A Multigenerational Investigation of Voice Onset Time in English-Hebrew Heritage Speakers

Abstract

This study investigates the speech acoustics of two generations of U.S. olim (immigrants) in Israel, first generation immigrants, whose first language (L1) is American English (AE), and their second-generation children, for whom English is a heritage language (HL), as well as L1 Modern Hebrew (MH) speakers. A specific HL accent has been demonstrated in the studies that have investigated the phenomenon, showing that heritage speakers have good control of phonetic/phonological contrasts between their two languages but demonstrate distinct patterns from both native speakers (NS) and second language (L2) learners (who both show L1 influence) (Godson 2004; Chang et al. 2009, 2011; Kupisch et al. 2014; Lein et al. 2016). The research focuses on issues of heritage language phonology and intergenerational multilingualism: What is the speech of HL speakers of AE in Israel like? How does this speech compare to the speech of their parents (their main source of input for AE)? How does MH, their L2 or primary language, respectively, affect their AE? These questions are investigated through a language questionnaire and a picture naming task targeting voice onset time (VOT) in the AE and MH stops /bdg ptk/, which differ in how phonological voicing/voicelessness is cued phonetically by VOT: AE voiced stops /bdg/ have short lag VOT (< 40 ms), while voiceless stops /ptk/ have long lag VOT (> 40 ms) (Lisker and Abramson 1964). In MH, voiced stops /bdg/ exhibit prevoicing (sometimes up to -100 ms or more), while /pt/ have short lag VOT (< 40 ms) and /k/ exhibits long lag VOT (often greater than 60 ms) (Laufer 1998). A total of 7 HS of AE; 10 American olim (immigrants; native speakers of AE); and 5 NS of MH participated in the experiment. Acoustic analysis demonstrates that HL speakers, echoing previous studies, have excellent control over phonetic and phonological contrasts in salient distinctions between their two languages, despite greater overall variability. VOT is within MH norms when speaking MH and within AE norms when speaking AE, results in line with Flege (1995)’s Speech Learning Model (SLM), which predicts that the younger the age of acquisition, the better phonetic discernment between the two languages will be, resulting in the formation of distinct phonetic categories for both languages. The American olim, rather than exhibiting purely L1 influence on the L2, show some L2 (MH) influence on the L1 (AE): Voiced stops, even in AE, tend to be produced as prevoiced, rather than short lag, a shift phenomenon that has been documented for other languages (Pavlenko 2000). This is especially apparent for /b/, but some speakers produce /dg/ with prevoicing as well. Language questionnaires taken by olim participants suggest that this L2 influence is the result of both extensive use of the L2 and cultural identification with their fellow Israeli Jews. NS of MH demonstrate typical L1 MH influence on their AE, but produce native-like AE values for voiceless stops /ptk/, likely because MH has long lag for /k/.