<<   >>   TOC

4. Semantic type: <<e, t>, t>


Within the framework of discourse representation theory (DRT), it quickly becomes apparent that proper names can explicitly denote sets of individuals instead of individuals. A theory that insists on assigning type <e> to proper names, while positing <<e, t>, t> for common nouns, would thus have to employ an ad hoc mechanism of type conversion. This mechanism would always expand the denotation of a proper name to a set when this is required by the syntactic-semantic environment of the lexical item. One may be tempted, for example, to propose that this mechanism is triggered by e.g. plural, partitive, restrictive adverbials. However, proper names are in overlapping distribution with common noun phrases that are typically assigned type <<e, t>, t>, as in (10):

(10)       I saw John, the dog, one of the chairs, a dinosaur.

The verb see and, for that matter, any other transitive verb, would have to have two subcategorization frames (cf. 11a and b) in order to allow proper names as complements.



        Since this would increase the complexity of the semantics, Montague (1974) proposed to treat all nominals alike as type <<e, t>, t>. Furthermore, the assumption of different semantic types within the one category of nominals would severely violate storage economy in the lexicon: virtually all lexical items that can subcategorize for nominal expressions would need two entries – one for proper names and one for common nouns.

        One may possibly propose a syntactically more refined structure that maintains <e> status for proper names:


        This strategy only shifts the problem from the subcategorization frames of verbs to those of determiners, since elements of category D can freely take common nouns or proper names as complements. However, it reduces the problem of storage economy by an order of magnitude, since D is a relatively small, closed class. Nevertheless, a formal semantic system with type ambiguities is still suboptimal.

        The DP hypothesis opens an interesting additional alternative: when English common nouns occur in a sentence, where they are interpreted as a set, they are always wrapped in a DP. The question may thus not be whether proper names are really sets, but whether common nouns are not, in fact, individuals, as in (13):


This alternative has some intuitive appeal: the set status would be decided at the DP node and not depend on information supplied by the complement, but rather on information supplied by the head node, i.e. the determiner.

        However, with Burge (1973), I would argue that both common nouns and proper names have a predicative quality that is not captured by the assignment <e> in  (12) and (13):

“A proper name occurring in a sentence used by a person [DRT: discourse participant] at a time designates an object [DRT: discourse referent] if and only if the person refers to that object at that time with that proper name, and the proper name is true of that object” (Burge 205-206).

        Before I investigate the syntactic details of proper names, I will first assess what “being true of a discourse referent” may mean for a proper name.

<<   >>   TOC

© Philipp Strazny 1998