I Think We Should… : Investigating Lexical Bundle Use in the Speech of English Learners Across Proficiency Levels

2.1. Spoken Bundles

The majority of comparative studies on the lexical use in SLA have been conducted on academic writing (e.g. Chen & Baker, 2010; Chen & Baker, 2016; Hyland, 2008). Of the relatively few recent studies that explored spoken lexical bundles, most have focused on contrasting native and native bundle use. Shirato and Stapleton (2007) compared the lexical bundle use of a NNS speech corpus with an established NS corpus, and found significant differences in the overuse and underuse of NNS vocabulary that deviate from NS norms including discourse markers and modal items. Sánchez Hernández (2013) examined 4-word lexical bundles in corpora of university students, both native and nonnative. In addition to structural and functional differences, the study revealed a larger number of lexical bundles in non-native speech. Comparing two non-native subcorpora during different stages of interaction, it was also found that classroom instruction can lead to increased use of lexical bundles (although no statistical tests have been performed).

One potential problem with the above approaches employing NS and NNS corpora is that such corpora are often collected in different contexts/registers with numerous uncontrollable variables, leading to reduced corpus comparability and difficulty in accounting for the observed differences (Gablasova et al., 2017a). From a developmental perspective, these approaches are also unable to provide insights into how NNS students of different proficiencies differ in the use of lexical bundles.