Anti-propositionalism -- as I understand it -- is the view that sentences don't express propositions. But if everyday sentences (like "The dog is behind the curtain") don't express propositions, what else could they express?
2 Answers
William G. Lycan summarize what he calls ”Use” theories of meaning like this in 'Philosophy of Language' (Routledge page 84):
”Use” theories have it that ”meanings” are not abstract objects like propositions; a linguistic expression's meaning is determined by the expression's characteristic function in human social behavior.
These views are closely connected with Wittgenstein's Language-game. In "Preformative Utterances" J.L. Austin state the following:
These performative utterances are not true or false, then. But they do suffer from certain disabilities of their own. They can fail to come off in special ways, […] the various ways in which a preformative utterance may be unsatisfactory we call, for the sake of a name, the infelicities; and an infelicity arises- that is to say the utterence is unhappy – if certain rules, transparently simple rules, are broken.
(Preformative Utterances in 'The Philosophy of Language', A.P Martinich, Oxford University Press page 292-293)
The utterances are what he calls either 'happy' or 'unhappy', not true or false. I found a PDF (Preformative Utterances, J.L. Austin)
When I say "I name this ship the Queen Elizabeth" I do not describe the christening ceremony, I actually preform the christening [...] Now these kinds of utterances are the ones we call preformative utterances.
(Preformative Utterances in 'The Philosophy of Language', A.P Martinich, Oxford University Press page 292-293)
Austin's pupil John Searle developed the idea under the label "speech acts". Their work greatly influenced pragmatics (from Wikipedia) H.P Grice have a great article called "Meaning" from 1957, which also can be found in 'The Philosophy of Language' by A.P Martinich.
Here is a good Wikipedia article about Meaning (Read section Usage and meaning).
If a utterance or statement fails to be truth-apt:
truth-apt denotes statements that could be uttered in some context (without their meaning being altered) and would then express a true or false proposition (from Wikipedia)
You can argue that the utterance or statement do not express a proposition.
As @GlenW notes, Wittgenstein rejected the notion that sentences, in isolation, expressed propositions, given that they require an entire context and rules of approval in order to actually have any effect. The sociological arguments that go along with the distinctions he lays out may not appeal to idealists, so here is another motivation.
Propositions are a mathematical idealization of 'steps' or 'moves' in what he called 'language-games'. In reality, the meaning of a symbol is deeply embedded in some system, and may or may not agree with something that has effective meaning. Ideal fields like mathematics seem to be the place that should be the safest to assume our idealizations are good enough, but even that fails.
Consider all of the mathematics referring to 'classes' that are not 'sets' to avoid the notions of set theoretic limitation. We happily talk about 'any group' or 'any topology' with such and such a 'property', but we still do not know whether those generalizations are propositions. The property could somehow specify something that leads to a paradoxical outcome in some odd corner. Then, are statements about such things propositions? Not all of our logic of propositions is going to apply.
We have to see propositions as an idealization and 'get behind them', even to discuss idealized things properly.