Citation
Client affective verbalization and therapist influence strategies in analogue psychotherapy interviews

Material Information

Title:
Client affective verbalization and therapist influence strategies in analogue psychotherapy interviews
Creator:
Jones, Stanley Elton, 1948-
Publication Date:
Language:
English
Physical Description:
viii, 116 leaves : ill. ; 28 cm.

Subjects

Subjects / Keywords:
Clinical psychology ( jstor )
Dyadic relations ( jstor )
Gene therapy ( jstor )
Manuals ( jstor )
Psychology ( jstor )
Psychotherapy ( jstor )
Self referential statements ( jstor )
Social interaction ( jstor )
Verbalization ( jstor )
Words ( jstor )
Dissertations, Academic -- Psychology -- UF ( lcsh )
Interviewing in psychiatry ( lcsh )
Psychology thesis Ph. D ( lcsh )
Psychotherapist and patient ( lcsh )
Psychotherapy ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis--University of Florida.
Bibliography:
Includes bibliographical references (leaves 109-115).
Additional Physical Form:
Also available online.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Stanley E. Jones.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
025655293 ( ALEPH )
03162650 ( OCLC )

Downloads

This item has the following downloads:


Full Text













CLIENT AFFECTIVE VERBALIZATION
AND THERAPIST INFLUENCE STRATEGIES IN
ANALOGUE PSYCHOTHERAPY INTERVIEWS












By

STANLEY E. JONES





















A THESIS PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA 1976



























DEDICATION



To Barbara

Whose patience and support have
made all the difference.














ACKNOWLEDGMENTS



I would like to thank my chairperson, Dr. Jacquelin

Goldman, for her willing and consistent guidance and assistance through a long period of often frustrating endeavor. I am also appreciative of the contributions of each of my other committee members: Dr. Audrey Schumacher, for her influence on my thinking about therapy and her always incisive critiques; Dr. Franz Epting for his encouragement to be more scholarly and to be creative in research applications; Dr. Marvin Shaw, for his assistance in solving the difficult design and analysis problems involved in the study; and Dr. Hannelore Wass, for her perspective from outside mv immediate area.

Many other individuals, faculty, staff, fellow graduate students, and undergraduate students, were involved in the various aspects of the work reported here, and though they are too numerous to mention, I would like to express my appreciation for their cooperation. A special word of thanks goes to my typist, Barbara Jones, for her hard work above and beyond the call of duty.














TABLE OF CONTENTS


Page

ACKNOWLEDGEMENTS . . . . . . . . . iii

LIST OF TABLES . . . . . . . . . . vi

ABSTRACT . . . . . . . . . . . vii

CHAPTER

1 INTRODUCTION . . . . . . . . 1

Communication as Relationship
Definition . . . . . . . 5
Therapist Goals as a Definition . . 6

2 DEFINING THE RESEARCH QUESTION . . . . 9

Communication Science: An Overview . 13 Defining Communication of Affect . . 16 Research Hypotheses . . . . . 22

3 TECHNICAL CONSIDERATIONS . . . . . 24

Method . . . . . . . . . 24
Rating Procedures . . . . . . 27

4 RESULTS . . . . . . . . . 36

Client Verbalization and Therapist
Strategies . . . . . . 36
Compatibility and Client Sex . . . .
Client Verbalization and Session Outcome. 61 Exploratory Analyses . . . . . 62

5 DISCUSSION . . . . . . . . . 65

Conclusion . . . . . . . . 69

APPENDICES . . . . . . . . . . . 71

APPENDIX A--Therapist Post-Session
Evaluation Form . . . . . 72
APPENDIX B--Client Post-Session Evaluation
Form . . . . . . . . 73


iv







APPENDICES-continued Page

APPENDIX C--Notice for Potential Client
Subjects ........................... 74
APPENDIX D--Verbatim Telephone Contact for
Informing Selected Clients ...... 75
APPENDIX E--Rules for Locating Segments Within
Full Interviews ... ........... 76
APPENDIX F--Training Manual for Unitization of
Transcripts by Proposition ...... 78
APPENDIX G--Training Manuals for Rating All
Variables ..... .............. 81
APPENDIX H--Listing of Primary Affect Words
and Limited Primary Affect Words. . 97
APPENDIX I--Listing of Secondary Affect Words 99
APPENDIX J--Listing of Affective Idioms and
Slang ...... ................ 100
APPENDIX K--Listing of Therapist Verbal
Reinforcers .... ............. 101
APPENDIX L--Examples of Graphs Plotting
Interaction Effects .. ......... 102
APPENDIX M--Examples of Score Distribution
Graphs ...... ................ 105

REFERENCE NOTES ........ ................... 108

REFERENCES ......... ...................... 109

BIOGRAPHICAL SKETCH ...... ................. 116




























v














LIST OF TABLES



Table Page

1 ANOVA-l Summary Table ...............39

2 ANOVA-2 Summary Table ...............40

3 Correlations of Modeling and Change Scores 42

4 Affective Measures Intercorrelated. .......45

5 Summary of Cross-lagged Panel Correlations 48

6 Tabulations of Following Versus Leading
Therapist Responses.................50

7 Correlations of Reinforcement Strategy
and Client Change Score ..............52

8 Correlations Between Number of Reinforcement
Responses and Client Change Scores. .......53

9 Correlations Between Number of Reinforcement
Responses and Reinforcement Strategy .. 55

10 ANOVA-3 Summary Table ...............59

11 A140VA-4 Summary Table ...............60




















vi














Abstract of Dissertation Presented
to the Graduate Council of the University of Florida
in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy



CLIENT AFFECTIVE VERBALIZATION
AND THERAPIST INFLUENCE STRATEGIES IN
ANALOGUE PSYCHOTHERAPY INTERVIEWS

By

Stanley E. Jones

December 1976

Chairperson: Jacquelin R. Goldman Major Department: Psychology

Psychotherapy theory emphasizing interpersonal interaction is used as a theoretical base for considering strategies that therapists might take to increase client affective verbalization in therapy situations. Previous research of analogue interviews is used for defining the strategies, Instruction, Modeling, and Reinforcement, and for defining the dependent measures, client affective verbalization and client affective self reference. Professional therapists conducted single interviews with experimental subjects, after the dyads were matched on interpersonal need compatibility and on all sex combinations. A rating system for determining affective verbalization and affective self reference was developed and applied to typed transcriptions of the interviews. Therapist strategies were also measured by



vii








" rating system. All rating systems were implemented through training manual and standard procedure. Hypotheses predicted that client affective verbalization and client affective self reference would increase across the interview, with the therapist strategies associated with larger increases. Interpersonal need compatibility, sex of client, and interactions between those two variables and the strategies were also predicted as important in the effects related to the dependent measure.

The rating procedure appeared adequate with rating checks producing good reliability. Data were analyzed through a split plot factorial (2x2x3) analysis of variance of compatibility, client sex, and segment separately for client affective verbalization and client affective self reference. one effect was significant: compatibility X client sex, with a simple main effects test indicating that the compatibility manipulation was effective only for female clients, as previous research has suggested. Other analyses were conducted through Pearson or Spearman correlations. Results suggested that clients tend to influence the affective verbalization level of therapists rather than vice versa, and that therapists in this sample do not actually use the strategies which analogue studies have researched. There was support for therapists believing that client affective verbalization is important to treatment outcome. Alternative hypotheses and alternatives for further research are discussed.



viii













CHAPTER 1
INTRODUCTION



Psychotherapy is an important phenomenon, both culturally and scientifically. Culturally, it is the practice of a helping profession by therapists and the seeking of assistance by clients. Scientifically, it is the interaction of two types of experimental subjects, where one subject is to effect some change in the other, often changes in very basic personality processes of the client subject. In both cases, a major question is that of how the therapist has his impact with the client. As Strupp (1973) puts it, "How does it come about that one person, by virtue of what he is or what he does, can exert such a lasting influence on another person that the changes that result from this encounter may be termed therapeutic?" (p. 19) The variations of this question and the more specific implications of it are numerous and complex, as indicated by the scope and number of reviews on research of psychotherapy (e.g., Bergin & Garfield, 1971; Bordin, 1974; Kiesler, 1973; Meltzoff & Kornreich, 1970). The present study will be focused on looking at one aspect of answering the above question.

One difficult task in beginning this focusing is specifying the way in which the interaction of therapist and client is to be conceptualized and described. The general


1




2


features of this task are essentially the same as those involved in trying to describe the interaction of any interpersonal dyad, and the controversy which attends any attempt to do this reflects the complexity of the problem. Despite the complexity, however, people do manage to relate with one another so as to reach a consensus that in their relationships some behaviors are okay and others are not okay. This mutual definition of the relationship is reached with sufficient clarity for each person to feel that he or she knows generally what to expect and what not to expect from the other person. Given that this is true, and it seems to be a reasonable postulate, it should be possible to describe some important aspects of relationships, including therapeutic relationships, and to do so using only the cues each person in the interaction has access to--the observable of communication.

Much of the complexity of communication arises out of its imprecision, and it is unreasonable to expect a methodology for describing communication to be any more precise than communication itself. In a sense, not only is the experimental observer inferring the "meaning" of the messages composing the interaction of a dyad, but each member of the dyad is making inferences about the meaning of the other member's messages as well as coming to understand more clearly what meaning he is trying to convey. This kind of process which has a basic nature of inference and variation of meaning can not be represented clearly and precisely in any other way than to reproduce it exactly, and our closest approximation to that is video-tape or sound movies. In itself, such







a reproduction does not get us any closer to understanding the process of communication, however. A representation apart from a reproduction is required, and this implies the operation of a theory providing a model for the representation and an artificial limitation of the complexity allowed in the representation. Numerous researchers recognize specifically the need for limiting the complexity of the process studied (Auld & Murray, 1955, Cartwright, 1966; Guetzkow, 1950; Kiesler, 1973; Marsden, 1971; and Strupp, 1962) and the need for being specific about the theory used. The influence of the theory on our selection of the particular phenomena studied cannot be over-emphasized, since the operation of a theoretical perspective in the selection of a representation is the primary determinant of the meaningfulness of the research no matter how sophisticated other aspects of the research may be.

Researching communication, then, involves the selection of a theoretical perspective which will dictate the aspects of communication to be described, but we cannot expect the application of a description system to render communication any clearer or more precise than it actually is. The actual selection of that theoretical perspective involves both the preference of the experimenter and the theory's support in the literature.

Many researchers focus on the nature of psychotherapy as a special case of a general dyadic communication model (Kiesler, 1973). Certainly there is ample theoretical pre-





4


cident for this focus. Sullivan (1953, 1954) is generally considered the first theorist to have emphasized the interpersonal-interactive nature of pathology and therapy, and numerous theorists and researchers since have developed interactive formulations, most focused on communication processes (Alexander, 1957; Buck & Cuddy, 1966, Buehler & Richmond, 1965; Davis, 1971; Frank, 1961; Greenhill, 1958; Heller, 1963; Hoch & Zubin, 1958; Jones & Thibaut, 1958; Riess, 1957; Rioch & Weinstein, 1964; Ruesch, 1961; Scheflin, 1965; Sebeok, Hayes, & Bateson, 1964; Watzlawick, Beavin, & Jackson, 1967; Williamson, 1959).

Another such theorist-therapist-researcher is Haley

(1963) who developed a framework based on his background in communication science (Ruesch, 1961; Reusch & Bateson, 1951) and on his work with families of schizophrenics in the Palo Alto group (Bateson, Jackson, Haley & Weakland, 1956). Haley's conceptualization appears to be a useful framework for psychotherapy research, even though it has apparently been used in only a few well-controlled studies of psychotherapy (Davis, 1971; Davis, Fisher, & Davis, 1973, 1975; Wadington, 1975). Its use in the research conducted by Haley's research group was occassionally systematic, but only informally so (e.g., Jackson, Riskin, & Satir, 1961).

This appears largely due to the complexity and subjectivity involved in applying the entirety of Haley's model to interactions, since it accounts for such complex phenomena as the influencing aspects of statements, the metacommunica-





5


tive function of communications, and the paradoxical nature of much therapy interaction. In these ideas lies the eventual promise of Haley's model for application to explaining and refining the process of therapy, but as Haley points out (1964), the research proceeding from a model must be based on highly reliable measurements, even if some of the meaningfulness of our initial description is lost as a result. In this particular paper, only the very basic ideas of Haley's model will be used, attempting by that limitation to develop adequate tests for those ideas.


Communication as Relationshi2 Definition


Haley contends that whenever two people interact they must engage in attempts to solve two mutual problems simultaneously. One problem requires answering the question: what behaviors (including messages) are to be allowed in this relationship? In other words the dyad attempts to reach a mutual definition of the relationship. The second problem requires an attempt to answer the question: who is to control the definition of the relationship? Here it is not control over behavior directly but over the definition of the relationship. This definition is very powerful, however, and provides indirect influence over behavior.

Haley argues that no matter what messages are exchanged or what behaviors engaged in, those messages and behaviors represent attempts to reach mutual solutions to the two problems. In other words, the interaction of any dyad, including





6


that of therapist and client, can be seen as involving a process of attempting to reach a mutual definition of what behaviors will and will not occur in the relationship, and a process of determining who will control the definition.

Control of the definition is an important issue, and can be seen as a competition between the two participants in a dyad. Each is engaging in purposive behavior directed toward obtaining some desired and satisfying state of affairs, and rarely do these purposes coincide so exactly as to allow an entirely cooperative venture. Since the definition of the relationship is essentially a statement about what behaviors are to be allowed in the interaction, the person who controls that definition is influencing the interaction and therefore the other person, and has an advantageous position. The person not in control is expected to attempt to take control by redefining the relationship in a more satisfactory way for him.


Therapist Goals as a Definition


Most of an effective therapist's "needs" in a session are directed toward therapeutic goals, and are essentially the needs the therapist is attempting to satisfy in his efforts to control the definition of the relationship. One way in which therapists frequently want to influence the interaction is by focusing the client into a particular style of interaction. Many theories of therapy emphasize the importance of styles of interaction, as for example in seeing




7


the client's expression of affect as a positive step in reaching therapeutic goals, either, for example, in clientcentered therapy as a major process, or in psychodynamic therapy as a cathartic and insight-generating process.

This control of style is most clearly a control of the definition problem when the client does not exhibit the desired behavior initially, as is usually the case. The therapist wants the relationship to be one in which it is acceptable for the therapist to respond in compatible ways (e.g., reflecting affect, making interpretations about the affect, etc.). If the client interacts this way naturally there is no problem, and the therapist and client get to business rather quickly. If the client does not, the therapist must help arrange for this to happen. The way this is attempted will be referred to as the therapist "strategy."

To summarize the rationale to this point, Haley's model describes the interaction between therapist and patient as like that of any dyad and consisting of the processes of determining the definition of the relationship and competing for control of that definition. It is through these processes that the interpersonal influence of therapy takes place.

One implication of Haley's theory as it is being considered is that the therapist is expected to work toward implementing the therapeutic goals as the definition of the relationship, insofar as they apply. In developing a research examination of this theoretical position, a specific




8


therapeutic goal and specific therapist interventions for increasing the occurrence of the goal behavior will be considered. In order to clearly indicate the relevance of the interventions, the goal behavior will be discussed first.













CHAPTER 2
DEFINING THE RESEARCH QUESTION



The general research question of the present study was stated above as: "How does the influence of therapy occur?" In the present formulation, a more specific question becomes: "How does the therapist behave so as to attempt instituting the therapy goal behaviors as a definition of the relationship?" The first step in operationalizing this question for the present study is selecting and defining a specific therapy goal for study.

As used in this paper, a "therapist goal" is defined

as a behavior which may be engaged in by the client and which is believed by the therapist to be related to the occurance of therapy. "Therapy" is defined as a client interacting with a therapist and working toward a change in the initial behavior, feelings, or belief system of the client, with that change in the direction of from less to more desirable as defined by the client, by society, or by the therapist. "Mutual definition of the relationship" is defined as the statement of acceptable behaviors desired by the therapist and exhibited by the client. Here we are considering only the case where the therapist is in control. If the client is not exhibiting the desired behaviors and the therapist stops making attempts to influence the client to do so, the


9




10


client could be said to be in control. Note that the focus here is on the therapist's belief about what client behavior is therapeutic, and in studying the therapist's process, it is important to look at client behaviors that are believed to lead to the occurrence of change and desired to be a characteristic of the client early in therapy.

For example, almost all psychotherapists would see "the client verbalizes" as such a goal. Except for the rare case, the client's talking is a pre-condition to the occurrence of therapy change. A related client behavior is that when the client talks the client talks about affect. This notion, that the client's verbalization of or about affective states is important to therapy, appears to be shared by many systems of therapy, either through the direct theoretical statement of the therapist or through the implication of the goal that is set (e.g., see Harper, 1959). Even when this is not a primary end, as it is in client-centered therapy for example, it is frequently a secondary end allowing the therapist and client to gain the information they need to proceed with other goals, as for example in rational-emotive, or various psychoanalytic therapies. This client behavior, the verbalization of affect, will be the therapist goal in this study.

Research has often focused on the variable, "client expression of affect," and operationally defined it in terms of judges' ratings (e.g., Waskow, 1962; Ellsworth, 1963) or in terms of systems making inferences about the client's







emotional state on the basis of his verbal behavior (e.g., Gottschalk & Gleser, 1969). In the present study as in other studies to be discussed later, the importance of the client's expression is that it is done directly, by verbalizing using feeling words, particularly since it facilitates a process of the therapist and client mutually coming to understand something about the client. There are, of course, other points at which the client is expressing or experiencing feelings, and the client's direct report of his feeling may or may not fit our judgements about his immediate affective state. In considering this study's specific goal for the client, however, it is the client's verbalization of affect words that is of interest.

Defining client affect expression in this way also facilitates the precision and usefulness of psychotherapy research as scientific data. Haley (1964) as well as others (see Kiesler, 1971, 1973) argue for the selection of discretely measurable behaviors in researching psychotherapy and other interpersonal processes. Doing so minimizes the inference inherent in the measurement process itself, and allows us to look formally at the inferences made in a systematic way.

The degree to which we can make inferences based on

this kind of data is limited by its measuring less general, though more reliably measured, behaviors. As long as the inferences made from the study are made after careful consideration of the limitations of the data, the problem is more one of frustration in being unable to deal with more rele-





12

vant and meaningful general constructs rather than a problem of experimental error. A particularly critical example of this potential error in this study, as suggested above, is that of confusing the variable "client verbalization of affect words" with the more general variable "client affective expression." The first refers to characteristics of the denotative meaning of words and idioms used by the client while the latter refers to inferences made about the state of the client in using whatever language he uses. In this study, then, it will be important to recognize that the data, and inferences made from it, are based on considerations of the kinds of language used by a person and not necessarily on whether or not that person is in touch with affective experience.

Although there is a significant amount of literature

dealing with verbalization of affective words as a dependent variable there appears to be little that has examined its validity as an indicator that important aspects of therapy are occurring when it occurs or as a predictor of successful outcomes. Levy (1967) provided support that in statements where judges agreed therapy was occurring in primarily analytic sessions there was a greater proportion of statements with affect words than in statements where therapy was judged not to be occurring. Isaacs and Haggard (1966) found that therapist attention to patient affective verbalizations resulted in the patient's responses having an increased affective orientation. Braaten (1961) demonstrated that increasing expression of feeling with reference to self as opposed





13


to nonself was related to therapist-rated success in clientcentered therapy. The validity of this variable seems theoretically and intuitively adequate, and its validity will be examined further in the present study by examining its relationship to session outcome.

Significant reliability between raters of interaction material is often used as a measure of the adequacy of data for process research, and in many studies it is sufficient to answer the questions posed. In the present study and for future extension to other aspects of Haley's model, however, it is important that "discrete" mean that with very few exceptions it can be agreed that a particular instance of a chosen behavior has or has not occurred. In order to attempt achieving this type of precision of measurement without losing the relevance of the behaviors in which we are interested, the communication situation must be considered in more detail and then the particular variables to be considered in this study can be defined. An overview of the field of communication science as it relates to psychotherapy research will be used to supply greater detail and establish perspective.


Communication Science: An Overview


Kiesler (1973) in presenting process studies of psychotherapy and proposing a structure for developing such studies suggests Markel's (1969) model of dyadic interaction. The basic communication situation is composed of an originator of a message, the ENCODER, a recipient of a message, the DE-




14


CODER, and a response of an encoder which may be the-stimulus for a decoder, the MESSAGE. COMMUNICATION results when a response of an encoder is received as a stimulus for a decoder. In further specifying the communication situation, Markel identifies six channels of human "face to face" communication by specifying the source in the encoder response. The channels with the source and destination for each are: speech (vocal tract, ear); kinesis (body movement, eye); odor (chemical processes, nose); touch (body surface, skin); observation (body surface, eye); and proxemics (body placement; eye). In psychotherapy research all three of the basic communication "entities" are important, and only three of the channels are studied to any significant extent at present: speech, kinesis, and proxemics. Odor appears to be studied the least, with observation and touch gaining attention. Speech is by far the most studied of all the channels.

Speech is divided in a number of ways which are important in specifying the way this study is going to deal with communication. Morris (1946) divides the science of language into three aspects: SYNTACTIC, formal relations of signs to each other; SEMANTIC, relations between signs and the objects to which the signs are applicable; and PRAGMATIC, the relations between signs and users of signs. This study will consider aspects of each of these areas.

Linguistics is one of the largest areas of communication science. Morris (1946) defines it as the descriptive study of the formal characteristics of language, and classi-




15


fies it under syntactics. Trager (1966) defines it as the scientific study of human language or speech; the study of messages once they are "on the air." Linguistics is composed of three areas: PHONOLOGY: the study of sounds, with units of increasing size called the phoneme, phone, syllable; MORPHOLOGY: the study of shapes (forms) constituted by sequences of sounds, with units of the morpheme, morph, and word; and SEMOLOGY (syntax or grammar): the study of the sense, the meaningful arrangements, from which the symbolizing functions arise, with basic units of the part of speech, constituent, and the sentence.

Psycholinguistics is the study of the relation between

messages on the speech channel and the cognitive or emotional states of human encoders and decoders who send and receive the messages (Markel, 1969). Linguistics studies language sounds, those sounds necessary to the production of meaningful speech, and psycholinguistics includes the study of nonlanguage speech sounds, those whose variation does not change the linguistic meaning of the speech.

This study will consider only one channel, speech, and within that only those aspects studied within linguistics proper. There are two primary reasons for this: 1) the variables to be defined deal with the meanings of language communications and the content of speech is the most important for measuring them; and 2) more highly valid and reliable measures can be made using a single channel and single type of meaning considering the present state of communication measurement methodology.




16


Defining Communication of Affect


A number of studies have utilized measures of verbalization of affect words, though with significant variations in its definition. Levy (1967) studied the "communication of affect" and defined it in terms of the verbalization of words from a list of affect words which he developed from analytic interviews. Moos and Clemes (1967) used "percentage of feeling words" as one dependent variable and defined it by "enumerating examples of words directly relevant to an individual's affective state" (p. 121).

Most other studies using this measure have been investigating the effects of various therapist behaviors on client verbalization of affect within operant conditioning paradigms. Here too the specific definition used has often varied. The most popular definition has also included the stipulation that the statement be a self reference and is accordingly called an "Affective Self Reference." Salzinger and Pisoni (1958, 1960) appear to have made the first formal use of this definition, which specifies that the response class statement must be an "I" or "We" statement describing or evaluating the client's own state, except for intellectual or physiological, and providing some rules for instances where the judgement is unclear. A number of other studies have used this definition (,:)r small modifications of it (Hekmat, 1971a, 1974; Hekmat & Lee, 1970; Hoffnung, 1969). "Affective Self Disclosure" has been used as the response class by Hekmat and Thiess (1971) and Hekmat (1971b). It




17


differs from the Affective Self Reference in using only "I" statements and considering affective expressions only if they have the form: "I feel ......" with an expression of affect. other definitions have included the "Feeling Statement" (Green & Marlatt, 1972) with extensive rules, and two Closely related definitions, the "Emotional Words" measure (Merbaum & Lukens, 1968) and the "Client Affective Word" measure (Barnabei, Cornier & Hye, 1974; Crowley, 1970).l

Variations, then, have included whether or not there

was a self-reference required and whether that included statements beginning with we; whether the affective expression was positive or negative or unscored for direction; whether only certain types of affective statements were considered or certain types excluded; whether both therapists and clients were scored; and whether statements were required to be with reference to present time or not (immediacy). The present study will use a definition modeled after Green and Marlatt's rules (Note 1), but reducing the amount of judgement to be exercised by raters by using a more extensive listing of affect words, and reducing the extent to which evaluation is included as feeling expression. In order to increase the applicability of this study to other studies, statements will be scored both with and without a self referenc-, requi r'I.

The inference is being made that a behavior like client 'Psycho-Dic, a content analysis dictionary, appears to have
the capability for defining this measure in any of these
ways, but no studies using it for this purpose were found.




18


verbalization of affect reflects a definition of the relationship, and the research question becomes, "How does the therapist behave so as to attempt increasing client affective verbalization?" The literature related to increasing client verbalization of affect indicates the particular effedtiveness of at least three strategies. One is that the therapist tells the client directly what kind of relationship he (the therapist) wants; i.e., the therapist instructs the client on how to behave. A second is that the therapist suggests a definition and behavior by modeling the desired client behavior. Third, the therapist utilizes differential verbal reinforcement of the desired behavior to increase its occurrence.

The first strategy is th e use of instructions by the

therapist, and has been studied and found effective in a number of studies (e.g., Green & Marlatt, 1972; Lukens, 1970; Merbaum & Lukens, 1968; and Whalen, 1969). Here the therapist communicates directly to the client that in the relationship it is the client's responsibility to engage in particular behaviors, and perhaps, not in other behaviors. For example, at the beginning of the interview, the therapist might say, "During the interview, you are to talk about your feelings about yourself, and not about the feelings other people have or subjects unrelated to yourself." In that case, the therapist would be making a very explicit definition of the relationship as to what he expected from the client.

The second strategy is for the therapist to be a model




19


for the client of the desired client behavior, and this has also been examined and found effective in increasing client affective verbalization in several studies (e.g., Green & Marlatt, 1972; Marlatt, 1971; Myrick, 1969; Whalen, 1969). In using this strategy, a therapist would be observed to use affective words, or in the case of modeling affective self references, to make statements with reference to the therapist's affect.

Various types of verbal conditioning techniques are the third strategy and the area to be considered here uses agreementn" as a social reinforcer, with specific instances as "mmm-hmn," "good," "wonderful," "yeah," "I see," etc. (e.g., Hekmat, 1971b, 1974; Hekmat & Lee, 1970; Moos & Clemes, 1967; Salzinger & Pisoni, 1960) A therapist using this strategy would be observed to give reinforcing responses after a large proportion of client statements containing affect words or idioms (or affective self references) and after a small proportion of other statements. Effective use of this strategy probably involves indiscriminant and frequent use of the reinforcing response(s) early in the session to establish its reinforcing power (Waskow, 1962), and the discriminant use of such responses subsequently.

Research has indicated that another important factor in considering variables in therapist-client control is the interpersonal need compatibility of the particular therapistclient dyad (Gassner, 1970; Sapoisky, 1960, 1965). The FIRO-B (Fundamental Interpersonal Relations Orientation-




20


Behavior, Schutz, 1966) is most used for assessing this variable, and has been demonstrated to be a reliable and valid indicator of interpersonal compatibility on a number of measures. Of the three interpersonal needs postulated by the FIRO scales, the most important for the present formulation is the need for inclusion and control, and of the various compatibility computations, that of reciprocal need compatibility is most appropriate. In formulating hypotheses about the use and effectiveness of the above therapist strategies the combination of reciprocal compatibility for Inclusion and Control of the therapist-client dyads will be considered.

One study (Mendelsohn & Rankin, 1969) has indicated that sex may also be an important variable. It was found that the compatibility measures work well in some ways with female clients, but poorly with male clients. Sex of therapist did not appear to be an important factor in the differences. Therefore, sex of client will also be considered as a variable in assessing the effects of the compatibility measure.

In summary then, client affective verbalization and client affective self reference are considered important variables, and three therapist strategies have been shown to be effective in increasing the frequency of occurrence for both these variables in quasi-therapeutic situations. Previous studies have apparently not examined the relationship of variables in situations more closely resembling actual therapists who have not been specifically instructed to engage in the strategy behaviors. The theory and literature re-




21


viewed above suggests that therapists may engage in these behaviors in actual interviews, and that when they do, clients' use of affective words increases across the interview.

The interviews will occur in a standardized situation

with each therapist serving as a control for himself or herself, and the literature suggests that two other variables require control and examination as independent variables. These variables are interpersonal need compatibility and client sex, and these can be both independent variables and controlled in other analyses by utilizing a completely balanced design with respect to them. In addition, therapist sex can also be included in the balancing to control for effects of that variable.

Although client affective verbalization and affective self reference were consistently assumed important in the literature reviewed, little evidence was cited which defended this assumption. In order to examine the validity of these variables in terms of session outcome, measures of session outcome will be examined.

Hypotheses can be conceptualized as falling into three groups. The first group deals with the question of whether or not therapists use the strategies as defined, whether or not the expected changes in client affective verbalization behavior occur, and whether or not they occur or are more pronounced when considering the strategies.

The second group of hypotheses deals with the variables of compatibility and client sex both as independent variables by themselves and also as they interact with the stra-




22


tegies and each other. The third group deals with the relationships between client affective verbalization measures and session outcome measures, with consideration given to the possible role of compatibility. The hypotheses are formally stated and labeled below.


Research Hy2otheses


It is hypothesized:

Group 1

1. that each therapist uses one or more of the above strategies;

2. that a) client affective verbalization and b) client affective self references will increase across the interview,

3. that greater use of a strategy is associated with larger

increases in client affective verbalization;

4. that greater use of a strategy is associated with larger

increases in client affective self reference; Group 2

5. that a) higher compatibility will be associated with enhancing the positive effect of the therapist strategy

behaviors, and b) that higher compatibility of the therapist-client dyad will be associated with (1) client affective verbalization and (2) client affective self reference increasing across the session, and with (3) total

client affective verbalization and (4) total client affective self reference;

6. that a) the positive effect of the therariqt strategy








of increasing client affective verbalization and client 23

affective self reference will be enhanced with female clients and not with male clients, and b) that female clients and not male clients will be associated with (1) client affective verbalization and (2) client affective self reference increasing across the session and

with (3) total client affective verbalization and (4)

total client affective self reference. Group 3

7. that a) client affective verbalization and client affective self reference will be positively associated with

session outcome, and b) higher compatibility will enhance

the positive association of client affective verbalization and client affective self reference to session outcome.














CHAPTER 3
TECHNICAL CONSIDERATIONS



Method



Subjects. There were two groups of subjects: 10

currently practicing psychotherapists, 5 male and 5 female, with at least 5 post-degree years of experience each; and 40 clients, 20 male and 20 female from the subject pool in the Psychology Department at the University of Florida.

Apparatus. Apparatus used in the experiment included standard videotape-recording equipment, and cassette transcribing equipment. The same room and seating arrangement were used in all interviews.

Paper and pencil instruments. Paper and Pencil scales used were the FIRO-B (Fundamental Interpersonal Relations Orientation-Behavior, Schutz, 1966), the Problem Pathological Potential Scale (Blumberg, 1968), a therapist postsession evaluation form (see Appendix A; Blumberg, 1969), and a client post-session evaluation form (see Appendix B; Blumberg, 1969).

Design. Through on test scores, client subjects were assigned to therapists such that each therapist had one client in each cell of two levels of compatibility by two levels of client sex. Interviews were considered as


24







being composed of three segments, and the design, therefore, can be viewed as a factorial design: 2 X 2 X 3, compatibility X sex of client X segment of interview (repeated measures). Where inappropriate to apply analysis of variance techniques, the independent variables and the dependent variables were examined by considering their correlational associations and graphs of interaction relationships.

Procedure. The raw data for this study were taken from transcripts of interviews videotaped as part of a study by Goldman and the author (Note 2). Therapists and potential clients were tested on the FIRO-B, and the combined reciprocal compatibilities for Inclusion and Control computed for all pairs. Potential clients had been obtained through their responding to a notice on the Psychology Department experiment bulletin board (see Appendix C), and the therapists were obtained by personal contact with the experimenters. High compatibility was operationally defined as existing in a pair when that pair's score fell within the bottom 5% of that therapist's scores with all potential clients. Low compatibility was defined as scores falling in the top 5%. If insufficient matches were available to fill all four cells [same- and opposite-sex clients and high and low compatibility], additional potential clients were tested until all 40 pairs were obtained. Clients were informed of their selection for the second part of the experiment through the phone message in Appendix D. Generally, each client was informed of his or her selection for the study,




26


and told that their task was to select a real problem that he or she faced and would be willing to discuss with a professional therapist, with the session being videotaped.

Therapists had been instructed that they would be conducting a single interview with each of four college students who had responded to a study entitled "An Analogue Study of Counseling," and who had been instructed by the experimenters to think about and select a real problem that he or she was facing and would be willing to talk about with a professional therapist. The therapists were also informed that the interview would be videotaped and that provisions had been made with the campus mental health services for referral of the client person if the therapist felt it was indicated.

At the time of the interview, the therapist and client completed the paper and pencil rating sheets, including for the client a pre- and post-Problem Pathological Potential Scale and a post-evaluation form, and for the therapist a post-evaluation form.

The interview was introduced by the experimenter operating the videotape equipment, and he or she informed the therapist and client jointly that they would have 50 minutes to talk and that the experimenter would knock on the door of the interview room at the end of 45 and 50 minutes. The interview was videotaped by the experimenter, with the present a(ithor accounting for about 25% of the procedure.

Three minute segments were reaped by the author and

two collaborators onto master video-tapes and cassette audio-




27


taoes. The cassettes were used to make typed transcriptions according to the rules for typists described by Gottschalk, Winget, and Gleser (1969) and were checked for accuracy by the experimenter. The segment locations were determined by taking the first 3 minutes of the interview, and the approximate middles of the middle and final thirds of the interview. Detailed rules for determining the segments are found in Appendix E.


Rating Procedures


The rating process was based on standard content

analysis procedures and terminology (Berelson, 1952). In the case of most variables, it operated within what has been termed the "classical model" (Marsden, 1971), meaning that what is measured is limited to the semantic and syntactic aspects of communication. Content analysis requires the definition of three ideas for the measurement of each variable, and these are 1) the unit, 2) the category, and 3) the indicator (Berelson, 1952). Each of these aspects of the content analysis process will be elaborated as needed for each variable to be measured.

Units. With the exception of the unit "WORD," the unit for rating each variable was scored separately and marked on the transcript. Units utilized were the "WORD," the "PROPOSITION," the "STATEMENT," the "SEGMENT," and the "INTERVIEW." WORD is essentially self explanatory, and includes hyphenated words as one WORD. PROPOSITION is a more complex unit, and was defined according to Lennard and Bernstein (1969)




28


as "a verbalization containing a subject and predicate whether expressed or implied" (p. 57). STATE14ENT was defined as an uninterrupted sequence of PROPOSITIONS by either therapist or client, and has been defined similarly by many researchers (Butler, Rice, & Wagstaff, 1962; Lennard & Bernstein, 1969; and Matarazzo, Saslow, & Matarazzo, 1956). SEGMENT was defined operationally above as a three-minute section of the interaction, and INTERVIEW is the composite of the three SEGDIE14TS of an interview session.

Raters. Raters in the study were volunteers from the clinical psychology introductory class, volunteers from the psychology club in the Department of Psychology, and subjects from the subject pool in the psychology department, all groups at.the University of Florida. All potential raters were screened and local references obtained to ensure that the confidentiality of the materials would be reasonably protected and to determine the motivation and reliability of the raters. Appropriate tasks were assigned for each person.

The experimenter scored the transcripts for the unit PROPOSITION independently of a random sample of 13% of the transcript scored by a separate rater. The rater was trained according to the training manual for scoring units (see Appendix F). Of over a total of 1802 PROPOSITIONS jointly scored, there were disagreements on 3.9%. PROPOSITIONS were indicated on these transcripts by separation with diagonal marks.




29


Affective Verbalization. Affective verbalization was scored utilizing three categories, labeled Primary Affect Word, Secondary Affect Word, and Affect Idiom. Each of these was scored separately in the content analysis procedure, but the same general definition holds for all three types of affective verbalization. Generally, an affective statement is one which is "expressive of a subjective reaction of an emotional tone" (Green & Marlatt, 1972), and in the present use, statements of evaluation without affective meaning, or statements referring to intellectual or physiological states were excluded.

Primary Affect Words were scored using the unit: WORD, the category: Primary Affect Word, and the indicator (rule for assigning a unit to the category): inclusion on the list of Primary Affect Words in Appendix H. This list was generated by selecting words which almost always refer denotatively to affective states. Examples are "angry," "sad," "happy," etc. Words on this list are in adjective form, but a word from this list was scored as a Primary Affect Word when it appeared in any of its various parts of speech forms: noun, verb, adjective, or adverb. The rater's task for this affective expression was to compare each WORD of a transcript with this list and to label any WORDs found by recording the WORD's code number. The rating and verification was handled in the same manner as with the unit rating, and the raters were trained using the manual in Appendix G. Three raters were used and in reliability




30


checks over 769, 353, and 343 PROPOSITIONS respectively, error rates of 0%, .5%, and 0% were obtained.

Secondary Affect Words were scored using the unit:

PROPOSITION, the category: Secondary Affect Word, and the indicator: inclusion on the list of Secondary Affect Words in Appendix I. This list was generated by selecting words which refer denotatively to affective states only when modifying the verb "feel" or the verb "sense." Examples are 11alert," "slow," "inhibited," etc. All part of speech forms of the words listed in Appendix I were scored. The rater's task for this affective expression was to note the subject's use of the verb "feel" or the verb "sense" in a PROPOSITION or adjacent PROPOSITION modifying it, either directly or by its being understood (as in an answer to a question using one of these verbs). When either verb was noted, the rater then used WORD as the unit, and compared each WORD in the verb's modifier with the list of Secondary Affect Words and labeled any WORDS found by recording the WORD's code number. The same rating and verification process was used, and the rater was trained using the manual in Appendix G. Two raters were used for this task and in reliability checks over approximately 350 PROPOSITIONS each, no errors were obtained.

Affective Idioms were scored using the unit: PROPOSITION, the category: Affective Idioms and Slang, and the indicator: inclusion on the list of Affective Idioms in Appendix J. This list was generated by selecting groups of




31


words which idiomatically refer denotatively to affective states. Examples are "sick and tired (of)," "get a bang out of," "hair stands on end," etc. Only part of speech changes from the listed forms which maintain the denotatively affective meaning were also scored. The rater's task for this affective expression was to compare each PROPOSITION of a transcript with list J, and to record the code number of that idiom's PROPOSITION. The same rating and verification process was used as in previous rating, and the rater was trained using the manual in Appendix G. Two raters were used, and in reliability checks of approximately 350 PROPOSITIONS each, error rates of 0% and .1% were obtained.

Self References. Self References were scored using

the unit: PROPOSITION, the category: Self Reference, and the indicator: "unit is a 'verbal response by a subject which expresses, descriptively or emotively, something about the subject in relation to himself, others, or the world'" (Green & Marlatt, 1972). This indicator was further specified by four rules utilized in training the rater, and are in the training manual for this scoring task (See Appendix G). The rater's task was to consider each PROPOSITION in the transcript and to determine the presence of any pronouns. If a pronoun was found, the PROPOSITION was compared to the scoring rules, and if found to be a Self Reference was labeled by recording the PROPOSITION code number(s) involved. Three raters were used to score this variable and reliability checks over 343, 769, and 353 PROPO-




32


SITIONS obtained, error rates of 7.0%, 3.1%, and 4.5% respectively.

Therapist Strategies. The three therapist strategy

behaviors were scored through a similar use of content analysis. Instruction was defined as a therapist statement or statements within the first 3 minute Segment of the interview in which the therapist communicates at least two of three things: 1) behavior the client is to include (i.e., "talk about your feelings"), 2) behavior the client is to exclude (i.e., "Don't talk exclusively about your thoughts"), and 3) examples of the general classes or areas desired to be talked about (i.e., "YOU might talk about your satisfactions and dissatisfactions, your concerns and confidences"). Instruction was scored using the unit: PROPOSITION, the categories: Instruction-1, Instruction-2, Instruction-3, and the indicator: unit includes defined communication. The rater's task on this rating was to consider each PROPOSITION within the transcript for an initial

3 minute Segment, and to label any Instruction categories found by recording that PROPOSITION's code number(s). Since this rating task was somewhat less precise than the earlier tasks, two non-experimenter raters were used, and the experimenter's rating was to be used to settle disagreements. The raters were trained using the training manual in Appendix G. Only one component of this strategy was found by raters, and no reliability assessment would be meaningful. This was extended to all three segments, and there was still only one found.




33


Modeling, the second therapist strategy, was defined earlier as the therapist's use of affective verbalization, both with and without self reference. Since every therapist used some affective words, this variable is a rating of how many affective verbalizations and how many affective self references the therapist made. No ratings in addition to those discussed above were necessary.

Reinforcement was scored by raters using the units: WORD and PROPOSITION, the category: Reinforcer, and the indicator: therapist makes one or two word expression of agreement or approval, and specifically uses only those words or expressions listed in Appendix K. Examples are "mmm-hmm," "yes," "that's nice," etc. The rater's task was to consider all therapist verbalizations, and to label all words or expressions from Appendix K by recording the Statement number of the therapist verbalization. Since the typist is the first "rater" to decide whether or not to place a questionable verbalization in the transcript, the typing by the typist and verification by the experimenter of the transcripts was considered the first stage of a twostage process of rating reinforcement. No assessment of reliability was made at the first stage. The raters were trained using the training manual in Appendix G. Two raters were used, and over approximately 350 PROPOSITIONS each, error rates of .4% and .7% were obtained.

Following the rating procedures, the necessary counts over various summarizing units and contingent conditions were made to obtain the following measures separately for





34


client and therapist: total number of WORDS, PROPOSITIONS, and STATEMENTS per segment; total number of Primary Affect Words, Secondary Affect Words, and Affect Idioms per segment; total number of Primary Affect Words, Secondary Affect Words, and Affect Idioms per segment when each occurred within a Self Referent PROPOSITION; number of different types of Instruction categories per segment; number of therapist Reinforcers preceded by client affective verbalization of any type per segment; number of client affective verbalizations not followed by therapist Reinforcement per segment.

The final measure for affective verbalization for both client and therapist is the sum of the number of Primary Affect Words plus the number of Secondary Affect Words plus the number of Affect Idioms per Segment divided by the total number of PROPOSITIONS in the segment being computed and given that only one affective expression per PROPOSITION has been counted in the sum. This gave a percentage score after being multiplied by 100, and is called the client (or therapist) affective verbalization score for Segment X. The total affective verbalization score for the interview is the sum of all affective expressions in the interview divided by the number of PROPOSITIONS in the interview. The same procedure was used for computing the affective self reference scores.

The final measure for the Instruction strategy is "yes" versus "no" as to use of that strategy according to the definition. The final measure for the Modeling strategy is







the therapist affective verbalization score as computed 3

above. The final measure for the Reinforcement strategy is most easily expressed as a formula which recognizes that Reinforcement might either encourage or discourage affective verbalizations:


Rij = [(rw ro)/(rw + ro)](x/8) (100),


where Rij is the Reinforcement variable score for client i in segment j; rw is the number of times the therapist used a verbal Reinforcer immediately after a client finished a statement which contained an affective verbalization; ro is the number of times the therapist did not Reinforce an affective verbalization; and x is a value equal to the larger of the two r's (rw or ro). (The x/8 factor attempts to correct the Reinforcement score for small numbers of reinforcers, which is an inference from Waskow's 1962 study.) The total Reinforcement score for the interview is the mean of the three Segment scores.













