Relationship of Change in Ego-State to Outcome of Stuttering Therapy: Preliminary Findings

William S. Rosenthal

California State University, Hayward

(Based on a paper presented at the 3rd World Congress on Fluency Disorders, August 2000, Nyborg, Denmark)

The relationship between treatment success and change in ego states was investigated in 32 young adult stutterers who completed an intensive eight-week therapy program. Preliminary results show a significant change in two of five ego states (Adapted Child and Nurturing Parent). Adapted Child decreased and Nurturing Parent increased in those subjects who improved most. Regression analysis shows that approximately 32% of the variance associated with treatment change was predicted by changes in ego states.


The exact mechanism of fluency breakdown in stuttering is not fully understood. Stuttering typically begins in early childhood and the pattern of dysfluencies among different children shows little variability. However, when stuttering persists into adolescence and adulthood, the patterns of dysfluency become individualized, elaborated and often bizarre. The variability of patterns seen in adult stuttering reflects the individual's struggle against social, emotional, and linguistic stresses that are associated with fluency failure.

Disagreement about the underlying causes of stuttering is often overshadowed by controversies about its treatment. There are two main approaches. The first is suppressive in nature. The stutterer is taught to re-train motor speech behaviors and to substitute fluency shaping techniques for stuttering. While this approach is often initially successful, there is a significant relapse rate, probably due to the necessity for a persistent and high degree of vigilance that most stutterers find impossible to maintain in routine speaking conditions. They grow weary of trying to manage in detail a process that should be free-flowing, with as casual acceptance of disfluencies as that shown by those who speak normally. Also, recipients of this kind of therapy typically complain that the resulting speech pattern sounds unnatural.

The second major treatment approach, and the one that is reported on here, is the modification of stuttering dysfluencies into a more acceptable and easier speech pattern. The therapy program has two main goals. First, to gradually modify the muscular patterns associated with stuttering by releasing hard articulatory contacts, unlocking the vocal cords, relaxing other tensed muscles in the tongue or lips, and letting the sound come out instead of holding back. Second, to reduce negative emotional responses that are associated with stuttering, such as fear, guilt, and embarrassment. An essential component of this approach requires the stutterer to closely examine not only the physical, but also the emotional and social aspects of stuttering, in order to reduce the impact of stressors that cause fluency breakdown (Naylor & Rosenthal, 1968). This is a decidedly confrontational approach, and its success is dependent in part on the ability of the stutterer to address life script issues that mold approaches to problem solving (Steiner, 1974) and to energize ego states that are compatible with problem solving (Berne, 1961; Dusay & Dusay, 1989). For example, when young children first encounter fluency breakdown, they do not possess the requisite analytical and logical thought processes necessary to accommodate or adjust to the problem. Ineffective childhood strategies follow, including avoidance, struggle and denial, and these strategies are carried into adolescence and adulthood. The adolescent and adult stutterer is capable of exerting more thoughtful and intentional control over episodes of fluency breakdown, but only if the appropriate ego states are energized. Stuttering therapy thus becomes a subtle blend of mechanical manipulation and supportive psychotherapy.

In previous work I have discussed the theoretical relationship between scripts, ego states, and stuttering, supported mainly by case studies, but also including a preliminary computational analysis that showed a change in ego state over the course of stuttering therapy (Rosenthal, 1998). The present research project extends that investigation by using archival data collected some 30 years ago as part of another research project (Naylor & Rosenthal, 1968; Rosenthal & Naylor, 1968).

Several investigators have described the use of ego state analysis in treatment (Dusay & Dusay, 1989; Marsh & Drennan, 1976; Mincis & McFarren, 1982; Williams & Williams, 1980). In the current analysis, each subject's predominant ego state was assessed from the visual record of pre and post therapy videotaped recordings. Ego state constructs as defined by transactional analysis theory were used. Berne (1961) has described ego states as coherent systems of thought and feeling, and their associated behaviors. They include Adult, Adapted Child, Natural Child, Critical Parent, and Nurturing Parent. The Adult ego state is computer-like, objectively assessing the environment and making decisions appropriate to those assessments. The Child ego state is the repository of feelings and behaviors that were acquired as a real child, but which persist into adulthood. There are two child states, Natural Child, which is exhibited in autonomous behavior, and Adapted Child, which reflects the dominance of parental influence. The Parent ego state is exhibited by behaviors and attitudes that are like the individual's own parents. There are two of these, Nurturing Parent, from which the child experiences positive feelings such as pride and accomplishment and love, and Critical Parent, which is the source of arbitrary, non-rational and prohibitory reactions. Normally there is fluid movement between ego states, although only one is active at a given moment. The sense of self derives from the ego state that is active at any particular moment.

