APPLYING COMPUTERIZED LINGUISTIC MEASURES TO EXPLORE THE CLINICAL FEATURES OF DIFFERENT DIAGNOSTIC GROUPS AND THE PSYCHOTHERAPY CHANGE PROCESS

Although mental and interaction processes like those emerging in psychotherapy and psychopathology should somehow be reflected in the vocabularies and linguistic style of the speakers involved, to date few studies investigated the linguistic manifestation of psychological functioning of patients with specific diagnoses in clinical or naturalistic contexts. This lack of studies is probably due partly to the difficulty of getting sessions and interviews transcripts, partly to the complexity of the interactions between verbal and non- verbal linguistic variables, and partly to the lacking clinical relevance of some linguistic variables so far investigated in the analysis of day-to-day conversation. The studies in this symposium describe two computerized linguistic analysis programs: The Computerized Reflective Functioning (CRF; Fertuck, Mergenthaler, Target, Levy, & Clarkin, 2012) and the Italian Discourse Attributes Analysis Program (IDAAP; Maskit, 2011; Maskit, Bucci, & Murphy, 2012) that calculates the linguistic measures of referential process phases as described by Bucci (1997, 2016). These programs have been used to empirically examine three linguistic styles that the authors associated to different psychological functions or states: emotional arousal; narrating/symbolization; and reorganization/reflective functioning. These functions have been investigated on narratives posted by borderline Instagram users; on the Adult Attachment Interview of women victims of intimate partner violence; on Relationship Anecdotes Paradigm Interview of patients suffering from unipolar and bipolar mood disorders; and on a large sample of psychotherapy sessions to which is possible to compare the single session or treatment. These studies are ones in a growing line of research exploring how patients speak rather than just the content of what they say, revealing aspects that are largely undetectable by both speakers and that bypass the biases of self-report or observer-based measures.