Abstract:

Under the framework of 'experimental historical phonology' (Ohala, 1974), this dissertation investigates two well-studied sound changes in the history of the Germanic language family: Verner's Law and the process of unstressed vowel reduction and loss.

I propose that the perceptual motivation for Verner's Law lies in a re-analysis of the Indo-European pitch-accent as an unintentional consequence of fricative voicing. To test this hypothesis, I model the Verner's Law environments in speech-perception experiments examining whether the pitch of vowels preceding fricatives can influence the perception of fricative voicing. Results support the hypothesis and show that preceding (and following) pitch can influence the perception of fricative voicing.

I propose that the unstressed vowel reduction and loss process is driven by target undershoot in unstressed vowels (Lindblom, 1963) and that this undershoot is attributable to the increased use of duration as a stress correlate in Germanic. Specifically, I posit that it was the shortened duration of unstressed syllables that brought about the reduction and loss of unstressed vowels that reshaped Germanic. If vowel reduction were a characteristic of non-prominent positions independently of durational differences, we would expect to find reduction in languages that do not use duration as a primary correlate of stress. To test the hypothesis that reduction develops only where there are substantial durational differences between stressed and unstressed vowels, I conducted a speech production study of Polish and Turkish, both languages that are not thought to use duration as a main correlate of stress. Results from Polish reveal a direct relationship between duration and reduction, while Turkish shows neither a strong use of duration nor any sign of reduction. Results from both languages support the claim that shorter duration in unstressed syllables was the factor that led to the reduction and loss process in Germanic. The interaction of shortening and reduction can also help to explain the persistence of reduction and loss into the modern Germanic daughter languages.

The dissertation shows how both sound changes can be modeled as resulting from the misperception of ambiguous acoustic cues in ambiguous environments. This illustrates the possibilities for more accurately portraying sound change through the application of synchronic acoustic and experimental methodologies.