|||||| i l \ llJ CUENT LANGUAGE PATTERNS DURING PSYCHOTHERAPY Thesis for the Degree of M. A. MICHlGAN STATE UNIVERSITY Joe! A. Darby 196:6 111111114 111111111111? [11111 3 1293 01093 0133 ABSTRACT CLIENT LANGUAGE PATTERNS DURING PSYCHOTHERAPY by Joel A. Darby Previous language style research indicated that a number of language and speech measures were capable of discriminating between verbal material produced by dis- turbed and non-disturbed individuals. A.review of this research generated a two-fold classification system within which the previous studies could be categorized. The types of measures used were classified as either diagnostic or dynamic. It was noted that content measures were gener- ally used in diagnostic studies and expressive measures generally employed in dynamic studies. The present research was concerned with changes in clients' verbal behavior from the beginning to the end of psychotherapy and, as such, utilized both content and expressive measures in a dynamic design. The content measures were the average number of syllables p25 word, the Eypg-token ratio, personal pponoun fregpency, gpalifi— cation.termhfregpency, allness term fre enc , and the verb-adjective gpotient; the expressive measures were the amount of repgtition and the filled pause fregpengz. Speech samples drawn from the first and last psychotherapy Joel he Darby interviews of 22 clients were analysed in terms of these measures. In addition, because little is known about the interrelation of the language measures used, a correlational analysis was carried out for both the first and last inter- view, and for the two interviews combined. The results were organized in terms of Osgood's (1954) levels of language habit organization. The three content measures reflecting the structural level of organi- zation exhibited no change from the first to the last inter- view. Two of the three content measures which reflected organization at the vocabulary level tended to change in the direction of increased vocabulary diversity and in- creased interpersonal familiarity but did not achieve sta- tistical significance. The measures at the expressive level of organization exhibited the greatest degree of change. Both expressive measures indicated a significant shift from the first to the last interview in the direction of increased fluency. Thus, for the group of clients as a whole, the greatest change occurred at the expressive level, less change was exhibited at the vocabulary level, and no change resulted at the structural level. In addition, one of the measures, allness terms, showed a marked difference in the direction of change between the successful and unsuccess- ful clients. The correlational analysis indicated a strong re- lation among average number of syllables pgr word, pypgr Joel A. Darby token ratio, pgrsonal pronoun fregpency, verb-adjective otient, and filled pause fregpengy. Various reasons for these relationships were suggested. REFERENCES Osgood, C. B. Psycholinguistics: A survey of theory and research problems. J. abnorm. soc. Psychol., 1954, "_T> Approved._£;2%zd4~ux;az7hééDClACJéZ Committee Ehairman Date 97 Gui-4 /9éé a d CLIENT LAIGUAGE PATTERHS DURING PSYCHOTHERAPY BY Joel A. Darby A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1965 dedicated to my wife, carolyn to my father, mother, and sister, diane and to robert n. hammer ACKNOWLEDGMENT The author wishes to express his sincere gratitude and appreciation for the patient guidance and assistance in the planning and execution of this research and the development of this manuscript to Professor Abram H. Barch, chairman of his committee. In addition he wishes to thank Professor John R. Hurley and Professor C. L. Winder who served on the author's Haster's conmittee and Dr. Hilliam Eueller of the Michigan State University Counseling Center for allowing the author use of the Counseling Center Tape Library. iii IITRODUCTIOI. Language Measures . . . . Experimental Design Purpose . IBTHOB. . . IABLE Source of Data. . . . . . Clients . Therapist Rating of the Outcome OF CONTENTS Hethod of Collecting Data . . . Coding of the Language Heasures RESULTS . . DISCUSSION. Language leasure Change . . . . Intercorrelational Analysis . . Amount of Repetition and Filled Frequency . SUMMARY . . BIBLIOGRAPHY. APPENDICES. O O O 0 iv 10 10 10 10 11 12 14 25 25 .30 33 34 37 40 LIST OF TABLES Page Changes in Language neasures from First to L‘stI‘thViC'seeeeeeeeeeeeee 15 Changes in Language leasures fron First to Luthtmi.w.eeeeeeeeeeeeee 16 comparison of Language Heasure Changes of Successful and Unsuccessful Clients . . . . 1? Correlation between First and.Last Interview Language measure Scores . . . . . . . . . . l9 Intercorrelations anong First Interview Language Measure Scores . . . . . . . . . . 20 Intercorrelations anong Last Interview Language Heasure Scores . . . . . . . . . . 21 Intercorrelations among Contined First and Last Interview Language leasure Scores. . . 22 Direction and lumber of Significant correla- tionseeeeeeeeeeeeeeeeeee 23 LIST OF APPENDICES APPEIDIX A. Manual for Scoring Language Measures . Average Number of Syllables per word Type-Token Ratio . . . . . . . Personal Prenoun Frequency . . Qualification Term Frequency . Allness Term Frequency . . . . Verqudjective Quotient. . . . Amount of Repetition . . . . . Filled Pause Frequency . . . . B. First Interview Language leasure C. Language Heasure Stability Coefficients within the First and.Last Interviews . vi Scores. Last Interview Language Measure Scores . Page 41 41 41 42 43 43 44 44 45 46 47 48 I. INTRODUCTION "new the patient talks is as indicative of his emotional state as what he says, often more so.” -Hah1, 1961 Psychotherapy has been, is today, and shows every indication of continuing to be, prinarily a verbal process. Therapists have in the nain focused their attention on what is said by the patient. In the above quotation, George Hahl (1961) swings the spotlight of attention to an aspect of the patient's speech which, while certainly not overlooked in therapy itself, has not been well explored. Recently, psycholinguistics has invaded the therapy inter- view in an attenpt to nore fully understand the “how” of speech in therapy (Pittman a Hynne, 1961; Gottschalk, 1961; Hchoun, 1957; Pittenger, Hochett, a Danehy, 1960). How a person expresses hinself in language has not been totally neglected in clinical psychology. lost of the language neasures used in the present study have a long and full history of use with anxious and disturbed individuals. Few have, however, been used in an analysis of the patient's speech in therapy. Langgage Measures The language neasures used in the present study are the average nunber of syllables pgr word, the‘gyggr 1 