CHAPTER 4
RESULTS



Results are presented in terms of the three groups of hypotheses discussed above and labeled: client verbalization and therapist strategies, compatibility and client sex, and client verbalization and session outcome. When one analysis is used for more than one hypothesis, it is presented to the extent needed to examine the results for that hypothesis. When it is apparent that a result is markedly different from the hypothesis and an alternative hypothesis is evident by virtue of b eing the converse of the predicted relationship, that alternative hypothesis is presented in this section with the analyses used to examine it.


Client Verbalization and Therapist Strategies


The first group of hypotheses (numbers 1 through 4)

dealt with therapist strategies and client affective verbalization and client affective self reference. Each of these variables was measured as described in Chapter 3. With respect to the therapist strategies, no full instances of the Instruction strategy were found, and only one type 3 component of that strategy (see Appendix G) was scored in the rating of all therapist responses. Consequently, this variable was dropped from further analyses. Also, although ther36




37


apist statements were scored for self references as part of the Modeling strategy measurement, only 31 therapist affective self references were found in all, making the inclusion of that variable meaningless, and it was eliminated from further analysis. This means that the strategy Modeling is equivalent to therapist affective verbalization in all analyses. Scores for the other strategies and for the client variables appeared adequate.

Hypothesis 1 predicted that each therapist would use one or more of the strategies, and was informally analyzed since some of the strategies were not dichotomous. Several factors suggest that this hypothesis was not supported. First of all, the Instruction strategy was clearly absent, as was the use of modeling of affective self reference behavior. Second, the frequency of modeling of affective verbalization behavior was at or below client frequencies in many cases and was generally low for therapists, inconsistent with our expectation of a Modeling strategy. Third, the Reinforcement behavior of therapists often did not involve a discriminant use of reinforces, but rather relatively high rates of reinforcing all client verbalization within a particular segment. Overall, therapists did not appear to use the defined strategy behaviors systematically, and each therapist clearly did not use at least one strategy as there were some therapists who did not approach using any of the strategies.

Hypothesis 2 predicted that both client affective verbalization and client affective self reference would increase across the session. This was analyzed by checking for a




38


significant segment effect in an analysis of variance. This was part of a larger analysis presented in the Group

2 hypotheses. For both client affective verbalization (see Table 1) and for client affective self reference (see Table 2), there was no significant segment effect, and in both cases, the trend was for the first segment to be the highest of the three segments and for the second segment to be lowest.

Hypotheses 3 and 4 predicted that there would be a significant positive relationship between the greater use of a therapist strategy and larger increases in client affective verbalization and client affective self reference. Since the therapist strategy variables occur within the same interactions which also provide the data for the client measures, statistics presuming independence were inappropriate. Consequently the associations predicted to occur in the data were analyzed by examining the correlations in the data. To determine the significance of a correlation for interpretation purposes, the correlation was compared to the null hypothesis that the correlation equals zero, with alpha = .05 for rejecting the null hypothesis. A correlation was defined as approaching significance if its significance level fell below .09. Unless otherwise noted correlations are Pearson r's. Spearman correlation coefficients were used in instances where comparisons were to be made which involved class variables (i.e., compatibility, sex of client, and sex of therapist).

As indicated above, some strategies did not occur in





39






U) co U) U) U) m m
04 Z z z z z z z



0 rn rn t cl r- m CD
H (-) r-4 v -1 CD M r
54 P r- (N 'IT LO r- ICT CN




cc -A m 110 rl) r- co r- Ln
cl C) 10 rl) cc Ln c) rn rD r- m (n CN r- 'IT CN m





44 C- N N N C*
m co r- r-I
r-A

In N M r-I M co 1-1 %lo Ln r- T all
rZ4 r4 r- m CD .D CD C) tlo -4 -1 r- CN r0 ce Ln r- c rn Ln cD Ln o, 4 co N Ln 0
c, 4 1 1 1 4 C r 1 :D 10 r-4 Ln r- 1-4 Ln en
cl (N N
fc x


E-1 4



U) >
r-q x V
u
E-4 tP
44

W P4 4J

0
K x f

x
0 In

z FA EA
a4 w "
D --I p N :
H 0 74 H


X :4 ro
u P
w
::D -P
M F--4 F-) H m r
t:) 0 x 4 : m _q x 4 a)
Ln H W H D H w H x 10
mmmmm mmm
u 74 H p p H
1:4 W E-1 P E-4 1"' P P P Q, Ln
:D < ,7, r H KC :, Kr W Q) C)
0 al W al P4 W a4
P n :T: P Z H ',s, u Pi
wo -qon"Wo -40W 0 A
muuum tmuuuw p 04
4J
,T Ln %,o r- m m c) -i CN 0 En
r-4 -4 r-q z z





40





U) U) Lo U) U) a4 U)
co Z Z z z z z Z F:4 Z



0 Ln tlo o -:31 CD 10 CD co r- "
H Ln m m rn (10 110 M Ln M CD
Ln (N Ln Ln -r (7) CD IqT
414 1


C4 rn -1 rn C) 110 Ln M co Ln 1-1 r- rz F: Ln m I'D N %10 Ln .10 co m -q CD M -4
F:4 D T C14 r- mT 110 ill Ln r- m -rr m kD
rLi CY
ni U) U C
rq CN f-A
0
.14
V
44 01) H ko CD c"I (N N m 0 r-I r-A r-i
ro m m co ro
4

:74 U) Ln (n r-I rn co Ln co C) Ln CD Ln 4 r- rD4 W IZT Ln M 10 cn a) (1) C) m U) H C) m r-A
c P cli m Ln m rn r- A mzjl >1 m T m %lo
F: -1-3 41 .
:F: D 1 (; 14 C r C 14 Oll rq f: m
Iq x D 0 01% N ko N CD -4 1-4 ft N r-q
ul m -I -i CN CN T -rq U
Q rA
-4 44
>I
cq P a4 0)
0 In
4 Q) C)
U) a) >
E-i r-A 4J -4
CN x 0444 4J

I1:4
> U) Ln 4-4
0 4-) 44
4 0 H 0 fl CL4
0 In 0 (1) I: 1-4 f:4
m W a4 44 4-4 4-)
H n 4-4
E-4 0 m 41 -4 C)
IZ4 P 41 -4
a4 0 (n X w
:Ei x x Q) Q) >4 >4
0 W :4 E-4 (n P E-4
U) H H (n C)
P -q
EA U) Z E-4 E-4 rx4 H F-I
:4 a4 W q) -- CQ m A
w :D E-i W rq F-i H r-4
F-I 0 U z F-4 r-i -4 C) E-4 El Q Qj
w w _q U) 11 fc: 4 fo
:F: u p U a4 P4 H
U u x > 4 U)
E-1 x z m x W x w 0 0 (15 z
u u u >
W E-4 X E-4 :D E-i EA Co-, 4-)
F-4 H F-4 t-D H F-4 ul F-I H
:D 0 x 0 1-- rn 0 x 0 -l x x
U) H w H D H W H X w w
w rQ m cl U) En M (n m CQ ro U) to
u ;-, E-4 E-4 H H E-4 H H
W E-i P P U 7, Fq p p Z &4 P Eq E-A 104
w z K w W 4 z K w KC F 4 :7, Z Q) Lr)
0 3. CL4 W P4 F-) rL4 W 134 124 04 W W CD
Lo P el H V M X* H -1 E-4 ; X H H
w 0 '1 0 D FA W 0 '1 0 w 0 0 0 0 -l
muuum 3:muuum uuuci
a,
V
M "o r- m M CD r-4 -4 N M 'tT 0
-1 -4 r-4 z




41


this sample, and this leaves Modeling and Reinforcement to be considered. The analyses of hypotheses 3 and 4 for Modeling will be presented first.

Since these hypotheses are dealing with changes in client behaviors, the Modeling score was correlated with scores representing the change of a client's behavior. Accordingly, correlations were computed between the various segment levels of therapist affective verbalization and the change scores for the two client measures across various parts of the interview (see Table 3). For client affective verbalization, one correlation was significant, this between therapist affective verbalization in the first segment and the change in client affective verbalization from segment 1 to 2, but in the opposite to predicted direction (r = -.44, P = .004). The correlation between therapist affective verbalization in the first segment and client affective verbalization change from the first segment to the last approached significance, again in the opposite to predicted direction, (r = -.29, p = .067). Therapist affective verbalization correlations with client affective self reference change scores showed the same phenomenon with both correlations significant (r = -.46, p = .003; r = -.41, p = .009). A scatter-plot of the segment 1 therapist affective verbalization by segment 1 to 2 client affective verbalization change scores shows a clear association between higher therapist scores and negative client change scores and between lower therapist scores and positive changes (see Figure 1). This suggests the need for a clearer under-




42







TABLE 3
CORRELATIONS OF MODELING AND CHANGE SCORES


MODELING: THERAPIST AFFECTIVE VERBALIZATION BY SEGMENT SEGMENT
CHANGE
12 3 Total


CLIENT AFFECTIVE VERBALIZATION


1 to 2 -.44** .04 -.01 -.13

2 to 3 .11 -.11 .07 .01

1 to 3 -.29 -.08 .06 -.11


CLIENT AFFECTIVE SELF REFERENCE


1 to 2 -.46** .08 -.08 -.14

2 to 3 .04 -.17 .01 -.08

1 to 3 -.41** -.11 -.06 -.23


** .01




43


+10








0
HCN
H


C A
>
>: A B
H
uo A A A

0 A A A A

E-4 C A A

A A B B C A A
1)
B A B

A A

-5 A A



0 5 10 15

THERAPIST AFFECTIVE VERBALIZATION IN SEGMENT 1


Legend: A = 1 observation; B = 2 observations, etc.




SCATTER PLOT OF A THERAPIST SCORE BY A CLIENT SCORE FIGURE 1




44


standing of the nature of the affective verbalization scores and their relationships.

Intercorrelations were run between all client and therapist affective expression scores, a total of 12 variables, and many highly significant correlations were present. (See Table 4 for specifics of the following discussion of effects.) Each segment level of client affective verbalization is highly correlated with its respective self reference variable Ws = .95, .93, .85). The affective verbalization levels are not, however, correlated with each other. Of the self reference levels, only segment 1 and 2 are correlated, but only modestly (r = .32). Therapist affective verbalization in segment 1 is strongly correlated with both client measures in segment 1 (r = .66 and .67) and with no other segment levels of either. Therapist affective verbalization in segment 3 is modestly correlated with all levels of client affective self reference only (r's = .34, .32, .40, .38), as well as being correlated with therapist affective verbalization in segments 1 and 2, (r's = .58, .34), which are not correlated with each other. The total scores for the therapist measure and both client measures are modestly to highly correlated with each other and all affective verbalization segment variables (range of r's = .32 to .92), with the exceptions of total client affective verbalization with therapist affective verbalization in segment 2 (r =.28, p = .076) and the total therapist measure with client affective self reference in segment 3 (r = .27, p = .089), both of which correlations approach significance.





45














FC
E-4












E-A




F:





E-4 Ln r-4
cq C 0*) C)
p 0 kD C)

H P4

Ln
C\j C\J
P D C"!


w 0

E-1 r-q
W F:t Ln c) C) C) CTN C) C\j C)
r- CD 110 CD Ln c) a) C
>

E-4
u ry. >
14 0
Ln
rT4 U co C) r- CC) tD U) Ln C)
r14 Ln cD 1-4 CN C) m 00 CD IT CD
KC >



r-4 Ln
cq r- C) -V 'IT ff) C-) m 0) C)
CN (D co Ln c) r-A m C) 'IT Ln c)




co r-i r-q Ln
C) CD r- co Ln (D U-) C) co 00 m %10 m C)
rq -A cq r- (D M








rq

>
F: KC
U) z u u u






46







m C)
00 CD 4




C) (D
110 C) 4
Q)

0

a) U) 00 C) m C1 Q)
Ln C) r- C)




00 C) N TT cq CD Ln C
fn C) C) U) C)


-4 4J
44
co ":I, C\j m 44 M
m C -q Ln r- 00 a) .11
r-4 CN --i Ln rl,) CD 0
u


0 11
CD 01) Ln C\l
r-A un m CN N CD rn CI

Q) m

4
Ln r-I 0 Q)
r- C) T3, CO r-4 Ln C) u 'C
t1c CD cq 1-1 co C Ln C J 0) 0
Q) 0
.1'- 04
V >1 0
C: -A 4
m C) cc r- N C) (Y) CD -1 -1 (d 44 r-i
Ln C) C) Ln C) 110 1-4 N Q) m
4 J -1 (24 Q
-1 4
(d LH Q)
Q r-i >
4 Q)
CN 0) V Q) (n Q)
CN f--4 C) Ln m C) m C) > >

Q) ro 0) > 4J
Ic: > --4 0
t" rA 4J q)
Ln m ko ko 0 -P U 4A
m co, C) (N TT Ln clq >i 0 Q) 4-4
C) Ul m C) ry) C) 4 4J a) 4-4
0) -rq LH 44 >
r-4 44 F:4 4J 4

"i F: M a)
Q 4J -H r-4 Cq M .0
fy 4J r 04 r
ko CD -A m I:T CN rn CD _Q F-I Q) rq -P 4J -W -4
I'D CD CN r-4 ro C> Ln c) 4 0 Q) 4 -: r
Q) 4 -4 W Q) C) Q) -i
04 04 -4 U &' rj a F, M
la4 p Uri 1:3) U) 41
it 0 0 w 0
it 11 U) En U) p Q) 4J P
> U) > Ln
4 F: K4 CD
u u p r-4 C-4 m
C14 rl) v

> > > 4-)
F:c - P E-4 1




47


The lack of correlation within segment levels of each of the client measures is consistent with the lack of a segment effect in the earlier analyses, and the high correlations between so many variables suggests an alternative hypothesis. This is that clients are influencing the therapists' frequencies of use of affective verbalizations, or equivalently, that in at least segment 1, therapists are using a following style which results in a high correlation between client and therapist affective verbalization scores. Two analyses were completed to examine these alternative hypotheses: cross-lagged panel correlations and tabulating therapist and client use of the same affective expressions.

Cross-lagged panel correlation (Campbell & Stanley,

1963) would predict that the "causal" force (if other than a separate underlying factor) in changing affective verbalization will have a higher correlation between its earlier measurement and the "effect" factor's later measurement than the correlation between the effect's earlier measurement and the cause's later measurement. In all three cases to consider given three time periods (segments 1 to 2, 2 to 3, and 1 to 3), the correlation between the earlier client variable and later therapist variable is larger than the correlation between the earlier therapist variable and the later client variable (see Table 5 of cross-lagged panel correlations).

To attempt further clarification of the notion that

therapists might be "following" client affective verbalization, ratings from a sample of 12 dyads (10% of the total segments) were examined to see what words the therapist in





48






TABLE 5
SUMMARY OF CROSS-LAGGED PANEL CORRELATIONS


CLIENT OR TIME LAG EXPRESSED
THERAPIST a AS SEGMENT DIFFERENCE
MEASURE AS _ _ _ _ _ _ _ _ _ _
VCAUSE~b
SEGMENT SEGMENT SEGMENT
IlAND 2 2 AND 3 1lAND 3


CLIENT AFFECTIVE VERBALIZATION


THERAPIST .09 .09 .24

CLIENT .21 .32* .34*


CLIENT AFFECTIVE SELF REFERENCE


THERAPIST .10 .11 .17

CLIENT .24 .27 .38*


Note: No correlation pairs are significantly different
using p b "Cause" indicates that the measure for that person is
for the earlier of the two time periods.

.--0




49


each pair had used for affective verbalization scores and then to determine whether the client had used that word, and if so whether before or after the therapist response and whether immediately or not. Immediately is defined here as being within the first statement following or preceding the one containing the affective verbalization.

The sample was chosen randomly, with the stipulation that each therapist must be chosen once and that there be four dyads chosen from each segment. Table 6 summarizes the results of the tabulation. It can be seen from the table that although the total number of responses in each segment does not vary greatly, the distribution of the responses in terms of their relationship to client responses using the same word or phrase is quite different. In segment 1, therapists could be described as "following" the client response. In segment 2, clients could be described as "following" the therapists. In segment 3, there is no apparent relationship between the particular affective words and phrases used by the therapists and clients. The sample is of course small, and should be interpreted conservatively, but suggests a trend toward differential leading and following therapist styles in different segments.

overall, it must certainly be concluded that the data clearly disconfirm the third and fourth hypotheses, with respect to the Modeling strategy.

Reinforcement is the second strategy to be considered

in hypotheses 3 and 4. Here, correlations were obtained between segment levels of the Reinforcement variable and cli-





50







TABLE 6
TABULATION4 OF FOLLOWING VERSUS
LEADING THERAPIST RESPONSES


S EGIMENT

TYPE OF
RESPONSE 1 2 3


IMMEDIATELYa AFTER 0 0 0
CLIENT

DELAYEDb AFTER 5 0 0
CLIENT

IMMEDIATELY BEFORE 0 4 0
CLIENT

DELAYED BEFORE 0 2 1
CLIENT

NO SAME CLIENT 0 2 5
RESPONSE


TOTAL 5 8 6


Note: Responses are therapist affective verbalizations,
and tabulations reflect the type of relationship to
client's use of the same word or phrase within a given
Segment of an interview.

a "Immediately" is defined as the statement right before
or right after the statement containing the affective
expression.
b "Delayed" is defined as any position within the same
Segment other than "immediately" before or after.




51


ent change scores. Modest, though significant, negative correlations were observed between some variable pairs (see Table 7). Reinforcement in the first segment was negatively correlated with client affective verbalization change from the second to the third segment and from the first to the third segments Wrs = -.35 and -.44), suggesting that greater and more discriminating use of verbal reinforcement of client affective verbalization in the first segment was related to decreases in that client measure from segment 1 to 3 and from 2 to 3. This was also true for client affective self reference Wrs = -.53, -.46). Second segment Reinforcement was negatively correlated with change scores for both client measures (affective verbalization and self reference) from segment 2 to 3,(r's = -.32 and -.51) Total Reinforcement was also negatively correlated to both measures from segments 1 to 3 (respectively r's = -.31, -.35) and to affective self reference from segments 2 to 3 (r = .50).

To examine the possible effects of simple number of reinforcers or quantity of verbalization, these variables were also correlated with the change scores (see Table 8). For client affective verbalization, number of reinforcers followed the same pattern as the Reinforcement variable, though at a lower correlation coefficient level such that only reinforcers in the first segment with client change from segments 1 to 3 and total reinforcers with change from 1 to 3 were significant negative correlations (r's = -.33, -.33). Client affective self reference showed exactly the same pattern with the number of reinforcers as with the Reinforce-




52







TABLE 7
CORRELATIONS OF REINFORCEMENT STRATEGY
AND CLIENT CHANGE SCORES


THERAPIST REINFORCEMENT STRATEGY BY SEGMENT CHANGE
BY
SEGMENTS 1 2 3 TOTAL


CLIENT AFFECTIVE VERBALIZATION


1 to 2 -.04 .12 -.08 .01

2 to 3 -.35* -.32* .16 -.29

1 to 3 -.44** -.25 .10 -.31"


CLIENT AFFECTIVE SELF REFERENCE


1 to 2 .12 .28 .02 .20

2 to 3 -.53** -.51** .05 -.50**

1 to 3 -.46** -.29 .07 -.35*


* P--.05
< .01





53






TABLE 8
CORRELATIONS BETWEEN NUMBER OF
REINFORCEMENT RESPONSES AND CLIENT CHANGE SCORES


NUMBER OF THE-RAPIST REINFORCEMENTS BY SEGMENT
SEGMENT
CHANGE
1 2 3 TOTAL


CLIENT AFFECTIVE VERBALIZATION


1 to 2 -.01 -.03 -.11 -.05

2 to 3 -.28 -.20 -.13 -.25

1 to 3 -.33 -.26 -.26 -.33*


CLIENT AFFECTIVE SELF REFERENCE


1 to 2 .11 .11 .08 .12

2 to 3 -.45** -.34* -.28 -.43**

1 to 3 -.38* -.26 -.22 -.34*


* .05

**---.01




54


ment variable, though with somewhat lower correlation coefficients.

An important incidental finding of these analyses was that, with the exception of Reinforcement in the third segment, all the Reinforcement variables are correlated with all segment levels of number of reinforces, number of statements in a segment, and number of therapist PROPOSITIONS in a segment (the closest measure available to quantity of verbal output). (See Table 9 on correlations of raw scores and Reinforcement measures.)

Overall, the data did not support the third and fourth hypotheses with respect to the Reinforcement strategy, and suggested an opposite to predicted trend.

The analyses of the Group 1 hypotheses, then, have

generally disconfirmed hypotheses 1 through 4, with support being demonstrated for an alternative hypothesis regarding the direction of influence between therapist and client.


Compatibility and Client Sex


The primary focus of the hypotheses in Group 2 is hypotheses 5a and 6a predicting interaction effects of compatibility and client sex with therapist strategies and client affective expression change scores. The examination of these hypotheses is in some ways a moot point since as stated, these hypotheses relied on supporting hypotheses 2, 3, and 4 above. In order to assess the general relationships predicted by these hypotheses, however, plots of the variables' relationships of interest were made and examined




55










TABLE 9
CORRELATIONS BETWEEN NUMBER OF
REINFORCEMENT RESPONSES AND REINFORCEMENT STRATEGY


NUMBER OF
THERAPIST THERAPIST REINFORCEMENT STRATEGY
REINFORCEMENTS BY SEGMENT
BY SEGMENT

1 2 3 TOTAL


1 .54** .53** .10 .57**

2 .49** .69** .28 .67**

3 .43** .53** .36* .59**

TOTAL .58** .68** .28 .71**


* <.05

** ~p<.01





56


for effects. (For examples, see Appendix L.) Generally, it was observed that the plots of mean scores for various variable groups suggested support for many complex interactions, but of the ones of those which could be tested, only the one interaction from the analyses of variance reported below was significant. The variability within each cell of the data for any of these plots is quite large, and these data must be approached very cautiously, and with the assumption that all of the effects are statistically non-significant.

Interpretation is further complicated by what appears to be a lack of independence of Modeling, Reinforcement, compatibility, client sex, and level of client affective verbalization, with an underlying factor of what appears to be a high-low dimension on level of client affective verbalization in segment 1. In other words, female clients tended to be higher.than males overall, especially when in the high compatible group. When high Modeling dyad clients or high Reinforcement dyad clients are separated from the total sample, they also appear more likely to be these high compatible female clients than any others. This implies that the creation of class variables from the continuous strategy measures (which was done to create plots) is questionable since it is likely to create groups which are similar in unknown ways with respect to their relationship with the dependent measures, while their class labels would suggest we are dealing with different variables. In other words, splitting the sample on the basis of the Modeling strategy, for




57


example, means that therapist affective verbalization scores, determine the groups. These groups, however, also split unevenly with respect to client affective verbalization scores. If the scores were determined independently, then we could interpret that as a significant effect. In the present case, it only means there are effects of several variables contained in any examination of the variable "Modeling."

Examination of effects of compatibility and client sex

without reference to therapist strategy could be appropriately handled by analysis of variance. The normalities of the distributions of client affective verbalization and client affective self reference were plotted and a square root transformation for scores including low scores was applied to analysis of variance data as suggested by Kirk (1968). (For examples, see Appendix M.) The transformation applied was


xv (X)1/2 + (x + 1)1/2.


Four analyses of variance were computed. The first two were split-plot factorial designs with two between block treatments (compatibility and client sex) and one within block treatment (segment or repeated measurements), using client affective verbalization as the dependent measure in analysis of variance 1 and client affective self reference in analysis of variance 2. The second two analyses were

2 X 2 factorials (compatibility and client sex), using client total affective verbalization as the dependent measure in analysis of variance 3 and client total affective




58


self reference in analysis of variance 4.

Hypothesis 5b(l) was disconfirmed by the non-significant compatibility X segment effect in analysis of variance 1 (see Table 1), indicating that compatibility alone did not account for changes in client affective verbalization over the session. Hypothesis 5b(2) predicted this same relationship for client affective self reference and was also disconfirmed in analysis of variance 2 (see Table 2). Hypothesis 5b(3) predicted that compatibility would show a significant effect for client total affective verbalization, and hypothesis 5b(4) made this prediction for client total affective self reference. Neither hypothesis was supported as indicated by analysis of variance 3 in the former case (see Table 10) and by analysis of variance 4 in the latter case (see Table 11). This effect did approach significance, however, in this last analysis, suggesting that compatibility alone has some influence on the total frequency of clients' use of affective self reference.

The analysis of these hypotheses for client sex was parallel to the analyses for compatibility. Hypothesis 6b(l), predicting that female clients but not male clients would increase across the session, was analyzed by the client sex X segment effect in analysis of variance 1 for client affective verbalization (see Table 1), and 6b(2), in analysis of variance 2 for client affective self reference (see Table 2). Neither hypothesis was confirmed. Hypotheses 6b(3) and 6b(4) predicted that female clients would be higher than males in total affective verbalization





59





U) U) m W z z z




H C) C m
r4 m 1 41 co




cq r-q C) m C) 0*) %D CD m Ln
m CN r-A kD r--A r-I


44 A .o m
0
.rl
4J
'A C) C) m
C? m CD 110
0 Z klo C)
D Cl -A r-A ro r- 4
C14
>

C) >4

41
X

F: U) E-4 44



> 4J
0 E-4 0
:4 F: E-4
F a4
x 4J
0




U)
E-4
z



4
x
>4 >1 E-A




E-4 P 04 Ln
Kil ;,; K
0 CL4 W 124 0
U)
0 0 0 z 0 A


r-4 CII mj* m 0
:71 z





60





a4 W P4 ll: z F:4



0
H 0) r- C)
44 p I'D -i
r4


(N Ln C) Ln c) 4 rC) C) r- ol W CY
C r C4 L



4-4 r-A rA .o m
rn M

4
Q)
N Ln CD Ln 4-4
fZ4 W m C:) zv co CC)
0 all, 0 CD r C) co
N q m Ln r- 44
D CY N r-i r-q Ln -4
K4 ;7. U)
E-4 W
H

4J
w x
D >4 44
44
E-4


>
0 p 0
04
:E: 4J
0


x
M




ro
4
x

E-1 EA 4-)
-q x -l


E-4 E-4 E-4 tir) (D
:D F: Z IC Z (1) CD 1-4
0 a4 W a4 0
:E: H z 04 000% 0 A V
u u u W p

r-q N rn Ln 0 PL4
z 4




61


and affective self reference respectively. These hypotheses were tested in analyses of variance 3 and 4 respectively, and were not supported (see Tables 10 and 11).

Though not stated as a specific hypothesis, the interaction effect of compatibility X client sex X segment for client affective self reference was significant (see Table 2), though it was not significant for client affective verbalization (see Table 1). Simple main effects tests on the significant interaction (see Table 2) indicated that compatibility was a significant effect only when considering female clients. Sex of client approached being a significant effect when considering clients in highly compatible dyads only.


Client Verbalization and Session Outcome


The third group of hypotheses deals with examining

the relationship between the client verbalization measures and the measures of session outcome. Hypothesis 7a predicted that for both client affective verbalization and client affective self reference there would be a significant correlation with client evaluation, therapist evaluation, and changes in pre- to post-session change scores on the problem rating scale. Of these only two correlations were significant, and they partially confirmed the prediction. Total client affective verbalization and this verbalization in the second segment were correlated with therapist evaluation (r = .34, P = .031; r = .37, p = .020). The relative independence of client affective verbalization in seg-





62


ment 2 from the multiple variables correlating with it in segment 1 makes this finding particularly interesting, since it raises the possibility that therapists do assess the helpfulness of an interview in terms of client affective expression. This is only in the context of other variables, which were not identified in this study, as the variance accounted for in those correlations is less that 14%.

Part b of hypothesis 7, predicting an improvement in the above correlation when considering compatibility, was examined by computing the correlation of client affective verbalization total and for segment 2 with therapist evaluation separately for each level of compatibility. This hypothesis was supported as the correlations for high compatible dyad clients are r = .697 and .594, while in the low compatible clients, these correlations are .413 and .221 respectively. Both differences are significant at the .05 level using a one-tailed test.


Exploratory Analyses


The significance of compatibility and client sex has been supported in this study, and additional exploratory analysis was undertaken. One aspect of this was to explore further possible associations in the data between compatibility, sex of client, and sex of therapist. This was done through the computation of Spearman correlations with all affective verbalization measures, the Reinforcement measures, affective verbalization change scores, and the three outcome measures. Two modest correlations were significant, between




63


sex of therapist and client affective self reference in segment 1 (r = .32) and Reinforcement in segment 1 (r -.40). The negative direction of the correlation indicated that the association was between female therapist and higher levels of client affective self reference in the second segment (or male therapist and lower Reinforcement measures). Sex of therapist approached significance in the same direction with client affective verbalization in segment 1 (r = -.27, p = .088), total client affective self reference (r = -.28, P = .083), and total Reinforcement (r = -.31, P = .053). Sex of therapist approached significance in the opposite direction on client affective self reference change from first to last segment (r = .30, P = .057), indicating that male therapists were somewhat more associated with higher overall change in client affective self reference scores than female therapists. Sex of client approached correlation with client first segment affective verbalization and self reference (r = -.28, P = .078; r = -.29, p = .069), with female clients tending to be associated with higher scores.

The second exploratory analysis was to examine the components of the compatibility score. Since the interview sample includes only individuals in certain matched pairs, the significance of any correlations are very suspect for interpretation, but may suggest directions for further exploration of the compatibility dimension. All basic compatibility measures were computed and were correlated with all affective verbalization measures including change scores, Reinforcement measures, and outcome measures. Several cor-




64


relations were significant, though no particular patterns were apparent. All compatibility variables had at least one significant correlation except for reciprocal compatibility on Inclusion and Affection and interchange compatibility on Inclusion. All significant correlations with affective verbalization measures were for second segment measures and with originator compatibilities. originator compatibilities for inclusion and for affection were negatively correlated with therapist affective verbalization Ws = -.35, -.34) and originator compatibility for control was correlated with client affective self reference (r = .34). Reciprocal compatibility for control was correlated with client affective verbalization change from segment 2 to 3 (r = -.37), and interchange for affection is correlated negatively with client affective self reference change from segments 2 to 3 and from 1 to 3 Ws = -.36, -.32). Interchange for affection is also correlated positively with Reinforcement in segment 1, 2, and total (r's = .38, .34, and .37). originator compatibility for control is correlated positively with total Reinforcement (r = .32). Interchange for control is correlated positively with the Problem Pathological Potential Scale change score (r = .31).













CHAPTER 5
DISCUSSION



The primary focus of this study was on defining client affective verbalization and client affective self reference, on defining therapist strategies which previous research has indicated are effective in increasing the above client measures across an analogue interview, and on examining actual therapists in experimental interviews with respect to their use of these strategies and the effects resulting from the use of these strategies. Results of the study indicate that the defining functions of the study appeared adequate, but that therapists in this sample do not use the defined stratagies, in any systematic form at least, and that the client measures do not change consistently across the interviews.

Although there was a lack of significant changes in client scores across the interview, there was a trend toward changes being important in accounting for the variance in the sample. The variation in segment changes was large within groups of the treatment blocks and suggests that important variables were not identified.

Not only did the results not support therapists as

utilizing the predicted strategies, but there was also evidence suggesting that clients are more in control of the amount of affective verbalization than are therapists. This 65




66


was most marked in the first segment where therapists used a following style, with a trend toward therapists assuming a more directive role in the second segment. Additional investigation of this phenomenon might prove helpful in clarifying these issues.

It is possible that in this sample a ceiling effect occurred where clients' affective verbalization rates were so high from the beginning of the interview that it did not necessitate therapists using strategies to increase that frequency. The data indicates that the rates were substantial, and that therapist interventions of higher modeling or reinforcement behaviors could only have inhibited client behaviors. Well controlled and highly specific strategies, as used in previous research with confederate therapists, might increase rates even in this sample of clients, but such an increase is unlikely when therapists use relatively unspecific strategies and when they are possibly content with the level of client affective verbalization as emitted. The client sample in this study was similar to that used in previous research (college students), but given the differences in inclinations of the therapists not to work toward increasing reasonably high rates of verbalization, a less articulate population of clients may have elicited different strategies from these same therapists. The relationship between this study and previous research on these therapist strategies might be clarified by another study utilizing such a sample of clients or perhaps more easily, by utilizing confederate therapists in the research format of the present study.





67


Analyses of the group 2 hypotheses resulted in one significant effect, the interaction of compatibility, client sex, and segment, indicating that in terms of segment changes, the compatibility manipulation was effective only for female clients. This is consistent with the literature cited earlier which also demonstrated this effect (Mendelsohn & Rankin, 1969).

Correlations between therapist evaluation and client

affective verbalization supported the contention that therapists believe client affective verbalization is important to treatment outcome. Other outcome measures, client evaluation and problem rating scale change scores, were not related to client affective expression measures. Limited ranges of scores for these two outcome measures were a problem and more adequate measures should be used in future research.

Various analyses and correlations offer some help in

the problems of designing further studies to investigate the implications of the present data. The large error variance found so often in the analyses suggests that 1) this must be expected and only very powerful effects can be examined, 2) the flexibility of the interview situation must be restricted in order to provide better control, or 3) a broader range of clients should be studied. The other alternative is to consider more variables, but this is probably prohibitive on the one hand in terms of investment of resources or on the other, in terms of an N that is too small for so many variables.

There is a danger inherent in increasing control in the




68


present type of study. On the well-controlled end of the spectrum, there are the studies which generated the literature used to define the therapist strategies and the dependent variables. The dependent variables seemed adequate, but the "therapist strategies" were apparently foreign to actual therapists in a similar situation! On the other hand, the present study was, by design, loose with respect to these variables, and consequently is very limited in its richness for interpretation or generalizability, vis-a-vis theoretical precision. A compromise between these two issues might be a study similar to the present one, with more attention to the construct validity of the variables and some assurance, by manipulation of therapist or client, that the issue of strategies to increase client affective verbalization is clearly defined in the data.

The present data are also helpful in demonstrating associations between affective verbalization and variables such as client sex, therapist sex, compatibility, quantity of verbalization of all types, inconsistency of change across standard segments, and other affective verbalization measures. Each of these must either be controlled or measured as a variable in subsequent research if it is to result in more significant, interpretable findings.

This appears particularly true of compatibility, since not only did the compatibility manipulation seem reasonably sound (though not significant overall), but the calculated compatibility components also received some support as potentially related to the other variables of the study. Fur-




69


ther investigation of this would require either separate samples matched on each type of compatibility component, or an adequately-sized group constructed randomly without regard to compatibility, and then analyzed in terms of compatibility measures. Economy for investigating more than one type of compatibility probably requires the latter approach.


Conclusion


This study grew out of a review of psychotherapy literature as conceptualized within a general communications model. The model was translated into the terms of research which has considered therapist behaviors and their relationships to clients' use of affective language. A study was designed for testing these relationships in a sample of clients similar to these previous studies, but using a sample of therapists which were different in not being confederates and also in being practicing, professional psychotherapists. This study was also different in including the variables compatibility and client sex as independent variables.

This study's contributions appear to lie in two areas. The first area is in its development of methodological tools for the relatively objective measurement of client and therapist use of words and phrases denoting affective experience both with and without reference to the self. Previous research which has provided similar tools has either maintained significantly subjective elements, failed to retain materials for use on other studies, or failed to adapt materials to




70


the specific purpose of measuring the use of such language by clients and therapists.

The second area of contribution is in applying the

above methodology to the examination of a group of variables in a sample of analogue psychotherapy interviews. The variables considered had been found in previous research to have certain relationships to each other, but these relationships were generally absent in this study. Two significant relationships were demonstrated. Compatibility was found to be effective under some conditions for female clients, but not for male clients. Client affective verbalization in the second segment and for the total interview were found to be significantly correlated with therapist evaluation, which provided modest support for the validity of the client affective measure as an indicator of some important processes in the interview. Therapist strategy variables were either found not to be utilized by therapists in the present sample, or to result in inhibitory effects, with clients more in apparent control of the frequency of affective verbalization in the session.

Suggestions for future research are made, and argue for the use of more control than in the present study. Caution is also made, however, for such strict control that the research looses its relevance to applied practice.



































APPENDICES








APPENDIX A
THERAPIST POST-SESSION EVALUATION FORM




Name:



Please answer the following questions as candidly as you can. Feel free to make additional comments where you wish to. Circle one of the alternatives.



1. Did you like the person whom you interviewed?

very much slightly not at all

2. If this interview had occurred in the context of your
clinical practice, would you:

want to refer the client unsure continue to see client 3. Do you think this experience was valuable for the client?

very much slightly not at all

4. Do you think your client is now better able to solve or
live with his or her problem?

yes unsure no

Additional comments:



















72







APPE14DIX B
CLIENT POST-SESSION EVALUATION FORM



Name:



Your participation in this experiment is now concluded. Please answer the remaining few questions as candidly as you can. Feel free to make additional comments where you wish to. Circle one of the alternatives.



1. Did you like the person who interviewed you?

very much slightly not at all

2. Did you enjoy the experience of being interviewed?

very much slightly not at all

3. Do you think the experience was valuable?

very much slightly not at all

4. Did it help you to solve, or live better with, your
problem?

yes unsure no

5. Do you think participating in this experiment has made
you more likely to seek professional help should the
occasion ever arise?

yes unsure no

Additional comments:













73














APPENDIX C
NOTICE FOR POTENTIAL CLIENT SUBJECTS





An Analogue Study of Counseling


We are conducting a study about counseling and psychotherapy, and need a number of subjects to be clients who will talk with therapists. The first part of the study involves taking some paper-and-pencil tests, and only a part of the subjects who take the tests will go on to the second part of the study. The first part of the study will take about 30 minutes, and you will receive 1 half-hour credit. The second part of the study, if you are selected, will take about an hour-and-a-half, and you will receive 3 half-hour credits. If you would like to participate in this study, please sign up below for only one of the testing times.


















74







APPENDIX D
VERBATIM TELEPHONE CONTACT
FOR INFORMI14G SELECTED CLIENTS





We're calling about the Psychology Study you're in. Your record was selected to be matched to a person in the other group. Now we would like to arrange a time for you and the other person to continue in the study. Your part of the study is to think about and select a real problem that you face. It should be something that you don't mind discussing with this second person. The second person is one of a number of therapists who consented to participate and who will discuss this area with you. You will meet at

--- (time) --- to complete a few brief paper and pencil questions and then have a conversation which will be videotaped. After the discussion there will be a few more paper and pencil items. The entire time for this part of the study will take about 1 1/2 hours.





















75







APPENDIX E
RULES FOR LOCATING SEGMENTS
WITHIN FULL INTERVIEWS



Master tapes are composed of three-minute segments of interviews with interviews and segments within interviews in randomized order. There are a total of 120 segments with

3 segments from each of the 40 approximately 50 minute interviews. Position of segments within each interview was determined by these rules.


Rule 1. Segment 1 begins at the beginning of the interview and continues as described in Rule 4. Specifically, the beginning of the interview is defined as the first videoresolved frame with clear audio. At that point the counter on the play-back recorder is set to 000.

Rule 2. Segment 2 begins with the first statement (as defined in the Method section) which begins after 17 minutes of interview time have elapsed. Specifically, the playback recorder is run forward to 425. It is then played past 430, and the first statement beginning after that point (430) is noted. The playback recorder is then rewound, played forward, and recording started at the identified statement. All machines used in this recording of master tapes were identical SONY units. (Other machines would need to have different counter numbers specified.)

Rule 3. Segment 3 begins with the first statement (as defined in the method section) which begins after 17 minutes of interview time have elapsed. The same procedure is used


76




77


as in Rule 2, with counter settings 595, and 600. If interview ends prior to 3 minutes elapsed time, rewind 100 counter units and use that number as the starting point. If the interview was too long, use 650 or 700 as the starting point.

Rule 4. on all segments, stop recording at the first sentence or clause end after 180 seconds or at 185 seconds.








APPENDIX F
TRAINING MANUAL FOR UNITIZATION OF
TRANSCRIPTS BY PROPOSITION



Your role in this study is to assist in the preparation of transcripts of interviews for scoring by other raters like yourself. It is most important that your work be done accurately, and it will require that you concentrate well on it. This training manual will help you learn the tasks, and generally, the tasks are specific and 'should not be too difficult. Your task is to go through the transcript and to use diagonal marks to separate phrases defined below as PROPOSITIONs.

A PROPOSITION is defined as a sequence of words in the transcript containing a subject and predicate, whether expressed or implied, and including the modifiers of that subject-predicate group. This is essentially the definition of a clause, which may be either coordinating (or independent) or subordinating (or dependent) Any clause is scored as a PROPOSITION with two additional stipulations. Parenthetical clauses, clauses which come in the middle of a larger clause, are not scored as separate PROPOSITIONs but are considered PARENTHETICALs within a PROPOSITION. Also, incomplete clauses, as, for example, when a person changes the message mid-sentence, are scored as PROPOSITIONs.

The following suggestions will help in making these

scorings. In the Harbrace Colleg Handbook, read the definition of conjunction, p. 446, and the discussion of conjunctions, p. 12-13. Conjunctions are, of course included 78




79


with the clause they introduce. Infinitives are not predicates. When auxiliary verbs (Harbrace, pp. 3-4) occur, make sure the PROPOSITION includes the entire predicate. Phrases which interrupt the predicate (e.g. "sort of" in "he would sort of be nice") do not split un the PROPOSITION. "You know" in its use as a sort of filling phrase, is not scored as a PROPOSITION unless it occurs clearly outside of any other PROPOSITION. Here are some examples:

1. /How much of this is a proposition?/
2. T: /How do you feel?/
C: /Bad./ I just really feel bad./
3. /He wants it this way,/and so do we./
4. /Whenever it happens, /we all know about it./
5. /This, that he wants, is going well./

Now note the subject-predicate in each PROPOSITION:

1. /this is/
2. T: /you do feel/
C: / (I feel) / I feel/
3. he wants / we do (want)
4. it happens / we know
5. this is going /

Words in parentheses indicate they were understood, and note that in #5 there is a parenthetical phrase that is not scored as a PROPOSITION even though there is a subjectpredicate: "he wants."

Now mark the diagonals on this reproduction of part of a transcript, and have that practice checked by the experimenter before continuing past the practice.



T: I guess the reason I said that, Ray, was because you

seemed to express such a sense of freedom when she was

gone.




80


C: Well that's just how it affected me. Like suddenly I

felt like I was just free.

T: So she is after you pretty frequently. C: Yep.

T: Kind of asking you to shape up. C: Well, it just seems like a constant hassle trying to

account for places to go and things I do and making all

the normal trivial talk that really doesn't make any

difference. I'd just as soon skip anyway. It's a

bother.





If you have any questions, discuss them with the experimenter when he checks your practice rating.







APPE14DIX G
TRAINING MANUALS FOR
RATING ALL VARIABLES



Your role in this study is to locate instances of certain verbal behavior in typed transcripts of interviews. It is critically important that your work be done accurately, and it will require that you concentrate well on it. Generally, your rating task is quite specific and should not be too technically difficult.

