Research talk

November 25, 16-17.30 h.

Abstract: Many languages allow reduced nominals (those lacking determiners, case marking, possibly also inflection and modification) as inner arguments. This talk is concerned with the semantics of constructions involving such reduced nominals. The primary focus is on Hindi incorporation, which is known to manifest a peculiar mismatch between syntax and semantics (Mohanan 1995, Dayal 1999, Wescoat 2002). The reduced nominal of Hindi incorporation behaves syntactically like a complement of the verb, with respect to agreement for example. At the same time, its interpretive possibilities align it with nominals inside compounds. Hindi can therefore be considered a pseudo-incorporation language. This talk explores the possibility of deriving the interpretation associated with compounds without necessitating a syntactic analysis combining an N0 with V0. It is shown that the semantics of pseudo-incorporation must capture at least four properties, which I illustrate with familiar examples from English:

1.     Kim is a book-seller.
2.     The baby ate *(something). It was a piece of fruit lying on the floor.
3.     Sue didn’t eat apples.
4.     Mary went apple-picking. #They/The apples were delicious.

Although the nominal in (1) is ‘singular’, there is no implication that only one book is involved in selling. We can call this the property of number-neutrality. Number-neutrality is related to, but distinct from, the second property which we can call prototypicality. As (2) shows, the intransitive version of eat restricts the interpretation of the understood direct object to prototypical themes such as meal. The third property is obligatory narrow scope, shown by bare plurals in examples such as (3). Finally, pronominal anaphora to compounds is disallowed, as shown in (4). Hindi pseudo-incorporation is shown to manifest all four of these properties.
    The two detailed proposals that have been made for the semantics of incorporation (Bittner 1984, Van Geenhoven 1997) derive existential interpretations, addressing only the properties of number neutrality and obligatory narrow scope. They are therefore inadequate for the construction at hand. A third proposal (Porterfield & Srivastav 1988, Ramchand 1997, Dayal 1999), in terms of complex-predicates, attempts to capture the other relevant properties but remains programmatic. This talk takes up the challenge of fleshing out the details of this proposal, deriving some of the desired effects via restrictions on the rule of complex predicate formation. This is compared with another recent proposal for incorporation by Farkas and de Swart (2003).
    The talk concludes with a discussion of the applicability of the proposed semantics to other constructions involving reduced inner arguments. This includes pseudo-incorporation in other languages (Albanian, Hungarian, Niuean) as well as to compounding (English) and canonical noun incorporation (Mohawk, Greenlandic).