Speaker specific characteristics in Maltese : potential indicators for use in forensic settings

Abstract

The main purpose of this research is to examine speech data with a view to identifying speaker specific characteristics which might be used for speaker verification and identification in forensic linguistic contexts. The characteristics which the study will focus on are speech disfluencies such as filled pauses, prolongations and articulation errors in spoken Standard Maltese. Although the filled pauses available (ee, em, mm, qq in Maltese) are likely to be language (or variety) specific (Cruttenden, 1997), the specific fillers used by different speakers, and their distribution in connected speech, may be speaker-specific. Other disfluencies may also have a speaker-specific basis and it is the choices and distribution of these which this study examines. Research has shown (e.g. Nolan, 1997) that phoneticians can put together a profile of a particular speaker based on the identification of specific characteristics in a speaker's speech. However, more research is needed on what elements of phonetic analysis can be used as evidence in legal contexts. Annotated Map Task and simulated conversational data from the MalToBI corpus involving four speakers served as the starting point for the analysis. A number of adaptations to the annotation guidelines were implemented. Further data from the same four speakers, engaged in a monologue task in this case, were collected specifically for use in this study. These data were annotated using the adapted guidelines. Results suggest that disfluency types which occur in speech for reasons to do with, for example planning one's discourse, are unlikely to be speaker-specific: it is the less-used disfluency types which may have a more speaker-specific basis. Moreover, an element of speaker-specificity may also be derived from examination of the specific implementation made by speakers of commonly used disfluency types. A tendency towards differences in the use of disfluencies across female as compared to male speakers was also noted.