There are three definitions that will specify the

score you make in each situation. The first definition is called the "unit," the second, the "category," and the third, the "indicator" or set of rules for assigning a unit to a specified category.

In all of the rating tasks in this study, several ideas are important. One is the manner in which the pages of the transcript are labeled. The largest division labeled is the "interview segment" and this is the second number in the label (the first number is an arbitrary grouping number). The third number is the page number within a particular segment. The last number is the page number within the entire group of all transcripts. In other words, the page labeled: 1-04-2-008, is the second page of the fourth segment in the first group of segments, and it is the eighth page of the entire 288 pages of the transcript. INote: Each rater had a training manual with this introduction, the section corresponding to his or her task, and one or more of the lists of words or phrases from
Appendices H, I, J, or K as specified.

81




82


Also, each time the speaker changes from the patient to therapist or vice versa, this defines a "statement" and statements are numbered down the left hand margin of the page beginning with one at the beginning of each "segment." Each "proposition" is set off by diagonal marks, and is numbered starting with one (1) at the beginning of each
it statement.





83


Training Manual: Primary Affect Word (PAW)



In scoring PAWs, the unit is the WORD, the category is PAW, and the indicator (or rule) is that the word is found on the attached list of PAW words. WORD is defined as the basic blocks of letters constituting units within the transcript, and is set off at its beginning and end by a space or punctuation. (Note: if a typographical error results in no space being left, it is considered that the words are separate for scoring and counting purposes.) The category is noted by entering a code number for the WORD scored, and labeling the scoring page, PAWs. The code number is composed of the page label plus the statement number containing the word, the PROPOSITION number containing the WORD, and the number of the WORD counting from the beginning of the PROPOSITION.

The words on the list of PAW words are almost all in

the adjective form, but any part-of-speech form of the woii is to be scored. Generally these words are words which almost always denotatively refer to affective, emotional, or feeling experience. Examples are angry, sad, happy, etc. (See list at this time.) Some of the words are called "limited PAWs," and these are to be scored only when the meaning of the word as used is equivalent to the meaning as prescribed on the list. "Feeling" is marked with an asterisk, and its code number should be marked with an asterisk. It should not be coded at all, however, when it has the meaning of "think" rather than "feel." Here are some examples:




84


Ex. 1: a) He is angry. b) His anger is strong.

c) He spoke angrily. d) He angered them.

Ex. 2: She is a moody person, who is really weird. Ex. 3: I'm tired and depressed. Ex. 4: a) I was really hurt when she left.

b) I was really hurt in the accident.

Ex. 5: a) I feel like I've been through a wringer.

b) I feel like he's a real dummy. Notes:

Ex. 1: all forms of "angry" are marked as PAWs: adjective,
noun, adverb, and verb.

Ex. 2: "Moody," but not "weird" is found on the list of
PAWs.

Ex. 3: "Depressed" but not "tired" is found on the list. Ex. 4: "Hurt" is scored only when it has the limited meaning, emotionally injured and not physically injured. Ex. 5: In "a" "feel" denotes emotional experience, but in
"b" it denotes thinking.


Now try this reproduction of part of a transcript, circling each WORD you find on the list of PAWs. T: But you have hurt them. C: Sure. Whenever you fight a lot you always hurt somebody, particularly a child.

T: You sound a little bit guilty. C: I feel guilty because; it's not my kid's fault. She

has nothing to do with it.


If you have any questions, discuss them with the experimenter when he checks your practice rating.




85


Training Manual: Secondary Affect Word (SAW)


In scoring SAWs, the unit is the WORD, the category is SAW, and the indicator (or rule) is that the word is found on the attached list of SAW words and that it occurs within a PROPOSITION containing (explicitly or implicitly) a form of the word "feel" or "sense," or within a PROPOSITION modifying such a PROPOSITION. PROPOSITION was defined above. WORD is defined as the basic blocks of letters constituting units within the transcript, and is set off at its beginning and end by a space or punctuation. (Note: if a typographical error results in no space being left, it is considered that the words are separate for scoring and counting purposes.) The category is noted by entering a code number for the WORD scored, and labeling the scoring page, SAWs. The code number is composed of the page label plus the statement number containing the WORD, and the number of the WORD counting from the beginning of the PROPOSITION. One code number is recorded for the "feel" or "sense" WORD, and a second number, for the SAW WORD.

The words on the list of SAW words are almost all in

the adjective form, but any part-of-speech form of the word is to be scored. Generally these SAW words are words which denotatively refer to affective, emotional, or feeling experience when used with the word "feel" (usually as a verb or noun) or the word "sense." Examples are alert, helpless, inhibited, etc. (See list at this time.)

Here are some examples:




86


Ex. 1: T: How do you feel?

P: Inhibited.

Ex. 2: This is an impatient feeling, but then I'm an

impatient person.

Ex. 3: 1 have a sense of uncertainty sitting here. NotesEx. 1: "Inhibited" is circled to indicate it is a SAW

since it is on the list of SAWS and the patient

statement implies "I feel".

Ex. 2: "Impatient" is only scored when it falls within a

PROPOSITION with "feel." Also, the second PROPOSITION does not modify the first, which would have

been the case had it read: "This feeling is

strange, as I am very impatient."

Ex. 3: "Uncertainty" is on the list and is in PROPOSITION

with "sense."


Now try this reproduction of part of a transcript,

circling each WORD you find on the list of SAWs which fits the criteria.

T: What do you feel like today? P: Really tired. I feel really depressed, and I wish I

weren't so insecure. I really feel insecure today,

especially today.

T: Today especially, you're feeling really depressed and

insecure. And really tired.


If you have any questions, discuss them with the experimenter when he checks your practice rating.





87


Training Manual: Affective Idioms (AI)


In scoring AIs, the unit is the PROPOSITION, the category is AI, and the indicator (or rule) is that a group of words occurring in the PROPOSITION is also found in a list of AIs attached to this manual. PROPOSITION has been defined earlier. The category is noted by entering a code number for the PROPOSITION scored, and labeling the scoring page, AIs. The code number is composed of the page label plus the statement number containing the phrase and the PROPOSITION number.

Generally AIs are groups of words which together denote affective, emotional, or feeling experience, but separately do not. Examples are "sick and tired" and "peace of mind." (See list at this time.) Flexibility must be given for the exact arrangement of the words and variations in the parts of speech used. In each case the test is whether the phrase as used in the transcript denotes affective experience. Here are some examples:


Ex. 1: 1 have peace of mind. Ex. 2: He must feel screwed up. Ex. 3: That really gives me the creeps.


Now try this reproduction of part of a transcript, circling each phrase you find from the list of AIs. P: Well, I just feel like there's something wrong. I'm

just sick and tired of living. That's really been

making me down for the last couple of months.




88


T: You're really feeling depressed, huh? P: Yah. I just have no taste for anything. And now this

is really giving me the creeps; my not feeling like

living.



If you have any questions, discuss them with the experimenter when he checks your practice rating.




89


Training Manual: Self References (SR)


In scoring SRs, the unit is the PROPOSITION, the category is SR, and the indicator (or rule) is defined by four rules discussed below. PROPOSITION has been defined earlier. The category is noted by entering a code number for the PROPOSITION scored, indicating the rule number used, and labeling the scoring page, SR. The code number is composed of the page label plus the statement number plus the PROPOSITION number. Since several SRs may follow one another, the code number may be entered as a range as long as only SRs are included. If this is done, each scoring rule used at least once within the sequence is indicated.

Generally, a self reference (SR) is a verbal response

by a subject which expresses descriptively or emotively something about the subject in relation to himself, others, or the world. This indicator is more specifically defined by the following four rules.1 Examples of each follow the rule. Rule 1: A self reference (SR) statement must begin with or
include a first person singular personal uronoun (e.g., I, myself, mine) unless it is covered by one of the criteria
below.

Rule 2: First person plural pronouns (e.g., we, our, ourself, ours) are counted as self references when the group
referred to is intimately related to the subject (e.g.,
family, roommate, girlfriend).

Rule 3: Second person pronouns (e.g., youl your, yours) are
counted as self references when they are clearly used as
a substitute for a first person pronoun form and not as an !Note: These four rules are taken directly from materials
received from Green and Marlatt (Note 1).




90


impersonal reference to a group. They can be checked, if necessary, by comparing the sense of the statement when substituting a first person pronoun with the sense of the statement when substituting a group reference (e.g., they, everybody, the world, etc.).

"You get a good feeling with these people." is counted.
"You can't solve anything that way" is not counted.

Rule 4: A response may be a self reference without explicitly using a personal pronoun if it is closely related to
a previous self reference and is contingent upon it.
These responses can be checked by adding a self reference phrase such as "to me" to the statement.

"I have met a lot of people here. It certainly has been
satisfying." (Both sentences are self references.)


Here are some more examples, with SRs underlined. Note that each example illustrates one of the rules.


Ex. 1: /1 like this. /This is mine./ Ex. 2: a) /My family is a lot of fun. /We do things together./
b) /American citizens celebrated the bicentennial./
/It was very interesting to us all./ Ex. 3: a) /You really like coming here./
b) /You do this,/ and it will make a lot of difference./

Ex. 4: /1 have really tried hard./ It has just about done
it./

Ex. 5: /1 like doing things like this/that are a lot of fun./


Now try this reproduction of part of a transcript, underlining any PROPOSITION that should be scored SR.


T: But you have hurt them. C: Sure. Whenever you fight a lot you always hurt somebody, particularly a child.

T: You sound a little bit guilty.




91


C: I feel guilty because; it's not my kid's fault. She has

nothing to do with it.



If you have any questions, discuss them with the experimenter when he checks you practice rating.




92


Training Manual: Therapist Instruction Strategy (I)


In scoring Is, the unit is the PROPOSITION, the category is I with subcategories of 1-1, 1-2, and 1-3, and the indicator or rule is specified by a set of rules defined earlier. The category is noted by entering a code number for the PROPOSITION scored, indicating the type of Instruction (1-1, -2, or -3), and labeling the scoring page, I. The code number is composed of the page label plus the statement number plus the PROPOSITION number.

I-I has the general meaning: "Talk about your feelings." There are three rules for assigning a therapist PROPOSITION (or PROPOSITION Group) to category 1-1. 1) The PROPOSITION is in the imperative mood, or can be translated into the imperative mood without significantly altering the PROPOSITION's meaning. 2) The PROPOSITION contains some form of the word "feel," "emotion," "affect," or "sense." 3) The PROPOSITION (or group of PROPOSITIONS) has the general meaning: "Talk about your feelings."

1-2 has the general meaning: "Do not talk about subject X (where X is a subject other than the general topic of feelings)." The rules here are: (1) The PROPOSITION is in the imperative mood or can be translated as above. 2) the PROPOSITION's imperative is a negation of something. 3) The PROPOSITION does not contain any form of the words listed under #2 of the indicator for 1-1. 4) The general meaning is as at the beginning of this paragraph.

1-3 has the general meaning "Talk about subject V or




Full Text
63
sex of therapist and client affective self reference in seg
ment 1 (r = .32) and Reinforcement in segment 1 (r = -.40).
The negative direction of the correlation indicated that the
association was between female therapist and higher levels
of client affective self reference in the second segment (or
male therapist and lower Reinforcement measures). Sex of
therapist approached significance in the same direction with
client affective verbalization in segment 1 (r = -.27,
¡D = .088), total client affective self reference (r = -.28,
p = .083), and total Reinforcement (r = -.31, ¡3 = .053).
Sex of therapist approached significance in the opposite di
rection on client affective self reference change from first
to last segment (r = .30, g_ = .057), indicating that male
therapists were somewhat more associated with higher over
all change in client affective self reference scores than
female therapists. Sex of client approached correlation
with client first segment affective verbalization and self
reference (r = -.28, p = .078; r = -.29, £ = .069), with fe
male clients tending to be associated with higher scores.
The second exploratory analysis was to examine the com
ponents of the compatibility score. Since the interview
sample includes only individuals in certain matched pairs,
the significance of any correlations are very suspect for
interpretation, but may suggest directions for further ex
ploration of the compatibility dimension. All basic com
patibility measures were computed and were correlated with
all affective verbalization measures including change scores,
Reinforcement measures, and outcome measures. Several cor-


a rating system. All rating systems were implemented through
a training manual and standard procedure. Hypotheses pre
dicted that client affective verbalization and client affec
tive self reference would increase across the interview,
with the therapist strategies associated with larger in
creases. Interpersonal need compatibility, sex of client,
and interactions between those two variables and the strate
gies were also predicted as important in the effects related
to the dependent measure.
The rating procedure appeared adequate with rating
checks producing good reliability. Data were analyzed
through a split plot factorial (2x2x3) analysis of variance
of compatibility, client sex, and segment separately for cli
ent affective verbalization and client affective self ref
erence. One effect was significant: compatibility X client
sex, with a simple main effects test indicating that the
compatibility manipulation was effective only for female cli
ents, as previous research has suggested. Other analyses
were conducted through Pearson or Spearman correlations.
Results suggested that clients tend to influence the affec
tive verbalization level of therapists rather than vice
versa, and that therapists in this sample do not actually
use the strategies which analogue studies have researched.
There was support for therapists believing that client af
fective verbalization is important to treatment outcome. Al
ternative hypotheses and alternatives for further research
are discussed.
vm


ACKNOWLEDGMENTS
I would like to thank my chairperson, Dr. Jacquelin
Goldman, for her willing and consistent guidance and assis
tance through a long period of often frustrating endeavor.
I am also appreciative of the contributions of each of my
other committee members: Dr. Audrey Schumacher, for her
influence on my thinking about therapy and her always
incisive critiques; Dr. Franz Epting for his encouragement
to be more scholarly and to be creative in research appli
cations; Dr. Marvin Shaw, for his assistance in solving
the difficult design and analysis problems involved in the
study; and Dr. Hannelore Wass, for her perspective from
outside my immediate area.
Many other individuals, faculty, staff, fellow graduate
students, and undergraduate students, were involved in the
various aspects of the work reported here, and though they
are too numerous to mention, I would like to express my
appreciation for their cooperation. A special word of
thanks goes to my typist, Barbara Jones, for her hard
work above and beyond the call of duty.
iii


CHAPTER 1
INTRODUCTION
Psychotherapy is an important phenomenon, both cul
turally and scientifically. Culturally, it is the practice
of a helping profession by therapists and the seeking of as
sistance by clients. Scientifically, it is the interaction
of two types of experimental subjects, where one subject is
to effect some change in the other, often changes in very
basic personality processes of the client subject. In both
cases, a major question is that of how the therapist has his
impact with the client. As Strupp (1973) puts it, "How does
it come about that one person, by virtue of what he is or
what he does, can exert such a lasting influence on another
person that the changes that result from this encounter may
be termed therapeutic?" (p. 19) The variations of this ques
tion and the more specific implications of it are numerous
and complex, as indicated by the scope and number of reviews
on research of psychotherapy (e.g., Bergin & Garfield, 1971;
Bordin, 1974; Kiesler, 1973; Meltzoff & Kornreich, 1970).
The present study will be focused on looking at one aspect
of answering the above question.
One difficult task in beginning this focusing is speci
fying the way in which the interaction of therapist and
client is to be conceptualized and described. The general
1


92
Training Manual: Therapist Instruction Strategy (I)
In scoring Is, the unit is the PROPOSITION, the cate
gory is I with subcategories of 1-1, 1-2, and 1-3, and the
indicator or rule is specified by a set of rules defined
earlier. The category is noted by entering a code number
for the PROPOSITION scored, indicating the type of Instruc
tion (1-1, -2, or -3), and labeling the scoring page, I.
The code number is composed of the page label plus the state
ment number plus the PROPOSITION number.
1-1 has the general meaning: "Talk about your feel
ings." There are three rules for assigning a therapist
PROPOSITION (or PROPOSITION Group) to category 1-1. 1) The
PROPOSITION is in the imperative mood, or can be translated
into the imperative mood without significantly altering the
PROPOSITION'S meaning. 2) The PROPOSITION contains some
form of the word "feel," "emotion," "affect," or "sense."
3) The PROPOSITION (or group of PROPOSITIONS) has the gen
eral meaning: "Talk about your feelings."
1-2 has the general meaning: "Do not talk about sub
ject X (where X is a subject other than the general topic
of feelings)." The rules here are: (1) The PROPOSITION is
in the imperative mood or can be translated as above.
2) the PROPOSITION'S imperative is a negation of something.
3) The PROPOSITION does not contain any form of the words
listed under #2 of the indicator for 1-1. 4) The general
meaning is as at the beginning of this paragraph.
1-3 has the general meaning "Talk about subject X" or


83
Training Manual: Primary Affect Word (PAW)
In scoring PAWs, the unit is the WORD, the category is
PAW, and the indicator (or rule) is that the word is found
on the attached list of PAW words. WORD is defined as the
basic blocks of letters constituting units within the tran
script, and is set off at its beginning and end by a space
or punctuation. (Note: if a typographical error results
in no space being left, it is considered that the words are
separate for scoring and counting purposes.) The category
is noted by entering a code number for the WORD scored, and
labeling the scoring page, PAWs. The code number is composed
of the page label plus the statement number containing the
word, the PROPOSITION number containing the WORD, and the
number of the WORD counting from the beginning of the
PROPOSITION.
The words on the list of PAW words are almost all in
the adjective form, but any part-of-speech form of the word
is to be scored. Generally these words are words which al
most always denotatively refer to affective, emotional, or
feeling experience. Examples are angry, sad, happy, etc.
(See list at this time.) Some of the words are called
"limited PAWs," and these are to be scored only when the
meaning of the word as used is equivalent to the meaning as
prescribed on the list. "Feeling" is marked with an
asterisk, and its code number should be marked with an
asterisk. It should not be coded at all, however, when it
has the meaning of "think" rather than "feel." Here are
some examples:


13
to nonself was related to therapist-rated success in client-
centered therapy. The validity of this variable seems theo
retically and intuitively adequate, and its validity will be
examined further in the present study by examining its re
lationship to session outcome.
Significant reliability between raters of interaction
material is often used as a measure of the adequacy of data
for process research, and in many studies it is sufficient
to answer the questions posed. In the present study and for
future extension to other aspects of Haley's model, however,
it is important that "discrete" mean that with very few ex
ceptions it can be agreed that a particular instance of a
chosen behavior has or has not occurred. In order to attempt
achieving this type of precision of measurement without
losing the relevance of the behaviors in which we are inter
ested, the communication situation must be considered in
more detail and then the particular variables to be consid
ered in this study can be defined. An overview of the field
of communication science as it relates to psychotherapy re
search will be used to supply greater detail and establish
perspective.
Communication Science: An Overview
Kiesler (1973) in presenting process studies of psycho
therapy and proposing a structure for developing such studies
suggests Market's (1969) model of dyadic interaction. The
basic communication situation is composed of an originator
of a message, the ENCODER, a recipient of a message, the DE-


18
verbalization of affect reflects a definition of the rela
tionship, and the research question becomes, "How does the
therapist behave so as to attempt increasing client affec
tive verbalization?" The literature related to increasing
client verbalization of affect indicates the particular ef
fectiveness of at least three strategies. One is that the
therapist tells the client directly what kind of relation
ship he (the therapist) wants; i.e., the therapist instructs
the client on how to behave. A second is that the therapist
suggests a definition and behavior by modeling the desired
client behavior. Third, the therapist utilizes differential
verbal reinforcement of the desired behavior to increase its
occurrence.
The first strategy is the use of instructions by the
therapist, and has been studied and found effective in a num
ber of studies (e.g., Green & Marlatt, 1972; Lukens, 1970;
Merbaum & Lukens, 1968; and Whalen, 1969). Here the thera
pist communicates directly to the client that in the rela
tionship it is the client's responsibility to engage in par
ticular behaviors, and perhaps, not in other behaviors. For
example, at the beginning of the interview, the therapist
might say, "During the interview, you are to talk about your
feelings about yourself, and not about the feelings other
people have or subjects unrelated to yourself." In that
case, the therapist would be making a very explicit defini
tion of the relationship as to what he expected from the cli
ent .
The second strategy is for the therapist to be a model


48
TABLE 5
SUMMARY OF CROSS-LAGGED PANEL CORRELATIONS
CLIENT OR
THERAPIST3
MEASURE AS
"CAUSE"b
TIME LAG EXPRESSED
AS SEGMENT DIFFERENCE
SEGMENT
1 AND 2
SEGMENT
2 AND 3
SEGMENT
1 AND 3
CLIENT AFFECTIVE
VERBALIZATION
THERAPIST
.09
.09
.24
CLIENT
.21
. 32*
.34*
CLIENT AFFECTIVE
SELF REFERENCE
THERAPIST
. 10
. 11
. 17
CLIENT
.24
.27
. 38*
Note: No
correlation pairs are
significantly
different
using £<.05 as criterion in a one-tailed test.
a Therapist measures are all Therapist Affective Verbal
ization scores.
b "Cause" indicates that the measure for that person is
for the earlier of the two time periods.
*
£<.05


CLIENT AFFECTIVE SELF REFERENCE
104
6.0 -
5.5 -
5.0 -
4.5 -
4.0 -
3.5 -
3.0 -
2.5 -
Legend: HM
HF
LM
LF
12 3
SEGMENT
High compatible dyad males.
High compatible dyad females.
Low compatible dyad males.
Low compatible dyad females.
INTERACTION GRAPH:
COMPATIBILITY X CLIENT SEX X SEGMENT
FIGURE L-2


79
with the clause they introduce. Infinitives are not predi
cates. When auxiliary verbs (Harbrace, pp. 3-4) occur,
make sure the PROPOSITION includes the entire predicate.
Phrases which interrupt the predicate (e.g. "sort of" in
"he would sort of be nice") do not split up the PROPOSITION.
"You know" in its use as a sort of filling phrase, is not
scored as a PROPOSITION unless it occurs clearly outside of
any other PROPOSITION. Here are some examples:
1. /How much of this is a proposition?/
2. T: /How do you feel?/
C: /Bad./ I just really feel bad./
3. /He wants it this way,/and so do we./
4. /Whenever it happens, /we all know about it./
5. /This, that he wants, is going well./
Now note the subject-predicate in each PROPOSITION:
1. /this is/
2. T: /you do feel/
C: / (I feel) / I feel/
3. / he wants / we do (want) /
4. /it happens / we know /
5. / this is going /
Words in parentheses indicate they were understood, and
note that in #5 there is a parenthetical phrase that is not
scored as a PROPOSITION even though there is a subject-
predicate: "he wants."
Now mark the diagonals on this reproduction of part of
a transcript, and have that practice checked by the experi
menter before continuing past the practice.
T: I guess the reason I said that, Ray, was because you
seemed to express such a sense of freedom when she was
gone.


TABLE 10
ANOVA-3 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
COMPATIBILITY
19.072
1
19.072
2.90
NS
2
CLIENT SEX
2.601
1
2.601
. 40
NS
3
COMPATIBILITY X CLIENT SEX
11.990
1
11.990
1.83
NS
4
ERROR
236.500
36
6.569
5
TOTAL
270.163
39
Note: Dependent variable is Client Total Affective Verbalization.
NS 05
On


42
TABLE 3
CORRELATIONS OF MODELING AND CHANGE SCORES
SEGMENT
CHANGE
MODELING: THERAPIST AFFECTIVE
VERBALIZATION BY SEGMENT
1
2
3
Total
CLIENT AFFECTIVE VERBALIZATION
1 to
2
-.44**
.04
-.01
-.13
2 to
3
.11
-.11
.07
.01
1 to
3
-.29
-.08
.06
-.11
CLIENT AFFECTIVE SELF REFERENCE
1 to
2
-.46**
.08
-.08
-.14
2 to
3
.04
-.17
.01
-.08
1 to
3
-.41**
-.11
-.06
-.23
** £

88
T: You're really feeling depressed, huh?
P: Yah. I just have no taste for anything. And now this
is really giving me the creeps; my not feeling like
living.
If you have any questions, discuss them with the experimen
ter when he checks your practice rating.


37
apist statements were scored for self references as part of
the Modeling strategy measurement, only 31 therapist affec
tive self references were found in all, making the inclusion
of that variable meaningless, and it was eliminated from
further analysis. This means that the strategy Modeling is
equivalent to therapist affective verbalization in all analy
ses. Scores for the other strategies and for the client
variables appeared adequate.
Hypothesis 1 predicted that each therapist would use
one or more of the strategies, and was informally analyzed
since some of the strategies were not dichotomous. Several
factors suggest that this hypothesis was not supported. First
of all, the Instruction strategy was clearly absent, as was
the use of modeling of affective self reference behavior.
Second, the frequency of modeling of affective verbalization
behavior was at or below client frequencies in many cases
and was generally low for therapists, inconsistent with our
expectation of a Modeling strategy. Third, the Reinforcement
behavior of therapists often did not involve a discriminant
use of reinforcers, but rather relatively high rates of re
inforcing all client verbalization within a particular seg
ment. Overall, therapists did not appear to use the defined
strategy behaviors systematically, and each therapist clearly
did not use at least one strategy as there were some thera
pists who did not approach using any of the strategies.
Hypothesis 2 predicted that both client affective ver
balization and client affective self reference would increase
across the session. This was analyzed by checking for a


86
Ex. 1:
T: How do you feel?
P: Inhibited.
Ex. 2 :
This is an impatient feeling, but then I'm an
impatient person.
Ex. 3:
I have a sense of uncertainty sitting here.
Notes:
Ex. 1:
"Inhibited" is circled to indicate it is a SAW
since it is on the list of SAWS and the patient
statement implies "I feel".
Ex. 2:
"Impatient" is only scored when it falls within a
PROPOSITION with "feel." Also, the second PROPO
SITION does not modify the first, which would have
been the case had it read: "This feeling is
strange, as I am very impatient."
Ex. 3:
"Uncertainty" is on the list and is in PROPOSITION
with "sense."
Now try this reproduction of part of a transcript,
circling each WORD you find on the list of SAWs which fits
the criteria.
T: What do you feel like today?
P: Really tired. I feel really depressed, and I wish I
weren't so insecure. I really feel insecure today,
especially today.
T: Today especially, you're feeling really depressed and
insecure. And really tired.
If you have any questions, discuss them with the experi
menter when he checks your practice rating.


CLIENT AFFECTIVE VERBALIZATION
AND THERAPIST INFLUENCE STRATEGIES IN
ANALOGUE PSYCHOTHERAPY INTERVIEWS
By
STANLEY E. JONES
A THESIS PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1976


56
for effects. (For examples, see Appendix L.) Generally, it
was observed that the plots of mean scores for various var
iable groups suggested support for many complex interactions,
but of the ones of those which could be tested, only the
one interaction from the analyses of variance reported be
low was significant. The variability within each cell of
the data for any of these plots is quite large, and these
data must be approached very cautiously, and with the assump
tion that all of the effects are statistically non-signif
icant .
Interpretation is further complicated by what appears
to be a lack of independence of Modeling, Reinforcement,
compatibility, client sex, and level of client affective
verbalization, with an underlying factor of what appears to
be a high-low dimension on level of client affective verbali
zation in segment 1. In other words, female clients tended
to be higher than males overall, especially when in the
high compatible group. When high Modeling dyad clients or
high Reinforcement dyad clients are separated from the total
sample, they also appear more likely to be these high com
patible female clients than any others. This implies that
the creation of class variables from the continuous strategy
measures (which was done to create plots) is questionable
since it is likely to create groups which are similar in un
known ways with respect to their relationship with the de
pendent measures, while their class labels would suggest we
are dealing with different variables. In other words, split-
ting the sample on the basis of the Modeling strategy, for


27
tapes. The cassettes were used to make typed transcriptions
according to the rules for typists described by Gottschalk,
Winget, and Gleser (1969) and were checked for accuracy by
the experimenter. The segment locations were determined by
taking the first 3 minutes of the interview, and the approx
imate middles of the middle and final thirds of the inter
view. Detailed rules for determining the segments are found
in Appendix E.
Rating Procedures
The rating process was based on standard content
analysis procedures and terminology (Berelson, 1952). In
the case of most variables, it operated within what has
been termed the "classical model" (Marsden, 1971), mean
ing that what is measured is limited to the semantic
and syntactic aspects of communication. Content analysis
requires the definition of three ideas for the measurement
of each variable, and these are 1) the unit, 2) the cate
gory, and 3) the indicator (Berelson, 1952). Each of these
aspects of the content analysis process will be elaborated
as needed for each variable to be measured.
Units. With the exception of the unit "WORD," the unit
for rating each variable was scored separately and marked on
the transcript. Units utilized were the "WORD," the "PROPO
SITION," the "STATEMENT," the "SEGMENT," and the "INTERVIEW."
WORD is essentially self explanatory, and includes hyphe
nated words as one WORD. PROPOSITION is a more complex unit,
and was defined according to Lennard and Bernstein (1969)


85
Training Manual: Secondary Affect Word (SAW)
In scoring SAWs, the unit is the WORD, the category is
SAW, and the indicator (or rule) is that the word is found
on the attached list of SAW words and that it occurs within
a PROPOSITION containing (explicitly or implicitly) a form
of the word "feel" or "sense," or within a PROPOSITION mod
ifying such a PROPOSITION. PROPOSITION was defined above.
WORD is defined as the basic blocks of letters constituting
units within the transcript, and is set off at its beginning
and end by a space or punctuation. (Note: if a typograph
ical error results in no space being left, it is considered
that the words are separate for scoring and counting pur
poses.) The category is noted by entering a code number for
the WORD scored, and labeling the scoring page, SAWs. The
code number is composed of the page label plus the state
ment number containing the WORD, and the number of the WORD
counting from the beginning of the PROPOSITION. One code
number is recorded for the "feel" or "sense" WORD, and a
second number, for the SAW WORD.
The words on the list of SAW words are almost all in
the adjective form, but any part-of-speech form of the word
is to be scored. Generally these SAW words are words which
denotatively refer to affective, emotional, or feeling ex
perience when used with the word "feel" (usually as a verb
or noun) or the word "sense." Examples are alert, helpless,
inhibited, etc. (See list at this time.)
Here are some examples:


30
checks over 769, 353, and 343 PROPOSITIONS respectively,
error rates of 0%, .5%, and 0% were obtained.
Secondary Affect Words were scored using the unit:
PROPOSITION, the category: Secondary Affect Word, and the
indicator: inclusion on the list of Secondary Affect Words
in Appendix I. This list was generated by selecting words
which refer denotatively to affective states only when mod
ifying the verb "feel" or the verb "sense." Examples are
"alert," "slow," "inhibited," etc. All part of speech
forms of the words listed in Appendix I were scored. The
rater's task for this affective expression was to note the
subject's use of the verb "feel" or the verb "sense" in a
PROPOSITION or adjacent PROPOSITION modifying it, either
directly or by its being understood (as in an answer to a
question using one of these verbs). When either verb was
noted, the rater then used WORD as the unit, and compared
each WORD in the verb's modifier with the list of Secondary
Affect Words and labeled any WORDS found by recording the
WORD'S code number. The same rating and verification pro
cess was used, and the rater was trained using the manual
in Appendix G. Two raters were used for this task and in
reliability checks over approximately 350 PROPOSITIONS each,
no errors were obtained.
Affective Idioms were scored using the unit: PROPO
SITION, the category: Affective Idioms and Slang, and the
indicator: inclusion on the list of Affective Idioms in
Appendix J. This list was generated by selecting groups of


DEDICATION
To Barbara
Whose patience and support have
made all the difference.


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Hannelore L. Waas, Professor of
Foundations of Education
This dissertation was submitted to the Graduate Faculty of
the Department of Psychology in the College of Arts and
Sciences and to the Graduate Council, and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
December, 1976
Dean, Graduate School


TABLE 4-continued
TAV
1
.66
.001*
.09
. 5.85
.24
. 126
.55
.001*
.67
.001*
.10
.526
.17
.298
.48
.002*
TAV
2
.21
.197
.26
. 109
.09
.566
.28
.076
.24
.141
.32
.047*
. 11
.504
.32
.042*
.19
.250
TAV
3
. 34
. 029*
. 32
.046*
. 39
.011*
.52
. 001*
. 38
.015*
.27
.096
. 31
.052
.42
. 006*
.58
.001*
. 34
. 034*
TAV
T
.53
.001*
. 35
. 026*
. 36
.024*
.63
.001*
.55
.001*
.35
.027*
.27
.089
.55
.001*
.73
.001*
.60
.001*
.89
. 001*
Note: The upper number of each pair is the correlation coefficient. The lower number
is the probability that the null hypothesis (r = 0) is true.
Key: CAV = Client Affective Verbalization
CASR = Client Affective Self Reference
TAV = Therapist Affective Verbalization
1 = Segment 1
2 = Segment 2
3 = Segment 3
T = Total interview.
* p< .05


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EY5MQC2BO_7QPCVS INGEST_TIME 2014-10-07T00:26:56Z PACKAGE AA00025756_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


77
as in Rule 2, with counter settings 595, and 600. If inter
view ends prior to 3 minutes elapsed time, rewind 100 coun
ter units and use that number as the starting point. If the
interview was too long, use 650 or 700 as the starting point.
Rule 4. On all segments, stop recording at the first
sentence or clause end after 180 seconds or at 185 seconds.


87
Training Manual: Affective Idioms (AI)
In scoring AIs, the unit is the PROPOSITION, the cate
gory is AI, and the indicator (or rule) is that a group of
words occuring in the PROPOSITION is also found in a list
of AIs attached to this manual. PROPOSITION has been de
fined earlier. The category is noted by entering a code
number for the PROPOSITION scored, and labeling the scoring
page, AIs. The code number is composed of the page label
plus the statement number containing the phrase and the
PROPOSITION number.
Generally AIs are groups of words which together denote
affective, emotional, or feeling experience, but separately
do not. Examples are "sick and tired" and "peace of mind."
(See list at this time.) Flexibility must be given for the
exact arrangement of the words and variations in the parts
of speech used. In each case the test is whether the phrase
as used in the transcript denotes affective experience.
Here are some examples:
Ex.
1:
I have peace
of mind.
Ex.
2:
He must feel
screwed
up.
Ex.
3:
That really
gives me
the creeps.
Now try this reproduction of part of a transcript, cir
cling each phrase you find from the list of AIs.
P: Well, I just feel like there's something wrong. I'm
just sick and tired of living. That's really been
making me down for the last couple of months.


21
viewed above suggests that therapists may engage in these
behaviors in actual interviews, and that when they do, cli
ents' use of affective words increases across the interview.
The interviews will occur in a standardized situation
with each therapist serving as a control for himself or her
self, and the literature suggests that two other variables
require control and examination as independent variables.
These variables are interpersonal need compatibility and cli
ent sex, and these can be both independent variables and
controlled in other analyses by utilizing a completely bal
anced design with respect to them. In addition, therapist
sex can also be included in the balancing to control for ef
fects of that variable.
Although client affective verbalization and affective
self reference were consistently assumed important in the
literature reviewed, little evidence was cited which defended
this assumption. In order to examine the validity of these
variables in terms of session outcome, measures of session
outcome will be examined.
Hypotheses can be conceptualized as falling into three
groups. The first group deals with the question of whether
or not therapists use the strategies as defined, whether or
not the expected changes in client affective verbalization
behavior occur, and whether or not they occur or are more
pronounced when considering the strategies.
The second group of hypotheses deals with the variables
of compatibility and client sex both as independent varia
bles by themselves and also as they interact with the stra-


APPENDIX D
VERBATIM TELEPHONE CONTACT
FOR INFORMING SELECTED CLIENTS
We're calling about the Psychology Study you're in.
Your record was selected to be matched to a person in the
other group. Now we would like to arrange a time for you
and the other person to continue in the study. Your part
of the study is to think about and select a real problem
that you face. It should be something that you don't mind
discussing with this second person. The second person is
one of a number of therapists who consented to participate
and who will discuss this area with you. You will meet at
(time) to complete a few brief paper and pencil ques
tions and then have a conversation which will be videotaped.
After the discussion there will be a few more paper and
pencil items. The entire time for this part of the study
will take about 1 1/2 hours.
75


15
fies it under syntactics. Trager (1966) defines it as the
scientific study of human language or speech; the study of
messages once they are "on the air." Linguistics is composed
of three areas: PHONOLOGY: the study of sounds, with units
of increasing size called the phoneme, phone, syllable; MOR
PHOLOGY: the study of shapes (forms) constituted by sequen
ces of sounds, with units of the morpheme, morph, and word;
and SEMOLOGY (syntax or grammar): the study of the sense,
the meaningful arrangements, from which the symbolizing func
tions arise, with basic units of the part of speech, constit
uent, and the sentence.
Psycholinguistics is the study of the relation between
messages on the speech channel and the cognitive or emotional
states of human encoders and decoders who send and receive
the messages (Markel, 1969). Linguistics studies language
sounds, those sounds necessary to the production of meaning
ful speech, and psycholinguistics includes the study of non
language speech sounds, those whose variation does not change
the linguistic meaning of the speech.
This study will consider only one channel, speech, and
within that only those aspects studied within linguistics
proper. There are two primary reasons for this: 1) the
variables to be defined deal with the meanings of language
communications and the content of speech is the most impor
tant for measuring them; and 2) more highly valid and reli
able measures can be made using a single channel and single
type of meaning considering the present state of communica
tion measurement methodology.


APPENDIX A
THERAPIST POST-SESSION EVALUATION FORM
Name:
Please answer the following questions as candidly as
you can. Feel free to make additional comments where you
wish to. Circle one of the alternatives.
1. Did you like the person whom you interviewed?
very much slightly not at all
2. If this interview had occurred in the context of your
clinical practice, would you:
want to refer the client unsure continue to see client
3. Do you think this experience was valuable for the client?
very much slightly not at all
4. Do you think your client is now better able to solve or
live with his or her problem?
yes unsure no
Additional comments:
72


34
client and therapist: total number of WORDS, PROPOSITIONS,
and STATEMENTS per segment; total number of Primary Affect
Words, Secondary Affect Words, and Affect Idioms per seg
ment; total number of Primary Affect Words, Secondary
Affect Words, and Affect Idioms per segment when each oc
curred within a Self Referent PROPOSITION; number of dif
ferent types of Instruction categories per segment; number
of therapist Reinforcers preceeded by client affective ver
balization of any type per segment; number of client affec
tive verbalizations not followed by therapist Reinforcement
per segment.
The final measure for affective verbalization for both
client and therapist is the sum of the number of Primary
Affect Words plus the number of Secondary Affect Words plus
the number of Affect Idioms per Segment divided by the to
tal number of PROPOSITIONS in the segment being computed
and given that only one affective expression per PROPOSITION
has been counted in the sum. This gave a percentage score
after being multiplied by 100, and is called the client (or
therapist) affective verbalization score for Segment X.
The total affective verbalization score for the interview
is the sum of all affective expressions in the interview
divided by the number of PROPOSITIONS in the interview. The
same procedure was used for computing the affective self
reference scores.
The final measure for the Instruction strategy is "yes"
versus "no" as to use of that strategy according to the def
inition. The final measure for the Modeling strategy is


115
Whalen, C. Effects of a model and instructions on group
verbal behaviors. Journal of Consulting and Clini
cal Psychology, 1969, 3_3, 509-521.
Williamson, E. G. The meaning of communication in coun
seling. Personnel and Guidance Journal, 1959, 88,
6-14.


44
standing of the nature of the affective verbalization scores
and their relationships.
Intercorrelations were run between all client and thera
pist affective expression scores, a total of 12 variables,
and many highly significant correlations were present. (See
Table 4 for specifics of the following discussion of effects.)
Each segment level of client affective verbalization is
highly correlated with its respective self reference vari
able (r's = .95, .93, .85). The affective verbalization
levels are not, however, correlated with each other. Of the
self reference levels, only segment 1 and 2 are correlated,
but only modestly (r = .32). Therapist affective verbaliza
tion in segment 1 is strongly correlated with both client
measures in segment 1 (r = .66 and .67) and with no other
segment levels of either. Therapist affective verbalization
in segment 3 is modestly correlated with all levels of cli
ent affective self reference only (r's = .34, .32, .40, .38),
as well as being correlated with therapist affective verbal
ization in segments 1 and 2, (r's = .58, .34), which are not
correlated with each other. The total scores for the thera
pist measure and both client measures are modestly to highly
correlated with each other and all affective verbalization
segment variables (range of r's = .32 to .92), with the ex
ceptions of total client affective verbalization with ther
apist affective verbalization in segment 2 (r =.28, £ = .076)
and the total therapist measure with client affective self
reference in segment 3 (r = .27, jd = .089), both of which
correlations approach significance.


APPENDICES-continued Page
APPENDIX C--Notice for Potential Client
Subjects 74
APPENDIX D--Verbatim Telephone Contact for
Informing Selected Clients 75
APPENDIX E--Rules for Locating Segments Within
Full Interviews 76
APPENDIX F--Training Manual for Unitization of
Transcripts by Proposition 78
APPENDIX G--Training Manuals for Rating All
Variables 81
APPENDIX H--Listing of Primary Affect Words
and Limited Primary Affect Words. . 97
APPENDIX I--Listing of Secondary Affect Words . 99
APPENDIX J--Listing of Affective Idioms and
Slang 100
APPENDIX K--Listing of Therapist Verbal
Reinforcers 101
APPENDIX L--Examples of Graphs Plotting
Interaction Effects 102
APPENDIX M--Examples of Score Distribution
Graphs 105
REFERENCE NOTES 108
REFERENCES. 109
BIOGRAPHICAL SKETCH 116
v


TABLE 4
AFFECTIVE MEASURES INTERCORRELATED:
CORRELATIONS AND PROBABILITIES
t_n
VARIABLE
AND
VARIABLE
AND
SEGMENT
NUMBER
SEGMENT
NUMBER
CAV
CASR
TAV
1
2
3 T
1
2
3
T
1
2
3
CAV 2
.20
. 207
CAV 3
. 17
-.02
. 288
. 877
CAV T
. 75
. 56
.58
. 001*
.001*
. 001*
CASR
1
. 95
.001*
.24
. 140
. 17
.280
. 75
.001*
CASR
2
. 28
.083
.93
.001*
.01
.965
.60
.001*
. 32
. 045*
CASR
3
. 18
. 265
.11
. 495
. 85
.001*
.59
. 001*
.22
. 161
.13
.405
CASR
T
.69
.001*
.59
.001*
.45
.003*
.92
. 001*
.77
.001*
.69
.001*
.60
.001*


110
Buehler, R. E. & Richmond, J. F. Interpersonal communi
cation theory. Corrective Psychiatry Journal of
Social Therapy, 1965, 1_1, 204-216 .
Butler, J. M., Rice, L. N. & Wagstaff, A. K. On the def
inition of variables: an analogue of clinical
analysis. In H. H. Strupp and L. Luborsky (Eds.),
Research in Psychotherapy (Vol. 2). Washington D. C.:
American Psychological Association, 1962.
Campbell, D. T. & Stanley, J. C. Experimental and Quasi-
Experimental Designs for Research. Chicago: Rand
McNally and Co., 1963.
Cartwright, D. S. Analysis of qualitative material. In
L. Festinger and D. Katz (Eds.), Research Methods in
the Behavioral Sciences. New York: Holt, Rinehart,
and Winston, 1966.
Crowley, T. The conditionability of positive and negative
self-reference emotional affect statements in a coun
seling type interview (Doctoral dissertation, Univer
sity of Massachusetts, 1970). Dissertation Abstracts
International, 1970, 31, 2100A-2101A. (University
Microfilms No. 70-23,026.)
Davis, J. D. The Interview as Arena: Strategies in
Standardized Interviews and Psychotherapy. Stanford,
Calf.: Stanford University Press, 1971.
Davis, J. D., Fisher, L., & Davis, M. Therapeutic inter
ventions of clinicians in standardized interviews: a
test of the social model. Journal of Consulting and
Clinical Psychology, 1973, LI, 16-26.
Davis, J. D., Fisher, L., & Davis, M. L. Stability and
change in strategies of psychiatric patients and nor
mal interviewees. Journal of Consulting and Clinical
Psychology, 1975, £3, 205-215.
Ellsworth, S. G. The consistency of counselor feeling-ver
balization. Journal of Counseling Psychology, 1963,
10, 356-361.
Frank, G. H. On the History of the objective investigation
of the process of psychotherapy. Journal of Psychology,
1961, 15_, 89-95.
Gassner, S. M. Relationship between patient-therapist com
patibility and treatment effectiveness. Journal of
Consulting and Clinical Psychology, 1970, ~34, 408-4~14.


