Found in Translation: Measuring Language Relatedness Through Cognates

Abstract

This paper examines semantic cognates and their ability to measure the relatedness of various languages. Our project builds upon previous work using the Levenshtein algorithm as a tool for cognate detection. The goal was to utilize the Levenshtein algorithm to discover semantic cognates in English, Romance languages, and Germanic languages, and to determine whether this calculation could accurately reveal the relatedness of English to each of these language families. The resulting data shows through extrapolation that (1) Romance languages are more closely related to one another than Germanic languages are, and that (2) the English lexicon is more closely related to Romance languages than it is to Germanic languages.