I was thinking how to explain things to myself. Maybe someone else will find this helpful, or can give comments that may be helpful to me.
I think it's easy to use the toy Rylean ancestors of Empiricism and the Philosophy of Mind. We can always just add the Myth of Jones to the Myth of our Rylean ancestors and get a better account.
So the Rylean ancestors in question have Rylean views about the world and human beings. They roughly adopt the following picture:
*thinking that-p ➡ *saying that-p
Roughly, intentionality dwells primarily in language. Propositions are only expressed by language. A thought is the inner state which accompanies or causes the saying which expresses the proposition.
So the Ryleans here think that baby Bailey has no intrinsic thoughts. Baby Bailey is taught via direct method / stimulus-response English by his parents, mommy Bailey and daddy Bailey. Baby Bailey begins to make noises, and then after a while begins to associate contents with his sayings. When baby Bailey is making noises, he has certain brain states in his brain. Eventually, he has some thinking that-p brain states whenever he has a saying that-p utterance. His saying that-p utterances also come to express the proposition that-p.
The Ryleans basically have a verbal behaviorism. We can think to ourselves because we can have the brain states which cause speaking, but fail to have the speaking. But we always just analyze brain states as if they came to fruition in verbal behavior.
It seems that with this stuff in mind, we can give a good account of what proto-intentional states are. When the baby learns to make noises, he isn't tokening sentences. It seems like he's tokening noises. So these noises seem to be proto-linguistic to me. They aren't tokens of English sentences, really. They are tokens of the same noises that English sentences are noises of. Something like that.
And so when baby Bailey is making proto-linguistic noises, he will be having proto-thoughts in his brain. His brain will be making states which cause the proto-linguistic noises. At some point, when the noises become English sentences, those proto-thoughts will become thoughts.
I guess the question is how thoughts or sentences come to express propositions at all. I suppose that the thoughts or sentences provide structures to which we pencil in semantic contents. Like a map, it seems that we need an agent to bestow an interpretation upon the structure. No map-reader, and then the map is just a sort of proto-map.
Perhaps it is like this:
This inscription might represent something. I have something in mind which it does represent, but if I don't tell you, it seems you can only imagine that it represents something. It is a sort of candidate for representation, or perhaps, a candidate for interpretation.
Maybe because we are all skilled and grown up, being told what it means allows us to jump right into taking it to be representational. Baby Bailey may not be so quick, and so may have an extended period of mastering inscription interpretation. Learning by direct method seems to have this sort of extended period of mastery as well. This period would seem to be the period of proto-states talked about above.
I guess it's like teaching someone to play chess. I can play effortlessly, so I play full-blooded chess. But Mr Noob has to do a lot of thinking and wondering. He isn't so sure of himself. He's always checking up rules in the rule book, say. So he's only proto-playing. He's in the midst of learning a game, so he's missing a lot of things that are obvious to others because he hasn't been conditioned to look for them.