King's College London
Browse

AVATAR Therapy Dialogue Corpus

dataset
posted on 2025-02-17, 11:40 authored by Clementine EdwardsClementine Edwards, Thomas Ward, Mark HuckvaleMark Huckvale, Sinead JacksonSinead Jackson, Philippa A Garety, Thomas Craig, Miriam Fornells-AmbrojoMiriam Fornells-Ambrojo, Mar Rus-Calafell, Sandra Bucci, Moya Clancy, Andrew Gumley, Gillian Haddock, Jeffrey Mcdonnell, Hamish McLeod, Alice MontagueAlice Montague, Nikos Xanidis

AVATAR therapy is an innovative form of relational therapy for the treatment of distressing auditory verbal hallucinations, or voice-hearing. AVATAR therapy involves the creation of a digital simulation (‘avatar’) of an individual’s most prominent or distressing voice. The avatar is used in a series of three-way dialogues between participant, avatar, and therapist, in which an initially hostile avatar becomes increasingly conciliatory over time.

This paper introduces the Avatar Therapy Dialogues Corpus, a one-million-word specialised corpus containing orthographic transcriptions of AVATAR therapy session recordings collected from a multi-centre clinical trial. We offer an overview of the design and construction of the corpus, presenting a five-step semi-automated transcription process with a complete description of the processes and specialised tools created. We describe the transcription mark-up and conventions used, which were designed to capture para-linguistic and non-speech features which may have clinical relevance. We include some illustrative analysis of interlocutor turn count and length to highlight the value of a longitudinal dataset of therapeutic interactions. Finally, we discuss the potential value of the corpus for clinical research, existing as the first resource of its kind for the analysis and evaluation of the language of this unique therapeutic approach.

Funding

Optimising AVATAR therapy for distressing voices: a multi-centre randomised controlled trial

Wellcome Trust

Find out more...

History

Data collection from date

2021/01/01

Data collection to date

2023/07/31

Language

English

Usage metrics

    Institute of Psychiatry, Psychology & Neuroscience

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC