The Discriminant Power of Interjections in Forensic Speaker Identification

Sinjini Chakraborty | Grace Suneetha Didla ^[2]

Abstract: While interjections have been looked at by Traditional Grammarians as an emotional language, other researchers have argued that interjections are deictic items (Poggi, 2008) which are essentially verbal tokens of the speaker’s emotional state in praesentia (Hockett, 1960). Interjections have been categorised in several ways but all interjections can be said to be a codified signal that is stored in our long-term memory as emotion-signal tokens. These tokens are neither universal nor meaningless. On the contrary, they are language-specific, and they are meaningful (Wierzbicka, 2003). Interjections are spontaneous utterances that are deeply rooted in the verbal repertoire of the speakers, and can thereby, be identifying markers of cultures and individuals. In that light, interjections hold relevance to Forensic Speaker Identification. The primary objective of this study was to identify whether interjection usage is influenced more by (a) linguistic background, (b) geographical region, or (c) idiosyncratic factors. To this effect, interjections for 9 different emotions were looked at (a) in Arabic, as spoken across 3 different countries - Lebanon, Iraq and Palestine and (b) in India, across 3 different languages - Telugu, Bengali and Malayalam. The outcome of the study suggests that interjection usage is influenced more by linguistic commonality than geographic belonging. However, the contribution of geographic factors cannot be ruled out.

Keywords: Interjections, emotions, speaker profiling, Forensic Speaker Identification, LADO

Edition: Volume 9 Issue 6, June 2020,

Pages: 491 - 498