This new project seeks to investigate the role of gender in prominence hierarchies across different languages, modalities, and registers.

We primarily use corpora to account for the multi-dimensionality of prominence across different languages and registers in naturally occurring data. To this end, we combine multifactorial corpus-driven methods with Bayesian hierarchical mixed effects regression modelling to account for the non-independence of data points in corpora, inter-personal variation, and lexical effects. This approach allows us to model the extent to which gender interacts with multiple linguistic variables whilst also factoring in important contextual variables such as register, priming effects and, where such data is available, sociolinguistic variables such as the gender of the speakers/writers.

We examine the role of gender in hierarchies of prominence in a language with no grammatical gender (WP 1: English), one where grammatical gender largely matches social/referential gender (WP 2: French), and one with more than two grammatical genders (WP 3: German). With respect to the latter two, we will also investigate the role of grammatical as opposed to referential gender as a prominence lending feature by taking a closer look at referential chains headed by DPs with hybrid nouns, i.e., nouns where grammatical and referential gender (optionally or necessarily) diverge, e.g., das Mädchen and la sentinelle (WP 4), using a combination of corpus and experimental methods.

The final objective of our project is to develop a theoretical model (WP 5) that integrates probabilistic statistical modelling with insights from theoretical linguistics to describe how social and/or grammatical gender hierarchies interact with other implicational hierarchies in determining language users’ choice of morphosyntactic structures across different registers and languages, whilst factoring in all other potential relevant variables including other prominence-lending features, discourse-specific, and user-specific factors.