An Introduction to Applied Linguistics

4 MAGNITUDE ESTIMATION
A way round the definition problem may be to acknowledge that we can at least agree
on who is a non-native speaker (NNS). This in essence is the approach Sorace takes
in the experiments she has reported in a number of papers (Sorace 2003). What
she does is to study advanced NNS (whom she terms near-natives) judgements of
grammatical acceptability using magnitude estimation. This methodology, which
she has pioneered with colleagues and students for interlanguage grammars, is a
highly sensitive and reliable technique for the measurement of change and therefore
for establishing stages in the development of an interlanguage grammar. Her
conclusion, based on experiments with a number of languages, is that ‘even learners
who are capable of native-like performance often have knowledge representations
that differ systematically from those of native speakers’ (Sorace and Robertson 2001:
267). In other words, there is an absolute distinction between NSs and NNSs, a gap
that can never be bridged (but see Birdsong 2004). Although I have suggested that
Sorace’s approach is unlike that proposed by Escudero and Sharwood-Smith, in
fact they are not so different. Sorace’s magnitude estimations are not a world away
from prototypes; and although her study is of near-natives, she needs for all her
experimental studies to obtain comparative judgements also for native speakers. She
uses students drawn from secondary schools and universities. Now what Sorace,
unlike Escudero and Sharwood-Smith, does is to take for granted that we know
what/who a native speaker is. Does she therefore avoid the failure to define which
Escudero and Sharwood-Smith grapple with? I think not: in the first place, her NNS
studies provide results not unlike prototypes and in the second place her choice of
educated NS subjects biases her claims for the knowledge representations of NSs.
152
An Introduction to Applied Linguistics
02 pages 001-202:Layout 1 31/5/07 09:31 Page 152

---