Functions of Language - Volume 11, Issue 1, 2004

The paper focuses on the relation between the determiner (D) and the common noun (N) in a noun phrase (NP). Four facts show that D depends on N: only N is relevant to whether NP can be used as an adjunct; possessive determiners are similar to clearly dependent possessives e.g. in Dutch and German; N decides whether or not D is obligatory; and in English only one D is possible per N. Three other facts show the converse, that N depends on D: in many languages D sometimes fuses with a preceding preposition (e.g. French de le = du; English for each = per); D decides whether or not N is obligatory; the ellipsis of N is a regular example of dependent ellipsis. Therefore D and N are mutually dependent, a relation which requires the structural flexibility offered by Word Grammar. This does not mean that NP has two heads, but rather that either D or N may be the head.

In this paper, I investigate the theoretical status of noun phrases without nouns, i.e. noun phrases that do not contain a noun or pronoun, but only words that otherwise occur as modifiers of nouns. I investigate six possible analyses for such noun phrases: (1) that they are elliptical, (2) that the apparent modifiers are nouns, (3) that the apparent modifiers are heads, (4) that the determiner is the head, (5) that they are headless, (6) that all noun phrases are headless. Although the answers vary depending on the language investigated, I argue that the last hypothesis is generally the most plausible one.

The problems posed by nominal structure and nominal reference are notoriously subtle and complex. Though fairly extensive, their treatment in Cognitive Grammar (CG) has thus far been partial, preliminary, and scattered in numerous publications. The synthesis attempted here can hardly overcome the first two limitations. It may however provide a useful way of framing the issues and suggest some promising lines of attack.

This study revisits the classic problem posed by the meaning of proper names, and proposes a procedural approach to this problem, by analysing the meaning of proper names as an instruction to find in long-term memory the referent that carries the proper name in question.This is a revision of my earlier theory of ‘naming predicates’ (Kleiber 1981), which captures the meaning of proper names like Louis in terms of paraphrases of the form “the x who is called Louis”. The concept of ‘naming predicate’ was meant to provide an alternative to the inadequacies of the two classic approaches to the meaning of proper names, viz. theories that analyse proper names as semantically empty (e.g. Mills, Kripke 1972) and theories that analyse proper names in terms of uniquely identifying descriptions (Frege, Russell 1956). An analysis in terms of naming predicates (‘the X called Louis’) gives proper names an abstract type of meaning, thus avoiding the disembodied sign that results from analysing them as semantically empty, and at the same time does not go to the other extreme of incoporating aspects of the referent in the proper name’s meaning, thus avoiding the well-known problems with referential identity (e.g. Tullius = Cicero) and the related puzzles of transparence and opacity.In spite of these descriptive advantages, further research has shown that there are a number of problems with the notion of ‘naming predicate’. One of these problems concerns the status of proper names in ‘naming constructions’ like I am called Louis. Applying a naming predicate analysis to such constructions either leads to infinite regression (Wilmet 1995), or — if Louis in the naming predicate ‘the x called Louis’ is regarded as a phonetic form rather than a proper name — to a denial of proper name status in the very construction that expresses the naming link between proper name and referent (Jonasson 1982). Another problem concerns the cognitive naturalness of an analysis in terms of ‘naming predicates’. While this analysis is quite natural in contexts like There is no Louis in this office, it works less well in contexts like This painting is a real Picasso and, most importantly, in prototypical uses like Louis is a painter and a sculpturer, where a naming predicate analysis solely identifies the referent as the carrier of the proper name.These problems have led me to propose a revision to the theory of naming predicates. The descriptive advantages of using the naming relation between proper name and referent as the basis of the semantic description are obvious, which means that this aspect of the theory needs to be maintained. What causes most of the problems, however, is associating this naming relation with a predicate. As an alternative, I propose to reanalyse it in a procedural sense, not as a predicate describing the referent but as a procedural instruction to look for the referent that carries the proper name. This puts proper names in the domain of indexical signs like deictic elements. Common nouns, on the other hand, are not indexical in this sense but stand for concepts, which means that indexicality only comes into the picture when deictic elements are added.