gradual accumulation of the parts until the whole structure of the language has been
built up’. However, language does not seem to function in such a way as, according to
Lewis (1996: 13), ‘the success of the whole is much more than the success of the
apparent component parts’. Newmark (1966), stressing that classroom language
learning is not an additive process, considers the isolation of parts from wholes
‘artificial’ and recommends the provision of ‘natural contexts’ where no attempt is
made to select and grade the input for the learner. Yalden (1987: 28), commenting on
the consequences of the synthetic approach stresses that it leads to a grammatical
syllabus, i.e. one which is structured and graded according to linguistic criteria only.
As Ellis (1990: 27) criticises, ‘Implicit in audio-lingual learning theory was the belief
that patterns and items could be learnt in the order they were practiced…No thought
was given to the inherent learnability of specific linguistic features’.
Finally, a third implication which arises from the notion of language learning as a
habit formation is that, within the audio-lingual framework, second language learning
is a process completely different from that of learning a first language, as the learner
has to forget a number of old habits and to acquire some new ones. Fries (1945),
quoted in Richards and Rodgers (1986: 46), stresses that ‘…the problems of learning
a foreign language were attributed to the conflict of different structural systems (i.e.
differences between the grammatical and phonological patterns of the native tongue
and the target language)’. Thus, the notion of ‘interference’, acquires a central role in
behaviourist accounts of SLA. Bright and McGregor (1970: 236) see the grammatical
system of the first language as interfering with the smooth acquisition of the second.
According to Ellis (1999: 21-22), two notions are becoming central in behaviouristic
accounts of SLA, those of ‘transfer’ and ‘interference’. According to these, when
there are no differences between the L1 and L2 there is positive transfer of structures
and the learning of L2 is actively aided by the knowledge of L1. However, when there
are differences between L1 and L2, there is negative transfer of old habits which thus
interfere with the emerging new ones and this is how errors occur. Lado (1957: 2)
expresses such a view clearly by claiming that ‘Those elements that are similar to his
[the learner’s] native language will be simpler for him, and those elements that are
different will be difficult’. As, according to behaviouristic learning theory (see section
1.1.1), errors are considered undesirable, a concern of the audio-lingual method was
to predict the errors the learners will make by comparing the learner’s first language
with the second language she is trying to learn, a method known as ‘contrastive
19