Treatment Notes: Objective Measures of Language Style Point to Clinical Insights

Abstract

Application of a computerized text analysis procedure is proposed that has the potential for use by psychoanalytic and psychodynamic clinicians: the systematic examination of linguistic style as reflected by clinicians in their ongoing process and case notes, which are ubiquitous in the mental health field. The studies reported here are, as far as is known, the first attempts to study treatment notes systematically using such procedures. Linguistic measures are used to track the trajectory of the clinical process throughout the treatment in two contrasting cases, one rated successful, the other not. The computerized linguistic analysis used here focuses on two analytically relevant linguistic variables: Mean High Referential Activity (MHW), a measure of the degree to which language is connected to emotional processing, and Reflection (REF), the use of words referring to logical functions. Changes in the relative position of these measures indicate nodal points in the treatment that might be analytically or therapeutically problematic, and that might be overlooked in a solely clinical reading. The analyst’s activity as reported in notes during such nodal periods is clinically examined to see how it may have affected the course of the analysis. This method has the potential for use in ongoing treatments, and may help clinicians refine their interventions.

In many of the health care professions there is a gap between science and practice, and for far too long this gap has been widest in psychoanalysis and psychodynamic psychotherapy. In recent years increasing attempts have been made to narrow this gap (e.g., Luyten, Blatt, and Corveleyn 2006; Levy and Ablon 2010; Kächele, Schachter, and Thomä 2009; Freedman et al. 2011; Midgley 2012). Yet clinicians and students of clinical psychoanalysis still find it difficult to incorporate research methodology in their clinical work and education (see Schachter and Luborsky 1998; Werbart 2009; Preter, Algus, and Hoffman 2012). It is essential, then, that psychoanalytic and psychodynamic researchers attempt to demonstrate to clinicians the value of systematic empirical studies for their work with patients and in teaching their students. Over the years, Wilma Bucci and Bernard Maskit have been engaged in a long-term project that tries to integrate systematic linguistic analysis of psychoanalytic and psychotherapy sessions with the clinical evaluation of those sessions (see, e.g., Bucci and Maskit 2007; Bucci, Maskit, and Hoffman 2012).

In this, the second in a series of studies, we propose a systematic method with potential for use by psychoanalytic and psychodynamic clinicians: the examination of clinicians’ linguistic styles as reflected in their ongoing process and case notes, which are ubiquitous in the mental health field.

Ongoing treatment notes (process notes or periodic summaries of treatments) written by analysts and dynamic therapists as a treatment progresses constitute the traditional database for psychoanalysis and psychodynamic psychotherapy: for training and supervision, for writing case reports, for confirming an accepted theory or demonstrating the value of an alternative or new one (Wallerstein and Sampson 1971; Tuckett 1993; Michels 2000; Colombo and Michels 2007; Glick 2008; Wolpert and Fonagy 2009).

Colombo and Michels (2007) maintain that “no other method of studying psychoanalysis can give us this same information, fragmented and distorted as it inevitably is, about the psychoanalytic process”(p. 640). Glick (2008) stresses that “writing can offer access to unconscious aspects of the analyst’s experience, defensive behavior, thinking, emotion, values, and attitudes. It makes available for reflection many of the dimensions of lived countertransference, and provides some perspective on the analyst’s role in therapeutic process and change, as both a transference object and a new object” (p. 1228).

To our knowledge, our studies are the first attempts to study treatment notes systematically using computerized text analysis procedures developed in the context of a psychological model of the psychoanalytic process. This method can be useful both scientifically and educationally, for clinicians and their supervisors, and can further the systematic scientific study of the comparative effectiveness of different interventions as implemented in everyday clinical treatment. In a series of studies we address the question of how notes can be used in a manner that satisfies the clinical needs of students and their teachers, while adhering to modern scientific criteria.

Aim of the Present Study

In this exploratory study we illustrate the potential clinical applicability of a systematic examination of the linguistic measures of ongoing clinical notes by analysts. We note how we can observe in graphic form the trajectory of the linguistic measures through the course of treatment and how the variations in these measures can be used as guides in clinically assessing the ongoing clinical notes. We suggest that variations of the linguistic measures provide a diagram of the clinical course described by the analyst and that these variations can help clinicians recognize potentially critical areas in the treatment (e.g., turning points that we call “nodal periods”) that may be missed when the material is viewed only from the clinical perspective.

To illustrate our proposed method we will compare the ongoing treatment notes of two cases, one successful and the other not. In both these cases we show how the language measures pointed to nodal periods in the analytic work. These periods were then examined in depth in an effort to understand why one case progressed to a successful termination, whereas the other ended in a disruption and a forced interruption of the treatment by the analyst.

The Overall Project

In a series of studies we have systematically examined treatment notes from fourteen analyses carried out by psychoanalytic candidates under supervision at the New York Psychoanalytic Society and Institute Treatment Center during the latter half of the twentieth century. Because of the great size and unsystematic nature of the collection of case material at the center, devising methods for studying it has been difficult. The development of automated techniques has finally made it possible to begin its systematic study. We describe our use of these techniques in detail in Bucci, Maskit, and Hoffman (2012).

Our Initial Study

In our initial study (Bucci, Maskit, and Hoffman 2012) we discussed the scientific problems and pragmatic value of using treatment notes as data, and the theoretical and empirical background of Bucci’s multiple code theory. In that theory, Referential Activity (RA) is an empirically derived measure that assesses the degree to which language (verbal symbolic) is connected to the nonverbal elements of emotional processing (nonverbal symbolic and subsymbolic), including the degree to which a person’s use of language is connected to his or her emotional experience.

This theoretical approach, which has been employed in many clinical and experimental studies (e.g., Bucci 1997; Bucci and Maski, 2007; Nelson, Moskovitz, and Steiner 2008), was applied for the first time to the study of notes written by psychoanalysts by Bucci, Maskit, and Hoffman (2012).

In this first study we described in detail the methods of the study, including selection of cases¹ from the Treatment Center archives and confidentiality protection. The selection procedure resulted in a sample of fourteen cases over the last five decades of the twentieth century, including at least one successful and one unsuccessful case from each decade. The selection was made using both ratings by clinicians at the time of treatment and current verification by expert clinicians using the Global Assessment of Functioning scale of DSM-IV (GAF). The scale was applied to beginning and end segments of the treatment notes; the segments were randomized, and raters were blind to the clinical evaluation at time of treatment. In this initial assessment, a GAF level of 61 or better at the close of treatment was set as a minimal cutoff point for positive outcome; for the full assessment, change measures were also included.

The fourteen cases selected for this study were also evaluated using the Psychodynamic Functioning Scales (PFS; Høglend et al. 2000, 2006). The PFS is a good measure for judging the outcome of a psychoanalytic treatment because it “seems to measure a construct that is discriminable from general symptoms and dysfunction” (Høglend et al. 2000, p. 194). The PFS has six scales: Quality of Family Relations; Quality of Friendships; Romantic/Sexual Relationships; Tolerance for Affects; Insight; Problem Solving and Adaptive Capacity. The scales of the PFS, like the GAF, are scored between 0 and 100, with anchors every ten points, interpreted as meaningful clinical differences.

All scales were scored by three raters. Reliability was measured as intra-class correlation (ICC) representing level of agreement among all three raters. For the PFS scales, ICC of .6 or better was achieved for five of the six scales; adequate reliability was not achieved for the Quality of Friendships scale, which therefore was not included in our assessment. ICCs for the PFS ranged from .633 to .80 for the other five scales, with a mean ICC of .699. ICC was .874 for the GAF.

