Verbal behavior of interviewees: The effects of several situational variables on verbal productivity, disfluency, and lexical diversity

JOURNAL

OF COMMUNICATION

DISORDERS

9 (1!976), 211-225

VERBAL BEHAVIOR OF INTERVIEWEES: THE EFFECTS OF SEVERAL SITUATIONAL VARIABLES ON VERBAL PRODUCTIVITY, DISFLUENCY, AND LEXICAL DIVERSITY JAMBS Communication

J. BRADAC

Research Division, Department ofSpeech and Dramatic Art, The University oflowa, Iowa City, Iowa 52240

CATHERINE Information

Sciences Department,

NORMAN Communication

W. KONSKY

Illinois State University,

Normal,

Illinois 61761

D. ELLIOTT

Research Division, Department ofSpeech and Dramatic Art, The University of Iowa, Iowa City, Iowa 52240

Interviewer status (high vs. low) and anticipated evaluation (positive vs. negative) were manipulated orthogonally to test competing predictions derived from HulLSpence and Mahl-Osgood drive theory hypotheses. The effects of question specificity, interview topic, and interview segment were examined concomitantly. Verbal behavior measures were total words per interviewee response, mean segmental TTR (25), partial mean segmental TTR (25), and ratio of disfluency per response. Results for disfluency tended to support the Mahl-Osgood hypothesis. Lexical diversity results offered weak support for the Hull-Spence hypothesis. Results for verbal productivity failed to support either of the hypotheses and were interpreted as possibly supporting tbe Duffy-Malmo inverted U hypothesis. Generally, the results suggested that the unitary construct “drive” may be inadequate for explaining the effects of specific situational variables upon verbal behavior.

Introduction The interview is a pervasive communication context in our society. Persons are interviewed by potential employers, opinion researchers, counselors, therapists, doctors, and linguists, among others. Inferences about an interviewee’s personality, health, competence, or emotional state are often made on the basis of verbal performance, and in turn critical decisions may be made on the basis of these inferences. More specifically, an interviewer’s judgments may be affected by quantity of speech (Bass, 1954; Bavelas, Hastorf, Gross, and Kite, 1965; Jaffee and Lucas, 1969) language style (Lambert, Anisfeld, and Yeni-Komshian, 1965; Triandis, Loh, and Levin, 1966; Hopper and Williams, 1973), and disfluency level (Miller and Hewgill, 1964; Sereno and Hawkins, 1967; McCroskey and Mehrley, 1969), as well as by the manifest content of utterances. These aspects of verbal behavior and others have been studied widely in recent years. Perhaps in response to a proclaimed neglect (Goffman, 1964)) studies have o American

Elsevier Publishing

Company,

Inc.,

1976

211

212

JAMES

J. BRADAC

et al.

focused increasingly on situational determinants of performance, especially on interviewer differences (see, for example, Pope and Siegman, 1965; Siegman and Pope, 1965; Heller, Davis, and Myers, 1966; Matarazzo and Wiens, 1967; Jourard and Jaffee, 1970; Davis and Skinner, 1974). Researchers and theorists have begun to examine in detail the general hypothesis that an interviewee’s speech may be affected by situational as well as personal factors. We believe the most impressive, programmatic research on situational determinants of interviewees’ speech to be that of Siegman, Pope, and associates (Siegman and Pope, 1972). These investigators have explored effects of arousalinducing situational variables on verbal behavior. Although not completely unambiguous, the results of some of their studies offer support for two of the several variants of drive theory. Our study grew out of an experiment by Pope and Siegman (1972) that probed the effects of two levels of interviewer status on several dimensions of interviewee verbal performance. We believed that some of their results would be qualified in important ways by another variable, namely, the type of evaluation anticipated by interviewees. Also we examined concomitantly the effects of three other situational variables: interview topic, interview segment, and specificity level of interviewer questions. We thought that a relatively complex design employing orthogonal manipulations of several situational variables might afford a better test of the competing predictions of the variants of drive theory than had been made previously. Furthermore, quite apart from theoretical considerations, these variables are ubiquitous in interviews, and we felt that information about their effects would be useful. Related Research and Rationale Power is fundamental in interpersonal relations (Russell, 1938, 10; Jacobson, 1972). In a given social system, high-status persons have the power to reward or punish others of lower status. Power of this type has been shown to affect both verbal and nonverbal behavior. For example, when confronted by high-status others may exhibit increased conformity to demands persons, lower-status (Critchlow, Herrup, and Dabbs, Jr., 1968), decreased ideational fluency (Collaros and Anderson, 1969), inhibited motor responses (Doob and Gross, 1968), and altered language styles (Cowan, Weber, Hoddinott, and Klein, 1967). In the study cited above, Pope and Siegman (1972) demonstrated a relationship between perceived interviewer status and interviewee verbal performance. Specifically, high-status interviewers elicited more speech from interviewees than did lowstatus interviewers. Also, interviewees were more disfluent in the low-status condition, although this was true only for one of two interview topics. Regardless of status relationships, in many contexts persons anticipate evaluative reactions from others as a result of prior learning, i.e., we have expectations