The hypothesis addressed in this study is that success in achieving the goals of treatment will be associated with changes in ego states. In the present study, ego states were assessed by observing facial expressions, gestures, and body posture. In addition, separate clinical ratings were made of each subject, before and after therapy, that reflected the degree of severity of the features addressed in this particular stuttering modification treatment program (see Appendix). In particular, it was predicted that treatment success would be associated with increased functioning of Adult, Nurturing Parent, and Natural Child ego states. Conversely, it was predicted that treatment success would be associated with decreased functioning of Adapted Child and Critical Parent ego states. This report presents preliminary findings of the investigation.



Subjects and Media

The subjects for this study were 32 young adult stutterers who were on active duty in military service. There were 31 males and 1 female. The subjects were those participants of a larger research project for whom complete recorded data of both pre and post therapy was available (Naylor & Rosenthal, 1968). The ages of the subjects in the larger study ranged from 18 to 35 with a mean age of 24 years. The subjects, who were referred for stuttering therapy at the Army Audiology and Speech Center at the Walter Reed Army Medical Center in Washington, D.C., participated in an eight-week, intensive in-patient treatment program that was based on a modification approach similar to that described by Van Riper (1973). Each participant was recorded under several controlled speaking conditions, before and at the conclusion of the therapy program. These included one-minute of reading, two minutes of spontaneous speech, and a one-minute telephone call. Samples of speech were originally collected for each subject on 16mm film that was produced from video taped recordings. For the present study, the original films were converted to digital media and both the video and audio signals were enhanced electronically. New VHS master tapes were produced for all subjects, resulting in a set of 64 pre and post therapy segments of approximately five minutes each.

Treatment Ratings

For this phase of the study, the 64 segments were arranged sequentially, so that the pre therapy segments occurred first and post therapy segments occurred second for each of the subjects. Three raters evaluated each of the segments, the author and two graduate students in speech-language pathology. The graduate students were also stutterers, who had participated in treatment programs similar to the one described in this study. Ratings were made of each segment immediately after viewing, and followed a seven point, equal appearing interval scale from "Very Mild" to "Very Severe". Seven criteria were applied to the rating of each subject and segment: (1) Indications of excessive muscle tension while stuttering, (2) Indications of inefficiency, or undue delay in releasing tension, (3) Attempts to avoid or postpone stuttering, (4) Inappropriate movements while stuttering that are conspicuous and distracting, (5) Instances of inappropriate rate associated with stuttering, (6) Instances of failure to initiate or maintain eye contact with the camera while stuttering, (7) Instances of inappropriate loudness, pitch, or voice quality associated with stuttering (see Appendix).

Ego State Ratings

Ego state ratings were made by the author and another speech-language pathologist, who was familiar with ego state identification and who was experienced in applying the theory and technique of transactional analysis in the treatment of speech disorders. For this phase of the study, the 64 segments were arranged in a pseudo random order, so that the pre therapy segments occurred first for half of the subjects, and post therapy segments occurred first for the other half. Pre and post therapy segments for a given subject were separated by at least four unrelated segments. Only the spontaneous speech segments were presented. The segments were viewed without sound, so that ratings would not be influenced by comments from the subjects about the success or failure of their therapy experience. Therefore, ego state ratings were based on non-verbal information only, mainly facial expressions, gestures, and body posture. The relative strength of each ego state, Adult, Adapted Child, Natural Child, Critical Parent, and Nurturing Parent, was assessed and assigned a score that totaled 20 points for all ego states combined.



Reliability of Measurements

There is substantial disagreement about how and what to measure when assessing treatment success in stuttering. That fact notwithstanding, even so-called objective measures of stuttering have somewhat poor reliability (Cordes & Ingham, 1994; Lewis, 1995). The treatment ratings used in this study are no exception. The present study used a seven point, equal appearing interval scale (see Appendix). When both pre and post therapy ratings were evaluated for all combinations of the three raters, inter-rater reliability coefficients (r) ranged from .59 to .75. These results suggest that each of the raters used somewhat different criteria when assessing the subjects' status before and after therapy. Since the Treatment Success scores are the means of the three rater's pre- and post- difference scores, the Treatment Success scores represent, then, a rather diverse set of criteria. Two of the raters repeated their ratings for all subjects within an interval of from one to three weeks. The intra-rater reliability coefficients for these raters ranged from .75 to .83. Even the repeated application of the same evaluation criteria for the same set of subjects results in ratings that are not robustly consistent.

