Background
Routine monitoring of clients’ functioning has shown great promise in contributing to therapeutic progress. However, most studies rely on clients’ self-report measures which may be affected by clients’ insight regarding their symptoms as well as their motivation to complete the questionnaires. Self-report measures are usually completed before or after the session and thus are limited in their ability to explore the structure of verbal exchanges between the clients and the therapists that are the essence of psychotherapy. The remarkable potential of automated text analytic techniques to explore the important information hidden in the huge amount of words spoken in psychotherapy sessions, had been recognized recently by a few psychotherapy researchers (e.g., Imel et al., 2016). Research outside the clinical domain that used advanced text analytic techniques, repeatedly recognized several textual markers, such as higher use of first person singular, higher use of negative emotional words and lower use of positive emotional words, to be associated with levels of distress. However, these previous studies have tended to focus on a single time point and did not examine change over time. The current study aims to use advanced text analytic techniques to examine whether change in textual markers is associated with change in clients’ well-being session by session throughout treatment.
Methods
Transcripts of 729 psychotherapy sessions from a sample of 58 clients treated by 52 therapists were analyzed. Prior to each session, clients self reported their functioning level. To explore linguistic markers we employed YAP (More & Tsarfaty, 2016) for first-person singular identification and created new positive and negative emotion words lexicons. Because our data has a multilevel structure (session nested within clients), we used multilevel models.
CHANGE IN LANGUAGE SESSION-BY-SESSION DURING PSYCHOTHERAPY 3
Results
Changes in all three textual markers – decreased levels of first person singular, decreased levels of negative emotional words and increased levels of positive emotional words - were associated with improvement in clients’ functioning levels from session to session.
Conclusions
The current study is the first to explore changes in linguistic markers of clients’ distress session by session throughout psychotherapy. Our findings indicate that linguistic markers, such as the amount of use of first person singular and positive and negative emotional words, are sensitive to nuanced fluctuations in clients’ well-being. These findings suggest that linguistic markers which represent more implicit measure of the individuals’ experience are associated with explicit measures of the clients’ well-being.These results may have important implications for research and practice. The computerized measures that analyze the raw linguistic data of therapy sessions could constitute an important addition to existing feedback systems that rely on self-reports. More research that would use computerized text analytic techniques and will focus on the raw linguistic data of psychotherapy sessions is needed in order to asses which mechanisms of change are most pertinent in advancing clients’ well-being.
References
More, A. & Tsarfaty, R. (2016, December). Data-driven morphological analysis and disambiguation for morphologically rich languages and universal dependencies. In Proceedings of coling 2016. Osaka.
Pace, B., Tanana, M., Xiao, B., Dembe, A., Soma, C., Steyvers, M., & Imel, Z. (2016). What about the words? natural language processing in psychotherapy. Psychotherapy Bulletin, 51(1), 17–18.