La connaissance des interventions des psychothérapeutes en situation d'entretien représente un enjeu essentiel dans le champ de la recherche en psychothérapie. Cet article d'ordre méthodologique vise à comparer et mettre en commun trois grilles d'analyse de discours afin d'élaborer un nouvel instrument. Mille soixante-cinq interventions verbales issues de huit entretiens (Investigation Psychodynamique Brève) effectués auprès de deux patientes ont été codées à l'aide des grilles de Blanchet (1991), de Porter (1950), et du Psychotherapy Intervention Rating Scale (1999). L'analyse des fréquences des 25 catégories d'intervention à partir de tableaux croisés a permis de dégager des catégories communes et spécifiques aux trois grilles et de construire un instrument synthétique de 15 catégories. Cette étude montre qu'un rapprochement entre des instruments issus d'univers théoriques différents s'avère réalisable et fructueux.

An essential issue for research in psychotherapy is a good knowledge of psychotherapists' interventions. Empirical studies have demonstrated that therapists' interventions are directly related to the results of the treatment and that they influence predictive factors of therapeutic success such as therapeutic alliance. For some fifty years, therapeutic interviews and their components have been studied in the light of various therapeutic theories and methodologies. Establishing a connection between investigation techniques would be useful in order to improve our understanding of these elements common to the various therapeutic treatments. This methodological study proposes to move forward along these lines. The first part of this article is devoted to the presentation of the principal methods of psychotherapists' speech analysis. We have drawn on currently available knowledge in the fields of linguistics and of psychotherapy research and presented our results in the form of a typology (non-linguistic approaches: hermeneutic and classification; formal linguistic approaches; functional linguistic approaches: pragmatic micro-analysis, linguistic clues, procedural approach, conversational analysis). The second part of our work aims to present an exploratory empirical study of psychotherapists' interventions. The objective of this empirical research is to build a new tool for analyzing psychotherapists' interventions on the basis of three instruments largely recognized in the field. Blanchet's grid (1991), Porter's classification (1950), and the Psychotherapy Intervention Rating Scale (PIRS, Milbrath et al., 1999) were applied to the same corpus (1,065 verbal interventions resulting from 8 therapeutic interviews carried out with two patients by a psychiatrist within the framework of Short Psychodynamic Investigation). The analysis of the frequencies of the 25 categories of the three grids starting from cross tables allowed us to identify common categories as well as categories specific to the three grids. The common categories were compiled and the specific ones were kept separate in order to shape a new synthetic instrument made up of 15 mutually exclusive categories. Three categories result from Porter's classification, seven from Blanchet's grid and five from the PIRS. These 15 categories were then grouped together into four classes according to formal and thematic criteria: 1) the “empathic interventions” group together four categories of Blanchet's grid (referential reiteration, modal reiteration, referential declaration, modal declaration); 2) the “search for information” interventions group together three categories of Blanchet's grid (instruction, referential interrogation, modal interrogation) and a category of Porter's classification (invitation with saying); 3) the “clinical interventions” group together two categories of Porter's classification (support, solution) and three categories of the PIRS (association, interpretation of defenses, interpretation of transfer); 4) the interventions relating to the “therapeutic framework” group together two categories of the PIRS (contractual arrangement, improvement of work). This new instrument shows that in spite of conceptual divergences and methodological specificities, the grids of analysis of therapists' speech share a certain common base. Thus, comparisons and connections between tools seem possible and very useful. Our new instrument of analysis of therapists' interventions has several advantages: integration of clinical and linguistic approaches ensuring internal and external validity, representation of various psychotherapeutic theories After validation (study in progress), this tool could be applied easily and contribute to support the development of research on linguistic processes in psychotherapy. Such a tool could prove particularly useful for studying the specificity of psychotherapeutic interviews and differences in interventions according to therapeutic school.