汉语词汇产生中语义、语音层次之间的交互作用

采用3种方法考察汉语词汇产生中语义激活（词条选择）和音位编码这两个阶段之间的关系。实验一、二分别采用图片命名法和干扰字命名法，发现目标图片（“牛”）的语义相关词（“羊”）的语音信息得到了激活，说明汉语词汇产生中存在多重语音激活，音位编码在词条选择完成之前既已开始。实验三采用语义范畴判断法，考察语音信息的激活是否会向上反馈到词条和语义层次，发现目标图片（“羊”）的同音字（“阳”）不会促进对图片的语义判断。研究表明，多重语音激活与语音激活向词条和语义层次反馈是两个独立的，可以分割的问题。虽然实验一和二的结果符合相互作用理论，实验三的结果符合模块化观点，但就总体而言，本研究更加支持词汇产生中的模块化观点

THE INTERACTION BETWEEN SEMANTICS AND PHONOLOGY IN THE SPEECH PRODUCTION OF CHINESE

Three sets of experiments were conducted to investigate the relationship between semantic activation (lexical selection) and phonological encoding in the speech production of Chinese. The issue of whether semantic activation and phonological encoding is interactive or discrete can be divided further into whether there is multiple phonological activation and whether there is feedback from phonological activation to lemma and semantic representation. Experiments 1 and 2 employed a picture-word interference paradigm in which a Chinese character was superimposed on a picture. In Experiment 1, picture naming was delayed both by a character semantically related to the picture name and by a homophonic character of the semantic competitor. In Experiment 2, naming of the semantically related character and its homophone was both facilitated by the picture. These results were consistent with the interactive view that phonological encoding does not wait until lemma selection is finished and phonological information of both the target word and its close semantic competitors is activated in speech production. Experiment 3 used a semantic categorization task in which the difficulty of categorization was varied across three sub-experiments. Subjects were asked to make speeded semantic judgment to pictures onto which characters semantically related to the pictures or characters homophonic to the picture names were superimposed. Facilitatory effects were observed for the semantic condition but no effect was observed for the homophone condition. Thus phonological activation had no feedback influence on the semantic activation of the picture, consistent with the modular view. These findings suggest that the issue of multiple phonological activations and the issue of phonological feedback to lemma and semantics should be differentiated in theories of speech production. While the modular view could accept the suggestion of multiple phonological activation, the interactive view has more difficulties in accommodating the present findings.