Second language speakers' use of multi-word verbs in spoken communication : evidence from the Trinity Lancaster corpus

Marin Cervantes, Irene and Gablasova, Dana and McEnery, Tony (2019) Second language speakers' use of multi-word verbs in spoken communication : evidence from the Trinity Lancaster corpus. PhD thesis, Lancaster University.

Text (2019marin-cervantesphd)
2019marin_cervantesphd.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (2MB)

Abstract

The present study investigates L2 speakers’ use of multi-word verbs (MWVs) in spoken, interactive communication. It focuses on two specific types of MWVs: phrasal verbs (e.g. carry out, get up) and phrasal prepositional verbs (e.g. look up to, get along with). Its main aim is to use corpus methods in order to examine i) how L2 speakers use MWVs in terms of frequency, lexical verb and particle productivity, number and type of non-canonical MWV forms as well as polysemy, and ii) whether learner variables (i.e. L2 proficiency and L1 background) and situational variables (i.e. task type) mediate L2 speakers’ use of MWVs. The study draws on data from the Trinity Lancaster Corpus (Gablasova, Brezina & McEnery, forthcoming), a large-scale POS-tagged corpus of L2 spoken production consisting of 4.2 million running words. More specifically, the speech of 1,348 L2 speakers was analysed. The speakers were at three different levels of L2 proficiency (i.e. B1, B2 and C1-C2) according to the Common European Framework of Reference and came from three different L1 backgrounds: Chinese, Italian and Spanish. MWVs were automatically extracted via the Sketch Engine (Kilgarriff et al., 2014) using tags based on Corpus Query Language (CQL). The automatic extraction was followed by a manual scrutiny of the results to discard false positives. For the analysis of data, both descriptive and inferential statistics were used. With respect to overall MWV frequency patterns, the results indicated that there was a relatively low MWV representation in L2 production, with a few high-frequency MWVs accounting for the majority of MWV occurrences in the corpus. A variety of non-canonical MWVs were also found; most of these forms were one-off occurrences produced mainly by the most advanced L2 speakers. Four patterns of non-canonical MWV use were identified including, for example, the combination of a lexical verb with a redundant particle and the creation of novel MWVs. The study also provided evidence of a frequency-polysemy phenomenon in which the most polysemous MWVs that the L2 speakers produced featured among the most frequent MWVs in the corpus. In terms of the relationship between learner variables and L2 MWV production, the results showed that the main effect for L2 proficiency on MWV use was not statistically significant, which indicated that gains in proficiency did not seem to translate into a high MWV frequency. In contrast, L1 background was found to be a significant factor mediating MWV use. From the three L1 backgrounds studied, Chinese L1 speakers appeared to use more MWVs per thousand words than Italian and Spanish L1 speakers. Moreover, the results revealed a large inter-speaker variation across proficiency levels and L1 backgrounds, and they also provided evidence of topic effects on MWV use given that particular MWVs recurred in the context of specific topics. Regarding task-related variables, the effect of speaking task type on MWV production was found to be statistically significant. The only monologic, pre-planned task contained not only a higher number but also a wider range of MWVs than the dialogic, unplanned tasks that were analysed. The MWVs in the monologic task tended to function as part of transitions in the context of delivering oral presentations (e.g. now let’s move on to the next point). The same MWVs found in the monologic task also occurred in the dialogic tasks. However, rather than serving as part of transitions in those tasks, these MWVs expressed a variety of meaning senses. The findings of this study both complement and add to previous research on MWVs in L2 contexts, particularly by showing important trends of MWV use in a spoken, interactive context and across a variety of L2 proficiency levels, L1 backgrounds and speaking tasks. The study also contributes to broaden our understanding of the MWV knowledge that L2 users possess, which has been found to enhance L2 speakers’ communicative competence and facilitate both language production and comprehension.