Abstract

Frequency effects are powerful mechanisms of human cognition and learning, including language acquisition. The study explores the possible contribution of frequency effects on the acquisition of the simple past in English. The focus of the study was the distribution of simple past in recurrent multi-word utterances (RMUs) in both native speaker and second language learner speech. The objective was to determine whether there is an association between RMUs and the semantic categories of verbs in NS speech, and if so, whether this association would in turn predict the accuracy rate of simple past in L2 speech. The prediction, informed by the Aspect Hypothesis, was the association between RMUs and telic predicates, and, hence, their greater accuracy in L2 speech. The study examined the distributional frequency of semantic categories of verbs in descriptions of past weekend produced by 29 native speakers and 33 ESL learners, as well as the accuracy of past morphological markers for the learners. The findings suggest that telics are significantly more predominant in L1 RMUs than in non-RMUs. Telics also prevail in L2 speech, and they received most accurate markings in all past multi-word utterances produced by learners. As to the prediction of more accurate morphological markers for telic predicates in RMUs versus those in non-RMUs, only advanced proficiency learners showed significant advantage of RMUs. The findings are interpreted in relation to the research in formulaic sequences and the AH. The issues of verb type frequency and L2 proficiency need to be further investigated. Pedagogical implications for instruction are suggested in conclusions.