中文摘要：

采取内隐启动范式研究汉语词汇产生中的音韵编码过程。实验一和二探测单字词产生中首音相同、韵和声调都相同时的启动效应，结果未发现首音启动效应，却发现了韵和声调相同时的抑制效应。实验三至五探测双字词产生中音节相同、声调相同以及音节和声调都相同时的促进效应，实验结果发现了显著的音节、音节和声调的促进效应，未得到显著的声调促进效应。上述结果表明，音节或“音节＋声调”是汉语词汇产生中的编码单元，音节促进效应发生在音韵编码水平；声调只是节奏框架的一部分，它只能与音节结合在一起才能起作用。结合汉语的特点讨论了研究结果，并与英语或荷兰语的研究结果进行了对比。最后提出了两种可能产生“韵母＋声调”抑制效应的原因，并讨论了记忆可能对实验结果产生的影响。

英文摘要：

A number of different models have been proposed to account for how a word is processed at the stage of phonological encoding. Based on speech error data, Dell＇s model assumes that the phonological unit includes a phoneme, segnent, and syllable. In contrast, Levelt et al.＇s model postulates that the phonological unit might include an onset, nucleus, and coda. Previous studies suggest that different languages might have different phonological encoding units in speech production. Moreover, the role of the syllable in production across different languages is still a rather controversial issue. Speech error data in Chinese indicate that the syllable plays a more important role in production in Chinese than it does in English or Dutch. Therefore, using an implicit priming paradigm, the present study investigated the stage of phonological encoding, especially focusing on the role of the syllable in Chinese word production. Four sets of word pairs served as experimental stimuli. Each set consisted of four pairs. The first word of a pair was the cue word, while the second was the response word. In the homo~aeous conditions, four response words in a set shared certain phonological properties. For instance, the four response words in a set were 夕阳 （sunset,/xilyang2/）, 媳 （wife, /xi2ful/）, 喜事 （good news, /xi3shi4/）, and 细胞 （cell, /xi4baol/）, which shared the first syllable. The heterogeneous conditions used the same word pairs, but assigned them to sets in which there were no shared properties. Participants were asked to memorize the sets of four pairs of associated words, after which they were presented with the cue words, and were required to produce the response words. Five different homogeneous and heterogeneous conditions were used in experiments 1 to 5, respectively. Experiments 1 and 2 examined implicit priming for the shared onset and shared time and tone, respectively, in monosyllabic word production. The results indicated that a reliable shared rime and tone inhibition effect wa