25.12.2013 Views

Was sollen wir tun? Was dürfen wir glauben? - bei DuEPublico ...

Was sollen wir tun? Was dürfen wir glauben? - bei DuEPublico ...

Was sollen wir tun? Was dürfen wir glauben? - bei DuEPublico ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

A Single-Type Ontology for Natural Language<br />

Kristina Liefke<br />

In (Montague 1970a), Richard Montague defines a formal theory of linguistic meaning which<br />

interprets a small fragment of English through the use of two basic types of objects:<br />

individuals and propositions. In (Partee 2006), Barbara Partee conjectures the possibility of<br />

a comparable semantic theory, which only uses one basic type of object (hence, single-type<br />

semantics). This paper supports Partee’s conjecture by identifying two suitable single-type<br />

candidates. To single out the latter, we first introduce a set of semantic requirements on any<br />

single basic type. The application of these requirements to the familiar types from (Montague<br />

1973) reduces this set to two members. The paper closes with a comparison of Partee’s<br />

preliminary single-type choice and our newly identified single basic types.<br />

1. Introduction<br />

Natural languages presuppose a rich semantic ontology. To provide an interpretation for, e.g.,<br />

English, we require the existence of individuals (e.g. Bill), propositions (Bill walks), first- and<br />

higher-order properties (walk, rapidly), relations (find), and many other kinds of objects.<br />

Theories of formal linguistic semantics (paradigmatically (Montague 1970a; 1970b; 1973))<br />

tame this ontological ‘zoo’ by casting its members into a type structure, and generating<br />

objects of a more complex type from objects of a simpler type via a variant of Church’s typeforming<br />

rule (Church 1940), (cf. Gallin 1975):<br />

Type-forming rule (CT)<br />

If α and β are the types for two (possibly different) kinds of objects, then is the<br />

type for functions from objects of the type α to objects of the type β.<br />

In this way, Montague (1970a) reduces the referents of the small subset of English from<br />

(Montague 1973) to constructions out of two basic types of objects: individuals (or entities,<br />

type e) and propositions (or functions from indices to truth-values, type ). Proper<br />

names (e.g. Bill) and sentences (Bill walks) are then interpreted as entities, respectively<br />

propositions, intransitive verbs (walk) as functions from entities to propositions (type ), and transitive verbs (find) as functions from entities to functions from entities to<br />

propositions (type ).<br />

Montague’s distinction between entities and propositions (or between entities, indices, and<br />

truth-values) has today become standard in formal semantics. This is due to the resulting<br />

semantics’ modelling power, and the attendant possibility of explaining a wide range of<br />

syntactic and semantic phenomena. However, recent findings in language development<br />

(Carstairs-McCarthy 1999; Cheney and Seyfarth 1990; Snedeker et al. 2007) suggest the<br />

possibility of an even simpler semantic basis for natural language. The latter lies in a single<br />

basic type (dubbed ‘o’) whose objects encode the semantic content of entities and<br />

propositions. From them, objects of a more complex type are constructed via a variant of the<br />

rule CT:<br />

Single-type-forming rule (ST)<br />

If o is the single basic type and α and β are single-type types, then is a singletype<br />

type.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!