Errors that related to the lack of referent in the theses involved either an item where it did not have a presupposed item in the text, or a plural noun with a definite article where the intended reference was generic. The latter type was more characteristic of low rated theses. This may be due to a proficiency problem: the writers of the theses might not be aware of the fact that plural nouns with definite articles are not used for generic reference and that using them that way can be misleading. Frequent examples of such items were, for example, the students, the teachers, the EFL classes and the foreigners. The lack of a specific presupposed item was almost equally likely to occur in both corpora. Two main reasons might lie behind this. One is that thesis writers tended to write in very vague and general terms about the study they were to explain in detail in subsequent sections of their paper, for example, some/most of the cases without specifying what cases are being discussed; the other skills without a description what they might be; or the questionnaires, the interviews, the school without any specification. In contrast, RA writers (and high-rated MA thesis writers) specify all the crucial details (e.g., type of research, research tools, participants or the location) very early in the article, either in the abstract or the introduction sections. The reason for not specifying such details could be the writers’ over-reliance on their audience’s familiarity with their research – it appeared that they only had their supervisor as an audience in mind.