Open AccessITEM PERFORMANCE ACROSS NATIVE LANGUAGE GROUPS ON THE TEST OF ENGLISH AS A FOREIGN LANGUAGE
Author(s) -
Alderman Donald L.,
Holland Paul W.
Publication year - 1981
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/j.2333-8504.1981.tb01251.x
Subject(s) - test of english as a foreign language , test (biology) , foreign language , language proficiency , linguistics , language assessment , differential item functioning , disadvantage , psychology , first language , variety (cybernetics) , natural language processing , computer science , mathematics education , item response theory , artificial intelligence , psychometrics , developmental psychology , paleontology , philosophy , biology
The Test of English as a Foreign Language (TOEFL) was examined for instances in which the item performance of examinees with comparable scores differed according to their native languages. A chi‐square procedure, sensitive to deviations of less than ten percent from the expected frequencies of correct item responses across several language groups, revealed significant differences on seven‐eighths of the TOEFL items. Reviewers familiar with particular languages could attribute the relative advantage or disadvantage of those language groups on a specific item to linguistic similarities or dissimilarities with the English language. Reviewers could not, however, identify which items would exhibit differential performance across groups based upon inspection of a test form and answer key alone. These findings suggest that examinees' performance on given items in a test of proficiency in a second language will vary according to linguistic contrasts with their native language and that statistical procedures will be necessary for identifying items with exaggerated or unexplained differences across language groups.