EFFECTS

OF SITUATIONAL

VARIABLES

ON VERBAL

BEHAVIOR

213

regarding the likelihood of positive or negative evaluation. Although many studies have explored the effects of actual evaluation (see, for example, Gibb, Smith, and Roberts, 1955; Cieutat, 1962; Stolz and Tannenbaum, 1963; Miller 1964; Vlandis, 1964; Bavelas, Hastorf, Gross, and Kite, 1965; Davis, 1967; Blubaugh, 1969; Gardiner, 1971; Heller, 1972), no research has been done on the effects of anticipated evaluation upon subsequent verbal performance. Labov (1970) offers some anecdotal evidence that suggests that anticipated negative evaluation may decrease verbal output and increase structural redundancy. More generally, Griffitt (1968) found that persons who anticipated positive evaluation from a stranger were more attracted to him than were persons who anticipated negative evaluation, and increased interpersonal attraction has been associated with increased communicative activity in dyads (Homans, 1950; Newcomb, 1956). Both interviewer status and anticipated evaluation can be related to the construct “drive. ” High-status interviewers are likely to produce higher levels of drive than are low-status interviewers as a function of their greater evaluation potential (Sasfy and Okun, 1974). Interviewees anticipating negative evaluation are likely to be more anxious than are interviewees anticipating positive evaluation (cf. Schachter, 1959). l Drive should be highest when interviewees anticipate negative evaluation from high-status interviewers and lowest when positive evaluation is anticipated from low-status interviewers. Unfortunately, contradictory findings abound in the literature on drive and verbal behavior (Murray, 1971). Some studies show that high levels of drive are associated with low verbal productivity, low lexical diversity, and high disfluency (Fairbanks, 1944; Mahl, 1956; Mednick, 1958; Levin, Baldwin, Gallwey, and Paivio, 1960; Gottschalk, 1961; Feldstein, 1962; Gardner and Sugerman, 1964; Kasl and Mahl, 1965; Pflager, 1969; Howeler, 1972). Thesefindingslendsupport to the Mahl-Osgood hypothesis which indicates that anxiety is cognitively disruptive (Mahl, 1956; Kasl and Mahl, 1965; Maclay and Osgood, 1959; Osgood, 1960). Specifically, Mahl and Osgood suggest that cognitive disruption will result in stereotyped, confused, and inhibited performance, the verbal manifestations of which are low lexical diversity, high disfluency, and low verbal productivity, respectively. On the other hand, some studies show that high levels of drive are associated with high diversity, low disfluency, and high productivity (Davids and Ericksen, 1955; Krause, l%l; Siegman and Pope, 1965; Eisenman, 1966; Bemick and Oberlander, 1968; Pope and Siegman, 1968; Siegman and Pope, 1972). If we assume that in relatively formal communication contexts persons have learned that verbal precision, fluent speech, and copious output are demanded, then the results of the latter studies support the Hull-Spence drive theory hypothesis, which

‘Following precedent, we will use the terms “drive,” stylistic purposes.

“anxiety,”

and “arousal”

synonymously

for

214

JAMES

J. BRADAC

et al.

indicates that dominant or well learned response tendencies are energized by high arousal (Hull, 1943; Spence 1956; Lajonc, 1965). In terms of the major variables of our study, interviewer status and anticipated evaluation, and in light of the competing drive theory predictions discussed above, six hypotheses can be offered. The first three (Hi-H2) were derived from the Mahl-Osgood hypothesis, whereas the second three (HJ-Hs) stem from the HullSpence formulation. Interviewees responding to a high-status interviewer (HS) will be less Hi: productive verbally (a), will exhibit lower lexical diversity (b), and will exhibit more disfluency (c) than will interviewees responding to a low-status interviewer (LS). H2: Interviewees anticipating negative evaluation (- E) will be less productive verbally (a), will exhibit lower lexical diversity (b), and will exhibit more disfluency (c) than will interviewees anticipating positive evaluation (+ E). H3: If the relationship between interviewer status and anticipated evaluation is additive, then for verbal productivity (a) HS - E < HS + E < LS - E < LS + E or HS-E < LS-E < HS+E < LS+E, for disfluency (b) HS-E > HS+E > LS-E> LS+EorHS-E> LS-E> HS+E> LS+E,andforlexicaldiversity (c)HS-E HS + E > LS- E > LS+ E or HS-E > LS-E > HS+E > LS+E, for disfluency (b) HS-E < HS+E < LS-E< LS+EorHS-EC LS-E< HS+E< LS+E,andforlexicaldiversity (c)HSsE> HS+E> LS-E> LS+EorHS-E> LS-E> HS+E> LS+E. All of the hypotheses, most notably H2 and Hs, assume that, as drive levels increase, behavior is affected monotonically, i.e., drive is conceptualized as a unidimensional construct increasing or decreasing as a function of external stimulation and producing corresponding increases or decreases in verbal productivity, etc. The assumption of unidimensionality and monotonicity underlies both the Mahl-Osgood and Hull-Spence hypotheses. We thought that our status and anticipated evaluation inductions would produce four levels of drive and hence would afford a good test of the monotonic hypotheses.2 Also, we used interview topics that varied in their arousal potential

