Campus News Society & Culture

NSF linguistics grant to support study of Southern language variation

Athens, Ga. – With the aid of a grant from the National Science Foundation, University of Georgia linguistic researchers will be isolating and identifying the specific variations in speech that make Southerners sound Southern.

Despite the uniformity found in dictionaries, extensive variation exists in the way that people actually use language. Linguists can break down differences in speech sounds-found in accents-to isolate and identify variations in spoken language.

The researchers will use computer software to analyze 64 interviews with speakers from Florida, Georgia, Tennessee, Alabama, Mississippi, Louisiana, Arkansas and Texas recorded from 1968 to 1983.

“We hope to document the wide range of pronunciations in the South, both to provide a database of that variation and to demonstrate how to model that variation,” said William Kretzschmar, Harry and Jane Willson Professor in the Humanities in the UGA department of English and principal investigator on the grant. “We think there are good chances for industrial uses of the database, to make automated airline phone systems work better for example, with all the different ways of speaking their clients have.”

Transcriptions of the interviews, the vowel pronunciation data, and the visualizations will be presented on the website of the Linguistic Atlas Project.

“We will be using methods of computer analysis that have not previously been fully exploited to study the way Southerners’ language varies,” said Margaret Renwick, assistant professor jointly appointed in the department of Romance languages and the Linguistics Program at UGA and co-principal investigator on the project. “A lot of research exists on dialectal speech based on small samples, but we now have the tools and technology to do it on a larger scale, and the data will provide much greater detail about how people talk.”

The wider range of information from the automated phonetic analysis will show the full range of variation in the way an individual speaks. The researchers will, in turn, apply the results to sociolinguistic questions on a broader scale. For example, do speakers in Georgia, Arkansas and Texas say words like ride or wide in the same distinctly Southern way? Or, how have speech patterns among African-American communities changed across space and time?

The project will employ undergraduates through the Research Experience for Undergraduates program to transcribe the interviews. The team will then explore the transcriptions and corresponding audio. The result will be a large-scale body of annotated, dialectal speech with many potential applications.

“We’ll be able to determine what historical speech was like, particularly in the South, and we’ll be able to test different linguistic models including language change, language diffusion and speech patterns,” Renwick said.

The 64 hours of audio recordings that will be analyzed include interviews up to 10 hours long with individual speakers, including participants born in the 19th century. Other recordings include two speakers from the same region but of different generations, enabling analysis of how some Southern speech patterns have developed over time. In all, there are 372 hours of digital audio interviews.

In the first stage of the research, vowel pronunciations will be extracted from a list of 78 words that were elicitation targets in the interviews, plus additional words found to occur frequently in the interviews such as color terms, up to a total of 300 words.

The second stage of the project will create visualizations to determine the dimensions of variation in vowels per speaker, social category and geographic area. The science of complex systems will model the results.

“Complexity science suggests that, while there will always be a wide range of variation in language, our speech should follow the 80/20 rule. That is, about 80 percent of what we actually say comes from only 20 percent of the different ways we have of saying it, and vice versa, only 20 percent of what we actually say comes from 80 percent of the different ways we have of saying it,” Kretszchmar said. “Our automated phonetic analysis will, for the first time, give us enough data to test this idea in depth.”

“Here at UGA we were early to start with automated phonetic processing, so it will be great to go full scale with this grant. No research thus far has processed conversational interviews quite so far from what most people would consider to be normal pronunciation, so it will be quite an achievement when we do it.”

The project has been supported by the Complex Systems and the Humanities research cluster, part of the Digital Humanities Initiative of the Willson Center, the UGA Libraries, and the UGA Press,