Voicing contrasts in the stops of Indian English produced by Assamese speakers

Speakers of most American and British varieties of English contrast word-initial stops using aspiration, with long-lag Voice Onset Time (VOT) for /ptk/ and short-lag or lead voicing for /bdg/ (Docherty 1992, Chodroff & Wilson 2017, etc.). However, Indian English (IndE) speakers of Hindi and Telugu backgrounds living in the US produce a short-lag versus lead voicing contrast instead (Davis & Beckman 1983, Sirsa & Redford 2013). We recorded 10 L1 Assamese speakers residing in India reading English wordlists, all bilinguals who attended English-medium schools starting at age 4.3 (±1.3), young enough to acquire the target English of their community (Flege 1991). We are measuring the VOT of ten tokens each of /ptkbdg/ in word-initial position. Results thus far support the interpretation of the contrast in IndE as voicing not aspiration, as unaspirated voiceless stops versus pre-voiced stops. We also measure two expected consequences of using [voice] for the contrast: full voicing of sonorant consonants after voiceless obstruents in onset clusters and full (>75% of closure), rather than passive, intervocalic voicing. Though neither occurs in aspirating languages (Beckman et al. 2013), we find both in the IndE of Assamese speakers, suggesting IndE has developed a distinct contrast from AmE/BrE.