AN ALTERNATIVE EXPLANATION FOR LANGUAGE LEARNING
As outlined earlier, Popper pointed out that (objective) knowledge can be attained by postulating testable hypotheses using strict universal statements. These hypotheses, in order for them to be useful, and in order to maximise its truth-value, ought to make bold and risky predictions. This epistemological stance can be applied when trying to explain how children acquire language. Fromkin and Rodman’s Introduction to Language say that “children seem to form the simplest and most general rule they can from the language input they receive, and to be so ‘pleased’ with their ‘theory’ that they use the rule wherever they can”. The child has a kind of predisposition to put forth all kinds of conjectures and (rudimentary) hypotheses, which will either turn out to be false, or not. Through trial and error, the child decides which is which. That is, the child begins with no reliable knowledge, but builds up a grammar “by formulating and testing a series of fallible conjectures, choosing between alternative hypotheses compatible with his data by reference to general methodological criteria (notably, strength and simplicity)”. The conjectures that are not falsified are retained and these provide the foundation for further epistemological advances. Sampson theorises that we have an innate ability to reason like this, but this kind of innate reasoning has nothing to do with Chomskyan nativism. Obviously, there must be innateness of some kind in the human brain, as even an extreme empiricist like John Locke says, but it is certainly not as detailed and precise as Chomsky claims it to be.
The fact that the child starts off with no reliable knowledge does not mean that he learns based on a tabula rasa. It means only that his [rudimentary] hypotheses are tentative and subject to empirical testing. As mentioned earlier, even modern-day behaviourists agree that for learning to take place, at least some innate endowment is needed. To repeat a point made earlier, Chomsky admitted back in 1976 that “every learning theory that is even worth considering incorporates an innateness hypothesis”, and later on that same page, that “the question is not whether learning presupposes innate structure – of course it does; that has never been in doubt – but rather what these innate structures are in particular domains”. My contention is exactly that: how much is innate; certainly not as much as Chomsky claims.