32. Someone coming into a strange country will sometimes learn the language of the inhabitants from ostensive definitions that they give him; and he will often have to guess the meaning of these definitions; and will guess sometimes right, sometimes wrong.
And now, I think, we can say: Augustine describes the learning of human language as if the child came into a strange country and did not understand the language of the [that particular] country; that is, as if it already had a language, only not this one. [Or again: as if the child could already think, only not yet speak. And "think" would here mean something like "talk to itself".]



  1. B, as a real starting-from-scratch learner, should not be confused with B', a pretend-learner, who is learning 'sepia', but who has already been provided with the whole structure of language, including 'naming' and 'colour-words', and just is missing this word - this piece in the jigsaw, that has been artificially removed.
  2. B cannotx learn his first language by just labelling alone. Labelling is part of the learning process (3, 6, 26 "naming is something like attaching a label to a thing") - it just isn't all of it.
  3. W. later, not surprisingly, thought better of the last two sentences. Some editions have a message from Anscombe to this effect.
  4. His target continues to be his old Tractatus self, not Augustine. There is no point in trying to find out what Augustine actually meant. W. is trying  to exorcise his old demons - to understand how he could have been so foolish. Once W.'s language is fully formed, he could fail to notice the place-setting aspect of it. He could fail to notice the necessity for this. He could just focus on the nouns, adjectives, and connectives, as though each one - 'cup', 'table', 'red', and 'on' - works in isolation, with an invisible dotted line linking it to an aspect of the world (the 'ostensive definitions of the words') such that the structure of the world is then copied (more or less, because this is not a fabled 'atomic sentence') by the structure of the sentence: "The red cup is on the table". He could thus think that when A teaches B our language, he just uses ostensive teaching of these definitions - pointing out the dotted linkage lines - pointing at and saying "Table".
  5. Language Learning: How exactly human language learning takes place is not his theme. He is merely presenting a rough model. Therefore problems with the model that he has sketched are not really relevant. For example, we could ask how B manages to evade the problem of conceptual underdetermination (Rabbit parts, gavagai, radical translation...). Are we supposing that the child somehow simultaneously is trying to learn specific words, and guess, from various contextual clues, the place-holders (for example, that 'sepia' is a colour word? Are we supposing that humans have a natural genetic tendency to think of the same place-holders? None of these questions matter. His theme is not how humans learn language but how language works, what it does, and what it cannot do. The process of learning may illuminate this; for example, realising that the teaching process is deeply unreliable might suggest that it is not likely to result - perhaps cannot result - in the creation of a unique, inhuman, metaphysical link (a rule-governed association) between the word and either an aspect of the world, or an aspect of our feelings/sensations.