If people attach something to a word within their minds, then this is a fact of their individual psychologies, not capable of establishing the different fact that the word actually means something within their language.
Here Wittgenstein is arguing that, despite appearances, words may become, and in fact are, animated in a way very different from a chunk of mind being stuck to them.
But this, I think, is not the most important lesson (in fact, as I will try to indicate later, such an outcome is not so surprising given the pragmatic nature of the turn); a more important lesson is that meanings, at least as usually conceived, are perhaps less crucial for semantic theory than previously thought.
In fact, if we can pinpoint this out without getting hold of any entity which we could call the meaning of the word, we would not seem to miss anything.
Before continuing, let me point out, by way of digression, another important aspect of the 'pragmatic turn': the fact that it brings about a certain amount of semantic holism.
In fact, once we accomplish the 'pragmatic turn', it is forthcoming.
And how would this square with the fact that I know what my words mean; that I can retrieve their meanings from my memory etc. – without being interpreted by somebody else?
In particular, how can it be reconciled with the fact that I perceive meanings?
Does this mean that people 'in fact' do not mean anything by their words?
The fact that meters are not actually inside a tree also does not preclude it from being five meters high.