Lai, J. & Poletiek, F.
Cognitive Psychology, Leiden University, Leiden, the Netherlands
Hierarchical center-embedded structures, such as AnBn, cause difficulties for language learners due to their complexity (Bach, Brown & Marslen-Wilson, 1986; Chomsky, 1957; Corballis, 2007). Recent artificial grammar learning (AGL) studies (Lai & Poletiek, 2011) demonstrated a starting small (SS) effect. In particular, sufficient exposure to zero-level-of-embedding exemplars and a staged-input were the critical conditions in learning AnBn structures. The present 2 AGL experiments aim to replicate the SS effect and test another possible facilitating effect of the input, i.e. the frequency distribution of the input stimuli. Participants were exposed to a set of non-words consisting of CV syllables generated by a hierarchical recursive grammar, and were required to deliver grammaticality judgments over novel items. We propose that learning is facilitated most when SS works under other conditional cues, such as a skewed frequency distribution with simple stimuli being more numerous than complex ones (Poletiek & Chater, 2006)