Abstract: At present, there are two accounts of the semantic interference effect in picture-word interference paradigm , which are lexical selection by competition and response exclusion hypothesis .The main difference between them is that the former suggests that semantic interference occurs in lexical phase while the latter accuses that it takes place in the post-lexical phase. Janssen et al（2008）created delayed naming task, in which the presentation of the picture preceded the presentation of the distractor word by 1,000ms , and until the distractor word was present participants could name the picture or read aloud the distractor word depending on the color of the distractor word . Their findings showed that there was semantic interference effect but not frequency effect in delayed naming condition. As a result, Janssen et al（2008）concluded that the semantic interference effect arose at a post-lexical level of processing, not at lexical level. The purpose of the current study was to explore whether the processing in the delayed naming mixed up lexicalization processing rather than a pure post-lexical level of processing. On the basis of the Janssen et al（2008）delayed naming paradigm, The current experiment changed the picture-word superimposed presentation to separation presentation , in which distractors were presented below the pictures, and made use of eye movement technology to investigate the effects in the delayed naming task. The results of the experiment were found: (1) there existed semantic interference effect and frequency effect in the reaction time; (2) compared with the filler item condition in which the distractor words were read aloud , the rate of fixation returning from distractor word to picture was more and the fixation time of picture before named was longer in the delayed picture naming condition ; (3) In the delayed picture naming condition, the fixation time of picture before naming was not only longer in the semantic related condition than in the semantic unrelated condition, but also longer in the low-frequency picture name than in the high-frequency picture name. Thus it can be seen that the delayed naming approach did not guarantee that the information processing completely happened at a post-lexical stage, namely, at a response output stage and it perhaps mixed up lexicalization stage. From the relation of fixation time with semantic interference effect and frequency effect, semantic interference effect maybe arose at the level of lexical selection rather than at a post-lexical level of processing in the delayed naming condition, so the lexical selection by competition hypothesis was supported.

Key words: picture-word interference effect, lexical selection by competition, response exclusion hypothesis, delayed naming, eye movement

摘要： 采用图-词分离呈现及眼动技术探讨在延时命名条件下图-词干扰效应发生的位置。结果发现：（1）延时命名图片时，仍出现图-词干扰效应和频率效应；（2）与命名干扰词相比，命名图片时图片的回视率高，注视时间长；（3）图片的注视时间也出现频率效应和图-词干扰效应。这说明即使在延时命名图片的条件下，目标词汇仍需要再次激活，频率效应和图-词干扰效应是发生在词汇化过程中，反驳反应排除理论的图-词干扰反应选择阶段的观点，支持了词汇化选择竞争的理论。

关键词: 图-词干扰效应, 词汇选择竞争, 反应排除理论, 延时命名, 眼动。