Do phonological features exist? Phonologists have provided answers from generalizations that can be made within and across languages. We look for experimental and computational evidence.
Computational evidence is that acoustic front–back asymmetries in five-vowel inventories may result from featural front–back asymmetries:
|2011||Paul Boersma & Kateřina Chládková:|
Asymmetries between speech perception and production reveal phonological structure.
Proceedings of the 17th International Congress of Phonetic Sciences, Hongkong, 17–21 August 2011. 328–331.
Evidence for vowel height features is found when computationally modelling the behaviour of how real Czech listeners discriminate central vowels that do not occur in their language:
|2015||Kateřina Chládková, Paul Boersma, & Titia Benders:|
The perceptual basis of the feature vowel height.
Proceedings of the 18th International Congress of Phonetic Sciences, Glasgow, 10–14 August 2015. Paper 711 (4 pages).
Asymmetries in ERP measurements with French listeners to rounded front and back vowels reveal that the feature [HIGH] may be privative:
|2015||Mirjam J.I. de Jonge & Paul Boersma:|
French high-mid vowels are underspecified for height.
Proceedings of the 18th International Congress of Phonetic Sciences, Glasgow, 10–14 August 2015. Paper 948 (4 pages).
Computationally, features can emerge substance-freely in a neural network:
|2021/10/14||Paul Boersma, Kateřina Chládková & Titia Benders:|
Phonological features emerge substance-freely from the phonetics and the morphology.
Go to Paul Boersma’s home page