Attention to client metaphors has been asserted as a way of enhancing cognitive behaviour therapy. Metaphors can be part of the shared language that is co-constructed between clients and therapists. Recent advances in cognitive linguistics have provided the most clearly operationalized method yet to identify metaphors in conversations, allowing analysis of how shared language develops. This study aims to explore how metaphoric shared language develops in early cognitive behaviour therapy sessions. Based on 12 transcripts of early cognitive behaviour therapy sessions, an iterative process led to the identification of a range of therapist and client responses to each other's metaphors, and identification of whether therapists or clients initiated metaphoric exchanges. Types of responses to therapist or client metaphors within three speaking turns were found to be: repetition, rephrasing, exploration, elaboration/extension, or agreement. Bursts of metaphoric exchange were initiated and taken up by therapists and clients at a similar rate. To conclude, therapists need to attend to the occurrence of metaphors and be aware of the range of potential responses that can engage the client in the ‘metaphoric dance’ (co-construction of shared metaphors). This has the potential to enhance engagement and outcomes though more salient reformulating of the problem for the client.

Background: Metaphors are common in psychotherapy and have potential to enhance therapy in numerous ways. However, the empirical study of metaphors in cognitive behaviour therapy (CBT) has tended to be put in the “too hard basket”, confined to being part of the art rather than the science of therapy. The lack of research is largely due to problems with definition, lack of a consistent, reliable approach to metaphor identification and the challenges of finding appropriate methodology to study this language-based activity. Aims: This study aimed to assess the frequency of metaphors in CBT in a large sample of therapy sessions and to evaluate the reliability and utility of the discourse dynamics approach to metaphor identification. Method: The discourse dynamics approach, recently developed by linguists, was used to identify metaphors in 48 CBT session transcripts (from 12 clients and 3 therapists) and the reliability of this approach was evaluated, using an independent rater. Results: The total frequency of metaphors was 31.5 (range 17–49) per 1000 words of therapy conversation. Therapists produced metaphors twice as often (21.2, range 7–36) as clients (10.3, range 3–24). Reliability of the Discourse Dynamics approach was adequate. Conclusions: Metaphors clearly occur in CBT sessions, with therapists using them at a higher rate than clients. While Discourse Dynamics is currently the most detailed identification approach available for investigating metaphor in CBT sessions, it is challenging to acquire skill in it and we found only adequate reliability. Ways to improve reliability and future research possibilities are discussed.