is, switching pauses, are referred to as interpersonal, because they signal a change in who holds the floor). Those speakers characterized as low on anxiety were the more frequent initiators

396

J. WELKOWITZ

et al

of simultaneous speech, both interruptive and noninterruptive. These results are particularly clear in the unmasked condition. In a study described by Feldstein, Alberti, Ben Debba, and Welkowitz (1974), frequencies with which participants initiated simultaneous speech were compared with their personality characteristics as indexed by the Cattell Sixteen Personality Questionnaire or the 16PFQ (Cattell, Eber, and Tatsuoka, 1970). These results indicated that those speakers characterized on the 16PFQ as relaxed, complacent, and secure tended to initiate more simultaneous speech than did those characterized as self-reproaching, tense, and frustrated. The present study indicates that the correlations between ratings of independent observers and frequency of simultaneous speech are consistent with the relationship previously found between the objective personality tests and simultaneous speech. The cues used by the listener to rate speaker dominance were fairly consistent across conditions. Generally, those speakers who paused longer and engaged in more frequent interruptive simultaneous speech were perceived as more dominant. Those engaged in more noninterruptive simultaneous speech were perceived as less dominant. As was found in previous studies, in which content was not masked (reviewed by Siegman, 1978), ratings of dominance yielded the highest interrater reliabilities. Ratings of warmth were not related to the rhythmic parameters. In a previous study by Welkowitz and Kuc (1973), similarity or congruence of mean durations of speakers’ switching pauses was found to be significantly correlated with independent ratings of warmth for the dyad. In the present study, subjects did not rate the dyad itself, but rather the characteristics of each participant. It may be that ratings of some personality variables, such as warmth, empathy, and genuineness, do indeed depend on the relationship that evolves between the speakers, that is, on the similarity of their speech patterns, whereas other characteristics depend primarily on the individual rhythmic patterns. In the present study, congruence was not estimated. Certainly it is an important dimension meriting further study. It would be difficult, if not impossible, to make direct comparisons between the present study and many previous studies. Not only have the conditions varied, most importantly, masking the content, but definitions of vocal parameters varied, depending on the researchers. Early studies relied on subjective evaluation of the physical parameters of speech, whereas more recent studies, using objectively computed physical measures, differed in defining such parameters as vocalizations, pauses, latency, and so forth. Using the computer to count and time simultaneous speech as well as pauses and vocalizations has provided additional nonlexical indices that appear to be significantly related to judgments of specific personality characteristics. It appears that these indices not only convey information to the speakers, but influence the judgments of independent listeners about the psychological attributes of the conversational partners.

ATTRIBUTION

OF PSYCHOLOGICAL

CHARACTERISTICS

391

References Cassotta, L., Feldstein, S., and Jaffe, J. (1967). The stability and modifiability of individual vocal characteristics in stress and nonstress interviews. Research Bulletin No. 2. New York: William Alanson White Institute. Cattell, R. B., Eber, H. W., and Tatsuoka, M. M. (1970). Handbookfor the Sixteen Personality Factor Questionnaire. Champaign, IL: Institute for Personality and Ability Testing. Digman, J. M., and Takemoto, N. K. (1978). Factors in the natural language of personality: Re-analysis and comparison of seven studies. Paper presented at the meeting of the Society of Multivariate Experimental Psychology, Cincinnati. Feldstein, S., and Welkowitz, J. (1978). A chronography of conversation: In defense of an objective approach. In A. W. Siegman and S. Feldstein (eds.), Nonverbal Behavior and Communication. Hillsdale, NJ: Lawrence Erlbaum Associates, pp. 329-378. Feldstein, S., Alberti, L., Ben Debba, M., and Welkowitz, J. (1974). Personality and simultaneous speech. Paper presented at the annual meeting of the American Psychological Association, New Orleans, August. Jatfe, J., and Feldstein, S. (1970). Rhythms of dialogue. New York: Academic Press. Kramer, E. (1963). Judgment of personal characteristics and emotions from nonverbal properties of speech. Psychol. Bull. 60: 408-420. Marcus, E. S., Welkowitz, J., Feldstein, S., and Jaffe, J. (1970). Psychological differentiation and the congruence of temporal speech patterns. Paper presented at the meeting of the Eastern Psychological Association, Atlantic City, April. Markel, N. N. (1965). The reliability of coding paralanguage: Pitch, loudness, and tempo. J. Verb. Learn. Verb. Behav. 4: 406-408. Martz, J., and Welkowitz, J. (1977). wmhraa-Computer programs to analyze dialogic’pattems. Percept. Mot. Skills 45: 531-537. Norman, W. T. (1963). Toward an adequate taxonomy of personality attributes; replicated factor structure in peer nomination personality ratings. J Abnorm. Sot. Psychol. 66: 574-583. Ramsey, R. W. (1966). Personality and speech. J Pers. Sot. Psychol. 4: 116-118. Ramsey, R. W. (1968). Speech patterns and personality. Language and Speech 11: 56-63. Sadd, S., Welkowitz, J., and Feldstein, S. (1978). Judgments of characteristics of speakers in a natural stress situation. Percept. Mot. Skills 47: 47-54. Scherer, K. R. (1972). Judging personality from voice: A cross-cultural approach to an old issue in interpersonal perception. J Pers. 40: 191-210. Siegman, A. W. (1978). The telltale voice: Nonverbal messages of verbal communication. In A. W. Siegman and S. Feldstein (eds.), Nonverbal behavior and communication. Hillsdale, NJ: Lawrence Erlbaum Associaties, pp. 329-378. Tupes, E. C., and Christal, R. E. (1958). Stability of personality trait rating factors obtained under diverse conditions. USAF WADC Technical Note 58-61. Welkowitz, J., and Feldstein, S. (1969). Dyadic interaction and induced differences in perceived similarity. Proc. 77th Ann. Conv. Am Psychol. 4: 343-344. Welkowitz, J., and Feldstein, S. (1970). Relation of experimentally manipulated interpersonal perception and psychological differentiation to the temporal patterning of conversation. Proc. 78th Ann. Conv. 4m. Psych&. Assoc. 5: 387-388. Welkowitz, J., and Feldstein, S. (1973). Rhythms of dialogue and therapy: Experience questionnaire. Paper presented at the annual meeting of the Society for Psychotherapy Research, Philadelphia, June. Welkowitz, J., Feldstein, S., Finkelstein, M., and Aylesworth, L. (1972). Changes in vocal intensity as a function of interspeaker influence. Percept. Mot. Skills 35: 715-718. Welkowitz, J., and Kuc, M. (1973). Interrelationships among warmth, genuineness, empathy, and temporal speech patterns in interpersonal interaction. J. Consult. Clin. Psychol. 41: 472-473.