Learning French Liaison with Gradient Symbolic Representations
Errors, Predictions, Consequences
Bibliographic Data
| ID | 19572797 |
|---|---|
| Authors | Anne-Michelle Tessier, Anne‐michelle Tessier (0000-0001-8261-9955, University of British Columbia), Karen Jesney |
| Year | 2021 |
| Volume | 9 |
| Publication date | 2021-05-01 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Proceedings of the Annual Meetings on Phonology (JOURNAL) |
| Journal identifiers | ISSN: 2377-3324 • E-ISSN: 2377-3324 |
| Publisher | Linguistic Society of America (PUBLISHER • US) |
| DOI | 10.3765/amp.v9i0.4940 |
| OpenAlex | W3158064982 |
| Language | EN |
| Citations received | 6 |
Smolensky & Goldrick (2016) first made the case for Gradient Symbolic Representations (GSRs) as the inputs to phonological grammar using the phenomena of French liaison. Under this view, many common French words are stored underlyingly with partially-activated word-final consonants, and others with gradient blends of partially-activated word-initial consonants. In this paper, we follow up some of that view's predictions and consequences, focusing on the acquisition of French liaison using GSRs. We compare our simulations of error-driven GSR learning with observed errors made by French-learning children, and find the results to be encouragingly similar. We also compare predictions about the end state of GSR learning with a pilot study reporting adult French speakers' use of liaison in nonce words, where we find a rather less good explanatory fit. The paper emphasizes the role of word and collocation frequency in the development of phonological patterns by a GSR learner, and outlines many future avenues for research
Cognitive psychology · Cryptographic nonce · Grammar · Linguistics · Machine learning · Natural language processing · Computer Science · Language Development and Disorders · Natural Language Processing Techniques · Philosophy · Phonetics and Phonology Research · Psychology · Artificial Intelligence
| Unique citing works | 6 |
|---|---|
| Citations per year | 1.5 |
| Citation span | 2022 - 2026 (5) |
| Citation velocity | current |
| Highly cited | No |
| Citation types | Neutral: 5 |