A longitudinal learner corpus investigation of the use and development of formulaic sequences in argumentative essays of EFL learners

Abstract

The current thesis investigates the usage patterns of three- to four-word sequences in longitudinal learner corpora composed of two semesters written data from 85 English as a Foreign Language learners (EFL). The analysis is twofold: the first one is a group analysis that made available to observe collective trends in terms of usage patterns of formulaic sequences across different time intervals, and the second one is an individual analysis that assisted in observing the use of formulaic sequences of the same 8 individual learners across two semesters. Based on the frequency approach, the most frequent three and four-word recurrent formulaic sequences were extracted from each sub-corpus in two groups of longitudinal learner corpora and classified structurally and functionally. Then, the use of these sequences was compared across native (LOCNESS) and non-native corpora. The individual analysis involved identifying unique sequences used frequently by particular learners across two semesters observation in the five sub-corpora. Quantitative analysis of the longitudinal learner corpora findings was followed by a qualitative analysis of the learners' responses to retrospective protocols. The frequency analyses, correlation statistical test, and the structural and functional analyses of the formulaic sequences showed that the number and the range of FSs seemed to show an increasing pattern in number and type, as the learners were given more instruction and teacher feedback regarding their essays for each week during two semesters. Structurally, the majority of formulaic sequences were found to be verb phrase fragments, followed by noun phrase and prepositional phrase fragments. Functionally, most formulaic sequences analysed in this study include referential expressions, followed by stance expressions like in previous literature. The Pearson correlation test also showed that the frequent FSs in the two learner corpora seemed to moderately correlate, in Group 1, with the native learner corpora in the five sub-corpora unlike Group 2. The results of the individual analysis showed some formulaic sequences that were unique to the particular learners. In other words, more than half of the formulaic sequences shared word-for-word or partially with longitudinal learner corpora whereas they shared less formulaic sequences with native written corpus (LOCNESS). Last but not least, the results of the retrospective protocol showed that participants' exposure and awareness of FSs increased through teacher feedback, and they used more FS across time intervals.