So what does this evidence of stages of development in second language acquisition tell us? It seems to suggest a universal process. Pienemann claims that 'sequences of acquisition may be caused by the ease with which certain structures can be processed by the mind' (Pienemann, 1989, cited in Cook, 2001: 29).

Is there a parallel with L1 acquisition?
Krashen and his research colleagues, Dulay and Burt (reviewed in Dulay, Burt and Krashen, 1982: 211-14) compared stages of development in first and second language acquisition of English (negation, question formation and grammatical morphemes) and found that there were broad similarities but also differences. They did not offer much in the way of explanation for their findings, although as we shall see (see section 2.3.2), Krashen has put forward a model which claims second language acquisition is essentially driven by a process which is 'similar, if not identical, to the way in which children develop ability in their first language' (Krashen, 1982: 10)

What is the impact of teaching?
Research (reviewed by Ellis, R, 1994: 628-30) confirms that stages of development are very similar, whether learners have received instruction or not. So does this mean that teaching has no effect on interlanguage development in these areas? Not quite… In general, studies looking at classroom learners have found that they often overproduce particular structures which they have been taught, compared to naturalistic learners. For example, Weinert's (1987) learners of German were initially able to produce correct negative phrases from memory after teaching, but when memory lapsed, their utterances resembled those of naturalistic learners. Similarly, Lightbown's (1983, 1987) francophone learners of English in a Canadian classroom tended to overuse constructions which had been taught (eg the present continuous tense), but with time, this overproduction disappeared and they reverted to forms reminiscent of naturalistic learners. Pica (1983), meanwhile, comparing classroom and naturalistic learners of English from Spanish-speaking backgrounds, found that the classroom learners were more accurate on some forms (eg plural -s) but less accurate on others (eg V+ing). Investigating the acquisition of German word order rules, Ellis (1989) found that English classroom learners showed a similar sequence of development to naturalistic learners, but seemed to make more progress on a complex rule than seemed typical of naturalistic learners.

Rod Ellis's (1994: 635) conclusion is that while taught learners may develop along the same route as naturalistic learners, teaching may help speed up the rate at which they acquire features of the grammar, and their accuracy. It may also be that different learners benefit in different ways from teaching, and that different grammatical structures are more or less amenable to teaching. Pienemann (1984), for example, reports that as a result of instruction on the inversion rule in German, one learner acquired the rule effectively while others did not, and teaching on the use of the copular verb (sein) had no effect on any of the learners. Ellis (1984, cited in Ellis, R, 1994: 624) similarly reports that communicative teaching of wh-question forms to ESL learners in Britain had no clear impact on accuracy of production for the group as a whole, but some individuals did progress significantly.