How use Theories of Meaning can Accommodate Shared Meanings: A Modal Account of Semantic Deference

Abstract

Use theories of meaning (UTMs) seem ill-equipped to accommodate the intuition that ignorant but deferential speakers use natural kind terms (e.g. ‘zinc’) and technical expression (e.g. ‘credit default swap’) with the same meanings as the experts do. After all, their use deviates from the experts', and if use determines meaning, a deviant use ordinarily would determine a deviant meaning. Yet the intuition is plausible and advocates of UTMs believe it can be accommodated. I examine Gilbert Harman's and Paul Horwich's views, and argue that they do not offer a satisfactory reconciliation of the intuition with the theory. I propose an accommodation based on a novel account of semantic deference, and show that it is consistent with UTMs that a speaker may use a word with a certain meaning without fully or adequately knowing it.

Notes

¹Cf. the work of semantic externalists like Hilary Putnam 1973; 1975, Saul Kripke 1980, and Tyler Burge 1979; 1982; 1986; 1989; 1993.

²The ‘e’ is meant to suggest ‘expert’ or ‘expertise’, although natural kind terms may have currency in the absence of serious expertise concerning the kind, and technical terms may survive after the technical expertise related to them has largely disappeared (think about terms for old farming tools).

³The problem is not limited to technical expressions (not a sharply defined category). For any type of term, UTMs need to explain how semantic fragmentation and individualism can be avoided. In spite of this, discussion is often centred on technical terms. For reasons that will become evident, I limit myself to this type of terms.

⁴As this quote suggests, use theorists are (correctly) not prone to dismissing the problem on the purely technical grounds that a function from uses to meanings can be many-one. It is a substantial problem to determine the conditions under which different uses determine the same meaning.

⁵Mark Greenberg and Gilbert Harman point out that CRS ‘supposes that meaning or content is determined by (and so supervenes on) conceptual role, but that does not imply … that any difference in conceptual role entails a difference in meaning’ 2006: 297; see also 312, 317 n. 7]. It is not hard to imagine, however, that the role of a lay person's concept of an oak tree may be quite deviant, constantly misapplied, involved in inferences that no expert would make, and inducing behaviour that no expert would replicate.

⁶See Harman 1982: 248; 1987: 68].

⁷I use ‘to defer’ to mean the same as ‘to have a deferential attitude/stance/disposition’. One may defer without engaging in actual episodes of correcting one's usage.

⁸The types of sentences must be individuated syntactically and acceptance, a relation to a sentence (type), is to be described without resorting to semantic notions; see Horwich 1998: 94–6; 2005: 30–1, 40–1].

⁹Conditions i)–iii) above are stated slightly differently in Horwich 2005: 52], where it is not clear whether deference requires experts (it is suggested).

¹⁰Michael Devitt 2002 complains about this and Horwich 2004; 2005: 53] acknowledges as much. I will come back to Devitt's criticism in §6.

¹¹Personal communication.

¹²What really matters is true belief.

¹³The counterfactual may need to be refined, and my account can be modified accordingly. But it is not the ultimate refinement that matters, and my project is not to look for one. I only need the assumption that there is such a counterfactual.

¹⁴A handy expert may not always suggest a replacement for the misapplied e-word; she may simply point out the word's inapplicability. The incorporation on the fly of a doctor's expertise might be manifested, for example, by an utterance of ‘I am afraid that I have arthritis, or whatever, in my thigh; it really hurts in the morning’.

¹⁵This means that for virtually every speaker and every e-word they use, meaning supervenes on purely counterfactual use, since the actual scenario (which is still part of the ceteris paribus scenarios) will not be among the selected ones, in the end—in the actual scenario we rarely interact with experts.

¹⁶So, we take the existence of a deferential stance as an important datum in deciding whether to consider the attribution of a certain disposition falsified by actual use; we do not simply register the facts about use. This may raise a worry: if we are free to decide what data about use we can keep and what data we can discard, we are free to decide what disposition to ascribe, and discretionary authority of this kind threatens to trivialize the attribution of a disposition. I think this worry is unjustified. One has to argue for any proposed filter. The existence of a deferential stance seems to be a legitimate one; others may not have a similar standing. This is a good place to bring up the common remark that deference alone does not justify attributing the shared meaning to someone's use of an e-word. Most authors require a minimal degree of competence: the lay speaker may use the e-word erratically but within limits (or her use must pass muster, in some way). Defining the limits is not easy. I do not address this side of the matter (see Goldberg [forthcoming] for some comments), except to say that the standing of deference as a filter may be defeated if further conditions are not met. I assume they can be met.

