A richer model is not always more accurate: the case of French and Portuguese plurals

Michael Becker, Lauren Eby Clemens, and Andrew Nevins

Article (pdf)

Michael Becker, Lauren Eby Clemens, and Andrew Nevins (2011) A richer model is not always more accurate: the case of French and Portuguese plurals. (pdf)

Abstract

We present two cases of morphophonological alternations in the plural of nouns, one from French and one from Brazilian Portuguese. In both of them, monosyllabic items are protected from alternations more than polysyllabic items are. Two large-scale nonce word tasks confirm the productivity of the trend in both languages. We analyze the results in terms of a discrete grammatical distinction between monosyllables and polysyllables, comparing this approach to continuous, more detail-rich models that rely on lexical statistics or phonetic duration. We show that the discretized approach is the most accurate predictor of participants' responses, as informationally richer models introduce irrelevant variance that diminishes their predictive force. We conclude that a relatively simple model of morphophonological alternations that describes the data in discrete, grammatical terms is closer to the one in fact employed by native speakers. A richer model is thus not always a more accurate representation of cognitive reality.

Materials and results

lexicons and experimental results for French and Brazilian Portguese (zip)

Contact info

Michael Becker, michael.becker@phonologist.org
Lauren Eby Clemens, lreby@fas.harvard.edu
Andrew Nevins, a.nevins@ucl.ac.uk