This paper provides evidence that there is a link between pronominal choice and the linguistic context. Thus, the variant they is generally preceded by quantifiers in both types of writing, whereas the pronominal forms he and he or she are mostly used with definite and indefinite antecedents.
Although the emergence of singular they as the predominant form and the decline of generic he as the normative form can be well argued for, specific linguistic contexts are likely to trigger the use of certain epicene forms and not others. For example, he is unlikely to occur with quantifiers and they is less frequently used with definite or indefinite antecedents.
Also, the occurrence of non-referential antecedents of the type a reader or the advertiser, which can be subject to multiple interpretations, contributes to the problematic of pronominal choice and therefore to the complexity of analytical scrutiny. Equally complex for the analysis is the treatment of those nouns or antecedents which are stereotypically interpreted as more masculine or more feminine, and those which tend to be understood as more masculine than feminine, even though they are epicene.
This is somehow reflected in the fact that there is no clear uniformity of pattern regarding the overall choice of pronoun in either of the two corpora of written language. In the newspapers analysed they is the most favoured option, followed by he or she and he, and it is the preferred form especially with quantifiers and with indefinite antecedents. However, as has been mentioned above, the large number of quantifier-specified antecedents found in the texts may account for the high frequency of they in the whole corpus. Normative he, while clearly dispreferred in journalistic writing, is the first option in the social sciences textbooks, although the proportions of the other variants are very similar.
Against this background, it seems logical to interpret the link between the choice of epicene pronoun and the type of genre as an indirect relation where certain types of antecedents may be more or less favoured by one genre or the other. So the preference for a particular type of antecedent is what ultimately appears to influence pronominal choice. However, a word of caution is in order here as journalistic and academic genres should not be regarded as homogeneous forms of communication but rather as text- type categories within which it is possible to find different types of subgenres that may exploit specific rhetorical devices that can be linked to specific communicative purposes.
The tendencies revealed in our study could be explained in the light of more general patterns of linguistic and social behaviour. So, we could argue that newspapers are more readily and more likely to reflect ongoing changes in society as they are aimed at a general audience whereas academic writing is a far more formal type of communication aimed at a particular discourse community and traditionally regarded as a male enterprise.
Future research should provide a more detailed analysis of the use of avoidance strategies such as pluralisation and the repetition of noun phrases, and this analysis should include statistical procedures in order to corroborate the significance of the findings.
Moreover, other written and even spoken language styles are worth exploring with a view to examining stylistic variation in the use of epicene pronouns. Finally, given the nature of the topic, future studies should focus on gender as an independent variable in order to find out whether the choice of pronouns made by male authors differ from the choice of pronouns made by female authors. This line of inquiry may prove fruitful as genres are socially constructed in part through association with the gender of their producers.