Skip to main content
Skip to main content

General Meeting - Lydia Quevedo / To agree or not to agree?

A young woman playing table tennis casually, concentrating on hitting the ball with her paddle.

General Meeting - Lydia Quevedo / To agree or not to agree?

Linguistics Friday, September 12, 2025 3:00 pm - 4:30 pm Marie Mount Hall, 1108B

Friday September 12 our General Meeting has Lydia Quevedo, discussing her work on consonant harmony for the all-department audience. 


Many phonological phenomena, such as consonant harmony, are understood to operate over phonemes which are similar in some way. What does it mean to be similar? In this talk, I address this question through two studies of consonant similarity and whether these measures are predictive of harmony patterns. The first study is a typological survey of similarity in phonological features for the consonant inventories of thirty languages with attested consonant harmony patterns. It has been proposed, however, that phonological similarity is an emergent property of acoustic, articulatory, and/or perceptual systems (e.g. Mielke 2008). The second study aims to probe this relationship by examining similarity in terms of low-level representations extracted from self-supervised speech models for three languages. I show that while featural similarity is not predictive of participation in harmony, low-level similarity does remarkably well. These findings suggest a need to refine our theoretical understanding of phonological similarity, and that low-level representations provide a fruitful avenue through which to do so. They also indicate that the relationship between featural and low-level representations is not as close as may sometimes be assumed. 

Add to Calendar 09/12/25 15:00:00 09/12/25 16:30:00 America/New_York General Meeting - Lydia Quevedo / To agree or not to agree?

Friday September 12 our General Meeting has Lydia Quevedo, discussing her work on consonant harmony for the all-department audience. 


Many phonological phenomena, such as consonant harmony, are understood to operate over phonemes which are similar in some way. What does it mean to be similar? In this talk, I address this question through two studies of consonant similarity and whether these measures are predictive of harmony patterns. The first study is a typological survey of similarity in phonological features for the consonant inventories of thirty languages with attested consonant harmony patterns. It has been proposed, however, that phonological similarity is an emergent property of acoustic, articulatory, and/or perceptual systems (e.g. Mielke 2008). The second study aims to probe this relationship by examining similarity in terms of low-level representations extracted from self-supervised speech models for three languages. I show that while featural similarity is not predictive of participation in harmony, low-level similarity does remarkably well. These findings suggest a need to refine our theoretical understanding of phonological similarity, and that low-level representations provide a fruitful avenue through which to do so. They also indicate that the relationship between featural and low-level representations is not as close as may sometimes be assumed. 

Marie Mount Hall false