Quantifying cognateness: A comparison of methods for determining cognate status

Cognates, words with overlapping form and meaning, are used extensively in psycholinguistic investigations of bilingual processing. Previous research varied in the use of different qualitative and quantitative descriptions of cognates. Definitions have relied upon researchers’ perceptions, and decisions on cognate status have been defined by scores from either bilingual perceptual data or theoretical cross-linguistic differences. To determine which theoretical method for comparison most closely aligns with perceptual ratings of spoken words, we asked Spanish-English bilinguals to rate the similarity of sets of cognate pairs they heard. Rating scores were compared to 5 quantitative methods: standard methods such as orthographic and IPA Levenshtein distances, Van Orden’s, and Weber’s; plus, a novel method weighting phonological features. Comparisons of methods showed that orthographic differences were more predictive than phonological differences, suggesting that orthographic comparison methods are valid, and that orthographic representations are activated during auditory processing.