Multiword units in academic writings by native speakers of English and Turkish EFL learners: A lexical bundles approach

Corpus linguistics has clearly demonstrated that authentic language use consists of repeated expressions, namely formulaic expressions (Byrd & Coxhead, 2010). These expressions, which can also be described as multiword units, serve as essential "building blocks of discourse in both spoken and written registers" (Biber & Barbieri, 2007, p. 263). Formulaic sequences constitute a large portion of oral and written language production (Schmitt & Carter, 2004) and "provide a steppingstone into language development" (Chenu & Jisa, 2009, p. 27). Being an important type of formulaic sequences, lexical bundles are "recurrent expressions, regardless of their idiomaticity, and regardless of their structural status" (Biber et al., 1999, p. 990). The current study investigates the use of the most frequent 3- and 4-word lexical bundles in the Turkish International Corpus of Learner English (TICLE) which is sub-corpus of the International Corpus of Learner English (ICLE) and Louvain Corpus of Native English Essays (LOCNESS). The TICLE corpus contains written argumentative essays by Turkish EFL learners, while the LOCNESS corpus comprises American students' written essays (Granger et al., 2009). The study employs the corpus-driven approach (Sinclair, 1987), also referred to as frequency-based approach (Nesselhauf, 2004). The most frequent expressions have been classified according to their structures and functions in the discourse using Biber et al.'s (2003; 2004) structural and functional taxonomy of lexical bundles framework. An interpretative contrastive analysis of these expressions has been conducted between Turkish EFL learners and native English speakers. And lastly, semantic domains of the multiword expressions in both corpora have been revealed using UCREL Semantic Analysis System (USAS) semantic tag sets utilizing Wmatrix (Version 4) semantic analysis platform (Rayson, 2008) to identify the most salient features of the multiword choices of learners. The study concludes that Turkish EFL learners overuse verb phrase fragments while they underuse the noun phrase and prepositional phrase fragments in relation to the native speakers of English. In terms of the functional classification, even though Turkish EFL learners produce less functional bundles, they overuse a limited set of repetitive bundles. Log-likelihood results also indicate a lack of lexical variety and richness in TICLE corpus in relative to the LOCNESS corpus. Additionally, semantic analysis of these phrases uncovers both common and specific domains uniquely used in both corpora. Current study adds our understanding of the use of English lexical bundles by Turkish EFL learners and offers valuable insights for SLA researchers, English language teachers, instructors and material developers and publishers.