Vol. 36:2 (2023) ► pp.436–466
Vol. 36:2 (2023) ► pp.436–466
Constructional complexity as a predictor of Korean EFL learners’ writing proficiency
Based on usage-based constructionist approaches to language development, this study investigates the validity of constructional complexity as a predictor of writing proficiency of Korean EFL learners. We analyzed argumentative essays produced by Korean EFL learners and compared a prediction model that uses English argument structure constructions with a conventional T-unit-based model. We first tested the predictive power of a discriminant function model with argument structure constructions as predictors for assessing writing proficiency of lower- and higher-level learners. We then compared the construction-based model to another model that included T-unit measures as predictors. The results validated the contribution of argument structure constructions for predicting L2 writing proficiency: The construction-based model yielded higher prediction accuracy than the T-unit-based model did, confirming that constructional complexity is closely aligned with writing proficiency. These findings demonstrate the significance of a construction-based model as a reliable tool for diagnosing EFL learners’ writing proficiency.
Article outline
- 1.Introduction
- 2.Syntactic complexity measures in L2 writing
- 3.Constructional approaches to language development
- 4.L2 acquisition of English ASCs
- 5.Present study
- 5.1Research question 1: Constructional model
- 5.1.1Method
- 5.1.1.1Data collection
- 5.1.1.2Variable selection and coding
- 5.1.2Results and discussion
- 5.1.2.1Multivariate analysis of variance
- 5.1.2.2Discriminant function analysis
- 5.1.1Method
- 5.2Research question 2: Testing model validity
- 5.2.1Method
- 5.2.2Results and discussion
- 5.2.2.1Multivariate analysis of variance
- 5.2.2.2Discriminant function analysis
- 5.2.2.3Comparison of the models
- 5.1Research question 1: Constructional model
- 6.General discussion and conclusion
-
References
https://doi.org/10.1075/resla.20066.kim