Social dimensions of language testing

Abstract

Toward the end of the academic year, a man with a clipboard turned up at one of the capital city's best high schools. He sauntered from classroom to classroom, ignoring the students and instead engaged in seemingly trivial chitchat with the teachers, twenty minutes at a time. Tell me, what subjects are your specialties? How long have you worked here? Can you explain to me a little about how you prepare your lessons? The inspector didn't seem to be particularly interested in what the teachers said. He only cared about how they said it. Olga Muravyova teaches biology and geography. She laughed nervously as she recalled her meeting with the inspector. 'He wrote a report saying that r understood all the questions, that I answered all the questions, but that r made some errors. That is actually what he claimed,' Ms. Muravyova said. 'Of course that is hard to hear.' After the inspector told her that she had failed the test, he told her to attend Estonian classes, which she has tried to do. But she is 57, an age when it is not easy to pick up a new language. This vignette, based on a stmy in the New York Times (Levy, 201 0), describes part of a test of spoken Estonian required since 2008 of teachers and other civil servants. In this small fmmer Soviet republic on the Baltic Sea, the government has been mounting a determined campaign to elevate the status ofits native language and to marginalize Russian, the tongue ofitsfmmer colonizer. Public schools, where students have long been taught in Russian, are now linguistic battlegrounds and the test itself is a skirmish betv.reen Estonian and Russian. The test and Ms. Muravyova's expe1ience illustrate two social dimen­ sions oflanguage testing: The first is the construct oflanguage knowledge on which the test is built and on which test results are inte1preted; the second is what happens to individuals, societies, and institutions when the test is used-the social consequences of assessment. In this chapter, I describe both of these dimensions and argue that, until quite recently, language tests have been built on incomplete knowledge of the social ground oflanguage in interaction and the social consequences oflanguage assessment.