In 1999, Professor Steven Weinberger’s phonetics students started a project. Each student would bring in an audio tape of a non-native speaker talking in English.
The class would analyze and transcribe the speech and “It was quite instructive,” Weinberger said. “Most of my English phonetics students are interested in teaching English to non-natives.” In examining the different accents of non-native speakers, these students learned how to best teach them English and soon the number of transcripts began to multiply.
“Then, the university provided server space, and I had to learn HTML,” said Weinberger. “We posted the phonetic transcriptions on the site for everyone to see and comment upon and soon we began to get noticed.”
More than 10 years later, the former class project has been renovated by professional web-builders, earned funding and relied on a collaborative effort from students, faculty and others to grow into a unique learning tool and internationally-recognized database called the The Speech Accent Archive. The online database has 1,111 different language samples -- audio clips that are paired with a transcript. The archive receives more than one million hits per month, according to Weinberger.
“The goal is to continue to gather accents from non-natives and natives alike,” said Weinberger. “It's free, it’s accessible and it provides valuable data. I think the most unique part of it is that it has more than 1,000 talkers speaking their accents with the same paragraph. This way, anyone can do an instant and systematic comparison.”
The archive is particularly useful for English teachers, actors who need to learn an accent for an upcoming role in a film, engineers who train speech recognition machines, linguists, phoneticians, speech pathologists and others.
“We would like to demystify the speech accent,” Weinberger said. “When we listen to somebody talk to us, we immediately and automatically determine if their speech is like ours or not. We may not be able to geographically place their speech, but we can tell how different the speech is from our own. At the same time, and not so automatically, we make certain biased social judgments about this talker. When listening to how they speak English, we may judge a British English speaker as ‘intelligent’, or a French speaker as ‘alluring,’ or a German speaker as ‘unfriendly.’ These types of judgments are learned from our society. “
The archive is due for a renovation, which will include a Google mapping system and an upgraded user-friendly search function that will enable users to search for individual sounds.
“We are also in the process of designing a computational tool for automatically comparing two different accents,” said Weinberger. “The output of such a device will allow us to instantly analyze various accents and we will be able to tell the precise differences between, say, a Zulu accent and a Vietnamese accent.”
Accent Archive Transcript
“Please call Stella. Ask her to bring these things with her from the store: Six spoons of fresh snow peas, five thick slabs of blue cheese, and maybe a snack for her brother Bob. We also need a small plastic snake and a big toy frog for the kids. She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.”
Each of 1,111 audio clips feature a native speaker reading this text. Explore the Speech Accent Archive for yourself.
February 19, 2009