In the South, “the devil’s beating his wife.” In the Northeast, a “sunshower” is falling.

These are both popular figures of speech for rain on a sunny day, according to an online survey being conducted by Associate Professor of Linguistics Bert R. Vaux.

Other questions on the survey asked about lexical differences like “sneakers” versus “gym shoes,” “sub” versus “grinder” or “hoagie” and “water fountain” versus “bubbler.”

The online survey has allowed Vaux and to gather a large amount of data inexpensively and with a relatively small amount of effort, said Scott A. Golder ’03, who works with Vaux. The survey will help Vaux compile an atlas of regional dialects that will be the first such work in decades.

The survey includes 122 questions ranging from the proper pronunciation of the word “poem” (with one syllable or two) to the name of the grassy area in the middle of a street (boulevard, midway or island, among others).

More than 25,000 self-selected respondents from around the world have taken the survey since it went online three years ago.

According to Vaux, the idea for a new and updated atlas occurred to him in 1997 when he began teaching Linguistics 80, “Dialects of English.”

“I…realized that none of the existing dialect grammars or dictionaries actually contained forms that were relevant today; they were all based on the speech of old white farmers from the 1920s…how many students today know what a whiffletree or bonny-clabber are?” Vaux wrote in an e-mail.

According to the Oxford English Dictionary, a whiffletree is a crossbar on a plough, while bonny-clabber is a type of sour milk.

A new atlas would assist dialectologists by allowing them to track regional dialect trends and population migrations, said North Carolina State University linguistics professor Walt Wolfram.

Because migration patterns have changed since the early 20th century, there’s a need for updated lexical information, Wolfram said.

“Dialect atlases are useful tools for mapping movements of the changing social landscape,” he said.

Despite the homogenizing influence of national television networks, linguists say that regional differences in dialect appear to be increasing. According to Golder, new figures of speech and pronunciations arise more quickly than population migration and television exposure can dilute them.

Another reason for the persistence of regional differences is that personal acquaintances are more compelling models for dialect acquisition than are television characters.

“Children appear to find the social and linguistic prestige of their immediate role models to be more salient and compelling than that of TV role models,” Vaux wrote. “Buffy may be cooler than the leader of the cheerleaders at your school. But Buffy is much less likely to make fun of you in front of your peers on the playground.”