製作新文類：晚清報刊的社會調查

The Production of New Literary Categories: Social Surveys within Late Qing Dynasty Periodicals

本文梳理社會調查文類於晚清報刊出現的歷史，探討「調查」如何作為「新文類」出現於留日學生的同人刊物上，為興利除弊，追求準確、實用性的工具。此文類的出現，不僅表述為一種新的知識形式、溝通方式、也是社會行動，包括訂定規範、組織調查會、設置專欄，以實證、表列、數字的文本型態，廣邀民眾參與、發布。
《浙江潮》、《江蘇》、《雲南》等三份刊物發行於1903-1907年，是較早發布調查公約，並具有規模的期刊。作為調查報導的濫觴，它們為調查建立較明確的定義，卻屢在實際收稿與編輯時遭遇困難，表現在專欄界定上的游移不定，與新聞、訪函、雜文的區別模糊，也正是在此摸索過程，研究者可見「新文類」碰撞、試錯的具體社會實踐。
本文借助數位研究工具，運用共現詞網絡的機器分群，協助找出調查報導主題，並輔以人工詞彙分類，解讀文本類型內涵。此外，我們透過分詞技術，找出日語借詞（新詞），進一步解讀新詞在新文類上的應用情形。在單篇文章平均超過三成的使用基礎上，我們可推論，新詞在新文類的製作過程中，功不可沒。此外，我們在新養分內尋找到舊資源，包括表列、繪圖及敘事方式，都有傳統地方志的痕跡，由此，新舊嫁接的調查文類才得以成形，並逐漸為人所接受、流傳。

The present paper investigates the history and contexts surrounding the emergence of social surveys within periodicals of the late Qing dynasty, primarily discussing how surveys as a“new literary category”operated as a device from which to pursue accuracy and practicality while dismantling what was considered harmful or backwards in the publications of students abroad in Japan. The development of this genre not only expressed new forms of knowledge and communication but was also a social act, namely prescribing norms, organizing societies revolving around surveys, creating specialized columns within publications, and inviting public participation.
Zhejiang chao《浙江潮》, Jiangsu《江蘇》, and Yunnan《雲南》, all issued around 1903-1907, were the earliest periodicals to publish survey conventions of any kind. Being the forebearers of survey reporting, these publications attempted to establish clear definitions of the new category. The receiving and editing of related articles, however, were rife with difficulties, and the distinctions between survey writings, news reports, letters, and opinion pieces were ambiguous, demonstrating the shifting boundaries of columns. But it is precisely within this process that we can observe the collisions of new literary categories and the trial and error behind the establishment of concrete social practices.
This paper employs digital research tools to identify shared themes across survey reports by using machine clustering of co-occurrence networks, themes which are supplemented by lexical classifications performed manually to decipher the connotations of textual forms. We have also discovered that“new”and“old”are far from diametric in these processes. The tables, figures, and narrative styles, all of which contain traces of traditional local gazetteers, used by social surveys grafted new and old together to form a new category of survey, one which was gradually accepted and circulated.