Recognition of Non-Native Speakers

Integrated Training Solutions, Inc., proposes a Phase I SBIR project to determine the feasibility of integrating computational sociolinguistics, social theory, and machine translation technology for a standalone, state of the art system to assist in the determination of native speaker status and potential native language of a speaker. The Sociolinguistic Identification and Rapid Extraction of Native Speaker Status (SIRENSS) technology rapidly ascertains whether an individual is a native language speaker or a foreign language speaker of another language. The SIRENSS technology consists of data elements and software which can rapidly extract information from speech and compare it to data elements correlating with native linguistic performance . The SIRENSS, when operated by human(s) in the loop, becomes a system that takes recorded or live speech and extracts phonological indicators of language (accent) and culture. These indicators are compared against the data elements in question. Each indicator will be marked as either matching or not matching the data elements to which it corresponds. When this process is complete, the system will compare the results to the criteria for native speaker status and provide a numerical indication of the probability of that individual being a native speaker of the language of interest.