FORENSIC SPEAKER PROFILING FROM THE SAMPLE IN ЕNGLISH AS A FOREIGN LANGUAGE – VOWEL DURATION ANALYSIS

Abstract

Forensic speaker profiling is a procedure employed in criminal cases where there is a voice recording of the criminal, but there is no suspect. It encompasses determining the age, gender, origin or socio- economic status of the recorded speaker (Rose 2002; Kašić, Đorđević 2009a; Jessen 2010). One of the challenges of modern forensic phonetic science is speaker profiling from the voice sample in a foreign language. In the current research, we analyzed the vowel duration of five speakers from Novi Sad and five speakers from Niš, when they were speaking spontaneously in their mother tongue, Serbian, and in a foreign lan- guage, English. We compared the quantity of vowels of each group of speakers within-language and across languages. The acoustic analysis of vowels was performed manually in Praat (Boersma, Weenink 2018), by looking at the spectrogram and waveform of the recordings. To test the difference in means of two groups of data, we used the Welch t-test (Welch 1947). Our results show that urban speakers from Niš and Novi Sad do not exhibit statistically significant differences in the duration of their English vowels. However, certain duration relations that exist between vowels may be indicative of one’s native dialect.