INSTRUCTIONS FOR SEGMENTING TEXT FOR THE APS (September 4, 2000)

therapist segments…patient segments…joint segments (2 types)

The primary purpose of segmenting sessions for rating on the Analytic Process Scales (APS) is to track the relationships between patient and therapist work as they develop through the session, and to study the process of mutual influence. Whole- session ratings, while they may give an idea of what the treatment is like, cannot inform us about the process of mutual influence. In order to arrive at meaningful ratings of what analyst and/or patient are contributing at any one time, we must divide the sessions into segments. Then, by tracking what happens subsequently, we can evaluate cause-and-effect relationships.
From studying the varying nature of the clinical exchanges, we have derived four kinds of segments in our classification, designated as:

“T” (therapist segment)

“P” (patient segment),

“J” (joint segment initiated by the patient)

“O” (for “Other” joint segment initiated by the therapist)
The”J” and the “O” segments are back-and-forth dialogues in which both patient and therapist contribute sufficiently to warrant rating on the APS. For these segments, both sets of scales are used.
How large should a segment be? If the whole session were taken as one joint segment, pooling everything in the same time period, there would be no analytic power in investigating how each party influences the other.
Turns of speech offer a natural unit which can be starting points for segmentation. On the other hand, when there are many rapid exchanges between patient and therapist, to segment at each turn of speech would produce too many small units, and the raters, in applying a dozen or more scales to each fragment, would be engaged in such repetitious work that they would lose focus. If the therapist were exceptionally silent, a single very large turn of speech as a patient segment might include too much variety to be readily kept in mind by the rater. When the same speaker changes topic during one turn of speech, lumping is encouraged unless the “lump” created becomes too unwieldy or heterogeneous. You might therefore choose to start a new segment when a change of topic occurs during the same turn of speech. (Please note that we do not consider a therapist’s extension of a clarification or an interpretation into another area to be a change of topic.) Finally, if the segmenting process isolates words or sentences, before or after, which are too fragmented to merit rating, it is preferable to create a larger segment than to “orphan” these comments.
The ideal length of segment will have coherence, and will provide enough material for meaningful rating, and yet should be small enough so that interplay between patient and therapist will be reflected in the sequence of ratings. [This is particularly important, since we use time-series methods of analysis.] Moreover, many of the APS variables are designed to allow for direct comparison of therapist and patient contributions, for example, the patient expressing self-esteem issues and the therapist addressing self-esteem issues. Thus, we can track the way the therapist addresses the patient contribution, and the ways in which the patient’s contribution influences the therapist’s interventions.
Taking these factors into consideration, we have evolved two overarching principles to be taken into account when segmenting sessions:
* the need for fewer rather than more segments (lumping versus splitting), once sufficient segmentation is made to capture the flow of work between the two participants.

* the need for segments which are inherently interesting to rate. If a segmentation under

consideration does not seem meaningful, it should be pooled with other material. For example, it might be a repetition or extension of something already discussed. However, it is not intended that the segmenters “cherry-pick” only the positive or contributory remarks.

* It is vitally important to read through the Analytic Process Scales Coding Manual in order

to understand the ratings which will be made, and thereby how the material can be most usefully segmented. It may be helpful to look at the “flower” pictograms showing the array of different variables.
How to notate: Ends of segments are marked with a large, double vertical bar mark | | . At the same time, for visual clarity, a prominent horizontal line should be drawn in the margin after each segment, with the letter of the segment just ending (P, T, J, or O) written prominently above the line where the new segment begins, and with the letter of the next segment (P, T, J, or O) written prominently underneath the line where the new segment begins, in the left margin.
When comparing the text with the other rater, write in the other rater’s marking but with a circle around the letter. Then use the right margin to mark the final agreed-upon rating, which will then be entered in the transcript.
Pauses should be included with the preceding segment, except for a pause before the initial segment.
THERAPIST SEGMENTS (T): All therapist remarks which are more than incidental are rated as a T segment or part of a T segment, unless they are part of a joint segment. The definition of “incidental” is difficult, because even contributions such as “hmmm”, “mmm-hmmm” and so forth do have significant impact on the patient’s experience and convey a sense of therapeutic participation. Almost all therapist interventions are important to the quality of the therapist-patient relationship. However, the APS focuses on the quality of the work being done and not as much on the quality of the relationship except for a few variables. Thus, any therapist intervention that can be rated in terms of type and quality, the nature of core analytic activity, affective quality and so on, should be designated as a T segment (See the APS_Therapist pictogram). Encouragements to elaborate are usually not eliminated as incidental, since they are an important aspect of what we measure.