The ratings of ego state change showed even less agreement, with statistically significant inter-rater reliability coefficients ranging from .40 to .61. In this instance, it is clear that the requirement of rating ego states, based solely on visual information, was a quite difficult task.

Test of Hypothesis

Subjects were divided into two groups based on their Treatment Success ratings, a Low Success group comprised of subjects whose change score was 1 or less (Mean = .52), and a High Success group comprised of subjects whose change score was greater than 1 (Mean = 2.49). A two-way ANOVA with repeated measures was used to compare the two groups with respect to changes in each of the ego states over the course of treatment. A significant interaction between Treatment Success and Ego State was found (F =2.49, p <.05). The mean change in each ego state is shown for both success groups in Table 1. Significant differences between the two success groups were found for Adapted Child and Nurturing Parent ego states. Figure 1 shows these relationships more clearly. Although only two of these group differences were statistically significant, all differences between the two success groups were in the expected direction. That is, when compared to the Low Success group, the High Success group showed greater increases in Adult, Natural Child, and Nurturing Parent ego states, and decreases in Adapted Child and Critical Parent ego states.

The use of parametric statistics in the above analysis provides a degree of descriptive detail not available in non-parametric procedures. However, the use of parametric analysis in this instance may be questioned because of the peculiar characteristics of ego state measures, and the possibility of lack of independence of, and interaction between, those measures. In addition, the rating scheme used in the study to measure treatment success may be subjected to the same criticism. Therefore, an additional, non-parametric analysis was done to compare the median ego state change scores of the two success groups. A Kruskal-Wallis one-way ANOVA using ranks, showed that the pattern demonstrated by Figure 1 is significant (Chi-square = 36.4, df = 9, p < .0001).


Table 1. Means, Standard Deviations, and Significance of Changes of Treatment and Ego State Ratings for High and Low Change Groups

GROUP   Treatment Change Adult Ego State Adapted Child Ego State Natural Child Ego State Nurturing Parent Ego State Critical Parent Ego State
High Success M 2.49 1.64 -2.60 .60 .74 -.38
(N=21) SD .95 2.59 3.12 2.18 .85 1.89
Low Success M .52 .68 - .32 -.27 .05 .09
(N-11) SD .43 1.71 2.27 1.46 .85 1.97
  t= -8.101 -1.256 2.359 -1.340 2.193 .652
  p <.0001* .231 .031* .193 .039* .525

* Statistically significant



Graph of Changes in Ego States of Low and High Treatment Success Groups (see text).


Predictive Model

Correlation analysis revealed the following significant relationships between ego state measures and severity ratings. 1) High ratings of pre-treatment severity are associated with high loading of

Adapted Child (r = .56) and low loading of Adult (r = -.50) ego states. 2) Treatment success is associated with high, pre-therapy ratings of Adapted Child (r = .48) and low loading of Adult (r = -.36). These results, together, suggest the not surprising finding that subjects who began treatment with a higher level of severity showed the most dramatic improvement, and that the improvement was associated with shifts in ego states.

In order to further examine the predictive relationship between ego states and treatment outcome, a step-wise linear multiple regression was performed, with the dependent variable Treatment Success, and the independent variables the changes in ego state ratings from pre to post therapy. The variables of Adapted Child, Natural Child, and Nurturing Parent resulted in a multiple R of .57. The addition of the remaining variables did not significantly increase the predictability of treatment success.



The results of the present study go in the predicted direction, but with statistical significance for only two of the five ego states. Moreover, the predictive power of the ego state model is limited. The results of the multiple regression analysis suggests that slightly more than 32% of the variance associated with treatment success is accounted for by changes in ego states. This is a sensible finding, since there are surely other factors that are associated with treatment success. Those factors may include the initial severity of the problem, the individual's responsiveness to particular types of intervention, and perhaps cognitive factors that are not yet entirely clear to us. There are two main criticisms to which the present results are subject. The measures of treatment success are based on ratings of descriptive criteria rather than on objective measurements of speech data. In addition, the measurements of ego states, limited as they were to non-verbal data, produced low inter rater agreement. Nevertheless, the present results are sufficiently encouraging to proceed with the remaining data analysis. That analysis will include SSI (Riley, 1994) type analyses of the pre and post treatment data. Those data will be compared to ego state measurements that are based on an utterance-by-utterance analysis of the spontaneous speech portion of the pre and post therapy transcripts.



