Features of speech related to fluency such as filled and silent pauses, sound prolongations, repetitions and self-interruptions exhibit considerable variation among speakers, yet the speaker-specificity of such features has received little attention inforensic phonetic research. The present study investigates the extent to which individual differences in disfluency behaviour are preserved across different speaking styles, a key concern for forensic speaker comparison cases. Disfluency phenomena in the speech of 20 male speakers of Standard Southern British English undertaking a simulated police interview task are compared with the occurrence of the same set of phenomena in the speech of the same speakers participating in a telephone conversation with an 'accomplice'. The speakers' disfluency features are analysed using TOFFA 'Taxonomy of Fluency Features for Forensic Analysis' (McDougall and Duckworth 2017). Individuals exhibit a wide range of variation in their overall rate of production of disfluency features, and these rates are relatively consistent within-speaker across interview and telephone styles. The results for each specific disfluency feature type also show patterns of relatively consistent behaviour within-speaker across-style for most features. For both interview and telephone styles, discriminant analyses based on speaker profiles of disfluency features demonstrate that disfluency features carry speaker-specific information which could be considered alongside other analyses in forensic speaker comparison cases.
from #Head and Neck by Sfakianakis via simeraentaxei on Inoreader https://ift.tt/2L1Rpdb