For this earlier study, a composite measure called the Composite Clinical Evaluation (CCE) was computed based on final scores and change scores for the GAF and PFS measures (see Bucci, Maskit, and Hoffman 2012). This procedure followed the approach used by Luborsky et al. (2001) in their evaluation of treatment outcome based on archival transcripts in the Penn Psychoanalytic Treatment Collection. For both GAF and PFS, if the end segment received a score of at least 61, but not greater than 70, one point was added to the CCE; if the end segment score was at least 71, two points were added. Also, for both the GAF and the PFS, if the difference between the end score and the beginning score was positive but less than 10, 1 point was added to the CCE; if this difference was 10 or greater, 2 points were added. Thus, each of the fourteen treatments received a CCE score between 0 and 8, with 0 representing no clinical improvement and 8 representing considerable improvement. A correlation of .733 (p = .003) was found between the Mean High WRAD (MHW; for WRAD, see below) of the candidates’ notes and the CCE as reported by Bucci, Maskit, and Hoffman (2012).

In the final assessment of treatment effectiveness, three cases showed an increase, with an end point above the clinical criterion of 60 for both assessment measures; seven cases showed mixed results. What is particularly interesting for the purposes of this clinical study is that four cases showed a decline in both GAF and overall PFS ratings from the beginning to the end of treatment to a level below the clinical criteria scores of both measures. GAF end scores ranged from 45 to 55, with an average decline of 6.25 points; overall PFS end scores ranged from 46.4 to 57.4, representing an average decline of 5.6. Here we have four patients, in treatment for an average of 38.5 months, who were not helped and, according to our measures, may have been hurt, at least symptomatically, by the treatments received. Table 1 shows that of the fourteen cases in our study, three showed considerable improvement, with two CCE scores of 8 and one of 7; four showed moderate improvement, with two CCE scores each of 3 and 2; seven showed essentially no improvement, with two CCE scores of 1, and five of 0.

Table 1.

Relation of clinical effectiveness and language measures for 14 cases

Treatment	CCE	MHW	REF	MHW/REF
50T1	0	.0657	.0717	−.006
50T2	7	.1098	.0801	.030
60T1	8	.1345	.0776	.057
60T2	3	.1133	.0620	.051
70T1	0	.0676	.0737	−.006
70T2	1	.0781	.1004	−.023
70T3	8	.0974	.0786	.019
80T1	0	.0781	.0868	−.009
80T2	0	.0830	.0903	−.007
80T3	1	.0803	.0907	−.010
90T1	3	.0833	.0852	−.002
90T2	2	.0575	.0915	−.034
90T3	2	.0932	.0951	−.002
90T4	0	.0907	.1117	−.021

The rationale for this project is the need to identify what is going amiss in a treatment at a point when it might be possible either to rectify it or to recognize that the patient is engaged in a treatment that is not helping, so that a referral elsewhere can be made. Since treatment notes are available for supervised treatments, we wondered if examination of the notes would allow us to identify such points. In making such assessments, computerized analysis allows us to track the full trajectory of a treatment, either in notes or in transcribed session material.

It is important to note that in this project we are looking at analytic material that has already been collected and archived; it is the kind of material customarily collected in psychoanalytic and psychotherapy centers. The notes reflect the nature of the treatment from the perspective of the analyst. Audiotapes of psychoanalytic sessions, which would allow the analysis of patient speech as well, are generally unavailable, and where they are, they require transcription before computerized linguistic measures can be applied. We also point out that the treatment notes permit examination of the analyst’s own experience in a way that is not possible with session tapes and transcripts alone.

Linguistic Measures

The linguistic measures used in the study have been described in detail in Bucci, Maskit, and Hoffman (2012), particularly the computerized linguistic measures of emotional engagement and affective language developed in the framework of Bucci’s multiple code theory (1997) and the referential process; these had already been applied to session transcripts (Bucci and Maskit 2007).

In brief, the stylistic features of a person’s language production (how a person speaks or writes) can be differentiated using a variety of empirically derived measures. The two most central are Referential Activity (RA) and Reflection (REF). Measures of RA assess the degree to which language is connected to the nonverbal elements of emotional processing and indicate the degree of affective connection of the language.

In broad terms, high-RA language is vivid, imagistic, and detailed and contains sensory information; low-RA language is abstract and general. High-RA language has a quality of immediacy; it evokes vivid and specific experiences and evinces an active and direct connection between imagery and words. The computerized Weighted Referential Activity Dictionary (WRAD; Bucci and Maskit 2006) was developed by empirically identifying lexical items that figure in the RA ratings.

In the construction of the WRAD, the importance of common function words (articles, conjunctions, pronouns) emerged empirically. For example, words such as and, in, on, the, each with weight of +1 are indicative of high-RA language, in contrast to also, much, quite, whether, each with weight of 0, indicative of low-RA language. It is important to note that these small words are used without specific awareness of any underlying linguistic dimensions and without intent regarding lexical choice. The following examples are taken from the treatment notes used in Bucci, Maskit, and Hoffman (2012).

In the first example, which illustrates high-RA language, the analyst reports a dream the patient has told:

That night she had a dream that she had walked into the luncheonette near the institute but it was located near her work. She wanted a hot drink. She also rented a spare room in order to sleep. One of the men there asked her whether she was in Q’s department or the dark room. She said she was not in the dark room. There was a girl there who serves in the cafe who said, why ask her who she is? She then went to the room and heard a baby crying upstairs.

In the following example, which illustrates low RA, a different analyst describes an interchange with the patient:

He starts talking spontaneously about the discussion yesterday and the waving incident, how when he leaves here he has to remind himself that he is an assistant manager in order to regain his status. I recall to him his remarks about me that “you would be a good analyst some day” and so forth. He then confesses that he has been thinking about his remarks about me ever since he left and wonders why he did talk that way to me. He is very confused by reality, since after all it is a fact that I am not finished with my training yet, so what’s wrong with what he says and yet he feels that there is something wrong. He goes into an obsessive rumination about what is real and about what isn’t here.

The differences in language style between these two passages are generally apparent to readers (or listeners); the elements contributing to these differences are not so obvious. To deconstruct the application of the WRAD measure, in the first passage the five most frequent words are she, the, in, and room, each with a WRAD weight of 1, and the article a, with a WRAD weight of 0.8125. These are the kind of words that people

use when describing events or images involving other people in specific spatiotemporal settings. In the second segment, the five most frequent words are he, about, that, and, and is; only the first of these, he, and the conjunction and have high WRAD weights; the other three are weighted zero or close to zero. Details of the WRAD list and the development of the weighting procedures are presented in Bucci and Maskit (2006). Optimally, in treatment notes we would expect to see language indicating engagement by the therapist with the patient and with the treatment, along with but not dominated by language indicating reflection on the material and its explication. A computerized measure of Reflection has also been developed composed of words that people use to describe how they think and communicate thought. The Reflection dictionary (REF) includes words of basic logic, such as if and but, words referring to cognitive or logical functions, such as think and believe, or referring to logical entities, such as reason and cause. The Reflection measure (REF) is the average number of words matching the Reflection dictionary in a text or text segment.

The measures of Referential Activity and Reflection are applied using the Discourse Attributes Analysis Program (DAAP).² The main measure of referential activity used here is the Mean High WRAD (MHW), the mean of referential activity when it is above the midpoint).³ This measure provides a measure of the average degree of intensity of emotional engagement. Whereas MHW is interpreted as a measure of emotional engagement by the speaker or writer, Reflection (REF) serves as an indicator of regulating or distancing emotional experience. By looking at the MHW/REF difference, we conjecture that high positive levels of this difference (relatively higher MHW and lower REF) indicate intense emotional engagement by the speaker or writer as the experience is described. The speaker or writer is living in the experience immediately, in the moment, and not standing back to reflect on it. The power of highly engaged narratives to express and activate emotion is widely recognized as a basis for exposure therapy (Foa and Meadows 1997) and in psychological and neuroscience research (Damasio 2003; Ellsworth and Tong 2006; Harber and Pennebaker 1992). Such intense engagement may be discouraged in forms of supportive treatment in which the therapeutic goal is to modulate and downregulate the patient’s affective intensity. Relatively lower MHW and higher REF indicate less emotional intensity and more emotional distance from the contents by the speaker or writer.

