High frequency collocations and second language learning

Abstract

This thesis explores the implications of high frequency collocation for adult second language learners. It addresses three main questions. First, it asks to what extent high frequency of occurrence in a corpus indicates that collocations are independently represented in the minds of native speakers. A word association study indicates that high frequency of occurrence is a fairly reliable predictor of mental representation, though this methodology does not allow us to determine the precise strength of the relationship. A series of lexical decision studies also show a relationship between frequency and representation, but effects are limited to those collocations which are sufficiently salient to also register as associates. This suggests that psycholinguistic 'priming' models may not be the best way of understanding collocation. Second, the thesis examines the idea that adult second language learners usually fail to retain the collocations to which they are exposed. This is tested through a lab-based training study and a learner-corpus study. Results suggest that adult learners are capable of learning collocations from input, but that 1) the relatively low levels of input to which most learners are exposed mean that they nevertheless tend not to attain native-like profiles of collocation use, and 2) input which provides repeated exposure to collocations can dramatically improve learning. Third, the thesis asks whether a useful pedagogical listing of frequent 'academic collocations' can be compiled. Results suggest that an academic collocation list is viable, but that important caveats need to be made concerning the nature of the collocations included and the range of disciplines for which such a listing will be useful. Moreover, listings of two-word collocations should be seen only as a starting point for more comprehensive phraseological listings. Suggestions will be made for ways in which we might go beyond such two-word listings.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 76 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.