?Self-reported anxiety ratings obtained from a sample of the target population support this claim. It should be noted that we conceived of our research as a study of “behavioral” as opposed to “subjective” drive, however (Lacey, 1967).

EFFECTS

OF SITUATIONAL

VARIABLES

ON VERBAL

BEHAVIOR

215

and interview questions that exhibited four levels of specificity, a variable that has been related to drive (Siegman and Pope, 1972). Furthermore, each interview comprised four segments that we hoped would allow us to examine the verbal effects of fluctuations in arousal across conditions. We did not form hypotheses regarding the topic, question specificity, and interview segment variables; rather, we asked the general research question: Do these other situational variables qualify the effects of interviewer status and anticipated evaluation in ways that have implications for the monotonic hypotheses?

Method Subjects Ss were 64 undergraduate students at The University of Iowa who were randomly assigned to one of four conditions: HS + E, HS- E, LS + E, and LS - E.

Interviewers Interviewers were graduate and undergraduate students at The University of Iowa. Each interviewer worked an equal number of times in each condition in order to prevent confounding idiosyncratic variables (response tendencies, appearance, etc.) with treatments.

Topics On the basis of pilot study results, four interview topics were used: the value of grading (Grades), sexual behavior on campus (Sex), the need for insurance (Insurance), and problems of home ownership (Home). Students in the pilot study indicated on opinion scales that the first two topics would produce anxiety and that the latter two would not. Topic order was counterbalanced within conditions which made it possible to examine separately the effects of interview segment and topic.

Status Induction Cues associated with status differences were ascertained in a second pilot study. On the basis of pilot study results and previous research on status differences, three attributes were manipulated to produce perceptions of high or low interviewer status: position in the academic hierarchy (teacher vs. student), dress (suit or dress vs. casual wear), and seating position (interviewer behind desk vs. interviewer in front of desk).3 The interviewer’s academic position was established orally by a confederate prior to the interview. 3The distance between interviewers and interviewees

was constant across status conditions.

JAMES

216

Anticipated

Evaluation

J. BRADAC

et al.

Induction

Anticipated positive or negative evaluation was induced by a written statement presented toss after the oral status induction. This statement ostensibly contained information about the purpose of the study. In the negative condition, Ss were told that they were participating in the first phase of an experiment that would be completed the following semester; that the second phase would require highly critical, intelligent, and demanding interviewers; and that the researchers were interested in retaining those interviewers whom Ss perceived to be extremely critical. Ss were then informed that the person who would soon interview them had attained a very high score on an IQ test and had demonstrated a clearly critical nature on Johnson’s Test of Personal Tendencies, a bogus instrument. After reading the induction, Ss responded to two questions designed to strengthen the negative response set, an example being: “How do you feel about a person whom you perceive to be cold?-(a) tense, (b) noncommunicative, (c) worried, or (d) hostile.” In the positive condition, the induction was the same except the interviewer was described as a tolerant, flexible person of moderate intelligence, andSs responded to two “positive” questions. Question Speczjicity Four types of questions were asked about each of the four topics. The first type we called “yes-no” (Are you aware of the sexual revolution?), the second “labeling” (What percentage of college students engage in premarital sex in your opinion?), the third “explanatory” (Assuming that sexual practices and attitudes are more open now, why do you think this is so?), and the fourth “open-ended” (What are your feelings about premarital sex?). The first two question types are relatively focused, the second two relatively unfocused (cf. Williams and Naremore, 1969). In all cases questions were presented in the order indicated above. Measures All interviews were audiotaped. Transcripts were obtained from the tapes and these were subsequently punched word for word onto computer cards. FREQDIC, a lexical statistics program, was used to obtain raw scores for lexical diversity (Swigger and Wachal, 1973). We used total and partial mean segmental type-token ratios as measures of diversity.4 A 25-word sampling unit was chosen, for this allowed us to include data in the effects analyses that would have been ‘The mean segmental type-token ratio (MSTTR) = the average of N different words/N words forK samples. ‘IIe partial mean segmental type-token ratio adjusts for the deflation of the MSTTR produced by repetitions (adjacent identical words) and hence is a somewhat purer measure of vocabulary range.

