教育實務情境常涉及評量或評鑑的決策過程，舉凡教材決定、教師甄試與教學評量等皆是。決策或判斷行為常是多準則(multi-criteria)或多屬性(multi-attribute)的特性，而評量或評鑑活動訊息的獲得，可分為兩部份。一為較明確客觀的數據，例如傳統的紙筆測驗分數、師生人數、修課分數等客觀標準依據；另一則為較不明確且含有主觀成份的判斷，例如學生的參與活動態度、教學示範的表現等。 對於不明確且含有主觀成份的判斷行為而言，是將個人潛在的(latent)心理感受或特質，轉化成外顯的(manifest)的決策，此決策可用語意變數(linguistic variable)表示出來，且為利於計量的分析，此語意變數以數值表示。傳統上常用李克特式量尺(Likert scale)或語意差異量尺(semantic differential scale)等量化方式，來衡量欲研究對象的心理認知，並將語意變數轉化成明確值(crisp value)之等距變數(interval variable)。此種資料分析方式，有下列兩項值得商榷： (一)人類思考具模糊性與不確定，而此傳統上將語意變數轉換成等距變項數值，過於簡化且欠缺理由依據。 (二)本質上思維是一種多元邏輯的過程，將語意變數轉換明確值，對於質化且主觀判斷、具不確定且模糊性的語意變數而言，實有不妥。 因此，本研究試圖提出基於模糊語意變數(fuzzy linguistic variable)的多準則評量模式，期能對教學評量或課程教材評估提供另類的思考方向。本文可分為二部分，第一部分是提出模糊語意變數多準則的評量模式，此模式考慮評分的可信水準以及評分者的嚴苛度等因素；在第二部份中，以實際資料為例，說明此一模式的資料分析過程。在實例資料中，評量準則和權重可經由群體決策獲得，且可得知在不同可信水準、評分嚴苛度下的評分結果。此外，從實例中亦可知此評分模式有較佳的鑑別度。最後，作者對於教育實務應用和未來研究上，提出心得和建議。

　　　　　In the education environment, we often take decisions according to some criteria. For example, learning assessment and teaching evaluation are very important for teachers. We always make decisions according to some attributes or criteria. The results of the decision for one criterion come from two parts of information. One is a kind of clear and objective information, such as paper test scores, the number of teachers and students, the credit hours, etc. The other is a kind of uncertain and subjective information, such as the attitude of students' participation, the performance of teaching, etc. As for the uncertain and subjective information, it is a decision which is transformed from the latent trait to the manifest one. And the decision could be expressed by linguistic variables. For the convenience of calculating, we often record linguistic variables by real numbers. Traditionally, most researchers use "Likert Scale" or "Semantic Differential Scale" to measure the psychological cognition of the research participants, and record linguistic variables by interval real numbers. As for this kind of recording, there are two things worthy of discussing. They are as follows: 1.Human thinking is fuzzy and uncertain. We always transform linguistic terms into linguistic variables, and transform these linguistic variables into crisp values. This way is too simplified and lack of reasons. 2.The process of thinking is multi-logic in nature. It isn't appropriate to transform linguistic variables into crisp values. The reason is that the linguistic variable is subjective, uncertain and fuzzy. And it is appropriate to express the linguistic variable by fuzzy number. In view of the research findings, the researcher wants to investigate "The Multi-Criteria Assessment Model of Fuzzy Linguistic Variable". And this model will give another way of research for education assessment and educational evaluation. There are two parts in this research. One is the model which is developed by the researcher, the other is the explanation of real data analysis. In the real data analysis, we can realize that the items and weights of criteria can be developed by group decision-making. Besides the above finding, the score is meaningful under different confidence level and scoring strictness. Finally, based upon the findings of this study, some recommendations for further research are suggested.