token ratio, pggsonal pronoun freggency, ggalification term fre enc , allness tern freggency, verb-adjective otient, amount of re tition, and filled pause freggency. The average nunber of syllables 225 word is a neas- are of vocabulary diversity. Although a relatively old measure, it has alnost no history in our present context. It was originally devised as a readability neasure by Flesch (1948) and has received sone attention in that fern (Gilinsky, 1948; Flesch, 1959). Zipf (1949) denonr strated that there was an inverse relationship between the length of words and their frequency. Since longer, rarer words typically have more syllables, it follows that the greater the average nunber of gyllables pgr word in a passage, the more rare are the uords used, and conse- quently the nore diversified the vocabulary. The Eypg-token ratio (TTR) was devised by fiendell Johnson (1944) and is also considered a measure of language diversification. It is the ratio of the number of differ- ent words to the total nunber of words in a language sample. Pronoun usage, too, has been of interest for the past twenty years or so. Studies have been conducted using frequency of total pronoun usage, frequency of personal pronoun usage, and frequency of first-person pronoun usage. Qualification terns are words or phrases which allow the individual to nodify the anount of connitment he makes in his state-ants or which directly express am- bivalence. Allness terms, on the other hand, are extrene or polarised words, phrases, or statenents; they are expres- sions which permit no exception. The oldest of the language approaches is the 13527 adjective gpotient (VAQ) which is defined as the quotient of the total nunber of verbs in a language sample divided by the total number of adjectives in the sanple. Boder (1940) conducted the first American study with the 1352: adjective gpotient when he compared various types of writ- ing such as scientific, legal, and fiction, different authors writing the same type of natarial, and writings of the same author over a span of sixty years. The amount of repetitive material in a language sample has become of interest in the past ten years and can be categorized separately free the previous neasures as being more concerned with the process, or nechanics, of vocalization. .The definition of the neasure varies widely across studies but generally includes such occur- rences as stutter, one word repetition, and phrase repeti- tion. The filled pause fregpengy is also a relatively recent developnent and like the amount of repptition neas- ure focuses prinarily on the mechanics of speech, being generally defined as vocalizations which intrude upon the ongoing flow of neaningful speech. Expprimental Design of Previous Research In viewing the above language neasures and the experinental designs in which they are employed, one sees two classification systens. The measures can be grouped into two separate categories and the experimental designs can also be grouped into two separate categories. Six of the measures are "content-analysis” measures and two are “expressive“ measures. Avera e nunber of s llables,‘pypg¢ token ratio, sonal ronouns, allness terms, gpalifica- tion terns, and the verb-adjective gpotient all quantify verbal content. The amount of repptition and filled pause fre enc , on the other hand, quantify aspects of the verbal- isation which are not strictly content. One of the experimental designs could properly be labeled “diagnostic," and the other "dynamic." Diagnostic studies compare the language patterns of two er nore groups which differ with regard to some relatively stable charac- teristics. In these studies the language neasures function as the dependent variables and the diagnostic categories (nornal vs. schizophrenic, successful vs. unsuccessful, etc.) are the independent variables. Dynamic studies, on the other hand, conpare the individual's language patterns under varied conditions or at different points in time. when the two classification systens are superin- posed, one finds that the content-analysis measures were used primarily in diagnostic studies while the expressive measures were used in dynamic studies. Diagnostic Studies. Fairbanks (1944) obtained 3006 word spoken language sanples from schizophrenic and high I.Q. college freshmen subjects. The speech of the college group displayed higher TTR's, contained fewer per- sonal pronouns, and had lower YAQ's than the schizophrenics' speech. nann (1944), in a parallel study, reported similar results with written language samples. The written pas- sages of the college freshnen displayed higher TTR's and lower VAQ's than did the schizophrenics' language. The college freshmen also used more personal pronouns. Lorenz and Cobb (1954) analyzed spontaneous spoken interview natarial obtained from normals, hysterics, eb- sessive-conpulsives, manics, and paranoid schizophrenics. The different groups of psychiatric patients used total pronouns and the word '1' to a greater extent than did the normals. Their speech also displayed lower TTR's and higher VAQ's than did the normal subjects' speech. Rainy (1948), analyzing his clients' speech in the final counseling interview, found that his unsuccessful clients used nore qualifying terns than did his successful clients. Conrad and Conrad (1956), in a sinilar study, reported that patients who exhibited greater progress and involvement in group psychotherapy used nore personal pro- nouns than did those patients who were less involved and who exhibited less progress. Wagner and Hilliams (1961) compared spoken language sanples obtained free groups differing in Achievement Imagery, as neasured by the Iowa Picture Interpretation Test, and groups differing in defensiveness, as measured on the HHPI. They found.no significant differences in usage of self-reference pronouns or in qualification tern usage. They did report a difference in the number of a11- ness terns between high and low defensiveness groups, the nore defensive subjects using significantly fewer allness terns. ' Balkan and Nassernan (1940) compared spoken TAT stories elicited from conversion hysteria, anxiety state, and obsessive compulsive patients. They reported differ- ences among the three groups in personal pronoun usage, qualification and allness term usage, and verb-adjective quotient. gypamic Studies. Eldred and Price (1958) analyzed 15 therapy sessions with one patient and found that pauses occurred in those passages which expressed suppressed anger or suppressed depression and in those passages in which anxiety was communicated. Dibner (1956) analyzed the first psychotherapeutic interview of 39 hospitalized patients. He found that Cue Set I, a neasure which included both repetition