Session 5F

Collocations, pairs of words that tend to occur together, have been considered as an important aspect of L2 proficiency (e.g., Durrant & Schmitt, 2009; Paquot, 2019). Studies have generally suggested that L2 users, even at high proficiency, use erroneous collocations, and underuse less frequent, strongly associated ones (e.g., Durrant & Schmitt, 2009). Although research has repeatedly shown the importance of collocation, two issues need to be addressed to gain a more comprehensive understanding of L2 collocation production. First, there are very few studies that simultaneously examine collocations in different syntactic relationships, including verb-adverbial modifier and subject-verb (see Paquot, 2019). Second, due to the growing use of average association scores (e.g., mean Mutual Information) in recent studies (e.g., Paquot, 2019), the resulting snapshots of collocation use may have been obscured; that is, average scores discard the information about the spread of collocation strengths within individuals. This study aims to address these two issues. First, three types of collocations were investigated through conducting dependency analysis; namely, verb-direct object, verb-adverbial modifier, and noun-adjectival modifier (Paquot, 2019). Second, learner-produced collocations were classified into a total of six bins according to their collocation strengths in a large-scale reference corpus (COCA; Davies, 2009). Subsequently, the number of collocations in these categories, rather than a single average score, were used to examine how distributional profiles of L2 collocation would be associated with L2 proficiency. The current study used the International Corpus Network of Asian Learners of English (ICNALE; Ishikawa, 2018), which provides 680 essays written by 320 learners (A2–B2 levels on CEFR). ICNALE also provides expert ratings on vocabulary use for each essay, which was used as an indicator of proficiency. Collocations in the learner essays were identified using a dependency tagging model provided by spaCy (Explosion AI, 2016). More specifically, the collocations were classified into 6 categories based on the percentiles of Mutual Information, and the number of collocations in each category was tallied for each essay. Statistical analyses were conducted through multinomial count regression for each of four collocation types to see how proficiency score would predict the collocation profiles. Results highlighted three points. First, all types of collocations were meaningfully associated with proficiency. Second, as a general tendency, a high proficiency is associated with a) decreased proportions of non-collocated pairs (e.g., Negative MI ), b) increased proportion of relatively weak collocations (e.g., 1 ≤ MI < 3), and c) increases of highly associated pairs (MI > 3) at higher proficiency levels. The results would complement the current state of research on the relationships between proficiency and statistical properties of collocation. Potential reasons for the differential effect of proficiency on L2 collocation production will be discussed. 
Davies, M. (2009). The 385+ million word Corpus of Contemporary American English (1990–2008+): Design, architecture, and linguistic insights. International Journal of Corpus Linguistics, 14(2), 159–190. 
Durrant, P., & Schmitt, N. (2009). To what extent do native and non-native writers make use of collocations? IRAL - International Review of Applied Linguistics in Language Teaching, 47(2). Explosion AI. (2016). SpaCy (Version 2.0) [Computer software]. https://spacy.io
Ishikawa, S. (2018). The ICNALE Edited Essays; A dataset for analysis of L2 English learner essays based on a new integrative viewpoint. English Corpus Studies, 25, 117–130.
Paquot, M. (2019). The phraseological dimension in interlanguage complexity research. Second Language Research, 35(1), 121–145.

Collocations are a crucial feature of second/foreign (L2) language proficiency (e.g. Crossley, Salsbury, & Mcnamara, 2014), but they pose considerable challenges for L2 learners (e.g. Laufer & Waldman, 2011). A few studies have investigated incidental learning of collocations (e.g., Pellicer-Sánchez, 2017; Szudarski, 2012; Webb, Newton, & Chang, 2013), but previous studies were often one-off interventions, produced inconsistent findings and did not examine learners' perspectives. Additionally, little is known about learning collocations in different reading modes. In light of those research gaps, the present longitudinal study was conducted. Our longitudinal study compares the effects of three reading modes: reading-only, reading-while-listening, and reading with textual input enhancement (i.e. underlining) on learners' learning of collocations. Second, our study also focuses on the effect of frequency of occurrence and learners' vocabulary size on learning collocations. One-hundred pre-intermediate Vietnamese EFL learners participated in the study. We adopted a pretest-posttest, mixed design in which the three conditions (reading only, reading-while-listening, and reading with textual input enhancement) were counterbalanced across participants. A control group (n = 29) that was not exposed to the input was used to control for a test effect. Participants in the experimental group read three graded readers, each in a different condition, over nine weeks. Thirty-two target collocations were selected from the input: 16 verb-noun and 16 adjective-noun collocations. Learning gains were measured at the level of form recall. Prior to the treatment, participants took the form recall pretest and the updated Vocabulary Levels Test (Webb, Sasao, & Ballance, 2017). After the treatment, they took the form recall posttest and completed questionnaires as well as semi-structured interviews to give their perspectives on the learning treatment. A logistic regression with repeated measures was used to analyse the data from the posttest. The results showed that the experimental group performed significantly better than the control group (there was basically no learning in the control group). Second, reading mode also affected the learning of collocations. Reading with textual input enhancement resulted in significantly higher learning gains than the other reading modes. Reading-while-listening was also more beneficial for learning than reading only. In line with previous research, the findings showed that learners with a larger vocabulary size picked up more collocations. However, frequency of occurrence in the input was not related to the learning gains. In terms of learners' perspectives, the participants in the experimental group showed mixed attitudes toward different reading modes. This presentation will conclude by zooming in on the theoretical and pedagogical implications of these findings. References: Crossley, S. A., Salsbury, T., & Mcnamara, D. S. (2014). Assessing lexical proficiency using analytic ratings: A case for collocation accuracy. Applied Linguistics, 36(5), 570-590. Laufer, B., & Waldman, T. (2011). Verb?noun collocations in second language writing: A corpus analysis of learners' English. Language Learning, 61(2), 647-672. Pellicer-Sánchez, A. (2017). Learning L2 collocations incidentally from reading. Language Teaching Research, 21(3), 381-402. Szudarski, P. (2012). Effects of meaning-and form-focused instruction on the acquisition of verb-noun collocations in L2 English. Journal of Second Language Teaching & Research, 1(2), 3-37. Webb, S., Newton, J., & Chang, A. C. S. (2013). Incidental learning of collocation. Language Learning, 63(1), 91–120. Webb, S., Sasao, Y., & Ballance, O. (2017). The updated vocabulary levels test. ITL-International Journal of Applied Linguistics, 168(1), 33-69.