A cross-task investigation of lexical bundles in L2 speech
(2025) Register and task variation in Learner Corpus Research (VAR4LCR) conference — Location: Université catholique de Louvain (7.July.2025)
Files
No attached file found for this publication.
Details
- Authors
Granger, Sylviane 
- Abstract
- Lexical bundles are recurrent sequences of contiguous words such as on the other hand, as a result of or it is important to that play a crucial role in shaping spoken and written discourse (Biber et al. 1999). Their use has been extensively studied in learner corpus research (LCR). The dominant methodology has consisted in analysing and contrasting the lexical bundles used by L2 learners and native speakers, thereby uncovering patterns of overuse, underuse, and inappropriate use among learners, even the most advanced ones (Chen & Baker 2016; Hasselgård 2020; Lu & Deng 2019). The large majority of lexical bundle studies in LCR have relied on corpora of L2 writing. With a few exceptions (e.g. De Cock 2004; Hougham et al. 2024), research on spoken lexical bundles has been very limited so far. In addition, the influence of task type on lexical bundle use has been significantly neglected. Our study aims to address this twofold gap by exploring the use of lexical bundles in different tasks included in the Louvain International Database of Spoken English Interlanguage (LINDSEI; Gilquin et al. 2010). LINDSEI contains informal interviews with intermediate to advanced learners of English as a foreign language from various mother tongue backgrounds. The interviews in LINDSEI are made up of three tasks: a monologic narrative based on a set topic (an experience that taught them a lesson, a country that impressed them, or a film or play they liked/disliked), a free discussion about students’ lives, and a short picture description (based on the same sequence of pictures). The LINDSEI data have been marked up to make it possible to study the use of linguistic phenomena across the three different tasks. Six L2 English varieties from LINDSEI are investigated: three from learners with Romance mother tongue backgrounds (LINDSEI-French, LINDSEI-Italian, and LINDSEI-Spanish) and three from learners with Germanic mother tongue backgrounds (LINDSEI-Dutch, LINDSEI-German, and LINDSEI-Swedish). Each variety represents 50 interviews and between c. 60,000 and 90,000 words of interviewee speech. In addition, the Louvain Corpus of Native English Conversation (LOCNEC; De Cock 2004), a comparable corpus containing the same type of interviews and the same three tasks but with L1 English students, is used as a reference corpus when relevant. Three subcorpora per variety are created, corresponding to each of the three tasks. Three-word lexical bundles are retrieved by means of LancsBox (Brezina et al. 2021), applying thresholds of 5 tokens (for frequency) and 3 interviews (for range). We measure the degree of ‘bundleness’ (De Cock & Granger 2021) of the different tasks as well as the degree of overlap between lexical bundles across tasks. The shared and task-specific lexical bundles are investigated as to their functions (relying on the taxonomy proposed by Biber et al. 2004) and their structures (relying on Altenberg 1998). Attention is also paid to lexical bundles that are prompt-based or topic-dependent (cf. Chen & Baker 2010). The main focus is on a qualitative analysis of the lexical bundles.
- Affiliations
Citations
De Cock, S., Gilquin, G., & Granger, S. (2025). A cross-task investigation of lexical bundles in L2 speech. Register and task variation in Learner Corpus Research (VAR4LCR) conference, Université catholique de Louvain. https://hdl.handle.net/2078.5/274951