and pauses, was related to situational anxiety. Similarly, nahl (1956, 1961) reported that his lonPAh Ratio, which included repe- tition, and his Speech Disturbance measure, which included both filled pauses and repetition, discriminated 'sonething' between sessions and within sessions for two different patients, the "something" presunably being anxiety. The repetition measure used by Krause and Pilisuk (1961) was unable to discriminate material generated under stressful conditions from that resulting fron non-stressful conditions. Their filled pause neasure, however, did dis- criminate successfully between naterial generated under the two conditions. A study reported by Ssgood and Talker (1959), which was the genesis of the present study, does not readily fit our classification scheme. They compared letters written by their friends and suicide notes written by persons just prior to committing suicide. Their measures were content- analysis measures and their design was essentially diag- nostic (they compared different groups), but their inde- pendent variable was presumably transitory stress rather than relatively permanent diagnostic differences. They reported that letters written under high stress (just prior to committing suicide) contained fewer syllables per word and nore allness and qualification terms than did letters written under normal conditions. The suicide notes also displayed lower TTR's and higher nounsverb/adjective-adverb quotients than the ordinary letters. Purppse The psychotherapeutic interview has only recently been accepted as a valid area for psychological research. As yet, little is really known about just what it is that happens in psychotherapy. Present research endeavors center heavily on the development of objective neasures which can be used for the examination of the therapy process. This constituted one of the objectives of the present research. There are, of course, nany researchable aspects of psychotherapy. The juncture of two of these aspects is the focus for the present study. First, psychotherapy is primarily a verbal phenomenon. Second, it is a process phenomenon, as opposed to a static one. One avenue of re- search into the process of therapy thus becomes the analy- sis of changes in verbal behavior. One of the purposes of the present study was to deternine whether or not the clients' verbal behavior, as reflected by the above language measures, changes over the course of psychotherapy. Inasmuch as one of the objectives of psychotherapy is the reduction of stress within the in- dividual, we would expect the difference in stress present at the first and final interview of therapy to be particu- larly reflected by the expressive neasures which have been shown to be sensitive to changes in stress. .If the content- analysis neasures are to be useful tools in the investiga- tion of psychotherapy, we should expect changes to occur with them also. Although no directional predictions are nade, the changes are expected to occur in directions con- gruent with changes from stressful to non-stressful condi- tions in the case of the expressive neasures, and from psychiatric diagnostic categories to normal in the case of the content-analysis neasures. The second portion of the study was concerned with the interrelatedness of the different language measures. In a recent study, Feldstein and Jaffe (1961) found a sig- nificant negative correlation between the type-token ratio and lahl's non-ah ratio for normal subjects but not for schizophrenic patients. With the exception of this study, there is no available infernation on the interrelatedness of the various measures used in the present research. In- asmuch as they all appear to be related to intrapersonal states, the assessment of their interrelatedness seems inportant. II. HETHOD Source of Data The lichigan State University Counseling Center is developing a library of tape—recorded therapy sessions. This library consists of tape-recorded interview sessions with self-referred undergraduate students who have ap- proached the Counseling Center seeking help with personal- social problems. Only those persons who were considered to be appropriate cases for the Counseling Center, as de- termined in an intake interview, and who had had no pre- vious psychotherapy were included. .At the time the study was carried out, this library consisted of recorded inter- views with approximately 50 clients. Clients The data used in the present study were obtained from tape-recorded psychotherapy interviews of 22 female clients selected from the above-mentioned tape library. These 22 subjects conprised the entire population of female clients for when both the first and last interview record- ings were available. Therapist Rating of the Outcome of Therapy .At the ternination of therapy with a client, the therapist involved rated the treatment as "successful,” 10 11 ”partially successful," ”partially unsuccessful,“ or ”now successful.“ For two of the 22 clients no therapist rating was available. For the purposes of analysis the clients rated "successful'I and ”partially successful" were grouped together and labeled "successful“ (R . 12) while the cli- ents rated “unsuccessful“ and ”partially unsuccessful” were labeled ”unsuccessful" (N a 8). Method of Collecting Data Three language sanples of 100 words each were tran- scribed fron each interview session. A sample was taken free the beginning, niddle, and end of each session by setting the tape indicator on '0' as the session began and starting to transcribe a sanple when the indicator reached '20,” "160,” and “300.” Each transcription began with the first word of the first sentence originating past the indicator point. If interruptions occurred in the client's speech which were caused by factors not under. control of the speaker, the entire phrase or clause in which this occurred was not transcribed. These included, prinarily, instances of therapist interruption or poor sound quality of the tape which prohibited accurate trans- lation of the client's speech. All vocalizations were transcribed including words, stutters, laughs, sighs, coughs, and miscellaneous vocalizations as “ah,” "uh," and "an." 12 Coding of the Language heasures Average number of Syllables ggr word. The total number of syllables in each language sample was counted and sunned for the three sanples from each interview. The sun total of syllables for each interview was then divided by 540, the total nunber of words in three sanples from one interview, to arrive at the average nunber of syllables per word. slag-Token Ratio. Type-token ratios were tallied for the first 100 words of each language sample by count- ing the nunber of different words used and dividing by the total number of words used. These were then averaged over the three sanples obtained fron each session to yield a single TTR for each interview session. Personal Pronoun Preggengy. All personal pronouns occurring in any context were counted for each language sanple and sun-ed for the three sanples from each inter- view. rhe sun total of personal pronouns for each inter- view was then divided by 540 and nultiplied by 100 to yield a frequency of usage per 109 words. This nethod of obtain- ing a frequency per 100 words was also used for ggalifica- tion tern fre en , allness tern fre enc , anount of re tition, and filled pause freggency. ggalification Tern Preggency. This neasure in- cluded all words and phrases of anbiguity such as almost, he s, a roxinatel , nggbe, nore-or-less, and about. 