Home Linguistics & Semiotics A data-driven approach to finding significant changes in language use through time series analysis
Chapter
Licensed
Unlicensed Requires Authentication

A data-driven approach to finding significant changes in language use through time series analysis

  • Andrew Kehoe , Matt Gee and Antoinette Renouf
View more publications by John Benjamins Publishing Company

Abstract

This paper conducts a diachronic study of language change in a corpus covering almost 30 years of mainstream UK news text. In our previous studies, several databases were compiled from the corpus, including diachronic records of word frequency, collocation and morphological analysis. Upon user enquiry, our WebCorp Linguist’s Search Engine produced tailored output from these resources. The system was therefore passive, requiring a word or phrase to be specified before querying the databases. The aim now is to extend the data-driven functionality to track the frequency of words in the corpus across time automatically and alert users to statistically significant change patterns. Three tests are employed to find upward and downward trends, sudden jumps in frequency, and seasonal variation.

Abstract

This paper conducts a diachronic study of language change in a corpus covering almost 30 years of mainstream UK news text. In our previous studies, several databases were compiled from the corpus, including diachronic records of word frequency, collocation and morphological analysis. Upon user enquiry, our WebCorp Linguist’s Search Engine produced tailored output from these resources. The system was therefore passive, requiring a word or phrase to be specified before querying the databases. The aim now is to extend the data-driven functionality to track the frequency of words in the corpus across time automatically and alert users to statistically significant change patterns. Three tests are employed to find upward and downward trends, sudden jumps in frequency, and seasonal variation.

Downloaded on 20.9.2025 from https://www.degruyterbrill.com/document/doi/10.1075/scl.105.10keh/html
Scroll to top button