EFFECTS

OF SITUATIONAL

VARIABLES

ON VERBAL

BEHAVIOR

217

excluded with a larger unit. Also, 25-word samples have proved sensitive in previous studies, and Jaffee (1956) has demonstrated the comparability of 25-, SO-, 75-, and loo-word samples. Verbal productivity was a second measure. This is simply the total number of words spoken by a subject in response to each interview question. A disfluency ratio was the third measure. “Non-ah” disfluencies were scored according to Mahl’s system (1956). A ratio of number of disfluencies to number of words was obtained for each response. Procedure Ss reported to a central waiting room and were taken to the interview room by a confederate. The confederate asked five standardized questions of the subject, one of each of the first three types and two of the fourth type. (What is your major? Why did you come to The University of Iowa? etc.). These questions established a baseline of verbalization for each subject before administration of the inductions. Data from this preinduction interview provided covariates corresponding to the dependent measures. After Ss responded to the introductory questions, the status and anticipated evaluation variables were induced. Interviewers were then introduced and they proceeded to ask their questions. Interviewers were instructed to use head nodding, “urn hum,” and other responses that were neither overtly positive nor negative. They were also instructed to ask only the questions provided them and not to probe for information or comment upon responses. Design ana’ halysis Five independent variables are described above. Topic and Interview Segment were analyzed separately, since Ss did not respond to all topics during each segment. Thus, two 2 X 2 X 4 X 4 factorial analyses of covariance were run for each of the dependent measures using the BMDP2V “Repeated Measures Analysis of Variance with Covariates” program (U.C.L.A., 1973). Covariance analysis is appropriate for many verbal behavior variables, especially productivity, given the large differences between individuals in normal response tendencies. (We will come back to this point below.) The 0.05 level of significance was set for all comparisons.

Correlations

Among

Measures

Partial and total MSTTR scores were highly correlated; for this reason we chose to use the partial MSTTR measure for the lexical diversity analyses, this being a somewhat purer measure of vocabulary range. Disfluency ratios were negatively

218

JAMES J. BRADAC et al.

correlated with total MSTTR’s. None of the other correlations were significant, hence independent analyses for each measure were deemed acceptable (see Table 1). TABLE 1 Correlations Among Dependent Measures

Verbal Prod. (1) Disfluency (2) Partial MSTTR (3) MSTTR (4)

1

2

3

4

X

+0.26 X

f0.11 -0.11 X

+o. 10 -0.53 f0.96 X

Lexical Diversity

The analyses included scores of 48 Ss for the third and fourth question types only. The remaining 16Ss and the first and second question types in some cases did not yield 25-word responses. Cell N’s ranged from 11 to 14. A main effect was found for anticipated evaluation (F = 4.26; df = 1,43). Ss in the -E conditions had a higher mean partial MSTTR than did Ss in the + E conditions (x’s = 0.85 1 and 0.830, respectively). 5 This was the only significant difference for the lexical diversity measure. Disjluenq

A topic x anticipated evaluation interaction was obtained (F = 4.01; df = 3, 180). Simple effects tests suggested that this interaction was produced by - ESs exhibiting more disfluency than +E Ss for the Home topic (x’s = 0.031 and 0.019, respectively). A main effect was found for status (F = 2.92, df = 1, 59;~ > 0.05 < 0.08). High-status interviewers produced more disfluency than did low-status interviewers (x’s = 0.045 and 0.029, respectively). Interview topic produced a main effect (F = 4.85; df = 3, 180). Duncan’s Multiple Range Test indicated that Grades produced more disfluency o( = 0.040) than did Insurance ( 0.33), Sex (0.033), or Home (0.025). The latter three did not differ. A main effect was obtained for question specificity (F = 21.01; df = 3, 180). Duncan’s Test showed that the explanatory questions produced more disfluency o( = 0.048) than did the yes-no (0.016), labeling (0.030), or open-ended questions (0.037). Open-ended questions produced more disfluency than did yes-no questions, as did labeling questions.

5All of the mean scores reported for each of the three measores are covariance adjusted.

EFFECTS

OF SITUATIONAL

VARIABLES

ON VERBAL

BEHAVIOR

219

Verbal Productivity A four-factor interaction was obtained for the interview segment analysis: segment X status X anticipated evah,tation X question specificity (F = 2.14; df = 9,540). This complex interaction can be interpreted most usefully in the context of a significant three-factor interaction described next. A segment X status x anticipated evaluation interaction was obtained (F = 3.17; df = 3, 180). A test for simple effects showed that HS+E Ss were more productive o( = 35.75) than were LS- E Ss (25.84) during the first interview segment. During the second segment, HS+ E Ss were more productive (43.00) thanwereHS-ESs(28.94), LS+ESs(24.30),andLS-ESs(31.06). Duringthe third segment, HS+ E Ss were more productive (40.48) than were LS+ E SS (26.97). During the fourth segment, HS-ESs were more productive (45.80) than were LS+E Ss (30.60) and LS-E Ss (23.41); HS+E Ss (39.26) were more productive than were LS- E Ss . (see Fig. 1). Graphing of the means and simple effects analyses suggested that these three-factor interactive effects are bound to the third and fourth question types primarily; the yes-no and labeling questions produced relatively little between-groups variability. Therefore, during the first, second, and third interview segments, HS+ ESs produced the most verbal output. During the fourth segment HS Ss in both anticipated evaluation conditions were more productive than were LS Ss.