Construct and Content Validation of the MHW and Ref Measures

Murphy (2012, 2013) has presented evidence that the WRAD measures are robust predictors of language style associated with describing vivid and emotionally significant events and have moderate to high temporal stability. The fluctuations of the MHW and REF measures, and differences between them, have been examined in several studies using session notes and transcripts. High levels of MHW in analyst speech were significantly correlated with clinical ratings of session effectiveness in a study of sixteen psychoanalytic sessions (Bucci and Maskit 2007). In an experimental study, high levels of the Reflection measure were associated with judges’ ratings of language segments as representing processes of distancing from immersion in emotional experience (Kingsley 2009). A high level of MHW compared to REF (high MHW/REF difference score) has been associated with uncovering phases of treatment; a low or negative difference score has been associated with containing, and also in avoidant or distancing phases, both in patient speech and therapist reports (Andrei et al. 2008; Crisafulli et al. 2011).

Focus on the Individual Case:The Present Study

Our initial study provided a first line of evidence that the language style of treatment notes, indicating engagement of the therapist with the patient and in the treatment, can provide some indication of how well a treatment is going and is likely to go. As noted, only three of the fourteen cases in our sample showed clear-cut improvement at the end of treatment (their current clinical evaluation), and four became worse (see Table 1).

Table 1 shows the relation of clinical effectiveness (CCE) and language measures for the fourteen cases. Cases 50T2, 60T1, and 70T3, all of which were judged successful by current criteria, with a CCE of 7 or 8, all have positive MHW/REF difference scores, whereas the others have negative MHW/REF difference scores. As noted, cases written about with greater emotional intensity by the analyst were judged more successful.

For clinical purposes, in supervised treatments (and in all treatments), the next question we wished to address was whether these measures could be applied to the overall trajectory of individual cases that differed in outcome, to identify points in treatment where problems arose and how these might have been managed in cases that were considered effective.

Method

For this purpose we chose to examine the trajectories of two of our basic linguistic measures across the entire course of treatment using the treatment notes of two sharply contrasting cases,⁴ one from the 1950s (50T1) in which the patient became worse according to our measures, showing declines in both GAF and PFS aggregate scores and ending below the clinical criteria scores, and one from the 1970s (70T3) in which the patient became better according to all measures. Case 50T1 had a beginning GAF of 53 and a beginning average PFS of 52.6. He had an ending GAF of 45 and an ending average PFS of 48.8. Case 70T3 had a beginning GAF of 65 and a beginning average PFS of 56.4. She had an ending GAF of 80 and an ending average PFS of 78. The composite of these measures, the CCE, yielded the minimum score of 0 for 50T1 and the maximum score of 8 for 70T3.

Clinical Description of Patient 50T1

This patient, a professional man in his mid- to late 30s, was considered by experienced evaluators to be suitable for analysis and eminently analyzable. One senior interviewer who screened him said, “I would consider him an excellent and worthwhile candidate for psychoanalysis at the Treatment Center. It is my impression that he could effectively undertake and utilize such treatment, not only for his own benefit, but also in the service of his work.”

Another evaluator said, “He is an intelligent man, obviously sensitive and introspective. He speaks fluently and well. He makes excellent personal contact. . . . The brief content given probably does not adequately convey the excellent personal impression that [he has] made. The intellectual qualities were mentioned before; to these may be added what seems to be a real capacity for emotional warmth.”

The patient’s initial diagnosis was psychoneurosis. In terms of analyzability, the evaluators commented that he had an excellent prognosis.

After twenty-two months of treatment, the patient was seen as having a severe ego disturbance that made analysis impossible. The termination summary states that he was unable to see himself as a separate object and was unable to examine himself objectively. “He will require constant supportive treatment to give him some continuity with an object between visits and to preserve his ‘façade’ that he is examining himself objectively.” His GAF score was 53 at the start of treatment and 45 at the end; the PFS average score was 52.6 at the start and 48.8 at the end.

What happened in a treatment in which the patient appeared to show such a promising prognosis at intake, and then deteriorated so severely during the subsequent years?

We first attempted to see if the language measures applied to the treatment notes could point to where the treatment began to show signs of being ineffective or even having a negative effect. Figure 1 shows a comparison of Mean High WRAD (MHW) and Reflection (REF) for the treatment as a whole, divided into natural segments covering the two years. In the beginning of the treatment, in the fall of 1951, the Reflection line is slightly above the MHW line, yielding an MHW/REF difference score of –.005, indicating some degree of distancing of the patient’s material. The difference score diminishes during the next two segments, reaching –.001 in spring of 1952. Thereafter, however, the MHW declines and the REF increases, in an almost mirror image of one another, and the differences score increases and remains high during the remainder of the treatment.

Figure 1.

Language scan of treatment 50T1

Figure 2.

Language scan of treatment 70T3

This “reading” of the graph can bring us to examine particular points in the treatment that might reveal the difficulties that emerged. Looking first at the initial session, in which the figure shows a mild distancing process, we see that the treating analyst’s notes already differ from the tone of the intake. One can speculate on the reason for this:

The patient is an aggressive, outspoken, neat, emphatic speaker. He begins telling about his potency disturbance, which has been present for the past three or four years. He has been going with a girl named G. for the past seven or eight years. She has been chasing him and then dropped him once. The potency disturbance appeared. At this point he breaks off and tells me he is very upset by my youth. He did not expect to see someone so young, he was prepared to see somebody who was well established, and well advanced in the field and in years, who would be taking care of his problem. It is such a serious problem to him he wants to be sure he has the best help available. Without waiting for any comments from me, he begins to wind himself up, talking more quickly and progressively louder until he is just about shouting. The content concerns my apparent lack of experience. Is the psychoanalytic institute any good? Is psychoanalysis any good, and so on?

The notes indicate a dramatic difference between this patient’s presentation to the evaluators and his presentation to the analyst. The analyst did not address the anger, either directly or possibly as in some way defensive (perhaps a more modern construction), nor did he address the patient’s concerns about his youth and inexperience and the patient’s need for the best help since his problems were so “serious.”

Throughout the analysis, from the very beginning, there were discussions about a variety of sexual issues: his potency problem, his masturbation fantasies, and dreams with a great deal of sexual content. A significant historical event that came up many times was witnessing a sexual scene when in his twenties, which both excited and frightened him. From the beginning, as in the first session and in the second session described below, the analyst seemed not to respond to the patient’s communication and instead seemed to have his own agenda. In the second session, according to the analyst’s notes, the patient

has been thinking all about this. He did the same thing with Dr. Z. and kept it up from time to time with all kinds of demands that Dr. Z. answer questions that he had already answered before. However, he does know what this means. He feels that only a part of it is related to what he sees as my youth. He has a disdain for younger people and he gives the example of “a little boy” of twenty-one in a bar he was in. He was giving his opinion of the slaying of the man by two teenage boys who stomped him to death. The patient was completely unable to listen to the boy and felt it necessary to disregard him and to belittle what he was saying. Somehow he feels that this kind of situation is related to the strong father who dominates him.

The analyst reports that he then responded, “I tell him that I would like some more data about his symptoms and he then proceeds to tell me about the potency disturbance.”

In other words, the patient clearly continued to describe his angry feelings: during the first session he expressed anger at the analyst; in the second he described his anger toward the previous analyst and his general attitude to younger people, including the example of his feelings toward the “little boy” of twenty-one. The patient makes a genetic connection to his domineering father, saying that it is almost as if he dominates younger people the way his father dominated him. The analyst changes the topic.

We then come to the period, in the spring of 1952, when, as seen in Figure 1, the REF and the MHW are virtually identical, indicating that the analyst is writing in a more emotionally engaged manner than in first two segments (fall 1951 and winter 1952). It seems possible that the treatment may take a different turn at this point; in spring 1952 it seems as if the MHW will cross over and become greater than the MR. However, the MHW declines in subsequent phases of the treatment, reaching its nadir in fall 1952, after the summer break, while the REF score increases in a mirroring pattern, indicating that the greater emotional connection that seemed in reach in spring 1952 was not achieved.