13 Verb-Adjective ggotien . lAll verbs in each lan- guage sample were counted, including auxiliary verbs. All adjectives in each sanple were sinilarly tallied. The total verb count for each sanple was divided by the total adjec- tive count for each sanple and the quotient carried out to two decinal places. The verb-adjective ggotients were then averaged over the three sanples obtained fron each session to yield a single VAQ for each interview session. Anount of Repetition. This neasure included such occurrences as stutter (dr-drive), one-word repetition (I don't - don't know), two-word repetition (I an - I an going), and phrase repetition (I don't know - I don't know). Each word repeated in these fashions was counted as one repeti- tion. Filled Pause Ereggencz. Any verbal or non-verbal sound which intruded upon the client's speech and inter- rupted the ongoing flow was counted as a filled pause. This included laughs, sighs, coughs, vocalizations such as “ah” and ”an," and procrastinations such as "well“ and ”oh" (when not used as enclanations). It did not include “uh-huh" and 'nn-hnn' which were construed as having nean- ing, assent and dissent, respectively. III. RESULTS Each language neasure was obtained free both the first and last interviews of each client. These data ape pear in Appendix.3. Sign tests were used to analyse the changes in the language measures fron the first to the last interview (see Table l). [A significant change (p < .05) occurred in two of the neasures. Both anount of repetition and filled pause fregpency decreased significantly from the first to the last interview. Changes approaching significance occurred in two other measures. Average nunber of syllables per word tended to increase fron the first to the last interview (p - .06), as did pepsonal pronoun freggengy (p - .08). Analysis of the sane data with t-tests for paired- conparisons also showed a significant change free the first to the last interview for the anount of repgtition and filled pause neasures (see Table 2). The tendency of the neasures to increase or decrease free the first to the last interview was analysed in terns of therapy outcone, “successful“ and "unsuccessful.“ As seen in Table 3, the language patterns of the "successful" and “unsuccessful“ clients varied in a sinilar fashion from 14 15 Table 1 Changes in Language leasures fron First to Last Interviews W Frequency of Directional Change' Measure Sign Test Increase Decrease (two-tailed) Syl 14 5 n.s.(.06) TTR 8 ll n.s. PP 15 6 n.s.(.08) dual 8 l4 n.s. All 9 l3 n.s. VAQ 12 19 n.s. Rep 5 15 .94 PP 5 16 .63 ‘Ties result in frequencies that do not total 22 in sone cases. 16 Table 2 Changes in Language Measures fron First to Last Interviews First Int. Last Int. leasure t p f 8.13. Y 3.9. (two-tailed) Syl 1.23 .03 1.25 .04 .20 n.s. TTR .60 .04 .60 .04 .44 n.s. PP 18.53 2.78 18.82 2.18 .56 n.s. dual 4.84 .92 4.45 1.32 1.42 n.s. All 1.96 .90 1.78 .80 .69 n.s. 110 4.01 2.61 3.88 1.86 .43 n.s. Rep 3.15 1.25 2.60 1.30 2.54 .02 PP 5.01 2.56 3.99 1.86 2.20 .05 17 Table 3 Conparison of Language heasure Changes of Successful and Unsuccessful Clients heasure Clients‘ Increase Decrease X2 p Syl 3 g g 0.16 n.s. TTR 3 g i o . 25 n. s . PP g z i 0.35 n.s. dual 3 g 2 1.44 n.s. A11 3 g g 9.93 .01 VAQ 3 Z i 0.13 n.s. Rep 3 § 3 1.52 n.s. PP g i z 0.35 n.s. aS a successful outcone; U - unsuccessful outcone. 18 the first to the last interview for 7 of the neasures. ,Allness tern freggency was the exception. Pregnancy of allness tern usage decreased for 10 of the 12 "successful“ clients while increasing for 6 of the 8 "unsuccessful” cli- ents. The conputed value of Chi-square (X? - 9.93) was significant at the .01 level. An indication of the stability of each language neasure was obtained by correlating the first and last interview scores (see Table 4). Wide variation anong cli— ents in the tine span between the first and last interview sessions prevents these coefficients fron accurately re- flecting the stability of the neasures. These data are thus presented only as indications of the stability. In addition, coefficients of stability for each language neas- are within interviews appear in Appendix C. The correlational analysis showed a consistent grouping of significant correlations anong 5 of the lan- guage neasures (see Tables 5, 6, & 7). Twenty-six of the, 84 conputed correlation coefficients achieved significance at the .05 level or above. Of these 26 significant corre- lations, 23 occurred anong the Syl, TTR, PP, the, and PP neasures. Table 8 presents a sunnary of the nunber and direction of the significant correlations anong these 5 neasures. The high intercorrelations anong these 5 neasures indicated that they were not independent of each other. The four neasures which had not exhibited a significant 19 Table 4 Correlation between First and.Last Interview Language Heasure Scores Syl TIE PP dual All VAC Rep PP .44 .94 .52 .36 .02 .35 .68 .56 20 Table 5 Intercorrelations anong First Interview Language Measure Scores TIE PP alll All YAQ Rep PP Syl .21 -.45‘ .33 .07 -.37 .09 -.45' TTR -.28 -.10 .17 -.27 -.03 -.50‘ PP .21 .13 .74. -.11 .45. anal .02 .09 .18 -.25 All .58. .43' -.07 VAQ .07 .71' Rep .02 'Significant at alpha - .05 21 Table 6 Intercorrelations anong Last Interview Language Measure Scores TTR PP dual A11 FAQ Rep FP Syl .67‘ -.66° -.06 .00 -.55' -.09 -.58‘ TTR -.44° .09 -.16 -.50’ -.11 -.52‘ PP -.21 .19 .48‘ .00 .32 dual -.47‘ .08 .06 .12 All .30 -.19 .07 FAQ -.21 .36 Rep .25 ‘Significant at alpha - .05 22 Table 7 Intercorrelations anong Conbined First and Last Interview Language Measure Scores TTR PP anal A11 FAQ Rep PP Syl .38 -.60‘ .19 -.23 -.54‘ -.07 —.60‘ TTR -.42‘ .03 -.26 -.70' -.18 -.53‘ PP -.13 .34 .72‘ -.05 .45‘ final -.23 .09 .22 -.05 A11 .37 .18 ".17 FAQ -.04 .66‘ Rep .15 'Significant at alpha - .05 23 Table 8 Direction and lunber of Significant Correlations TTR PP FAQ PP Syl (4») 1 (-) 3 (-) 2 (-) 3 TTR (-) 2 (-) 2 (-) 3 PP (+) 3 (+) 2 FAQ (+) 2 lote.