Fig. 1.

Interaction of interviewer productivity

status, anticipated

eqluation,

and interview segment for verbal

220

JAMES

J. BRADAC

et al.

Status interacted with question specificity (F = 3.21; df = 3, 180). Simple effects tests indicated that the explanatory and open-ended questions were implicated primariIy. For these question types HS Ss produced more verbal output k’s =55.55and50.52,respectively)thandidLSSs(37.58and38.82). HSSsandLS Ss did not differ in their responses to the yes-no and labeling questions. A significant main effect was obtained for status (F = 6.28; df = 1, 59), although this effect is qualified strongly by the interactions described above. HS Ss were more productive than were LS Ss k’s = 37.08 and 29.97, respectively). Question specificity produced a significant main effect (F = 78.04; df = 3, 180). Duncan’s Test showed that yes-no questions produced less verbal output o( = 8.52) than did labeling questions (30.37), explanatory questions (46.57), and open-ended questions (44.67). Types three and four did not differ from each other; both differed from one and two. The topical analysis produced a significant interaction between topic and question specificity (F = 10.20; df = 9,540). Verbal output for topics did not differ for yes-no questions. For labeling questions Grades produced the most verbalization o( = 55.28), Sex the least (11.88). Inskance (30.33) and Home (23.92) fell between these extremes. For explanatory questions Grades (50.67) and Sex (51.03) produced more verbal output than did Insurance (44.06) and Homes (40.36). For open-ended questions Grades (62.99) produced more verbalization than did Sex (32.06), Insurance (41.1 I), and Homes (41.53)Thelatter two differed from Sex. A main effect for topic was obtained (F = 22.43; df = 3, 180). Grades o( = 44.8 1) produced more speech than did Insurance (32.52), Home (27.56), and Sex (24.00). Insurance produced more speech than did Sex. Discussion It would be misleading to suggest that the results offer support for any of the research hypotheses with the exceptions of Hi(b) and Hs(b). The disfluency results are consistent with the general Mahl-Osgood hypothesis, however. Decreased question specificity, an arousing topic, and high-status interviewers Hi(b) produced relatively high levels of interviewee disfluency; anticipated negative evaluation produced more disfluency than did anticipated positive evaluation, although this effect was bound to one topic only. The lexical diversity results offer some support, albeit weak, for the Hull-Spence hypothesis and for Hs(b). High-status interviewers elicited more speech from interviewees than did lowstatus interviewers, a result consistent with that obtained by Pope and Siegman (1972). These researchers explained this result in terms of the Hull-Spence hypothesis. However, our interactive effects indicate that the verbally facilitating effects of high status are bound to low-specificity questions and to particular combinations of anticipated evaluation and interview segment. These interactive

EFFECTS OF SITUATIONAL

VARIABLES

ON VERBAL BEHAVIOR

221

effects suggest that the assumption of additivity, which underlies both H3 and He, is incorrect and, more generally, that increases in drive do not affect verbal productivity monotonically. During the first, second, and third interview segmerits, the verbal productivity of HS + E Ss was highest; the output of Ss in the other conditions was rather uniformly depressed. During the fourth segment, a dramatic increase in output occurred in the HS- E condition. This result seems anomalous from the standpoints of both the HullSpence and Mahl-Osgood hypotheses. If in fact the HS-E condition produced the highest level of arousal, why wereSs in this condition not maximally or minimally productive throughout? Or, more particularly, assuming that drive decreased across interview segments (Pope, Siegman, Blass, and Cheek, 1972), why was not the relationship among HS-E and the other conditions at least a stable function of that decrease? The Duffy-Malmo inverted U hypothesis may provide an answer (Duffy, 1962; Malmo, 1966). This hypothesis indicates that both high and low arousal are inhibiting, whereas moderate arousal is releasing. This would seem to square with the results for the first three segments, if we assume that low-status interviewers produced low arousal regardless of the type of evaluation that interviewees anticipated from them, i.e., anticipated evaluation became salient only when the interviewer possessed high evaluation potential. It is likely that the high arousal of Ss in the HS-E condition decreased to a more moderate level by the fourth segment, for no actual negative evaluation was offered by interviewers. Possibly the interviewers’ neutral responses came to seem positive when contrasted with interviewees’ negative expectations. So perhaps the decreased arousal of HS - E Ss resulted in increased verbal output. Thus, the change in HS - ESs behavior may support the Duffy-Malmo hypothesis also. This interpretation is, of course, speculative. A more direct test of the U hypothesis can be made by systematically increasing drive in one group of interviewees (low to moderate to high) and decreasing it in a second group, as opposed to allowing drive to fluctuate uncontrolled across interview segments. To summarize, the disfluency results tend to support the Mahl-Osgood hypothesis, the lexical diversity results offer weak support for the Hull-Spence hypothesis, and the verbal productivity results fail to support either of the hypotheses. Taken together, the results suggest that increased drive serves to improve or to deteriorate verbal performance, depending on which aspect of performance is being scrutinized. Ideally, there would have been a perfect correspondence among outcomes for the three measures, and there was not. The findings regarding lexical diversity are especially deviant. To the extent that the lack of correspondence is a result of measurement error, it is uninteresting except from the standpoint of research technology. There is another possibility that is more interesting, however. Theorists and researchers have begun to question the unidimensionality of the general construct “drive” (Bradac and Elliott, 197.5). Lacey (1967), for example,

