Speaker: Boban Arsenijević (University of Graz)
Title: Deriving lexical categories: valued and unvalued classifiers and classifier-relativization
Date: Thursday 9 April
Venue: Online (sign up for our mailing list for details)
Time: 15.15 – 16.30 hrs
I tackle the status of the controversial lexical categories: adjectives (A), adverbs (Adv) and prepositions (P), and argue that they make one lexical category together, which I label PAd. This yields a system with only three lexical categories: nouns (N), verbs (V) and PAds. Unlike Ns and Vs, which establish reference to entities in general and eventualities, respectively, PAds are characterized by the inability to refer.
Considering that adverbs are usually classified with adjectives (Cook & Newson 1988, Radford 1990), my focus is on manifesting the identity between prepositions and adjectives, deriving the apparent differences from the fact that Ps are semantically light Pads, while As (and Advs) have acquired richer semantic content by incorporating their complement.
I propose a model of lexical categories coached in Distributed Morphology (Halle & Marantz 1993), where roots are fully intensional, and that the only way for them to obtain extensional semantics is to combine with functional material: The only functional feature that may combine with a root is the classifier feature. The classifier feature itself supplies the capacity for extension, but leaves the expression intensional. A potential value carried by this feature further restricts the way of reference of the expression, in particular regarding the ontological class and the properties of quantity, and determines the extended projection of the expression. This provides the expression with an extension (within a given reference domain). The value of the classifier feature is either lexically specified on the classifier (as in count classifier languages), or supplied by gender for Ns, i.e. by lexical aspect for Vs. An unvalued classifier feature yields a PAd. It needs to move to the left edge – effectively acting as a classifier-relative pronoun. It thus turns the expression into a predicate (hence Aps, AdvPs and PPs are all predicates), and receives a value from the head of the classifier-relative (i.e. the modified expression).
The proposed view straightforwardly captures the similarities in distributions and meanings between adjectives and PPs. The fact that the set of syntactic environments in which adjectives occur is a subset of those where PPs are found is explained by the fact that the incorporation of the complement through which adjectives are derived is subject to certain restrictions – syntactic, semantic and pragmatic. Hence, there are configurations where even in the most liberal languages – incorporation is not possible, and the only way of realization is in the form of a PP, i.e. of a non-reduced PAdP. The fact that prepositions are more likely to be transitive follows straightforwardly from the fact that adjectives incorporate the immediate complement of the categorized root, and if they end up transitive – it is only because the incorporated item was transitive itself. Core prepositions are those members of the PAd category which do not incorporate their complements, so their complements must receive overt realization. The fact that adjectives are more likely to inflect is an instance of a more general tendency that within the same category, items with more semantic content are more likely to inflect than those with less.