—-Each cell represents 3 possible significant correlations. Total possible significant correlations - 30 Total obtained significant correlations - 23 24 change from the first to the last interview were thus con- bined. It was felt that the conbined neasure night he nore sensitive to whatever changes in intrapersonal states had occurred. A t-test for paired conparisons was perforned on the conbined neasure scores of the first and last inter- views. The obtained t-value (t n .84) was not significant. IV. DISCUSSION Langpage Measure Change Osgood (1954), in a review of psycholinguistics, described three levels of habit organization within the process of speech. The language neasures used in the pres-- ent research were selected such that each of Osgood's three levels was represented. The first level is the point at which the speaker decides the overall structure of the utterance he intends to nake. Qualification terns, allness terms, and verb- adjective quotient reflect this level of habit organization. lone of these neasures changed significantly fron the first to the last interview of psychotherapy. .At the second level of organization, the speaker selects the verbal tokens necessary to express the utter- ance he structurally fornulated at Level I. This level of organization is reflected in the average nunber of syl- lables per word, type-token ratio, and personal pronoun neasures. Although none of these neasures changed signif- icantly fron the first to the last interview, average nun- ber of syllables per word and personal pronoun frequency exhibited changes which approached statistical significance. The speaker is concerned with the nechanics of 25 26 vocalization at the third level of organisation. Reflect- ing this level are the anount of repetition and filled pause neasures. Both of these neasures changed significantly fron the first to the last interview session. The results indicate that whatever is occurring between the two interviews has the greatest effect on the clients' speech at the fluency level. Less effect is shown at the vocabulary level and no effect is exhibited at the structural level. These results are in close agreanent with the results of previous studies when viewed fron within the conceptual franework of Osgood's levels of organisation. The expressive neasures have been found to reflect transitory states of anxiety and stress of a.nild degree. Significant differences using content neasures, however, have nest generally occurred in conparisons nade between clients with severe pathological disturbances, or in one case people with suicidal intent, and nornal subjects. Hith psychotic patients, it is safe to assune that an ex- tensive reorganization of behavior patterns has occurred. In the Osgood and walker (1959) study, it was assuned that the suicide victins, just prior to taking their own lives, were functioning under very extrene duress. we can thus tentatively conclude that ninor fluc- tuations in the intrapersonal states are enough to cause variations at the expressive level of speech. For changes at the vocabulary level to occur, greater changes in 27 intrapersonal states are necessary, and najor changes are required for habits at the structural level to be affected. In the present case, the clients were neither suf- fering fron severe pathological disorders nor, presunably, functioning under the duress characteristic of the suicide victins. Under these circunstances only those habits at the expressive level were being naxinally affected, with sone effect also being observed at the vocabulary level. The changes which either approached or achieved significance did so in the expected directions; that is, consistent with a decrease in stress. The clients' speech in the final intervieu, as reflected by the repetition and pause neasures, was nore fluent. Their speech in the final interview also tended to be nore diversified as character- ised by the increased average nunber of syllables per word. Previous research actually gave no basis for pre- dicting a direction of change regarding personal pronoun usage. The change which occurred was in the direction of increased usage in the final interview. This nay well have been as such, or nere, a function of the increased intimacy of the therapeutic relationship at the final interview as it was a function of decreased stress (if the two are sep- arable). In other words, the clients nay well have used nore personal pronouns because they were nore freely able to talk about thenselves and interpersonal relationships and because they knew that the therapist would recognize 28 the referents of the pronouns they used. One of the neasures at the structural level, all- ness terns, while not showing a significant change for the group of clients as a whole, did show a narhed difference in the direction of change between the successful and un- successful clients. The direction of change for the suc- cessful clients was in the expected direction, fewer all- ness terns in the final interview. This one significant test cannot be considered too seriously because of the nunber of tests conducted, because sone therapist ratings were nade nonths after ternination, and because no criteria of success or failure were used other than the therapists' personal opinions. The result, however, is not inconsistent with the concdption of organi- sational levels. While such ninor changes_as presunably occurred with all clients over the course of therapy were enough to effect sone changes at the other levels in the najority of clients, it would take a najor change, such as presunably occurs in successful therapy, to effect lane guage habits at this level. The present research suffers fron nany linitations. The sanple was snall and consisted solely of young adult fenale college students who were highly sophisticated ver- bally. The clients were not suffering fron severe disturb- ances and only the first and last psychotherapy interviews uere analyzed. In addition, as noted in the hethod section, 29 unintelligible portions of the clients' speech were not transcribed. It is possible that the unintelligibility of sono of those portions was due to the high enotional stress under which the clients were functioning at these points in tine. If this is true, those portions of the clients' speech which were nost affected by enotional states were not included in the data. Nonetheless, the results, in conjunction with Osgood's (1954) conceptualized levels of language habit organisation, provide the basis fron which to attach the current, rather singular, direction of research into speech in psychotherapy. Mahl (1959) summarized, or perhaps provided the genesis for, this direction when he stated that the most fruitful channel of research into the relationship between speech and intrapersonal states was that utilizing expressive measures. . Certainly nuch of value renains to be discovered concerning such expressive aspects of speech as fluency, rate, intonation, and pitch and their relation to intra- personal states in psychotherapy. However, this does not seen reason enough to forego research into other levels of language behavior. Understanding of the psychotherapy process would be advanced by increased knowledge at all levels of lan- guage behavior. New, better, and more theoretically based neasures are needed at the deeper levels of language habits. 