All Talks
All Talks
Fascinating new systems of nominal classification keep being found, but the tools for analysis have not kept pace. We therefore propose a typology of nominal classification, encompassing gender and the various classifiers. Earlier it made sense to oppose gender and classifiers (Dixon 1982), but the opposition cannot be maintained. Miraña has characteristics of gender and of classifiers (Seifart 2005); Reid’s (1997) account of Ngan’gityemerri provides further evidence against a sharp divide, since classifiers can grammaticalize into gender, through intermediate types.
Relinguishing the opposition gender vs classifiers allows a clearer picture of the possibilities. We pull apart traditional gender characteristics, and traditional classifier characteristics, and see that these characteristics combine in many ways. This motivates a canonical persprective: we define the notion of canonical gender, and use this idealization as a baseline from which to calibrate the theoretical space of nominal classification. This allows us to situate the interesting combinations we find.
According to the Canonical Gender Principle “each noun has a single gender value.” (Corbett & Fedden 2016: 503; cf. Dixon 1982). Under this principle there are two specific criteria:
Criterion “constant”: canonically a noun takes the same gender agreement in all domains; nouns taking different agreements (hybrids) are non-canonical. German comes close to being canonical, while Russion has many hybrids.
Criterion “lexical”: gender can be read unambiguously off the lexical entry. Nearest to canonical are strictly semantic systems, e.g. Bagvalal (male human / female human / other, Kibrik 2001: 64-66); then we find systems like Mawng (Singer 2016), where recategorization is readily available. Furthest from canonical are traditional numeral classifiers which can highlight different semantic aspects of referents.
Canonical agreement on which canonical gender is based, contributes helpful criteria:
Criterion “obligatory”:since canonical agreement is obligatory, this is a component of a canonical gender system, as in languages like Russian. Less canonical are those where agreement can be optional, as in Ngan’gityemerri (Reid 1997). And classifiers of various types are frequently optional.
Criterion “obligatory values”: in addition to agreement being obligatory, canonical systems use the most specific feature value associated with the controller; such systems include Mian. Less canonical, since it exhbits’superclassing’, is Jingulu (Meakins & Pensalfini 2016).
Criteriaon “orthogonal”: as with any morphosyntactic feature, gender is canonical to the extent that it is orthogonal to parts of speech. We find systems where all parts of speech show gender agreement (Archi), and at the other extreme, systems where agreement is limited: in North Ambyrm only the relational classifiers show relevant inflection (Franjieh 2016).
As we calibrate carefully, using these criteria, traditional gender and traditional classifier languages are close in some respects, more distant in others. This is exactly right: Russian, Archi, Mawng, and Ngan’gityemerri are profoundly similar in parts of their nominal classification system, and profoundly different in others. Our typology helps reveal the great diversity of systems (greater than suspected even a few years ago), and the shared phenomena that play out as variations on similar themes.