We attempted to understand reasons for the precipitous drop in MHW, along with a sharp increase in REF, between summer and fall 1952. To this end we examined the clinical material beginning in the spring through the fall of 1952 to try to ascertain the nature of the analytic work reported by the analyst. Since the analyst’s notes are extensive, a detailed clinical analysis is possible. It is probably not coincidental that this nodal sequence (during spring and summer 1952) occurs as the analytic work is approaching the first summer hiatus.

Two Clinical Vignettes from Case 50T1

Spring 1952

A sequence in the middle of May illustrates how the clinical discussion confirms the linguistic measure of increasing MHW in the analyst’s emotional connection to the patient, as reported in the notes. The analyst communicates that he understands the patient’s affective state and his conflicts about his close feelings toward the analyst.

At the beginning of the week, the patient described interactions with a woman with whom he had made several unsuccessful attempts at intercourse. This woman has been in treatment with an analyst the patient himself had once consulted. In his associations the patient discussed a variety of issues referring to anal stimulation and fears of homosexuality. When the analyst touched on these homosexual references, the patient became defensive yet continued with associations from his past, including a dream of walking into a bathroom and seeing a gray-haired woman on the toilet who reminded him of his mother.

The patient came to the session the next day and said he felt good facing a lot of these unpleasant associations, perhaps because the analyst was “forcing” him to look at himself. After greater elaboration of a variety of sexual fantasies and themes of being forced, the analyst noted to the patient that comments about being forced coincide with his being more open in the sessions; he was acting as if he was not responsible for communicating these fantasies, as if the analyst was forcing him.

The patient reported the next day that he was troubled about the previous session and that even when he is talking freely, there are thoughts he is withholding. The patient thought that his speech is a cover for things he doesn’t want to talk about. The analyst touched on the transference situation by asking whether that withholding was going on at the moment. The patient agreed and reported a dream with a variety of sexual scenes taking place on a trip he was planning to take. This association about a trip, along with other associations, led the analyst to “wonder if [the patient] isn’t concerned about the vacation.” The next day, the patient reported that he doesn’t like to think he would miss the analyst or would need him. The patient associated to his brother and made a negative comparison about the size of his own penis. This led the analyst to say, “So if you rely on your brother, it means that you are less than him. Therefore you must make your brother less than you. Otherwise you will feel you have a small penis.”

In this sequence the analyst communicates an important thread: that the patient is concerned about the vacation. The analyst made several references to transference manifestations: the patient being more open in the sessions and then feeling forced, asking whether the patient was withholding at that moment in the session, and connecting the patient’s associations to the upcoming summer vacation.

In this material, the analyst certainly does communicate his understanding that the reason the patient cannot feel reliance on other men is that he would feel less manly. To avoid feeling dependent on or connected to other men, he needed to build himself up by making them inferior to him. The dynamic that is explicitly discussed in these sessions is reminiscent of the transference interaction in the first session and the connection to Dr. Z. and the young man in the second session. The patient’s problems with men and his defensiveness against feeling close to them, implicit in the first sessions, were made explicit in this material: the patient wanted to avoid feeling dependent on the analyst in order to protect himself from feeling that his was the smaller penis.

The significance of this piece of analytic work was confirmed the next day when the patient, in the context of his concern about the size and presence/absence of his penis, revealed for the first time that he doesn’t touch his penis when he urinates. This led to an important genetic root in the patient’s early adolescence, his suffering an injury that made him very anxious.

Given our contemporary interest in understanding and addressing the transference most explicitly, it is striking to read the analyst’s notes as he continues to write: “These injuries are not associated with real traumas and I am puzzled about them. He digresses to talk about how much more comfortable he is with me.” The use of the word “digresses” is notable. Today we would not consider a patient’s focus on the transference a digression. Whether this view of transference as resistance to exploration of genetic material was idiosyncratic to this analyst and his supervisor or expressed a more pervasive theoretical stance is, of course, impossible to tell.

The analytic week ended with the patient bringing up Miss E. (another patient in the clinic, with whom he has had a variety of flirtatious interactions and who comes up again just before the summer vacation). The patient, according to the analyst, said “he is waiting for me to tell him why he uses her. He wants the answers from me. He knows I won’t let him lean on me.”

In this analytic work in May, despite some limitations, the analyst communicated that he understood the patient’s dynamics and how they were being played out in the transference. As indicated by Figure 1, this coincides with increase in the MHW/REF difference score, which approaches the positive zone for the first and only time in the treatment.

Summer 1952

This gain is lost in the next segment. The description of the analytic work in July, just before the summer break, indicates that the analyst did not continue to communicate to the patient that he understood how upset the patient was about the summer break and that the patient had to deny his closeness and dependence on the analyst. The analyst focused on the patient’s aggression without communicating to him that it would be valuable to understand the meaning of the aggression, including a possibly defensive stance against missing feelings stimulated by the impending summer break.

In the beginning of July the patient reported he had been

upset all weekend. Everyone was away and he was alone. This is always disturbing to him. He was thinking it didn’t pay to come back here for the next two weeks and had a fantasy that he would stop and that I would call him and that he wouldn’t come anyway.

The analyst does not report how he responded to these comments; however, in sessions shortly after this communication he describeshow he focused on the patient’s “provocative remarks,” which, he writes, were

intended to get an angry response from me. I point this out to him, and he agrees but much too quickly. It as though he wants to have me fit into his picture of me as the cutting person, but this also frightens him. He talks about J.’s vagina being hidden, that he can’t see it, and I suggest that it is as though she has a penis but he can’t see it.

There is no reference to the upcoming summer vacation. By the middle of July, the analyst reports that the patient

is more aware of his need to provoke me, to get me to speak to reassure himself; I think actually to attack him anally. He had a dream Sunday night after wandering around downtown and denied to himself that he had any homosexual urges. In the dream he was with an unknown man and he gets the man to agree to lay for him. Then he cannot find the vagina and looks for the rectum. Just as the patient is about to insert his penis there, he wakes up. On awakening, he has a sudden feeling that something is about to enter his rectum, and he became very panicky.

The patient continued obsessively about sexual issues, and the analyst states:

I utilize this to stress the anal material that has come out and his preoccupation with his rectum and the pleasurable experiences he has had with it. I am able once again to stress his provocative behavior with me lately. Last night he took Miss E. up to his apartment for the first time. All during the evening, although he did not have any intercourse with her, his thoughts were about what I would say if I hear about it. I point this out as acting out of his need to provoke me. His fear of my reaction is his fear of his own aggressive wishes toward me—the acting out is also a denial of his passive wishes toward me.

On the last day before the summer vacation, the analyst writes that the patient

spends the whole hour in praising me—how good he realizes I really am, how much he has learned, how he has learned to be able to face his aggression and so forth. He is even willing to see his passivity with Miss E. now. It is her responsibility, not his. He relates it to his impotence in withdrawal from danger situations. Even during intercourse the woman has to put his penis in the vagina. In this way she is doing it, not him. It is her responsibility. I think this also refers to a desperate attempt on his part to say that I am leaving on my own, that he isn’t killing me off at this vacation.

Several themes in these sessions stand out, particularly the analyst’s lack of addressing the connection between the patient’s associations and feelings and their possible connection to the vacation. It is striking to note that the analyst was able to do that in May but couldn’t (or at least did not report it) just prior to the vacation.

Certainly homosexual feelings are on the patient’s mind, and one can only conjecture the connection between those feelings and the upcoming break from the analyst. What is most apparent is that the analyst focuses on the sexual content and the patient’s aggression. He discusses the patient’s anal pleasures and focuses on his “provocative behavior.”