JAMES

222

3. BRADAC

et al.

has postulated three discrete types of drive: behavioral, electrocortical, and autonomic. Pflager (1969) provides evidence that ratio measures of lexical diversity are sensitive to trait anxiety and that distributional measures are sensitive to situational anxiety. These observations indicate that qualitatively different drive states have correspondingly different behavioral manifestations. Our results suggest that different drive sources (status and anticipated evaluation, for example) may similarly produce correspondingly different verbal effects, that certain aspects of verbal behavior may be more sensitive to specific situational differences than are others. We are currently investigating this hypothesis directly. The possibility that drive levels and situational cues interact in complex ways has been noted previously, although not with reference to verbal behavior (Berkowitz, 1968). We will conclude with two methodological recommendations. In spite of our suggestion regarding a more direct test of the U hypothesis, it should be noted that the interview segment analysis was very revealing in our study; it showed that the relationship between interviewer status and anticipated evaluation changed over the course of the interview. Segmental or phasic analyses have yielded unusually interesting results in research on small group communication, and we encourage their increased use in interview research (Bales and Strodtbeck, 1951; Scheidel and Crowell, 1964; Fisher, 1970; Bell, 1974; Marr, 1974). We also encourage strongly the use of covariance analysis, especially in research on verbal productivity. Even though we randomly assignedSs to conditions, analyses of our preinduction interviews demonstrated significant differences between conditions for the productivity variable @ < 0.001). Many of the studies on productivity have failed to obtain a base-line measure; the great variability in individual verbal response tendencies and the accidental confounding with treatments which this may produce suggest that this failure may be responsible for some of the conflicting findings that abound in the literature.

References Bales, R. F., Stodtbeck,

F. L. Phases in group problem-solving.

J. Abnorm. Sm. Psychol.,

195 1,46,

485.495.

Bass, B. M. The leaderless group discussion. Psychol. Bull., 1954, 51, 470-495. Bavelas, A., Hastorf, A. H., Gross, A. E., Kite, W. R. Experiments on the alteration of group structure. J. Exp. Sot. Psychol., 1965, S-70. Bell, M. A. The effects of substantive and affective conflict in problem-solving groups. Speech Monogrs. 1974, 19-23. Berkowitz, L. The motivational status of cognitive consistency theorizing. In R. P. Abelson et al. (Eds.), Theories of cognitive comistency. a source book. Chicago: Rand McNally, 1968, pp. 303-310. Bemick, N., Oberlander, M. Effect of verbalization and two different modes of experiencing on pupil size. Percept. Psychophys., 1968, 3, 327-330.

EFFECTS

OF SITUATIONAL

VARIABLES

ON VERBAL

BEHAVIOR

223

Blubaugh, J. A. Effects of positive and negative audience feedback on selected variables of speech behavior. Speech Monogr, 1969, 36, 131-137. Bradac, J. J., Elliott, N. D. Verbal behavior and drive: a theory in crisis. Paper presented at the meeting of the Midwest Modem Language Association, Chicago, 1975. Cieutat, V. J. Sex differences and reinforcement in the conditioning and extinction of conversational behavior. Psychol. Rep., 1962, 18, 691694. Collaros, P. A., Anderson, L. R. Effect of perceived expertness upon creativity of members of brainstorming groups. J Appl. Psychol., 1969, 53, 159163. Cowan, P. A., Weber, J., Hoddinott, B, A., Klein, J. Mean length of spoken response as a function of stimulus, experimenter, and subject. Child Dev., 1967, 38, 191-203. Critchlow, K. F., Herrup, R., Dabbs, Jr., J. M. Experimenter influence in a conformity situation. Psychol. Rep., 1968, 23,40%410. Davids, A. H., Erickson, C. W. The relationship of manifest anxiety to association productivity and intellectual attainment. J Consult. Psychol., 1955, 19, 219-222. Davis, J. D., Skinner, A. E. G. Reciprocity of self-disclosure in interviews: modeling or social exchange?. .I. Personal. Sot. Psychol., 1974, 29, 77%784. Davis, J. W. Variations in verbal behavior in dyads as a function of varied reinforcing conditions. Speech

Monogr.,

1967,

Doob, A. N., Gross, Psychol.,

34, 443-447.