23
of increasing client affective verbalization and client
affective self reference will be enhanced with female
clients and not with male clients, and b) that female
clients and not male clients will be associated with
(1) client affective verbalization and (2) client af
fective self reference increasing across the session and
with (3) total client affective verbalization and (4)
total client affective self reference.
Group 3
7. that a) client affective verbalization and client affec
tive self reference will be positively associated with
session outcome, and b) higher compatibility will enhance
the positive association of client affective verbaliza
tion and client affective self reference to session out
come .


69
ther investigation of this would require either separate
samples matched on each type of compatibility component, or
an adequately-sized group constructed randomly without re
gard to compatibility, and then analyzed in terms of com
patibility measures. Economy for investigating more than
one type of compatibility probably requires the latter
approach.
Conclusion
This study grew out of a review of psychotherapy litera
ture as conceptualized within a general communications model.
The model was translated into the terms of research which
has considered therapist behaviors and their relationships
to clients' use of affective language. A study was designed
for testing these relationships in a sample of clients simi
lar to these previous studies, but using a sample of thera
pists which were different in not being confederates and
also in being practicing, professional psychotherapists.
This study was also different in including the variables com
patibility and client sex as independent variables.
This study's contributions appear to lie in two areas.
The first area is in its development of methodological tools
for the relatively objective measurement of client and thera
pist use of words and phrases denoting affective experience
both with and without reference to the self. Previous re
search which has provided similar tools has either maintained
significantly subjective elements, failed to retain materials
for use on other studies, or failed to adapt materials to


67
Analyses of the group 2 hypotheses resulted in one sig
nificant effect, the interaction of compatibility, client
sex, and segment, indicating that in terms of segment changes,
the compatibility manipulation was effective only for female
clients. This is consistent with the literature cited earlier
which also demonstrated this effect (Mendelsohn & Rankin,
1969).
Correlations between therapist evaluation and client
affective verbalization supported the contention that thera
pists believe client affective verbalization is important to
treatment outcome. Other outcome measures, client evaluation
and problem rating scale change scores, were not related to
client affective expression measures. Limited ranges of
scores for these two outcome measures were a problem and more
adequate measures should be used in future research.
Various analyses and correlations offer some help in
the problems of designing further studies to investigate the
implications of the present data. The large error variance
found so often in the analyses suggests that 1) this must be
expected and only very powerful effects can be examined,
2) the flexibility of the interview situation must be re
stricted in order to provide better control, or 3) a broader
range of clients should be studied. The other alternative
is to consider more variables, but this is probably prohibi
tive on the one hand in terms of investment of resources or
on the other, in terms of an N that is too small for so many
variables.
There is a danger inherent in increasing control in the


CLIENT AFFECTIVE VERBALIZATION
CHANGE FROM SEGMENT 1 TO 2
43
+ 10
+ 5
A
0 5 10
THERAPIST AFFECTIVE VERBALIZATION
IN SEGMENT 1
15
Legend: A 1 observation; B = 2 observations, etc.
SCATTER PLOT OF A THERAPIST
SCORE BY A CLIENT SCORE
FIGURE 1


14
CODER, and a response of an encoder which may be the stimu
lus for a decoder, the MESSAGE. COMMUNICATION results when
a response of an encoder is received as a stimulus for a de
coder. In further specifying the communication situation,
Markel identifies six channels of human "face to face" com
munication by specifying the source in the encoder response.
The channels with the source and destination for each are:
speech (vocal tract, ear); kinesics (body movement, eye);
odor (chemical processes, nose); touch (body surface, skin);
observation (body surface, eye); and proxemics (body place
ment; eye). In psychotherapy research all three of the
basic communication "entities" are important, and only three
of the channels are studied to any significant extent at
present: speech, kinesics, and proxemics. Odor appears to
be studied the least, with observation and touch gaining
attention. Speech is by far the most studied of all the
channels.
Speech is divided in a number of ways which are impor
tant in specifying the way this study is going to deal with
communication. Morris (1946) divides the science of lan
guage into three aspects: SYNTACTIC, formal relations of
signs to each other; SEMANTIC, relations between signs and
the objects to which the signs are applicable; and PRAGMATIC,
the relations between signs and users of signs. This study
will consider aspects of each of these areas.
Linguistics is one of the largest areas of communica
tion science. Morris (1946) defines it as the descriptive
study of the formal characteristics of language, and classi-


32
SITIONS obtained, error rates of 7.0%, 3.1%, and 4.5%
respectively.
Therapist Strategies. The three therapist strategy
behaviors were scored through a similar use of content an
alysis. Instruction was defined as a therapist statement
or statements within the first 3 minute Segment of the in
terview in which the therapist communicates at least two
of three things: 1) behavior the client is to include
(i.e., "talk about your feelings"), 2) behavior the client
is to exclude (i.e., "Don't talk exclusively about your
thoughts"), and 3) examples of the general classes or areas
desired to be talked about (i.e., "YOU might talk about
your satisfactions and dissatisfactions, your concerns and
confidences"). Instruction was scored using the unit:
PROPOSITION, the categories: Instruction-1, Instruction-2,
Instruction-3, and the indicator: unit includes defined
communication. The rater's task on this rating was to con
sider each PROPOSITION within the transcript for an initial
3 minute Segment, and to label any Instruction categories
found by recording that PROPOSITION'S code number(s). Since
this rating task was somewhat less precise than the earlier
tasks, two non-experimenter raters were used, and the ex
perimenter's rating was to be used to settle disagreements.
The raters were trained using the training manual in Appen
dix G. Only one component of this strategy was found by
raters, and no reliability assessment would be meaningful.
This was extended to all three segments, and there was still
only one found.


93
"You might talk about subject X" where X is from one of the
three lists of words and phrases attached to this manual.
Rules for assigning to this category are: 1) The PROPOSI
TION is in the imperative or second person conditional mood.
2) The PROPOSITION contains some form of one or more words
from the attached list of words and phrases. 3) The PROP
OSITION contains as a predicate one or more of the follow
ing verbs: begin, report, talk, tell, say, speak, and think.
4) The general meaning is as above.
Here are some examples. Note that 1, 2, and 3 would
be assigned to those respective categories, and that 4 and
5 would
not be assigned to any
category.
Ex.
1:
Tell me how you feel.
Ex.
2:
Don't talk about your
relatives.
Ex.
3:
You might talk about anger or sadness.
Talk about anger or sadness.
Ex.
4:
How were you feeling?
Ex.
5:
Don't talk about your
feelings.
Now try this part of a transcript, putting the category num
ber to the left of any statements containing scored PROPOSI
TIONS. Remember that more than one PROPOSITION may need to
be grouped to assign it to a category.
T: /How are you feeling?/
C: /Okay. /I'm really worried about this thing./
T: /Talk about your feelings about it./
C: /I have to go take a test, /and if I don't pass it/
/I'll have to take it over./


96
C: I like football. I feel great when I'm playing.
T: Mm-hmm.
C: I play every day.
T: Oh?
If you have questions, discuss them with the experimenter,
when he checks your practice rating.


I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
lfacquelin R. Goldman, Chairperson
Professor of Clinical Psychology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Util iy *J>- /\ cl- A/--*-
Audrey 6. Schumacher
Professor Emeritus of Psychology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Marvin E. Shaw
Professor of Psychology


16
Defining Communication of Affect
A number of studies have utilized measures of verbali
zation of affect words, though with significant variations
in its definition. Levy (1967) studied the "communication
of affect" and defined it in terms of the verbalization of
words from a list of affect words which he developed from
analytic interviews. Moos and Ciernes (1967) used "percent
age of feeling words" as one dependent variable and defined
it by "enumerating examples of words directly relevant to an
individual's affective state" (p. 121).
Most other studies using this measure have been inves
tigating the effects of various therapist behaviors on cli
ent verbalization of affect within operant conditioning para
digms. Here too the specific definition used has often var
ied. The most popular definition has also included the sti
pulation that the statement be a self reference and is ac
cordingly called an "Affective Self Reference." Salzinger
and Pisoni (1958, 1960) appear to have made the first formal
use of this definition, which specifies that the response
class statement must be an "I" or "We" statement describing
or evaluating the client's own state, except for intellectu
al or physiological, and providing some rules for instances
where the judgement is unclear. A number of other studies
have used this definition or small modifications of it
(Hekmat, 1971a, 1974; Hekmat & Lee, 1970; Hoffnung, 1969).
"Affective Self Disclosure" has been used as the response
class by Hekmat and Thiess (1971) and Hekmat (1971b). It


53
TABLE 8
CORRELATIONS BETWEEN NUMBER OF
REINFORCEMENT RESPONSES AND
CLIENT CHANGE SCORES
SEGMENT
CHANGE
NUMBER OF THERAPIST REINFORCEMENTS
BY SEGMENT
1
2
3
TOTAL
CLIENT
AFFECTIVE
VERBALIZATION
1 to
2
-.01
-.03
-.11
-.05
2 to
3
-.28
-.20
-.13
-.25
1 to
3
-.33
-. 26
-.26
-.33*
CLIENT
AFFECTIVE
SELF REFERENCE
1 to
2
.11
. 11
.08
.12
2 to
3
-.45**
-.34*
-.28
-.43**
1 to
3
-.38*
-.26
-.22
-.34*
* £<.05
** £

64
relations were significant, though no particular patterns
were apparent. All compatibility variables had at least one
significant correlation except for reciprocal compatibility
on Inclusion and Affection and interchange compatibility on
Inclusion. All significant correlations with affective ver
balization measures were for second segment measures and
with originator compatibilities. Originator compatibilities
for inclusion and for affection were negatively correlated
with therapist affective verbalization (r's = -.35, -.34)
and originator compatibility for control was correlated with
client affective self reference (r = .34). Reciprocal com
patibility for control was correlated with client affective
verbalization change from segment 2 to 3 (r = -.37), and in
terchange for affection is correlated negatively with client
affective self reference change from segments 2 to 3 and
from 1 to 3 (r's = -.36, -.32). Interchange for affection
is also correlated positively with Reinforcement in segment
1, 2, and total (r's = .38, .34, and .37). Originator com
patibility for control is correlated positively with total
Reinforcement (r = .32). Interchange for control is corre
lated positively with the Problem Pathological Potential
Scale change score (r = .31).


Limited Primary Affect Words
ATTRACTED = LIKE
BLUE = SAD
BREATHLESS = THRILLED
BUGGED = BOTHERED
CENTERED = EMOTIONALLY SETTLED
COOL = CALM
CROSS = SULKY
DIG = LIKE
DISLIKE = EMOTIONALLY UNATTRACTED
DOWN = DEPRESSED
ELECTRIFYING = STIMULATING
FLUSHED = BLUSHED
HEARTY = VIGOROUS
HIGH = EUPHORIC
HURT = EMOTIONALLY INJURED
IMPRESSED = AFFECTED
LIKE = ATTRACTED TO
MOVED = EMOTIONALLY AFFECTED
RAVING = ANGRY
RELISHED = ENJOYED
REPELLING = REPULSING
SEETHING = ANGRY
SHOCKING = AFFECTING
SMARTING = HURTING EMOTIONALLY
SPEECHLESS = SHOCKED
STEWED = IRED
TEMPER = ANGER
TENDER = AFFECTIONATE
TIRED = BORED
TOGETHER = EMOTIONALLY SETTLED
TOUCHED = AFFECTED
WARM = EMOTIONAL
WINCED = EMOTIONALLY HURT


20
Behavior, Schtz, 1966) is most used for assessing this vari
able, and has been demonstrated to be a reliable and valid
indicator of interpersonal compatibility on a number of meas
ures. Of the three interpersonal needs postulated by the
FIRO scales, the most important for the present formulation
is the need for inclusion and control, and of the various
compatibility computations, that of reciprocal need compati
bility is most appropriate. In formulating hypotheses about
the use and effectiveness of the above therapist strategies
the combination of reciprocal compatibility for Inclusion
and Control of the therapist-client dyads will be considered.
One study (Mendelsohn & Rankin, 1969) has indicated
that sex may also be an important variable. It was found
that the compatibility measures work well in some ways with
female clients, but poorly with male clients. Sex of thera
pist did not appear to be an important factor in the differ
ences. Therefore, sex of client will also be considered as
a variable in assessing the effects of the compatibility
measure.
In summary then, client affective verbalization and cli
ent affective self reference are considered important vari
ables, and three therapist strategies have been shown to be
effective in increasing the frequency of occurrence for both
these variables in quasi-therapeutic situations. Previous
studies have apparently not examined the relationship of
variables in situations more closely resembling actual thera
pists who have not been specifically instructed to engage
in the strategy behaviors. The theory and literature re-


7
the client's expression of affect as a positive step in
reaching therapeutic goals, either, for example, in client-
centered therapy as a major process, or in psychodynamic
therapy as a cathartic and insight-generating process.
This control of style is most clearly a control of the
definition problem when the client does not exhibit the de
sired behavior initially, as is usually the case. The thera
pist wants the relationship to be one in which it is accept
able for the therapist to respond in compatible ways (e.g.,
reflecting affect, making interpretations about the affect,
etc.). If the client interacts this way naturally there is
no problem, and the therapist and client get to business
rather quickly. If the client does not, the therapist must
help arrange for this to happen. The way this is attempted
will be referred to as the therapist "strategy."
To summarize the rationale to this point, Haley's model
describes the interaction between therapist and patient as
like that of any dyad and consisting of the processes of de
termining the definition of the relationship and competing
for control of that definition. It is through these pro
cesses that the interpersonal influence of therapy takes
place.
One implication of Haley's theory as it is being con
sidered is that the therapist is expected to work toward im
plementing the therapeutic goals as the definition of the
relationship, insofar as they apply. In developing a re
search examination of this theoretical position, a specific


Ill
Gottschalk, L. A. & Gleser, G. C. The Measurement of Psy
chological States Through the Content Analysis of Ver
bal Behavior. Berkeley: University of California
Press, 1969.
Gottschalk, L. A., Winget, C. N., & Gleser, G. C. Manual of
Instruction for Using the Gottschalk-Gleser Content
Analysis Scales. Berkeley: University of California
Press, 1969.
Green, A. H. & Marlatt, A. Effects of instructions and
modeling upon affective and descriptive verbalization.
Journal of Abnormal Psychology, 1972, 8JD, 189-196.
Greenhill, M. H. The focal communication concept. American
Journal of Psychotherapy, 1958, 1_2, 30-41.
Guetzkow, A. Unitizing and categorizing problems in coding
qualitative data. Journal of Clinical Psychology,
1950, 6, 47-58.
Haley, J. Strategies of Psychotherapy. New York: Grue
and Stratton, 1963.
Haley, J. Research on family patterns: an instrument
measure. Family Process, 1964 3^, 41-51.
Harper, R. A. Psychoanalysis and Psychotherapy: 36 systems.
Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1959.
Hekmat, H. Reinforcing values of interpretation and reflec
tions in a quasi-therapeutic interview. Journal of
Abnormal Psychology, 1971a, 77, 25-31.
Hekmat, H. Extraversin, neuroticism, and verbal condition
ing of affective self disclosures. Journal of Counsel
ing Psychology, 1971b, 1_8, 64-69.
Hekmat, H. Three techniques of reinforcement modification:
a comparison. Behavior Therapy, 1974, 5^, 541-548.
Hekmat, H. & Lee, Y. B. Conditioning of affective self-ref
erences as a function of semantic meaning of verbal
reinforcers. Journal of Abnormal Psychology, 1970, 76,
427-433.
Hekmat, H. & Thiess, M. Self actualization and modification
of affective self disclosures during a social condition
ing interview. Journal of Counseling Psychology, 1971,
18, 101-105.
Heller, K. Experimental analogues of psychotherapy: the
clinical relevance of laboratory findings of social
influence. Journal of Nervous and Mental Disease, 1963,
137, 420-426^ ~


41
this sample, and this leaves Modeling and Reinforcement to
be considered. The analyses of hypotheses 3 and 4 for Model
ing will be presented first.
Since these hypotheses are dealing with changes in
client behaviors, the Modeling score was correlated with
scores representing the change of a client's behavior.
Accordingly, correlations were computed between the various
segment levels of therapist affective verbalization and the
change scores for the two client measures across various
parts of the interview (see Table 3). For client affective
verbalization, one correlation was significant, this between
therapist affective verbalization in the first segment and
the change in client affective verbalization from segment 1
to 2, but in the opposite to predicted direction (r = -.44,
^ = .004). The correlation between therapist affective ver
balization in the first segment and client affective ver
balization change from the first segment to the last ap
proached significance, again in the opposite to predicted
direction, (r = -.29, £ = .067). Therapist affective ver
balization correlations with client affective self reference
change scores showed the same phenomenon with both correla
tions significant (r = -.46, p = .003; r = -.41, p = .009).
A scatter-plot of the segment 1 therapist affective verbal
ization by segment 1 to 2 client affective verbalization
change scores shows a clear association between higher
therapist scores and negative client change scores and be
tween lower therapist scores and positive changes (see
Figure 1). This suggests the need for a clearer under-


84
Ex. 1: a) He is angry. b) His anger is strong.
c) He spoke angrily, d) He angered them.
Ex. 2:
She is a moody person, who is really weird.
Ex. 3:
I'm tired and depressed.
Ex. 4 :
a) I was really hurt when she left.
b) I was really hurt in the accident.
Ex. 5:
a) I feel like I've been through a wringer.
b) I feel like he's a real dummy.
Notes:
Ex. 1:
all forms of "angry" are marked as PAWs: adjective,
noun, adverb, and verb.
Ex. 2:
"Moody," but not "weird" is found on the list of
PAWs.
Ex. 3:
"Depressed" but not "tired" is found on the list.
Ex. 4 :
"Hurt" is scored only when it has the limited mean
ing, emotionally injured and not physically injured.
Ex. 5:
In "a" "feel" denotes emotional experience, but in
"b" it denotes thinking.
Now try this reproduction of part of a transcript,
circling each WORD you find on the list of PAWs.
T: But you have hurt them.
C: Sure. Whenever you fight a lot you always hurt some
body, particularly a child.
T: You sound a little bit guilty.
C: I feel guilty because; it's not my kid's fault. She
has nothing to do with it.
If you have any questions, discuss them with the experi
menter when he checks your practice rating.


Abstract of Dissertation Presented
to the Graduate Council of the University of Florida
in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy
CLIENT AFFECTIVE VERBALIZATION
AND THERAPIST INFLUENCE STRATEGIES IN
ANALOGUE PSYCHOTHERAPY INTERVIEWS
By
Stanley E. Jones
December 1976
Chairperson: Jacquelin R. Goldman
Major Department: Psychology
Psychotherapy theory emphasizing interpersonal interac
tion is used as a theoretical base for considering strate
gies that therapists might take to increase client affective
verbalization in therapy situations. Previous research of
analogue interviews is used for defining the strategies,
Instruction, Modeling, and Reinforcement, and for defining
the dependent measures, client affective verbalization and
client affective self reference. Professional therapists
conducted single interviews with experimental subjects,
after the dyads were matched on interpersonal need compati
bility and on all sex combinations. A rating system for
determining affective verbalization and affective self ref
erence was developed and applied to typed transcriptions of
the interviews. Therapist strategies were also measured by
vii


CHAPTER 4
RESULTS
Results are presented in terms of the three groups of
hypotheses discussed above and labeled: client verbaliza
tion and therapist strategies, compatibility and client sex,
and client verbalization and session outcome. When one ana
lysis is used for more than one hypothesis, it is presented
to the extent needed to examine the results for that hypothe
sis. When it is apparent that a result is markedly different
from the hypothesis and an alternative hypothesis is evident
by virtue of being the converse of the predicted relationship,
that alternative hypothesis is presented in this section
with the analyses used to examine it.
Client Verbalization and Therapist Strategies
The first group of hypotheses (numbers 1 through 4)
dealt with therapist strategies and client affective verbali
zation and client affective self reference. Each of these
variables was measured as described in Chapter 3. With re
spect to the therapist strategies, no full instances of the
Instruction strategy were found, and only one type 3 compo
nent of that strategy (see Appendix G) was scored in the
rating of all therapist responses. Consequently, this vari
able was dropped from further analyses. Also, although ther-
36


10
client could be said to be in control. Note that the focus
here is on the therapist's belief about what client behavior
is therapeutic, and in studying the therapist's process, it
is important to look at client behaviors that are believed
to lead to the occurence of change and desired to be a char
acteristic of the client early in therapy.
For example, almost all psychotherapists would see "the
client verbalizes" as such a goal. Except for the rare case,
the client's talking is a pre-condition to the occurrence of
therapy change. A related client behavior is that when the
client talks the client talks about affect. This notion,
that the client's verbalization of or about affective states
is important to therapy, appears to be shared by many sys
tems of therapy, either through the direct theoretical state
ment of the therapist or through the implication of the
goal that is set (e.g., see Harper, 1959). Even when this
is not a primary end, as it is in client-centered therapy
for example, it is frequently a secondary end allowing the
therapist and client to gain the information they need to
proceed with other goals, as for example in rational-emotive,
or various psychoanalytic therapies. This client behavior,
the verbalization of affect, will be the therapist goal in
this study.
Research has often focused on the variable, "client ex
pression of affect," and operationally defined it in terms
of judges' ratings (e.g., Waskow, 1962; Ellsworth, 1963) or
in terms of systems making inferences about the client's


APPENDIX F
TRAINING .MANUAL FOR UNITIZATION OF
TRANSCRIPTS BY PROPOSITION
Your role in this study is to assist in the prepara
tion of transcripts of interviews for scoring by other
raters like yourself. It is most important that your work
be done accurately, and it will require that you concentrate
well on it. This training manual will help you learn the
tasks, and generally, the tasks are specific and should not
be too difficult. Your task is to go through the transcript
and to use diagonal marks to separate phrases defined below
as PROPOSITIONS.
A PROPOSITION is defined as a sequence of words in the
transcript containing a subject and predicate, whether ex
pressed or implied, and including the modifiers of that sub
ject-predicate group. This is essentially the definition
of a clause, which may be either coordinating (or indepen
dent) or subordinating (or dependent). Any clause is
scored as a PROPOSITION with two additional stipulations.
Parenthetical clauses, clauses which come in the middle of
a larger clause, are not scored as separate PROPOSITIONS
but are considered PARENTHETICALs within a PROPOSITION.
Also, incomplete clauses, as, for example, when a person
changes the message mid-sentence, are scored as PROPOSITIONS.
The following suggestions will help in making these
scorings. In the Harbrace College Handbook, read the def
inition of conjunction, p. 446, and the discussion of con
junctions, p. 12-13. Conjunctions are, of course included
78


22
tegies and each other. The third group deals with the re
lationships between client affective verbalization measures
and session outcome measures, with consideration given to
the possible role of compatibility. The hypotheses are
formally stated and labeled below.
Research Hypotheses
It is hypothesized:
Group 1
1. that each therapist uses one or more of the above strat
egies;
2. that a) client affective verbalization and b) client af
fective self references will increase across the inter
view;
3. that greater use of a strategy is associated with larger
increases in client affective verbalization;
4. that greater use of a strategy is associated with larger
increases in client affective self reference;
Group 2
5. that a) higher compatibility will be associated with en
hancing the positive effect of the therapist strategy
behaviors, and b) that higher compatibility of the thera
pist-client dyad will be associated with (1) client af
fective verbalization and (2) client affective self ref
erence increasing across the session, and with (3) total
client affective verbalization and (4) total client af
fective self reference;
6. that a) the positive effect of the therapist strategy


62
ment 2 from the multiple variables correlating with it in
segment 1 makes this finding particularly interesting, since
it raises the possibility that therapists do assess the help
fulness of an interview in terms of client affective expres
sion. This is only in the context of other variables, which
were not identified in this study, as the variance accounted
for in those correlations is less that 14%.
Part b of hypothesis 7, predicting an improvement in
the above correlation when considering compatibility, was
examined by computing the correlation of client affective
verbalization total and for segment 2 with therapist evalu
ation separately for each level of compatibility. This hy
pothesis was supported as the correlations for high compat
ible dyad clients are r = .697 and .594, while in the low
compatible clients, these correlations are .413 and .221 re
spectively. Both differences are significant at the .05
level using a one-tailed test.
Exploratory Analyses
The significance of compatibility and client sex has
been supported in this study, and additional exploratory
analysis was undertaken. One aspect of this was to explore
further possible associations in the data between compati
bility, sex of client, and sex of therapist. This was done
through the computation of Spearman correlations with all
affective verbalization measures, the Reinforcement measures,
affective verbalization change scores, and the three outcome
measures. Two modest correlations were significant, between


91
C: I feel guilty because; it's not my kid's fault. She has
nothing to do with it.
If you have any questions, discuss them with the experimen
ter when he checks you practice rating.


49
each pair had used for affective verbalization scores and
then to determine whether the client had used that word, and
if so whether before or after the therapist response and
whether immediately or not. Immediately is defined here as
being within the first statement following or preceding the
one containing the affective verbalization.
The sample was chosen randomly, with the stipulation
that each therapist must be chosen once and that there be
four dyads chosen from each segment. Table 6 summarizes the
results of the tabulation. It can be seen from the table
that although the total number of responses in each segment
does not vary greatly, the distribution of the responses in
terms of their relationship to client responses using the
same word or phrase is quite different. In segment 1, ther
apists could be described as "following" the client response.
In segment 2, clients could be described as "following" the
therapists. In segment 3, there is no apparent relationship
between the particular affective words and phrases used by
the therapists and clients. The sample is of course small,
and should be interpreted conservatively, but suggests a
trend toward differential leading and following therapist
styles in different segments.
Overall, it must certainly be concluded that the data
clearly disconfirm the third and fourth hypotheses, with
respect to the Modeling strategy.
Reinforcement is the second strategy to be considered
in hypotheses 3 and 4. Here, correlations were obtained be
tween segment levels of the Reinforcement variable and cli-


APPENDIX J
LISTING OF AFFECTIVE
IDIOMS AND SLANG
ABIDE
BANG
BESIDE
BLOW
BOTTOM
BURNS
CARE
CARE
CHARGE
CONSUMED
CREEPS
CREEPS
EATS
EATS
EDGE
ENDURE
ENTER
FED
FORWARD
FREAKED
FUCKED
HAIR
HEAD
HEART
HEAVY
KICKS
LOOK
LUMP
PEACE
PINS
PISSED
SICK
SCREWED
SEIZED
SOUL
STAND
TAKE
TAKE
TASTE
TEAR
THINK
TICKED
TIRED
TURN
TURNS
VEGETATES
WELL-BEING
WRONG
(can't abide someone or something)
(get a bang out of)
(be beside oneself)
(blow one's mind)
(from the bottom of one's heart)
(burns one up = angers)
(care for someone or something)
(care nothing for someone or something)
(get a charge out of)
(consumed with something = zealous)
(flesh creeps)
(gives me the creeps)
(eats away at one)
(eats one's heart out)
(be on edge)
(can't endure)
(enter the spirit of)
(fed up)
(looking forward to)
(freaked out)
(fucked up)
(hair stands on end)
(head over heels)
(heart and soul)
(heavy experience)
(do it for kicks)
(look up to)
(lump in the throat)
(peace of mind)
(on pins and needles)
(pissed off, be pissed)
(sick and tired of)
(screwed up)
(seized with emotion)
(soul stirring)
(can't stand)
(can't take it anymore)
(take to someone = attracted)
(have a taste for; or not for)
(tears one up; all torn up)
(think much of someone; or little of)
(ticked off = mad)
(tired of living)
(person turns color; turn green, black, pale)
(turns one on)
(vegetates one; vegetates one out)
(sense of well-being)
(feel something's wrong)
100


38
significant segment effect in an analysis of variance.
This was part of a larger analysis presented in the Group
2 hypotheses. For both client affective verbalization (see
Table 1) and for client affective self reference (see Table
2), there was no significant segment effect, and in both
cases, the trend was for the first segment to be the highest
of the three segments and for the second segment to be
lowest.
Hypotheses 3 and 4 predicted that there would be a sig
nificant positive relationship between the greater use of
a therapist strategy and larger increases in client affec
tive verbalization and client affective self reference.
Since the therapist strategy variables occur within the same
interactions which also provide the data for the client
measures, statistics presuming independence were inappro
priate. Consequently the associations predicted to occur in
the data were analyzed by examining the correlations in the
data. To determine the significance of a correlation for
interpretation purposes, the correlation was compared to the
null hypothesis that the correlation equals zero, with
alpha = .05 for rejecting the null hypothesis. A correla
tion was defined as approaching significance if its signif
icance level fell below .09. Unless otherwise noted corre
lations are Pearson r's. Spearman correlation coefficients
were used in instances where comparisons were to be made
which involved class variables (i.e., compatibility, sex of
client, and sex of therapist).
As indicated above, some strategies did not occur in


17
differs from the Affective Self Reference in using only "I"
statements and considering affective expressions only if
they have the form: "I feel....," with an expression of af
fect. Other definitions have included the "Feeling State
ment" (Green & Marlatt, 1972) with extensive rules, and two
closely related definitions, the "Emotional Words" measure
(Merbaum & Lukens, 1968) and the "Client Affective Word"
measure (Barnabei, Cornier & Hye, 1974; Crowley, 1970).^
Variations, then, have included whether or not there
was a self-reference required and whether that included state
ments beginning with we; whether the affective expression
was positive or negative or unscored for direction; whether
only certain types of affective statements were considered
or certain types excluded; whether both therapists and cli
ents were scored; and whether statements were required to be
with reference to present time or not (immediacy). The pre
sent study will use a definition modeled after Green and
Marlatt's rules (Note 1), but reducing the amount of judge
ment to be exercised by raters by using a more extensive
listing of affect words, and reducing the extent to which
evaluation is included as feeling expression. In order to
increase the applicability of this study to other studies,
statements will be scored both with and without a self ref
erence required.
The inference is being made that a behavior like client
^Psycho-Die, a content analysis dictionary, appears to have
the capability for defining this measure in any of these
ways, but no studies using it for this purpose were found.


95
Training Manual: Therapist Reinforcement (R)
In scoring Rs, the unit is the WORD with certain stip
ulations, the category is R, and the indicator (or rule) is
that the WORD or WORDS are found on the attached list of
therapist reinforcers. WORD is defined as the basic block
of letters constituting units within the transcript and is
set off at its beginning and end by a space or punctuation.
The stipulations are that there may be one or two WORDS in
the unit and that they must occur at the beginning of a
therapist statement as defined earlier. The category is in
dicated by entering a code number for the WORD or WORD pair
scored, and labeling the scoring page R. The code number
is composed of the page label plus the statement number.
Generally, the words on the list are brief expressions
of approval, agreement, or attention by the therapist. Spe
cifically, the indicator is that the unit is included on
the list attached. Examples are "good," "mm-hmm," or "yes."
(Read over the list at this time.)
Here are some examples:
Ex.
1:
I see.
You really think that's the way it is
Ex.
2:
Mm-hmm.
I see.
Ex.
3:
That's
terrific.
Now try this reproduction of part of a transcript, cir
cling each R WORD or WORDS.
C: I feel great.
T: Good.


8
therapeutic goal and specific therapist interventions for
increasing the occurence of the goal behavior will be con
sidered. In order to clearly indicate the relevance of the
interventions, the goal behavior will be discussed first.


80
C: Well that's just how it affected me. Like suddenly I
felt like I was just free.
T: So she is after you pretty frequently.
C: Yep.
T: Kind of asking you to shape up.
C: Well, it just seems like a constant hassle trying to
account for places to go and things I do and making all
the normal trivial talk that really doesn't make any
difference. I'd just as soon skip anyway. It's a
bother.
If you have any questions, discuss them with the experi
menter when he checks your practice rating.


APPENDIX H
LISTING OF PRIMARY AFFECT WORDS
AND LIMITED PRIMARY AFFECT WORDS
Primary Affect Words
ABHORANT
DREADING
AFFECTIONATE
AFRAID
EAGER
AGHAST
EARNEST
AGITATED
ECSTATIC
AMAZED
ELATED
ANGRY
EMBARRASSED
ANGUISHED
EMOTIONAL
ANXIOUS
ENJOY
ANNOYED
ENRAGED
APPALLED
ENRAPTURED
APPREHENSIVE
ENTHUSED
AWED
ENVIOUS
EXCITED
BASHFUL
EXHILARATED
BITTER
EXHUBERANT
BLUSHED
BORED
FANATICAL
BOTHERED
FEARFUL
FEELING
CALM
FELICITOUS
CHEERFUL
FERVENT
CHERISHED
FIDGETY
COMPOSED
FLUSTERED
CONCERNED
FRETFUL
CONFIDENT
FRIGHTENED
CONFUSED
FRUSTRATED
CONSTERNATION
FURIOUS
CONTENTED
CRUEL
GAY
CURIOUS
GLAD
GLEEFUL
DEJECTED
GLOOMY
DELIGHTED
GROUCHY
DEPRESSED
GUSTO
DESPERATE
DESPONDENT
HAPPY
DETEST
HATEFUL
DISAPPOINTED
HORRIFIED
DISGUSTED
HOSTILE
DISCONTENTED
HYSTERICAL
DISCOURAGED
DISENCHANTMENT
IMPETUOUS
DISILLUSIONED
IMPRESSED
DISMAYED
INCENSED
DISPLEASED
INFURIATED
DISSATISFIED
INSPIRED
INTIMIDATED
SAD
IREFUL
SATISFIED
IRRITABLE
SCARED
IRRITATED
SEDATE
SENSUOUS
JEALOUS
SENTIMENTAL
JITTERY
SERENE
JOLLY
SHY
JOYFUL
SINCERE
JOYOUS
SORROWFUL
SQUEAMISH
LIGHTHEARTED
STARTLING
LONELY
SUFFERING
SULLEN
MAD
SURPRISED
MALEVOLENT
MEEK
SYMPATHETIC
MISERABLE
TENSE
MISGIVING
TERRIFIED
MISTRUST
THRILLED
MOODY
TIMID
TRANQUIL
NERVOUS
OPTIMISTIC
TREPIDATION
TRUSTING
OVERWHELMED
UNEASY
UNHAPPY
PANICKY
UNNERVED
PASSIONATE
UNPERTURBED
PEACEFUL
UNRUFFLED
PEPPY
PERTURBED
UPSET
PESSIMISTIC
VEHEMENT
PIQUED
PITY
VIGOR
PLACID
WISTFUL
PLEASANT
PROUD
WORRIED
ZEALOUS
RAPT
RAPTUROUS
RELAXED
REPOSED
REPULSED
RESENTFUL
RESTFUL
RESTLESS
ZESTY
97


CHAPTER 5
DISCUSSION
The primary focus of this study was on defining client
affective verbalization and client affective self reference,
on defining therapist strategies which previous research has
indicated are effective in increasing the above client meas
ures across an analogue interview, and on examining actual
therapists in experimental interviews with respect to their
use of these strategies and the effects resulting from the
use of these strategies. Results of the study indicate that
the defining functions of the study appeared adequate, but
that therapists in this sample do not use the defined strata
gies, in any systematic form at least, and that the client
measures do not change consistently across the interviews.
Although there was a lack of significant changes in cli
ent scores across the interview, there was a trend toward
changes being important in accounting for the variance in
the sample. The variation in segment changes was large
within groups of the treatment blocks and suggests that im
portant variables were not identified.
Not only did the results not support therapists as
utilizing the predicted strategies, but there was also evi
dence suggesting that clients are more in control of the
amount of affective verbalization than are therapists. This
65


57
example, means that therapist affective verbalization scores
determine the groups. These groups, however, also split un
evenly with respect to client affective verbalization scores
If the scores were determined independently, then we could
interpret that as a significant effect. In the present
case, it only means there are effects of several variables
contained in any examination of the variable "Modeling."
Examination of effects of compatibility and client sex
without reference to therapist strategy could be appropriate
ly handled by analysis of variance. The normalities of
the distributions of client affective verbalization and
client affective self reference were plotted and a square
root transformation for scores including low scores was ap
plied to analysis of variance data as suggested by Kirk
(1968). (For examples, see Appendix M.) The transformation
applied was
x' = (x)l/2 + (x + 1)^/2.
Four analyses of variance were computed. The first two
were split-plot factorial designs with two between block
treatments (compatibility and client sex) and one within
block treatment (segment or repeated measurements), using
client affective verbalization as the dependent measure in
analysis of variance 1 and client affective self reference
in analysis of variance 2. The second two analyses were
2X2 factorials (compatibility and client sex), using
client total affective verbalization as the dependent mea
sure in analysis of variance 3 and client total affective


50
TABLE 6
TABULATION OF FOLLOWING VERSUS
LEADING THERAPIST RESPONSES
SEGMENT
TYPE OF
RESPONSE
1
2
3
IMMEDIATELY3 AFTER
CLIENT
0
0
0
DELAYED13 AFTER
CLIENT
5
0
0
IMMEDIATELY BEFORE
CLIENT
0
4
0
DELAYED BEFORE
CLIENT
0
2
1
NO SAME CLIENT
RESPONSE
0
2
5
TOTAL
5
8
6
Note: Responses are therapist affective verbalizations,
and tabulations reflect the type of relationship to
client's use of the same word or phrase within a given
Segment of an interview.
a "Immediately" is defined as the statement right before
or right after the statement containing the affective
expression.
b "Delayed" is defined as any position within the same
Segment other than "immediately" before or after.


12
vant and meaningful general constructs rather than a prob
lem of experimental error. A particularly critical example
of this potential error in this study, as suggested above,
is that of confusing the variable "client verbalization of
affect words" with the more general variable "client affec
tive expression." The first refers to characteristics of
the denotative meaning of words and idioms used by the cli
ent while the latter refers to inferences made about the
state of the client in using whatever language he uses. In
this study, then, it will be important to recognize that the
data, and inferences made from it, are based on considera
tions of the kinds of language used by a person and not nec
essarily on whether or not that person is in touch with af
fective experience.
Although there is a significant amount of literature
dealing with verbalization of affective words as a dependent
variable there appears to be little that has examined its
validity as an indicator that important aspects of therapy
are occurring when it occurs or as a predictor of successful
outcomes. Levy (1967) provided support that in statements
where judges agreed therapy was occurring in primarily anal
ytic sessions there was a greater proportion of statements
with affect words than in statements where therapy was judged
not to be occurring. Isaacs and Haggard (1966) found that
therapist attention to patient affective verbalizations re
sulted in the patient's responses having an increased affec
tive orientation. Braaten (1961) demonstrated that increas
ing expression of feeling with reference to self as opposed


26
and told that their task was to select a real problem that
he or she faced and would be willing to discuss with a pro
fessional therapist, with the session being videotaped.
Therapists had been instructed that they would be con
ducting a single interview with each of four college stu
dents who had responded to a study entitled "An Analogue
Study of Counseling," and who had been instructed by the
experimenters to think about and select a real problem that
he or she was facing and would be willing to talk about
with a professional therapist. The therapists were also
informed that the interview would be videotaped and that
provisions had been made with the campus mental health
services for referral of the client person if the therapist
felt it was indicated.
At the time of the interview, the therapist and client
completed the paper and pencil rating sheets, including for
the client a pre- and post-Problem Pathological Potential
Scale and a post-evaluation form, and for the therapist a
post-evaluation form.
The interview was introduced by the experimenter oper
ating the videotape equipment, and he or she informed the
therapist and client jointly that they would have 50 minutes
to talk and that the experimenter would knock on the door
of the interview room at the end of 45 and 50 minutes. The
interview was videotaped by the experimenter, with the pre
sent author accounting for about 25% of the procedure.
Three minute segments were retaped by the author and
two collaborators onto master video-tapes and cassette audio-


CHAPTER 2
DEFINING THE RESEARCH QUESTION
The general research question of the present study was
stated above as: "How does the influence of therapy occur?"
In the present formulation, a more specific question becomes:
"How does the therapist behave so as to attempt instituting
the therapy goal behaviors as a definition of the relation
ship?" The first step in operationalizing this question for
the present study is selecting and defining a specific ther-
apy goal for study.
As used in this paper, a "therapist goal" is defined
as a behavior which may be engaged in by the client and which
is believed by the therapist to be related to the occurance
of therapy. "Therapy" is defined as a client interacting
with a therapist and working toward a change in the initial
behavior, feelings, or belief system of the client, with
that change in the direction of from less to more desirable
as defined by the client, by society, or by the therapist.
"Mutual definition of the relationship" is defined as the
statement of acceptable behaviors desired by the therapist
and exhibited by the client. Here we are considering only
the case where the therapist is in control. If the client
is not exhibiting the desired behaviors and the therapist
stops making attempts to influence the client to do so, the
9


66
was most marked in the first segment where therapists used a
following style, with a trend toward therapists assuming a
more directive role in the second segment. Additional in
vestigation of this phenomenon might prove helpful in clari
fying these issues.
It is possible that in this sample a ceiling effect oc
curred where clients' affective verbalization rates were so
high from the beginning of the interview that it did not ne
cessitate therapists using strategies to increase that fre
quency. The data indicates that the rates were substantial,
and that therapist interventions of higher modeling or rein
forcement behaviors could only have inhibited client behaviors.
Well controlled and highly specific strategies, as used in
previous research with confederate therapists, might increase
rates even in this sample of clients, but such an increase is
unlikely when therapists use relatively unspecific strategies
and when they are possibly content with the level of client
affective verbalization as emitted. The client sample in
this study was similar to that used in previous research
(college students), but given the differences in inclinations
of the therapists not to work toward increasing reasonably
high rates of verbalization, a less articulate population of
clients may have elicited different strategies from these
same therapists. The relationship between this study and pre
vious research on these therapist strategies might be clari
fied by another study utilizing such a sample of clients or
perhaps more easily, by utilizing confederate therapists in
the research format of the present study.


