Grammatical collocations in English exam texts written by Czech secondary-school students

Abstract

This paper focuses on the frequency and accuracy of five types of grammatical collocations G8E-G8I (Benson et al. 1986) in CZEMATELC, an English language learner corpus (8,338 types; 211,503 tokens) consisting of 1,841 English exam texts from the written part of the national school-leaving exams between 2015 and 2019. The findings reveal the prevalence of A1-A2 CEFR level colligations relying on a limited number of verb lemmas, a wide incorrect pattern variation and preference for patterns which are also the most frequent patterns of their Czech equivalents.