ABSTRACT
There is a sense that personal pronoun use is somewhat non-academic and that to write successfully in an academic register one needs to carefully ration any use of pronouns. Pronouns are a closed class set of features, which means that (in contrast to open class items) we do not generally create new pronouns over time, and this means that we have a very well-defined set of items, an ideal situation for a corpus-based analysis. The issue of normalisation is important, and it is worth reflecting on how often this needs to be taken into account, particularly when working with naturally occurring language data. Since the length of the texts cannot be constrained by the researcher, there is inevitably variation. For a t-test to be reliable, both sets of data – the pronoun distribution for Genome Research and for the Journal of Modern History – should be normally distributed.