47
The lack of correlation within segment levels of each
of the client measures is consistent with the lack of a
segment effect in the earlier analyses, and the high corre
lations between so many variables suggests an alternative
hypothesis. This is that clients are influencing the
therapists' frequencies of use of affective verbalizations,
or equivalently, that in at least segment 1, therapists are
using a following style which results in a high correlation
between client and therapist affective verbalization scores.
Two analyses were completed to examine these alternative
hypotheses: cross-lagged panel correlations and tabulating
therapist and client use of the same affective expressions.
Cross-lagged panel correlation (Campbell & Stanley,
1963) would predict that the "causal" force (if other than
a separate underlying factor) in changing affective verbali
zation will have a higher correlation between its earlier
measurement and the "effect" factor's later measurement than
the correlation between the effect's earlier measurement and
the cause's later measurement. In all three cases to con
sider given three time periods (segments 1 to 2, 2 to 3, and
1 to 3), the correlation between the earlier client variable
and later therapist variable is larger than the correlation
between the earlier therapist variable and the later client
variable (see Table 5 of cross-lagged panel correlations).
To attempt further clarification of the notion that
therapists might be "following" client affective verbaliza
tion, ratings from a sample of 12 dyads (10% of the total
segments) were examined to see what words the therapist in


APPENDIX B
CLIENT POST-SESSION EVALUATION FORM
Name:
Your participation in this experiment is now concluded.
Please answer the remaining few questions as candidly as you
can. Feel free to make additional comments where you wish
to. Circle one of the alternatives.
1. Did you like the person who interviewed you?
very much slightly not at all
2. Did you enjoy the experience of being interviewed?
very much slightly not at all
3. Do you think the experience was valuable?
very much slightly not at all
4. Did it help you to solve, or live better with, your
problem?
yes unsure no
5. Do you think participating in this experiment has made
you more likely to seek professional help should the
occasion ever arise?
yes unsure no
Additional comments:
73


REFERENCES
Alexander, F. In H. D. Kruse (Ed.)/ Integrating the
Approaches to Mental Disease. New York: Hoeher-
Hayser, 1957.
Auld, F. & Murray, E. J. Content analysis studies of
psychotherapy. Psychological Bulletin, 1955, 52,
377-395.
Barnabei, F., Cornier, W. H., & Hye, L. S. Determining
the effects of three counselor verbal responses on
client verbal behavior. Journal of Counseling Psy
chology, 1974, 21, 355-359.
Bateson, G., Jackson, D., Haley, J., & Weakland, J. Toward
a theory of schizophrenia. Behavioral Science, 1956,
1, 251-264.
Berelson, B. Content Analysis in Communications Research.
Glencoe, Ill.: Free Press, 1952.
Bergin, A. E. & Grafield, S. L. Handbook of Psychotherapy
and Behavior Change: an Empirical Analysis. New York
John Wiley and Sons, 1974.
Blumberg, R. W. A scale for problems in living research.
Psychological Reports, 1968, 22_, 161-162.
Blumberg, R. W. Client dogmatism, therapist leadership,
and the psychotherapeutic relationship (Doctoral dis
sertation, University of Florida, 1969). Dissertation
Abstracts International, 1970, 31^, 389B. (University
Microfilms No. 76-12,227.)
Bordin, E. S. Research Strategies in Psychotherapy. New
York: Wiley and Sons, 1974.
Braaten, L. J. The movement from non-self to self in cli
ent-centered psychotherapy. Journal of Counseling
Psychology, 1961, 13, 20-24 .
Buck, L. A. & Cuddy, J. M. A theory of communication in
psychotherapy. Psychotherapy: Theory, Research, and
Practice, 1966, 3, 7-13.
109


APPENDIX L


2
features of this task are essentially the same as those in
volved in trying to describe the interaction of any interper
sonal dyad, and the controversy which attends any attempt to
do this reflects the complexity of the problem. Despite the
complexity, however, people do manage to relate with one
another so as to reach a consensus that in their relation
ships some behaviors are okay and others are not okay. This
mutual definition of the relationship is reached with suffi
cient clarity for each person to feel that he or she knows
generally what to expect and what not to expect from the other
person. Given that this is true, and it seems to be a reason
able postulate, it should be possible to describe some impor
tant aspects of relationships, including therapeutic relation
ships, and to do so using only the cues each person in the
interaction has access tothe observables of communication.
Much of the complexity of communication arises out of
its imprecision, and it is unreasonable to expect a method
ology for describing communication to be any more precise
than communication itself. In a sense, not only is the ex
perimental observer inferring the "meaning" of the messages
composing the interaction of a dyad, but each member of the
dyad is making inferences about the meaning of the other mem
ber's messages as well as coming to understand more clearly
what meaning he is trying to convey. This kind of process
which has a basic nature of inference and variation of mean
ing can not be represented clearly and precisely in any other
way than to reproduce it exactly, and our closest approxima
tion to that is video-tape or sound movies. In itself, such


106
16 -
14 -
12 uuu
uuu
10 -
tttuuu
8 -
ttt ttt ttt
6 -
4 -
uuu ttt
2 ttt
uuu uuu ttt
uuu ttt
0 -
uuu
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
(1.0) (2.1) (3.1) (4.2) (5.2) (6.3) (7.4) (8.4) (9.5)
CLIENT SCORE INTERVALS3
a Intervals for transformed scores are in parentheses,
u Indicates untransformed scores,
t Indicates transformed scores.
DISTRIBUTION OF SCORES FOR CLIENT AFFECTIVE
VERBALIZATION IN SEGMENT 1
FIGURE M1


51
ent change scores. Modest, though significant, negative cor
relations were observed between some variable pairs (see
Table 7). Reinforcement in the first segment was negatively
correlated with client affective verbalization change from
the second to the third segment and from the first to the
third segments (r's = -.35 and -.44), suggesting that greater
and more discriminating use of verbal reinforcement of cli
ent affective verbalization in the first segment was related
to decreases in that client measure from segment 1 to 3 and
from 2 to 3. This was also true for client affective self
reference (r's = -.53, -.46). Second segment Reinforcement
was negatively correlated with change scores for both client
measures (affective verbalization and self reference) from
segment 2 to 3 (r's = -.32 and -.51). Total Reinforcement
was also negatively correlated to both measures from segments
1 to 3 (respectively r's = -.31, -.35) and to affective self
reference from segments 2 to 3 (r = -.50).
To examine the possible effects of simple number of re
inforcers or quantity of verbalization, these variables were
also correlated with the change scores (see Table 8). For
client affective verbalization, number of reinforcers fol
lowed the same pattern as the Reinforcement variable, though
at a lower correlation coefficient level such that only re
inforcers in the first segment with client change from seg
ments 1 to 3 and total reinforcers with change from 1 to 3
were significant negative correlations (r's = -.33, -.33).
Client affective self reference showed exactly the same pat-
with the number of reinforcers as with the Reinforce


94
T: /Don't just tell me/what is happening; /how do you feel
about it?/
C: /What do you mean?/
T: /Well, you might talk about the feeling you have when
you go to take the test /or when you fail,/like feeling
anxious, upset, frustrated./
Discuss any questions you have with the experimenter when
he checks your practice rating.


107
16 uuu
14 -
12 -
uuu
ttt ttt
10 -
ttt
8 -
6 -
4 uuu
2 -
ttt
uuu
uuu
ttt
0 ttt uuu uuutttuuuttt
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
(1.0) (2.1) (3.1) (4.2) (5.2) (6.3) (7.4) (8.4) (9.5)
CLIENT SCORE INTERVALS3
a Intervals for transformed scores are in parentheses,
u Indicates untransformed scores,
t Indicates transformed scores.
DISTRIBUTION OF SCORES FOR CLIENT AFFECTIVE
VERBALIZATION TOTAL SCORE
FIGURE M-2


112
Hoch, P. A. & Zubin, J. (Eds.). Psychopathology of Commu
nication New York: Grue and Stratton, 1958.
Hodges, J. C. & Whitten, M. E. Harbrace College Handbook
(6th ed.). New York: Harcourt, Brace & World, Inc.,
1967.
Hoffnung, R. J. Conditioning and transfer of affective self
references in a role-played counseling interview.
Journal of Consulting Clinical Psychology, 1969, 33,
527-531.
Isaacs, K. S. & Haggard, E. A. Some methods used in the
study of affect in psychotherapy. In L. A. Gottschalk
& A. H. Auerbach (Eds.), Methods of Research in Psycho
therapy New York: Appleton-Century-Crofts, 1966.
Jackson, D. D., Riskin, J. & Satir, V. A method of analysis
of a family interview. Archives of General Psychiatry,
1961, 5, 321-339.
Jones, E. E. & Thibaut, J. W. Interaction goals as bases
of inference in interpersonal perception. In R.
Tagiuri and L. Petrullo (Eds.), Person Perception and
Interpersonal Behavior. Stanford: Stanford Univiver-
sity Press, 1958.
Kiesler, D. J. Experimental designs in psychotherapy re
search. In A. E. Bergin and S. L. Garfield (Eds.),
Handbook of Psychotherapy and Behavior Change: An
Empirical Analysis. New York: John Wiley and Sons,
Inc., 1971.
Kiesler, D. J. The Process of Psychotherapy. Chicago:
Aldine, 1973.
Kirk, R. E. Experimental Designs: Procedures for the Be
havioral Sciences. Belmont, Calf.": Brooks/Cole Pub-
lishing Company, 1968.
Lennard, H. L. & Bernstein, A. The Anatomy of Psychotherapy:
Systems of Communication and Expectation. New York:
Columbia University Press, 1960.
Lennard, H. L. & Bernstein, A. Pattern in Human Interaction.
San Francisco: Jossey-Bass, 1969.
Levy, W. H. The communication of affect in psychotherapy.
Archives of General Psychiatry, 1967, 1£, 102-106.
Lukens, H. C. The emission of affective verbal behavior
as function of two modes of experimenter intervention:
a partial replication and extension. Dissertation
Abstracts International, 1970, 31, 398B^ (University
Microfilms No. 70-12,392.)


28
as "a verbalization containing a subject and predicate
whether expressed or implied" (p. 57). STATEMENT was
defined as an uninterrupted sequence of PROPOSITIONS by
either therapist or client, and has been defined similarly
by many researchers (Butler, Rice, & Wagstaff, 1962;
Lennard & Bernstein, 1969; and Matarazzo, Saslow, & Mataraz-
zo, 1956). SEGMENT was defined operationally above as a
three-minute section of the interaction, and INTERVIEW is the
composite of the three SEGMENTS of an interview session.
Raters. Raters in the study were volunteers from the
clinical psychology introductory class, volunteers from the
psychology club in the Department of Psychology, and sub
jects from the subject pool in the psychology department,
all groups at the University of Florida. All potential
raters were screened and local references obtained to en
sure that the confidentiality of the materials would be
reasonably protected and to determine the motivation and
reliability of the raters. Appropriate tasks were assigned
for each person.
The experimenter scored the transcripts for the unit
PROPOSITION independently of a random sample of 13% of the
transcript scored by a separate rater. The rater was
trained according to the training manual for scoring units
(see Appendix F). Of over a total of 1802 PROPOSITIONS
jointly scored, there were disagreements on 3.9%. PROPO
SITIONS were indicated on these transcripts by separation
with diagonal marks.


4
cident for this focus. Sullivan (1953, 1954) is generally
considered the first theorist to have emphasized the inter-
personal-interactive nature of pathology and therapy, and
numerous theorists and researchers since have developed in
teractive formulations, most focused on communication proces
ses (Alexander, 1957; Buck & Cuddy, 1966, Buehler & Richmond,
1965; Davis, 1971; Frank, 1961; Greenhill, 1958; Heller, 1963
Hoch & Zubin, 1958; Jones & Thibaut, 1958; Riess, 1957; Rioch
& Weinstein, 1964; Ruesch, 1961; Scheflin, 1965; Sebeok,
Hayes, & Bateson, 1964; Watzlawick, Beavin, & Jackson, 1967;
Williamson, 1959) .
Another such theorist-therapist-researcher is Haley
(1963) who developed a framework based on his background in
communication science (Ruesch, 1961; Reusch & Bateson, 1951)
and on his work with families of schizophrenics in the Palo
Alto group (Bateson, Jackson, Haley & Weakland, 1956).
Haley's conceptualization appears to be a useful framework
for psychotherapy research, even though it has apparently
been used in only a few well-controlled studies of psycho
therapy (Davis, 1971; Davis, Fisher, & Davis, 1973, 1975;
Wadington, 1975). Its use in the research conducted by
Haley's research group was occassionally systematic, but
only informally so (e.g., Jackson, Riskin, & Satir, 1961).
This appears largely due to the complexity and subjec
tivity involved in applying the entirety of Haley's model to
interactions, since it accounts for such comolex phenomena
as the influencing aspects of statements, the metacommunica-


LIST OF TABLES
Table Page
1 ANOVA-1 Summary Table 39
2 ANOVA-2 Summary Table 40
3 Correlations of Modeling and Change Scores 42
4 Affective Measures Intercorrelated 45
5 Summary of Cross-lagged Panel Correlations 48
6 Tabulations of Following Versus Leading
Therapist Responses 50
7 Correlations of Reinforcement Strategy
and Client Change Score 52
8 Correlations Between Number of Reinforcement
Responses and Client Change Scores 53
9 Correlations Between Number of Reinforcement
Responses and Reinforcement Strategy .... 55
10 ANOVA-3 Summary Table 5 9
11 ANOVA-4 Summary Table 6 0
vi


82
Also, each time the speaker changes from the patient
to therapist or vice versa, this defines a "statement" and
statements are numbered down the left hand margin of the
page beginning with one at the beginning of each "segment."
Each "proposition" is set off by diagonal marks, and is num
bered starting with one (1) at the beginning of each
statement.


5
tive function of communications, and the paradoxical nature
of much therapy interaction. In these ideas lies the even
tual promise of Haley's model for application to explaining
and refining the process of therapy, but as Haley points out
(1964), the research proceeding from a model must be based
on highly reliable measurements, even if some of the meaning
fulness of our initial description is lost as a result. In
this particular paper, only the very basic ideas of Haley's
model will be used, attempting by that limitation to develop
adequate tests for those ideas.
Communication as Relationship Definition
Haley contends that whenever two people interact they
must engage in attempts to solve two mutual problems simul
taneously. One problem requires answering the question:
what behaviors (including messages) are to be allowed in this
relationship? In other words the dyad attempts to reach a
mutual definition of the relationship. The second problem
requires an attempt to answer the question: who is to con
trol the definition of the relationship? Here it is not con
trol over behavior directly but over the definition of the
relationship. This definition is very powerful, however,
and provides indirect influence over behavior.
Haley argues that no matter what messages are exchanged
or what behaviors engaged in, those messages and behaviors
represent attempts to reach mutual solutions to the two prob
lems. In other words, the interaction of any dyad, including


TABLE 11
ANOVA-4 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
COMPATIBILITY
22.052
1
22.052
3.69
AP
2
CLIENT SEX
1.005
1
1.005
.17
NS
3
COMPATIBILITY X CLIENT SEX
19.740
1
19.740
3.30
AP
4
ERROR
215.085
36
5.974
5
TOTAL
257.882
39
Note: Dependent variable is Client Total Affective Self Reference.
NS d > 0 5
AP £<.10
cn
o


29
Affective Verbalization. Affective verbalization was
scored utilizing three categories, labeled Primary Affect
Word, Secondary Affect Word, and Affect Idiom. Each of
these was scored separately in the content analysis pro
cedure, but the same general definition holds for all three
types of affective verbalization. Generally, an affective
statement is one which is "expressive of a subjective
reaction of an emotional tone" (Green & Marlatt, 1972), and
in the present use, statements of evaluation without affec
tive meaning, or statements referring to intellectual or
physiological states were excluded.
Primary Affect Words were scored using the unit: WORD,
the category: Primary Affect Word, and the indicator (rule
for assigning a unit to the category): inclusion on the
list of Primary Affect Words in Appendix H. This list was
generated by selecting words which almost always refer de
notatively to affective states. Examples are "angry,"
"sad," "happy," etc. Words on this list are in adjective
form, but a word from this list was scored as a Primary
Affect Word when it appeared in any of its various parts of
speech forms: noun, verb, adjective, or adverb. The
rater's task for this affective expression was to compare
each WORD of a transcript with this list and to label any
WORDs found by recording the WORD'S code number. The rating
and verification was handled in the same manner as with the
unit rating, and the raters were trained using the manual
in Appendix G. Three raters were used and in reliability


54
ment variable, though with somewhat lower correlation coef
ficients .
An important incidental finding of these analyses was
that, with the exception of Reinforcement in the third seg
ment, all the Reinforcement variables are correlated with
all segment levels of number of reinforcers, number of state
ments in a segment, and number of therapist PROPOSITIONS in
a segment (the closest measure available to quantity of ver
bal output). (See Table 9 on correlations of raw scores
and Reinforcement measures.)
Overall, the data did not support the third and fourth
hypotheses with respect to the Reinforcement strategy, and
suggested an opposite to predicted trend.
The analyses of the Group 1 hypotheses, then, have
generally disconfirmed hypotheses 1 through 4, with support
being demonstrated for an alternative hypothesis regarding
the direction of influence between therapist and client.
Compatibility and Client Sex
The primary focus of the hypotheses in Group 2 is
hypotheses 5a and 6a predicting interaction effects of
compatibility and client sex with therapist strategies and
client affective expression change scores. The examination
of these hypotheses is in some ways a moot point since as
stated, these hypotheses relied on supporting hypotheses
2, 3, and 4 above. In order to assess the general relation
ships predicted by these hypotheses, however, plots of the
variables' relationships of interest were made and examined


55
TABLE 9
CORRELATIONS BETWEEN NUMBER OF
REINFORCEMENT RESPONSES AND
REINFORCEMENT STRATEGY
NUMBER OF
THERAPIST
REINFORCEMENTS
BY SEGMENT
THERAPIST REINFORCEMENT
BY SEGMENT
STRATEGY
1
2
3
TOTAL
1
.54**
.53**
o
r1
.57**
2
.49**
.69**
to
00
.67**
3
.43**
.53**
. 36*
.59**
TOTAL
.58**
.68**
.28
.71**
*

. 0 5
¡3 < 01


58
self reference in analysis of variance 4.
Hypothesis 5b (1) was disconfirmed by the non-signif
icant compatibility X segment effect in analysis of variance
1 (see Table 1), indicating that compatibility alone did not
account for changes in client affective verbalization over
the session. Hypothesis 5b(2) predicted this same relation
ship for client affective self reference and was also dis
confirmed in analysis of variance 2 (see Table 2). Hypoth
esis 5b (3) predicted that compatibility would show a signif
icant effect for client total affective verbalization, and
hypothesis 5b (4) made this prediction for client total
affective self reference. Neither hypothesis was supported
as indicated by analysis of variance 3 in the former case
(see Table 10) and by analysis of variance 4 in the latter
case (see Table 11). This effect did approach significance,
however, in this last analysis, suggesting that compatibil
ity alone has some influence on the total frequency of
clients' use of affective self reference.
The analysis of these hypotheses for client sex was
parallel to the analyses for compatibility. Hypothesis
6b(1), predicting that female clients but not male clients
would increase across the session, was analyzed by the
client sex X segment effect in analysis of variance 1 for
client affective verbalization (see Table 1), and 6b(2),
in analysis of variance 2 for client affective self refer
ence (see Table 2). Neither hypothesis was confirmed.
Hypotheses 6b(3) and 6b(4) predicted that female clients
would be higher than males in total affective verbalization


APPENDIX I
LISTING OF SECONDARY AFFECT WORDS
ACTIVE
ADVENTUROUS
AGGRESSIVE
AGREEABLE
ALERT
AMBITIOUS
APPRECIATIVE
AWFUL
BAD
CLOSE
COMFORTABLE
CONSIDERATE
CRAPPY
CRUDDY
DEFENSIVE
DESIRE
DOUBTFUL
DULL
EASY
ENERGETIC
FREE
FRIENDLY
GOOD
GRATEFUL
GRATIFIED
GREAT
GUILTY
HELL
HELPLESS
HOPELESS
IMPATIENT
INDIFFERENT
INFLAMED
INHIBITED
INSECURE
INTENSE
KIND
LEISURELY
LIVELY
LOATHE
LOST
LOVING
MEAN
NEGATIVE
NUMB
PLODDING
POSITIVE
PROVOKED
QUARRELSOME
REBELLIOUS
RO1ANTIC
ROTTEN
SAVAGE
SECURE
SETTLED
SHAKY
SHITTY
SLEEPY
SLOW
SLUGGISH
SOLEMN
STEADY
THOUGHTFUL
THREATENED
TIRED
TOGETHER
TOLERANT
TROUBLED
UNCERTAIN
UNCONCERNED
UNDISTURBED
UNSURE
USELESS
WEARY
WEIGHTY
WEIRD
WITHDRAWN
WONDERMENT
WORTHLESS
99


APPENDICES


113
Markel, N. N. Psycholinguistics: an Introduction to the
Study of Speech and Personality. Homewood, Ill.:
Dorsey Press, 1969.
Marlatt, H. G. Exposure to a model and task ambiguity as
determinants of verbal behavior in an interview.
Journal of Consulting and Clinical Psychology, 1971,
36., 268-276 .
Marsden, G. Content-analysis studies of psychotherapy:
1954 through 1968. In A. E. Bergin & S. L. Garfield
(Eds.), Handbook of Psychotherapy and Behavior Change:
An Empirical Analysis. New York: John Wiley & Sons,
Inc., 1971.
Matarazzo, J. D., Saslow, G., & Matarazzo, R. G. The inter
action chronograph as an instrument for objective meas
urement of interaction patterns during interviews.
Journal of Psychology, 1956 41_, 347-367.
Meltzoff, J. & Kornreich, M. Research in Psychotherapy.
New York: Atherton, 1970.
Mendelsohn, G. A. & Rankin, N. 0. Client-counselor compati
bility and the outcome of counseling. Journal of Ab
normal Psychology, 74, 1969.
Merbaum, M. & Lukens, H. C. Effects of instructions, elici
tations, and reinforcements in the manipulation of af
fective verbal behavior. Journal of Abnormal Psychol
ogy, 1968, 73, 376-380.
Moos, R. H. & Ciernes, S. R. Multivariate study of the
patient-therapist system. Journal of Consulting
Psychology, 1967, 31, 119-130.
Morris, C. Signs, Language, and Behavior. Englewood Cliffs,
N. J.: Prentice Hall, 1946.
Myrick, R. D. Effect of a model on verbal behavior in
counseling. Journal of Counseling Psychology, 1969,
16, 185-190.
Riess, B. F. Communication in psychotherapy. American
Journal of Psychotherapy, 1957, .LI, 474-489 .
Rioch, D. & Weinstein, E. A. (Eds.). Disorders of Communi
cation Baltimore: Williams and Wildins, 1964.
Ruesch, J. Therapeutic Communication. New York: Norton,
1961.
Salzinger, K. & Pisoni, S. Reinforcement of affect responses
of schizophrenics during the clinical interview.
Journal of Abnormal and Social Psychology, 1958, 57,


APPENDIX K
LISTING OF THERAPIST VERBAL REINFORCERS
MM-HMM*
GOOD
WONDERFUL
YEAH
I SEE
YES
TRUE
THAT'S TRUE
AH-HAH*
YEP
RIGHT
UH-HUH*
OKAY (OR 0. K. )
*Slight variations in these (e.g., Um-huh), should also be
scored.
101


CHAPTER 3
TECHNICAL CONSIDERATIONS
Method
Subjects. There were two groups of subjects: 10
currently practicing psychotherapists, 5 male and 5 female,
with at least 5 post-degree years of experience each; and
40 clients, 20 male and 20 female from the subject pool in
the Psychology Department at the University of Florida.
Apparatus. Apparatus used in the experiment included
standard videotape-recording equipment, and cassette trans
cribing equipment. The same room and seating arrangement
were used in all interviews.
Paper and pencil instruments. Paper and pencil scales
used were the FIRO-B (Fundamental Interpersonal Relations
Orientation-Behavior, Schtz, 1966), the Problem Patholog
ical Potential Scale (Blumberg, 1968), a therapist post
session evaluation form (see Appendix A; Blumberg, 1969),
and a client post-session evaluation form (see Appendix B;
Blumberg, 1969).
Design. Through matching on test scores, client sub
jects were assigned to therapists such that each therapist
had one client in each cell of two levels of compatibility
by two levels of client sex. Interviews were considered as
24


CLIENT AFFECTIVE VERBALIZATION
AND THERAPIST INFLUENCE STRATEGIES IN
ANALOGUE PSYCHOTHERAPY INTERVIEWS
By
STANLEY E. JONES
A THESIS PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1976

DEDICATION
To Barbara
Whose patience and support have
made all the difference.

ACKNOWLEDGMENTS
I would like to thank my chairperson, Dr. Jacquelin
Goldman, for her willing and consistent guidance and assis
tance through a long period of often frustrating endeavor.
I am also appreciative of the contributions of each of my
other committee members: Dr. Audrey Schumacher, for her
influence on my thinking about therapy and her always
incisive critiques; Dr. Franz Epting for his encouragement
to be more scholarly and to be creative in research appli
cations; Dr. Marvin Shaw, for his assistance in solving
the difficult design and analysis problems involved in the
study; and Dr. Hannelore Wass, for her perspective from
outside my immediate area.
Many other individuals, faculty, staff, fellow graduate
students, and undergraduate students, were involved in the
various aspects of the work reported here, and though they
are too numerous to mention, I would like to express my
appreciation for their cooperation. A special word of
thanks goes to my typist, Barbara Jones, for her hard
work above and beyond the call of duty.
iii

TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS iii
LIST OF TABLES vi
ABSTRACT vii
CHAPTER
1 INTRODUCTION 1
Communication as Relationship
Definition 5
Therapist Goals as a Definition 6
2 DEFINING THE RESEARCH QUESTION 9
Communication Science: An Overview ... 13
Defining Communication of Affect 16
Research Hypotheses 22
3 TECHNICAL CONSIDERATIONS 24
Method 2 4
Rating Procedures 27
4 RESULTS 36
Client Verbalization and Therapist
Strategies 36
Compatibility and Client Sex 54
Client Verbalization and Session Outcome. 61
Exploratory Analyses 62
5 DISCUSSION 65
Conclusion 69
APPENDICES 71
APPENDIX A--Therapist Post-Session
Evaluation Form 72
APPENDIX B--Client Post-Session Evaluation
Form 73
IV

APPENDICES-continued Page
APPENDIX C--Notice for Potential Client
Subjects 74
APPENDIX D--Verbatim Telephone Contact for
Informing Selected Clients 75
APPENDIX E--Rules for Locating Segments Within
Full Interviews 76
APPENDIX F--Training Manual for Unitization of
Transcripts by Proposition 78
APPENDIX G--Training Manuals for Rating All
Variables 81
APPENDIX H--Listing of Primary Affect Words
and Limited Primary Affect Words. . 97
APPENDIX I--Listing of Secondary Affect Words . 99
APPENDIX J--Listing of Affective Idioms and
Slang 100
APPENDIX K--Listing of Therapist Verbal
Reinforcers 101
APPENDIX L--Examples of Graphs Plotting
Interaction Effects 102
APPENDIX M--Examples of Score Distribution
Graphs 105
REFERENCE NOTES 108
REFERENCES. 109
BIOGRAPHICAL SKETCH 116
v

LIST OF TABLES
Table Page
1 ANOVA-1 Summary Table 39
2 ANOVA-2 Summary Table 40
3 Correlations of Modeling and Change Scores 42
4 Affective Measures Intercorrelated 45
5 Summary of Cross-lagged Panel Correlations 48
6 Tabulations of Following Versus Leading
Therapist Responses 50
7 Correlations of Reinforcement Strategy
and Client Change Score 52
8 Correlations Between Number of Reinforcement
Responses and Client Change Scores 53
9 Correlations Between Number of Reinforcement
Responses and Reinforcement Strategy .... 55
10 ANOVA-3 Summary Table 5 9
11 ANOVA-4 Summary Table 6 0
vi

Abstract of Dissertation Presented
to the Graduate Council of the University of Florida
in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy
CLIENT AFFECTIVE VERBALIZATION
AND THERAPIST INFLUENCE STRATEGIES IN
ANALOGUE PSYCHOTHERAPY INTERVIEWS
By
Stanley E. Jones
December 1976
Chairperson: Jacquelin R. Goldman
Major Department: Psychology
Psychotherapy theory emphasizing interpersonal interac
tion is used as a theoretical base for considering strate
gies that therapists might take to increase client affective
verbalization in therapy situations. Previous research of
analogue interviews is used for defining the strategies,
Instruction, Modeling, and Reinforcement, and for defining
the dependent measures, client affective verbalization and
client affective self reference. Professional therapists
conducted single interviews with experimental subjects,
after the dyads were matched on interpersonal need compati
bility and on all sex combinations. A rating system for
determining affective verbalization and affective self ref
erence was developed and applied to typed transcriptions of
the interviews. Therapist strategies were also measured by
vii

a rating system. All rating systems were implemented through
a training manual and standard procedure. Hypotheses pre
dicted that client affective verbalization and client affec
tive self reference would increase across the interview,
with the therapist strategies associated with larger in
creases. Interpersonal need compatibility, sex of client,
and interactions between those two variables and the strate
gies were also predicted as important in the effects related
to the dependent measure.
The rating procedure appeared adequate with rating
checks producing good reliability. Data were analyzed
through a split plot factorial (2x2x3) analysis of variance
of compatibility, client sex, and segment separately for cli
ent affective verbalization and client affective self ref
erence. One effect was significant: compatibility X client
sex, with a simple main effects test indicating that the
compatibility manipulation was effective only for female cli
ents, as previous research has suggested. Other analyses
were conducted through Pearson or Spearman correlations.
Results suggested that clients tend to influence the affec
tive verbalization level of therapists rather than vice
versa, and that therapists in this sample do not actually
use the strategies which analogue studies have researched.
There was support for therapists believing that client af
fective verbalization is important to treatment outcome. Al
ternative hypotheses and alternatives for further research
are discussed.
vm

CHAPTER 1
INTRODUCTION
Psychotherapy is an important phenomenon, both cul
turally and scientifically. Culturally, it is the practice
of a helping profession by therapists and the seeking of as
sistance by clients. Scientifically, it is the interaction
of two types of experimental subjects, where one subject is
to effect some change in the other, often changes in very
basic personality processes of the client subject. In both
cases, a major question is that of how the therapist has his
impact with the client. As Strupp (1973) puts it, "How does
it come about that one person, by virtue of what he is or
what he does, can exert such a lasting influence on another
person that the changes that result from this encounter may
be termed therapeutic?" (p. 19) The variations of this ques
tion and the more specific implications of it are numerous
and complex, as indicated by the scope and number of reviews
on research of psychotherapy (e.g., Bergin & Garfield, 1971;
Bordin, 1974; Kiesler, 1973; Meltzoff & Kornreich, 1970).
The present study will be focused on looking at one aspect
of answering the above question.
One difficult task in beginning this focusing is speci
fying the way in which the interaction of therapist and
client is to be conceptualized and described. The general
1

2
features of this task are essentially the same as those in
volved in trying to describe the interaction of any interper
sonal dyad, and the controversy which attends any attempt to
do this reflects the complexity of the problem. Despite the
complexity, however, people do manage to relate with one
another so as to reach a consensus that in their relation
ships some behaviors are okay and others are not okay. This
mutual definition of the relationship is reached with suffi
cient clarity for each person to feel that he or she knows
generally what to expect and what not to expect from the other
person. Given that this is true, and it seems to be a reason
able postulate, it should be possible to describe some impor
tant aspects of relationships, including therapeutic relation
ships, and to do so using only the cues each person in the
interaction has access tothe observables of communication.
Much of the complexity of communication arises out of
its imprecision, and it is unreasonable to expect a method
ology for describing communication to be any more precise
than communication itself. In a sense, not only is the ex
perimental observer inferring the "meaning" of the messages
composing the interaction of a dyad, but each member of the
dyad is making inferences about the meaning of the other mem
ber's messages as well as coming to understand more clearly
what meaning he is trying to convey. This kind of process
which has a basic nature of inference and variation of mean
ing can not be represented clearly and precisely in any other
way than to reproduce it exactly, and our closest approxima
tion to that is video-tape or sound movies. In itself, such

3
a reproduction does not get us any closer to understanding
the process of communication, however. A representation
apart from a reproduction is required, and this implies the
operation of a theory providing a model for the representa
tion and an artificial limitation of the complexity allowed
in the representation. Numerous researchers recognize spe
cifically the need for limiting the complexity of the process
studied (Auld & Murray, 1955, Cartwright, 1966; Guetzkow,
1950; Kiesler, 1973; Marsden, 1971; and Strupp, 1962) and
the need for being specific about the theory used. The in
fluence of the theory on our selection of the particular phe
nomena studied cannot be over-emphasized, since the operation
of a theoretical perspective in the selection of a represen
tation is the primary determinant of the meaningfulness of
the research no matter how sophisticated other aspects of
the research may be.
Researching communication, then, involves the selection
of a theoretical perspective which will dictate the aspects
of communication to be described, but we cannot expect the
application of a description system to render communication
any clearer or more precise than it actually is. The actual
selection of that theoretical perspective involves both the
preference of the experimenter and the theory's support in
the literature.
Many researchers focus on the nature of psychotherapy
as a special case of a general dyadic communication model
(Kiesler, 1973). Certainly there is ample theoretical pre-

4
cident for this focus. Sullivan (1953, 1954) is generally
considered the first theorist to have emphasized the inter-
personal-interactive nature of pathology and therapy, and
numerous theorists and researchers since have developed in
teractive formulations, most focused on communication proces
ses (Alexander, 1957; Buck & Cuddy, 1966, Buehler & Richmond,
1965; Davis, 1971; Frank, 1961; Greenhill, 1958; Heller, 1963
Hoch & Zubin, 1958; Jones & Thibaut, 1958; Riess, 1957; Rioch
& Weinstein, 1964; Ruesch, 1961; Scheflin, 1965; Sebeok,
Hayes, & Bateson, 1964; Watzlawick, Beavin, & Jackson, 1967;
Williamson, 1959) .
Another such theorist-therapist-researcher is Haley
(1963) who developed a framework based on his background in
communication science (Ruesch, 1961; Reusch & Bateson, 1951)
and on his work with families of schizophrenics in the Palo
Alto group (Bateson, Jackson, Haley & Weakland, 1956).
Haley's conceptualization appears to be a useful framework
for psychotherapy research, even though it has apparently
been used in only a few well-controlled studies of psycho
therapy (Davis, 1971; Davis, Fisher, & Davis, 1973, 1975;
Wadington, 1975). Its use in the research conducted by
Haley's research group was occassionally systematic, but
only informally so (e.g., Jackson, Riskin, & Satir, 1961).
This appears largely due to the complexity and subjec
tivity involved in applying the entirety of Haley's model to
interactions, since it accounts for such comolex phenomena
as the influencing aspects of statements, the metacommunica-

5
tive function of communications, and the paradoxical nature
of much therapy interaction. In these ideas lies the even
tual promise of Haley's model for application to explaining
and refining the process of therapy, but as Haley points out
(1964), the research proceeding from a model must be based
on highly reliable measurements, even if some of the meaning
fulness of our initial description is lost as a result. In
this particular paper, only the very basic ideas of Haley's
model will be used, attempting by that limitation to develop
adequate tests for those ideas.
Communication as Relationship Definition
Haley contends that whenever two people interact they
must engage in attempts to solve two mutual problems simul
taneously. One problem requires answering the question:
what behaviors (including messages) are to be allowed in this
relationship? In other words the dyad attempts to reach a
mutual definition of the relationship. The second problem
requires an attempt to answer the question: who is to con
trol the definition of the relationship? Here it is not con
trol over behavior directly but over the definition of the
relationship. This definition is very powerful, however,
and provides indirect influence over behavior.
Haley argues that no matter what messages are exchanged
or what behaviors engaged in, those messages and behaviors
represent attempts to reach mutual solutions to the two prob
lems. In other words, the interaction of any dyad, including

6
that of therapist and client, can be seen as involving a pro
cess of attempting to reach a mutual definition of what be
haviors will and will not occur in the relationship, and a
process of determining who will control the definition.
Control of the definition is an important issue, and can
be seen as a competition between the two participants in a
dyad. Each is engaging in purposive behavior directed toward
obtaining some desired and satisfying state of affairs, and
rarely do these purposes coincide so exactly as to allow an
entirely cooperative venture. Since the definition of the
relationship is essentially a statement about what behaviors
are to be allowed in the interaction, the person who con
trols that definition is influencing the interaction and
therefore the other person, and has an advantageous posi
tion. The person not in control is expected to attempt to
take control by redefining the relationship in a more satis
factory way for him.
Therapist Goals as a Definition
Most of an effective therapist's "needs" in a session
are directed toward therapeutic goals, and are essentially
the needs the therapist is attempting to satisfy in his ef
forts to control the definition of the relationship. One
way in which therapists frequently want to influence the in
teraction is by focusing the client into a particular style
of interaction. Many theories of therapy emphasize the im
portance of styles of interaction, as for example in seeing

7
the client's expression of affect as a positive step in
reaching therapeutic goals, either, for example, in client-
centered therapy as a major process, or in psychodynamic
therapy as a cathartic and insight-generating process.
This control of style is most clearly a control of the
definition problem when the client does not exhibit the de
sired behavior initially, as is usually the case. The thera
pist wants the relationship to be one in which it is accept
able for the therapist to respond in compatible ways (e.g.,
reflecting affect, making interpretations about the affect,
etc.). If the client interacts this way naturally there is
no problem, and the therapist and client get to business
rather quickly. If the client does not, the therapist must
help arrange for this to happen. The way this is attempted
will be referred to as the therapist "strategy."
To summarize the rationale to this point, Haley's model
describes the interaction between therapist and patient as
like that of any dyad and consisting of the processes of de
termining the definition of the relationship and competing
for control of that definition. It is through these pro
cesses that the interpersonal influence of therapy takes
place.
One implication of Haley's theory as it is being con
sidered is that the therapist is expected to work toward im
plementing the therapeutic goals as the definition of the
relationship, insofar as they apply. In developing a re
search examination of this theoretical position, a specific

8
therapeutic goal and specific therapist interventions for
increasing the occurence of the goal behavior will be con
sidered. In order to clearly indicate the relevance of the
interventions, the goal behavior will be discussed first.

CHAPTER 2
DEFINING THE RESEARCH QUESTION
The general research question of the present study was
stated above as: "How does the influence of therapy occur?"
In the present formulation, a more specific question becomes:
"How does the therapist behave so as to attempt instituting
the therapy goal behaviors as a definition of the relation
ship?" The first step in operationalizing this question for
the present study is selecting and defining a specific ther-
apy goal for study.
As used in this paper, a "therapist goal" is defined
as a behavior which may be engaged in by the client and which
is believed by the therapist to be related to the occurance
of therapy. "Therapy" is defined as a client interacting
with a therapist and working toward a change in the initial
behavior, feelings, or belief system of the client, with
that change in the direction of from less to more desirable
as defined by the client, by society, or by the therapist.
"Mutual definition of the relationship" is defined as the
statement of acceptable behaviors desired by the therapist
and exhibited by the client. Here we are considering only
the case where the therapist is in control. If the client
is not exhibiting the desired behaviors and the therapist
stops making attempts to influence the client to do so, the
9

10
client could be said to be in control. Note that the focus
here is on the therapist's belief about what client behavior
is therapeutic, and in studying the therapist's process, it
is important to look at client behaviors that are believed
to lead to the occurence of change and desired to be a char
acteristic of the client early in therapy.
For example, almost all psychotherapists would see "the
client verbalizes" as such a goal. Except for the rare case,
the client's talking is a pre-condition to the occurrence of
therapy change. A related client behavior is that when the
client talks the client talks about affect. This notion,
that the client's verbalization of or about affective states
is important to therapy, appears to be shared by many sys
tems of therapy, either through the direct theoretical state
ment of the therapist or through the implication of the
goal that is set (e.g., see Harper, 1959). Even when this
is not a primary end, as it is in client-centered therapy
for example, it is frequently a secondary end allowing the
therapist and client to gain the information they need to
proceed with other goals, as for example in rational-emotive,
or various psychoanalytic therapies. This client behavior,
the verbalization of affect, will be the therapist goal in
this study.
Research has often focused on the variable, "client ex
pression of affect," and operationally defined it in terms
of judges' ratings (e.g., Waskow, 1962; Ellsworth, 1963) or
in terms of systems making inferences about the client's

11
emotional state on the basis of his verbal behavior (e.g.,
Gottschalk & Gleser, 1969). In the present study as in other
studies to be discussed later, the importance of the client's
expression is that it is done directly, by verbalizing using
feeling words, particularly since it facilitates a process
of the therapist and client mutually coming to understand
something about the client. There are, of course, other
points at which the client is expressing or experiencing
feelings, and the client's direct report of his feeling may
or may not fit our judgements about his immediate affective
state. In considering this study's specific goal for the
client, however, it is the client's verbalization of affect
words that is of interest.
Defining client affect expression in this way also fa
cilitates the precision and usefulness of psychotherapy re
search as scientific data. Haley (1964) as well as others
(see Kiesler, 1971, 1973) argue for the selection of dis
cretely measureable behaviors in researching psychotherapy
and other interpersonal processes. Doing so minimizes the
inference inherent in the measurement process itself, and
allows us to look formally at the inferences made in a sys
tematic way.
The degree to which we can make inferences based on
this kind of data is limited by its measuring less general,
though more reliably measured, behaviors. As long as the in
ferences made from the study are made after careful consider
ation of the limitations of the data, the problem is more
one of frustration in being unable to deal with more rele-

12
vant and meaningful general constructs rather than a prob
lem of experimental error. A particularly critical example
of this potential error in this study, as suggested above,
is that of confusing the variable "client verbalization of
affect words" with the more general variable "client affec
tive expression." The first refers to characteristics of
the denotative meaning of words and idioms used by the cli
ent while the latter refers to inferences made about the
state of the client in using whatever language he uses. In
this study, then, it will be important to recognize that the
data, and inferences made from it, are based on considera
tions of the kinds of language used by a person and not nec
essarily on whether or not that person is in touch with af
fective experience.
Although there is a significant amount of literature
dealing with verbalization of affective words as a dependent
variable there appears to be little that has examined its
validity as an indicator that important aspects of therapy
are occurring when it occurs or as a predictor of successful
outcomes. Levy (1967) provided support that in statements
where judges agreed therapy was occurring in primarily anal
ytic sessions there was a greater proportion of statements
with affect words than in statements where therapy was judged
not to be occurring. Isaacs and Haggard (1966) found that
therapist attention to patient affective verbalizations re
sulted in the patient's responses having an increased affec
tive orientation. Braaten (1961) demonstrated that increas
ing expression of feeling with reference to self as opposed

13
to nonself was related to therapist-rated success in client-
centered therapy. The validity of this variable seems theo
retically and intuitively adequate, and its validity will be
examined further in the present study by examining its re
lationship to session outcome.
Significant reliability between raters of interaction
material is often used as a measure of the adequacy of data
for process research, and in many studies it is sufficient
to answer the questions posed. In the present study and for
future extension to other aspects of Haley's model, however,
it is important that "discrete" mean that with very few ex
ceptions it can be agreed that a particular instance of a
chosen behavior has or has not occurred. In order to attempt
achieving this type of precision of measurement without
losing the relevance of the behaviors in which we are inter
ested, the communication situation must be considered in
more detail and then the particular variables to be consid
ered in this study can be defined. An overview of the field
of communication science as it relates to psychotherapy re
search will be used to supply greater detail and establish
perspective.
Communication Science: An Overview
Kiesler (1973) in presenting process studies of psycho
therapy and proposing a structure for developing such studies
suggests Market's (1969) model of dyadic interaction. The
basic communication situation is composed of an originator
of a message, the ENCODER, a recipient of a message, the DE-

14
CODER, and a response of an encoder which may be the stimu
lus for a decoder, the MESSAGE. COMMUNICATION results when
a response of an encoder is received as a stimulus for a de
coder. In further specifying the communication situation,
Markel identifies six channels of human "face to face" com
munication by specifying the source in the encoder response.
The channels with the source and destination for each are:
speech (vocal tract, ear); kinesics (body movement, eye);
odor (chemical processes, nose); touch (body surface, skin);
observation (body surface, eye); and proxemics (body place
ment; eye). In psychotherapy research all three of the
basic communication "entities" are important, and only three
of the channels are studied to any significant extent at
present: speech, kinesics, and proxemics. Odor appears to
be studied the least, with observation and touch gaining
attention. Speech is by far the most studied of all the
channels.
Speech is divided in a number of ways which are impor
tant in specifying the way this study is going to deal with
communication. Morris (1946) divides the science of lan
guage into three aspects: SYNTACTIC, formal relations of
signs to each other; SEMANTIC, relations between signs and
the objects to which the signs are applicable; and PRAGMATIC,
the relations between signs and users of signs. This study
will consider aspects of each of these areas.
Linguistics is one of the largest areas of communica
tion science. Morris (1946) defines it as the descriptive
study of the formal characteristics of language, and classi-

15
fies it under syntactics. Trager (1966) defines it as the
scientific study of human language or speech; the study of
messages once they are "on the air." Linguistics is composed
of three areas: PHONOLOGY: the study of sounds, with units
of increasing size called the phoneme, phone, syllable; MOR
PHOLOGY: the study of shapes (forms) constituted by sequen
ces of sounds, with units of the morpheme, morph, and word;
and SEMOLOGY (syntax or grammar): the study of the sense,
the meaningful arrangements, from which the symbolizing func
tions arise, with basic units of the part of speech, constit
uent, and the sentence.
Psycholinguistics is the study of the relation between
messages on the speech channel and the cognitive or emotional
states of human encoders and decoders who send and receive
the messages (Markel, 1969). Linguistics studies language
sounds, those sounds necessary to the production of meaning
ful speech, and psycholinguistics includes the study of non
language speech sounds, those whose variation does not change
the linguistic meaning of the speech.
This study will consider only one channel, speech, and
within that only those aspects studied within linguistics
proper. There are two primary reasons for this: 1) the
variables to be defined deal with the meanings of language
communications and the content of speech is the most impor
tant for measuring them; and 2) more highly valid and reli
able measures can be made using a single channel and single
type of meaning considering the present state of communica
tion measurement methodology.

