Abstract
Abstract
English nonsense consonant-vowel-consonant syllables were presented at four different signal-to-noise ratios for recognition. Information theory methods are used to analyze the response data according to segment type and phonological feature, and are consistent with previous studies showing that the consonantal contrast of voicing is more robust than place of articulation, word-initial consonants are more robust than word-final consonants, and that vowel height is more robust than vowel backing. Asymmetrical confusions are also analyzed, indicating a bias toward front vowels over back vowels. The results are interpreted as parts of phonetic explanations for synchronic and diachronic phonological patterns.
Subject
Linguistics and Language,Acoustics and Ultrasonics,Language and Linguistics
Reference104 articles.
1. Allen, J.: How do humans process and recognize speech? IEEE Trans. Speech Audio Processing 2: 567–577 (1994).
2. Alwan, A.; Zhu, Q., Lo, J.: Human and machine recognition of speech sounds in noise. Proc. 13th World Multiconf. on Systems, Cybernetics, and Information, pp. 218–223 (2001).
3. Bahl, L.R.; Jelinek, F.; Mercer, R.L.: A maximum likelihood approach to continuous speech recognition. IEEE Trans. Pattern Analysis Machine Intelligence PAMI-5, pp. 179–190 (1983).
4. Barrett, R.: A grammar of Sipakapense Maya; PhD diss. University of Texas, Austin (1999).
5. Beckman, J.N.: Positional faithfulness; PhD diss. University of Massachusetts, Amherst (1998).
Cited by
60 articles.
订阅此论文施引文献
订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献