Volume 6, Number 2 (December 1999)

A. P. A. Broeders
Foreword
Full Text

Marjan Sjerps and Dirk B. Biesheuvel

The interpretation of conventional and ‘Bayesian’ verbal scales for expressing expert opinion: a small experiment among jurists

The Bayesian framework is a useful logical concept which potentially applies to a large number of areas in forensic science, including speaker identification. According to this line of reasoning, the forensic expert is not in a position to draw conclusions about the probability of a hypothesis, e.g. the probability that the unknown speaker on a tape recording is the suspect, unless the expert is absolutely sure. If there is any doubt, the expert can only say how likely it is for the similarities and differences between the voice on the tape and the suspect’s voice to occur if the suspect is indeed the speaker on the tape and if he/she is not. The Bayesian approach also implies that conventional verbal scales for expressing probability conclusions are logically incorrect. The Netherlands Forensic Institute is therefore considering introducing an alternative scale. We will report on a small-scale experiment among jurists on the interpretation of such scales.
Full Text

A. P. A. Broeders

Some observations on the use of probability scales in forensic identification

The increasing prominence of DNA technology as an evidential tool, coupled with the growing interest in the application of Bayesian statistics that has followed in its wake, is making itself increasingly felt in many fields of forensic expertise. It has given fresh impetus to the ongoing discussion about the expression of conclusions in those areas where interpretation of the findings plays an important role, as in handwriting analysis and speaker identification. Recent casework is used to illustrate the inadequacy of some of the present scales and for areas such as writer and speaker identification a modified probability scale is advocated to replace them. The implications of the continued use of ‘logically incorrect’ conclusions – those which address the probability of a hypothesis given the evidence rather than the probability of the evidence given a hypothesis – are examined with particular reference to handwriting and speaker identification.
Full Text

Johan Koolwaaij and Lou Boves

On decision making in forensic casework

In forensic applications of speaker recognition it is necessary to be able to specify a confidence level for a decision that two sets of recordings have been produced by the same speaker (or by different speakers). Forensic phoneticians are sometimes criticized because they find it impossible to provide ‘hard’ estimates of the confidence level of their expert opinions. This paper investigates to what extent the problem can be solved by deploying automatic speaker verification algorithms, to work alone or to support the work of forensic phoneticians. It is shown that, although heavily dependent on operating conditions, one of the advantages of automatic systems is that their performance is in fact measurable. We construct a confidence measure which takes into account the past performance of the automatic system, the operating conditions and the probative value of the speech evidence, as well as the non-speech evidence. It is very important to note that such a confidence measure will never lead to a fully automatic procedure, since it still requires human input to weigh the non-speech evidence as well as human explanation of the procedure followed, and, finally, human interpretation. However, when all conditions are met, this procedure is able to (1) provide an interpretative measure in the individual forensic case and (2) join together the strengths of the human interpretation of the non-speech evidence and the automatic interpretation of the speech evidence, so that finally the joint performance of human and machine is better than the performance of one of them in isolation.
Full Text

Geoff Lindsey and Allen Hirson

Variable robustness of nonstandard /r/ in English: evidence from accent disguise

In forensic phonetics it is valuable to identify speech features which are robust, that is, show little variability for a given speaker. This study was prompted by a case in which threatening telephone calls with apparent accent disguise exhibited standard English /r/, while the interviewed suspect exhibited /r/ pronunciations of a type traditionally described as ‘defective’. An experiment was conducted in which subjects with standard and nonstandard /r/ read sentences in their native accent and then in imitation of a different accent exhibiting standard /r/. Acoustic analysis and auditory assessment suggest that /r/ is nonstandard if its F3 is above about 2000 Hz for men and above about 2350 Hz for women. Of five subjects whose native /r/ was judged nonstandard, three were able to produce standard /r/ in the assumed accent. The results are discussed, including the possibility that degree of native nonstandardness in /r/ may be a predictor of ability to modify it.
Full Text

Duncan Markham

Listeners and disguised voices: the imitation and perception of dialectal accent

This paper presents an experimental investigation into whether a group of speakers could produce convincing text readings in various dialectal accents of Swedish, and the performance of listeners in identifying the accents and determining whether the accents were natural or a disguise. It was observed that individual speakers vary greatly in their ability to produce plausible imitations of accents and to mask their own dialectal background. Examination of the listeners’ perceptual strategies contributes an important dimension to the understanding of reasoning processes in earwitnesses. The linguistically trained listeners were found to use combinations of accent markers as cues to the degree of naturalness, although some of the judgements reflected misconceptions or preconceived ideas about the possible forms of specific accents. The hazards of using speakers with certain accent features in voice line-ups and the potential problems associated with earwitness accent identification are discussed.
Full Text

Augustin Simo Bobda, Hans-Georg Wolf and Lothar Peter

Identifying regional and national origin of English-speaking Africans seeking asylum in Germany

Faced with the influx of asylum seekers without reliable identification, some European countries have started to resort to the analysis of spoken texts produced by the applicants to trace their true origins. This has triggered a hot debate among politicians, non-governmental organizations and academics. Ethical and scientific considerations are the two main poles of the debate. On the scientific front, the question is whether it is possible to determine the asylum seekers’ origin from the analysis of the spoken texts. The present paper, probably the first purely scientific contribution to the debate, has an affirmative answer with respect to asylum seekers from anglophone Africa. It argues that the identification of the region and even the country of origin of the subject, is possible, from phonetic/phonological, sociolinguistic, socio-cultural and other clues. The paper contends that, while one of the clues, especially the linguistic, may be sufficient, a combination of several provides a higher degree of reliability.
Full Text

Book Reviews

Book reviews
Full Text

User: WEIMING LIU
Session: 18888

Forensic Linguistics is published by the University of Birmingham Press.