THE EFFECT OF LANGUAGE BACKGROUND ON FOREIGN-LANGUAGE TEST-PERFORMANCE - THE CASE OF CHINESE, ITALIAN, AND MODERN GREEK

This article considers the validity of applying common assessment inst ruments and scales to assess the language skills of Australian school- age language learners from different L1 backgrounds. I used data gathe red from standardized national tests of reading and listening proficie ncy in 3 heritage languages: Italian, Greek, and Chinese. The tests, k nown as the Australian Language Certificates (ALC), are administered a nnually to a large population of language learners at Year 8 or Year 9 . I applied Mann Whitney and Mantel-Haeszel procedures to data from th e 1993 administration to investigate score differences and differentia l item functioning (DIF) between those with and without a home backgro und in each of the 3 target languages. Content analyses of DIF items a ttempted to identify possible sources of group difference. Findings re veal a strong relationship between home exposure to the language and l evel of performance on the listening and, to a lesser extent, the read ing component of the tests. There is also evidence of differential ite m functioning, although this effect is stronger for Chinese and Italia n than for Creek. The content analysis suggests that the DIF effect ma y be due to ''true'' differences in language ability, rather than to t est bias. Nevertheless, the use of common instruments and reporting pr ocedures for L1 and L2 learners may be inappropriate and demotivating and, in the later years of schooling, could lead to serious misuse of test scores.