Persistent errors in spoken English among Taiwanese and Czech learners at CEFR B2 and C1

Abstract

The study provides empirical evidence illustrating the development of grammatical accuracy in spoken learner English from CEFR's B2 to C1 levels of accuracy. It uses the CEFR-rated and error-tagged Taiwanese and Czech sub-corpora of LINDSEI containing altogether 54 B2 and 36 C1 speakers.

The analysis of the 5,120 identified errors revealed a threefold increase in accuracy at the C1 level. However, some errors which might be labelled persistent were identified, namely errors in some aspects of the use of articles and errors in the use of some verb tenses.

The study contributes to the discussion of criterial features for proficiency levels.

Keywords

• learner corpus
• computer-aided error analysis
• speaking accuracy
• LINDSEI
• CEFR