Linguistic Discovery (Jan 2016)
Exploring Grammatical Complexity Crosslinguistically
Abstract
This paper proposes a set of principles and methodologies for the crosslinguistic investigation of grammatical complexity and applies them to the in-depth study of one grammatical domain, gender. The complexity of gender is modeled on the basis of crosslinguistically documented properties of gender systems and by taking into consideration interactions between gender and two other grammatical domains: nominal number and evaluative morphology. The study proposes a complexity metric for gender that consists of six features: “Gender values”, “Assignment rules”, “Number of indexation (agreement) domains”, “Cumulative exponence of gender and number”, “Manipulation of gender assignment triggered by number/countability”, and “Manipulation of gender assignment triggered by size”. The metric is tested on a sample of 84 African languages, organized in subsamples of genealogically related languages. The results of the investigation show that: (1) the gender systems of the sampled languages lean towards high complexity scores; (2) languages with purely semantic gender assignment tend to lack pervasive gender indexation; (3) languages with a high number of gender distinctions tend to exhibit pervasive gender indexation; (4) some of the uses of manipulable gender assignment are only attested in languages with a high number of gender distinctions and/or pervasive indexation. With respect to the distribution of the gender complexity scores, the results show that genealogically related languages tend to have the same or similar gender complexity scores. Languages that display exceedingly low or high gender complexity scores when compared with closely related languages exhibit distinctive sociolinguistic profiles (contact, bi- or multilingualism). The implications of these findings for the typology of gender systems and the crosslinguistic study of grammatical complexity and its distribution are discussed.
Keywords
