Hindi Speech Rate Effects and the Phonology of Voiced Aspirates Eli Asikin-Garmager
Thursday, October 2, 2014
For languages with two- and three-way stop contrasts, there is an asymmetrical effect of speaking rate on Voice Onset Time (VOT). As speakers slow down, prevoiced and aspirated stops are produced with more prevoicing and aspiration, respectively. Conversely, stops produced with short-lag VOT show little to no change in their VOTs (Summerfield 1981, Pind 1995, Magloire & Green 1999, Beckman 2014). Kessinger and Blumstein (1997) show that this asymmetry is found both in languages with two-way stop contrasts (e.g., English, French), and a three-way contrast (e.g., Thai). For Swedish, a language with a two-way contrast between prevoiced and voiceless aspirated stops (Helgason & Ringen 2007), Beckman et al. (2011) found more prevoicing and aspiration as speakers slowed down. They hypothesize that in all of these cases, it is the phonologically-specified categories’ VOT values that undergo a change, assuming privative [voice] and [spread glottis].
The current study is the first to examine speaking rate effects in a language with a four-way stop contrast, Hindi, and the first investigation of rate effects on breathy voiced stops (voiced aspirates). I examine how speech rate affects stop cues in Hindi, and then bring these phonetic data to bear on previous phonological accounts of voiced aspirates, for which authors have posited a range of features, varying both in number and type (Chomsky & Halle 1968, Ladefoged 1971, Davis 1994). I recorded 5 native speakers at different speaking rates, including data for the full four-way stop paradigm. Results show that for plain voiced stops and voiceless aspirated stops, both prevoicing and aspiration increase as a function of speech rate, respectively. The voiceless unaspirated stops show no changes in their VOT values. Crucially, both the prevoicing and breathy voice of voiced aspirates increase as speakers slow down. These results provide empirical support for a privative [voice, spread glottis] analysis of voiced aspirates, which naturally explains why the speech rate effects on the production of the post-release noisy period (breathy voice) of these stops are parallel to the changes found for aspiration of voiceless aspirated stops (e.g., English, German, Swedish). Moreover, such a phonological representation serves to form a natural class of aspirates, which is desirable based on phonological data from closely related Indic languages with four-way stop contrasts.
Please join us for refreshments before the talk at 3:30 pm in 571 EPB.