Phonological Activeness Biases in Language Structuring and Acquisition

Undergraduate: William Carter

Faculty Advisor: Elliott Moreton
Department: Linguistics

---

To make a noun plural, one writes an ¿¿¿s¿¿¿ at the end, but the pronunciation of ¿¿¿s¿¿¿ actually depends on the word ¿¿¿ sometimes it¿¿¿s buzzy like ¿¿¿z¿¿¿ as in ¿¿¿dogs¿¿¿. This is determined by the sound-patterns that English-speakers acquire. These patterns target sounds that are grouped by common properties and features. The pronunciation of ¿¿¿s¿¿¿ depends on whether the final sound of a word has vibrating vocal cords.

Prior studies find that language-learners show biases towards acquiring certain patterns in second languages. This study proposes feature activeness to be one source of linguistic bias. Active features are those used in a large number of sound-patterns.

This study consists of a corpus study testing for within-language activeness biases, and an online artificial language-learning task where 100 subjects learn sound patterns using features of various activeness in the English language. This second task assesses whether activeness biases apply in second-language learning. The experiment will be completed in early-mid March.

In the corpus study, frequency distributions generated by a rich-get-richer algorithm were fitted to the data and tested for equivalence with the observed feature distribution for five unrelated languages. Results suggest within-language feature frequency distributions for each of these languages exhibit rich-get-richer effects, and that languages prefer to reimplement active features. This offers a new perspective on the mental organization of language.