Phonemics as the new thing in language testing

Alan Davies posting the following to the L-Test mailing list.

Robert Lado (1961) mentions three approaches to the role of pronunciation: ignoring it is an error; reaching for native pronunciation is wishful thinking; intelligibility is attractive but hard to define.

'Phonemics' he claims ' offers the way out of this problem ... testing pronunciation with a phonemic criterion of accuracy is the new thing' (: 140). He deals with both recognition and production of the sound segments in chapters 7 and 8 and in later chapters with stress and intonation. Because, he says, production presents more problems than testing recognition, his advice is to test production through what he calls partial production, which is wholly paper based. Yesterday's print is today's machine?

In the 1960s, recognition tests of pronunciation by means of phonemic contrasts were popular. I used two in the English Proficiency Test Battery (1965), the precursor of ELTS and IELTS. The first of these contained 65 triplets. Results indicated:

"phonemic discrimination bore little if any relation to the other components (grammar, reading and listening comprehension).... The practice of insisting on the inclusion of a phoneme test seemed rather like requiring a knowledge of astrology for students of meteorology. Hindsight has not dealt kindly with phonemic discrimination, nor indeed with phonetics more generally, displaced as it is from its central role in the linguistic sciences, a place now occupied by theories of social interaction and cognitive science."

(Davies 2008, Assessing Academic English: 21)