He does not report that the provocative behavior may have meaning, perhaps retaliation against the analyst for leaving or a defense against the patient’s attachment, love, and feelings of missing someone. The homosexual dream with the “unknown man” may well represent fantasies about the analyst, yet that is not reported in the material. The patient’s complex feelings about the analyst—the closeness, the sexual fantasies, the retaliatory and/or defensive aggression—were not addressed analytically. Little wonder the patient took another patient to his room. All the analyst could communicate to the patient was that this action was “provocative” to the analyst, with no understanding of the complex meaning of this action. On the day before the vacation, the patient felt almost coerced to agree with the analyst: he felt he had to please him, echoing the analyst’s formulations back to him, before they separate for the summer.

The analyst’s lack of appreciation of the patient’s real closeness to him continued in September (fall 52a), when the MHW reaches its nadir. In the first week of that month, the analyst reports how the patient stressed he did not need the analyst over the summer. The analyst writes: “He recognizes that he protested too much. ‘You are afraid of your need to submit which is the only way you feel you can get your father to love you.’” The analyst focuses on the word “submit” rather than love or connection, and on the father rather than himself.

In short, as in the first session of the analysis and unlike the May sessions, in July the analyst focuses simply on the patient’s aggression without noting the attachment, love, and sexual feelings he has toward the analyst, or his defenses against those feelings. This lack of focus on feelings of missing the analyst continues in September.

The clinical comparison between May and July through September is mirrored in the linguistic measures. The MHW is rising in May and then declines in the summer toward its nadir when the analytic pair return from vacation in the fall.

It is significant that the understanding the analyst was able to communicate in spring 1952 was temporary. Linguistically, the MHW stayed down until spring 1953, when the analyst told the patient that analysis should not continue and terminated the treatment at the end of the second year.

Clearly the patient was correct in his assessment that the analyst would not let the patient “lean on him.” In a contemporary formulation, it would be conjectured that the analyst could directly address neither the homosexual transference nor the close bond the patient felt for him, and that this resulted in termination of the analysis.

This conjecture is confirmed in April 1953, when the analyst writes that the patient “wishes that the analyst do something” and “wants to fight instead of analyzing.” In the midst of attacking the analyst, the patient suddenly brings in his thoughts of being a woman. The analyst concludes a session by telling the patient that “homosexual thoughts seem to be so much on his mind and so frightening, that we should examine them. They come up often but he seems to run away from them.” From our perspective today, it seems this is the first time in the analysis that the analyst actively addresses the patient’s defenses against a homosexual transference. However, immediately after this session, the analyst writes that “in supervision it was decided to continue the analysis no longer than the summer because the patient seemed unable to tolerate it. Meanwhile it was suggested that I continue to abstain from his provocations to talk.”

Can one safely conjecture that in supervision, given the tenor of the times, there was a discussion about homosexuality that led to a decision that the analysis be terminated? Certainly this sequence illustrates Kantrowitz’s idea (2002) that the supervisor has an effect on the analyst-patient dyad.

In short, the patient expressed feelings of connection to the analyst. The analyst reinforced the patient’s fantasies that connectedness is a sign of weakness and castration. The patient became defensive and fought with the analyst to counter his feelings of weakness. When the analyst did not seem to understand that he needed to communicate in July, as he did in May, that the patient needed and wanted him, the patient displaced those needs to Miss E. The analyst in essence did not communicate that he understood either the patient’s needs or his defenses against them.

We suggest that the availability of language scan graphs at the time might have communicated to both therapist and supervisor that the treatment was working well in May (when the analyst reported the material in an emotionally engaged way) but entered a troubled area in the summer and fall of 1952, when the analyst reported the material in a distant, less emotionally engaged way. Awareness of the linguistic measures might have allowed recovery and progression of the analysis.

Clinical Description of Patient 70T3

Case 70T3 provides a striking contrast to 50T1, in terms of initial presentation, initial assessment, and course of treatment. The treatment, evaluated as successful at the time, showed increases in both the GAF and PFS aggregate scores, ending well above the clinical criteria scores on both measures.

The patient was a married professional woman in her early thirties, with two children. She had had recurrent depressions that at times would become severe. She was unhappy and dissatisfied with her life but wasn’t sure why. The summer before the analysis was to begin, she had found herself on vacation at home, unable to leave the house for approximately three weeks. In contrast to the uniformly positive initial evaluations of patient 50T1, the senior interviewers had mixed feelings about this patient. One evaluator considered it “difficult to evaluate the severity or modifiability of the patient’s pathology. She did, however, seem like an intelligent, motivated person who might be able to tolerate an analytic therapy, without too severe disorganization and with considerable benefit.” A second evaluator found

an acting-out tendency which might cause concern that the patient would leave analysis if her depression got either much better or much worse. Yet the patient herself fears the acting out, and my guess would be that a stable transference would develop and function much like the relationship to her husband. As to prognosis, analysis may not be able entirely to resolve the patient’s chronic depressiveness, but there is reason to hope that the severity of the depression might be greatly mitigated with analytic insights. Because of this as well as pedagogic interest I would recommend the patient for acceptance as a difficult second or third case.

The student analyst was concerned about the amount of acting out the patient had displayed in the past, in terms of both promiscuity and homosexual behavior. There seemed to be a possibility that this propensity to action might take the patient out of analysis before she had a chance to examine her feelings. The supervising analyst was less concerned about this and took as a good prognostic sign the fact the patient had been able to make a good marriage and had a trusting relationship with her husband. An especially good prognostic sign, he thought, was that she had maintained a relationship with her parents despite her feelings of being constantly rejected by them. One may speculate about the positive impact of this supervisor’s opinion on the course of the analysis, in contrast to the negative impact of the supervisor in case 50T1 (see Kantrowitz 2002.)

This patient’s diagnosis was depression. Regarding analyzability, her prognosis was good, though there were questions about it; she appears to have been given a less positive prognosis than patient 50T1.

At termination, the patient remembered how angry and fearful she had been at the beginning of the analysis. She said she was pleased to have been able to work with the analyst; she said she had other things to say that she couldn’t say directly to him, but that she hoped at some point to be able to put them in writing. She was able to talk in a nuanced way about many issues in her family relationships that worried her, and about her concerns regarding her ability to manage on her own. She was pleased with a new job that would begin in the fall, and excited about preparations for a family celebration.

Utility of Language Scans in Trying to Understand the Progression of This Case

In parallel to the discussion of case 50T1, we examined the scan of the language measures to try to understand the achievement of a positive outcome in this much longer treatment (5 years, 9 months, compared to 22 months for 50T1).

Looking at the language scan, the beginning of the analysis was marked by high MHW and a positive difference score, which then declined during the next two phases of the treatment, dropping to a first low point (and negative difference score) from October 1972 through March 1973.

Clinical Evaluation Directed by Language Scans and Comparison with Failed Case

In this discussion, we illustrate the nature of the analytic work during the second year, comparing a segment from the beginning of the second year, when MHW is low (October 1972 through March 1973), and the analysis is in danger of following a course similar to that of 50T1, with the following segment (April through September 1973), when MHW returns to a higher level than REF. MHW then remains elevated for three years until discussion of termination is begun. At that point there is another decline in MHW and an increase in REF, leading to the second negative difference score of this treatment (represented as a crossover of the language measures. Here, as following the first such pattern, there is recovery of the treatment, allowing for successful termination.

We will focus on the first nodal period with a negative difference score (October 1972–March 1973) that was then repaired in the subsequent phase. We note that this crossover period occurred at a comparable time early in the analysis as the period in case 50T1 (spring 1952), where MHW and REF converged and there appeared to be some potential for repair of the treatment.

In contrast to patient 50T1, who was aggressively assertive, this patient presented at the beginning of treatment as inhibited and anxious. The analyst noted that because she was frightened to use the couch she jumped and lay down immediately; she created a ritual way of preparing herself every time to lie down and preparing herself to leave at the end of the session. She often complained of the difficult trip to the analyst’s office.

