A COMPUTATIONAL APPROACH TO TEXT COMPLEXITY & EXAM PERFORMANCE: IMPLICATIONS FOR EFL LEARNERS
The aim of the present paper is to delineate a range of linguistic features that characterize the reading comprehension texts used at the B2 (Independent User) and C1 (Proficient User) level of the Greek State Certificate of English Language Proficiency (KPG) exams in order to examine whether specific text variables have an impact on actual KPG test-takers reading comprehension exam performance. Although a lot of research has been conducted in the field of second language acquisition with specific reference to ways of reading and text processing strategies, Alderson (2000: 104) stressed language testers' lack of success "to clearly define what sort of text a learner of a given level of language ability might be expected to be able to read or define text difficulty in terms of what level of language ability a reader must have in order to understand a particular text". Such information would be particularly useful in providing empirical justification for the kinds of reading texts test-takers sitting for various language exams are expected to process, which to date have been arrived at mainly intuitively by various exam systems. By making use of advanced Computational Linguistics and Machine Learning systems, the present study has, thus, been designed to fill this void and further add to our present state of knowledge on English as a Foreign Language (EFL) text difficulty in general. To this end, the impact of 135 text variables on the mean reading performance per text and per level of competence of a total number of 152,039 B2 and 36,517 C1 test-takers that had taken part in the KPG English language exams in English over a period of 10 years was investigated. The findings of the study provide practical guidance to EFL learners, teachers, material developers and test designers as to the kind of lexicogrammatical features a learner of an expected level of language ability might be able to handle for a successful exam performance.