3O Personality and therapy theories, such as those proposed by Sullivan (1947), Ellis (1962), Berne (1961), and others, which explicitly deal with language behavior, seen to be fertile soil from which to derive new measures of this type. Intercorrelational Analysis The analysis showed five measures to be highly and consistently intercorrelated. Verb-adjective quotient, filled pause frequency, and personal pronoun frequency were all positively correlated. Each correlated negatively with average number of syllables per word and type-token ratio which were, in turn, positively correlated. In attempting to explain the interrelationship anong these measures, the verb-adjective quotient was found to be the best point of departure. This was also consis— tent with the tenporal aspect of Osgood's conception of levels of organization. Structural decisions on the part of the speaker occur first in tine, followed by vocabulary choices, and then by expressive choices. Therefore, the logical point at which to begin the explanation of the interrelationships was with the verb-adjective quotient, which represented activity at the first point in the ten- poral sequence . he can state, then, that a person who uses nany verbs relative to the nunber of adjectives enpleyed tends to use nore personal pronouns, shorter words, and fewer different words, and tends to have sore filled pauses. 31 Or, to put it differently, the sore active, as opposed to descriptive, the language, the less diversified is the vo- cabulary esployed and the less fluent is the speech in terns of filled pauses. Pros this frase of reference the first relationship which needs to be explained is that between the active- descriptive and the vocabulary dinensions. Characteristics of the English language can be effectively used to explain such of this relationship. verbs in our language typically contain fewer syllables than do adjectives. we would thus expect the speech of the person who uses sore verbs relap tive to adjectives to contain a fewer nusber of syllables. Sisilarly, the English language contains sany sore adjectives than verbs and the speaker has a greater choice asong synonysous adjectives than asong synonynous verbs when expressing any given intent. Thus, the individual who uses sore verbs has a ssaller nusber of different words available to his and his speech is expected to be charac- terised by a lower type-token ratio. The relationship between verb-adjective quotient and personal pronoun frequency is not as clear-cut and needs to be viewed at a sosewhat sore abstract level. The individual who uses sore verbs relative to adjectives is enploying language in a less descriptive fashion than is one who uses fewer verbs and sore adjectives. we sight thus expect his language to be less descriptive in other 32 respects. One aspect of descriptiveness can be defined as naking, or inplying, distinctions or discrininations. The use of pronouns is just the opposite; it is a process of generalization. It is a way of taking nany, and in sone ways very distinct, entities and subsuning then under one verbal label. we would expect, then, speech which is nore active and less descriptive to contain a higher number of personal pronouns. the interrelationship between the three vocabulary neasures can be seen quite clearly at this point.- High frequency personal pronouns tend to be nonosyllabic. Con- sequently, the sore personal pronouns used, the fewer nun- her of syllables enployed. Again, pronouns are used to subsune a nunber of different entities under one label and consequently the sore personal pronouns used, the fewer the nunber of different words required. Both the length of words used and thetype-token ratio are directly related to the size of the vocabulary available to the speaker. The probability that the person's vocabulary includes longer, rarer words increases as the size of the vocabulary increases. Sinilarly, the likeli- hood of the speaker using nany different words increases as the nunber of different words available to hin increases. we thus expect speech characterized by a higher nunber of syllables per word to also characteristically contain nany nore different words. Up to this point we have seen that the speech of 33 'the sore verbally sophisticated individual displays a lower 'verb-adjective quotient and personal pronoun frequency and a.nigher average nunber of syllables per word and type-token .ratio. In other words, the sore sophisticated speaker dis- plays a nore descriptive style and a greater con-and of vocabulary. we would expect this sane person to be sore sophisticated at the fluency level also; that is, to exhibit a fewer nunber of filled pauses. Anount of Repetition and Filled Pause Preggency Hahl (1956) stated that, although he had no defin- itive data to the effect, some people sealed to character- istically use 'ah' while others characteristically repeated words and/or phrases, and that these speaking differences O night in some way be related to personality differences. Our data allow a sonewhat cursory look into the first part of uahl's statenent, that some people are 'ah-er's' while others are "repeaters." Although both the anount of repetition and number of filled pauses were relatively stable across interviews, there was no correlation between the neasures for either interview. hahl's conjecture was that those people who tended to use "ah” tended not to "repeat“ and vice-versa, inplying a negative correlation between the two neasures. The present results, although linited because of snall sanple size, indicate that no such negative relationship exists. V. SUMMARY Previous language style research indicated that a nunber of language and speech.neasures were capable of discrininating between verbal naterial produced by disturbed and non-disturbed individuals. A review of this research generated a two-foldclassification systen within which the previous studies could be categorised. The types of neasures used were classified as either content or expres- sive neasures while the types of experinental designs en- ployed were classified