Therefore, each e-mail response and word-processed essay was examined for evidence of writers' attempts at providing context initially in their writing for the reader.
Thus, it was assured that the target audience was the same in both e-mail and word-processed situations, and divergent results would not be due to divergent audiences.
The potential influence of a practice effect was considered; in other words, the students' e-mail writing may have affected students' word-processed writing.
If all students had been present on all occasions for e-mail as well as word-processed writing for the three topics, a total of 42 essay pairs (14 e-mail and word-processed pairs for each topic) would have been obtained.
The present study represents an exploration into the differences between ESL students' e-mail and word-processed writing.
The tallies were then added and averaged for each group of e-mail and word-processing texts (per topic), then for all e-mail responses and word-processed essays together.
As a result, it was expected that different patterns of grammatical and lexical cohesion would emerge in e-mail and word-processed texts.
However, no obvious differences between e-mail and word-processed writing can be observed when the normalized averages for each feature are considered (see Appendix B).
These average frequencies, however, erroneously give the impression that all cohesive devices were used by all writers in their e-mail writing as well as their word-processed writing to similar extents.
Since both e-mail and word-processed writing took place in the context of an academic learning environment, and since the majority of participating students are Asian, one might in fact expect similarities in the use of cohesive devices across the two media.
An examination of whether or not students' e-mail writing contains more unsupported pronouns than their word-processed writing was not done in this study, but might warrant investigation.