A longitudinal corpus of Swedish university students’ written English, some problems and some results

To understand how writing skills are acquired requires longitudinal studies, which should focus not only on correctness, but also on style and vocabulary development. In this project I got first-semester students in the English Department of Stockholm University to write early and late in the semester on the same topic. This gave me two samples which should only have differed due to development in the subjects’ proficiency. To analyse them I made use of the facilities in the concordancing program Wordsmith Tools (Scott 1996). The first set of essays contained more different word-types than the second., probably because the writers later on had more uniform ideas of what was expected and perhaps were trying less hard to be original. The code used in the second essays is less personal. This is partly because of changes in the content, but the form of expression is also less personal, with, for example, fewer instances of I think, an index both of less personal writing and of a more written style. The second essays also conform more closely to the superficial conventions of the academic written code, changes which make the writing more acceptable to an international adult audience. Only two features reflect any deeper development: the small increase in discussion of what text means, and the increasingly selective use of we, which starts to focus on the key function of persuading the audience. There are a number of problems with the comparability of the samples and with the nature of the essays, but this approach could be combined with more naturalistic longitudinal studies to provide more reliable and valid findings than either could produce on its own.