It would be helpful to shed more light on quite how what is fixed – for example by word meaning – and what varies, according to occasions of utterance, is apportioned in Travis’ picture.
Here is one such indication. In ‘To represent as so’, Travis criticises one element of Frege’s thinking.
For Frege, a concept is a function, namely, one from objects to truth values. (See Frege 1891.) If words ‘Sid grunts’ decompose into an element ‘Sid’, which names an object, and an element ‘grunts’, which names a function from objects to truth-values, then the whole, ‘Sid grunts’, names the value of that function for a certain argument, namely, Sid. Which is to say that it names a truth-value: true if Sid grunts, false if he does not. Which is to say that for parts jointly to play these roles is, ipso facto, for them to decide a unique and determinate truth condition for their whole. Mutatis mutandis for propositions, of which words ‘Sid grunts’ could be but one instance, or expression. [Travis 2011: 172]
According to this view, if an element in a proposition (which has pride of place in both Frege’s and Wittgenstein’s thinking) has the role of naming an object or naming a concept its contribution to the truth condition and thus truth value of the whole is fixed by that alone.
Inspired by the later Wittgenstein’s deployment of language games, Travis considers two different contexts for the sentence ‘The room is dark’. In one, it is used to say that books cannot be found by vision alone. In the other, it indicates that undeveloped film can be removed from a canister.
One might correctly say of either game that in it ‘is dark’ names (speaks of) being dark. But the role it plays in naming this differs from the one game to the other. In the one game, but not the other, it contributes to a condition on being as said which is not satisfied if, where whether to remove film turns on whether the room is dark, removing film is not the thing to do. So if a move consists (on an analysis) of parts, for each of which there is a such-and-such it names, those facts about the move are compatible with any of indefinitely many mutually exclusive conditions on correctness (answers to the question when things would be as thus said). [ibid: 173]
So if we think – with Frege – of words naming objects and concepts, still a suitable combination of objects and concepts does not fix the condition under which the result is true, contra the earlier summary of Frege.
That words name such-and-such determines no unique contribution which is that which such words make to conditions on the correctness of the whole they thus are part of. Such is just part of what naming is. It holds equally for naming in the context of a move in a language game, and naming in the context of a proposition. The fact of my speaking of being dark is compatible with my saying any of many things in, and by, doing so. There are many different things, each of which being dark may, sometimes correctly, be taken to be (or come to). Being dark admits of understandings. [ibid: 173]
So the fact that words name such and such is fixed across occasions but its role in the truth of the whole varies between occasions of utterance. Returning to the earlier truism, Travis suggests that the following is the correct thing to say:
It is on an application of a concept to an object, on an occasion, that one says the object to be thus and so. The rest of the truism then holds. The concept as such admits of many applications, each excluding others. So it alone cannot assign an object, in being as it is, a truth value. [ibid: 173]
Sticking with the case of the room being dark, it seems that the same concept is named by ‘dark’ in the different language games, but it has a different application. The concept remains the same across occasions, but its application not, and it is the application which matters to truth. The same concept, I think, admits of ‘understandings’, which correspond to the fine grained applications. I think that the same applies to another Travis phrase for part of the conceptual: ‘a way for things to be’.
What of propositions? These could be common between different occasions but with different applications. Or they could be fine grained: distinct between occasions. But given the connection between propositions and truth, and given the connection between occasions of utterance and the truth of sentences, propositions look to be fine grained (unlike concepts). (Cf also ‘A proposition makes a demand on the world: to decide, in being as it is, the proposition’s fate—true or false. The proposition fixes a way for the world to decide this—if the proposition is that the setting sun is red, then what about the world would make this so or not.’ [ibid: 218]. This, I think, is evidence for my interpretation.)
There’s an added complication in that Travis is using Wittgenstein to correct Frege and so it’s hard to know how much is invested in preserving the idea of propositions rather than exploring word use in language games. But in ‘The proposition’s progress’ there is both a clear suggestion that they still find a place in Travis’ own approach and on the above question:
Propositions are one device for carving up exposure to risk. Language games are another. Such games are used in the Investigations to make a particular point, as above. This is not to banish the notion proposition. In suitable circumstances I may say to you, ‘Pia said that there is wine on the rug’, where this admits the response, ‘And is there?’. We may then go on to discuss ‘the proposition that there is wine on the rug’. What proposition is this? When would it be true? I mentioned it in words ‘there is wine on the rug’. Things would be as they are according to that proposition when they would be as those words speak of things being. When that would be is fixed by the operation of parochial equipment on my words in just the way described, above, for Pia’s. The above model of representing finds just this application here. Such a proposition, one might say, is what it is to us. Nor does it thereby speak of a way for things to be which admits of no divergent understandings. (Cf. §§429-465.) [Travis 2011: 215]
The one thing which is confusing here is the last line. It almost sounds as though the proposition admits of divergent understandings. But my hunch is that different propositions are the results of divergent understandings of utterances.
Travis, C. (2011) ‘The proposition’s progress’ in Objectivity and the Parochial, Oxford: Oxford University Press pp 193-228
Travis, C. (2011) ‘To represent as so’ in Objectivity and the Parochial, Oxford: Oxford University Press pp 165-192