Traditionally, British English has been considered the norm for EFL teaching in Europe (Trudgill & Hannah, 1994:1); however, recent studies in the Swedish context have indicated a shift towards an increasing preference for American English (e.g. Modiano & Söderlund, 2002). The present quantitative corpus-based study investigates whether Swedish university students of English show a preference for British English or American English in terms of spelling, as well as to what extent they are consistent in their choice. The Stockholm University Student English Corpus (SUSEC), compiled in 2007, comprising essays from students of English linguistics and literature from four student levels (first through fourth semester), was examined and then compared to the student essays from the Swedish subcorpus of the International Corpus of Learner English (ICLE), compiled in the 1990s. The results of the Swedish subcorpora were subsequently compared to the results of the Bulgarian and Italian subcorpora of the ICLE to gain a broader perspective on preference and degree of consistency outside of Sweden.

The results showed a clear preference for British English spelling for all the investigated subcorpora. There was no statistically significant difference between the different student levels of the SUSEC, nor was there any statistically significant difference between the SUSEC and the Swedish subcorpora of ICLE (SWICLE); the preference seemed therefore not to have changed notably over the years separating the SWICLE from the SUSEC. When comparing the findings from the Swedish subcorpora to the Bulgarian and Italian data, an even stronger preference for British English was found in the Italian subcorpus, whereas the Bulgarian subcorpus showed no statistically significant difference compared to the Swedish results. In terms of consistency, the students were generally consistent in their use of one variety, although the results from the SUSEC showed a slightly higher incidence of inconsistent essays compared to the Bulgarian and Italian subcorpora; the difference between the SUSEC and the SWICLE was not statistically significant. There was, furthermore, no significant difference between the different student levels within the SUSEC. These findings could provide a basis for further studies on how advanced EFL learners actually use the language, and add to the discussion of which variety – or varieties – should be taught in schools (Levin, 2002:59).