Analyzing and interpreting automatically learned rules across dialects
Nancy F. Chen, Wade Shen, Joseph P. Campbell · 2012
In this paper, we demonstrate how informative dialect recogni-tion systems such as acoustic pronunciation model (APM) help speech scientists locate and analyze phonetic rules efficiently. In particular, we analyze dialect-specific characteristics auto-matically learned from APM across two American English di-alects. We show that unsupervised rule retrieval performs sim-ilarly to supervised retrieval, indicating that APM is useful in practical applications, where word transcripts are often unavail-able. We also demonstrate that the top-ranking rules learned from APM generally correspond to the linguistic literature, and can even pinpoint potential research directions to refine existing knowledge. Thus, the APM system can help phoneticians ana-lyze rules efficiently by characterizing large amounts of data to postulate rule candidates, so they can reserve time to conduct more targeted investigations. Potential applications of informa-tive dialect recognition systems include forensic phonetics and diagnosis of spoken language disorders. Index Terms: informative dialect recognition, rule retrieval, phonological rules, forensic phonetics