¹⁷There is no need to think that scenarios with experts include people who do not exist in the actual scenarios. They are best thought of as scenarios in which knowledge has been acquired (by some of the people who in the actual scenarios are ignorant).

¹⁸Furthermore, we are free to take the content of the speaker's concept of felony as identical with the content of the word, if we follow Harman's suggestions that for e-words the content of the concept is determined by the content of the word. Although we are free to do so, there does not seem to be any compelling reason for doing so. If we follow Harman's suggestion, we have as much reason to say that the content of the lay person's concept of felony is different from that of the expert's. What counts is that it is not the former content that determines/is the content of the word as used by the lay person. After all, Harman has severed the intra-personal connection between e-word content and concept content. The former is not determined by the latter. Harman himself observes that ‘Of course, the content of the more ignorant person's concept of an oak tree is not as rich as, so not the same as, the content of the expert's concept’[1987: 68].

¹⁹Devitt focuses on proper names, but he explicitly recognizes that these challenges arise for the type of words I am discussing (‘zinc’, ‘felony’, ‘arthritis’, and so forth). I set aside proper names because I think they need to be treated somewhat differently. For one thing, the notion of expertise that may apply to them must be articulated in a specific way (for an author who is aware of the peculiar character of names in this respect, see Goldberg [forthcoming: n. 22]).

²⁰This is perhaps too strong; I can concede that there are such speakers, and simply claim that they use an e-word without a determinate meaning. See below for more on indeterminacy.

²¹I am considering cases in which the object and the experts were contemporaries for a while, then the object ceased to exist, while the experts did not (or managed to hand down their expertise). Cases in which expertise is acquired about an object that has long ceased to exist before the acquisition of knowledge began (think about species of dinosaurs) pose a different challenge. I am going to set it aside for reasons of space.

²²Not disagreement in the relevant sense. But not every disagreement is relevant: physicists can disagree on whether gold is beautiful, and, more importantly, on its more esoteric properties.

²³This holds for the lay speakers, not necessarily for the experts; the latter may well divide into sub-communities whose members do use the word according to a basic pattern, thus with a determinate meaning.

²⁴Theologians (‘cherubim’), and philosophers (‘apeiron’) probably should be added to the list. Some of these terms could be called ‘supernatural kind terms’, but it would be better to simply recognize that we are not compelled to read ‘natural’ as ‘naturalistic’ in ‘natural kind’. In general I am thinking of terms for putative elements of reality; they include empty natural kind terms.

²⁵The community need not be actually identifiable by the deferential speakers; it is sufficient that it be identifiable by anybody who comes to reflect on these matters.

²⁶One example of the kind of questions that should be addressed is this: suppose a lay speaker has a strong allegiance to a specifically identified group of experts who use a certain word w uniformly, but there is no knowledge to be had about w(s). Why should this result in indeterminacy of meaning rather than in the meaning with which the group uses the word? I owe this question to Rob Streiffer. More in general, we need to consider the implications of choosing between two interpretations of deference to whoever is an expert: i) narrow interpretation: deference to whoever belongs to this specifically identified group of experts (where the group is identified without identifying all, or even any, of the members, e.g., as the Royal Society, or the academics who convene in this building every Wednesday); ii) wide interpretation: deference to whoever belongs to whatever group of experts there is.

²⁸I assume that knowledge of word meaning is sufficient for word understanding. Dean Pettit 2002 argues it is not necessary. I do not take a position on the matter of necessity.

²⁷See Greenberg and Harman 2006: 298]. See also Horwich [1998: 89]. Horwich is a bit circumspect on the whole matter. He concedes that understanding a word is a practical ability, but he thinks it is better characterized as an implicit form of knowing-that [1998: 18].

²⁹Demotion is not elimination. The ignorant speaker's idiosyncratic concept of oak tree might help to explain the cognitive significance of the expression ‘oak tree’ for her. However, I do not wish to make any definitive claim here.

³⁰I wish to thank Dennis Stampe, Martha Gibson, Robert Streiffer, Paul Horwich, Gilbert Harman, Deborah Bernhardt, and two anonymous referees for their comments on various drafts of this paper.