as either diagnostic or dynanic. It was noted that content neasures were generally used in diagnostic studies and expressive neasures generally en- ployed in dynanic studies. The present research was concerned with changes in clients' verbal behavior free the beginning to the end of psychotherapy and, as such, utilised both content and expressive neasures in a dynanic design. The content neas- ures were the avera e nunber of s llables word, the tymtoken ratio, arsenal pronoun freggengy, ggalification tern fre en~ , allness tern fre enc , and the verb-adjec- tive otient; the expressive neasures were the anount of reggtition and the filled pguse frquensy. Speech sanples drawn free the first and last psychotherapy interviews of 34 35 22 clients were analysed in terns of these neasures. In addition, because little is known about the interrelation of the language neasures used, a correlational analysis was carried out for both the first and last interview, and for the two interviews conbined. .The results were organised in terns of Osgood's (1954) levels of language habit organisation. The three content neasures reflecting the structural level of organi- sation exhibited no change free the first to the last inter- view. Two of the three content neasures which reflected organisation at the vocabulary level tended to change in the direction of increased vocabulary diversity and increased interpersonal faniliarity but did not achieve statistical significance. The neasures at the expressive level of or- ganisation exhibited the greatest degree of change. Both expressive neasures indicated a significant shift free the first to the last interview in the direction of increased fluency. Thus, for the group of clients as a whole, the greatest change occurred at the expressive level, less change was exhibited at the vocabulary level, and no change resulted at the structural level. In addition, one of the neasures, allness terns, showed a.narked difference in the direction of change between the successful and unsuccess- ful clients. The correlational analysis indicated a strong re- lation anong average nunber of syllables 225 word,‘§zp§f 36 token ratio, msonal pronoun fregency, verb-adjective gpotient, and filled pause fregpency. Various reasons for these relationships were suggested. BIBLIOGRAPHY Balkan, Eva, a nassernan, J. H. The language of phantasy: III. The language of phantasies of patients with conversion hysteria, anxiety states, and obsessive- conpulsive neurosis. J. Psychol., 1946, 10, 75-86. Berne, E. Transactional analysis in psychotherapy. New York: Grove Press, Inc., 1. Boder, D. P. The adjective-verb quotient. Psychol hec., 1940, 3, 310-343. Chotlos, J. w. A statistical and conparative analysis of individual written language sanples. Psychol. Honogr. , 194‘, 56, “Ge 2 ('Ml. lo. 25 , -111e Conrad, Dorothy, a Conrad, R. The use of personal pronouns as categories for studying snall group interaction. J. abnorn. soc. Psychol., 1956, 52, 277-279. Dibner, A. S. Cue-counting: .A ueasure of anxiety in Inter- views. J. consult. Psychol., 1956, 20, 475-478. Dittnan, A. T., a Uynne, L. C. Linguistic techniques and the analysis of enotionality in interviews. g, abnorn. soc. Psychol., 1961, 63, 201-294. Eldred, S. H., a Price, B. B. [A linguistic evaluation of feeling states in psychotherapy. Ps chiatr , 1958, 21, 115-121. Ellis, A. Reason and enotion in psychotherapy. low York: LYl. Stunt 9 e Fairbanks, Helen. The quantitative differentiation of sanples of spoken language. Ps chol hone r., 1944, 56, lo. 2 (whole no. 255), lS-Ig. Peldstein, 8., a Jaffe, J2 A note about speech disturbances and vocabulary diversity. J. Connunication, 1962, 12 (3), 166-170. Flesch, R. A new readability yardstick. J. appl. Psychol., 37 38 y_. leasuring the level of abstraction. J. appl. PSZChele’ 1950, 34, 384-3900 Gilinsky, A. 8. How valid is the Flesch readability for- nula. Aner. Psychologist., 1948, 3, 261. Gottschalk, L. A. (Ed.). Con arative sycholin -istic an- al sis of two s choEEEra u~ c interviews. lew York: InternaEional Universities Press, Inc., 1961. Johnson, H. Studies in language behavior. Psychol. Honogr., 1944, 56, no. 2 (whole lo. 255). Krause, I. 8., & Pilisuk, n. .Anxiety in verbal behavior. J. consult. Ppychol., 1961, 25, 414-419. Lorenz, Maria, & Cobb, 3. Language behavior in psychotic and neurotic subjects. Arch. laurel. Psychiat., 1954, 72, 665-673. Hahl, G. P. Disturbances and silences in the patient's speech in psychotherapy. J. abnorn. soc. Psychol., 1956, 53, 1-15. _yfi_. Exploring enotional states by content analysis. In I. Pool (Ed.), Trends in content anal sis. Hrbana: University of Iliinois Press, i955. Pp. 83‘130e . Measures of two expressive aspects of a patient's speech in two psychotherapeutic interviews. In L. A. Gottschalk (Ed.), Con arative :sycholingpistic analysis of two psychotherapgu- c interviews. New Yer : Interna onal Universities Press, Inc., 1961. Mann, nary. The quantitative differentiation of sanples of written language. Ps chol. mono r., 1944, 56, "be 2 ('hO1. "be 255), 1- hcauown, l. A. Linguistic transcription and specification of psychiatric interview naterials. Psychiatry, 1957, 20, 79-66. Osgood, C. E. Psycholinguistics: A survey of theory and research problens. J. abnorn. soc. Psychol., 1954, 49 (‘)9 1-293. Osgood, C. 3., a walker, Evelyn G. Motivation and language behavior: .A content analysis of suicide notes. J. abnorn. soc. Psychol., 1959, 59, 58-67. 39 Pittenger, R. 8., Hockett, C. r., a Danehy, J. J. The first five ninutes: A san la of nicrosco ic interview analysis. Ithaca, New York: Paul uarEiEeau, i960. Rainy, V. C. Self-reference in counseling interviews. J. consult. Psychol., 1948, 12, 155-163. Siegel, 5. Hon aranetric statistics. law York: HcGraw- Hill, IESE. . Sullivan, H. S. Conceptions of nodern psychiatgy. William Alanson White Foundat on, 1947. wagner, R. r., a Uillians, J. E. An analysis of speech behavior in groups differing in achieve-ent inagery and.defensiveness. J. Pers., 1961, 29, 1-10. walker, Helen 11., & Lev, J. Statistical inference. lew York: Holt, Rinehart, anaifiinstzn, I955. Zipf, G. K. Hunan behavior and the inci 1e of least of- fort. Ezibridge, hass.: Adgison-Hesiey, I519. APPEIDICES APPENDIX.A -- PROEEDURES 3529 FOR SCORIIG LAIGUAGB P1881635 Average lunber of Syllables 225 word (Syl) Syllabification in all cases was carried out accord- ing to wgbsgeg's 33w world Digtiogagy of the Anerican Lan- a e, 1960. gypg-Token Ratio' (1:2) 1. Each group of letters separated by spaces on both sides fron adjacent groups of letters was counted as a word, even though it night be part of a place nane or an initial. 2. titles and nodes of address (Dr., lr., etc.) were counted as separate words. 