By Andries W. Coetzee
Feb 17, 2013
Over the past several years, the focus of phonetics research in our department has expanded significantly to now also include detailed study of articulation. In order to facilitate this new research direction, we have added new equipment to the Phonetics Lab, including an ultrasound machine (for imaging tongue movement) and a high speed digital camera (for capturing lip movement). Several research projects in in the Lab have made use of this new equipment, including the dissertation projects of our recent alum Susan Lin (PhD 2011) and current graduate student Jon Yip.
In this Friday's Phondi meeting, Jon will present the initial results of his dissertation research. In his dissertation, Jon is investigating the inter-gestural timing patterns in word-initial clusters in Modern Greek. His data collection protocol has him simultaneously capturing tongue movement (with the ultrasound machine), lip movement (with the lip camera), and the acoustic signal. Having access to these three channels of information in a time co-ordinated structure enables Jon to investigate the relative timing of the consonantal constriction gestures in great detail. Jon's dissertation speaks directly to currently ongoing debates in the literature on whether inter-gestural timing is driven primarily by considerations of perception (accommodation to the listener) or production (self-accommodation by the speaker). Full information about his presentation, including a title and abstract, is given below.
What: Phondi Presentation by Jon Yip
When: Friday 2/22, 1 pm
Where: 473 Lorch Hall
Linguistics effects on gestural-coordination timing in Greek CC sequences
Kinematic measures of the relative timing of articulatory movements in the production of consonant clusters in several languages have revealed systematic patterns of articulatory overlap. Specifically, CC sequences exhibit greater overlap between labial and lingual constrictions (1) for front-to-back clusters (e.g., [bg]) than back-to-front clusters (e.g. [gd]; Chitoran et al., 2002; Chitoran & Goldstein, 2006), and (2) for stop-lateral clusters (e.g., [pl]) than for stop-stop clusters ([pt] or [pn]; Chitoran & Cohn, 2009; Kühnert et al., 2006). The source of these timing patterns is of importance to theories of speech production, and to theories of the relation between production and perception: Is overlap primarily constrained by biomechanical factors, such as constraints on the coordination of the oral articulators, or by perceptual factors, such as the use of production timing strategies to improve perceptual recoverability?
This study of Modern Greek speakers’ CC productions investigates the effects of place order (front-to-back, back-to-front), C1 manner (plosive, fricative), and C2 manner (plosive, fricative, lateral) on gestural overlap. Using a combination of ultrasound imaging and lip-camera video, I collected articulatory data on lip, tongue tip, and tongue dorsum movements for clusters [pt ps pl ft kt ks kl xt] and measured gestural overlap in terms of the duration of lag between the release of C1 and the achievement of C2. A perceptual recoverability account predicts that lingual gestures should overlap less in, for example, [kt] than in [ks] or [kl] because the acoustic release of [k] should be masked more by overlap with complete closure ([t]) than partial closure ([s] or [l]); similarly, differences in acoustic masking mean that gestures should overlap less in [kt] (back-to-front) than [pt] (front-to-back). Alternatively, the biomechanical account simply predicts that the interdependent lingual (tongue tip and dorsum) gestures in [kt ks kl xt] (back-to-front) should overlap less than the separate labial and tongue tip gestures in [pt ps pl ft] (front-to-back) and that variation in overlap due to differences in C1 or C2 manner should be relatively small.
For all speakers, the lag between C1 and C2 was significantly greater in back-to-front sequences than front-to-back sequences. In terms of interaction effects between place order and manner, results show two patterns across speakers. For half of the speakers, the order of place do not interact with manner. However, the other half of the speakers produced the opposite of the expected interaction effect predicted by the perceptual hypothesis; that is, C1-release-to-C2-achievement lag is actually greater in [pt] than in [ps pl] but not greater in [kt] than in [ks kl]. These results for gestural overlap are easily explained by a biomechanical approach to gestural timing and cast doubt on the possibility that gestural timing between consonants (at least in Modern Greek) is readily explained in terms of perceptual recoverability.