Sexual dimorphism in language, and the gender shift hypothesis of homosexuality

Abstract

Psychological sex differences have been studied scientifically for more than a century. They are also the focus of much current research, debate, and public interest. Sex difference research is conducted at increasing rates in psychology, behavioural science, neuroscience, genetics, and biomedical science, but the way in which psychological sex differences may be reflected in language use has received surprisingly limited attention. Psychologically informed corpus linguistic studies on sex differences are scant, while such studies focusing on sexual orientation are virtually non-existent. The aim of this thesis is to connect and synthesise contemporary research in psychology, cognitive science, developmental and behavioural neuroscience, evolutionary science, and computational linguistics by analysing sex differences and sexual orientation differences in a corpus of 694 novels, and comparing the findings with existing research from those fields. Extending the study of psychological sex differences and sexual orientation differences to literary fiction reveals whether contemporary sex difference findings in psychology and cognitive neuroscience have equivalents in novels that stretch the study period from the present day back to the beginning of the 19th century. This approach has the added benefit of tapping into psychological material which is vastly different from that used in most psychological research. I made predictions for sex differences and sexual orientation differences based on prior research, both theoretical and empirical, conducted in the abovementioned research fields, and used Linguistic Inquiry and Word Count (LIWC 2015) to extract quantitative psycholinguistic data from a corpus of literary texts comprising 66.9 million words. I analysed these data using multilevel hierarchical models to account for non-independence of novels by the same author and to control for variation in publication year and authors’ age. The results indicate the existence of substantial psycholinguistic sexual dimorphism in a sample of canonical and prize-winning English-language novels (n = 304) written by heterosexual authors, spanning more than two centuries, in a way that is consistent with the theoretical predictions which arose from the multidisciplinary research synthesis. The psycholinguistic sex differences in this sample of novels largely align with known psychological sex differences, such as empathising–systemising, and people–things orientation. A few unexpected results include the higher use of verbs and cognitive process words by female authors, which was not predicted by theory. Canonical and contemporary novels (n = 158) written by lesbian authors show signs of psycholinguistic masculinisation, particularly with anger-related words (d = 0.52) and sexual words (d = 0.42), when compared with novels by heterosexual female authors. Novels (n = 167) written by homosexual men are psycholinguistically feminine, showing a significant shift from male-typical patterns of language use towards a female-typical pattern of writing. This study extends prior research from behavioural, psychological, and evolutionary sciences to 19th to 21st century literary art, finding confirmatory evidence for psychological sex differences and sexual orientation differences using psycholinguistic data extracted from a large corpus of novels. The added benefit of the multidisciplinary approach adopted here is that it connects research from corpus linguistics, computerised text analysis, and biocultural literary theory with psychological science, evolutionary science, and developmental, cognitive, and behavioural neuroscience, thus providing a broader, more convincing, and more evidence-based research synthesis than is possible with any unidisciplinary approach. In the aggregate, literary Big Data gathered with the help of computerised text analysis feeds back into thinking about literature as human behaviour and how cultural products are shaped by biological factors, such as neurodevelopmental sexual differentiation of the brain, which have deep roots in mammalian evolutionary history.