Colloquium: Luke Adamson
September 30 @ 3:00 pm
Speaker: Luke Adamson (Rutgers)
Title: A noun's gender is locally determined: Evidence from gender and possession
What determines a noun’s grammatical gender? Often this question is posed in terms of how gender is ‘assigned’, with one answer being that a noun’s gender can depend on semantic criteria (e.g. animacy, sociocultural gender), nominalizing morphology, arbitrary lexical requirements, and possibly phonological criteria (Corbett 1991, Kramer 2020, a.o.). However, the conceptualization of gender features as having a syntactic locus, such as on a nominalizing head n (Kramer 2015, Adamson and Šereikaite 2019, a.o.), raises the possibility that we can define a domain of gender determination structurally. Previous work has demonstrated that covariation between number and gender is possible, where a noun’s gender appears to covary with number values, specifically when number is ‘low’, correlating with irregular morphology and/or semantics (Acquaviva 2008, Kramer 2016, a.o.), though not when number is ‘high’.
I propose that the number pattern should be subsumed under a more general hypothesis: gender determination of the noun is local, confined to an nP constituent low within the nominal. In this talk, I provide evidence in favor of this hypothesis from the interaction between gender and possessors, which like number features, have also been characterized by a ‘low’ vs. ‘high’ dichotomy, with low possessors being inalienable and high possessors being alienable (Myler 2016 and references therein). Building on previous observations from the descriptive literature, I show that inalienable possessors are implicated in gender alternations in several unrelated languages, including Teop (Austronesian, Oceanic), Jarawara (Arawan), Yanyuwa (Pama-Nyungan), and Coastal Marind (Anim). Alienable possessors are not implicated in any such alternations.
This work has several important implications which I discuss in the talk, as the hypothesis is restrictive in a way that rules out conceivable interactions: i) it rules out interactions between gender determination and features higher in the nominal domain, such as definiteness and case; and ii) it rules out putative cases of long-distance gender valuation, such as nominal predicate agreement with a subject (pace Kučerova et al 2021). The hypothesis fits into a more general pattern of domain-restricted feature interaction; for example, ‘low’ vs. ‘high’ distinctions have also been claimed to be relevant for ergative case splits (Legate 2017).
This event will be held in person in room 108 at the Linguistics Building.