16
Defining Communication of Affect
A number of studies have utilized measures of verbali
zation of affect words, though with significant variations
in its definition. Levy (1967) studied the "communication
of affect" and defined it in terms of the verbalization of
words from a list of affect words which he developed from
analytic interviews. Moos and Ciernes (1967) used "percent
age of feeling words" as one dependent variable and defined
it by "enumerating examples of words directly relevant to an
individual's affective state" (p. 121).
Most other studies using this measure have been inves
tigating the effects of various therapist behaviors on cli
ent verbalization of affect within operant conditioning para
digms. Here too the specific definition used has often var
ied. The most popular definition has also included the sti
pulation that the statement be a self reference and is ac
cordingly called an "Affective Self Reference." Salzinger
and Pisoni (1958, 1960) appear to have made the first formal
use of this definition, which specifies that the response
class statement must be an "I" or "We" statement describing
or evaluating the client's own state, except for intellectu
al or physiological, and providing some rules for instances
where the judgement is unclear. A number of other studies
have used this definition or small modifications of it
(Hekmat, 1971a, 1974; Hekmat & Lee, 1970; Hoffnung, 1969).
"Affective Self Disclosure" has been used as the response
class by Hekmat and Thiess (1971) and Hekmat (1971b). It

17
differs from the Affective Self Reference in using only "I"
statements and considering affective expressions only if
they have the form: "I feel....," with an expression of af
fect. Other definitions have included the "Feeling State
ment" (Green & Marlatt, 1972) with extensive rules, and two
closely related definitions, the "Emotional Words" measure
(Merbaum & Lukens, 1968) and the "Client Affective Word"
measure (Barnabei, Cornier & Hye, 1974; Crowley, 1970).^
Variations, then, have included whether or not there
was a self-reference required and whether that included state
ments beginning with we; whether the affective expression
was positive or negative or unscored for direction; whether
only certain types of affective statements were considered
or certain types excluded; whether both therapists and cli
ents were scored; and whether statements were required to be
with reference to present time or not (immediacy). The pre
sent study will use a definition modeled after Green and
Marlatt's rules (Note 1), but reducing the amount of judge
ment to be exercised by raters by using a more extensive
listing of affect words, and reducing the extent to which
evaluation is included as feeling expression. In order to
increase the applicability of this study to other studies,
statements will be scored both with and without a self ref
erence required.
The inference is being made that a behavior like client
^Psycho-Die, a content analysis dictionary, appears to have
the capability for defining this measure in any of these
ways, but no studies using it for this purpose were found.

18
verbalization of affect reflects a definition of the rela
tionship, and the research question becomes, "How does the
therapist behave so as to attempt increasing client affec
tive verbalization?" The literature related to increasing
client verbalization of affect indicates the particular ef
fectiveness of at least three strategies. One is that the
therapist tells the client directly what kind of relation
ship he (the therapist) wants; i.e., the therapist instructs
the client on how to behave. A second is that the therapist
suggests a definition and behavior by modeling the desired
client behavior. Third, the therapist utilizes differential
verbal reinforcement of the desired behavior to increase its
occurrence.
The first strategy is the use of instructions by the
therapist, and has been studied and found effective in a num
ber of studies (e.g., Green & Marlatt, 1972; Lukens, 1970;
Merbaum & Lukens, 1968; and Whalen, 1969). Here the thera
pist communicates directly to the client that in the rela
tionship it is the client's responsibility to engage in par
ticular behaviors, and perhaps, not in other behaviors. For
example, at the beginning of the interview, the therapist
might say, "During the interview, you are to talk about your
feelings about yourself, and not about the feelings other
people have or subjects unrelated to yourself." In that
case, the therapist would be making a very explicit defini
tion of the relationship as to what he expected from the cli
ent .
The second strategy is for the therapist to be a model

19
for the client of the desired client behavior, and this has
also been examined and found effective in increasing client
affective verbalization in several studies (e.g., Green &
Marlatt, 1972; Marlatt, 1971; Myrick, 1969; Whalen, 1969).
In using this strategy, a therapist would be observed to use
affective words, or in the case of modeling affective self
references, to make statements with reference to the thera
pist's affect.
Various types of verbal conditioning techniques are the
third strategy and the area to be considered here uses
"agreement" as a social reinforcer, with specific instances
as "mmm-hmm," "good," "wonderful," "yeah," "I see," etc.
(e.g., Hekmat, 1971b, 1974; Hekmat & Lee, 1970; Moos & Ciernes,
1967; Salzinger & Pisoni, 1960). A therapist using this
strategy would be observed to give reinforcing responses af
ter a large proportion of client statements containing af
fect words or idioms (or affective self references) and af
ter a small proportion of other statements. Effective use
of this strategy probably involves indiscriminant and fre
quent use of the reinforcing response(s) early in the ses
sion to establish its reinforcing power (Waskow, 1962), and
the discriminant use of such responses subsequently.
Research has indicated that another important factor in
considering variables in therapist-client control is the in
terpersonal need compatibility of the particular therapist-
client dyad (Gassner, 1970; Sapolsky, 1960, 1965). The
FIRO-B (Fundamental Interpersonal Relations Orientation -

20
Behavior, Schtz, 1966) is most used for assessing this vari
able, and has been demonstrated to be a reliable and valid
indicator of interpersonal compatibility on a number of meas
ures. Of the three interpersonal needs postulated by the
FIRO scales, the most important for the present formulation
is the need for inclusion and control, and of the various
compatibility computations, that of reciprocal need compati
bility is most appropriate. In formulating hypotheses about
the use and effectiveness of the above therapist strategies
the combination of reciprocal compatibility for Inclusion
and Control of the therapist-client dyads will be considered.
One study (Mendelsohn & Rankin, 1969) has indicated
that sex may also be an important variable. It was found
that the compatibility measures work well in some ways with
female clients, but poorly with male clients. Sex of thera
pist did not appear to be an important factor in the differ
ences. Therefore, sex of client will also be considered as
a variable in assessing the effects of the compatibility
measure.
In summary then, client affective verbalization and cli
ent affective self reference are considered important vari
ables, and three therapist strategies have been shown to be
effective in increasing the frequency of occurrence for both
these variables in quasi-therapeutic situations. Previous
studies have apparently not examined the relationship of
variables in situations more closely resembling actual thera
pists who have not been specifically instructed to engage
in the strategy behaviors. The theory and literature re-

21
viewed above suggests that therapists may engage in these
behaviors in actual interviews, and that when they do, cli
ents' use of affective words increases across the interview.
The interviews will occur in a standardized situation
with each therapist serving as a control for himself or her
self, and the literature suggests that two other variables
require control and examination as independent variables.
These variables are interpersonal need compatibility and cli
ent sex, and these can be both independent variables and
controlled in other analyses by utilizing a completely bal
anced design with respect to them. In addition, therapist
sex can also be included in the balancing to control for ef
fects of that variable.
Although client affective verbalization and affective
self reference were consistently assumed important in the
literature reviewed, little evidence was cited which defended
this assumption. In order to examine the validity of these
variables in terms of session outcome, measures of session
outcome will be examined.
Hypotheses can be conceptualized as falling into three
groups. The first group deals with the question of whether
or not therapists use the strategies as defined, whether or
not the expected changes in client affective verbalization
behavior occur, and whether or not they occur or are more
pronounced when considering the strategies.
The second group of hypotheses deals with the variables
of compatibility and client sex both as independent varia
bles by themselves and also as they interact with the stra-

22
tegies and each other. The third group deals with the re
lationships between client affective verbalization measures
and session outcome measures, with consideration given to
the possible role of compatibility. The hypotheses are
formally stated and labeled below.
Research Hypotheses
It is hypothesized:
Group 1
1. that each therapist uses one or more of the above strat
egies;
2. that a) client affective verbalization and b) client af
fective self references will increase across the inter
view;
3. that greater use of a strategy is associated with larger
increases in client affective verbalization;
4. that greater use of a strategy is associated with larger
increases in client affective self reference;
Group 2
5. that a) higher compatibility will be associated with en
hancing the positive effect of the therapist strategy
behaviors, and b) that higher compatibility of the thera
pist-client dyad will be associated with (1) client af
fective verbalization and (2) client affective self ref
erence increasing across the session, and with (3) total
client affective verbalization and (4) total client af
fective self reference;
6. that a) the positive effect of the therapist strategy

23
of increasing client affective verbalization and client
affective self reference will be enhanced with female
clients and not with male clients, and b) that female
clients and not male clients will be associated with
(1) client affective verbalization and (2) client af
fective self reference increasing across the session and
with (3) total client affective verbalization and (4)
total client affective self reference.
Group 3
7. that a) client affective verbalization and client affec
tive self reference will be positively associated with
session outcome, and b) higher compatibility will enhance
the positive association of client affective verbaliza
tion and client affective self reference to session out
come .

CHAPTER 3
TECHNICAL CONSIDERATIONS
Method
Subjects. There were two groups of subjects: 10
currently practicing psychotherapists, 5 male and 5 female,
with at least 5 post-degree years of experience each; and
40 clients, 20 male and 20 female from the subject pool in
the Psychology Department at the University of Florida.
Apparatus. Apparatus used in the experiment included
standard videotape-recording equipment, and cassette trans
cribing equipment. The same room and seating arrangement
were used in all interviews.
Paper and pencil instruments. Paper and pencil scales
used were the FIRO-B (Fundamental Interpersonal Relations
Orientation-Behavior, Schtz, 1966), the Problem Patholog
ical Potential Scale (Blumberg, 1968), a therapist post
session evaluation form (see Appendix A; Blumberg, 1969),
and a client post-session evaluation form (see Appendix B;
Blumberg, 1969).
Design. Through matching on test scores, client sub
jects were assigned to therapists such that each therapist
had one client in each cell of two levels of compatibility
by two levels of client sex. Interviews were considered as
24

25
being composed of three segments, and the design, therefore,
can be viewed as a factorial design: 2X2X3, compati
bility X sex of client X segment of interview (repeated
measures). Where inappropriate to apply analysis of var
iance techniques, the independent variables and the depen
dent variables were examined by considering their correla
tional associations and graphs of interaction relationships.
Procedure. The raw data for this study were taken from
transcripts of interviews videotaped as part of a study by
Goldman and the author (Note 2). Therapists and potential
clients were tested on the FIRO-B, and the combined recipro
cal compatibilities for Inclusion and Control computed for
all pairs. Potential clients had been obtained through
their responding to a notice on the Psychology Department
experiment bulletin board (see Appendix C), and the therapists
were obtained by personal contact with the experimenters.
High compatibility was operationally defined as existing in
a pair when that pair's score fell within the bottom 5% of
that therapist's scores with all potential clients. Low
compatibility was defined as scores falling in the top 5%.
If insufficient matches were available to fill all four
cells [same- and opposite-sex clients and high and low
compatibility], additional potential clients were tested
until all 40 pairs were obtained. Clients were informed of
their selection for the second part of the experiment
through the phone message in Appendix D. Generally, each
client was informed of his or her selection for the study,

26
and told that their task was to select a real problem that
he or she faced and would be willing to discuss with a pro
fessional therapist, with the session being videotaped.
Therapists had been instructed that they would be con
ducting a single interview with each of four college stu
dents who had responded to a study entitled "An Analogue
Study of Counseling," and who had been instructed by the
experimenters to think about and select a real problem that
he or she was facing and would be willing to talk about
with a professional therapist. The therapists were also
informed that the interview would be videotaped and that
provisions had been made with the campus mental health
services for referral of the client person if the therapist
felt it was indicated.
At the time of the interview, the therapist and client
completed the paper and pencil rating sheets, including for
the client a pre- and post-Problem Pathological Potential
Scale and a post-evaluation form, and for the therapist a
post-evaluation form.
The interview was introduced by the experimenter oper
ating the videotape equipment, and he or she informed the
therapist and client jointly that they would have 50 minutes
to talk and that the experimenter would knock on the door
of the interview room at the end of 45 and 50 minutes. The
interview was videotaped by the experimenter, with the pre
sent author accounting for about 25% of the procedure.
Three minute segments were retaped by the author and
two collaborators onto master video-tapes and cassette audio-

27
tapes. The cassettes were used to make typed transcriptions
according to the rules for typists described by Gottschalk,
Winget, and Gleser (1969) and were checked for accuracy by
the experimenter. The segment locations were determined by
taking the first 3 minutes of the interview, and the approx
imate middles of the middle and final thirds of the inter
view. Detailed rules for determining the segments are found
in Appendix E.
Rating Procedures
The rating process was based on standard content
analysis procedures and terminology (Berelson, 1952). In
the case of most variables, it operated within what has
been termed the "classical model" (Marsden, 1971), mean
ing that what is measured is limited to the semantic
and syntactic aspects of communication. Content analysis
requires the definition of three ideas for the measurement
of each variable, and these are 1) the unit, 2) the cate
gory, and 3) the indicator (Berelson, 1952). Each of these
aspects of the content analysis process will be elaborated
as needed for each variable to be measured.
Units. With the exception of the unit "WORD," the unit
for rating each variable was scored separately and marked on
the transcript. Units utilized were the "WORD," the "PROPO
SITION," the "STATEMENT," the "SEGMENT," and the "INTERVIEW."
WORD is essentially self explanatory, and includes hyphe
nated words as one WORD. PROPOSITION is a more complex unit,
and was defined according to Lennard and Bernstein (1969)

28
as "a verbalization containing a subject and predicate
whether expressed or implied" (p. 57). STATEMENT was
defined as an uninterrupted sequence of PROPOSITIONS by
either therapist or client, and has been defined similarly
by many researchers (Butler, Rice, & Wagstaff, 1962;
Lennard & Bernstein, 1969; and Matarazzo, Saslow, & Mataraz-
zo, 1956). SEGMENT was defined operationally above as a
three-minute section of the interaction, and INTERVIEW is the
composite of the three SEGMENTS of an interview session.
Raters. Raters in the study were volunteers from the
clinical psychology introductory class, volunteers from the
psychology club in the Department of Psychology, and sub
jects from the subject pool in the psychology department,
all groups at the University of Florida. All potential
raters were screened and local references obtained to en
sure that the confidentiality of the materials would be
reasonably protected and to determine the motivation and
reliability of the raters. Appropriate tasks were assigned
for each person.
The experimenter scored the transcripts for the unit
PROPOSITION independently of a random sample of 13% of the
transcript scored by a separate rater. The rater was
trained according to the training manual for scoring units
(see Appendix F). Of over a total of 1802 PROPOSITIONS
jointly scored, there were disagreements on 3.9%. PROPO
SITIONS were indicated on these transcripts by separation
with diagonal marks.

29
Affective Verbalization. Affective verbalization was
scored utilizing three categories, labeled Primary Affect
Word, Secondary Affect Word, and Affect Idiom. Each of
these was scored separately in the content analysis pro
cedure, but the same general definition holds for all three
types of affective verbalization. Generally, an affective
statement is one which is "expressive of a subjective
reaction of an emotional tone" (Green & Marlatt, 1972), and
in the present use, statements of evaluation without affec
tive meaning, or statements referring to intellectual or
physiological states were excluded.
Primary Affect Words were scored using the unit: WORD,
the category: Primary Affect Word, and the indicator (rule
for assigning a unit to the category): inclusion on the
list of Primary Affect Words in Appendix H. This list was
generated by selecting words which almost always refer de
notatively to affective states. Examples are "angry,"
"sad," "happy," etc. Words on this list are in adjective
form, but a word from this list was scored as a Primary
Affect Word when it appeared in any of its various parts of
speech forms: noun, verb, adjective, or adverb. The
rater's task for this affective expression was to compare
each WORD of a transcript with this list and to label any
WORDs found by recording the WORD'S code number. The rating
and verification was handled in the same manner as with the
unit rating, and the raters were trained using the manual
in Appendix G. Three raters were used and in reliability

30
checks over 769, 353, and 343 PROPOSITIONS respectively,
error rates of 0%, .5%, and 0% were obtained.
Secondary Affect Words were scored using the unit:
PROPOSITION, the category: Secondary Affect Word, and the
indicator: inclusion on the list of Secondary Affect Words
in Appendix I. This list was generated by selecting words
which refer denotatively to affective states only when mod
ifying the verb "feel" or the verb "sense." Examples are
"alert," "slow," "inhibited," etc. All part of speech
forms of the words listed in Appendix I were scored. The
rater's task for this affective expression was to note the
subject's use of the verb "feel" or the verb "sense" in a
PROPOSITION or adjacent PROPOSITION modifying it, either
directly or by its being understood (as in an answer to a
question using one of these verbs). When either verb was
noted, the rater then used WORD as the unit, and compared
each WORD in the verb's modifier with the list of Secondary
Affect Words and labeled any WORDS found by recording the
WORD'S code number. The same rating and verification pro
cess was used, and the rater was trained using the manual
in Appendix G. Two raters were used for this task and in
reliability checks over approximately 350 PROPOSITIONS each,
no errors were obtained.
Affective Idioms were scored using the unit: PROPO
SITION, the category: Affective Idioms and Slang, and the
indicator: inclusion on the list of Affective Idioms in
Appendix J. This list was generated by selecting groups of

31
words which idiomatically refer denotatively to affective
states. Examples are "sick and tired (of)," "get a bang out
of," "hair stands on end," etc. Only part of speech changes
from the listed forms which maintain the denotatively af
fective meaning were also scored. The rater's task for this
affective expression was to compare each PROPOSITION of a
transcript with list J, and to record the code number of
that idiom's PROPOSITION. The same rating and verification
process was used as in previous rating, and the rater was
trained using the manual in Appendix G. Two raters were
used, and in reliability checks of approximately 350 PROP
OSITIONS each, error rates of 0% and .1% were obtained.
Self References. Self References were scored using
the unit: PROPOSITION, the category: Self Reference, and
the indicator: "unit is a 'verbal response by a subject
which expresses, descriptively or emotively, something
about the subject in relation to himself, others, or the
world'" (Green & Marlatt, 1972). This indicator was fur
ther specified by four rules utilized in training the rater,
and are in the training manual for this scoring task (See
Appendix G). The rater's task was to consider each PROPO
SITION in the transcript and to determine the presence of
any pronouns. If a pronoun was found, the PROPOSITION was
compared to the scoring rules, and if found to be a Self
Reference was labeled by recording the PROPOSITION code
number(s) involved. Three raters were used to score this
variable and reliability checks over 343, 769, and 353 PROPO-

32
SITIONS obtained, error rates of 7.0%, 3.1%, and 4.5%
respectively.
Therapist Strategies. The three therapist strategy
behaviors were scored through a similar use of content an
alysis. Instruction was defined as a therapist statement
or statements within the first 3 minute Segment of the in
terview in which the therapist communicates at least two
of three things: 1) behavior the client is to include
(i.e., "talk about your feelings"), 2) behavior the client
is to exclude (i.e., "Don't talk exclusively about your
thoughts"), and 3) examples of the general classes or areas
desired to be talked about (i.e., "YOU might talk about
your satisfactions and dissatisfactions, your concerns and
confidences"). Instruction was scored using the unit:
PROPOSITION, the categories: Instruction-1, Instruction-2,
Instruction-3, and the indicator: unit includes defined
communication. The rater's task on this rating was to con
sider each PROPOSITION within the transcript for an initial
3 minute Segment, and to label any Instruction categories
found by recording that PROPOSITION'S code number(s). Since
this rating task was somewhat less precise than the earlier
tasks, two non-experimenter raters were used, and the ex
perimenter's rating was to be used to settle disagreements.
The raters were trained using the training manual in Appen
dix G. Only one component of this strategy was found by
raters, and no reliability assessment would be meaningful.
This was extended to all three segments, and there was still
only one found.

33
Modeling, the second therapist strategy, was defined
earlier as the therapist's use of affective verbalization,
both with and without self reference. Since every thera
pist used some affective words, this variable is a rating
of how many affective verbalizations and how many affective
self references the therapist made. No ratings in addition
to those discussed above were necessary.
Reinforcement was scored by raters using the units:
WORD and PROPOSITION, the category: Reinforcer, and the
indicator: therapist makes one or two word expression of
agreement or approval, and specifically uses only those
words or expressions listed in Appendix K. Examples are
"mmm-hmm," "yes," "that's nice," etc. The rater's task was
to consider all therapist verbalizations, and to label all
words or expressions from Appendix K by recording the State
ment number of the therapist verbalization. Since the
typist is the first "rater" to decide whether or not to
place a questionable verbalization in the transcript, the
typing by the typist and verification by the experimenter
of the transcripts was considered the first stage of a two-
stage process of rating reinforcement. No assessment of
reliability was made at the first stage. The raters were
trained using the training manual in Appendix G. Two raters
were used, and over approximately 350 PROPOSITIONS each,
error rates of .4% and .7% were obtained.
Following the rating procedures, the necessary counts
over various summarizing units and contingent conditions
were made to obtain the following measures separately for

34
client and therapist: total number of WORDS, PROPOSITIONS,
and STATEMENTS per segment; total number of Primary Affect
Words, Secondary Affect Words, and Affect Idioms per seg
ment; total number of Primary Affect Words, Secondary
Affect Words, and Affect Idioms per segment when each oc
curred within a Self Referent PROPOSITION; number of dif
ferent types of Instruction categories per segment; number
of therapist Reinforcers preceeded by client affective ver
balization of any type per segment; number of client affec
tive verbalizations not followed by therapist Reinforcement
per segment.
The final measure for affective verbalization for both
client and therapist is the sum of the number of Primary
Affect Words plus the number of Secondary Affect Words plus
the number of Affect Idioms per Segment divided by the to
tal number of PROPOSITIONS in the segment being computed
and given that only one affective expression per PROPOSITION
has been counted in the sum. This gave a percentage score
after being multiplied by 100, and is called the client (or
therapist) affective verbalization score for Segment X.
The total affective verbalization score for the interview
is the sum of all affective expressions in the interview
divided by the number of PROPOSITIONS in the interview. The
same procedure was used for computing the affective self
reference scores.
The final measure for the Instruction strategy is "yes"
versus "no" as to use of that strategy according to the def
inition. The final measure for the Modeling strategy is

35
the therapist affective verbalization score as computed
above. The final measure for the Reinforcement strategy is
most easily expressed as a formula which recognizes that
Reinforcement might either encourage or discourage affec
tive verbalizations:
Rij = t(rw rQ)/(rw + r0)](x/8)(100),
where Rij is the Reinforcement variable score for client i
in segment j; rw is the number of times the therapist used
a verbal Reinforcer immediately after a client finished a
statement which contained an affective verbalization; rQ is
the number of times the therapist did not Reinforce an af
fective verbalization; and x is a value equal to the larger
of the two r's (rw or rQ). (The x/8 factor attempts to cor
rect the Reinforcement score for small numbers of reinfor
cers, which is an inference from Waskow's 1962 study.) The
total Reinforcement score for the interview is the mean of
the three Segment scores.

CHAPTER 4
RESULTS
Results are presented in terms of the three groups of
hypotheses discussed above and labeled: client verbaliza
tion and therapist strategies, compatibility and client sex,
and client verbalization and session outcome. When one ana
lysis is used for more than one hypothesis, it is presented
to the extent needed to examine the results for that hypothe
sis. When it is apparent that a result is markedly different
from the hypothesis and an alternative hypothesis is evident
by virtue of being the converse of the predicted relationship,
that alternative hypothesis is presented in this section
with the analyses used to examine it.
Client Verbalization and Therapist Strategies
The first group of hypotheses (numbers 1 through 4)
dealt with therapist strategies and client affective verbali
zation and client affective self reference. Each of these
variables was measured as described in Chapter 3. With re
spect to the therapist strategies, no full instances of the
Instruction strategy were found, and only one type 3 compo
nent of that strategy (see Appendix G) was scored in the
rating of all therapist responses. Consequently, this vari
able was dropped from further analyses. Also, although ther-
36

37
apist statements were scored for self references as part of
the Modeling strategy measurement, only 31 therapist affec
tive self references were found in all, making the inclusion
of that variable meaningless, and it was eliminated from
further analysis. This means that the strategy Modeling is
equivalent to therapist affective verbalization in all analy
ses. Scores for the other strategies and for the client
variables appeared adequate.
Hypothesis 1 predicted that each therapist would use
one or more of the strategies, and was informally analyzed
since some of the strategies were not dichotomous. Several
factors suggest that this hypothesis was not supported. First
of all, the Instruction strategy was clearly absent, as was
the use of modeling of affective self reference behavior.
Second, the frequency of modeling of affective verbalization
behavior was at or below client frequencies in many cases
and was generally low for therapists, inconsistent with our
expectation of a Modeling strategy. Third, the Reinforcement
behavior of therapists often did not involve a discriminant
use of reinforcers, but rather relatively high rates of re
inforcing all client verbalization within a particular seg
ment. Overall, therapists did not appear to use the defined
strategy behaviors systematically, and each therapist clearly
did not use at least one strategy as there were some thera
pists who did not approach using any of the strategies.
Hypothesis 2 predicted that both client affective ver
balization and client affective self reference would increase
across the session. This was analyzed by checking for a

38
significant segment effect in an analysis of variance.
This was part of a larger analysis presented in the Group
2 hypotheses. For both client affective verbalization (see
Table 1) and for client affective self reference (see Table
2), there was no significant segment effect, and in both
cases, the trend was for the first segment to be the highest
of the three segments and for the second segment to be
lowest.
Hypotheses 3 and 4 predicted that there would be a sig
nificant positive relationship between the greater use of
a therapist strategy and larger increases in client affec
tive verbalization and client affective self reference.
Since the therapist strategy variables occur within the same
interactions which also provide the data for the client
measures, statistics presuming independence were inappro
priate. Consequently the associations predicted to occur in
the data were analyzed by examining the correlations in the
data. To determine the significance of a correlation for
interpretation purposes, the correlation was compared to the
null hypothesis that the correlation equals zero, with
alpha = .05 for rejecting the null hypothesis. A correla
tion was defined as approaching significance if its signif
icance level fell below .09. Unless otherwise noted corre
lations are Pearson r's. Spearman correlation coefficients
were used in instances where comparisons were to be made
which involved class variables (i.e., compatibility, sex of
client, and sex of therapist).
As indicated above, some strategies did not occur in

TABLE 1
ANOVA-1 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX X SEGMENT
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
BETWEEN SUBJECTS
168.572
39
2
COMPATIBILITY
4.798
1
4.798
1.133
NS
3
CLIENT SEX
.901
1
.901
.213
NS
4
COMPATIBILITY X CLIENT SEX
10.365
1
10.365
2.447
NS
5
SUBJECTS WITHIN GROUPS
152.508
36
4.236
6
WITHIN SUBJECTS
271.001
80
7
SEGMENT
3.566
2
1.783
. 513
NS
8
COMPATIBILITY X SEGMENT
4.915
2
2.457
. 707
NS
9
CLIENT SEX X SEGMENT
10.417
2
5.208
1.499
NS
10
COMPATIBILITY X CLIENT SEX X SEGMENT
1.874
2
.937
.270
NS
11
SEGMENT X SUBJECTS WITHIN GROUPS
250.229
72
3.475
12
TOTAL
439.573
119
Note: Dependent variable is Client Affective Verbalization.
NS p> .05
OJ
V£>

TABLE 2
ANOVA-2 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX X SEGMENT
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
BETWEEN SUBJECTS
196.345
39
2
COMPATIBILITY
5.453
1
5.453
1.155
NS
3
CLIENT SEX
.691
1
.691
.146
NS
4
COMPATIBILITY X CLIENT SEX
20.263
1
20.263
4.293
*
5
SUBJECTS WITHIN GROUPS
169.938
36
4.720
6
WITHIN SUBJECTS
223.595
80
7
SEGMENT
8.932
2
4.466
1.596
NS
8
COMPATIBILITY X SEGMENT
1.308
2
. 654
.234
NS
9
CLIENT SEX X SEGMENT
8.730
2
4.365
1.560
NS
10
COMPATIBILITY X CLIENT SEX X SEGMENT
3.167
2
1.583
.566
NS
11
SEGMENT X SUBJECTS WITHIN GROUPS
201.455
72
2.798
12
TOTAL
419.940
119
Tests
for Simple
Compatibility and
Client Sex Effects after
Significant Interaction
1
COMPATIBILITY
@ C=1 (MALE)
2.315
1
2.315
. 490
NS
2
COMPATIBILITY
@ C=0 (FEMALE)
23.401
1
23.401
4.958

3
CLIENT SEX @
COMPATIBILITY =1
(HIGH)
14.337
1
14.337
3.037
AP
4
CLIENT SEX @
COMPATIBILITY =0
(LOW)
6.617
1
6.617
1.402
NS
Note: Dependent variable is Client Affective Self Reference.
* P < 05 NS £>.10 AP £<.10
O

41
this sample, and this leaves Modeling and Reinforcement to
be considered. The analyses of hypotheses 3 and 4 for Model
ing will be presented first.
Since these hypotheses are dealing with changes in
client behaviors, the Modeling score was correlated with
scores representing the change of a client's behavior.
Accordingly, correlations were computed between the various
segment levels of therapist affective verbalization and the
change scores for the two client measures across various
parts of the interview (see Table 3). For client affective
verbalization, one correlation was significant, this between
therapist affective verbalization in the first segment and
the change in client affective verbalization from segment 1
to 2, but in the opposite to predicted direction (r = -.44,
^ = .004). The correlation between therapist affective ver
balization in the first segment and client affective ver
balization change from the first segment to the last ap
proached significance, again in the opposite to predicted
direction, (r = -.29, £ = .067). Therapist affective ver
balization correlations with client affective self reference
change scores showed the same phenomenon with both correla
tions significant (r = -.46, p = .003; r = -.41, p = .009).
A scatter-plot of the segment 1 therapist affective verbal
ization by segment 1 to 2 client affective verbalization
change scores shows a clear association between higher
therapist scores and negative client change scores and be
tween lower therapist scores and positive changes (see
Figure 1). This suggests the need for a clearer under-

42
TABLE 3
CORRELATIONS OF MODELING AND CHANGE SCORES
SEGMENT
CHANGE
MODELING: THERAPIST AFFECTIVE
VERBALIZATION BY SEGMENT
1
2
3
Total
CLIENT AFFECTIVE VERBALIZATION
1 to
2
-.44**
.04
-.01
-.13
2 to
3
.11
-.11
.07
.01
1 to
3
-.29
-.08
.06
-.11
CLIENT AFFECTIVE SELF REFERENCE
1 to
2
-.46**
.08
-.08
-.14
2 to
3
.04
-.17
.01
-.08
1 to
3
-.41**
-.11
-.06
-.23
** £
CLIENT AFFECTIVE VERBALIZATION
CHANGE FROM SEGMENT 1 TO 2
43
+ 10
+ 5
A
0 5 10
THERAPIST AFFECTIVE VERBALIZATION
IN SEGMENT 1
15
Legend: A 1 observation; B = 2 observations, etc.
SCATTER PLOT OF A THERAPIST
SCORE BY A CLIENT SCORE
FIGURE 1

44
standing of the nature of the affective verbalization scores
and their relationships.
Intercorrelations were run between all client and thera
pist affective expression scores, a total of 12 variables,
and many highly significant correlations were present. (See
Table 4 for specifics of the following discussion of effects.)
Each segment level of client affective verbalization is
highly correlated with its respective self reference vari
able (r's = .95, .93, .85). The affective verbalization
levels are not, however, correlated with each other. Of the
self reference levels, only segment 1 and 2 are correlated,
but only modestly (r = .32). Therapist affective verbaliza
tion in segment 1 is strongly correlated with both client
measures in segment 1 (r = .66 and .67) and with no other
segment levels of either. Therapist affective verbalization
in segment 3 is modestly correlated with all levels of cli
ent affective self reference only (r's = .34, .32, .40, .38),
as well as being correlated with therapist affective verbal
ization in segments 1 and 2, (r's = .58, .34), which are not
correlated with each other. The total scores for the thera
pist measure and both client measures are modestly to highly
correlated with each other and all affective verbalization
segment variables (range of r's = .32 to .92), with the ex
ceptions of total client affective verbalization with ther
apist affective verbalization in segment 2 (r =.28, £ = .076)
and the total therapist measure with client affective self
reference in segment 3 (r = .27, jd = .089), both of which
correlations approach significance.

TABLE 4
AFFECTIVE MEASURES INTERCORRELATED:
CORRELATIONS AND PROBABILITIES
t_n
VARIABLE
AND
VARIABLE
AND
SEGMENT
NUMBER
SEGMENT
NUMBER
CAV
CASR
TAV
1
2
3 T
1
2
3
T
1
2
3
CAV 2
.20
. 207
CAV 3
. 17
-.02
. 288
. 877
CAV T
. 75
. 56
.58
. 001*
.001*
. 001*
CASR
1
. 95
.001*
.24
. 140
. 17
.280
. 75
.001*
CASR
2
. 28
.083
.93
.001*
.01
.965
.60
.001*
. 32
. 045*
CASR
3
. 18
. 265
.11
. 495
. 85
.001*
.59
. 001*
.22
. 161
.13
.405
CASR
T
.69
.001*
.59
.001*
.45
.003*
.92
. 001*
.77
.001*
.69
.001*
.60
.001*

TABLE 4-continued
TAV
1
.66
.001*
.09
. 5.85
.24
. 126
.55
.001*
.67
.001*
.10
.526
.17
.298
.48
.002*
TAV
2
.21
.197
.26
. 109
.09
.566
.28
.076
.24
.141
.32
.047*
. 11
.504
.32
.042*
.19
.250
TAV
3
. 34
. 029*
. 32
.046*
. 39
.011*
.52
. 001*
. 38
.015*
.27
.096
. 31
.052
.42
. 006*
.58
.001*
. 34
. 034*
TAV
T
.53
.001*
. 35
. 026*
. 36
.024*
.63
.001*
.55
.001*
.35
.027*
.27
.089
.55
.001*
.73
.001*
.60
.001*
.89
. 001*
Note: The upper number of each pair is the correlation coefficient. The lower number
is the probability that the null hypothesis (r = 0) is true.
Key: CAV = Client Affective Verbalization
CASR = Client Affective Self Reference
TAV = Therapist Affective Verbalization
1 = Segment 1
2 = Segment 2
3 = Segment 3
T = Total interview.
* p< .05

47
The lack of correlation within segment levels of each
of the client measures is consistent with the lack of a
segment effect in the earlier analyses, and the high corre
lations between so many variables suggests an alternative
hypothesis. This is that clients are influencing the
therapists' frequencies of use of affective verbalizations,
or equivalently, that in at least segment 1, therapists are
using a following style which results in a high correlation
between client and therapist affective verbalization scores.
Two analyses were completed to examine these alternative
hypotheses: cross-lagged panel correlations and tabulating
therapist and client use of the same affective expressions.
Cross-lagged panel correlation (Campbell & Stanley,
1963) would predict that the "causal" force (if other than
a separate underlying factor) in changing affective verbali
zation will have a higher correlation between its earlier
measurement and the "effect" factor's later measurement than
the correlation between the effect's earlier measurement and
the cause's later measurement. In all three cases to con
sider given three time periods (segments 1 to 2, 2 to 3, and
1 to 3), the correlation between the earlier client variable
and later therapist variable is larger than the correlation
between the earlier therapist variable and the later client
variable (see Table 5 of cross-lagged panel correlations).
To attempt further clarification of the notion that
therapists might be "following" client affective verbaliza
tion, ratings from a sample of 12 dyads (10% of the total
segments) were examined to see what words the therapist in

48
TABLE 5
SUMMARY OF CROSS-LAGGED PANEL CORRELATIONS
CLIENT OR
THERAPIST3
MEASURE AS
"CAUSE"b
TIME LAG EXPRESSED
AS SEGMENT DIFFERENCE
SEGMENT
1 AND 2
SEGMENT
2 AND 3
SEGMENT
1 AND 3
CLIENT AFFECTIVE
VERBALIZATION
THERAPIST
.09
.09
.24
CLIENT
.21
. 32*
.34*
CLIENT AFFECTIVE
SELF REFERENCE
THERAPIST
. 10
. 11
. 17
CLIENT
.24
.27
. 38*
Note: No
correlation pairs are
significantly
different
using £<.05 as criterion in a one-tailed test.
a Therapist measures are all Therapist Affective Verbal
ization scores.
b "Cause" indicates that the measure for that person is
for the earlier of the two time periods.
*
£<.05

49
each pair had used for affective verbalization scores and
then to determine whether the client had used that word, and
if so whether before or after the therapist response and
whether immediately or not. Immediately is defined here as
being within the first statement following or preceding the
one containing the affective verbalization.
The sample was chosen randomly, with the stipulation
that each therapist must be chosen once and that there be
four dyads chosen from each segment. Table 6 summarizes the
results of the tabulation. It can be seen from the table
that although the total number of responses in each segment
does not vary greatly, the distribution of the responses in
terms of their relationship to client responses using the
same word or phrase is quite different. In segment 1, ther
apists could be described as "following" the client response.
In segment 2, clients could be described as "following" the
therapists. In segment 3, there is no apparent relationship
between the particular affective words and phrases used by
the therapists and clients. The sample is of course small,
and should be interpreted conservatively, but suggests a
trend toward differential leading and following therapist
styles in different segments.
Overall, it must certainly be concluded that the data
clearly disconfirm the third and fourth hypotheses, with
respect to the Modeling strategy.
Reinforcement is the second strategy to be considered
in hypotheses 3 and 4. Here, correlations were obtained be
tween segment levels of the Reinforcement variable and cli-

50
TABLE 6
TABULATION OF FOLLOWING VERSUS
LEADING THERAPIST RESPONSES
SEGMENT
TYPE OF
RESPONSE
1
2
3
IMMEDIATELY3 AFTER
CLIENT
0
0
0
DELAYED13 AFTER
CLIENT
5
0
0
IMMEDIATELY BEFORE
CLIENT
0
4
0
DELAYED BEFORE
CLIENT
0
2
1
NO SAME CLIENT
RESPONSE
0
2
5
TOTAL
5
8
6
Note: Responses are therapist affective verbalizations,
and tabulations reflect the type of relationship to
client's use of the same word or phrase within a given
Segment of an interview.
a "Immediately" is defined as the statement right before
or right after the statement containing the affective
expression.
b "Delayed" is defined as any position within the same
Segment other than "immediately" before or after.

