Monday, 3 March 2008

Semantic normativity #2

Having recently looked at Martin Kusch’s (2006) A Sceptical Guide to Meaning and Rules: Defending Kripke’s Wittgenstein, I’ve been reading Anandi Hattiangadi’s Oughts and Thoughts: Rule-Following and the Normativity of Content.

Hattiangadi was supervised by Kusch and shares with him hostility to the idea that linguistic meaning (or mental content) is normative. Unlike him, she does not try to rehabilitate Kripke and does not think that Kripke’s invocation of normativity is merely part of an immanent critique. On her account, Kripke goes down with the normative ship.

The arguments against the normativity of meaning are intriguing. The main thrust is this. Meaning might be governed by constitutive rules but merely constitutive rules are not prescriptive and one needs prescriptivity to get a sceptical argument about meaning. (If facts about meaning are genuinely normative or prescriptive then such facts would have to be essentially motivating. But, either, such facts could not be reduced to descriptive terms because of Hume’s Law (no ought from an is); or the very idea of sui generis motivating facts would be ‘queer’ indeed.)

The main arguments against full blown prescriptivity seem to be that one simply cannot offer a plausible prescription in the neighbourhood of meaning. Thus, for example, it simply isn’t true that if ‘red’ means red then one ought to call all red things ‘red’ because ‘ought’ implies ‘can’ and no one can call all red things ‘red’. In general, there is no intrinsic obligation to speak the truth. Thinking true thoughts inherits whatever limited obligation it does merely instrumentally from the fact that true beliefs combined with desires increase the chances of getting what one wants etc.

I do not have any knock down arguments against the case Hattiangadi offers. Despite this, I am not, however, rushing to embrace it for three reasons. (These reasons hopefully explain my suspicion. They are not yet arguments against her. I hope such arguments will follow.)

Firstly, the obligations in play in the book look like moral obligations but would have to be semantic obligations for the claims of semantic normativity to work. Thus, for example, one option considered is that semantic normativity amounts to the claim that one ought to speak the truth and yet there is no such general obligation. But this looks to assume that semantic normativity both is and is not a species of moral obligation and such a species may be hard to come by.

Secondly, the obligations are supposed to be all or nothing but no such plausible obligation is forthcoming. This seems, however, to be the wrong sort of norm for meaning. It would seem strange to think that semantic norms would not function holistically. So just as precise articulation of mental content might be part of an explanation of action but only in the context of holistic facts about other beliefs and meanings so I would not expect semantic normativity to issue action guiding prescriptions that could be stated without at least meaning-related conditional antecedents.

Thirdly, the distinction between regulative and constitutive rules does not seem quite right for the case made. Recall the distinction from Searle and summarised by Kusch thus:
Regulative: In conditions C you ought to perform action A. These regulate antecedently or independently existing forms of behaviour.
Constitutive: Doing A counts as B in context C. These create or define new forms of behaviour.

On one reading of the latter, an anthropologist might baptise a form of behaviour and so constitute a way of going on as, say, ‘passive aggressive’ without the agents self ascribing that phrase or even being aware of it. The concept would shape a form of behaviour in whoever’s eyes. This minimal reading would denude constitutive rules of any power to do work against semantic reductionism perhaps because it has no prescriptive force.

But Hattiangadi (perhaps following Searle - ages since I read him) takes an example to contrast with a regulative rule which does seem to be normative.

A group of people might go through the motions with marked tiles, but if they are in ignorance of the rules of Mah-jong, they are not playing Mah-jong. Before the rules of Mah-jong were invented, nobody played Mah-jong; nobody could have played Mah-jong without first inventing the rules. In contrast… the Indian rule of etiquette that prohibits eating with one’s left hand… regulates the activity of eating, which certainly predates the invention of the rule, and could just as well outlive it. [Hattiangadi 2007: 54]

The behaviour which is constituted as the behaviour it is by the rules of Mah-jong is surely so constituted in part by the idea that, if one is playing the game, then one ought to play in accord with both the legal rules and the norms of good strategy. It would still be impossible to derive a quasi moral ‘all out’ obligation. No one need try to win at Mah-jong. But that seems to me to indicate that there is something wrong with the strategy of the book rather than showing that neither Mah-jong nor semantics is normative.

See also this later entry.