Olav Mueller-Reichau - How indefinite singular NPs can be functional concepts

Functional concepts manifest themselves (among others) as definite NPs, which serve to denote the unique object or the unique kind within the utterance context. In the latter case, the NP-referent will be the kind named by the common noun. The paper shows that indefinite NPs can likewise serve to denote the (unique) kind named by the common noun, but only in cases where the definite NP cannot be used; in all other cases the availability of the definite article blocks the indefinite. This is because the two articles constitute a Horn scale, with the definite article forming the more informative expression (cf. Hawkins 1991).
   While the indefinite article requires the context to provide at least one suitable kind, the definite article requires that the context provides exactly one kind (“uniqueness presupposition”) and that this has realisations in the world of speaker/hearer (“existence presupposition”). Given this, the hearer will tell from the indefinite article that the definite article has not been used because the context does not meet either the existence condition or the uniqueness condition. I propose that non-uniqueness is a last resort (arguably because there is the lexical item one which is specialised in non-uniqueness contexts). Therefore, indefinite NPs can be functional concepts only in contexts where the existence presupposition of the definite article is not met. There are three relevant cases:
1. Novelty: The speaker assumes that the hearer does not know of the named kind, let alone of realisations of it (cf. Dayal 2004):

(1)     John invented a pumpkin crusher.

2. Fictionality: The speaker assumes that the hearer believes that there are no realisations of the named kind:

(2)     John invented a time machine.

3. Rules: The speaker wants to state a rule (cf. Cohen 2001), which means that whether or not there are realisations of the kind in the world is irrelevant to what the speaker wants to communicate.

(3a) A lion is a dangerous feline.
(3b)     A mermaid is a woman with a fish’s tail.

It has been observed that indefinites display subkind readings in classic kind predications (cf. Krifka et al. 1995):

(4a) A dinosaur is extinct.
(4b) Babbage invented a computer

This is explicable as follows: In (4a), the subkind reading is due to the lexical semantics of the predicate. Since be extinct demands the previous existence of realisations of the argument kind, i.e. of real dinosaurs, the indefinite article can only be motivated by presuming non-uniqueness of the kind, which yields the subkind reading.
The speciation predicate invent does not presuppose the existence of argument realisations. However, as computers are well-known and omnipresent, it is not natural assume that the speaker thinks that the named kind is new or fictional for the hearer. So here the availability of the definite construction blocks the functional use of the indefinite construction, and we understand that the speaker has one computer subkind in mind.

References

Cohen, A. (2001): On the Generic Use of Indefinite Singulars. Journal of Semantics 18, 183-209.

Dayal (2004): Number Marking and (In)definiteness in Kind Terms. Linguistics & Philosophy 27, 393-450.

Hawkins, J.A. (1991): On (In)definite Articles: Implicatures and (Un)grammaticality Prediction. Journal of Linguistics 27, 405-442.

Krifka, M. et al. (1995): Genericity: An Introduction. In: G. Carlson & F. Pelletier (eds.), The Generic Book, Chicago, London: University of Chicago Press, 1-124.