A. E. Status of frustrator

1968,

as an inhibitor of horn-honking

Duffy, E. Activation and behavior. Wiley: New York, 1962. Eisenman, R. Birth order, anxiety, and verbalizations in group psychotherapy. 1966,30,

Fairbanks,

56,

J. Sot.

J. Consult. Psychol.,

521-526.

H. The quantitative

1944,

responses.

76, 213-218.

differentiation

of samples of spoken language.

Psychol.

Monogr.,

lF38.

Feldstein, S. The relationship of interpersonal involvement and affectiveness of content to the verbal communication of schizophrenic patients. J. Abnorm. Sot. Psychol., 1962, 64, 39-45. Fisher, B. A. Decision emergence: phases in group decision-making. Speech Monogr., 1970, 37, 5>66.

Gardiner, J. C. A synthesis of experimental studies of speech communication feedback. J. Commun., 1971, 21, 17-35. Gardner, R. C., Sugarman, E. D. Effects of anxiety and word fluency on verbal diversification. Unpublished manuscript cited in M. P. Pflager, The eflects ofanxiety on verbalproductivity and vocabulary diversity. M.A. thesis, University of Victoria, 1969. Gibb, J. R., Smith, E. E., Roberts, A. H. Effects of positive and negative feedback upon defensive behavior in small problem-solving groups. Am Psychol., 1955, 10, 335. Goffman, E. The neglected situation. In J. J. Gumperz and D. Hymes (Eds.), The ethnography of communication.

Am. Anfhro.,

1964,

66,

133-136.

L. A. (Ed). Comparative psycholinguistic analysis of two psychotherapeutic interviews. New York International Universities Press, 1961. Griffitt, W. B. Anticipated reinforcement and attraction. Psychonom. Psychol., 1968, 11 355. Heller, K. Interview structure and interviewer style in initial interviews. In A. W. Siegman and B. Pope (Eds.), Studies in dyadic communication. New York: Pergamon Press, 1972, pp. 9-28. Heher, K., Davis, J. D., Myers, R. A. The effects of interviewerstyle inastandardized interview../.

Gottschalk,

Consult.

Psychol.,

1966,

30, 501-508.

Homans, G. C. The human group. New York: Harcourt, Brace, & World, 1950. Hopper, R. W., Williams, F. Speech characteristics and employability. Speech Monogr.,

1973,40,

296302.

Howeler,

M. Diversity of word usage as a stress indicator in an interview situation. J. Psychol. Res.. 1, 243-248.

1972,

224

JAMES

J. BRADAC

et al.

