Comparing cohesive devices: a corpus-based analysis of conjunctions in written and spoken learner discourse

1.The aims of this paper are: to describe the process of developing two learner corpora of English at Sofia University; to mention some problems associated with the collection of the data as well as some of the applications of corpora of this type, and to present the findings from the research carried out so far. 2. The International Corpus of Learner English (ICLE) project, launched at the University of Louvain in Belgium to compile an international corpus of written learner language was joined by the Bulgarian team in 1996. To conform to the requirement for comparability all the teams participating in the project are collecting the same type of data, the only difference between the sub-corpora being the learners ’ mother tongue. At present the Bulgarian sub-corpus contains about 112 000 words, which is 55 % of the total amount of data required, and work on it is still going on. The data consist largely of argumentative essays written by Sofia University students at the beginning of their second academic year, so they can be described as adult advanced learners of English. Each of the 112 learners contributed about 1000 words on one or two of the essay topics suggested by the ICLE members. In 1995, a complementary project was conceived in Louvain to compile a corpus of spoken learne