Exploring The Use of Formulaic Language By EFL Saudi Students: A Corpus- Based Study

Abstract

This corpus-based study investigates the use of three types of English formulaic language (FL) by Saudi EFL learners, focusing on multi-word verbs (MWVs), delexical verbs (DLVs), and collocations. It examines how learners use FL in their writing across different years of study and proficiency levels, and it also explores the relationship between learners’ receptive knowledge of FL and their productive use in writing.

This study adopted a corpus-based mixed-methods design to analyse data from 21 male Saudi university students in a Department of Languages and Translation at a Saudi university. Data were collected using the Oxford Placement Test (OPT), which was used to determine participants’ proficiency levels, and the Phrasal Vocabulary Size Test (PVST), which was used to measure learners’ receptive knowledge of FL. Learners also completed five written classroom assignments over a 12-week period. All the assignments were compiled into the Saudi Students Corpus (the SSC corpus). To examine productive use, every instance of MWVs, DLVs, and lexical collocations in the SSC corpus was identified and manually coded (including attempted uses), supported by computational tools (Sketch Engine and Lextutor) for profiling and consistency checks, and the English Vocabulary Profile (EVP) was consulted to assign CEFR levels to the coded FL items. PVST scores were then compared with the coded corpus data to examine the relationship between receptive knowledge and productive use of FL.

The analysis identified 1,035 FL tokens in the SSC corpus. Collocations accounted for most tokens (78.6%), followed by MWVs (18.6%), while DLVs were least frequent (2.8%). Overall, FL use showed a non-linear pattern across both years of study and proficiency levels. Output increased and reached its highest point in Year 3, then decreased slightly in Year 4, where use appeared more controlled. A similar pattern was found across proficiency levels: output rose overall but attempted use was most concentrated at B1, suggesting a transitional stage in which learners expanded their use of FL before showing greater control at B2–C1. A total of 119 attempted uses were coded, with word choice, word form, and preposition problems accounting for most cases.

In terms of qualitative patterns, MWV use relied largely on high-frequency, semantically transparent combinations, while figurative or idiomatic MWVs were rarely attested in the SSC corpus. Regarding the receptive–productive relationship, PVST scores rose across years, but the correlation between PVST and written FL use was strong in Years 1 and 2 (r = 0.93, 0.92) and weaker in Years 3 and 4 (r = 0.50, 0.38). When the PVST item list was compared with the FL items actually produced in the SSC corpus, overlap was minimal: seven tokens matched six PVST items, mostly MWVs, with one DLV appearing as an attempted transfer, and no PVST-listed lexical collocations attested.

Overall, this thesis shows how Saudi EFL learners use MWVs, DLVs, and collocations in university writing, and how this use changes across academic years and proficiency levels. The findings suggest a non-linear pattern, with a mid-programme transition around Year 3 (and, in the proficiency-based analysis, around B1) where learners produce more FL and more attempted forms, followed by more controlled and selective use in Year 4 and at higher proficiency levels (B2-C1). These results also suggest that learners may recognise FL items in the PVST without using the same items in open-ended writing, so teaching and assessment may need to link recognition with guided output practice, feedback, and repeated retrieval in academic writing tasks.