LOC Colloquium- Linguistica ex Machina: Tone inventories as hypothesized by machine

UPDATED: This colloquium will be held in Woodward Lecture Hall #1 (WOOD 1)
Michael Fry (UBC Linguistics) will be presenting a colloquium for undergraduate students. This presentation will be eligible for LOC credit.
A fundamental task of linguists is to accurately describe the sound patterns of a language. In the field of phonology, this generally starts with identifying the set of contrastive sounds in the language, its phoneme inventory. If the language under investigation is a tone language, then identifying the contrastive tones in the language, its tone inventory, is also needed. Historically, phonologists have identified phoneme and tone inventories through lengthy consulting processes, relying on their own intuitions, or a combination of both. Yet, in light of the recent advances in machine learning, perhaps there is another way. In this presentation, I demonstrate machine learning’s efficacy in generating hypothesized tone inventories for several tone languages. In doing so, I also support the central tenet of Emergent Phonology, which is that phonology (i.e. phonological categories) emerges through the accumulation of phonetic experience.