This paper examines the washback effect of the Greek language test component, of the annual national university entrance examination in Greece. In order to explain the reasons behind student underachievement on the aforementioned language exam, we adopted a mixed methods approach combining (a) sample document analysis of national curricula and past examination papers from 2000 to date and (b) a quantitative and qualitative analysis of practice test papers by 4 focal students, produced during their final year at school. The results suggest that the assessment criteria produced by the Ministry of Education are rather vague and generic. They seem to target the ‘correct’ production of pseudo-genres with specific content and ideological agenda; implicitly, a single, ‘formal’ register is expected (indexed by the use of archaic morphology, subordination in the syntax, formal fixed expressions and vocabulary) and emphasis is placed on the ‘correct’ production of paragraph structure. Despite “teaching to the test”, progress in the students’ language skills examined in this study was only random. We suggest that this is ultimately an aspect of the washback effect as the criteria covertly implemented for exam preparation target a very narrowly defined subset of linguistic skills, presented out of context and taught through repetition, without honing (critical) metalinguistic awareness.
Keywords: test washback, EFL, document analysis, test papers, teaching to the test