Understanding dialects in the American Midwest
Cynthia G. Clopper, clopper.1@osu.edu
Ohio State University
Popular version of paper 1pSCb5
"Speech processing and dialect variation in the American Midwest"
Presented Monday June 30, 2008 at 2:20 p.m. in Room 250B
The Midwestern United States is divided into two dialect regions: Northern and Midland. The Northern dialect stretches from western upstate New York, around the Great Lakes, and into Minnesota. The Midland dialect stretches from Ohio to Nebraska and Kansas. In the map below, the Northern dialect region is highlighted in red and the Midland dialect region is highlighted in green.
The Midland dialect is similar to what is typically thought of as General American English. One of the characteristic features of the Midland dialect is that words like cot and caught or Don and Dawn are homophones, whereas in other varieties of American English cot and Don have a different vowel than caught and Dawn. The Northern dialect has many features that distinguish it from the Midland and General American dialects. For example, the vowel in bet is often pronounced similar to the vowel in bat, the vowel in bit is pronounced similar to the vowel in bet, and the vowel in bat is pronounced as a diphthong, like bay-at, or bee-at. The actor Dennis Franz, who played Andy Sipowicz on NYPD Blue, is from Chicago and you can hear examples of these vowels in his speech.
The goal of the current study was to explore how well listeners in central Ohio (in the Midland region) can understand the speech of talkers from the Northern and Midland dialects.
In the first experiment, individual words produced by female talkers from the Northern and Midland dialects were mixed with noise and presented to undergraduates in Ohio. The undergraduates were asked to write down each word that they heard.
For example:
Was the first word guess or gas? It was produced by a talker from the Northern dialect, and therefore may be ambiguous for some listeners between guess and gas. While many of the listeners in the study correctly identified the word as guess, many listeners also misidentified it as gas.
Was the second word fetch or fatch? It was produced by the same talker from the Northern dialect, and therefore may also be ambiguous. However, since fatch is not a real word in English, the listeners in the study were more likely to correctly identify the word as fetch. Thus, the potential for misunderstandings across dialects is shaped not only by the characteristics of the dialects themselves but also by the inventory of real words in English. When a word is ambiguous between two real words in English (like guess and gas), it is more likely to lead to misunderstandings than when a word is ambiguous between a real word and a nonsense word (like fetch and fatch).
In the second experiment, the words bed and bad produced by female talkers from the Northern and Midland dialects were presented to undergraduates in Ohio. The undergraduates were asked to identify each word as bad or bed as quickly as possible.
For example:
This task is fairly easy, and the listeners answered correctly more than 85% of the time for both dialects. However, they made more errors on the Northern words than the Midland words and they also responded more slowly to the Northern words than the Midland words. Thus, the potential for misunderstandings across dialects is not limited to noisy conditions, as in the first experiment, but may also emerge when listeners need to process and respond to speech quickly, as in the second experiment.