Linguistics PhD student Mathieu Paillé gave an invited talk online at Humboldt University of Berlin on November 9th, entitled “Derivational morphemes exhaustify roots: a hypothesis on the relationship between language and concepts.”
Abstract: Non-scalar content vocabulary taken from a particular conceptual domain is usually interpreted as mutually exclusive, as seen in examples like #This comedy is a tragedy or #Some animated films are live-action (Paillé 2020; cf. Cann 2011). This has been variously dealt with as a fact about the structure of the lexicon (de Saussure 1916) or of conceptual space (Gärdenfors 2000). I begin by showing that the mutual exclusivity is in fact a product of grammar; indeed, it can be removed through conjunction or additive particles. As such, I speculate that it is the effect of a grammatical Exh(aust) operator (Chierchia et al. 2012). If this account is accepted, it comes with the consequence that the Exh found with these predicates displays novel behaviour. Not only is it obligatory (cf. e.g. Magri 2009), but it also, at first approximation, necessarily has the predicate in its immediate scope. To understand these requirements on Exh, I turn to another linguistic phenomenon that has the same twin properties of being obligatory with, and always local to, content vocabulary. This is derivational morphology. As discussed by Boeckx (2011), derivational morphemes take concepts (qua roots) and make them mergeable — i.e., linguistically usable. My proposal is that Exh’s unusual behaviour with content words is due to these very morphemes not just selecting a root/concept, but also requiring an Exh operator in their immediate vicinity. I formalize this through an Agree relation between derivational morphemes and Exh; this explains both the obligatory nature of Exh and its locality requirement, assuming there is no upward Agree. Thus, in effect, derivational morphemes ‘clean up’ underlyingly messy conceptual spaces, hiding away any overlap between related concepts.