A clinical vignette in the beginning of the analysis is an example of the engagement of the therapist, as indicated by the vividness of the language.

In the analysis, the feeling mounted that I didn’t really care about her, especially after times of absence. Finally she went into a rage and insisted on sitting up for a session in which she told me just how angry she was at me. In this session she announced that the only way she could be sure that I cared about her is for me to be available at all times when she needed me. She was enraged that she had to ring the bell five times before I had answered it, that she had the feeling that “you didn’t pick me, you’re stuck with me from the institute.” Again, she had the feeling that she was upset either because it was her fault or my fault or her mother’s fault. Next session she felt embarrassed about her behavior. She could reflect that her anger at the analyst was excessive and she didn’t know why, marking the beginning of the analysis of the transference.

She went on to describe her chronic fears whenever she was confronted with her children or husband not being immediately available to her. She would often be fearful that something would happen to her children on their way to school or if they were late. When she left to go to an analytic session leaving her husband and children together in the house, she was concerned that something would happen while she was gone. She felt that her analysis might be “bad for them,” i.e., take something away from them.

Following this initial period of analytic work indicated by high MHW, the MHW declines below the REF. In the segment of October 1972 to March 1973, the patient discussed a variety of traumatic and potentially traumatic events in her life and in her family history: the severe illness of a friend’s child and thoughts about close relatives who had suffered traumas. During this time she wanted to feel closer to the analyst for protection, just as she felt protective of her friend whose child was ill, and became obsessed with images from stories she heard about the traumatic events suffered by her relatives. She reported how frightened she would become and how she would feel guilty for being alive, unlike those children who had died.

The analyst reported that over this time “it was very important for the patient to feel that she was doing the analysis herself and would often feel that the analyst’s interpretations were attempts to prove that she couldn’t handle herself alone or that the analyst knew more than she did. The patient reacted to separations from the analysis by insisting she did not need to be in analysis anymore. This may be compared to the comment by patient 50T1 that he should not come for the last few weeks before the analyst’s vacation.

Toward the end of this period the patient, reporting a dream of a dilapidated house where everything was falling apart, defiantly asked the analyst what it meant. When the analyst suggested she talk about whatever ideas she had about it she refused to do so, saying that dreams didn’t mean anything.

During this segment, in the face of disturbing associations from her current life and memories about her family, it seemed as if the analyst could say nothing helpful. “Over this time,” the analyst wrote, “it was very important for Mrs. X. to have the feeling that she was doing the analysis herself. She would often feel the analyst’s interpretations as an attempt to prove that she couldn’t handle herself alone or that the analyst knew more than she did.”

Both from a clinical reading of the analyst’s summary and looking at the language style, it seemed as if the connection to the analyst and the analysis was in danger, with many unexamined negative feeling states toward the analyst. The analyst did not report any associations that she herself might have had about the dream of the dilapidated house. With all the negativism present during this time, we might speculate that the dream image may in part represent the danger of the analysis falling apart, as the patient threatened to interrupt the analysis: she really didn’t need to be in analysis; she could do it herself.

During the next six months (April through September 1973), negative feeling states continued to pervade the notes. The major themes included the patient’s envy of other women, envy of the analyst, and competitive feelings in her professional life, together with a conviction that professional criticisms were personal attacks. At the same time, she castigated herself for her envy and depression and felt that the critiques of her work were warranted because she lacked sufficient credentials.

At one point the theme shifted to children, and she talked about her friends having babies. She was concerned that her husband would lose sexual interest in her if she did not have a third child. The analyst wrote the following:

Typical of her defensive character style was that in response to this she again became extremely angry and stated that “women just have third children to get away from themselves.” When the analyst pointed out that if she were so angry she must be upset about something, it increased her anger and defensiveness. What finally emerged was that she, indeed, did have wishes to have another child and was fearful that if she allowed herself to be aware of those wishes that she would have to do it. This was the beginning of her understanding that she could have mixed feelings about things and still make a choice; that she didn’t have to bury one of the feelings under the anger and defiance.

This conceptualization and interpretation by the analyst may be compared to the conceptualizations and communications by the analyst in 50T1 at a similar phase in the treatment. In contrast to that case, here the analyst directly addresses the defensive nature of the patient’s anger: “If you are so angry, you must be upset about something.” This line of intervention let the patient understand that one can have mixed feelings.

At a point before the summer vacation, patient 50T1 discussed his upset feelings about being abandoned and his wish to retaliate against the analyst. In contrast to the analyst in case 70T3, who focused on mixed feelings and the defensive use of anger, the analyst in case 50T1 focused solely on the patient’s provocation, “intended to get an angry response” from him, and avoided communicating the defensive nature of the anger, even after the summer break.

As the summer break of 1973 approached, patient 70T3 continued to discuss her fear of having good things for herself, or even to wish for positive things, like having a baby, because of her fear that they would be taken away.⁵ Her concern about missing the analyst was masked by anxiety around the possibility of her own pregnancy (which did not occur). The analyst noted that

upon her return in the fall, she spent a good deal of time longing for the “freedom” of the summer and stated that she was sad because she was back in the city. She talked about how she might want to move and how she certainly was not going to allow the analysis to make her stay. The analyst reminded her that this was similar to her response last summer when she had come back after a summer in which she clearly missed the analyst, saying that she was thinking that perhaps now she should leave treatment. Again, she was reminded of her difficulty in experiencing her warmer feelings about the analyst. She laughed and the feelings subsided.

We may interpret that the analyst here is in effect asking her to stay; this can be contrasted with the shift that occurred in the process for patient 50T1. As with that process, the major focus in September 1973 continued to be on the patient’s denial of needing the analyst over the summer. However, the analyst in case 70T3 was able to communicate an understanding of the patient’s need for closeness, in contrast to the analyst in case 50T1, who focused primarily on the patient’s fear of “submitting” and on his defensive aggression.

Throughout the analysis, particularly during the nodal points, the analyst in case 70T3 continued to interpret the patient’s mixed emotions. Later on, toward termination, during the second nodal period, for example, the patient had a dream in which she was leaving for college and her mother asked her to take a later train. “It was interpreted to her that she had mixed feelings about both her mother and the analyst asking her to stay; she wanted them to ask her to stay but protected herself against hurt feelings by insisting she wanted to leave.” The patient responded that “she was not aware of a wish on her part for the analyst to ask her to stay and then went on to describe how she felt that there was no reason for the analyst to care whether she was leaving or staying. She said that she understood that she was in Treatment Center analysis and that the analyst was getting an education out of her analysis. Again, it was interpreted how much she had to ward off her own feelings of wishing that the analyst care about her as she had the concern that her mother didn’t love her.”

Clinical Conclusion

The differences between the two analyses are quite dramatic, as indicated by the outcome measures and language style analysis presented in the figures and discussed in the text. At similar points of threatened rupture in the treatment, the analyst in case 50T1 does not seem to appreciate the patient’s connection to him, does not address conflict or ambivalence (wish to leave / wish to stay; aggression/love), and does not directly address defenses against wishes for closeness in the transference. After a brief period of emotional connection with the patient, the analyst could not continue the connection.

By contrast, the analyst in case 70T3 understands and interprets that the patient wants to be asked to stay but protects herself from hurt by threatening to leave.

In sum, it seems the analyst in case 50T1 did not like (or was afraid of) the patient’s connection to him (perhaps particularly the homosexual transference), whereas the analyst in case 70T3 appears to have appreciated the patient’s connection to her. Analyst 50T1 would not or could not understand this central dynamic and communicate it to the patient, whereas analyst 70T3 was able both to understand the patient’s central dynamic and to talk to her about it. Finally, analyst 50T1 focused on content interpretations, whereas analyst 70T3 focused on the process, particularly the transference relationship. The analysis of patient 50T1 was interrupted by the analyst shortly after a year and a half of work. The analysis of patient 70T3 continued for six years.

