Learner dynamics in a model of wug inflection: integrating frequency and phonology

AbstractA recent large-scale wug-task study found that non-native speakers of English tend to produce fewer regular past-tense -ed inflections than native speakers (Cuskley et al., 2015). In this paper we present a model that can account for this difference in behaviour as resulting from a difference in input amounts and distributions. This model attends to both frequency, using Bayesian non-parametric methods, and phonological similarity between words, using a neural model of word forms, and unifies these factors within a single probabilistic framework. We show that the general pattern of over-use of irregular inflections in non-native speakers can result simply from exposure to a smaller amount of input and does not require any model-internal distinction of natives and non-native speakers. Our model also captures the interaction between class frequency and phonological similarity that was evident across all participant productions.


Return to previous page