The BNT-38: Applying Rasch Analysis to Adapt the Boston Naming Test for Use With English and French Monolinguals and Bilinguals.

Abstract

Purpose Currently, there is no reliable instrument to measure naming abilities in bilingual speakers of English and French. The Boston Naming Test (BNT; Kaplan, Goodglass, & Weintraub, 1983 )is a widely used scale for clinical assessments of language function, but it is not suitable to assess bilinguals. Rasch analysis provides a unique and powerful method to establish measurement invariance across language groups that improves reliability of measurement. Method Rasch analysis was applied to a sample ( n = 215) of English or French monolingual and bilingual speakers completing the BNT in either language. Participants included English-French bilinguals ( n = 83), English monolinguals ( n = 72), and French monolinguals ( n = 60). Results The best overall Rasch model fit was obtained after the removal of 22 misfitting items, resulting in a 38-item BNT solution (BNT-38), with a modification of 7 items that showed differential item functioning by language factor. To increase the clinical utility of the BNT-38 in French speakers and bilinguals, we generated ordinal-to-interval conversion tables for monolinguals and bilinguals in English and French. Conclusions Use of the BNT-38 and the associated conversion tables will allow valid comparisons of naming abilities across bilingual and monolingual English and French speakers of different age groups. Applying these tools increases accuracy in measurement of naming ability and higher diagnostic precision in French and English monolinguals and bilinguals.