Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.773143
Title: Language and behaviour : pragmatics and parole
Author: Crimmins, Melissa
ISNI:       0000 0004 7960 560X
Awarding Body: Lancaster University
Current Institution: Lancaster University
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Thesis embargoed until 21 Jan 2024
Access from Institution:
Abstract:
This thesis applies linguistic analysis to one of the most difficult and under-theorised problems in criminology: recidivism on parole (Weisberg, Mukamal, & Segall, 2011, p. 3). During the parole hearing, inmates' suitability for early release is judged according to parole board members' intuitions and impressions, not evidence-based -- or even consistent -- criteria. The forensic linguistic approach of this study examines the same inmate statements considered by parole board members, but draws quantitative speech data from these, combining cutting-edge linguistic research with sophisticated data modelling to determine whether inmates exhibiting differentiated parole outcomes (i.e., recidivism or non-recidivism) also exhibit differentiated structural linguistic choices in the parole hearing. This study is inspired by the text analysis efforts of historic criminological and social cognitive research, contemporary psychological studies, and burgeoning interdisciplinary explorations toward connections between language and behaviour. It innovates by assessing pragmatic dimensions of language in naturally-occurring, high-stakes data, and is able to go further as a result. 182 successful parole hearings for violent criminals (i.e., murder, rape and sexual assault, and/or robbery) in four American states were transcribed and coded to form a parole hearing corpus of 750,000 words. Over thirty linguistic features, identified from criminological, social-cognitive, and psychological theories of violence, were operationalised for a thorough corpus analysis. The most significant linguistic features (p < 0.05) were included in a multiple logistic regression model alongside variables commonly used by criminologists, to ensure potentially significant and interfering factors were judged together, preventing bias or omission which could otherwise invalidate results -- a serious Type I error which commonly compromises linguistic research. The result is the first evidence-based linguistic assessment of interactions between criminal recidivism and language behaviours. The findings suggest significantly greater accuracy towards parole outcome (76%) than predictive measures currently in use. Theoretically, they provide new contours to the understanding of language and behaviour. In practical terms, they indicate that evaluating linguistic choices toward potential re-offense and behavioural outcome may be justified, while affirming the importance of evidence-based grounding for those judgments (Latessa & Lovins, 2010).
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.773143  DOI:
Share: