Staff Work in Progress Meeting with Professor Michael Devitt

Testing Theories of Reference

How should we test theories of reference? The accepted practice is to test them against the referential intuitions of philosophers. Machery et al (2004) wonder why it is appropriate to rely on the intuitions of philosophers rather than those of the folk. I wonder why it is appropriate to rely on referential intuitions at all.

We should not go along with the common philosophical view that these intuitions are a priori. Philosophers might follow linguists in thinking that linguistic intuitions are “the voice” of our linguistic competence. I have argued against this view and urged that referential intuitions are empirical theory-laden central-processor responses to linguistic phenomena, differing from many other such responses only in being fairly immediate and unreflective. So we should rely on the intuitions only to the extent that they are reliable indicators of the nature of linguistic reality. And, at best, they are only indirect evidence. We should be seeking more direct evidence by examining the linguistic reality that these intuitions are about: we need to examine linguistic usage. The results of this direct examination can then also be used to assess the reliability of referential intuitions.

Referential intuitions differ among themselves in several significant ways. There are the perceptual ones, memory ones, and ones formed in thought experiments. Among the latter there are ones about humdrum hypothetical cases and ones about fanciful hypothetical cases. And among them all there are differences in the degree to which their reliability depends on expertise. I conjecture that referential intuitions about humdrum cases, whether perceptual or not, are likely to be fairly reliable without much expertise about language. In contrast, referential intuitions about fanciful hypothetical cases, like the ‘Gödel’ cases tested by Machery et al, probably require a good deal of expertise.

Finally, I address the problem of testing theories of reference more directly against the evidence of linguistic usage. One source of this evidence is the corpus. I illustrate this with some material used by Genone and Lombrozo in their experiment testing theories of reference. Many of their uses of the invented term ‘tyleritis’ seem to be inconsistent with what description theories of ‘tyleritis’ would predict. But there are notorious difficulties in using the corpus as evidence. So, philosophers should follow linguistics in using the method of elicited production to test their theories of language. But it has so far proved difficult to come up with a satisfactory experimental test because of the ‘implicit-scare-quote’ problem. The problem is that elicited usage, indeed the corpus, provides the evidence we need only if speakers are not implicitly distancing themselves from their usage. We need experiments that control for this worry. The experiments conducted so far have not managed this.

• More information about Professor Michael Devitt can be found on his web pages at: http://www.gc.cuny.edu/Page-Elements/Academics-Research-Centers-Initiatives/Doctoral-Programs/Philosophy/Faculty-Bios/Michael-Devitt