Comparison Of The Realization Means Of Boundary Tone Between Mandarin And English Declarative Corpora

With the rapid development of telecommunication technology especially computer science, speech technology such as speech concatenation, speech recognition and TTS (text to speech) has achieved great progress in the last decade. Now, computer can already produce individual syllable or word with a high degree of intelligibility. However, when these individual syllables or words are concatenated to form a larger linguistic unit, e.g. a phrase or a whole sentence, the concatenated products are far from satisfactory because they sound mechanical, unnatural and boring to human ears and mostly difficult to understand, which has thrusted a tough block for the popularisation of speech technology. To tackle this problem, further understanding about the nature of supra-segmental features (of which, intonation is a major one) of speech is quite necessary. To cater for this need, relevant investigation has been carried out in some languages in recent years and some basic conclusions have been drawn. However up to now comparison of supra-segmental features between different languages has not been undertaken yet. Under this circumstance, this thesis aims to seek out the similarities and differences between Mandarin and English in the way of boundary tone realization with an eye to understand the supra-segmental features from a broader perspective. With reference to previous study results, five aspects are addressed in this thesis, i.e. pre-boundary lengthening, after-boundary speeding up, boundary tone distribution, boundary pitch resetting and pitch movement for boundary tone realization. In the end, the following conclusions are drawn: pre-boundary lengthening and after-boundary speeding up exist in both English and Mandarin, with the degree of lengthening and speeding up differing considerably mainly due to the influence of syllable numbers, semantic strength etc.; boundary tone is distributed differently between Mandarin and English: in Mandarin, L% enjoys absolute domination in all these declarative sentences; but in English, both H% and L% occur; pitch resetting is realized in different ways in Mandarin and English: in Mandarin, for different tone combination patterns, there are different pitch performances for pre-boundary and after-boundary matches( for detail, please refer to section IV 4.3 in the thesis); in English, it is realized mainly by the contrast of frequency valuebetween pre-boundary's last voiced segment and after-boundary's first voiced segment, with different performance for L% and H%; factors determining the tone type perception are different between English and Mandarin: the four factors ( details are in IV 4.4 of this thesis)determining English's intonation perception does not apply to Mandarin, which should be approached from some other perspectives, which is unfortunately beyond the power of this experiment; finally it is proposed that the mental process underlying the two languages' boundary tone perception should be the same one, i.e. both involve comparing with one unmarked standard status (i.e. the standard L% status in one language), however, this needs perceptual and psychological experiment in the future to provide further evidence.