Accuracy across proficiency levels : insights from an error-tagged EFL learner corpus
(2012)
Files
PHDJenniferThewissen.pdf
Closed Access - Adobe PDF
- 3.26 MB
Details
- Authors
- Supervisors
- Granger, Sylviane
- Abstract
- (en) This PhD thesis zooms in on the construct of accuracy among intermediate to advanced learners of English as a Foreign Language. Its overarching aim is to trace the behavioural development of errors across four levels of proficiency, viz. B1 (lower-intermediate), B2 (upper-intermediate), C1 (advanced), and C2 (mastery), as defined by the Common European Framework of Reference for Languages (henceforth CEF) (Council of Europe, 2001). Concretely, accuracy developmental profiles have been identified in terms of specific areas of progress, stabilisation and regression across the above-mentioned levels. Adopting such a developmental focus has helped fill a small part of the void in second language acquisition research which, to this day, has tended to favour the cross-sectional study of learner language. The methodology adopted in this PhD is based on data taken from the International Corpus of Learner English (henceforth ICLE) (Granger et al., 2009). A total of 223 learner argumentative essays were selected from the French, German, and Spanish components of ICLE, amounting to c. 150,000 tokens altogether (i.e. c. 50,000 words per L1 group). Each text was submitted to a two-tiered system of analysis: the error tagging phase involved annotating each script for errors on the basis of 46 distinct error types while the CEF rating phase assigned one final CEF score (either B1, B2, C1 or C2) to each essay. Following these two steps, each text was accompanied by a detailed error profile and a CEF score, hence enabling the study of errors from a developmental perspective. The error counting method adopted in our work, i.e. potential occasion analysis, involved counting the error types at each level out of a more refined denominator than de-facto the total number of words (e.g. the noun-number errors were counted out of the total number of nouns used in each text). The figures thus obtained were subsequently submitted to an ANOVA statistical test to identify areas of significant progress, stabilisation and regression. The results yielded by the developmental error analyses have implications for second language acquisition research as well as for language testing research. In terms of second language acquisition, the developmental trends pinpointed here revealed that EFL accuracy development from B1 to C2 is mainly characterised by a mixture of progress and stabilisation tendencies, with regression being rarer. The behavioural trends found in the learner corpus sample were subsequently compared with the development implied in the CEF descriptors for grammatical accuracy, vocabulary control, orthographic control, coherence and cohesion. It was revealed that using the development attested by learner corpus data could lead to stronger, more concrete language descriptors for specific learner groups at given proficiency levels.
- Affiliations
Citations
Thewissen, J. (2012). Accuracy across proficiency levels : insights from an error-tagged EFL learner corpus. https://hdl.handle.net/2078.5/45760