本研究主要目的在建立一套中文字形的相似性心理量表，藉以探討中 文字形的分類系統。所選用的中文字包括中等字頻的七大類別，並分成低、中、 高等三種筆劃數分別施測。實驗一利用長寬比例為2：1的倚天標準字形令受試 者將相似的字形分成一組，不相似的字形分在不同組，依此分組法的標準分成 二、三、四、五組。實驗二則是將長寬比例為2：1的倚天標準字及1：1的細明 體兩種字形以受試者內設計呈現。實驗三採用長寬比例為1：1的細明體字形， 與前二實驗不同的字群樣本。實驗四則採長寬比例為1：1的特黑體字形。所有 的實驗結果進行多向度量表(Multidimensional Scaling, MDS)分析。結果顯示，所 有字形主要可分成三軸，依序為上下╱左右、包圍╱開放、以及方╱不方等。不 同長寬比例和不同字群樣本所得的結果並無顯著差異，但不同字體則有些許差 異。此結果無法以特徵量表(feature list;Gibson ,1969)或中文字的永字八法特徵來 解釋，似乎中文字形的心理量表比較可能是採用整體性的，依整個字形的上下╱ 左右、包圍╱開放等特性來決定其相似性。

　　　　　Reading Chinese characters involves the perception of each character's shape,pronunciation, and meanings. The present study focuses on the form perception ofChinese characters. Using a course-to-fine grouping method to classify Chinesecharacters, a similarity scale was established through the analysis of multi-dimensionalscaling (MDS). Two sample groups of Chinese characters of seven categories, threeranges of stroke counts (low, medium and high) were used. The occurrence frequencyof all characters was held constant at a medium level. Font and aspect ratio of thecharacters were varied. The MDS results revealed three main axes for classifying Chinesecharacters: horizontal/vertical, close/open, and square/not-square. Aspect ratios andsample groups had no effect on the similarity judgment, whereas fonts had an effect onthe second and third dimensions. A feature-based analysis failed to explain the results.It seems that the similarity judgment of Chinese characters is more likely to be based onthe holistic structure of each character, rather than on the analysis of its featurecomponents.