Eleanor Chodroff’s Dissertation PresentationBy Krieger School of Arts & Sciences
3400 N Charles St, Baltimore, MD 21218
Eleanor Chodroff will present an open talk on her dissertation entitled, “Structured Variation in Obstruent Production and Perception.”
ABSTRACT: The acoustic-phonetic properties of speech sounds vary substantially across languages, and across talkers within a single language. The central thesis of this dissertation is that different aspects of phonetic realization are not independent but rather highly structured among phonetic categories across languages and talkers. Structured variation (defined as phonetic covariation among speech sounds) has important implications for the theory of phonetic realization as it applies to individual speakers and languages, and may also account for instances of generalized perceptual adaptation.
The dissertation argues for a principle of uniformity to account for phonetic covariation across talkers. This is formalized as three uniformity constraints that operate at the phonetics-phonology interface: pattern, target, and contrast uniformity. Pattern uniformity serves as a general constraint on phonetic implementation, requiring a similar structure of phonetic targets across talkers. Target and contrast uniformity directly influence the mapping from distinctive features to phonetic targets. Target uniformity requires similar (or identical) phonetic realization of a distinctive feature value, whereas contrast uniformity requires a comparable phonetic difference in sounds that contrast in a feature across talkers.
In this talk and dissertation, I present several case studies of structured variation in stop consonant voice onset time (VOT) and sibilant fricative spectral shape to evaluate the predictions of uniformity. Variation and covariation in VOT were examined across American English talkers, children, and cross-linguistically. Structured variation was also investigated among sibilant fricatives across American English and Czech talkers. The structure that emerged from our analyses provided strong evidence for target uniformity (and thus also for the more general notion of pattern uiformity).
Listeners may exploit phonetic covariation to generalize talker-specific phonetic properties from one sound to another. A series of experiments investigating generalized adaptation to talker VOT and fricative spectral shape were conducted. For fricatives, the phonetic covariation hypothesis was compared to a general auditory hypothesis based on spectral contrast and a cue-based normalization hypothesis. Generalization was observed for both VOT and fricative spectral shape in a manner consistent with phonetic covariation, but strong evidence was also found for spectral contrast, particularly in local contexts.