| 1 |
Computational Modeling of Learning Biases in Stress Typology
|
|
|
|
In: Doctoral Dissertations (2014)
|
|
BASE
|
|
Show details
|
|
|
| 2 |
Modeling Morphological Subgeneralizations
|
|
|
|
In: Proceedings of the Annual Meetings on Phonology; Proceedings of the 2013 Annual Meeting on Phonology ; 2377-3324 (2014)
|
|
Abstract:
Exceptions to morphological regularities often pattern together phonologically. In the English past tense, exceptions to the regular ‘Add /-d/’ rule frequently inhabit ‘Islands of Reliability’ (Albright & Hayes, 2003), in which a group of words take the same irregular past and also pattern together on a set of phonological characteristics. Adults seem to have implicit knowledge of both the overall pattern (the regular past) and the ‘subgeneralizations’.We model this knowledge of subgeneralizations through the interaction of a structured lexicon and a Maximum Entropy grammar. Words that pattern together with respect to a particular morphological process are grouped into a ‘bundle’, which is indexed to a constraint expressing the change that these words undergo to realize the morpheme. These ‘operational constraints’ compete with markedness and faithfulness in the phonological component. The phonological regularity of a bundle is represented by the average of constraint violations for members. Novel words are assigned a bundle on the basis of similarity to these averages.Our model shows promising correspondence with human data, including biases toward regularity and Island of Reliability effects. The model’s joint learning approach to phonology and morphology, as well as an inclusive concept of `context’, show promise for future application.
|
|
Keyword:
English past tense; learning; morphological learning; morphology; phonological learning; productivity
|
|
URL: http://journals.linguisticsociety.org/proceedings/index.php/amphonology/article/view/42
https://doi.org/10.3765/amp.v1i1.42
|
|
BASE
|
|
Hide details
|
|
|
|
|
