《日臺大辭典》（1907）是日人小川尚義編纂的台灣閩南語辭典之一，也是對漢語方言進行全面比較的先驅著作，其〈日臺大辭典緒言〉附於全書之前，乃具有重要參考價值的方言材料。本文共分五節，第一節前言；第二節介紹〈緒言〉之章節架構與內容；第三節討論〈緒言〉中古音的依據，亦即有關《韻鏡》的問題；第四節討論〈緒言〉中標音符號一致性；第五節結語。經由資料比較，本文發現〈緒言〉中的語料有些應用上值得注意之處：第一，小川所謂用以檢視古今音變的《韻鏡》，極可能是1815年太田全齋改編《韻鏡》而成的《漢吳音圖》，故不能以《韻鏡》或《廣韻》來直接檢討〈緒言〉的用例，必須先對《漢吳音圖》的內容與音系有所了解，才能正確理解小川所錄反切。第二，小川雖指出〈緒言〉中的羅馬標音按照一般羅馬字發音，然其中含有四種標音系統，容易產生誤讀。今人若欲對〈緒言〉內所有的方音對照表進行語言分析，宜先確定每個符號於各個標音系統中的內涵，且須對小川所引諸書皆有一定程度的掌握，才能對〈緒言〉有更深入的理解。

Ritai Dacidian (A Japanese-Taiwanese Dictionary, 1907), the pioneer book of Chinese dialect contrast, is one of Naoyoshi Ogawa's Taiwanese dictionaries. The ＂Preface＂（緒言） to the dictionary is not only an introduction, but research, because it includes abundant data about Chinese dialects, Taiwanese, and Japanese around the 19th century. In this article, the two main points within five sections are 1) discussing the question about whether the mid-ancient Chinese data that Ogawa has used came from ＂Yunjing＂（韻鏡）, and 2) discussing whether the four Romanization methods of Chinese dialects resulted in confusion. By observation, the main resource of mid-ancient Chinese sounds, declared by Ogawa to be the ＂Yunjing＂, might properly be known as ＂Hanwu Yintu＂（漢吳音圖，1815）, because Ogawa used many charts and Chinese characters that sounded the same as those in ＂Hanwu Yintu＂ instead of ＂Yunjing＂. In the other part of this article, it is also found that some contradictory situations in the ＂Preface＂ came from the mixture of four Romanization systems of Chinese dialects.