01.02.2015 Views

Presuppositions and Pronouns - Nijmegen Centre for Semantics

Presuppositions and Pronouns - Nijmegen Centre for Semantics

Presuppositions and Pronouns - Nijmegen Centre for Semantics

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.

Dynamic semantics 123<br />

implementation of this analysis, Kamp <strong>and</strong> Reyle adopt a purely lexicalist<br />

approach: they set up a lexical rule associated with <strong>and</strong>, which, by means of<br />

an indexing scheme introduced expressly <strong>for</strong> this purpose, ensures that Sl Sj is<br />

processed be<strong>for</strong>e S2. S 2 . Clearly, this analysis is very much in the spirit of<br />

dynamic semantics, <strong>and</strong> all objections against the dynamic treatment of<br />

conjunction apply to Kamp <strong>and</strong> Reyle's proposal, too. However, within the<br />

DRT framework their treatment is anything but inevitable.<br />

I have already hinted at what I take to be the right analysis of conjunction.<br />

It is the orthodox analysis, according to which the lexical content of <strong>and</strong><br />

merely says that the sentences flanking it are both true. In terms of Kamp<br />

<strong>and</strong> Reyle's version of DRT, this is just to say that the lexical rule associated<br />

with <strong>and</strong> is an instruction to interpret both conjuncts within the context of<br />

the current DRS. Hence, as far as the lexical meaning of <strong>and</strong> is concerned,<br />

conjunction is commutative. If the order of the members of a conjoined<br />

sentence matters, as it often does, it is not because the lexical meaning of <strong>and</strong><br />

is non-commutative, but because the process of interpretation is sensitive to<br />

the order in which it receives its input, as in general it has a preference <strong>for</strong><br />

incremental processing.<br />

This is just a sketch, to be sure, but it can be made quite precise by<br />

borrowing concepts from computer science, more especially from parsing<br />

theory. The construction rules of a system a la Kamp <strong>and</strong> Reyle may be<br />

viewed as a grammar, which is applied by a semantic parser, as one might say.<br />

Formulated in these terms, the notion of incremental interpretation may be<br />

defined by saying that, by default, the semantic parser processes its input<br />

depth-first <strong>and</strong> left-to-right.<br />

4.3 Forward reference<br />

Whereas I say that incremental interpretation is just a processing strategy,<br />

the dynamic semanticist says that it is encoded in the meaning of <strong>and</strong>. Our<br />

positions yield different predictions about the interpretative effects of linear<br />

order. If we adopt the DRT analysis that I advocate, it should be possible <strong>for</strong><br />

an anaphoric expression in the first half of a conjunction to have its<br />

antecedent in the second conjunct; <strong>for</strong> the expectation that this will not<br />

happen is based upon a processing strategy, <strong>and</strong> is there<strong>for</strong>e defeasible. If, on<br />

the other h<strong>and</strong>, the dynamic semanticist is right, then this should be<br />

impossible, since on his analysis it follows from the lexical meaning of <strong>and</strong><br />

that a pronoun in the first conjunct cannot have its antecedent in the second<br />

conjunct. It is the <strong>for</strong>mer prediction rather than the latter which is borne out<br />

by observations like the following, which I borrow from Bolinger (1977):

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

Saved successfully!

Ooh no, something went wrong!