PATIENT SEGMENTS (P): Unless there is a rapid exchange between patient and therapist, in which case the segment would be considered to be a joint segment (J or O), P segments tend to be considerably larger than therapist segments. This is because the patient’s work is to present material which opens up or elaborates a topic, whereas the therapist makes a more focused intervention. Each segment of patient material is rated on a number of patient characteristics in which we are interested, such as how well the patient conveys experiences, communicates feelings, is self-observant, and so forth (See APS-Patient pictogram). P segments can contain therapist statements as long as these are judged incidental.

As an aid in determining whether to divide a given patient turn of speech into more than one segment, it can be helpful to think whether you would expect the APS ratings to be substantially different from one part of the patient turn of speech to another. For example, if the patient from material that appeared not to have anything particularly to do with the therapist, to an expression of a transference fantasy, we know that the APS scores for each of those segments would be significantly different. Therefore, it might be preferable to divide such a patient turn of speech, so that raters are not obliged to try to average scores from two rather different segments. On the other hand, if the turn of speech represents several variants, but stays pretty coherently in the same general way, then we can spare the additional effort of having raters rate it twice.
JOINT SEGMENTS: Joint segments are characterized by a rapid exchange between patient and therapist where, through the “grouping” of contributions, there is mutual participation in an ongoing and unfolding way. (There is no implication here as to whether the unfolding is constructive or not.) Joint segments are important in that they allow us to consider complex exchanges in short passages as a whole rather than being obliged to rate individual fragments of their contributions. This leads to more meaningful ratings.
THE FIRST TYPE OF JOINT SEGMENT (J): A (J) segment is so designated when the “rapid exchange” aspect begins with a patient remark and is either interrupted or followed by one or more therapist remarks. The logic behind the designation of a J segment is that in such an exchange it would not be sufficiently meaningful or coherent to rate the initial patient contribution by itself, and all of the patient contributions in the exchange are actually part of a process of developing the same point. At the same time, the therapist’s contributions put together are considered more meaningful and coherent than standing by themselves in that they are elaborating on the same point in a step-by-step fashion. As soon as the “rapid exchange” aspect subsides, or a significant change of topic occurs, a new segment can begin.
THE SECOND TYPE OF JOINT SEGMENT (O): An O segment is so designated when the “rapid exchange” aspect of the interchange begins with a therapist remark and is either interrupted or followed by one or more incomplete or incidental patient remarks. The logic behind the designation of an O segment is that, in such an exchange, it would not be sufficiently meaningful to rate the initial therapist contribution by itself because it is better understood when taken together with subsequent remarks by the therapist, and that the patient contributions put together are considered more meaningful and coherent as well when they are rated together. As soon as the “rapid exchange” aspect of the interchange subsides, or a significant change of topic occurs, a new segment, whether a P, T, J, or O can begin.
Raters complete the patient APS scales first for a J segment, since the patient’s material begins the segment. The opposite is the case for an O segment. Do not form a joint segment if the result leads to just as many ratings as if you divided the material into P and T segments. For example, a P segment followed by an O segment would lead to three sets of ratings – P, T. and P. If there were only two turns of speech in the proposed O segment (first the therapist and then the patient), the result could just as well be expressed by P, T, then P segments, which is more straightforward than P followed by an O segment.

SUMMARY
In allocating turns of speech to segments, clinical coherence is the primary criterion. The net effect of our approach to segmentation will be that long pieces of text that stand alone will mostly end up as therapist or patient segments, but sometimes will become part of joint (J) or other (O) segments to help resolve problems in the surrounding material.