Refinement of the matched-guise technique for the study of the effect of non-native accents compared to native accents

Abstract

Finding and selecting representative matched guises of native and non-native English accents is challenging, because cannot be both a native and a non-native speaker of one language. In order to test a method to establish representative matched guises, matched-guise samples were recorded for the non-native accent, Dutch English, and standard British and American English accents. Four speakers, pre-selected by language experts, each produced all three accents. The samples were then evaluated on degree of ‘nativeness’ and ‘standardness’ by three ‘linguistically naive’ listener groups: 40 native British English speakers evaluated the British English samples, 40 native American English speakers evaluated the American English samples, and 40 native Dutch speakers evaluated the Dutch English samples. Despite the language experts’ assumptions regarding the representativeness of the accents produced by the four matched-guise speakers, only two of the four speakers were deemed representative speakers of all three accents. This suggests that representative matched guises for the purposes of (non-)native English accents research can be achieved, but that speaker evaluations by ‘linguistically naive’ native listeners are necessary to ensure valid matched-guise selections.