Ordinary language is all right.
One could divide humanity into two classes:
those who master a metaphor, and those who hold by a formula.
Those with a bent for both are too few, they do not comprise a class.
'In the dark there is emphatically "nothing" to see, although the world is still "there" more obtrusively.'
'Oh I did not say it in such limpid language. And when I say I said, etc., all I mean is that I knew confusedly things were so, without knowing exactly what it was all about. And every time I say, I said this, or, I said that, or speak of a voice saying, far away inside me, Molloy, and then a fine phrase more or less clear and simple, or find myself compelled to attribute to others intelligible words, or hear my own voice uttering to others more or less articulate sounds, I am merely complying with the convention that demands you either lie or hold your peace. For what really happened was quite different. And I did not say, Yet a little while, at the rate things are going, etc., but that resembled perhaps what I would have said, if I had been able. In reality I said nothing at all, but I heard a murmur, something gone wrong with the silence, and I pricked up my ears, like an animal I imagine, which gives a start and then pretends to be dead. And then sometimes there arose within me, confusedly, a kind of consciousness, which I express by saying, I said, etc., or, Don't do it Molloy, or, Is that your mother's name? said the sergeant, I quote from memory. Or which I express without sinking to the level of oratio recta, but by means of other figures quite as deceitful, as for example, it seemed to me that, etc., or, I had the impression that, etc., for it seemed to me nothing at all, and I had no impression of any kind, but simply somewhere something had changed, so that I too had to change, or the world too had to change, in order for nothing to be changed.'
'At last I began to think, that is to say to listen harder.'
When the teacher points at a thing while naming it, showing the child a thing and saying a word, it might seem as if he's doing two things, and thus that an analysis of the things being done results in a certain kind of problem to be solved, understanding how the associative connection between word and thing could enable the child to manage what he must. But why not say the teacher's doing one thing?
In some form, Augustine's description of how he learned to speak, quoted in §1 of the Investigations, does speak of his coming into a voice, learning to use his. So one subtle effect of Wittgenstein's shopkeeper example in §1d, which is (almost!) given without any reference to speech, and explicitly involves transmitting the order for 'five red apples' to the shopkeeper in writing, is that Wittgenstein has described a way of operating with words, using language, in which there is no call whatsoever to talk of 'speaking', nor even of 'sounds' which might or might not be understood. Voice has been omitted. Thus one difference between the shopkeeper example and the builders' language of §2 which, in terms of the text, seems functionally to replicate much of what was already contained in §1, is that in the builders' language, the voice is back: A and B use a language consisting of words which A 'calls out', in response to which B brings the stones he's learned to bring as called.
In §6, expanding scope to consider how training in such a language would proceed, Wittgenstein notes that it would include the teacher's pointing to an object in order to direct the child's attention to it, while uttering the word for the object. This sets up an 'associative connection' between word and thing, he says. Here, one of the earliest instances of his touchiness toward mental images follows almost immediately: one might think (angling toward elaborating some notion of 'meaning' thus established) that an associative connection's being set up between word and thing means that an image of the thing comes before the child's mind when he hears the word, but Wittgenstein pushes back by insisting that this would not, save certain special cases, be the purpose of that word, viz. to evoke that image; and in the builders' language in particular, it is not.
Why the pushback here? It's inconspicuous, but the idea suggested to explain what's meant by 'associative connection' posits that, upon hearing a word, something else comes before the child's mind; something that, more or less, also belongs to another sense modality. He hears the word, 'sees' the image of the object associated with it.
In the shopkeeper example of §1d, the shopkeeper responds to the written order, 'five red apples', by performing a sequence of actions: he opens the drawer containing the apples, looks up the color red in a table, and then counts out five apples from the drawer, each possessing the color of the sample in the table. The drawer in particular can seem like a curious detail, gratuitous. But barring a case where apples would be individually labeled with the word 'apple', in writing, say tagged with it as later suggested in §15, storing them in a drawer labeled with the word 'apples' seems the next best thing. It sets up a complex situation, though. The transposition of the order from speech (as we might expect in §1) to writing means that 'understanding' the order can, up to a point, be imagined to involve simply matching visual correspondences: these scratches on this slip of paper, the same scratches on this piece of cardstock slipped into a slot on the front of the set of drawers. At the same time, though, this suggests that in some sense the word is nothing like the thing: the shopkeeper can fill the order by matching visual correspondences between different inscriptions of the word, but the word doesn't look at all like the apples it refers to. The shopkeeper is lucky the drawer is there to help him out! In effect, it becomes a metaphor for the classificatory function of the word, or common nouns generally. Yet applying the same sort of reasoning to 'red' multiplies the problems: again, the inscriptions of 'red' on the order and the table look sufficiently alike for the shopkeeper to perform the lookup, but the sample next to 'red' in the table doesn't look at all like the word (and the table itself just barely has any visual similarity to the word, by virtue of having the word written somewhere on it!). Then the action the shopkeeper performs, to take an apple from the drawer which has the color of that sample, reaches back to awaken a crazy question about the first step: if it's by matching the color of a thing the shopkeeper has, to the color of a thing he has yet to have, that he knows which to take, then how is it that in picking the drawer to open in the first place, he knows that to fill the order means to take those, the apples there in the drawer? After all, he doesn't have an apple, they're in the drawer. Worse still!—taking the word 'five' into account, with no numerical aide-mémoire available, say with which to match inscriptions of 'five', the shopkeeper now has to do something which transposes the linguistic activity with written words into a spoken activity, counting to five; and with each number he says (he knows them by heart, Wittgenstein adds), he takes an apple of the appropriate color from the drawer—thus applying the word 'five' in such a way that not only does matching of the number-word with each apple seem out of the question (save maybe the first), since there's nothing oneish or twoish or fiveish about the apples he takes from the drawer to recommend a match, as there is with the red sample and red apples, but this step (which itself multiplies the steps, or consists of steps) awakens its own crazy question about the preceding step: how is it that in picking apples of the appropriate color in the first place, he knows that each has its own color, as the sample has its own color (each being its own thing, different), but these in turn can have, at once, 'the same' color? After all, if he counts them, the colors red, the red things, he comes up with two (or five, or six): or is it one, one red?
The images of things proposed in §6 to explain the idea of an associative connection between word, as heard, and thing, might appear as if they enabled the child to make the transposition and thus to reliably recognize things, given utterances of their names: the sound gets connected to an image (which is not at all like the sound) which then looks like the thing, and the child as it were matches the mental image to the thing, or matches the remembered image to the present image upon actually seeing the thing. But the complex of externalizations effected in the peculiar description of the shopkeeper underscores how, internal or external, positing some kind of performance of a 'match' still stands in need of some explanation of what is done, what one learns to do; and even the simplest examples of such doing quickly seem to make the very idea of matching appear of limited value, unenriched by some idea of multiform activity. Particularly a matching which is attracted to the need to pair sensory modalities with one another, inner mediator and outer object or experience.
In some sense, no such pairing is needed for, let's call it, the activity modality. To speak is already to do something; as are even to make sounds, noises. And so is to respond to what someone says.
So it might be that by omitting the voice from the shopkeeper example—which could even be completely done, with the shopkeeper counting silently, to himself—and thus bracketing all sense of the initiating order being merely some sounds understood in such-and-such a way to refer to such-and-such things, as in Augustine's description, Wittgenstein is able to bring out the surprising degree to which it condensed within itself, as an act, all the shopkeeper's complex activity, as an act in response.
(Condensed, or seemed to.)