A stash of rare Haitian Creole recordings made by
Carnegie Mellon researchers in the late 1990s is providing a valuable translation tool to rescue workers in Haiti.
Created by the university's
Language Technologies Institute (LTI) for a project sponsored by DARPA, the database has preserved otherwise hard to obtain speech recordings. While the official language in Haiti is French, Haitian Creole is the most widely spoken, having evolved since the Haitians overthrew the French colonists more than 200 years ago.
The street language is so different from French today that it's often difficult for a French speaker to understand, says Robert Frederking, senior systems scientist and a lead on the DARPA project. The data was recently rediscovered during LTI's move into the Gates Hillman Center.
"This is cutting edge research," he adds. "We don't want to insert ourselves into the crisis. Instead we decided that the best thing to do is to release the data and produce something that may be useful."
The recordings were made public
online last week. Microsoft Research is using them to develop an experimental, web-based system for translating between English and Haitian Creole;
Translators Without Borders, a Paris-based non-profit, is using it to create and distribute a medical triage dictionary.
LTI researchers are working on their own updated translation system for Haitian Creole that would use the latest translation technologies. LTI. with a focus on machine translation, speech processing, information retrieval, text mining and computer-assisted language learning, is part of Carnegie Mellon's School of Computer Science.
Sign up to receive Pop City each week.Writer:
Debra Diamond SmitSource: Robert Frederking, Carnegie Mellon University
Image courtesy of
flickr.com