Wyandot Phonology: Recovering the Sound System of an Extinct Language
Craig Kopris, State University of New York at Buffalo

To perform a phonological analysis of a language linguists normally phonetically transcribe utterances obtained from native speakers, then search the data for contrastive and complementary distribution, seek minimal pairs, return to the speakers for verification of hypotheses, and so on.  But what should be done when there are no longer any speakers of the language?  When the linguist cannot simply go to a speaker to round out a paradigm, or run a spectrogram?  One test case is Wyandot.

Wyandot is a Northern Iroquoian language which has lost all fluent native speakers.  However, there are over 250 pages of text in detailed transcription dating from the early 1910s, by Marius Barbeau.  The transcription is problematic in many ways.  First of all, the system used is idiosyncratic, using characters differently from the main traditions (IPA, Americanist, etc).  Second, the descriptions of the sounds the characters represent are often vague or ambiguous.  Third, there is a large amount of over-differentation in the transcription, treating some allophones separately.  Fourth, there is additionally a large amount of under-differentiation, with certain phonemic distinctions not being shown.

On the positive side, Wyandot's nearest relatives are still spoken and relatively well-described.  Using these languages comparatively, as well as standard phonological procedure, and historical insights, we can attempt to resolve the phonology of Wyandot.

Back to the Conference Schedule