Some acoustic correlates of Mandarin tones and their interaction have been tested. Production studies have demonstrated that lexical tones were most clearly demonstrated when produced in citation forms. The phonology specifies how lexical tones can be reduced by way of stress rules and different neutralization rules. Production data of Mandarin Chinese spontaneous speech were used to test these phonological specifications. Acoustic analysis of the production data made it possible to determine whether a one-to-one correlation could be found between the phonological prediction and the actual phonetic output. A low correlation (36.13%) was found. Complementary perception studies show that when forced to identify lexical tones produced in citations, listeners performed almost perfectly. When forced to identify lexical tones edited from spontaneous speech, listeners' performance was relatively poor (41% of correct identification). The results show that the actual production and perception of spontaneous speech is very likely a cognitive decision involving interacting linguistic/extra-linguistic factors. A native speaker's knowledge of the language enables him to produce and perceive only partially specified phonetic information in verbal communication. There appears to be a trade-off effect between the lower level phonetics, phonology and the other "higher" linguistic levels. Phonology and phonetics alone are insufficient in predicting the actual output of sentences regarding tones.; An initial study of the interaction between lexical tones and intonation demonstrates modulation of tones and intonation. Production data of Mandarin simple declarative sentences from spontaneous speech were utilized to test two intonation theories, i.e., the breath-group theory (Lieberman, 1967) and the declination theory (Maeda, 1976; Breckenridge and Liberman, 1977; Pierrehumbert (nee Breckenridge), 1979; Sorensen and Cooper, 1980). The results of the experiment showed a low portion (20%) of Mandarin sentences that could be characterized by the declination theory, whereas 74% of the data could be characterized by the breath-group theory. However, 26% of the data could not be characterized by either theory. We suspect, as Umeda and Coker (1980) have pointed out, that the declination effect is more of a discourse effect rather than a sentential phenomenon, and that a potential universal intonation theory should allow more room for flexibility.
展开▼
