Dynamic Variability in Speech

A Forensic Phonetic Study of British English

Progress Reports

 


DyViS Progress Report: 18 Months

 

At the halfway point of the DyViS project we are able to report that the project is making very good progress. We have completed all the recordings for the DyViS database, made considerable inroads in segmenting, labelling and transcribing the speech files and have undertaken analyses of a number of the recordings already. We have presented findings at the British Association for Academic Phoneticians (BAAP) Colloquium 2006 (2 papers), the International Association for Forensic Phonetics and Acoustics Annual Conference 2006 (3 papers) and the Australasian International Speech Science and Technology 2006 Conference in Auckland (1 paper). We have also written a book chapter and submitted three further conference papers. Details of all these presentations and publications are given below.

DyViS Database

Compilation of the DyViS database is well underway. Firstly, we developed the tasks and experimental materials needed for each recording, a major part of the foundation work for DyViS. Our simulated police interview task is a new and innovative technique for eliciting speech for forensic phonetic research, which took considerable thought, trialling and evaluation to develop. The technique has proved very effective for eliciting spontaneous (while partially phonetically-controlled) speech under ‘stressful’ circumstances. The data collection techniques for DyViS were outlined at BAAP in Nolan, de Jong and McDougall (2006) and at IAFPA in Nolan, McDougall, de Jong and Hudson (2006b), and a description of the simulated police interview task is provided in Nolan, McDougall, de Jong and Hudson (2006a). The recordings have also all been completed: we have recorded the full 100 speakers required. We have also made additional recordings of 20 of these speakers on a second occasion, 10-14 weeks after their initial recording session. Segmenting, labelling and transcribing the speech files is making good progress.

Research on Sound Change

One of the research aims of DyViS is to test diachronic change as a source of speaker idiosyncrasy. To this end we have investigated monophthongal vowels produced by 50 speakers in the database. We have analysed midpoint F1 and F2 frequencies of the vowels in HEED, HAD, HARD, HOARD, HOOD AND WHO’D. Results for the first 20 speakers were presented in Nolan, de Jong and McDougall (2006) at BAAP and in Nolan, McDougall, de Jong and Hudson (2006b) at IAFPA. Written reports are available in Nolan, McDougall, de Jong and Hudson (2006a) and de Jong, McDougall and Nolan (forthcoming), and further papers are planned.

Research on Formant Dynamics

Another research aim of DyViS is to quantify articulatory-acoustic dynamic features for individual speakers. Work on this aim includes analysis of intervocalic /r/ sequences (McDougall 2006b). Investigation of the vowel /u:/ in read speech for 20 speakers is also underway. Models of individuals’ formant dynamics using polynomial equations are being developed. Papers reporting these findings and further analyses are planned.

Research on ‘Speaker Space’

Together with the vowel formant analysis above, we have calculated f0 statistics for the full set of 100 speakers for the telephone call task. Papers reporting these findings and further analyses are planned.

Other Activities: Visiting Scholars and Workshops

The DyViS project has attracted several visiting scholars, who were keen to meet us to discuss our research. These interactions have been very useful for the project, stimulating interesting discussion and leading to further new ideas. Dr Michael Jessen of the Bundeskriminalamt in Germany visited the project for two weeks in May 2006. During this time he presented a departmental seminar entitled ‘Theoretical issues and empirical studies on the concept of “idiolect” in forensic phonetics’. We conducted a number of ‘structured discussions’ on forensic phonetic issues (both practical and research-oriented), as well as benefiting from many helpful informal discussions of the DyViS research. Martin Duckworth of the College of St Mark and St John, Plymouth, also visited during this period, contributing to the structured discussions and providing helpful feedback on the development of DyViS. In July 2006 the project was visited by Professor Suzanne Boyce of the University of Cincinnati, USA, and Dr Yuko Kinoshita of the University of Canberra. We took the opportunity to hold a departmental workshop at which Professor Boyce and Dr Kinoshita gave presentations, together with three presentations of DyViS work-in-progress. Professor John Gibbons of the University of New South Wales, Australia, visited the DyViS project in September 2006. He presented a departmental seminar entitled ‘Pressure points: how witnesses come to agree with what they do not believe’. Also in September 2006, the DyViS team visited York to meet with Peter French, Paul Foulkes, Phil Harrison and Louisa Cawley. The aim of this meeting was to discuss issues regarding the presentation of conclusions in forensic phonetic case reports. A position statement was drafted for circulation among the UK forensic phonetics community.

References

2006a, K. McDougall, ‘Characterisation of individuals’ formant dynamics using polynomial equations.’ Paper presented at the International Association for Forensic Phonetics and Acoustics Annual Conference, Gothenburg, 23-26 July 2006.

2006b, K. McDougall, ‘The effects of stress and neighbouring vowel on speaker-distinguishing properties of /r/ in SSBE.’ Paper presented at the British Association of Academic Phoneticians Colloquium, Edinburgh, 10-12 April 2006.

2006, F. Nolan, G. de Jong and K. McDougall, ‘Introducing the DyViS project: ‘Dynamic variability in speech: a forensic phonetic study of British English’.’ Paper presented at the British Association of Academic Phoneticians Colloquium, Edinburgh, 10-12 April 2006.

2006, F. Nolan, G. de Jong, K. McDougall and T. Hudson, ‘Diachronic change as a source of speaker idiosyncrasy: a study of Standard Southern British English monophthongs.’ Paper presented at the International Association for Forensic Phonetics and Acoustics Annual Conference, Gothenburg, 23-26 July 2006.

2006a, F. Nolan, K. McDougall, G. de Jong and T. Hudson, ‘A forensic phonetic study of ‘dynamic’ sources of variability in speech: the DyViS project.’ In: P. Warren and C.I. Watson (eds.), Proceedings of the 11th Australasian International Conference on Speech Science and Technology, 6-8 December 2006, Auckland: Australasian Speech Science and Technology Association, 13-18.

2006b, F. Nolan, K. McDougall, G. de Jong and T. Hudson, ‘Introducing DyViS: a dynamic study of British English for forensic purposes.’ Paper presented at the International Association for Forensic Phonetics and Acoustics Annual Conference, Gothenburg, 23-26 July 2006.

(forthcoming) G. de Jong, K. McDougall and F. Nolan, ‘Sound change and speaker identity: an acoustic study.’ In: Christian Müller and Susanne Schötz (eds.), Speaker Classification. Springer.