3. Any number was counted as one word. 4. Hyphenated words (webster's lew‘Iorld Dictionary was used as authority) were counted as single words; e.g., ”hitch-hiker” is one word. 5. Contractions were recorded as spoken; e.g., "didn't" is not to be changed to ”did not”: “didn't” is one word. 6. Each tine a word was used as a different part ‘Revised and reproduced in part fron Johnson et a1., Studies in language behavior, Psychol. nonogr., 1944, 56, so. 2 (whole lo. 255). 41 42 of speech it was counted as a different word. 7. Any two words are different unless spelled ex- actly alike, except: a. plurals and possessives, and contractions b. Ce involving apostrophes, are to be differen- tiated even though they are spelled alike. the third person singular fern of a verb is not counted different fron the other forn of the verb. different tenses of a verb having identical spellings are counted as different words; _e.g., “read.“ Personal Pronoun Preggency (PP) The following list includes all words counted as personal pronouns: I “Y nine nyself we our ours ourselves you your yours yourself yourselves be his hin hinself the-selves she her hers herself tit! their theirs then 43 Qualification Tern Preggency’ (anal) Qualification terns are words or phrases which ex- press anbivalence or enable the speaker to nodify the anount of connitnent he nakes in his statenent. Each such_word or phrase is counted as one qualification tern. Below is a list of frequently used words and phrases of this type: noro-or-less it-could-be possibly sort-of in-this-respect sonotines seens however but except appears or alnost if looks-like kind-of I'n-not-sure would guess probably naybe should surely night because really Allness Tern Preggency‘ (All) Allness terns are extrene and polarised state-ents or words which pernit no exception. Each such word or state- nent was counted as one allness tern. .A partial list follows: never perfectly all no-nere none-whatsoever always conpletely anything no-natter-what forever everybody wholly everyone infinite exactly anyone no-one ultinate nothing everything without-a-doubt definitely nobody regardless every-tine utter 'Reproduced in part fron Osgood and Walker, lotiva- tion and language behavior: .A content analysis of suicide notes, J. abnorn. soc. Psychol., 1959, 59, 56-67. 44 verb-Adjective ggotient' (VAQ) The verb and adjective counts were nade in accord- ance with the following rules: 1. Both attributivo and predicate adjectives were counted. 2. bonus used as adjectives were not counted un- less specifically recognised by Hebster's New world Diction- ary as also functioning in the adjective fern. 3. Adjectives used as nouns (the 'poor') were not counted. 4. Ordinal nunerals were not counted. 5. Verbs in all ferns including infinitivos and participles (except as described under point 6) were counted. 6. Participles used without nouns and preceded by an article or by the preposition “of“ were not counted. Anount of Repetition (Rep) (A repetition was defined as a word following closely upon itself, which was unnecessary and therefore redundant with regard to expressing the intent of the statement. Each word repeated in such fashion was counted as one repe- tition. Sone exanples follow: He dr-drove hone; he-he drove hone (l repetition) I don't - I don't know (2 repetitions) ‘Revised and reproduced in part from D. P. Boder, The adjective-verb quotient, Psychol. Rec., 1940, 3, 310- 343. 45 I love hin - I do love bin (3 repetitions) I don't - I don't - I just don't know (4 repetitions) tilled Pause Preggency (PP) _ A filled pause was defined as a break in the flow of neaningful verbalization which was filled with, or caused by, non-lexical vocalizations. Each such break, regardless of the number of vocalisations within it, was counted as one pause. For example, "I 'ah' 'ah' didn't do that" con- tains only one filled pauSe, while "I 25; didn't in}; do that“ contains two filled pauses. APPENDIX 3 FIRST INTERVIEW LAIGUAGE MEASURE SCORES Client 01 03 05 06 11 17 20 40 50 $1 52 02 10 15 19 30 32 33 38 42 47 48 511 T‘I‘R pp _gual All VAL Rep F? 1.24 .61 16.10 4.81 1.48 3.87 3.52 4.07 1.18 .50 26.27 5.00 3.14 15.08 3.70 12.58 1.23 .66 22.39 4.81 2.59 4.71 2.96 4.81 1.29 .54 17.39 5.74 0.56 2.76 2.22 3.70 1.25 .56 22.39 5.74 1.11 4.18 1.66 4.26 1.20 .59 21.28 4.26 1.30 4.66 0.92 7.58 1.25 .61 13.69 5.55 2.04 2.73 1.85 '2.78 1.20 .64 19.06 5.36 1.48 4.33 4.81 7.40 1.20 .59 19.80 4.44 3.14 3.07 2.96 4.44 1.25 .54 18.13 6.48 1.48 3.79 4.07 7.40 1.24 .60 16.28 4.07 2.78 2.68 3.88 2.59 1.20 .61 18.69 3.88 0.92 2.45 2.40 4.44 1.25 .62 15.73 3.52 2.40 2.41 2.59 4.44 1.26 .58 19.24 5.55 2.59 5.31 3.33 3.52 1.23 .63 18.32 5.00 1.48 3.25 3.52 2.96 1.18 .56 16.65 3.88 0.74 2.41 2.40 6.66 1.26 .68 19.06 5.00 1.11 3.83 1.30 4.26 1.20 .58 18.87 4.62 1.85 3.72 4.62 2.96 1.23 .65 17.58 3.88 2.59 3.74 1.85 5.55 1.27 .58 14.99 2.96 1.85 3.80 4.81 9.25 1.28 .65 17.58 5.74 4.07 2.64 5.18 2.04 1.25 .63 18.13 6.29 2.40 2.88 4.81 2.59 lote.--See Appendix A for the full nanes of the above abbreviated language neasures. 46 47 LAST INTERVIEW LANGUAGE MEASURE SCORES Client Syl TTR PP 9231 .All VAQ Rep FP 01 1.26 .59 16.23 4.26 2.22 5.33 2.04 3.33 63 1.17 .50 22.39 4.31 2.59 16.74 2.22 7.40 05 1.21 .53 23.63 3.70 2.22 4.52 0.92 3.33 06 1.30 .63 18.69 3.33 1.43 2.35 4.26 1.66 11 1.30 .63 17.76 4.62 2.22 2.69 1.66 2.40 17 1.23 .63 21.46 3.33 3.33 5.67 1.43 4.62 26 1.30 .62 13.13 6.10 6.55 4.65 1.30 2.59 46 1.25 .64 19.61 5.92 1.11 2.77 5.36 6.29 56 1.23 .63 19.93 3.33 1.11 3.23 3.76 2.73 51 1.25 .65 16.23 6.66 6.92 3.92 1.35 3.52 52 1.26 .59 16.84 7.03 2.04 2.46 2.46 5.36 62 1.26 .59 13.37 2.59 1.66 3.37 1.30 3.52 16 1.36 .62 15.73 1.36 2.73 1.74 -l.66 4.44 15 1.31 .59 17.95 4.26 2.40 4.76 2.96 4.31 19 1.23 .60 18.69 4.67 1.36 3.36 3.33 1.36 36 1.24 .63 13.56 5.66 2.22 4.66 1.66 2.46 32 1.30 .57 16.47 5.36 0.92 2.13 1.66 1.36 33 1.27 .54 21.46 3.14 2.73 3.76 3.33 3.33 33 1.23 .61 13.32 5.66 6.92 2.99 1.43 5.00 42 1.13 .52 21.33 4.62 0.56 4.06 3.33 3.32 47 1.25 .57 16.84 4.44 1.30 2.73 5.13 5.13 43 1.22 .56 13.32 4.44 2.59 3.59 4.67 5.66 lote.--See Appendix.A for the full noses of the above abbreviated language neasures. APPENDIX C Language Measure Stability Coefficients within the First and Last Interviews First Interview' Last Interview Measure a B-H H—E B-E B-H H-E B-E Syl -.20 -.04 .42 .55 -.07 -.11 TTR .12 .08 .47 -.19 .06 .23 PP .32 .31 .25 .49 .19 .05 dual -.09 -.38 .00 .30 .18 .22 All -.03 .14 .02 .67 .05 -.04 YAQ .66 .73 .59 .35 .68 .69 Rep -.06 .43 -.13 .23 .56 .18 PP .38 .56 .60 .48 .74 .59 a3 a beginning of interview language passages; H - niddle of interview language passes; E n and of inter- view passages. 48 MICHIGAN STQTE UNIV. LIBRQRIES I ll‘lWllHlllHl mumI1Immlunlmlmml 31293010930133