Abstract: Historical overview The general consensus in the 1950s and 60s was that learners’ errors could be predicted by comparing and contrasting the grammars of their L1 and of the target language; where there were di¤erences, there was likely to be error. This belief was rooted in a behaviourist theory of language learning whereby learning was equated with ‘habit forming’: the habits of the L1 were believed to be ‘transferred’, and regarded as ‘interfering with’ the newly-acquired habits of the L2.