本研究以四個實驗針對螢幕上所呈現的配對刺激，進行不同的精緻登錄，加強配對聯結的整體性，藉以討聯結整體性是否可以產生圖字配對新聯結的內隱記憶。實驗一的作業以抄錄加強知覺-動作的聯結整體性;實驗二的作業以造句加強語意的聯結整體性;實驗三的作業以產生互動心像加強心像整體性;實驗四將配對刺激的呈現結合為一物體以加強物理上的聯結整體性。結果與作者先前的研究相比，聯結整體性登錄雖然降低了項目的內隱記憶，但仍無法產生新聯結的內隱記憶。這些結果比較支持激發理論。根據激發理論，既存的心理表徵是內隱記憶的必要條件。新聯結本身並沒有任何既存的心理表徵，因此無法產生重複促發效果的內隱記憶。

We examined in four experiments the effects of unitization on implicit memory for picture-word pairs. Copying picture-word pairs was used in Experiment 1 to create perceptual-motor unitization, sentence-generation was used in Experiment 2 to create semantic unitization, interactive imagery was used in Experiment 3 to produce conceptual unitization in imagery, and physical integration was used in Experiment 4 to produce object-based physical unitization. In accord with previous findings that support the activation view of implicit memory, results show significant item-specific repetition priming but insignificant association-specific repertition priming. Suggestions for future research are also discussed. Repetition priming take place when better performance occurs in response to previously studied items compared to unstudied items, without conscious or intentional retrieval of these previous episodes. Two theoretical accounts have been proposed to explain the mechanism through which repetition priming manifests, with the activation view (Morton, 1979) explaining repetition priming as re-activation or modification of per-existing representations whereas the episode-based model (Jacoby, 1983; Tulving & Schacter, 1990) accounting the phenomenon as flunency in processing the reinstatement of a previous episode. The necessity of pre-existing representations in producing implicit memory thus is the crucial test between the two theoretical accounts. Results from studies of implicit memory for new associations that do not have preexisting representations, however, are equivocal in supporting either account. Furthermore, studies of repetition priming for newly learned picture-word associations fail to show any association specific repetition priming (Dean & Young, 1996; 1997; Chao, Yeh, & Huang, 2000). In contrast, explicit memory of new associations can be obtained after a single learning trial. That previous studies, using a matching task in both study and test phares, failed to fined any association-specific repetition priming in picture-word pairs may result from no association being formed in the study phase to show its influence. To strengthen the association between pictures and words, this study employed four methods that have been shown to enbance relational processing between unrelated items. From conveging results of the four experiments, we find significant item-specific repetition priming but fail to find significant association-specific repetition priming. Because representations pre-exist for items and not for new associations, the results hence support the activation view of implicit memory.