模仿与产生和感知密切相关，研究二语学习者的语音模仿有助于进一步了解模仿中的机制，并验证模仿与感知和产生之间的关系。本研究以国际中文的送气音和不送气音为测查材料（b[p]/p[pʰ]、g[k]/k[kʰ]、d[t]/t[tʰ]、j[tɕ]/q[tɕʰ]、zh[tʂ]/ch[tʂʰ]、z[ts]/c[tsʰ]），对35名美国大学的国际中文学习者的识别、朗读和模仿能力进行测查，探索模仿与感知和产生之间的关系以及模仿的机制，以期为之后的研究提供更多的实验结果，并对国际中文教学提供建议。

结果发现，（1）在国际中文的送气音和不送气音范围内，对于美国学习者而言，识别任务的准确率（81.10%）最低，朗读任务的准确率（95.44%）最高，模仿任务准确率（94.92%）居中，且任务间准确率影响显著（p<0.001），这说明模仿不是最难的任务，可能不需要调动感知和产生的全部技能。（2）对三个任务进行相关性统计分析后发现，根据被试在识别任务和朗读任务中的数据组建起的级联模型与模仿结果的总体相关性（ρ=0.571）最高，说明被试在感知和产生任务中的表现都与模仿任务的结果有密切关联，并且共同影响被试在模仿任务中的表现；感知模型（相关性：ρ=0.563）比产生模型（相关性：ρ=0.429）与模仿结果的相关性更高，说明被试的感知能力与模仿表现的关系更为紧密。（3）感知模型、产生模型和级联模型的预测结果表明模仿是一个复杂的过程，模仿任务中的错误与感知和产生任务中的错误并不完全匹配，支持了模仿可能会绕过部分音系编码的假设。（4）被试在不送气音上的准确率（93.97%）高于在送气音上的准确率（91.67%），且两者间准确率差异显著（p=0.021），说明对于美国的国际中文学习者而言，送气音的学习难度明显高于不送气音。（5）关于年级间的比较发现年级间差异显著（p<0.001），三年级学习者在识别任务、朗读任务和模仿任务中的准确率均显著高于一年级学习者和二年级学习者。在模仿任务中被试的表现随着年级的上升而逐渐提升（一年级学习者平均准确率：93.27%；二年级学习者平均准确率：95.83%；三年级学习者平均准确率：96.01%），说明学习经验对学习者在模仿任务中的表现有积极影响；但是在识别任务（一年级的平均准确率：91.35%；二年级的平均准确率：81.05%；三年级的平均准确率：95.14%）和朗读任务（一年级朗读平均准确率：96.47%，二年级的平均准确率：93.35%，三年级的平均准确率：97.40%）中，二年级学习者的准确率最低，被试的表现会随着年级的上升呈现出先降后升的趋势，这说明二语学习不是一蹴而就的，需要不断巩固和练习以避免退步。（6）关于韵律位置的比较结果表明被试在双音节前位置上的准确率最高，单音节的准确率与目标音在双音节前位置上的准确率差别不大，都在整体上高于双音节后位置上的准确率，但是并没有达到统计学意义上的显著。

综合上述结果，对国际中文教学提几点建议：语音学习不是一蹴而就的，需要反复温习巩固，避免退步；语音偏误有一定的规律性，教师需要多总结得出教学规律，针对不同偏误不能一概而论，需要“对症下药”；学习者的感知和产生能力不一定是均衡发展的，因此在语音教学中发音练习和听力练习都需要重视，当学习者的产生暂时领先于感知时，可以增加感知训练频率；需要重视教学方法的多样性和教师的流动性，增加学习者目的语输入的多样性；语音教学需要合理分类。

Imitation is closely related to production and perception, and studying imitation of the second language learners helps to further understand the mechanisms involved in imitation and to verify the relationship between imitation, perception and production. In the present study, aspirated and non-aspirated sounds were used as the measurement materials (b[p]/p [pʰ], g[k]/k [kʰ], d[t]/t [tʰ], j [tɕ]/q [tɕ ʰ], zh [tʂ]/ch [tʂʰ], z [ts]/c [tsʰ]) on American Chinese learners’ recognition, reading aloud and imitation abilities were measured to explore the relationship between imitation, perception and production as well as the mechanisms of imitation in order to provide additional experimental results for subsequent research and suggestions for international Chinese language teaching.

The research found that (1) within the range of aspirated and non-aspirated sounds in Chinese, for American learners, the accuracy rate of the recognition task (81.10%) was the lowest, the accuracy rate of the reading aloud task (95.44%) was the highest, and the accuracy rate of the imitation task (94.92%) was in the middle, and the difference in accuracy between tasks was significant (p<0.001) , indicating that imitation is not the most difficult task and may not require the mobilization of the full range of perceptual and producing skills. (2) A statistical analysis of the correlations of the three tasks revealed that the cascade model formed from the learners’ data in the recognition task and the reading aloud task had the highest overall correlation (ρ = 0.571) with the imitation results, indicating that the learners’ performance in both the perception and production tasks was closely related to the results of the imitation task and jointly influenced the learners’ performance in imitation; the perception model (correlation: ρ = 0.563) had a higher correlation with the imitation outcome than the generation model (correlation: ρ = 0.429), indicating that learners’ perceptual ability was more closely related to imitation performance. (3) The predicted results of the perception, production, and cascade models suggest that imitation is a complex process and that errors in the imitation task do not exactly match the errors in the perception and production task, supporting the hypothesis that imitation may bypass some of the phonological encoding. (4) The accuracy of learners’ performance on the non-aspirated sounds (93.97%) was significantly higher than that on the aspirated sounds (91.67%), and the difference in accuracy between the two was significant (p=0.021), indicating that the aspirated sounds are significantly more difficult to learn for international Chinese learners in the United States than the non-aspirated sounds. (5) Comparisons between grades revealed that the accuracy rates of third grade learners were significantly higher than those of first grade learners and second grade learners in the recognition task, the reading aloud task, and the imitation task. learners’ performance in the imitation task gradually increased with grade level (average accuracy rate of first-year learners: 93.27%; average accuracy rate of second-year learners: 95.83%; average accuracy rate of third-year learners: 96.01%), indicating that learning experience had a positive effect on learners' performance in the imitation task; however, in the recognition task (average accuracy rate of first-year learners: 91.35%; the average accuracy rate of second grade: 81.05%; the average accuracy rate of third grade: 95.14%) and the reading aloud task (the average accuracy rate of reading aloud in first grade: 96.47%, the average accuracy rate of second grade: 93.35%, and the average accuracy rate of third grade: 97.40%), the accuracy rate of second grade learners is the lowest, and the learners’ performance shows a trend that as the grade level rises the performance of the subjects tends to go down first and then up. (6) The comparison results regarding rhyme position showed that the learners’ accuracy was the highest in the pre-bisyllabic position, and the accuracy rates of monosyllables were not significantly different from those in the pre-bisyllabic position, both of which were overall higher than those in the post-bisyllabic position, but the accuracy of the three positions did not reach statistical significance.

The above results suggest a few suggestions for international Chinese teaching: phonetic learning is not easy, but needs to be repeatedly reviewed and consolidated to avoid regression; phonetic errors have certain regularity, and teachers need to summarize the teaching rules more often, and not to make generalizations for different errors, but to choose the most suitable teaching techniques; phonetic practice and listening practice need to be emphasized in phonetic teaching; in phonetics teaching, both pronunciation practice and listening practice need to be emphasized, and the frequency of perceptual training can be increased; the diversity of teaching methods and the mobility of teachers need to be emphasized; phonetics teaching needs to be reasonably classified.