Hull, C. L. Principles of behavior. New York: Appleton-Century, 1943. Jacobson, W. D. Power and interpersonal relations. Belmont, Calif.: Wadsworth, 1972. Jaffee, C. L., Lucas, R. L. Effects of rates of talking and correctness of decisions on leader choice in small groups. J. Sot. Psychol., 1969, 79, 247-254. Jaffee, J. The comparability of small segments of different lengths in the calculation of the Type-Token Ratio. Unpublished manuscript cited in A. W. Siegman and B. Pope (Eds.), Studies in dyadic communication. New York: Pergamon Press, 1972. Jourard, S. M., Jaffee, P. E. Influence of an interviewer’s disclosureon the self-disclosing behavior of interviewees. .I. Consult. Psychol., 1970, 17, 252-257. Kasl, S. V., Mahl, G. F. The relationship of disturbances and hesitations in spontaneous speech to anxiety. J. Personal Sot. Psychol.. 1965, 1,425-433. Krause, M. S. Anxiety in verbal behavior: an intercotrelational study.J. Consult. Psychol., 1961,25, 272. Labov, W. The logic of nonstandard English. In F. Williams (Ed.), L.anguageandpover@. Chicago: Markham, 1970, pp. 153-189. Lacey, J. L. Somatic response patterning and stress: some revisions of activation theory. In M. H. Appley and R. Trumbell (Eds.), Psychological stress. New York: Appleton-Century-Crofts, 1967, pp. 14-37. Lambert, W. E., Anisfeld, M., Yeni-Komshian, G. Evaluational reactions of Jewish and Arab adolescents to dialect and language variations. J. Personal. Soc..Psychol., 1965, 2, 84-90. Levin, H. Baldwin, A. L., Gallwey, M., Paivio, A. Audience stress, personality, and speech. J. Abnorm. Sot. Psychol., 1960, 61, 469-473. Maclay, H., Osgood, C. E. Hesitation phenomena in spontaneous English speech. Word, 1959,15, 1944. Mahl, G. F. Disturbances and silences in patient’s speech in psychotherapy. J. Abnorm. Sot. Psychol.. 1956, 53, l-15. MaImo, R. B. Studies of anxiety: some clinical origins of the activation concept. In C. D. Spielberger (Ed.), Anxiety and behavior. New York: Academic Press, 1966, pp. 151178. Marr. T. J. Conciliation and verbal responses as functions of orientation and threat in group interaction. Speech Monogr., 1974, 41, 6-18. Matarazzo, J. D., Wiens, A. N. Interviewer influence on duration of interviewee silence. J. Exp. Res. Personal, 1967, 2, 56-69. McCroskey, J. C., Mehrley, R. S. The effects of disorganization and nonfluency on attitude change and source credibility. Speech Monogr., 1969, 36, 1%21. Mednick, S. A. A learning theory approach to research on schizophrenia. Psychol. Bull., 1958,55, 316327. Miller, G. R. Variations in the verbal behavior of a second speaker as a function of varying audience responses. Speech Monogr., 1964,31, 109-115. Miller, G. R., Hewgill, M. The effect of variations in nonfluency on audience ratings of source credibility. Q. J. Speech, 1964, 50, 364%. Murray, D. C. Talk, silence, and anxiety. Psychol. Bull., 1971, 75, 244-260. Newcomb, T. M. The prediction of interpersonal attraction. Am Psychol., 1956, 11, 575-586. Osgood, C. E. Some effects of motivation on style of encoding. In T. A. Sebok (Ed.), Style in language. Cambridge: M.I.T. Press, 1960, pp. 293-306. Pflager, M. P. The effects of anxiety on verbal productivity and vocabulary diversity. M.A. thesis, University of Victoria, 1969. Pope, B., Siegrnan, A. W. Interviewer specificity and topical focus in relation to interviewee productivity. J. Verbal Learn. Verbal Behav., 1965,4, 188-192. Pope, B., Siegman, A. W. Interviewer warmth in relation to interviewee verbal behavior. J. Consult. Clin. Psychol., 1968, 22, 588-595.

EFFECTS

OF SITUATIONAL

VARIABLES

ON VERBAL

BEHAVIOR

225

Pope, B., Siegman, A. W. Relationship and verbal behavior in the initial interview. In A. W. Siegman and B. Pope (Eds.), Studies in dyadic communication. New York: Pergamon Press, 1972, pp. 6989. Pope, B., Siegman, A. W., Blass, T., Cheek, J. Some effects of discrepant role expectations on interviewee verbal behavior in the initial interview. J. Cottsult. Clin. Psychol., 1972, 39, 50 l-507. Russell, B. Power. London: George Allen & Unwyn, 1938. Sasfy, J., Okun, M. Form of evaluation and audience expertness as joint determinants of audience effects. .I. Exp. Sot. Psychol., 1974, 10,461-467. Schachter, S. The psychology of ajiliation. Stanford: Stanford University Press, 1959. Scheidel, T. M., Crowell, L. Idea development in small discussion groups. Q. J. Speech, 1964,50, 40-45. Sereno, K. K., Hawkins, G. J. The effects of variations in speaker’s nonfluency upon audience ratings of attitude toward the speech topic and speaker’s credibility. Speech Monogr., 1967,34,58-64. Siegman, A. W., Pope, B. Effects of question specificity and anxiety producing messages on verbal fluency in the initial interview. J. Personal Sot. Psychol.. 1965, 2, 522-530. Siegman, A. W., Pope, B. (Eds.). Studies in dyadic communication. New York: Pergamon Press, 1972. Spence, K. W. Behavior theory andconditioning. New Haven, Conn: Yale University Press, 1956. Stolz, W., Tannenbaum, P. H. Effects of feedback on oral encoding behavior. Lang. Speech, 1963,6, 2lg228. Swigger, K. B., Wachal, R. S. FREQDIC: a word frequency dicfionary and lexical statistics program. Iowa City, Iowa: The University of Iowa, 1973. Triandis, H. C., Loh, W. D., Levin, A. L. Race, status, quality of spoken English, and opinion about civil rights as determinants of interpersonal attitudes. J. Personal Sot. Psychol., 1966, 3, 468472. U.C.L.A. Health Sciences Computing Facility. BMDP2V repeated measures analysis of variance with covariates program. Los Angeles: University of California, 1973. Vlandis, J. Variation in the verbal behavior of a speaker as a function of varied reinforcing conditions. Speech Monogr., 1964,31, 116-119. Williams, F., Naremore, R. C. On the functional analysis of social class differences in modes of speech. Speech Monogr, 1969, 36, 77-102. hajonc, R. B. Social facilitation. Science, 1965, 149, 269-274.