I’ve wondered hither and thither through Danièle Moyal-Sharrock’s book on the train for the last two days and I’ve found it helpful. A third book on On Certainty seems a good thing even if the source text still doesn’t seem half as interesting as the Investigations.
I want to pick out just one thread. Unlike some other commentators, Moyal-Sharrock thinks that the bipolarity of propositions was a constant commitment of Wittgenstein throughout his writing. (Bipolarity: every proposition must be capable of being true and capable of being false. It is more restrictive than bivalence: every proposition is either true or false.) Thus only something that could be true and could be false counts as a proposition. So, in the context of the later Wittgenstein as usually described (the second Wittgenstein for Moyal-Sharrock), grammatical prescriptions, which set out conditions for playing particular language games (there is no such thing as reddish green; pink is lighter than red; 2 + 2 = 4), cannot be false and thus do not count as propositions. To put it that way is to mislead, however, – sorry! – because neither are they true. (At this point, some commentators are happier to say that they are true in a minimal or surd manner.) So what looked like metaphysical, supernatural truths are, instead, merely norms that prescribe the correct use of terms in empirical language games and perhaps license empirical inferences.
Moyal-Sharrock argues that the later, later Wittgenstein (the third Wittgenstein) realised that this status applied not only to sentences that looked metaphysical or supernatural but also to some that look merely empirical but very sure. Moore’s ‘This is a hand’ is one such. In so far as Moore is right - which is not very far, I hasten to add - it’s because he’s articulating a grammatical rule, something that lays out the conditions for empirical language games of doubt and knowledge, or reason giving. Since, when being used to articulate a grammatical rule, they cannot be false, they are not propositions. Neither do they have a positive empirical role (contra Moore).
If I say in the middle of a conversation: ‘I am here’, the sentence has no truth-value because, pronounced in such circumstances, it has no informational, hypothetical or descriptive use. [Moyal-Sharrock 2007: 93]
So there is no use which combines both their pre-language-game certainty and a fact stating role. But this point goes further.
Moore’s confusion is explained by the fact that the very same sentence can play a doppelganger role. Stripped of its certainty by the right context, it can be used to make an empirical claim (on an amputation ward, perhaps) and that explains our confusion in thinking that the rest of our empirical claims might be founded on very sure true propositional claims. So what look like hinge propositions are really grammatical rules which are neither true nor false. They can be used heuristically to explain correct empirical use. So they can be spoken. But they cannot, in a charged sense, be said: a form of vocalisation restricted to propositions. This idea of a background grammatical fits Marie McGinn’s account from twenty years ago here:
Thus, my certainty regarding, say, the judgement ‘This is a hand’ is to be seen as a pre-epistemic attitude that is in part constitutive of my practical ability to speak the language. the judgement that this is a hand is not a piece of knowledge – a true, justified belief, based on evidence – but an authoritative expression of my established mastery of English. [McGinn 1989: 144]
But not McGinn’s acceptance of the applicability of truth to hinges, at least in a derivative sense expressed here:
We are not, therefore, to think of Moore-type [ie framework] propositions as stating empirical truths, in the sense of something which has turned out to be so but which may have turned out otherwise. The judgements of the frame are not applying our language in propositions whose meaning is independent of their truth-value; for these judgements, their being true in part determines the meaning of the expressions being employed. [McGinn 1989: 142]
Still, I’m not sure that the difference is as great as all that. All turns on the next bit. This is just my first reaction and I may take it down later.
Moyal-Sharrock goes beyond what I’ve said so far in saying that hinge propositions, in not having bipolarity and thus not being propositions at all, are, in fact, nonsense in a technical sense. This fits with the idea that their use to make a claim in conversation but without the context that might make for doubt or knowledge would be bizarre (I am here; this is a token etc). But now the sense of ‘nonsense’ in play looks too strong to me.
It’s not just a Travisian point that use and meaning are highly contextually and pragmatically determined. The use is sufficiently reified that it can said to be nonsense (whereas the doppelganger use isn’t nonsense) but it is – if I’m reading this right – nonsense in virtue of the way the words are put together and the context in which they are launched. Moore isn’t just baffled by his own utterance (in the way that we may not know whether – to follow Travis – it’s right to say that ‘Smith is at home’ is true if he is in his house but it has been pushed downhill in a landslide; we know that there’s something misleading about just saying ‘Smith is at home’ without qualification).
Moore is led by the composition of the sentence to use just that sentence in that context even though it’s illicit. Whilst it would be fine for a philosopher-anthropologist to use it to chart Moore’s certainty in practice (it can play a heuristic role and the form of words matters for that), his certainty is ineffable and thus his attempt to put it into words, to say what should only be shown, results in nonsense. (I should refer back to posts on Adrian Moore here. Moore also has a notion that putting something ineffable into words results in nonsense and yet it has to be the right bit of nonsense, as it were, to count as such an attempt.) But if so, it looks very much like a two tier view of nonsense. The nonsense isn’t just of the ‘iggle wiggle piggle’ variety.
Now as Moyal-Sharrock emphasises, ‘nonsense’ is not a perjorative word for Wittgenstein. So we might simply keep an eye on the connection between sense and propositionality, sense and bipolarity, and not worry about calling everything else ‘nonsense’ but in a significant kind of way. But given that there are so many other perfectly ordinary uses of words which are not true or false, it seems excessive to deploy the label ‘nonsense’ for these cases and this corrective would, in any case, diminish our understanding of the ineffability of hinges. (Other things that would count as nonsense would not be ineffable. So the status as nonsense does not explain what fails when we try to express our certainty in words rather than actions.)
Maybe I’ve spent too long being confused by resolute readings but that experience has taken to this extent: now any sense of nonsense which isn’t actually nonsense seems a cheat.