The Optimality Theoretic Analysis Of Errors In Tone Sandhi Acquisition

This study conducts two experiments which explored the tone sandhi acquisition of the second language learners in different language levels and with different mother-tongue background. We had their errors recorded and tried on analyzing them in an Optimality Theory structure. In experiment 1, we examined the pronunciation of pseudo two-syllable words with the 3rd tones by the second language learners of different levels. The results show that the second langue learners acquisit the tone sandhi rule later than the four single tones, they cannot master the pronunciation of tone-combination in a long period after having mastered the four single tones. Their sequence of acquisition is different from child's mother-tongue acquisition. Their mastering better on true words than pseudo words argues that the subjects is not using the tone sandhi rule but a strategy of memorizing routinized words. The experiment 2 examine the pronunciation of all 16 tone-combinations by the second language learners of the same level but different mother-tongue background , and analyze the recorded errors. The result shows: The Thailand students with a tone-language mother-tongue and other students without any tone-language mother-tongues have similar main correct rate but some differences on each item. The strong errors recorded in the experiment are analyzed by the Optimality Theory and found a common explanation in the flaw of inter-language constraints ranking. This dissertation tries to derive the imperfect constraints ranking by analysing the errors, and reappear the process of their coming out. We tried to supply a new explanation of tone sandhi errors. In a word, the tone sandi rule is not a single rule for sandi, but a series of rules for all tone-combinations. While the rule—the constraints ranking system—has any flaws, they can appear in the output pronunciation as some strong errors.