A Recurrent Word Combination Approachto the Study of Formulae in the Speech of Native and Non-Native Speakers of English

This article reports on a pilot study into how corpus methods can be applied to the study of one type of phraseological unit, formulae, in native speaker and learner speech. Formulae, or formulaic expressions, are multi-word units performing a pragmatic and/or discourse-structuring function and have been characterised as being typically native-like. The methodology presented here is contrastive and involves the use of computerised corpora of both native and non-native speaker speech. It consists of two steps: (1) the automatic extraction of all recurrent word combinations to produce lists of potential formulae, and (2) a carefully specified manual filtering process designed to reduce these lists to lists of actual formulaic usage. The results of this process allow for the first genuine quantitative comparison of formulae in the speech of native and non-native speakers, which in turn has significant implications for SLA research. This paper focuses on methodology and does not present a full discussion of the results. However, selected example findings are presented to support the approach adopted.