Linguistics and Math: Using Math to Understand the History Between Languages
Abstract
We investigate the idea of calculating distances in higher dimensions and then mapping them to 2D data visualizations and observing the results. Applying the experiment to a real-world representation, we consider the linguistic concept of the Swadesh list, a list of 207 standard concepts across languages, and use mathematical methods to analyze the lexical distances between languages. A lexical distance is a measure that defines a method to analyze the similarity or difference between two distinct languages. In Linguistics, the Levenshtien distance represents the number of characters needed to transform one word into another, which fits perfectly with what we are trying to study, since we are looking at how closely written languages are. In this study, our criteria include the Levenstein distance as the defining metric, a minimum of 100,000 speakers of the language, and a required Latin Alphabet. Once we established our criteria, we filtered the initial data set of over 5,000 languages to 232 that fit the necessary prerequisites. An important note to include is that each language consists of its own individual Swadesh list, which allows us to control the words that are compared between the 232 languages. Next, we used the statistical programming language R to create code that took in the various data and generated matrices of the distances, after which we were able to produce various visualizations representing the data points. Our goal is to identify patterns in the grouping of the languages within these graphs to explore historical relationships among the languages.
Faculty Sponsors
Dr. Radleigh Santos
Project Type
Event
Location
Alvin Sherman Library
Start Date
4-3-2024 12:30 PM
End Date
4-4-2024 1:30 PM
Linguistics and Math: Using Math to Understand the History Between Languages
Alvin Sherman Library
We investigate the idea of calculating distances in higher dimensions and then mapping them to 2D data visualizations and observing the results. Applying the experiment to a real-world representation, we consider the linguistic concept of the Swadesh list, a list of 207 standard concepts across languages, and use mathematical methods to analyze the lexical distances between languages. A lexical distance is a measure that defines a method to analyze the similarity or difference between two distinct languages. In Linguistics, the Levenshtien distance represents the number of characters needed to transform one word into another, which fits perfectly with what we are trying to study, since we are looking at how closely written languages are. In this study, our criteria include the Levenstein distance as the defining metric, a minimum of 100,000 speakers of the language, and a required Latin Alphabet. Once we established our criteria, we filtered the initial data set of over 5,000 languages to 232 that fit the necessary prerequisites. An important note to include is that each language consists of its own individual Swadesh list, which allows us to control the words that are compared between the 232 languages. Next, we used the statistical programming language R to create code that took in the various data and generated matrices of the distances, after which we were able to produce various visualizations representing the data points. Our goal is to identify patterns in the grouping of the languages within these graphs to explore historical relationships among the languages.