Berne, E. (1961) Transactional Analysis in Psychotherapy. New York: Grove Press

Cordes, A.K. & Ingham, R.J. (1994) The reliability of observational data: II. Issues in the identification and measurement of stuttering events. Journal of Speech and Hearing Research, 37, 279-294.

Dusay, J.M.& Dusay, K.M. (1989) Transactional analysis. In Corsini, R.J. & Wedding, D. (Eds.) Current Psychotherapies (4th ed.) pp. 405-453. Itasca, IL: F. E. Peacock Publishers, Inc.

Lewis, K.E. (1995) Do SSI-3 scores adequately reflect observations of stuttering behaviors? American Journal of Speech-Language Pathology, 4, 46-59.

Marsh, C. & Drennan, B. (1976) Ego states and egogram therapy. Transactional-Analysis-Journal, 6, 135-137.

Mincis, P. & McFarren, C. (1982) TA in an alcoholic rehabilitation program. Transactional-Analysis-Journal, 4, 247-248..

Naylor, R.V. and Rosenthal, W.S. (1968) Clinical Investigations of Stuttering: II. Treatment and Follow-up of the Adult Stutterer. Final Report, Project No. 3A-015601A826-01-036, U.S. Army Medical Research and Development Command.

Riley, G.D. (1994) Stuttering Severity Instrument for Children and Adults (3rd ed.). Austin TX: PRO-ED.

Rosenthal, W.S. (1998) The transactional analysis of stuttering therapy: scripts and ego states. In E.C. Healy & H.F.M. Peters (Eds.) 2nd World Congress on Fluency Disorders Proceedings, San Francisco, California, 1997 (pp. 185-189). Nijmegen, The Netherlands: Nijmegen University Press.

Rosenthal, W.S. and Naylor, R.V. (1968) Clinical Investigations of Stuttering: I. Adjustment and Effectiveness of Stutterers in Military Service. Final Report, Project No. 3A-015601A826-01-036, U.S. Army Medical Research and Development Command.

Steiner, C. M. (1974) Scripts People Live; Transactional Analysis of Life Scripts. New York: Grove Press.

Van Riper, Charles (1973) The Treatment of Stuttering. Englewood Cliffs, N.J.: Prentice Hall

Williams, K.B. & Williams, J.E. (1980) The assessment of transactional analysis ego states via the adjective checklist. Journal of Personality Assessment, 44, 120-129.


Acknowledgement: Rex V. Naylor, Ph.D. reviewed this paper and made valuable recommendations that were incorporated into the final version.


MSWord download version egostatechange.doc





The tapes that you are going to view show individuals before and after an 8-week intensive therapy program for stuttering. Your task is to evaluate the success of each of these individuals. The objective of the therapy program was not necessarily to reduce the absolute number of dysfluencies, but rather for each individual to achieve an easier, more direct manner of stuttering by reducing or eliminating the following features:

Indications of excessive muscle tension while stuttering
Indications of inefficiency, or undue delay in releasing tension
Attempts to avoid or postpone stuttering
Inappropriate movements while stuttering that are conspicuous and distracting
Instances of inappropriate rate associated with stuttering
Instances of failure to initiate or maintain eye contact with the camera while stuttering
Instances of inappropriate loudness, pitch, or voice quality associated with stuttering


Each segment that you will see consists of three parts; reading, spontaneous speech, and a telephone call. Each segment is preceded by a code, (for example, TV17-3) that identifies each subject and whether that segment was recorded before or after therapy. It is important for you to match your ratings to the correct code for each individual and segment.

As you work through the tapes, always complete the rating of a single subject (both pre and post therapy segments) before stopping. You do not have to complete all 32 subjects (64 segments) at one sitting. It is better to work no more than 1 hour at a time. Take breaks if you find yourself becoming tired or inattentive.

1. For each subject, please view the first segment (pre therapy) in its entirety and then rate that subject overall on the features listed below. Use the following scale and mark your choice on the Subject Rating Sheet next to the subject's code number.

1-Very Mild 2-Mild 3-Milder Than Average 4-Average 5-More Severe Than Average 6-Severe 7-Very Severe

2. After you have rated the first segment, view the second (post therapy) segment in its entirety and then rate the subject overall on the features listed below. Use the following scale and mark your choice on the Subject Rating Sheet next to the subject's code number.

1-Very Mild 2-Mild 3-Milder Than Average 4-Average 5-More Severe Than Average 6-Severe 7-Very Severe