大學生物、化學和電腦科學原文教科書特定篇章之質性研究

A Qualitative Study on Texts Taken from English Textbooks of Biology, Chemistry and Computer Science

一篇文章可讀性的高低可以判定其難度，可讀性越低，文章就越難。傳統計算 可讀性的公式多是根據其字彙和句子的長度而定的。但篇章分析專家卻抨撃這些公 式無法解釋與文本有關的其他因素，例如句法的複雜度、語意密度和修辭結構。因 此本研究企圖從篇章分析切入探討其他會影響可讀性的因素。採用的研究方法如下: 將生物、化學和電腦學門共十五本原文教科書輸入電腦，用Coh-Metrix L2 Reading Index的公式計算文本中每一篇章的可讀性指數，根據這些指數挑出最易和最難的 篇章，各選其中一部分，採用Lautamatti(1987)以及Sympson(2000)的理論架構， 進行質性分析，檢視(一)篇章的主題是如何發展的？（二)比較兩篇的訊息結構和訊 息流動情形。（三)分析兩篇如何運用句法結構來安置各種訊息。（四)檢視這三個學門 裡最難的篇章彼此間有無顯著差異？ 結果發現(一)易讀篇章的主題發展傾向選用較單一且低層次的延續方式，而不 是複合式高層次的方式。反觀難讀篇章的主題結構明顯較複雜、主題轉換較頻繁、 銜接方式也較多樣性。（二)易讀篇章的訊息流動大致符合「從舊到新」的原則。而 難讀篇章則顯現「從半舊半新到新」的趨勢。（三)從訊息和句法結構如何互動的角 度而言，易讀篇章的訊息傳遞較直接，句法結構單純，篇章的資訊大致呈線性排列， 橫向的附帶說明較少。反觀難讀篇章的主要子句總數遠少於從屬子句，訊息結構錯 綜複雜，為了要達成連貫性，作者必須努力的把重點安插在主要子句中，其它較不 重要的補充資料、以及為文章布局而埋伏的種種訊息則安插在廣義的從屬子句中。 (四）就表達多樣性來看，生物最接近文學，有豐富的表達力；化學居中；電腦則接 近純科學，明確清晰卻不具表達力。

Readability is a powerful predictor of how comprehensible a text is. The traditional readability index of a text is based on its word and sentence lengths， but discourse analysts such as Davison and Kantor (1982) and Carrell (1987) criticized this practice， for it does not account for text-based factors such as syntactic complexity， rhetorical organization and propositional density. As a result， this study aims to explore readability in the light of discourse analysis. The study method is as follows. First， all the units of the 15 textbooks of biology， chemistry and computer science are keyed in to measure each unifs readability via Coh-Metrix L2 Reading Index. Based on these indices， the most and the least readable units are sorted out for analysis via the approach of Lautamatti (1987) and Sympson (2000). The purpose is (1) to examine each texfs topical structure; (2) to compare its thematic structure and information flow; (3) to analyze how each text uses syntax to ground various types of information; (4) to compare the 3 least readable texts. Important findings are as follows: (1) The least readable texts show more complicated topical structures and more diverse topical progressions， but the opposite is true of the most readable texts. (2) The overall information flow of the least readable texts follows the principle of “from accessible to new” rather than “from old to new.”（3) In terms of discourse grounding， the least readable texts use more subordinates than main clauses， but the reverse is true of the most readable texts. The information structure of the former is often so complicated that only a sophisticated writer with solid training in grounding can express it coherently， embedding important backbone information in main clauses and less important background information in subordinates. (4) The text of biology is found to be the most expressive， while the text of computer science is the most precise but the least expressive.