Framing vs. supporting evidence in L2 argumentative writing: a mixed-methods study of Chinese EFL learners
BackgroundEvidence integration is central to argumentative writing, yet the relationships between different evidence functions and L2 writing quality across proficiency levels remain under-examined.MethodsUsing an explanatory sequential mixed-methods design (QUAN → qual), we analyzed 542 classroom,...
Uloženo v:
| Vydáno v: | Frontiers in psychology Ročník 16 |
|---|---|
| Hlavní autor: | |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
Frontiers Media S.A
01.11.2025
|
| Témata: | |
| ISSN: | 1664-1078 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | BackgroundEvidence integration is central to argumentative writing, yet the relationships between different evidence functions and L2 writing quality across proficiency levels remain under-examined.MethodsUsing an explanatory sequential mixed-methods design (QUAN → qual), we analyzed 542 classroom, timed English argumentative essays by Chinese undergraduates (30 min; 120–180 words). Texts were functionally coded as framing evidence (constructing the inferential scaffold) and supporting evidence (verifiable data, examples, expert attribution, etc.). Inter-rater reliability for the evidence scheme was high; writing quality was represented by standardized rubric scores. We ran ordinary least squares (OLS) regressions stratified by proficiency (high/mid/low) to test whether evidence type predicted scores, and used qualitative close readings to illustrate typical evidence–reason coupling.ResultsOverall, framing evidence predominated. The mid-proficiency group showed the most balanced framing–supporting configuration; the low-proficiency group was weak on both types of evidence. Stratified regressions indicated that only in the mid-proficiency group did evidence type significantly predict writing scores (β ≈ 0.40, 95% CI ≈ 0.19–0.62); effects in other groups were not robust, and model fit was modest (low–moderate R2).ConclusionThe findings suggest a developmental shift from “having evidence” to “using evidence well.” Once writers can supply basic evidence, further gains in quality hinge less on adding types or quantity and more on selecting precise evidence, explaining it clearly, and aligning it tightly with the claim—that is, achieving functional fit and linking through explicit warrants. Instruction and assessment should therefore pivot from “whether/how much evidence” to how evidence is selected, interpreted, and embedded in the inferential chain. |
|---|---|
| ISSN: | 1664-1078 |
| DOI: | 10.3389/fpsyg.2025.1705232 |