Discussion

It is striking that the computerized language measures, based on two interacting variables, have pointed to corresponding moments in two contrasting treatments with starkly different outcomes. Clinical examination of these phases has suggested how the analyst’s activity may have affected the course of the analysis. The linguistic analysis allows for a microanalysis of clinical material highlighting points that might be missed in the customary clinical review. Use of such a method might have helped in the supervision of case 50T1 and in the conduct of this treatment of a patient who was initially seen as having an excellent prognosis but then deteriorated.

The utility of this method, of course, needs to be verified by further studies. If computerized linguistic analysis of process notes is done prospectively, this technique has the potential to provide analysts and therapists, and their supervisors, opportunities to refine interventions with patients. For example, had linguistic scans been available in the case of 50T1, the analyst and supervisor could have seen that something was going wrong in the sessions before the first summer break, something not clinically discerned. Analyst and supervisor may then have become more aware that the response of the patient to the separation was not being dealt with effectively.

Limitations and Future Directions

An important question that needs to be addressed is whether the finding in this exploratory study of two sample cases (a male analyst treating a male patient and a female analyst treating a female patient) can be replicated with other cases and other gender configurations. Can differences in the linguistic pattern of case write-ups reliably differentiate successful and unsuccessful analytic work and successful and less successful interventions? Our clinical analysis of the therapeutic work during nodal periods indicated that more effective clinical work occurred concurrently with shifts in the language curve. A more fine-grained analysis of the linguistic measures (e.g., a time series analysis) would be required to determine whether the clinical effectiveness followed, was concurrent with, or preceded those shifts.

Can our language measures provide a way to study the construct of analyzability and its problematic predictive capacity (Bachrach 1983), help us address the problem of patient dropouts (Hamilton, Wininger, and Roose 2009), and help us understand the effectiveness of the supervisor-analyst-patient triad (Kantrowitz 2002)?

Two cases that were dramatically different from each other—in terms of their language scans, their initial GAFs (but not their initial average PFS), and their clinical outcomes—were chosen for illustrative purposes. This choice of two cases, very different from one another and treated during different psychoanalytic eras, may confound the conclusion that the language measures can reliably differentiate success from failure. In future work the method would need to be tested on a large sample of different types of treatment. The clinical notes made available to us from the archives of the Treatment Center varied in both length and level of detail. Preliminary within-case comparisons did not show differences in the language measures among different kinds of notes (daily notes, weekly notes, six-month summaries). However, none of the material was in the form of process notes and did not include extensive descriptions of “he said, I said . . .” back-and-forth. It would be instructive to examine whether linguistic analysis of detailed process material yields results similar to or different from those drawn from more general case descriptions.

Conclusions and Clinical and Professional Implications

The clinical evaluations of the treatments used in this study, provisional as they are, point strongly to the need for a monitoring of treatment effects beyond the usual supervisory practices. The fourteen treatments included in this study were carried out by candidates in a highly selective psychoanalytic training institute, and were supervised by senior analysts. Yet seven of these patients showed essentially no improvement based on the Composite Clinical Evaluation (with a CCE of 0 or 1, Table 1), as reported in our initial study, and four in fact became worse. While it is possible that other clinical ratings might have provided a different picture, the proportion of success and failure is not so different from that reported in other studies of comparable treatments. As reported by Bucci, Maskit, and Hoffman (2012), six of the seventeen cases included in an archival study by Luborsky et al. (2001) showed little or no improvement; of the twenty-two psychoanalytic cases studied in the Menninger Foundation Psychotherapy Research Project, nine were considered equivocal or failed (Wallerstein 1986, p. 74).

Major clinical difficulties include problems in assessment and in predictions of analyzability (Bachrach 1983), problems with patient dropouts (Hamilton, Wininger, and Roose 2009), and the long-standing problem of therapeutic failure (Obendorf 1948; Gold and Stricker 2011). Clinical evaluations often are difficult and are often wrong in their prognoses, as seen in the cases reported here: one was considered an eminently analyzable patient, with many laudatory comments by evaluators, yet had a failed analysis within two years. In contrast, the evaluators of the other patient had mixed feelings about her capacity to undertake psychoanalysis and come to a satisfactory conclusion, yet she turned out to have a successful analysis. In addition, in training cases the nature of the supervisory experience, as well as the analyst-patient match, may play an important, yet often unexamined, role in the progress of the treatment, as Kantrowitz (2002) has pointed out.

Obendorf’s list of the types of failures with psychoanalysis (1948) is still relevant to contemporary practice. He identified four classes of failures. One group, which he found particularly challenging, comprised cases in which the patient’s personality and symptoms seemed to justify the application of classical psychoanalysis, yet the treatment ended with unsatisfactory results. The other classes of failures included cases where mild symptoms masked deeper pathology, where there were difficulties in applying the method, and where external conditions posed difficulties. Similar problems are seen in studies of other forms of psychotherapy; professional responsibility in the psychodynamic field, as in all treatments, requires that we address the fact that treatment fails many who seek our help.

Outcome research is not sufficient to address these problems; process research is required to address the nuances of the treatment interaction in a timely fashion. Optimally, for research purposes, the computerized analysis of treatment notes that has been illustrated in this paper should be embedded in a study including external evaluation of treatment effectiveness, ideally through measures administered prospectively, at intervals in the course of treatment, at its conclusion, and at follow-up. In such a “wish-list” study, sessions should be recorded and transcribed, and computerized measures applied to the transcripts. Such a study would enable us to address a range of questions concerning the relative value of language style analysis of notes and of transcript material in predicting outcome. We could also address the relation of notes and transcripts, asking what is covered in one that might be missing in the other.

The value of notes is that they are widely available; they are used in many or most supervised treatments and in treatments by practicing therapists and analysts at all levels of seniority. These days analysts and therapists can record their session notes in a password-protected, HIPAA-compliant manner on a computer, often close to the time of the session itself; by contrast, the transcription of audiotapes generally involves considerable cost and lengthy delays. The real-time digital processing of notes would in principle allow the language style analysis to contribute to the supervision, or to guide the practitioner’s recognition of a difficulty in the treatment as the situation is evolving rather than long after the fact. The method proposed here has the potential to help analysts understand where an analysis is going or has gone right, and where it is going or has gone wrong. With this understanding, we can modify our analytic technique in an ongoing way to fit a particular patient’s needs at a particular time.

Footnotes

Acknowledgements

The authors thank the New York Psychoanalytic Foundation for seed funds, the Fund for Psychoanalytic Research of the American Psychoanalytic Association for a major grant, and the Treatment Center of the New York Psychoanalytic Society and Institute, and its director, Peter Dunn, for their generosity and cooperation.

Leon Hoffman, Director, Pacella Parent Child Center, New York Psychoanalytic Society and Institute; Training and Supervising Analyst, New York Psychoanalytic Institute. Jane Algus, Clinical Assistant Professor, NYU Medical Center. Will Braun, Assistant Attending, Department of Psychiatry, Lenox Hill Hospital. Wilma Bucci, Professor Emerita, Derner Institute of Advanced Psychological Studies, Adelphi University; Director of Research, Pacella Parent Child Center, New York Psychoanalytic Society and Institute. Bernard Maskit, Professor Emeritus, Mathematics Department, Stony Brook University.

1.

As noted in Bucci, Maskit, and Hoffman (2012), “The initial design of this study called for selection of ten cases from these files, including one successful and one unsuccessful case from each of the five decades of the last half of the 20th century, based on outcome evaluations as recorded in the files at the time of treatment. The Treatment Center Director and staff, not connected with this project, first randomly selected one case from each decade, and provided notation as to outcome evaluation based on the information contained in the file. Five additional cases were then selected randomly, one from each decade, with the goal of providing a pair of cases with contrasting outcomes from each decade. If this was not achieved for any decade, the second case was returned to the file and another case from that decade randomly chosen. This process was repeated by Treatment Center staff until one successful and one unsuccessful case had been chosen from each decade based on the contemporaneous evaluations in the files” (pp. 316–317).

