| 摘要: | According to the prominent working memory model, verbal and non-verbal materials are stored in the forms of phonological and visual-spatial representations, respectively. However, previous studies of non-alphabetic languages (e. g., Chinese and Japanese) also found the visual similarity effect in verbal short-term memory (STM) tasks. It was unclear whether the effect resulted from the contribution from long-term memory or from the supplementary function of the visual-spatial sketchpad. The aim of the present study is to examine the nature of orthographic representation underlying Chinese characters, and to investigate how the orthographic representation is involved in verbal STM. In the first part of the study, three dual-task experiments were conducted to examine the dissociation between orthographic and visual-spatial representations of verbal STM for Chinese characters. The retention of orthographically similar and dissimilar Chinese characters was intervened by a task of choice-reaction (the control condition), matrix matching (visual-spatial interference), or character matching (orthographic interference). Although all intervening tasks decreased memory performance, the magnitude of the orthographic similarity effect in phonologically dissimilar characters was selectively reduced by interfering character matching but not by matrix matching. These results support that the orthographic representation for Chinese characters might be different from the nonverbal spatial information underlying the retention of Brook’s matrices. The second part of this study examined whether participants actually rely more on phonological representation in remembering the orthographically similar than dissimilar characters. Because the difference between the memory accuracy for orthographically similar and dissimilar conditions was not reduced by intervened homophone-judgment or concurrent articulatory suppression, it is suggested that the convergent effect of orthographic similarity does not result from more involvement of phonological representation in the orthographically similar condition. In the third part of the study, the functional magnetic resonance imaging (fMRI) technique was applied to determine whether the neural mechanisms supporting the retention of orthographic representations reside in the brain network of visual-spatial or verbal STM. In an experiment with blocked design, participants silently maintained three Brook’s matrices and four Chinese characters in the spatial and verbal STM tasks, respectively, with orthographic similarity manipulated in the verbal task. The results revealed typical neural networks associated with stimulus retention in the spatial and verbal STM tasks. More important, when contrasting the retention of orthographically similar to dissimilar characters, there were more activations in the left middle frontal gyrus and the right superior parietal lobule. The former activation partially overlapped with the brain regions that are more activated in verbal than visual-spatial STM, while the latter activation partially overlapped with the brain regions that are more activated in visual-spatial than verbal STM. Such results were consistent with the behavioral findings to suggest that Chinese orthography involves linguistic-specific knowledge in addition to spatial representation. In summary, the findings reported in the present study provide strong evidence for co-existing contributions from phonological and orthographic representations to verbal STM for Chinese characters. These results also indicate that the maintenance of Chinese orthographic representations involves abstract linguistic-specific knowledge in addition to visual-spatial analyses of word forms. Theoretical models of verbal STM should take characteristics of individual languages into account to accommodate the findings from linguistic materials with different degrees of homophony. 在過去與短期記憶相關的研究中，廣泛認為語言與非語言(例如視覺或空間材料)的短期記憶處理歷程是可以獨立的。語音表徵被認為是主要用來短暫記住語言材料的線索，只有當這個主要線索沒有辦法幫助受試者記憶時，才會觀察到字形這個與視覺相關的表徵對語言短期記憶的影響。由於中文具有高比例的同音字，因此本實驗室過去研究發現，即使在語音線索很容易使用也有幫助的狀況下，字形相似仍會降低短期記憶的表現。相關實驗進一步說明，字形相似性不能僅靠視覺相似性來解釋，其中中文使用者對文字結構的知識，例如:音旁、意旁的組合與語音的對應是字形處理關鍵。 本研究包括三個部分，分別採用行為實驗以及功能性磁振造影(fMRI)技術試圖澄清中文字形在短期記憶中的處理歷程，是否獨立於視覺與空間的處理歷程。實驗一採用雙作業派典(dual-task paradigm)，在操弄刺激材料之語音與字形相似性之情況下，同時操弄干擾作業的特性，間接探討字形表徵之處理與視覺空間處理之關係。實驗結果呼應過去的資料，發現即使在語音線索可幫助受試者記憶的狀況下，受試者的表現仍然受到字形相似性的影響。此外，本實驗結果發現，字形相似性效果會受到附加的筆畫判別作業影響而略為減小，但不受額外之空間判別所干擾。 實驗一的結果似乎支持我們所提出的假設，即中文字形處理包括語言知識的層面，不能單由視覺空間處理所解釋。然而，對這些資料的另一種解釋方式為，由於字形相似情況下，字形表徵不易使用，因此受試者實際上反而更加仰賴語音線索來幫助記憶。若此假設成立，則我們應該預期語音干擾作業會破壞所謂的字形相似性效果。實驗二採用同音字配對作業及咬音抑制這兩種不同處理歷程中的語音干擾作業，結果發現語音干擾作業中咬音抑制作業僅會大幅削弱語音相似性效果，而這兩種干擾作業皆不影響字形相似性效果。此實驗結果並不支持在字形相似狀況下，受試者更加仰賴語音表徵之假設。 由於行為實驗可做的操弄有限，因此實驗三採用功能性核磁共振造影技術，觀察受試者執行作業之腦部活化情形，以獲得較直接的生理證據。受試者在掃描過程中，分別執行空間短期記憶，以及文字短期記憶作業。其中文字短期記憶作業中隨機出現字形相似與字形不相似兩種題型。實驗三中所觀察到的腦部活化現象，大致呼應文獻中空間與文字短期記憶作業的處理網路。此外，對照字形相似與字形不相似兩種情形，在維持記憶的階段，字形相似組在左側前中央溝(left precentral gyrus, BA 6)、雙側額葉中迴(bilateral middle frontal gyrus, BA 9)、右側頂葉上迴(right superior parietal lobule, BA 7)及左側頂葉下迴(left inferior parietal lobule, BA 39 & 40)會有較高的活化。實驗結果似乎也支持，字形處理機制除了視覺空間處理之外，亦包含了語言知識的處理。 總結本研究，我們觀察到語音表徵與字形表徵同時參與中文短期記憶機制。而中文字形處理機制同時包含了視覺空間及語言知識之處理。語言短期記憶模型，應該同時考量不同語言系統之特性，使用多同音字之語言者的短期記憶處理機制，可能包括語音以外之表徵參與其運作。 |
