Cognitive psychologists explain this conundrum by referring to the difference between declarative knowledge (knowledge which we can state, such as 'The 3^rd person singular present simple tense form of the English verb ends in -s' or 'Paris is the capital of France') and procedural knowledge (the knowledge in our brains that underlies the performance of a skill, such as swimming or driving a car, which often we cannot state). Much of our procedural knowledge, it seems, may build up from declarative knowledge that we are able to automatize through practice: leading researchers in this field claim that declarative knowledge is 'a major avenue for the acquisition of procedural knowledge' (Anderson and Fincham, 1994 cited in DeKeyser, 1998: 48). However, this is not necessarily the only avenue for developing procedural knowledge: we seem able to 'pick up' some skills implicitly, ie just from observing others performing them and from practising ourselves.

Is using a second language such a skill? To the extent that there are examples of people 'picking up' second languages implicitly with little teaching, it must be. But many would also see second language learning as involving declarative knowledge: language use is based on a complex system of rules (a grammar) and a huge repertoire of items (vocabulary). In these respects, maybe learning a second language is like learning other 'school subjects' like biology and history.