51
ent change scores. Modest, though significant, negative cor
relations were observed between some variable pairs (see
Table 7). Reinforcement in the first segment was negatively
correlated with client affective verbalization change from
the second to the third segment and from the first to the
third segments (r's = -.35 and -.44), suggesting that greater
and more discriminating use of verbal reinforcement of cli
ent affective verbalization in the first segment was related
to decreases in that client measure from segment 1 to 3 and
from 2 to 3. This was also true for client affective self
reference (r's = -.53, -.46). Second segment Reinforcement
was negatively correlated with change scores for both client
measures (affective verbalization and self reference) from
segment 2 to 3 (r's = -.32 and -.51). Total Reinforcement
was also negatively correlated to both measures from segments
1 to 3 (respectively r's = -.31, -.35) and to affective self
reference from segments 2 to 3 (r = -.50).
To examine the possible effects of simple number of re
inforcers or quantity of verbalization, these variables were
also correlated with the change scores (see Table 8). For
client affective verbalization, number of reinforcers fol
lowed the same pattern as the Reinforcement variable, though
at a lower correlation coefficient level such that only re
inforcers in the first segment with client change from seg
ments 1 to 3 and total reinforcers with change from 1 to 3
were significant negative correlations (r's = -.33, -.33).
Client affective self reference showed exactly the same pat-
with the number of reinforcers as with the Reinforce

52
TABLE 7
CORRELATIONS OF REINFORCEMENT STRATEGY
AND CLIENT CHANGE SCORES
CHANGE
THERAPIST REINFORCEMENT
STRATEGY BY SEGMENT
BY
SEGMENTS
1
2
3
TOTAL
CLIENT
AFFECTIVE
VERBALIZATION
1 to
2
-.04
. 12
-.08
.01
2 to
3
-.35*
-. 32*
.16
-.29
1 to
3
-.44**
-.25
.10
-.31*
CLIENT
AFFECTIVE
SELF REFERENCE
1 to
2
. 12
.28
.02
.20
2 to
3
-.53**
-.51**
.05
-.50**
1 to
3
-.46**
-.29
.07
-.35*
* £<.05
** £<.01

53
TABLE 8
CORRELATIONS BETWEEN NUMBER OF
REINFORCEMENT RESPONSES AND
CLIENT CHANGE SCORES
SEGMENT
CHANGE
NUMBER OF THERAPIST REINFORCEMENTS
BY SEGMENT
1
2
3
TOTAL
CLIENT
AFFECTIVE
VERBALIZATION
1 to
2
-.01
-.03
-.11
-.05
2 to
3
-.28
-.20
-.13
-.25
1 to
3
-.33
-. 26
-.26
-.33*
CLIENT
AFFECTIVE
SELF REFERENCE
1 to
2
.11
. 11
.08
.12
2 to
3
-.45**
-.34*
-.28
-.43**
1 to
3
-.38*
-.26
-.22
-.34*
* £<.05
** £
54
ment variable, though with somewhat lower correlation coef
ficients .
An important incidental finding of these analyses was
that, with the exception of Reinforcement in the third seg
ment, all the Reinforcement variables are correlated with
all segment levels of number of reinforcers, number of state
ments in a segment, and number of therapist PROPOSITIONS in
a segment (the closest measure available to quantity of ver
bal output). (See Table 9 on correlations of raw scores
and Reinforcement measures.)
Overall, the data did not support the third and fourth
hypotheses with respect to the Reinforcement strategy, and
suggested an opposite to predicted trend.
The analyses of the Group 1 hypotheses, then, have
generally disconfirmed hypotheses 1 through 4, with support
being demonstrated for an alternative hypothesis regarding
the direction of influence between therapist and client.
Compatibility and Client Sex
The primary focus of the hypotheses in Group 2 is
hypotheses 5a and 6a predicting interaction effects of
compatibility and client sex with therapist strategies and
client affective expression change scores. The examination
of these hypotheses is in some ways a moot point since as
stated, these hypotheses relied on supporting hypotheses
2, 3, and 4 above. In order to assess the general relation
ships predicted by these hypotheses, however, plots of the
variables' relationships of interest were made and examined

55
TABLE 9
CORRELATIONS BETWEEN NUMBER OF
REINFORCEMENT RESPONSES AND
REINFORCEMENT STRATEGY
NUMBER OF
THERAPIST
REINFORCEMENTS
BY SEGMENT
THERAPIST REINFORCEMENT
BY SEGMENT
STRATEGY
1
2
3
TOTAL
1
.54**
.53**
o
r1
.57**
2
.49**
.69**
to
00
.67**
3
.43**
.53**
. 36*
.59**
TOTAL
.58**
.68**
.28
.71**
*

. 0 5
¡3 < 01

56
for effects. (For examples, see Appendix L.) Generally, it
was observed that the plots of mean scores for various var
iable groups suggested support for many complex interactions,
but of the ones of those which could be tested, only the
one interaction from the analyses of variance reported be
low was significant. The variability within each cell of
the data for any of these plots is quite large, and these
data must be approached very cautiously, and with the assump
tion that all of the effects are statistically non-signif
icant .
Interpretation is further complicated by what appears
to be a lack of independence of Modeling, Reinforcement,
compatibility, client sex, and level of client affective
verbalization, with an underlying factor of what appears to
be a high-low dimension on level of client affective verbali
zation in segment 1. In other words, female clients tended
to be higher than males overall, especially when in the
high compatible group. When high Modeling dyad clients or
high Reinforcement dyad clients are separated from the total
sample, they also appear more likely to be these high com
patible female clients than any others. This implies that
the creation of class variables from the continuous strategy
measures (which was done to create plots) is questionable
since it is likely to create groups which are similar in un
known ways with respect to their relationship with the de
pendent measures, while their class labels would suggest we
are dealing with different variables. In other words, split-
ting the sample on the basis of the Modeling strategy, for

57
example, means that therapist affective verbalization scores
determine the groups. These groups, however, also split un
evenly with respect to client affective verbalization scores
If the scores were determined independently, then we could
interpret that as a significant effect. In the present
case, it only means there are effects of several variables
contained in any examination of the variable "Modeling."
Examination of effects of compatibility and client sex
without reference to therapist strategy could be appropriate
ly handled by analysis of variance. The normalities of
the distributions of client affective verbalization and
client affective self reference were plotted and a square
root transformation for scores including low scores was ap
plied to analysis of variance data as suggested by Kirk
(1968). (For examples, see Appendix M.) The transformation
applied was
x' = (x)l/2 + (x + 1)^/2.
Four analyses of variance were computed. The first two
were split-plot factorial designs with two between block
treatments (compatibility and client sex) and one within
block treatment (segment or repeated measurements), using
client affective verbalization as the dependent measure in
analysis of variance 1 and client affective self reference
in analysis of variance 2. The second two analyses were
2X2 factorials (compatibility and client sex), using
client total affective verbalization as the dependent mea
sure in analysis of variance 3 and client total affective

58
self reference in analysis of variance 4.
Hypothesis 5b (1) was disconfirmed by the non-signif
icant compatibility X segment effect in analysis of variance
1 (see Table 1), indicating that compatibility alone did not
account for changes in client affective verbalization over
the session. Hypothesis 5b(2) predicted this same relation
ship for client affective self reference and was also dis
confirmed in analysis of variance 2 (see Table 2). Hypoth
esis 5b (3) predicted that compatibility would show a signif
icant effect for client total affective verbalization, and
hypothesis 5b (4) made this prediction for client total
affective self reference. Neither hypothesis was supported
as indicated by analysis of variance 3 in the former case
(see Table 10) and by analysis of variance 4 in the latter
case (see Table 11). This effect did approach significance,
however, in this last analysis, suggesting that compatibil
ity alone has some influence on the total frequency of
clients' use of affective self reference.
The analysis of these hypotheses for client sex was
parallel to the analyses for compatibility. Hypothesis
6b(1), predicting that female clients but not male clients
would increase across the session, was analyzed by the
client sex X segment effect in analysis of variance 1 for
client affective verbalization (see Table 1), and 6b(2),
in analysis of variance 2 for client affective self refer
ence (see Table 2). Neither hypothesis was confirmed.
Hypotheses 6b(3) and 6b(4) predicted that female clients
would be higher than males in total affective verbalization

TABLE 10
ANOVA-3 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
COMPATIBILITY
19.072
1
19.072
2.90
NS
2
CLIENT SEX
2.601
1
2.601
. 40
NS
3
COMPATIBILITY X CLIENT SEX
11.990
1
11.990
1.83
NS
4
ERROR
236.500
36
6.569
5
TOTAL
270.163
39
Note: Dependent variable is Client Total Affective Verbalization.
NS 05
On

TABLE 11
ANOVA-4 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
COMPATIBILITY
22.052
1
22.052
3.69
AP
2
CLIENT SEX
1.005
1
1.005
.17
NS
3
COMPATIBILITY X CLIENT SEX
19.740
1
19.740
3.30
AP
4
ERROR
215.085
36
5.974
5
TOTAL
257.882
39
Note: Dependent variable is Client Total Affective Self Reference.
NS d > 0 5
AP £<.10
cn
o

61
and affective self reference respectively. These hypothe
ses were tested in analyses of variance 3 and 4 respective
ly, and were not supported (see Tables 10 and 11).
Though not stated as a specific hypothesis, the inter
action effect of compatibility X client sex X segment for
client affective self reference was significant (see Table
2), though it was not significant for client affective verbal
ization (see Table 1). Simple main effects tests on the sig
nificant interaction (see Table 2) indicated that compati
bility was a significant effect only when considering female
clients. Sex of client approached being a significant
effect when considering clients in highly compatible dyads
only.
Client Verbalization and Session Outcome
The third group of hypotheses deals with examining
the relationship between the client verbalization measures
and the measures of session outcome. Hypothesis 7a
predicted that for both client affective verbalization and
client affective self reference there would be a significant
correlation with client evaluation, therapist evaluation,
and changes in pre- to post-session change scores on
the problem rating scale. Of these only two correlations
were significant, and they partially confirmed the predic
tion. Total client affective verbalization and this verbal
ization in the second segment were correlated with therapist
evaluation (r = .34, £ = .031; r = .37, £ = .020). The rela
tive independence of client affective verbalization in seg-

62
ment 2 from the multiple variables correlating with it in
segment 1 makes this finding particularly interesting, since
it raises the possibility that therapists do assess the help
fulness of an interview in terms of client affective expres
sion. This is only in the context of other variables, which
were not identified in this study, as the variance accounted
for in those correlations is less that 14%.
Part b of hypothesis 7, predicting an improvement in
the above correlation when considering compatibility, was
examined by computing the correlation of client affective
verbalization total and for segment 2 with therapist evalu
ation separately for each level of compatibility. This hy
pothesis was supported as the correlations for high compat
ible dyad clients are r = .697 and .594, while in the low
compatible clients, these correlations are .413 and .221 re
spectively. Both differences are significant at the .05
level using a one-tailed test.
Exploratory Analyses
The significance of compatibility and client sex has
been supported in this study, and additional exploratory
analysis was undertaken. One aspect of this was to explore
further possible associations in the data between compati
bility, sex of client, and sex of therapist. This was done
through the computation of Spearman correlations with all
affective verbalization measures, the Reinforcement measures,
affective verbalization change scores, and the three outcome
measures. Two modest correlations were significant, between

63
sex of therapist and client affective self reference in seg
ment 1 (r = .32) and Reinforcement in segment 1 (r = -.40).
The negative direction of the correlation indicated that the
association was between female therapist and higher levels
of client affective self reference in the second segment (or
male therapist and lower Reinforcement measures). Sex of
therapist approached significance in the same direction with
client affective verbalization in segment 1 (r = -.27,
¡D = .088), total client affective self reference (r = -.28,
p = .083), and total Reinforcement (r = -.31, ¡3 = .053).
Sex of therapist approached significance in the opposite di
rection on client affective self reference change from first
to last segment (r = .30, g_ = .057), indicating that male
therapists were somewhat more associated with higher over
all change in client affective self reference scores than
female therapists. Sex of client approached correlation
with client first segment affective verbalization and self
reference (r = -.28, p = .078; r = -.29, £ = .069), with fe
male clients tending to be associated with higher scores.
The second exploratory analysis was to examine the com
ponents of the compatibility score. Since the interview
sample includes only individuals in certain matched pairs,
the significance of any correlations are very suspect for
interpretation, but may suggest directions for further ex
ploration of the compatibility dimension. All basic com
patibility measures were computed and were correlated with
all affective verbalization measures including change scores,
Reinforcement measures, and outcome measures. Several cor-

64
relations were significant, though no particular patterns
were apparent. All compatibility variables had at least one
significant correlation except for reciprocal compatibility
on Inclusion and Affection and interchange compatibility on
Inclusion. All significant correlations with affective ver
balization measures were for second segment measures and
with originator compatibilities. Originator compatibilities
for inclusion and for affection were negatively correlated
with therapist affective verbalization (r's = -.35, -.34)
and originator compatibility for control was correlated with
client affective self reference (r = .34). Reciprocal com
patibility for control was correlated with client affective
verbalization change from segment 2 to 3 (r = -.37), and in
terchange for affection is correlated negatively with client
affective self reference change from segments 2 to 3 and
from 1 to 3 (r's = -.36, -.32). Interchange for affection
is also correlated positively with Reinforcement in segment
1, 2, and total (r's = .38, .34, and .37). Originator com
patibility for control is correlated positively with total
Reinforcement (r = .32). Interchange for control is corre
lated positively with the Problem Pathological Potential
Scale change score (r = .31).

CHAPTER 5
DISCUSSION
The primary focus of this study was on defining client
affective verbalization and client affective self reference,
on defining therapist strategies which previous research has
indicated are effective in increasing the above client meas
ures across an analogue interview, and on examining actual
therapists in experimental interviews with respect to their
use of these strategies and the effects resulting from the
use of these strategies. Results of the study indicate that
the defining functions of the study appeared adequate, but
that therapists in this sample do not use the defined strata
gies, in any systematic form at least, and that the client
measures do not change consistently across the interviews.
Although there was a lack of significant changes in cli
ent scores across the interview, there was a trend toward
changes being important in accounting for the variance in
the sample. The variation in segment changes was large
within groups of the treatment blocks and suggests that im
portant variables were not identified.
Not only did the results not support therapists as
utilizing the predicted strategies, but there was also evi
dence suggesting that clients are more in control of the
amount of affective verbalization than are therapists. This
65

66
was most marked in the first segment where therapists used a
following style, with a trend toward therapists assuming a
more directive role in the second segment. Additional in
vestigation of this phenomenon might prove helpful in clari
fying these issues.
It is possible that in this sample a ceiling effect oc
curred where clients' affective verbalization rates were so
high from the beginning of the interview that it did not ne
cessitate therapists using strategies to increase that fre
quency. The data indicates that the rates were substantial,
and that therapist interventions of higher modeling or rein
forcement behaviors could only have inhibited client behaviors.
Well controlled and highly specific strategies, as used in
previous research with confederate therapists, might increase
rates even in this sample of clients, but such an increase is
unlikely when therapists use relatively unspecific strategies
and when they are possibly content with the level of client
affective verbalization as emitted. The client sample in
this study was similar to that used in previous research
(college students), but given the differences in inclinations
of the therapists not to work toward increasing reasonably
high rates of verbalization, a less articulate population of
clients may have elicited different strategies from these
same therapists. The relationship between this study and pre
vious research on these therapist strategies might be clari
fied by another study utilizing such a sample of clients or
perhaps more easily, by utilizing confederate therapists in
the research format of the present study.

67
Analyses of the group 2 hypotheses resulted in one sig
nificant effect, the interaction of compatibility, client
sex, and segment, indicating that in terms of segment changes,
the compatibility manipulation was effective only for female
clients. This is consistent with the literature cited earlier
which also demonstrated this effect (Mendelsohn & Rankin,
1969).
Correlations between therapist evaluation and client
affective verbalization supported the contention that thera
pists believe client affective verbalization is important to
treatment outcome. Other outcome measures, client evaluation
and problem rating scale change scores, were not related to
client affective expression measures. Limited ranges of
scores for these two outcome measures were a problem and more
adequate measures should be used in future research.
Various analyses and correlations offer some help in
the problems of designing further studies to investigate the
implications of the present data. The large error variance
found so often in the analyses suggests that 1) this must be
expected and only very powerful effects can be examined,
2) the flexibility of the interview situation must be re
stricted in order to provide better control, or 3) a broader
range of clients should be studied. The other alternative
is to consider more variables, but this is probably prohibi
tive on the one hand in terms of investment of resources or
on the other, in terms of an N that is too small for so many
variables.
There is a danger inherent in increasing control in the

68
present type of study. On the well-controlled end of the
spectrum, there are the studies which generated the litera
ture used to define the therapist strategies and the depen
dent variables. The dependent variables seemed adequate,
but the "therapist strategies" were apparently foreign to
actual therapists in a similar situation! On the other
hand, the present study was, by design, loose with respect
to these variables, and consequently is very limited in its
richness for interpretation or generalizability, vis-a-vis
theoretical precision. A compromise between these two is
sues might be a study similar to the present one, with more
attention to the construct validity of the variables and
some assurance, by manipulation of therapist or client, that
the issue of strategies to increase client affective verbal
ization is clearly defined in the data.
The present data are also helpful in demonstrating as
sociations between affective verbalization and variables
such as client sex, therapist sex, compatibility, quantity
of verbalization of all types, inconsistency of change
across standard segments, and other affective verbalization
measures. Each of these must either be controlled or mea
sured as a variable in subsequent research if it is to re
sult in more significant, interpretable findings.
This appears particularly true of compatibility, since
not only did the compatibility manipulation seem reasonably
sound (though not significant overall), but the calculated
compatibility components also received some support as po
tentially related to the other variables of the study. Fur-

69
ther investigation of this would require either separate
samples matched on each type of compatibility component, or
an adequately-sized group constructed randomly without re
gard to compatibility, and then analyzed in terms of com
patibility measures. Economy for investigating more than
one type of compatibility probably requires the latter
approach.
Conclusion
This study grew out of a review of psychotherapy litera
ture as conceptualized within a general communications model.
The model was translated into the terms of research which
has considered therapist behaviors and their relationships
to clients' use of affective language. A study was designed
for testing these relationships in a sample of clients simi
lar to these previous studies, but using a sample of thera
pists which were different in not being confederates and
also in being practicing, professional psychotherapists.
This study was also different in including the variables com
patibility and client sex as independent variables.
This study's contributions appear to lie in two areas.
The first area is in its development of methodological tools
for the relatively objective measurement of client and thera
pist use of words and phrases denoting affective experience
both with and without reference to the self. Previous re
search which has provided similar tools has either maintained
significantly subjective elements, failed to retain materials
for use on other studies, or failed to adapt materials to

70
the specific purpose of measuring the use of such language
by clients and therapists.
The second area of contribution is in applying the
above methodology to the examination of a group of variables
in a sample of analogue psychotherapy interviews. The vari
ables considered had been found in previous research to have
certain relationships to each other, but these relationships
were generally absent in this study. Two significant re
lationships were demonstrated. Compatibility was found to
be effective under some conditions for female clients, but
not for male clients. Client affective verbalization in
the second segment and for the total interview were found
to be significantly correlated with therapist evaluation,
which provided modest support for the validity of the client
affective measure as an indicator of some important processes
in the interview. Therapist strategy variables were either
found not to be utilized by therapists in the present sam
ple, or to result in inhibitory effects, with clients more in
apparent control of the frequency of affective verbalization
in the session.
Suggestions for future research are made, and argue for
the use of more control than in the present study. Caution
is also made, however, for such strict control that the re
search looses its relevance to applied practice.

APPENDICES

APPENDIX A
THERAPIST POST-SESSION EVALUATION FORM
Name:
Please answer the following questions as candidly as
you can. Feel free to make additional comments where you
wish to. Circle one of the alternatives.
1. Did you like the person whom you interviewed?
very much slightly not at all
2. If this interview had occurred in the context of your
clinical practice, would you:
want to refer the client unsure continue to see client
3. Do you think this experience was valuable for the client?
very much slightly not at all
4. Do you think your client is now better able to solve or
live with his or her problem?
yes unsure no
Additional comments:
72

APPENDIX B
CLIENT POST-SESSION EVALUATION FORM
Name:
Your participation in this experiment is now concluded.
Please answer the remaining few questions as candidly as you
can. Feel free to make additional comments where you wish
to. Circle one of the alternatives.
1. Did you like the person who interviewed you?
very much slightly not at all
2. Did you enjoy the experience of being interviewed?
very much slightly not at all
3. Do you think the experience was valuable?
very much slightly not at all
4. Did it help you to solve, or live better with, your
problem?
yes unsure no
5. Do you think participating in this experiment has made
you more likely to seek professional help should the
occasion ever arise?
yes unsure no
Additional comments:
73

APPENDIX C
NOTICE FOR POTENTIAL CLIENT SUBJECTS
An Analogue Study of Counseling
We are conducting a study about counseling and
psychotherapy, and need a number of subjects to be
clients who will talk with therapists. The first part
of the study involves taking some paper-and-pencil tests,
and only a part of the subjects who take the tests will
go on to the second part of the study. The first part
of the study will take about 30 minutes, and you will
receive 1 half-hour credit. The second part of the study,
if you are selected, will take about an hour-and-a-half,
and you will receive 3 half-hour credits. If you would
like to participate in this study, please sign up below
for only one of the testing times.
74

APPENDIX D
VERBATIM TELEPHONE CONTACT
FOR INFORMING SELECTED CLIENTS
We're calling about the Psychology Study you're in.
Your record was selected to be matched to a person in the
other group. Now we would like to arrange a time for you
and the other person to continue in the study. Your part
of the study is to think about and select a real problem
that you face. It should be something that you don't mind
discussing with this second person. The second person is
one of a number of therapists who consented to participate
and who will discuss this area with you. You will meet at
(time) to complete a few brief paper and pencil ques
tions and then have a conversation which will be videotaped.
After the discussion there will be a few more paper and
pencil items. The entire time for this part of the study
will take about 1 1/2 hours.
75

APPENDIX E
RULES FOR LOCATING SEGMENTS
WITHIN FULL INTERVIEWS
Master tapes are composed of three-minute segments of
interviews with interviews and segments within interviews
in randomized order. There are a total of 120 segments with
3 segments from each of the 40 approximately 50 minute in
terviews. Position of segments within each interview was
determined by these rules.
Rule 1. Segment 1 begins at the beginning of the in
terview and continues as described in Rule 4. Specifically,
the beginning of the interview is defined as the first video-
resolved frame with clear audio. At that point the counter
on the play-back recorder is set to 000.
Rule 2. Segment 2 begins with the first statement (as
defined in the Method section) which begins after 17 minutes
of interview time have elapsed. Specifically, the play
back recorder is run forward to 425. It is then played past
430, and the first statement beginning after that point
(430) is noted. The playback recorder is then rewound,
played forward, and recording started at the identified
statement. All machines used in this recording of master
tapes were identical SONY units. (Other machines would need
to have different counter numbers specified.)
Rule 3. Segment 3 begins with the first statement (as
defined in the Method section) which begins after 17 minutes
of interview time have elapsed. The same procedure is used
76

77
as in Rule 2, with counter settings 595, and 600. If inter
view ends prior to 3 minutes elapsed time, rewind 100 coun
ter units and use that number as the starting point. If the
interview was too long, use 650 or 700 as the starting point.
Rule 4. On all segments, stop recording at the first
sentence or clause end after 180 seconds or at 185 seconds.

APPENDIX F
TRAINING .MANUAL FOR UNITIZATION OF
TRANSCRIPTS BY PROPOSITION
Your role in this study is to assist in the prepara
tion of transcripts of interviews for scoring by other
raters like yourself. It is most important that your work
be done accurately, and it will require that you concentrate
well on it. This training manual will help you learn the
tasks, and generally, the tasks are specific and should not
be too difficult. Your task is to go through the transcript
and to use diagonal marks to separate phrases defined below
as PROPOSITIONS.
A PROPOSITION is defined as a sequence of words in the
transcript containing a subject and predicate, whether ex
pressed or implied, and including the modifiers of that sub
ject-predicate group. This is essentially the definition
of a clause, which may be either coordinating (or indepen
dent) or subordinating (or dependent). Any clause is
scored as a PROPOSITION with two additional stipulations.
Parenthetical clauses, clauses which come in the middle of
a larger clause, are not scored as separate PROPOSITIONS
but are considered PARENTHETICALs within a PROPOSITION.
Also, incomplete clauses, as, for example, when a person
changes the message mid-sentence, are scored as PROPOSITIONS.
The following suggestions will help in making these
scorings. In the Harbrace College Handbook, read the def
inition of conjunction, p. 446, and the discussion of con
junctions, p. 12-13. Conjunctions are, of course included
78

79
with the clause they introduce. Infinitives are not predi
cates. When auxiliary verbs (Harbrace, pp. 3-4) occur,
make sure the PROPOSITION includes the entire predicate.
Phrases which interrupt the predicate (e.g. "sort of" in
"he would sort of be nice") do not split up the PROPOSITION.
"You know" in its use as a sort of filling phrase, is not
scored as a PROPOSITION unless it occurs clearly outside of
any other PROPOSITION. Here are some examples:
1. /How much of this is a proposition?/
2. T: /How do you feel?/
C: /Bad./ I just really feel bad./
3. /He wants it this way,/and so do we./
4. /Whenever it happens, /we all know about it./
5. /This, that he wants, is going well./
Now note the subject-predicate in each PROPOSITION:
1. /this is/
2. T: /you do feel/
C: / (I feel) / I feel/
3. / he wants / we do (want) /
4. /it happens / we know /
5. / this is going /
Words in parentheses indicate they were understood, and
note that in #5 there is a parenthetical phrase that is not
scored as a PROPOSITION even though there is a subject-
predicate: "he wants."
Now mark the diagonals on this reproduction of part of
a transcript, and have that practice checked by the experi
menter before continuing past the practice.
T: I guess the reason I said that, Ray, was because you
seemed to express such a sense of freedom when she was
gone.

80
C: Well that's just how it affected me. Like suddenly I
felt like I was just free.
T: So she is after you pretty frequently.
C: Yep.
T: Kind of asking you to shape up.
C: Well, it just seems like a constant hassle trying to
account for places to go and things I do and making all
the normal trivial talk that really doesn't make any
difference. I'd just as soon skip anyway. It's a
bother.
If you have any questions, discuss them with the experi
menter when he checks your practice rating.

APPENDIX G
TRAINING MANUALS FOR
RATING ALL VARIABLES^
Your role in this study is to locate instances of cer
tain verbal behavior in typed transcripts of interviews.
It is critically important that your work be done accurately
and it will require that you concentrate well on it. Gen
erally, your rating task is quite specific and should not
be too technically difficult.
There are three definitions that will specify the
score you make in each situation. The first definition is
called the "unit," the second, the "category," and the
third, the "indicator" or set of rules for assigning a unit
to a specified category.
In all of the rating tasks in this study, several ideas
are important. One is the manner in which the pages of the
transcript are labeled. The largest division labeled is
the "interview segment" and this is the second number in
the label (the first number is an arbitrary grouping number)
The third number is the page number within a particular seg
ment. The last number is the page number within the entire
group of all transcripts. In other words, the page labeled:
1-04-2-008, is the second page of the fourth segment in the
first group of segments, and it is the eighth page of the
entire 288 pages of the transcript.
iNote: Each rater had a training manual with this intro
duction, the section corresponding to his or her task,
and one or more of the lists of words or phrases from
Appendices H, I, J, or K as specified.
81

82
Also, each time the speaker changes from the patient
to therapist or vice versa, this defines a "statement" and
statements are numbered down the left hand margin of the
page beginning with one at the beginning of each "segment."
Each "proposition" is set off by diagonal marks, and is num
bered starting with one (1) at the beginning of each
statement.

83
Training Manual: Primary Affect Word (PAW)
In scoring PAWs, the unit is the WORD, the category is
PAW, and the indicator (or rule) is that the word is found
on the attached list of PAW words. WORD is defined as the
basic blocks of letters constituting units within the tran
script, and is set off at its beginning and end by a space
or punctuation. (Note: if a typographical error results
in no space being left, it is considered that the words are
separate for scoring and counting purposes.) The category
is noted by entering a code number for the WORD scored, and
labeling the scoring page, PAWs. The code number is composed
of the page label plus the statement number containing the
word, the PROPOSITION number containing the WORD, and the
number of the WORD counting from the beginning of the
PROPOSITION.
The words on the list of PAW words are almost all in
the adjective form, but any part-of-speech form of the word
is to be scored. Generally these words are words which al
most always denotatively refer to affective, emotional, or
feeling experience. Examples are angry, sad, happy, etc.
(See list at this time.) Some of the words are called
"limited PAWs," and these are to be scored only when the
meaning of the word as used is equivalent to the meaning as
prescribed on the list. "Feeling" is marked with an
asterisk, and its code number should be marked with an
asterisk. It should not be coded at all, however, when it
has the meaning of "think" rather than "feel." Here are
some examples:

84
Ex. 1: a) He is angry. b) His anger is strong.
c) He spoke angrily, d) He angered them.
Ex. 2:
She is a moody person, who is really weird.
Ex. 3:
I'm tired and depressed.
Ex. 4 :
a) I was really hurt when she left.
b) I was really hurt in the accident.
Ex. 5:
a) I feel like I've been through a wringer.
b) I feel like he's a real dummy.
Notes:
Ex. 1:
all forms of "angry" are marked as PAWs: adjective,
noun, adverb, and verb.
Ex. 2:
"Moody," but not "weird" is found on the list of
PAWs.
Ex. 3:
"Depressed" but not "tired" is found on the list.
Ex. 4 :
"Hurt" is scored only when it has the limited mean
ing, emotionally injured and not physically injured.
Ex. 5:
In "a" "feel" denotes emotional experience, but in
"b" it denotes thinking.
Now try this reproduction of part of a transcript,
circling each WORD you find on the list of PAWs.
T: But you have hurt them.
C: Sure. Whenever you fight a lot you always hurt some
body, particularly a child.
T: You sound a little bit guilty.
C: I feel guilty because; it's not my kid's fault. She
has nothing to do with it.
If you have any questions, discuss them with the experi
menter when he checks your practice rating.

85
Training Manual: Secondary Affect Word (SAW)
In scoring SAWs, the unit is the WORD, the category is
SAW, and the indicator (or rule) is that the word is found
on the attached list of SAW words and that it occurs within
a PROPOSITION containing (explicitly or implicitly) a form
of the word "feel" or "sense," or within a PROPOSITION mod
ifying such a PROPOSITION. PROPOSITION was defined above.
WORD is defined as the basic blocks of letters constituting
units within the transcript, and is set off at its beginning
and end by a space or punctuation. (Note: if a typograph
ical error results in no space being left, it is considered
that the words are separate for scoring and counting pur
poses.) The category is noted by entering a code number for
the WORD scored, and labeling the scoring page, SAWs. The
code number is composed of the page label plus the state
ment number containing the WORD, and the number of the WORD
counting from the beginning of the PROPOSITION. One code
number is recorded for the "feel" or "sense" WORD, and a
second number, for the SAW WORD.
The words on the list of SAW words are almost all in
the adjective form, but any part-of-speech form of the word
is to be scored. Generally these SAW words are words which
denotatively refer to affective, emotional, or feeling ex
perience when used with the word "feel" (usually as a verb
or noun) or the word "sense." Examples are alert, helpless,
inhibited, etc. (See list at this time.)
Here are some examples:

86
Ex. 1:
T: How do you feel?
P: Inhibited.
Ex. 2 :
This is an impatient feeling, but then I'm an
impatient person.
Ex. 3:
I have a sense of uncertainty sitting here.
Notes:
Ex. 1:
"Inhibited" is circled to indicate it is a SAW
since it is on the list of SAWS and the patient
statement implies "I feel".
Ex. 2:
"Impatient" is only scored when it falls within a
PROPOSITION with "feel." Also, the second PROPO
SITION does not modify the first, which would have
been the case had it read: "This feeling is
strange, as I am very impatient."
Ex. 3:
"Uncertainty" is on the list and is in PROPOSITION
with "sense."
Now try this reproduction of part of a transcript,
circling each WORD you find on the list of SAWs which fits
the criteria.
T: What do you feel like today?
P: Really tired. I feel really depressed, and I wish I
weren't so insecure. I really feel insecure today,
especially today.
T: Today especially, you're feeling really depressed and
insecure. And really tired.
If you have any questions, discuss them with the experi
menter when he checks your practice rating.

87
Training Manual: Affective Idioms (AI)
In scoring AIs, the unit is the PROPOSITION, the cate
gory is AI, and the indicator (or rule) is that a group of
words occuring in the PROPOSITION is also found in a list
of AIs attached to this manual. PROPOSITION has been de
fined earlier. The category is noted by entering a code
number for the PROPOSITION scored, and labeling the scoring
page, AIs. The code number is composed of the page label
plus the statement number containing the phrase and the
PROPOSITION number.
Generally AIs are groups of words which together denote
affective, emotional, or feeling experience, but separately
do not. Examples are "sick and tired" and "peace of mind."
(See list at this time.) Flexibility must be given for the
exact arrangement of the words and variations in the parts
of speech used. In each case the test is whether the phrase
as used in the transcript denotes affective experience.
Here are some examples:
Ex.
1:
I have peace
of mind.
Ex.
2:
He must feel
screwed
up.
Ex.
3:
That really
gives me
the creeps.
Now try this reproduction of part of a transcript, cir
cling each phrase you find from the list of AIs.
P: Well, I just feel like there's something wrong. I'm
just sick and tired of living. That's really been
making me down for the last couple of months.

88
T: You're really feeling depressed, huh?
P: Yah. I just have no taste for anything. And now this
is really giving me the creeps; my not feeling like
living.
If you have any questions, discuss them with the experimen
ter when he checks your practice rating.

89
Training Manual: Self References (SR)
In scoring SRs, the unit is the PROPOSITION, the cate
gory is SR, and the indicator (or rule) is defined by four
rules discussed below. PROPOSITION has been defined ear
lier. The category is noted by entering a code number for
the PROPOSITION scored, indicating the rule number used,
and labeling the scoring page, SR. The code number is com
posed of the page label plus the statement number plus the
PROPOSITION number. Since several SRs may follow one an
other, the code number may be entered as a range as long as
only SRs are included. If this is done, each scoring rule
used at least once within the sequence is indicated.
Generally, a self reference (SR) is a verbal response
by a subject which expresses descriptively or emotively some
thing about the subject in relation to himself, others, or
the world. This indicator is more specifically defined by
the following four rules.1 Examples of each follow the rule
Rule 1: A self reference (SR) statement must begin with or
include a first person singular personal pronoun (e.g., I,
myself, mine) unless it is covered by one of the criteria
below.
Rule 2: First person plural pronouns (e.g., we, our, our
self, ours) are counted as self references when the group
referred to is intimately related to the subject (e.g.,
family, roommate, girlfriend).
Rule 3: Second person pronouns (e.g., you, your, yours) are
counted as self references when they are clearly used as
a substitute for a first person pronoun form and not as an
1Note: These four rules are taken directly from materials
received from Green and Marlatt (Note 1).

90
impersonal reference to a group. They can be checked, if
necessary, by comparing the sense of the statement when
substituting a first person pronoun with the sense of the
statement when substituting a group reference (e.g., they,
everybody, the world, etc.).
"You get a good feeling with these people." is counted.
"You can't solve anything that way" is not counted.
Rule 4: A response may be a self reference without explic
itly using a personal pronoun if it is closely related to
a previous self reference and is contingent upon it.
These responses can be checked by adding a self reference
phrase such as "to me" to the statement.
"I have met a lot of people here. It certainly has been
satisfying." (Both sentences are self references.)
Here are some more examples, with SRs underlined. Note
that each example illustrates one of the rules.
Ex. 1: /I like this. /This is mine./
Ex. 2: a) /My family is a lot of fun. /We do things to
gether /
b) /American citizens celebrated the bicentennial./
/It was very interesting to us all./
Ex. 3:
a) /You really like coming here./
b) /You do this,/ and it will make a lot of dif
ference /
Ex. 4: /I have really tried hard./ It has just about done
itT7
Ex. 5: /I like doing things like this/that are a lot of fun./
Now try this reproduction of part of a transcript, un
derlining any PROPOSITION that should be scored SR.
T: But you have hurt them.
C: Sure. Whenever you fight a lot you always hurt some
body, particularly a child.
T: You sound a little bit guilty.

91
C: I feel guilty because; it's not my kid's fault. She has
nothing to do with it.
If you have any questions, discuss them with the experimen
ter when he checks you practice rating.

92
Training Manual: Therapist Instruction Strategy (I)
In scoring Is, the unit is the PROPOSITION, the cate
gory is I with subcategories of 1-1, 1-2, and 1-3, and the
indicator or rule is specified by a set of rules defined
earlier. The category is noted by entering a code number
for the PROPOSITION scored, indicating the type of Instruc
tion (1-1, -2, or -3), and labeling the scoring page, I.
The code number is composed of the page label plus the state
ment number plus the PROPOSITION number.
1-1 has the general meaning: "Talk about your feel
ings." There are three rules for assigning a therapist
PROPOSITION (or PROPOSITION Group) to category 1-1. 1) The
PROPOSITION is in the imperative mood, or can be translated
into the imperative mood without significantly altering the
PROPOSITION'S meaning. 2) The PROPOSITION contains some
form of the word "feel," "emotion," "affect," or "sense."
3) The PROPOSITION (or group of PROPOSITIONS) has the gen
eral meaning: "Talk about your feelings."
1-2 has the general meaning: "Do not talk about sub
ject X (where X is a subject other than the general topic
of feelings)." The rules here are: (1) The PROPOSITION is
in the imperative mood or can be translated as above.
2) the PROPOSITION'S imperative is a negation of something.
3) The PROPOSITION does not contain any form of the words
listed under #2 of the indicator for 1-1. 4) The general
meaning is as at the beginning of this paragraph.
1-3 has the general meaning "Talk about subject X" or

93
"You might talk about subject X" where X is from one of the
three lists of words and phrases attached to this manual.
Rules for assigning to this category are: 1) The PROPOSI
TION is in the imperative or second person conditional mood.
2) The PROPOSITION contains some form of one or more words
from the attached list of words and phrases. 3) The PROP
OSITION contains as a predicate one or more of the follow
ing verbs: begin, report, talk, tell, say, speak, and think.
4) The general meaning is as above.
Here are some examples. Note that 1, 2, and 3 would
be assigned to those respective categories, and that 4 and
5 would
not be assigned to any
category.
Ex.
1:
Tell me how you feel.
Ex.
2:
Don't talk about your
relatives.
Ex.
3:
You might talk about anger or sadness.
Talk about anger or sadness.
Ex.
4:
How were you feeling?
Ex.
5:
Don't talk about your
feelings.
Now try this part of a transcript, putting the category num
ber to the left of any statements containing scored PROPOSI
TIONS. Remember that more than one PROPOSITION may need to
be grouped to assign it to a category.
T: /How are you feeling?/
C: /Okay. /I'm really worried about this thing./
T: /Talk about your feelings about it./
C: /I have to go take a test, /and if I don't pass it/
/I'll have to take it over./

94
T: /Don't just tell me/what is happening; /how do you feel
about it?/
C: /What do you mean?/
T: /Well, you might talk about the feeling you have when
you go to take the test /or when you fail,/like feeling
anxious, upset, frustrated./
Discuss any questions you have with the experimenter when
he checks your practice rating.

95
Training Manual: Therapist Reinforcement (R)
In scoring Rs, the unit is the WORD with certain stip
ulations, the category is R, and the indicator (or rule) is
that the WORD or WORDS are found on the attached list of
therapist reinforcers. WORD is defined as the basic block
of letters constituting units within the transcript and is
set off at its beginning and end by a space or punctuation.
The stipulations are that there may be one or two WORDS in
the unit and that they must occur at the beginning of a
therapist statement as defined earlier. The category is in
dicated by entering a code number for the WORD or WORD pair
scored, and labeling the scoring page R. The code number
is composed of the page label plus the statement number.
Generally, the words on the list are brief expressions
of approval, agreement, or attention by the therapist. Spe
cifically, the indicator is that the unit is included on
the list attached. Examples are "good," "mm-hmm," or "yes."
(Read over the list at this time.)
Here are some examples:
Ex.
1:
I see.
You really think that's the way it is
Ex.
2:
Mm-hmm.
I see.
Ex.
3:
That's
terrific.
Now try this reproduction of part of a transcript, cir
cling each R WORD or WORDS.
C: I feel great.
T: Good.

96
C: I like football. I feel great when I'm playing.
T: Mm-hmm.
C: I play every day.
T: Oh?
If you have questions, discuss them with the experimenter,
when he checks your practice rating.

