Dr Mark Griffiths

FORENSIC LINGUISTICS:

"Did he have an accent?" Forensic speaker descriptions of unknown voices

 

 

Abstract:

This paper addresses the issue of forensic speaker description: cases where the voice of the suspect may form a part of a police investigation, but where no voice recording exists. There is currently no established reliable procedure for eliciting earwitness descriptions of voices. When dealing with earwitnesses, evidence gathering is restricted to the police asking for descriptions of the voice(s) encountered, which are then written down. There is no synthetic reconstruction of the voice. Much doubt has existed over nonlinguists’ abilities to provide voice descriptions, yet every day nonlinguist police officers engage nonlinguist earwitnesses in the process of eliciting linguistic data. This paper offers evidence that whilst nonlinguist ideologies of language are often at odds with those of the linguist, there are nevertheless consistencies and patterns in the

nonlinguist’s perceptions and descriptions of accents and voices. A case is made that it is possible, with the right framework, to elicit data that can be meaningful and make a contribution to the investigative process and that there is the potential to develop an audiofit

 

 

FORENSIC LINGUISTICS:

PhD Thesis - Exploring the forensic audiofit: non-linguist perceptions, conceptions and descriptions and evlautions of unfamiliar voices in a forensic context 

 

 

 

Summary of thesis:

Research in the field of forensic linguistics has in recent years advanced our understanding of the processes and practice of speaker identification, yet the area of forensic speaker description remains relatively under-researched. Drawing on research and methodologies from forensic linguistics, language attitudes and social psychology, this study investigates how non-linguists perceive and conceive of unknown speakers’ accents and voices, and how language ideologies and social stereotypes interact with these perceptions. The study presents two groups of non-linguist respondents with recordings of five speakers responding to interview questions in a pseudo legal context. The analysis considers three major aspects of the non-linguists’ responses: firstly, descriptions of speaker accent, accuracy of respondent estimations and possible mediating factors; secondly, non-linguists’ voice descriptions from the first cohort of respondents, and their presentation to the second cohort in the form of an audiofit framework; thirdly, the study compares speaker evaluations of voices against the findings of previous studies, uniting research on general speaker evaluations of accents with studies on the attributions of guilt. Interacting mechanisms that mediate these evaluations are discussed.

 

Through quantitative statistical analysis, the study finds that non-linguists from the south of the UK perform well at locating accents geographically and socially when provided with information through the vocal channel only, and that accuracy of estimations is further enhanced by providing an assisting framework. For description of voice characteristics, the study finds that many of the features that are salient to and articulated by non-linguists form a useful basis for the an audiofit descriptor of voice characteristics, and are enhanced by respondents’ consideration of possible soundalikes as a shorthand for conveying the salient characteristics. Finally, the study finds that the relationship between accent and attributions of guilt is more indirect and complex than previous literature may have indicated, noting that the common favouring or disfavouring of accents may be bound with unconscious evaluations of intelligence and aggression, and also with the voice characteristics of the individual being evaluated.