2.

The DAAP computer program, constructed by Maskit (), is a modern text analysis system designed to track the ebb and flow of psychological or linguistic variables. It does this without requiring or imposing segmentation.

3.

The mean WRAD is analogous to a hiker’s average speed for an entire trip (sometimes walking and sometimes running); the MHW is analogous to the hiker’s average speed while running, ignoring those times when he/she is walking.

4.

To ensure confidentiality, facts about these patients and the dates of their treatment have been changed (the dates remain within the same decade in which the treatment was conducted).

5.

The segmentation of this long treatment given by the six-month summaries left the periods before and after summer vacations within the six-month periods, so that the dynamics surrounding those phases are not reflected in the figures. A different segmentation structure would allow such an examination.

References

Andrei

Bucci

Keable

Maskit

Hoffman

(2008). Trajectory of a child analysis as seen through treatment notes. Poster presented at the international meeting of the Society for Psychotherapy Research, Barcelona.

Bachrach

H.M.

(1983). On the concept of analyzability. Psychoanalytic Quarterly 52:180–203.

Bucci

(1997). Psychoanalysis and Cognitive Science: A Multiple Code Theory. NewYork: Guilford Press.

Bucci

Maskit

(2006). A weighted dictionary for Referential Activity. In Computing Attitude and Affect in Text, ed. Shanahan

J.G.

Wiebe

Dordrecht, The Netherlands: Springer, pp. 49–60.

Bucci

Maskit

(2007). Beneath the surface of the therapeutic interaction: The psychoanalytic method in modern dress. Journal of the American Psychoanalytic Association 55:1355–1397.

Bucci

Maskit

Hoffman

(2012). Objective measures of subjective experience: The use of therapist notes in process-outcome research. Psychodynamic Psychiatry 40:303–340.

Colombo

Michels

(2007). Can (should) case reports be written for research use? Psychoanalytic Inquiry 27:640–649.

Crisafulli

Bucci

Hart

Hoffman

Kolod

Newirth

(2011). Multiple perspectives on the treatment process in a nine-session treatment. Panel, American Psychoanalytic Association, New York, January.

Damasio

(2003). Looking for Spinoza: Joy, Sorrow, and the Feeling Brain. Orlando: Harcourt.

10.

Ellsworth

P.C.

Tong

E.M.W.

(2006). What does it mean to be angry at yourself? Categories, appraisals, and the problem of language. Emotion 6:572–586.

11.

Foa

E.B.

Meadows

E.A.

(1997). Psychosocial treatments for posttraumatic stress disorder: A critical review. Annual Review of Psychology 48:449–480.

12.

Freedman

Hurvich

Ward

Geller

Hoffenberg

(2011). Another Kind of Evidence: Studies on Internalization, Annihilation Anxiety, and Progressive Symbolization in the Psychoanalytic Process. CIPS Series on the Boundaries of Psychoanalysis. London: Karnac Books.

13.

Glick

R.A.

(2008). Writing as psychoanalytic pedagogy: A primer (the Columbia Psychoanalytic Writing Program). Journal of the American Psychoanalytic Association 56:1261–1277.

14.

Gold

Stricker

(2011). Failures in psychodynamic psychotherapy. Journal of Clinical Psychology 67:1096–1105.

15.

Hamilton

Wininger

Roose

S.P.

(2009). Dropout rate of training cases: Who and when. Journal of the American Psychoanalytic Association 57:695–702.

16.

Harber

K.D.

Pennebaker

J.W.

(1992). Overcoming traumatic memories. In The Handbook of Emotion and Memory: Research and Theory, ed. Christianson

S.-A.

Hillsdale, NJ: Erlbaum, pp. 359–387.

17.

Høglend

Amlo

Marble

Bøgwald

K.-P.

Sørbye

Ø.

Sjaastad

M.C.

Heyerdahl

(2006). Analysis of the patient-therapist relationship in dynamic psychotherapy: An experimental study of transference interpretations. American Journal of Psychiatry 163: 1739–1746.

18.

Høglend

Bøgwald

K.-P.

Amlo

Heyerdahl

Sørbye

Ø.

Marble

Sjaastad

M.C.

Bentsen

(2000). Assessment of change in dynamic psychotherapy. Journal of Psychotherapy Practice & Research 9:190–199.

19.

Kächele

Schachter

Thomä

(2009). From Psychoanalytic Narrative to Empirical Single Case Research: Implications for Psychoanalytic Practice. New York: Routledge.

20.

Kantrowitz

J.L.

(2002). The triadic match. Journal of the American Psychoanalytic Association 50:939–968.

21.

Kingsley

(2009). The clinical validation of measures of the Referential Process. Dissertation, Adelphi University, Derner Institute of Advanced Psychological Studies.

22.

Levy

R.A.

Ablon

J.S.

, eds. (2010). Handbook of Evidence-Based Psychodynamic Psychotherapy: Bridging the Gap between Science and Practice. New York: Humana Press.

23.

Luborsky

Stuart

Friedman

Diguer

Seligman

D.A.

Bucci

Pulver

Krause

E.D.

Ermold

Davison

W.T.

Woody

Mergenthaler

(2001). The Penn Psychoanalytic Treatment Collection. Journal of the American Psychoanalytic Association 49:217–233.

24.

Luyten

Blatt

Corveleyn

(2006). Minding the gap between positivism and hermeneutics in psychoanalytic research. Journal of the American Psychoanalytic Association 54:571–610.

25.

Michels

(2000). The case history. Journal of the American Psychoanalytic Association 48:355–375.

26.

Midgley

(2012). The therapist’s perspective on participation in research: Learning from experience. Counselling & Psychotherapy Research 12:165–166.

27.

Murphy

S.M.

(2012). Components of the referential process as measures of therapeutic change: Development of normative and psychometric properties. Dissertation, Adelphi University, Derner Institute of Advanced Psychological Studies.

28.

Murphy

S.M.

(2013). Measuring the details that matter: The validity and psychometric properties of the Weighted Referential Activity Dictionary. Unpublished manuscript.

29.

Nelson

K.L.

Moskowitz

D.J.

Steiner

(2008). Narration and vividness as measures of event-specificity in autobiographical memory. Discourse Processes 45:195–209.

30.

Obendorf

C.P.

(1948). Failures with psychoanalytic therapy. In Failures in Psychiatric Treatment: The Proceedings of the Thirty-seventh Annual Meeting of the American Psychopathological Association Held in New York City, June 1947, ed. Hoch

P.H.

New York: Grune & Stratton, pp. 10–21.

31.

Preter

Algus

Hoffman

(2012). Psychoanalytic education: Evaluation of a course on psychoanalytic research using a Grounded Theory Approach. Poster presented at the American Psychoanalytic Association, January.

32.

Schachter

Luborsky

(1998). Who’s afraid of psychoanalytic research? Analysts’ attitudes towards reading clinical versus empirical research papers. International Journal of Psychoanalysis 79:965–969.

33.

Tuckett

(1993). Some thoughts on the presentation and discussion of the clinical material of psychoanalysis. International Journal of Psychoanalysis 74:1175–1189.

34.

Wallerstein

R.S.

(2001). The generations of psychotherapy research. Psychoanalytic Psycholology 18:243–267.

35.

Wallerstein

R.S.

Sampson

(1971). Issues in research in the psychoanalytic process. International Journal of Psychoanalysis 52:11–50.

36.

Werbart

(2009). Minding the gap between clinical practice and empirical research in psychoanalysis: Review of From Psychoanalytic Narrative to Empirical Single Case Research, by Kächele, Schachter, & Thomä. International Journal of Psychoanalysis 90:1459–1466.

37.

Wolpert

Fonagy

(2009). There is no place for the psychoanalytic case report. British Journal of Psychiatry 195:483–487.