Matthew J. Gordon
This chapter explores methodological and theoretical issues related to chain shifts and mergers, familiar concepts in historical phonology that have received renewed attention by sociolinguists and others engaged in the study of active changes. It is argued that investigating changes while they are in progress opens new windows on the mechanisms driving these linguistic innovations. This fresh perspective shines light on traditional questions and introduces new wrinkles to our understanding of the processes at work in sound change.
A. Daniel Yarmey
A perpetrator speaking over the telephone or one whose face was obscured or disguised are examples of incidents that might lead to testimony on voice identification. Earwitness identification is part of the general area of person identification, but refers specifically to victims' and witnesses' verbal descriptions of voices and speaker identification. Although many laypersons give significantly more credibility to the identification of speakers than is justified, experts generally agree that earwitness descriptions and identification should be treated by the criminal justice system with great caution. This article presents a scientific overview of factors that affect the accuracy of speaker identification, or what is referred to as aural-perceptual analysis, and discusses the reliability and validity of speaker recognition and identification. The police do not have the luxury of handpicking their witnesses (or culprits) but must interview any and all male and female victims or witnesses, all of whom can differ in age, race, expertise, and other characteristics. The article also considers showups and voice lineups.
Paul Foulkes and Peter French
Two of the tasks undertaken in forensic voice and speech analysis are speaker profiling and speaker or voice comparison (or speaker/voice identification). The first task involves an examination of features of the voice to ascertain information about the speaker's regional, social, and ethnic background. The second involves comparative analysis against a reference sample known to have been produced by a particular talker. Although analysis of vocal features cannot determine a speaker's identity, it can provide a wealth of information about the speaker, albeit to varying degrees of precision and confidence. It is for this reason that the label ‘speaker comparison’ is now preferred to identification. Globally, speaker comparison is carried out using three general methodologies: ‘voiceprinting’, analysis using automatic systems, and linguistic-acoustic analysis. This article first presents a brief historical review of forensic speaker comparison and then explores the use of the linguistic-acoustic method in forensic speaker comparison.