APPENDIX H
LISTING OF PRIMARY AFFECT WORDS
AND LIMITED PRIMARY AFFECT WORDS
Primary Affect Words
ABHORANT
DREADING
AFFECTIONATE
AFRAID
EAGER
AGHAST
EARNEST
AGITATED
ECSTATIC
AMAZED
ELATED
ANGRY
EMBARRASSED
ANGUISHED
EMOTIONAL
ANXIOUS
ENJOY
ANNOYED
ENRAGED
APPALLED
ENRAPTURED
APPREHENSIVE
ENTHUSED
AWED
ENVIOUS
EXCITED
BASHFUL
EXHILARATED
BITTER
EXHUBERANT
BLUSHED
BORED
FANATICAL
BOTHERED
FEARFUL
FEELING
CALM
FELICITOUS
CHEERFUL
FERVENT
CHERISHED
FIDGETY
COMPOSED
FLUSTERED
CONCERNED
FRETFUL
CONFIDENT
FRIGHTENED
CONFUSED
FRUSTRATED
CONSTERNATION
FURIOUS
CONTENTED
CRUEL
GAY
CURIOUS
GLAD
GLEEFUL
DEJECTED
GLOOMY
DELIGHTED
GROUCHY
DEPRESSED
GUSTO
DESPERATE
DESPONDENT
HAPPY
DETEST
HATEFUL
DISAPPOINTED
HORRIFIED
DISGUSTED
HOSTILE
DISCONTENTED
HYSTERICAL
DISCOURAGED
DISENCHANTMENT
IMPETUOUS
DISILLUSIONED
IMPRESSED
DISMAYED
INCENSED
DISPLEASED
INFURIATED
DISSATISFIED
INSPIRED
INTIMIDATED
SAD
IREFUL
SATISFIED
IRRITABLE
SCARED
IRRITATED
SEDATE
SENSUOUS
JEALOUS
SENTIMENTAL
JITTERY
SERENE
JOLLY
SHY
JOYFUL
SINCERE
JOYOUS
SORROWFUL
SQUEAMISH
LIGHTHEARTED
STARTLING
LONELY
SUFFERING
SULLEN
MAD
SURPRISED
MALEVOLENT
MEEK
SYMPATHETIC
MISERABLE
TENSE
MISGIVING
TERRIFIED
MISTRUST
THRILLED
MOODY
TIMID
TRANQUIL
NERVOUS
OPTIMISTIC
TREPIDATION
TRUSTING
OVERWHELMED
UNEASY
UNHAPPY
PANICKY
UNNERVED
PASSIONATE
UNPERTURBED
PEACEFUL
UNRUFFLED
PEPPY
PERTURBED
UPSET
PESSIMISTIC
VEHEMENT
PIQUED
PITY
VIGOR
PLACID
WISTFUL
PLEASANT
PROUD
WORRIED
ZEALOUS
RAPT
RAPTUROUS
RELAXED
REPOSED
REPULSED
RESENTFUL
RESTFUL
RESTLESS
ZESTY
97

Limited Primary Affect Words
ATTRACTED = LIKE
BLUE = SAD
BREATHLESS = THRILLED
BUGGED = BOTHERED
CENTERED = EMOTIONALLY SETTLED
COOL = CALM
CROSS = SULKY
DIG = LIKE
DISLIKE = EMOTIONALLY UNATTRACTED
DOWN = DEPRESSED
ELECTRIFYING = STIMULATING
FLUSHED = BLUSHED
HEARTY = VIGOROUS
HIGH = EUPHORIC
HURT = EMOTIONALLY INJURED
IMPRESSED = AFFECTED
LIKE = ATTRACTED TO
MOVED = EMOTIONALLY AFFECTED
RAVING = ANGRY
RELISHED = ENJOYED
REPELLING = REPULSING
SEETHING = ANGRY
SHOCKING = AFFECTING
SMARTING = HURTING EMOTIONALLY
SPEECHLESS = SHOCKED
STEWED = IRED
TEMPER = ANGER
TENDER = AFFECTIONATE
TIRED = BORED
TOGETHER = EMOTIONALLY SETTLED
TOUCHED = AFFECTED
WARM = EMOTIONAL
WINCED = EMOTIONALLY HURT

APPENDIX I
LISTING OF SECONDARY AFFECT WORDS
ACTIVE
ADVENTUROUS
AGGRESSIVE
AGREEABLE
ALERT
AMBITIOUS
APPRECIATIVE
AWFUL
BAD
CLOSE
COMFORTABLE
CONSIDERATE
CRAPPY
CRUDDY
DEFENSIVE
DESIRE
DOUBTFUL
DULL
EASY
ENERGETIC
FREE
FRIENDLY
GOOD
GRATEFUL
GRATIFIED
GREAT
GUILTY
HELL
HELPLESS
HOPELESS
IMPATIENT
INDIFFERENT
INFLAMED
INHIBITED
INSECURE
INTENSE
KIND
LEISURELY
LIVELY
LOATHE
LOST
LOVING
MEAN
NEGATIVE
NUMB
PLODDING
POSITIVE
PROVOKED
QUARRELSOME
REBELLIOUS
RO1ANTIC
ROTTEN
SAVAGE
SECURE
SETTLED
SHAKY
SHITTY
SLEEPY
SLOW
SLUGGISH
SOLEMN
STEADY
THOUGHTFUL
THREATENED
TIRED
TOGETHER
TOLERANT
TROUBLED
UNCERTAIN
UNCONCERNED
UNDISTURBED
UNSURE
USELESS
WEARY
WEIGHTY
WEIRD
WITHDRAWN
WONDERMENT
WORTHLESS
99

APPENDIX J
LISTING OF AFFECTIVE
IDIOMS AND SLANG
ABIDE
BANG
BESIDE
BLOW
BOTTOM
BURNS
CARE
CARE
CHARGE
CONSUMED
CREEPS
CREEPS
EATS
EATS
EDGE
ENDURE
ENTER
FED
FORWARD
FREAKED
FUCKED
HAIR
HEAD
HEART
HEAVY
KICKS
LOOK
LUMP
PEACE
PINS
PISSED
SICK
SCREWED
SEIZED
SOUL
STAND
TAKE
TAKE
TASTE
TEAR
THINK
TICKED
TIRED
TURN
TURNS
VEGETATES
WELL-BEING
WRONG
(can't abide someone or something)
(get a bang out of)
(be beside oneself)
(blow one's mind)
(from the bottom of one's heart)
(burns one up = angers)
(care for someone or something)
(care nothing for someone or something)
(get a charge out of)
(consumed with something = zealous)
(flesh creeps)
(gives me the creeps)
(eats away at one)
(eats one's heart out)
(be on edge)
(can't endure)
(enter the spirit of)
(fed up)
(looking forward to)
(freaked out)
(fucked up)
(hair stands on end)
(head over heels)
(heart and soul)
(heavy experience)
(do it for kicks)
(look up to)
(lump in the throat)
(peace of mind)
(on pins and needles)
(pissed off, be pissed)
(sick and tired of)
(screwed up)
(seized with emotion)
(soul stirring)
(can't stand)
(can't take it anymore)
(take to someone = attracted)
(have a taste for; or not for)
(tears one up; all torn up)
(think much of someone; or little of)
(ticked off = mad)
(tired of living)
(person turns color; turn green, black, pale)
(turns one on)
(vegetates one; vegetates one out)
(sense of well-being)
(feel something's wrong)
100

APPENDIX K
LISTING OF THERAPIST VERBAL REINFORCERS
MM-HMM*
GOOD
WONDERFUL
YEAH
I SEE
YES
TRUE
THAT'S TRUE
AH-HAH*
YEP
RIGHT
UH-HUH*
OKAY (OR 0. K. )
*Slight variations in these (e.g., Um-huh), should also be
scored.
101

APPENDIX L

CLIENT AFFECTIVE VERBALIZATION
103
6.0 -
5.5 -
5.0 -
4.5 -
4.0 -
3.5 -
3.0 -
2.5 -
Legend: HM
HF
LM
LF
12 3
SEGMENT
High compatible dyad males.
High compatible dyad females.
Low compatible dyad males.
Low compatible dyad females.
INTERACTION GRAPH:
COMPATIBILITY X CLIENT SEX X SEGMENT
FIGURE L-l

CLIENT AFFECTIVE SELF REFERENCE
104
6.0 -
5.5 -
5.0 -
4.5 -
4.0 -
3.5 -
3.0 -
2.5 -
Legend: HM
HF
LM
LF
12 3
SEGMENT
High compatible dyad males.
High compatible dyad females.
Low compatible dyad males.
Low compatible dyad females.
INTERACTION GRAPH:
COMPATIBILITY X CLIENT SEX X SEGMENT
FIGURE L-2

APPENDIX M

106
16 -
14 -
12 uuu
uuu
10 -
tttuuu
8 -
ttt ttt ttt
6 -
4 -
uuu ttt
2 ttt
uuu uuu ttt
uuu ttt
0 -
uuu
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
(1.0) (2.1) (3.1) (4.2) (5.2) (6.3) (7.4) (8.4) (9.5)
CLIENT SCORE INTERVALS3
a Intervals for transformed scores are in parentheses,
u Indicates untransformed scores,
t Indicates transformed scores.
DISTRIBUTION OF SCORES FOR CLIENT AFFECTIVE
VERBALIZATION IN SEGMENT 1
FIGURE M1

107
16 uuu
14 -
12 -
uuu
ttt ttt
10 -
ttt
8 -
6 -
4 uuu
2 -
ttt
uuu
uuu
ttt
0 ttt uuu uuutttuuuttt
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
(1.0) (2.1) (3.1) (4.2) (5.2) (6.3) (7.4) (8.4) (9.5)
CLIENT SCORE INTERVALS3
a Intervals for transformed scores are in parentheses,
u Indicates untransformed scores,
t Indicates transformed scores.
DISTRIBUTION OF SCORES FOR CLIENT AFFECTIVE
VERBALIZATION TOTAL SCORE
FIGURE M-2

REFERENCE NOTES
1. Green, A. H., & Marlatt, G. A. Scoring criteria for
feeling and content responses. Unpublished re
search materials, 1972. (Available from G. Alan
Marlatt, Department of Psychology, University of
Washington, Seattle, Washington, 98195.)
2. Goldman, J. R. Therapist-client interpersonal need
compatibility and interview process and outcome.
Study in progress, 1976.
108

REFERENCES
Alexander, F. In H. D. Kruse (Ed.)/ Integrating the
Approaches to Mental Disease. New York: Hoeher-
Hayser, 1957.
Auld, F. & Murray, E. J. Content analysis studies of
psychotherapy. Psychological Bulletin, 1955, 52,
377-395.
Barnabei, F., Cornier, W. H., & Hye, L. S. Determining
the effects of three counselor verbal responses on
client verbal behavior. Journal of Counseling Psy
chology, 1974, 21, 355-359.
Bateson, G., Jackson, D., Haley, J., & Weakland, J. Toward
a theory of schizophrenia. Behavioral Science, 1956,
1, 251-264.
Berelson, B. Content Analysis in Communications Research.
Glencoe, Ill.: Free Press, 1952.
Bergin, A. E. & Grafield, S. L. Handbook of Psychotherapy
and Behavior Change: an Empirical Analysis. New York
John Wiley and Sons, 1974.
Blumberg, R. W. A scale for problems in living research.
Psychological Reports, 1968, 22_, 161-162.
Blumberg, R. W. Client dogmatism, therapist leadership,
and the psychotherapeutic relationship (Doctoral dis
sertation, University of Florida, 1969). Dissertation
Abstracts International, 1970, 31^, 389B. (University
Microfilms No. 76-12,227.)
Bordin, E. S. Research Strategies in Psychotherapy. New
York: Wiley and Sons, 1974.
Braaten, L. J. The movement from non-self to self in cli
ent-centered psychotherapy. Journal of Counseling
Psychology, 1961, 13, 20-24 .
Buck, L. A. & Cuddy, J. M. A theory of communication in
psychotherapy. Psychotherapy: Theory, Research, and
Practice, 1966, 3, 7-13.
109

110
Buehler, R. E. & Richmond, J. F. Interpersonal communi
cation theory. Corrective Psychiatry Journal of
Social Therapy, 1965, 1_1, 204-216 .
Butler, J. M., Rice, L. N. & Wagstaff, A. K. On the def
inition of variables: an analogue of clinical
analysis. In H. H. Strupp and L. Luborsky (Eds.),
Research in Psychotherapy (Vol. 2). Washington D. C.:
American Psychological Association, 1962.
Campbell, D. T. & Stanley, J. C. Experimental and Quasi-
Experimental Designs for Research. Chicago: Rand
McNally and Co., 1963.
Cartwright, D. S. Analysis of qualitative material. In
L. Festinger and D. Katz (Eds.), Research Methods in
the Behavioral Sciences. New York: Holt, Rinehart,
and Winston, 1966.
Crowley, T. The conditionability of positive and negative
self-reference emotional affect statements in a coun
seling type interview (Doctoral dissertation, Univer
sity of Massachusetts, 1970). Dissertation Abstracts
International, 1970, 31, 2100A-2101A. (University
Microfilms No. 70-23,026.)
Davis, J. D. The Interview as Arena: Strategies in
Standardized Interviews and Psychotherapy. Stanford,
Calf.: Stanford University Press, 1971.
Davis, J. D., Fisher, L., & Davis, M. Therapeutic inter
ventions of clinicians in standardized interviews: a
test of the social model. Journal of Consulting and
Clinical Psychology, 1973, LI, 16-26.
Davis, J. D., Fisher, L., & Davis, M. L. Stability and
change in strategies of psychiatric patients and nor
mal interviewees. Journal of Consulting and Clinical
Psychology, 1975, £3, 205-215.
Ellsworth, S. G. The consistency of counselor feeling-ver
balization. Journal of Counseling Psychology, 1963,
10, 356-361.
Frank, G. H. On the History of the objective investigation
of the process of psychotherapy. Journal of Psychology,
1961, 15_, 89-95.
Gassner, S. M. Relationship between patient-therapist com
patibility and treatment effectiveness. Journal of
Consulting and Clinical Psychology, 1970, ~34, 408-4~14.

Ill
Gottschalk, L. A. & Gleser, G. C. The Measurement of Psy
chological States Through the Content Analysis of Ver
bal Behavior. Berkeley: University of California
Press, 1969.
Gottschalk, L. A., Winget, C. N., & Gleser, G. C. Manual of
Instruction for Using the Gottschalk-Gleser Content
Analysis Scales. Berkeley: University of California
Press, 1969.
Green, A. H. & Marlatt, A. Effects of instructions and
modeling upon affective and descriptive verbalization.
Journal of Abnormal Psychology, 1972, 8JD, 189-196.
Greenhill, M. H. The focal communication concept. American
Journal of Psychotherapy, 1958, 1_2, 30-41.
Guetzkow, A. Unitizing and categorizing problems in coding
qualitative data. Journal of Clinical Psychology,
1950, 6, 47-58.
Haley, J. Strategies of Psychotherapy. New York: Grue
and Stratton, 1963.
Haley, J. Research on family patterns: an instrument
measure. Family Process, 1964 3^, 41-51.
Harper, R. A. Psychoanalysis and Psychotherapy: 36 systems.
Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1959.
Hekmat, H. Reinforcing values of interpretation and reflec
tions in a quasi-therapeutic interview. Journal of
Abnormal Psychology, 1971a, 77, 25-31.
Hekmat, H. Extraversin, neuroticism, and verbal condition
ing of affective self disclosures. Journal of Counsel
ing Psychology, 1971b, 1_8, 64-69.
Hekmat, H. Three techniques of reinforcement modification:
a comparison. Behavior Therapy, 1974, 5^, 541-548.
Hekmat, H. & Lee, Y. B. Conditioning of affective self-ref
erences as a function of semantic meaning of verbal
reinforcers. Journal of Abnormal Psychology, 1970, 76,
427-433.
Hekmat, H. & Thiess, M. Self actualization and modification
of affective self disclosures during a social condition
ing interview. Journal of Counseling Psychology, 1971,
18, 101-105.
Heller, K. Experimental analogues of psychotherapy: the
clinical relevance of laboratory findings of social
influence. Journal of Nervous and Mental Disease, 1963,
137, 420-426^ ~

112
Hoch, P. A. & Zubin, J. (Eds.). Psychopathology of Commu
nication New York: Grue and Stratton, 1958.
Hodges, J. C. & Whitten, M. E. Harbrace College Handbook
(6th ed.). New York: Harcourt, Brace & World, Inc.,
1967.
Hoffnung, R. J. Conditioning and transfer of affective self
references in a role-played counseling interview.
Journal of Consulting Clinical Psychology, 1969, 33,
527-531.
Isaacs, K. S. & Haggard, E. A. Some methods used in the
study of affect in psychotherapy. In L. A. Gottschalk
& A. H. Auerbach (Eds.), Methods of Research in Psycho
therapy New York: Appleton-Century-Crofts, 1966.
Jackson, D. D., Riskin, J. & Satir, V. A method of analysis
of a family interview. Archives of General Psychiatry,
1961, 5, 321-339.
Jones, E. E. & Thibaut, J. W. Interaction goals as bases
of inference in interpersonal perception. In R.
Tagiuri and L. Petrullo (Eds.), Person Perception and
Interpersonal Behavior. Stanford: Stanford Univiver-
sity Press, 1958.
Kiesler, D. J. Experimental designs in psychotherapy re
search. In A. E. Bergin and S. L. Garfield (Eds.),
Handbook of Psychotherapy and Behavior Change: An
Empirical Analysis. New York: John Wiley and Sons,
Inc., 1971.
Kiesler, D. J. The Process of Psychotherapy. Chicago:
Aldine, 1973.
Kirk, R. E. Experimental Designs: Procedures for the Be
havioral Sciences. Belmont, Calf.": Brooks/Cole Pub-
lishing Company, 1968.
Lennard, H. L. & Bernstein, A. The Anatomy of Psychotherapy:
Systems of Communication and Expectation. New York:
Columbia University Press, 1960.
Lennard, H. L. & Bernstein, A. Pattern in Human Interaction.
San Francisco: Jossey-Bass, 1969.
Levy, W. H. The communication of affect in psychotherapy.
Archives of General Psychiatry, 1967, 1£, 102-106.
Lukens, H. C. The emission of affective verbal behavior
as function of two modes of experimenter intervention:
a partial replication and extension. Dissertation
Abstracts International, 1970, 31, 398B^ (University
Microfilms No. 70-12,392.)

113
Markel, N. N. Psycholinguistics: an Introduction to the
Study of Speech and Personality. Homewood, Ill.:
Dorsey Press, 1969.
Marlatt, H. G. Exposure to a model and task ambiguity as
determinants of verbal behavior in an interview.
Journal of Consulting and Clinical Psychology, 1971,
36., 268-276 .
Marsden, G. Content-analysis studies of psychotherapy:
1954 through 1968. In A. E. Bergin & S. L. Garfield
(Eds.), Handbook of Psychotherapy and Behavior Change:
An Empirical Analysis. New York: John Wiley & Sons,
Inc., 1971.
Matarazzo, J. D., Saslow, G., & Matarazzo, R. G. The inter
action chronograph as an instrument for objective meas
urement of interaction patterns during interviews.
Journal of Psychology, 1956 41_, 347-367.
Meltzoff, J. & Kornreich, M. Research in Psychotherapy.
New York: Atherton, 1970.
Mendelsohn, G. A. & Rankin, N. 0. Client-counselor compati
bility and the outcome of counseling. Journal of Ab
normal Psychology, 74, 1969.
Merbaum, M. & Lukens, H. C. Effects of instructions, elici
tations, and reinforcements in the manipulation of af
fective verbal behavior. Journal of Abnormal Psychol
ogy, 1968, 73, 376-380.
Moos, R. H. & Ciernes, S. R. Multivariate study of the
patient-therapist system. Journal of Consulting
Psychology, 1967, 31, 119-130.
Morris, C. Signs, Language, and Behavior. Englewood Cliffs,
N. J.: Prentice Hall, 1946.
Myrick, R. D. Effect of a model on verbal behavior in
counseling. Journal of Counseling Psychology, 1969,
16, 185-190.
Riess, B. F. Communication in psychotherapy. American
Journal of Psychotherapy, 1957, .LI, 474-489 .
Rioch, D. & Weinstein, E. A. (Eds.). Disorders of Communi
cation Baltimore: Williams and Wildins, 1964.
Ruesch, J. Therapeutic Communication. New York: Norton,
1961.
Salzinger, K. & Pisoni, S. Reinforcement of affect responses
of schizophrenics during the clinical interview.
Journal of Abnormal and Social Psychology, 1958, 57,

114
Salzinger, K. & Pisoni, S. Reinforcement of verbal affect
responses of normal subjects during the interview.
Journal of Abnormal and Social Psychology, 1960, 60,
127-130.
Sapolsky, A. Effects of interpersonal relationships on ver
bal conditioning. Journal of Abnormal and Social Psy
chology;, 1960, 60_, 241-246 .
Sapolsky, A. Relationship between patient-doctor compatibil
ity, mutual perception, and outcome of treatment.
Journal of Abnormal Psychology, 1965, 7_0, 70-76.
Scheflen, A. E. Communication systems such as psychotherapy.
In J. H. Masserman (Ed.), Current Psychiatric Therapies.
New York: Grue and Stratton, 1965.
Schtz, W. C. The Interpersonal Underworld. Palo Alto:
Science and Behavior Books, Inc., 1966.
Strupp, H. H. Patient-doctor relationships: the psychothera
pist in the therapeutic process. In A. J. Bachrach
(Ed.), Experimental Foundations of Clinical Psychology.
New York: Basic Books, 1962.
Strupp, H. H. Psychotherapy: Clinical Research and Theo
retical Issues. New York: Jason Aronson, 1973.
Sullivan, H. S. The Interpersonal Theory of Psychiatry,
New York: Norton, 1953.
Sullivan, H. S. The Psychiatric Interview. New York:
Norton, 1954.
Trager, G. L. Language and psychotherapy. In L. A.
Gottschalk and A. H. Auerbach (Eds.), Methods of Re
search in Psychotherapy. New York: Appleton-Century-
Crofts, 1966.
Wadington, J. Social decision schemes and two-person bar
gaining. Behavioral Science, 1975, 20^ 157-165.
Waskow, I. E. Reinforcement in a therapy-like situation
through selective responding to feelings or content.
Journal of Consulting Psychology, 1962, 26^, 11-19.
Watzlawick, P., Beavin, J. H. & Jackson, D. D. Pragmatics
of Human Communication: a Study of Interactional
Patterns, Pathologies, and Paradoxes. New York:
Norton, 1967.

115
Whalen, C. Effects of a model and instructions on group
verbal behaviors. Journal of Consulting and Clini
cal Psychology, 1969, 3_3, 509-521.
Williamson, E. G. The meaning of communication in coun
seling. Personnel and Guidance Journal, 1959, 88,
6-14.

BIOGRAPHICAL SKETCH
Stanley E. Jones was born August 21, 1948, at Big
Stone Gap, Virginia. In June, 1966, he was graduated from
Dobyns-Bennett High School in Kingsport, Tennessee. He
received the Bachelor of Arts degree in psychology from
Tennessee Wesleyan College in June, 1970. In September,
1970, he entered the Graduate School of the University of
Florida in the clinical psychology program, completing the
Master of Arts degree in June of 1972, and is presently
working toward the degree of Doctor of Philosophy. Since
September, 1975, he has worked as a psychologist for the
North Central Florida Community Mental Health Center.
Stanley E. Jones is married to the former Barbara-jeanne
Muceia.
116

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
lfacquelin R. Goldman, Chairperson
Professor of Clinical Psychology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Util iy *J>- /\ cl- A/--*-
Audrey 6. Schumacher
Professor Emeritus of Psychology
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Marvin E. Shaw
Professor of Psychology

I certify that I have read this study and that in my
opinion it conforms to acceptable standards of scholarly
presentation and is fully adequate, in scope and quality, as
a dissertation for the degree of Doctor of Philosophy.
Hannelore L. Waas, Professor of
Foundations of Education
This dissertation was submitted to the Graduate Faculty of
the Department of Psychology in the College of Arts and
Sciences and to the Graduate Council, and was accepted as
partial fulfillment of the requirements for the degree of
Doctor of Philosophy.
December, 1976
Dean, Graduate School



70
the specific purpose of measuring the use of such language
by clients and therapists.
The second area of contribution is in applying the
above methodology to the examination of a group of variables
in a sample of analogue psychotherapy interviews. The vari
ables considered had been found in previous research to have
certain relationships to each other, but these relationships
were generally absent in this study. Two significant re
lationships were demonstrated. Compatibility was found to
be effective under some conditions for female clients, but
not for male clients. Client affective verbalization in
the second segment and for the total interview were found
to be significantly correlated with therapist evaluation,
which provided modest support for the validity of the client
affective measure as an indicator of some important processes
in the interview. Therapist strategy variables were either
found not to be utilized by therapists in the present sam
ple, or to result in inhibitory effects, with clients more in
apparent control of the frequency of affective verbalization
in the session.
Suggestions for future research are made, and argue for
the use of more control than in the present study. Caution
is also made, however, for such strict control that the re
search looses its relevance to applied practice.


35
the therapist affective verbalization score as computed
above. The final measure for the Reinforcement strategy is
most easily expressed as a formula which recognizes that
Reinforcement might either encourage or discourage affec
tive verbalizations:
Rij = t(rw rQ)/(rw + r0)](x/8)(100),
where Rij is the Reinforcement variable score for client i
in segment j; rw is the number of times the therapist used
a verbal Reinforcer immediately after a client finished a
statement which contained an affective verbalization; rQ is
the number of times the therapist did not Reinforce an af
fective verbalization; and x is a value equal to the larger
of the two r's (rw or rQ). (The x/8 factor attempts to cor
rect the Reinforcement score for small numbers of reinfor
cers, which is an inference from Waskow's 1962 study.) The
total Reinforcement score for the interview is the mean of
the three Segment scores.


APPENDIX M


TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS iii
LIST OF TABLES vi
ABSTRACT vii
CHAPTER
1 INTRODUCTION 1
Communication as Relationship
Definition 5
Therapist Goals as a Definition 6
2 DEFINING THE RESEARCH QUESTION 9
Communication Science: An Overview ... 13
Defining Communication of Affect 16
Research Hypotheses 22
3 TECHNICAL CONSIDERATIONS 24
Method 2 4
Rating Procedures 27
4 RESULTS 36
Client Verbalization and Therapist
Strategies 36
Compatibility and Client Sex 54
Client Verbalization and Session Outcome. 61
Exploratory Analyses 62
5 DISCUSSION 65
Conclusion 69
APPENDICES 71
APPENDIX A--Therapist Post-Session
Evaluation Form 72
APPENDIX B--Client Post-Session Evaluation
Form 73
IV


6
that of therapist and client, can be seen as involving a pro
cess of attempting to reach a mutual definition of what be
haviors will and will not occur in the relationship, and a
process of determining who will control the definition.
Control of the definition is an important issue, and can
be seen as a competition between the two participants in a
dyad. Each is engaging in purposive behavior directed toward
obtaining some desired and satisfying state of affairs, and
rarely do these purposes coincide so exactly as to allow an
entirely cooperative venture. Since the definition of the
relationship is essentially a statement about what behaviors
are to be allowed in the interaction, the person who con
trols that definition is influencing the interaction and
therefore the other person, and has an advantageous posi
tion. The person not in control is expected to attempt to
take control by redefining the relationship in a more satis
factory way for him.
Therapist Goals as a Definition
Most of an effective therapist's "needs" in a session
are directed toward therapeutic goals, and are essentially
the needs the therapist is attempting to satisfy in his ef
forts to control the definition of the relationship. One
way in which therapists frequently want to influence the in
teraction is by focusing the client into a particular style
of interaction. Many theories of therapy emphasize the im
portance of styles of interaction, as for example in seeing


TABLE 2
ANOVA-2 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX X SEGMENT
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
BETWEEN SUBJECTS
196.345
39
2
COMPATIBILITY
5.453
1
5.453
1.155
NS
3
CLIENT SEX
.691
1
.691
.146
NS
4
COMPATIBILITY X CLIENT SEX
20.263
1
20.263
4.293
*
5
SUBJECTS WITHIN GROUPS
169.938
36
4.720
6
WITHIN SUBJECTS
223.595
80
7
SEGMENT
8.932
2
4.466
1.596
NS
8
COMPATIBILITY X SEGMENT
1.308
2
. 654
.234
NS
9
CLIENT SEX X SEGMENT
8.730
2
4.365
1.560
NS
10
COMPATIBILITY X CLIENT SEX X SEGMENT
3.167
2
1.583
.566
NS
11
SEGMENT X SUBJECTS WITHIN GROUPS
201.455
72
2.798
12
TOTAL
419.940
119
Tests
for Simple
Compatibility and
Client Sex Effects after
Significant Interaction
1
COMPATIBILITY
@ C=1 (MALE)
2.315
1
2.315
. 490
NS
2
COMPATIBILITY
@ C=0 (FEMALE)
23.401
1
23.401
4.958

3
CLIENT SEX @
COMPATIBILITY =1
(HIGH)
14.337
1
14.337
3.037
AP
4
CLIENT SEX @
COMPATIBILITY =0
(LOW)
6.617
1
6.617
1.402
NS
Note: Dependent variable is Client Affective Self Reference.
* P < 05 NS £>.10 AP £<.10
O


114
Salzinger, K. & Pisoni, S. Reinforcement of verbal affect
responses of normal subjects during the interview.
Journal of Abnormal and Social Psychology, 1960, 60,
127-130.
Sapolsky, A. Effects of interpersonal relationships on ver
bal conditioning. Journal of Abnormal and Social Psy
chology;, 1960, 60_, 241-246 .
Sapolsky, A. Relationship between patient-doctor compatibil
ity, mutual perception, and outcome of treatment.
Journal of Abnormal Psychology, 1965, 7_0, 70-76.
Scheflen, A. E. Communication systems such as psychotherapy.
In J. H. Masserman (Ed.), Current Psychiatric Therapies.
New York: Grue and Stratton, 1965.
Schtz, W. C. The Interpersonal Underworld. Palo Alto:
Science and Behavior Books, Inc., 1966.
Strupp, H. H. Patient-doctor relationships: the psychothera
pist in the therapeutic process. In A. J. Bachrach
(Ed.), Experimental Foundations of Clinical Psychology.
New York: Basic Books, 1962.
Strupp, H. H. Psychotherapy: Clinical Research and Theo
retical Issues. New York: Jason Aronson, 1973.
Sullivan, H. S. The Interpersonal Theory of Psychiatry,
New York: Norton, 1953.
Sullivan, H. S. The Psychiatric Interview. New York:
Norton, 1954.
Trager, G. L. Language and psychotherapy. In L. A.
Gottschalk and A. H. Auerbach (Eds.), Methods of Re
search in Psychotherapy. New York: Appleton-Century-
Crofts, 1966.
Wadington, J. Social decision schemes and two-person bar
gaining. Behavioral Science, 1975, 20^ 157-165.
Waskow, I. E. Reinforcement in a therapy-like situation
through selective responding to feelings or content.
Journal of Consulting Psychology, 1962, 26^, 11-19.
Watzlawick, P., Beavin, J. H. & Jackson, D. D. Pragmatics
of Human Communication: a Study of Interactional
Patterns, Pathologies, and Paradoxes. New York:
Norton, 1967.


TABLE 1
ANOVA-1 SUMMARY TABLE
COMPATIBILITY X CLIENT SEX X SEGMENT
SOURCE
SUM OF
SQUARES
df
MEAN
SQUARE
F
RATIO
P
1
BETWEEN SUBJECTS
168.572
39
2
COMPATIBILITY
4.798
1
4.798
1.133
NS
3
CLIENT SEX
.901
1
.901
.213
NS
4
COMPATIBILITY X CLIENT SEX
10.365
1
10.365
2.447
NS
5
SUBJECTS WITHIN GROUPS
152.508
36
4.236
6
WITHIN SUBJECTS
271.001
80
7
SEGMENT
3.566
2
1.783
. 513
NS
8
COMPATIBILITY X SEGMENT
4.915
2
2.457
. 707
NS
9
CLIENT SEX X SEGMENT
10.417
2
5.208
1.499
NS
10
COMPATIBILITY X CLIENT SEX X SEGMENT
1.874
2
.937
.270
NS
11
SEGMENT X SUBJECTS WITHIN GROUPS
250.229
72
3.475
12
TOTAL
439.573
119
Note: Dependent variable is Client Affective Verbalization.
NS p> .05
OJ
V£>


3
a reproduction does not get us any closer to understanding
the process of communication, however. A representation
apart from a reproduction is required, and this implies the
operation of a theory providing a model for the representa
tion and an artificial limitation of the complexity allowed
in the representation. Numerous researchers recognize spe
cifically the need for limiting the complexity of the process
studied (Auld & Murray, 1955, Cartwright, 1966; Guetzkow,
1950; Kiesler, 1973; Marsden, 1971; and Strupp, 1962) and
the need for being specific about the theory used. The in
fluence of the theory on our selection of the particular phe
nomena studied cannot be over-emphasized, since the operation
of a theoretical perspective in the selection of a represen
tation is the primary determinant of the meaningfulness of
the research no matter how sophisticated other aspects of
the research may be.
Researching communication, then, involves the selection
of a theoretical perspective which will dictate the aspects
of communication to be described, but we cannot expect the
application of a description system to render communication
any clearer or more precise than it actually is. The actual
selection of that theoretical perspective involves both the
preference of the experimenter and the theory's support in
the literature.
Many researchers focus on the nature of psychotherapy
as a special case of a general dyadic communication model
(Kiesler, 1973). Certainly there is ample theoretical pre-


89
Training Manual: Self References (SR)
In scoring SRs, the unit is the PROPOSITION, the cate
gory is SR, and the indicator (or rule) is defined by four
rules discussed below. PROPOSITION has been defined ear
lier. The category is noted by entering a code number for
the PROPOSITION scored, indicating the rule number used,
and labeling the scoring page, SR. The code number is com
posed of the page label plus the statement number plus the
PROPOSITION number. Since several SRs may follow one an
other, the code number may be entered as a range as long as
only SRs are included. If this is done, each scoring rule
used at least once within the sequence is indicated.
Generally, a self reference (SR) is a verbal response
by a subject which expresses descriptively or emotively some
thing about the subject in relation to himself, others, or
the world. This indicator is more specifically defined by
the following four rules.1 Examples of each follow the rule
Rule 1: A self reference (SR) statement must begin with or
include a first person singular personal pronoun (e.g., I,
myself, mine) unless it is covered by one of the criteria
below.
Rule 2: First person plural pronouns (e.g., we, our, our
self, ours) are counted as self references when the group
referred to is intimately related to the subject (e.g.,
family, roommate, girlfriend).
Rule 3: Second person pronouns (e.g., you, your, yours) are
counted as self references when they are clearly used as
a substitute for a first person pronoun form and not as an
1Note: These four rules are taken directly from materials
received from Green and Marlatt (Note 1).


APPENDIX E
RULES FOR LOCATING SEGMENTS
WITHIN FULL INTERVIEWS
Master tapes are composed of three-minute segments of
interviews with interviews and segments within interviews
in randomized order. There are a total of 120 segments with
3 segments from each of the 40 approximately 50 minute in
terviews. Position of segments within each interview was
determined by these rules.
Rule 1. Segment 1 begins at the beginning of the in
terview and continues as described in Rule 4. Specifically,
the beginning of the interview is defined as the first video-
resolved frame with clear audio. At that point the counter
on the play-back recorder is set to 000.
Rule 2. Segment 2 begins with the first statement (as
defined in the Method section) which begins after 17 minutes
of interview time have elapsed. Specifically, the play
back recorder is run forward to 425. It is then played past
430, and the first statement beginning after that point
(430) is noted. The playback recorder is then rewound,
played forward, and recording started at the identified
statement. All machines used in this recording of master
tapes were identical SONY units. (Other machines would need
to have different counter numbers specified.)
Rule 3. Segment 3 begins with the first statement (as
defined in the Method section) which begins after 17 minutes
of interview time have elapsed. The same procedure is used
76


19
for the client of the desired client behavior, and this has
also been examined and found effective in increasing client
affective verbalization in several studies (e.g., Green &
Marlatt, 1972; Marlatt, 1971; Myrick, 1969; Whalen, 1969).
In using this strategy, a therapist would be observed to use
affective words, or in the case of modeling affective self
references, to make statements with reference to the thera
pist's affect.
Various types of verbal conditioning techniques are the
third strategy and the area to be considered here uses
"agreement" as a social reinforcer, with specific instances
as "mmm-hmm," "good," "wonderful," "yeah," "I see," etc.
(e.g., Hekmat, 1971b, 1974; Hekmat & Lee, 1970; Moos & Ciernes,
1967; Salzinger & Pisoni, 1960). A therapist using this
strategy would be observed to give reinforcing responses af
ter a large proportion of client statements containing af
fect words or idioms (or affective self references) and af
ter a small proportion of other statements. Effective use
of this strategy probably involves indiscriminant and fre
quent use of the reinforcing response(s) early in the ses
sion to establish its reinforcing power (Waskow, 1962), and
the discriminant use of such responses subsequently.
Research has indicated that another important factor in
considering variables in therapist-client control is the in
terpersonal need compatibility of the particular therapist-
client dyad (Gassner, 1970; Sapolsky, 1960, 1965). The
FIRO-B (Fundamental Interpersonal Relations Orientation -


APPENDIX C
NOTICE FOR POTENTIAL CLIENT SUBJECTS
An Analogue Study of Counseling
We are conducting a study about counseling and
psychotherapy, and need a number of subjects to be
clients who will talk with therapists. The first part
of the study involves taking some paper-and-pencil tests,
and only a part of the subjects who take the tests will
go on to the second part of the study. The first part
of the study will take about 30 minutes, and you will
receive 1 half-hour credit. The second part of the study,
if you are selected, will take about an hour-and-a-half,
and you will receive 3 half-hour credits. If you would
like to participate in this study, please sign up below
for only one of the testing times.
74


CLIENT AFFECTIVE VERBALIZATION
103
6.0 -
5.5 -
5.0 -
4.5 -
4.0 -
3.5 -
3.0 -
2.5 -
Legend: HM
HF
LM
LF
12 3
SEGMENT
High compatible dyad males.
High compatible dyad females.
Low compatible dyad males.
Low compatible dyad females.
INTERACTION GRAPH:
COMPATIBILITY X CLIENT SEX X SEGMENT
FIGURE L-l


52
TABLE 7
CORRELATIONS OF REINFORCEMENT STRATEGY
AND CLIENT CHANGE SCORES
CHANGE
THERAPIST REINFORCEMENT
STRATEGY BY SEGMENT
BY
SEGMENTS
1
2
3
TOTAL
CLIENT
AFFECTIVE
VERBALIZATION
1 to
2
-.04
. 12
-.08
.01
2 to
3
-.35*
-. 32*
.16
-.29
1 to
3
-.44**
-.25
.10
-.31*
CLIENT
AFFECTIVE
SELF REFERENCE
1 to
2
. 12
.28
.02
.20
2 to
3
-.53**
-.51**
.05
-.50**
1 to
3
-.46**
-.29
.07
-.35*
* £<.05
** £<.01


REFERENCE NOTES
1. Green, A. H., & Marlatt, G. A. Scoring criteria for
feeling and content responses. Unpublished re
search materials, 1972. (Available from G. Alan
Marlatt, Department of Psychology, University of
Washington, Seattle, Washington, 98195.)
2. Goldman, J. R. Therapist-client interpersonal need
compatibility and interview process and outcome.
Study in progress, 1976.
108


31
words which idiomatically refer denotatively to affective
states. Examples are "sick and tired (of)," "get a bang out
of," "hair stands on end," etc. Only part of speech changes
from the listed forms which maintain the denotatively af
fective meaning were also scored. The rater's task for this
affective expression was to compare each PROPOSITION of a
transcript with list J, and to record the code number of
that idiom's PROPOSITION. The same rating and verification
process was used as in previous rating, and the rater was
trained using the manual in Appendix G. Two raters were
used, and in reliability checks of approximately 350 PROP
OSITIONS each, error rates of 0% and .1% were obtained.
Self References. Self References were scored using
the unit: PROPOSITION, the category: Self Reference, and
the indicator: "unit is a 'verbal response by a subject
which expresses, descriptively or emotively, something
about the subject in relation to himself, others, or the
world'" (Green & Marlatt, 1972). This indicator was fur
ther specified by four rules utilized in training the rater,
and are in the training manual for this scoring task (See
Appendix G). The rater's task was to consider each PROPO
SITION in the transcript and to determine the presence of
any pronouns. If a pronoun was found, the PROPOSITION was
compared to the scoring rules, and if found to be a Self
Reference was labeled by recording the PROPOSITION code
number(s) involved. Three raters were used to score this
variable and reliability checks over 343, 769, and 353 PROPO-


68
present type of study. On the well-controlled end of the
spectrum, there are the studies which generated the litera
ture used to define the therapist strategies and the depen
dent variables. The dependent variables seemed adequate,
but the "therapist strategies" were apparently foreign to
actual therapists in a similar situation! On the other
hand, the present study was, by design, loose with respect
to these variables, and consequently is very limited in its
richness for interpretation or generalizability, vis-a-vis
theoretical precision. A compromise between these two is
sues might be a study similar to the present one, with more
attention to the construct validity of the variables and
some assurance, by manipulation of therapist or client, that
the issue of strategies to increase client affective verbal
ization is clearly defined in the data.
The present data are also helpful in demonstrating as
sociations between affective verbalization and variables
such as client sex, therapist sex, compatibility, quantity
of verbalization of all types, inconsistency of change
across standard segments, and other affective verbalization
measures. Each of these must either be controlled or mea
sured as a variable in subsequent research if it is to re
sult in more significant, interpretable findings.
This appears particularly true of compatibility, since
not only did the compatibility manipulation seem reasonably
sound (though not significant overall), but the calculated
compatibility components also received some support as po
tentially related to the other variables of the study. Fur-


APPENDIX G
TRAINING MANUALS FOR
RATING ALL VARIABLES^
Your role in this study is to locate instances of cer
tain verbal behavior in typed transcripts of interviews.
It is critically important that your work be done accurately
and it will require that you concentrate well on it. Gen
erally, your rating task is quite specific and should not
be too technically difficult.
There are three definitions that will specify the
score you make in each situation. The first definition is
called the "unit," the second, the "category," and the
third, the "indicator" or set of rules for assigning a unit
to a specified category.
In all of the rating tasks in this study, several ideas
are important. One is the manner in which the pages of the
transcript are labeled. The largest division labeled is
the "interview segment" and this is the second number in
the label (the first number is an arbitrary grouping number)
The third number is the page number within a particular seg
ment. The last number is the page number within the entire
group of all transcripts. In other words, the page labeled:
1-04-2-008, is the second page of the fourth segment in the
first group of segments, and it is the eighth page of the
entire 288 pages of the transcript.
iNote: Each rater had a training manual with this intro
duction, the section corresponding to his or her task,
and one or more of the lists of words or phrases from
Appendices H, I, J, or K as specified.
81


33
Modeling, the second therapist strategy, was defined
earlier as the therapist's use of affective verbalization,
both with and without self reference. Since every thera
pist used some affective words, this variable is a rating
of how many affective verbalizations and how many affective
self references the therapist made. No ratings in addition
to those discussed above were necessary.
Reinforcement was scored by raters using the units:
WORD and PROPOSITION, the category: Reinforcer, and the
indicator: therapist makes one or two word expression of
agreement or approval, and specifically uses only those
words or expressions listed in Appendix K. Examples are
"mmm-hmm," "yes," "that's nice," etc. The rater's task was
to consider all therapist verbalizations, and to label all
words or expressions from Appendix K by recording the State
ment number of the therapist verbalization. Since the
typist is the first "rater" to decide whether or not to
place a questionable verbalization in the transcript, the
typing by the typist and verification by the experimenter
of the transcripts was considered the first stage of a two-
stage process of rating reinforcement. No assessment of
reliability was made at the first stage. The raters were
trained using the training manual in Appendix G. Two raters
were used, and over approximately 350 PROPOSITIONS each,
error rates of .4% and .7% were obtained.
Following the rating procedures, the necessary counts
over various summarizing units and contingent conditions
were made to obtain the following measures separately for


25
being composed of three segments, and the design, therefore,
can be viewed as a factorial design: 2X2X3, compati
bility X sex of client X segment of interview (repeated
measures). Where inappropriate to apply analysis of var
iance techniques, the independent variables and the depen
dent variables were examined by considering their correla
tional associations and graphs of interaction relationships.
Procedure. The raw data for this study were taken from
transcripts of interviews videotaped as part of a study by
Goldman and the author (Note 2). Therapists and potential
clients were tested on the FIRO-B, and the combined recipro
cal compatibilities for Inclusion and Control computed for
all pairs. Potential clients had been obtained through
their responding to a notice on the Psychology Department
experiment bulletin board (see Appendix C), and the therapists
were obtained by personal contact with the experimenters.
High compatibility was operationally defined as existing in
a pair when that pair's score fell within the bottom 5% of
that therapist's scores with all potential clients. Low
compatibility was defined as scores falling in the top 5%.
If insufficient matches were available to fill all four
cells [same- and opposite-sex clients and high and low
compatibility], additional potential clients were tested
until all 40 pairs were obtained. Clients were informed of
their selection for the second part of the experiment
through the phone message in Appendix D. Generally, each
client was informed of his or her selection for the study,


11
emotional state on the basis of his verbal behavior (e.g.,
Gottschalk & Gleser, 1969). In the present study as in other
studies to be discussed later, the importance of the client's
expression is that it is done directly, by verbalizing using
feeling words, particularly since it facilitates a process
of the therapist and client mutually coming to understand
something about the client. There are, of course, other
points at which the client is expressing or experiencing
feelings, and the client's direct report of his feeling may
or may not fit our judgements about his immediate affective
state. In considering this study's specific goal for the
client, however, it is the client's verbalization of affect
words that is of interest.
Defining client affect expression in this way also fa
cilitates the precision and usefulness of psychotherapy re
search as scientific data. Haley (1964) as well as others
(see Kiesler, 1971, 1973) argue for the selection of dis
cretely measureable behaviors in researching psychotherapy
and other interpersonal processes. Doing so minimizes the
inference inherent in the measurement process itself, and
allows us to look formally at the inferences made in a sys
tematic way.
The degree to which we can make inferences based on
this kind of data is limited by its measuring less general,
though more reliably measured, behaviors. As long as the in
ferences made from the study are made after careful consider
ation of the limitations of the data, the problem is more
one of frustration in being unable to deal with more rele-


BIOGRAPHICAL SKETCH
Stanley E. Jones was born August 21, 1948, at Big
Stone Gap, Virginia. In June, 1966, he was graduated from
Dobyns-Bennett High School in Kingsport, Tennessee. He
received the Bachelor of Arts degree in psychology from
Tennessee Wesleyan College in June, 1970. In September,
1970, he entered the Graduate School of the University of
Florida in the clinical psychology program, completing the
Master of Arts degree in June of 1972, and is presently
working toward the degree of Doctor of Philosophy. Since
September, 1975, he has worked as a psychologist for the
North Central Florida Community Mental Health Center.
Stanley E. Jones is married to the former Barbara-jeanne
Muceia.
116


61
and affective self reference respectively. These hypothe
ses were tested in analyses of variance 3 and 4 respective
ly, and were not supported (see Tables 10 and 11).
Though not stated as a specific hypothesis, the inter
action effect of compatibility X client sex X segment for
client affective self reference was significant (see Table
2), though it was not significant for client affective verbal
ization (see Table 1). Simple main effects tests on the sig
nificant interaction (see Table 2) indicated that compati
bility was a significant effect only when considering female
clients. Sex of client approached being a significant
effect when considering clients in highly compatible dyads
only.
Client Verbalization and Session Outcome
The third group of hypotheses deals with examining
the relationship between the client verbalization measures
and the measures of session outcome. Hypothesis 7a
predicted that for both client affective verbalization and
client affective self reference there would be a significant
correlation with client evaluation, therapist evaluation,
and changes in pre- to post-session change scores on
the problem rating scale. Of these only two correlations
were significant, and they partially confirmed the predic
tion. Total client affective verbalization and this verbal
ization in the second segment were correlated with therapist
evaluation (r = .34, £ = .031; r = .37, £ = .020). The rela
tive independence of client affective verbalization in seg-


90
impersonal reference to a group. They can be checked, if
necessary, by comparing the sense of the statement when
substituting a first person pronoun with the sense of the
statement when substituting a group reference (e.g., they,
everybody, the world, etc.).
"You get a good feeling with these people." is counted.
"You can't solve anything that way" is not counted.
Rule 4: A response may be a self reference without explic
itly using a personal pronoun if it is closely related to
a previous self reference and is contingent upon it.
These responses can be checked by adding a self reference
phrase such as "to me" to the statement.
"I have met a lot of people here. It certainly has been
satisfying." (Both sentences are self references.)
Here are some more examples, with SRs underlined. Note
that each example illustrates one of the rules.
Ex. 1: /I like this. /This is mine./
Ex. 2: a) /My family is a lot of fun. /We do things to
gether /
b) /American citizens celebrated the bicentennial./
/It was very interesting to us all./
Ex. 3:
a) /You really like coming here./
b) /You do this,/ and it will make a lot of dif
ference /
Ex. 4: /I have really tried hard./ It has just about done
itT7
Ex. 5: /I like doing things like this/that are a lot of fun./
Now try this reproduction of part of a transcript, un
derlining any PROPOSITION that should be scored SR.
T: But you have hurt them.
C: Sure. Whenever you fight a lot you always hurt some
body, particularly a child.
T: You sound a little bit guilty.