The popular and general Fregean (etc.) idea is that because no particular expression of a proposition is identical to that proposition, then a proposition must be distinguishable not only from a specific expression of it, but from all expressions of it.
The argument is usually that a proposition is distinguishable from a particular expression. Though perhaps the truth is that it is the sum of all possible linguistic expressions of it. That is, when all possible expressions of the proposition are in, as it were, there's nothing left out of the proposition behind it. There is nothing else to say about it. And if there's nothing else to say about the ostensibly expression-independent proposition that hasn't been said by the sum of all its linguistic expressions, then perhaps there is no distinction between them at all. Thus propositions may not even exist in the first place for them to be linguistically expressed.
Michael Dummett goes into detail as to why Frege believed that there is such a distinction between a proposition and its linguistic expressions – or, in Frege’s case, between a sentence and the ‘Thought’ which it expresses:
"[A thought is] not true for you and false for me, it’s not true at one time and false at another time." 
The idea being expressed here is that it may appear to be the case that a thought is true for me though false for you; or true now, though false tomorrow. However, that's only because its linguistic expression leaves something out. If expressions were more precise and more explicit, such things wouldn't be left out.
This leads us of course to the conclusion that the Fregean Thought itself is precise and explicit. It's strange that something that is ostensibly completely non-linguistic can be both precise and explicit at all. How can something abstract and non-linguistic be precise and explicit? What does it mean for an abstract thought or proposition to be precise and explicit? Surely only linguistic expressions can carry the requisite information that will allow a thought to be precise and explicit. How can it be precise and explicit without its – or any – linguistic expression? What would constitute its precision and explicitness?
I mentioned the expression’s lack of precision and explicitness. Dummett explains why this is the case:
"Those are sentences which contain indexical elements: “It’s very cold today” and so on. So he says rather vaguely, when such sentences are uttered, the sentence itself does not suffice to express the thought. The time of utterance, for example, enters into the expression of the thought." (220)
Why must we bring in thoughts or propositions into this argument? It is of course the case that my sentence
'It’s very cold today.'
won't be true tomorrow. It may be hot tomorrow. So, yes, it's imprecise and not explicit enough. However, we needn't bring in a thought to make it explicit and precise. All we need to do is reformulate the sentence thus:
'Today, the 11th June 2008, it is cold here in the Midlands.'
That will be still true tomorrow. However, even then we will be required to drop the word ‘today’ and write:
'On the 11th June 2008 it was cold in the Midlands.'
We don’t require a thought/proposition that is more explicit and precise.
Even if thoughts do exist, who’s to say that they must be more precise and explicit that their expressions anyway? Perhaps the utterer simply didn’t think of the time, date and location of his or her utterance at all. Unless the thought has nothing at all to do with what the utterer was thinking at the time. In that case, why call it a ‘thought’ at all – simply call it a ‘proposition’? In that case, because propositions are often unequivocally seen as abstract, then they needn't be thoughts at all.
Here again, we must we automatically assume that the abstract expression which is somehow expressed is more explicit and precise than the expression itself? What on earth is making it more precise and more explicit if it has nothing to do with thought at all? How can it be precise and explicit at all without linguistic expressions and therefore the communication of information?
Dummett himself sees some of these problems. As Julian Baggini puts it:
"Dummett sees part of the difficulty in working out how we determine what the thought is, given that many utterances which express a thought are ambiguous." (220)
Are the expressions of the thought, if it exists at all, ‘ambiguous’, or do they simply leave elements out? This may be more an example of ellipsis than ambiguity. Pragmatically speaking, we always need to leave things out of our utterances.
For example, many people say ‘Thanks’ rather than ‘Thank you’, or ‘What!’ rather than ‘What was that?’, or ‘The train’s arriving soon’ rather than ‘The trains arriving in ten minutes 45 seconds’. There are numerous examples of this. Can we really say that they are ambiguous rather than just elliptical?
In any case, what does it mean to "determine what the thought is"? How
c/would we do that? What would we be determining? If it's genuinely non-linguistic and abstract, how on earth can we translate it at all into something concrete and linguistic? People speak about the problem of mind and body interaction (or they did), what about the problem of the proposition-into-linguistic-expression relation? How is that possible? How can such unlike things have any kind of relation at all? In addition, when we get the thought or proposition right (or wrong), how do we know that we have got it right (or wrong)? We may have something more precise and explicit than any given expression. Though have we still got the proposition or thought wrong? How would we know? What is it, exactly, that we're getting right or wrong? How do we know that our utterances are ‘ambiguous’ or, for that matter, that they leave vital information out?
Perhaps all this ostensible ambiguity points to the fact that there is nothing to be ambiguous about. If we are ambiguous at all, it's simply because we haven't expressed what it is that we want to express. It may have nothing to do with the fact that we simply haven’t got the proposition or thought right. If we can’t know that we have right (without the help of a philosopher like Dummett), how do we know that we have got it wrong? And if we can’t know whether our expression is either right or wrong, then to speak of ‘right’ and ‘wrong’ in the first place may be suspect (to use an idea of Wittgenstein). In other words, there may be no proposition or thought to either get right or to get wrong. There may be no such things as propositions or thoughts. The only thing we get wrong is that we don’t say something clearer or we don’t say what it is we would want to say.
We can’t really say ‘what we want to say’ because that would presuppose the existence of a thought that we wanted to express correctly. Perhaps we only know we've gone wrong somewhere when we don’t communicate properly with either ourselves or with another person. That has nothing to do with the expression somehow matching the thought or proposition. If anything solves the problem of inadequate communication, it will be another linguistic expression, not a thought: an expression that effectively communicates what it is that we want to communicate. What it is we want to communicate isn't a thought at all: it's a satisfactory linguistic expression. The fact that we have gone wrong is a fact about our linguistic skills, not a fact about not matching an abstract thought or proposition. If we match our failed expression with anything, it will be with other linguistic expressions which will manage to communicate something more effectively. They will not achieve this feet simply by being faithful to an abstract proposition. They will do so only if they manage to effectively communicate with us or with other persons.
We must apply Occam’s razor here and simply get rid of propositions – we don't need them. We shouldn't posit more entities than we require. The only entities we do require are other linguistic expressions, not abstract thoughts or propositions. Why over-populate the ontological slums with yet more abstract entities when we don’t even require them in the first place? If we require objectivity and determinacy of sense or meaning, perhaps we can achieve it by comparing sentences with sentences so as to adopt the sentence which does the job that we want to do.
In any case, something abstract and non-linguistic can hardly do the job of giving us more determinacy and objectivity if, by Dummett’s own lights, it is "difficult to work out how we determine what the thought is". If that is the case, then why do many philosophers believe that things that are hard to determine somehow miraculously assure us of (more) determinacy and even objectivity? If anything is objective, it's a linguistic expression itself and the concrete objects they are often about. Abstract entities like meanings and thoughts, if anything, surely bring with them more indeterminacy and even subjectivity, not less.
Not many people disagree with what the word ‘cat’ means. Not many people know what to make of abstract entities like propositions, let alone how we gain access to them. The problems we have had with Platonic Forms or universals should give us pause for thought. After all, these mysterious entities have given philosophers seemingly unsolvable problems for over two millennia. They explain a lot, sure, though only at the expense of our having to believe in mysterious abstract entities, which themselves were taken to secure truth (if not meaning), determinacy and objectivity. What the philosophers have offered us in order to secure objectivity and determinacy has come at a price. That price is a belief in mysterious non-spatiotemporal entities which somehow spread themselves across the universes and over time.