Voyant Tools is a popular website that offers a series of text analysis tools for examining word frequencies, patterns, and trends of a document. Although Voyant Tools supports analyzed documents in multiple languages, current tutorials and articles mainly focus on the application of Voyant Tools to English (Rockwell and Sinclair 2017). In this project, I will be testing the application of Voyant Tools to Chinese language sources by analyzing Lienü zhuan 列女傳 (Biographies of Exemplary Women), a section that features in many Chinese official histories to document women’s words and deeds. Among the twenty-five official Chinese histories, thirteen contain Lienü zhuan 列女傳. These are the Book of the Later Han 後漢書, the Book of Wei 魏書, the Book of Jin 晉書, the Book of Sui 隋書, the History of the Northern Dynasties 北史, the Old Book of Tang 舊唐書, the New Book of Tang 新唐書, the History of Liao 遼史, the History of Jin 金史, the History of Song 宋史, the History of Yuan 元史, the History of Ming 明史, and the Draft History of Qing 清史稿, written by different dynasties between 445 and 1928. Specifically, I will focus on word frequencies, trends, and collocates of particular terms that describe women’s textile making activities in these histories. The objective of this project is to examine the role of female textile making over time in the narratives that were constructed by history writers.
Reliable digitized Lienü zhuan from the thirteen dynasties can be obtained from an established online library, the Chinese Text Project. This library allows its content to be downloaded as plain text; however, the downloaded text requires a preprocessing step for data cleaning before it can be imported into Voyant Tools. This preprocessing step is crucial to the success of analyzing Chinese language-based text as it removes noise, such as punctuation, from the text for better analysis results. For guidance on how to remove punctuation from Chinese text, please visit my previous piece Data Cleaning Chinese Text with Open Refine for details. After data cleaning, thirteen html documents are created ready for analysis.
The thirteen documents can be imported as a bundle into Voyant Tools. The default setting of Voyant Tools allows five digital tools to present the basic analysis results in five panels. These are Cirrus, Reader, Trends, Summary, and Contexts (as shown in Fig. 1). Users may select other tools, such as Bubblelines, Knots, and Terms Berry to replace any current tools for different types of analysis and visualization. The basic principle of many of these tools is based on two concepts – the frequencies of terms in the text and the relationship between different terms in the text. In this project, I will examine three textual analysis tools that are built on these two concepts – Trends, Collocates, and Terms Berry.
Fig. 1. Default text analysis results shown in five panels.
Word frequency is a basic concept in textual analysis. It is the occurrence of a particular term in the text and how many times said term appears. Many textual analysis tools are based on this concept, such as Trends in Voyant Tools. Trends uses a line graph to describe the frequencies of given terms in different segments of a document or in different documents (if multiple documents are uploaded). Thus, Trends is useful for evaluating the significance of a term since high frequency often suggests that a term is important. In this project, I will study four terms that describe textile making activities. They are weaving 織, sewing 紉, spinning 紡, and needlework 針黹.
Spinning 紡 and weaving 織 are essential fabric producing techniques. Spinning is the activity that twists natural fiber into threads while weaving forms threads to fabrics. Together, the two activities constitute crucial steps in turning raw materials into cloth. On the contrary, sewing 紉 and needlework 針黹 are techniques dealing with cloth that is already made. According to the nature of these four activities, I will investigate them in two groups. Fig. 2 presents the frequencies of weaving and spinning in the thirteen histories chronologically. Both terms appear persistently across most documents while there is a sharp plunge in the usage of both terms in the History of Liao 遼史 and the History of Jin 金史. A possible explanation for this drop could be the short length of both histories, which was partially due to the insufficient sources at the time the histories were compiled. Excluding these two histories from consideration, we see that the overall trends of for spinning and weaving is a steady, but decreasing presence from early to late period of imperial China. On the contrary, sewing and needlework only made their appearance in the Ming and Qing histories, as shown in Fig. 3.
Fig. 2. Trends of spinning 紡 and weaving 織.
Fig. 3. Trends of sewing 紉 and needlework 針黹.
The trends relating to terminology about the four textile making activities demonstrate that in the official Chinese histories, women’s textile work was limited to fabric producing for a long period, and that in late imperial China, women’s involvement in the textile industry extended to a variety of textile making activities including spinning, weaving, sewing, and embroidering. Francesca Bray argues in herTechnology and Gender: Fabrics of Power in late Imperial China that because of the development of technology and the specialization in textile industry in late imperial China, women’s dominance in fabric producing activities was challenged by men (Bray 1997, 206 -236). The Trends analysis demonstrated above – that women started to pick up sewing and embroidering in Ming and Qing times – was likely the result of women regaining their textile making skills after men took over weaving jobs. On the other hand, the Trends anaylsis also suggests that women in late imperials China were not entirely pushed out of textile producing. Fig. 4 shows the the two groups together and reveals the continued high frequencies of spinning and weaving in Ming and Qing times. Although women’s biographies in official histories were highly selective and are not identical to real life, they nonetheless reflect the criteria chosen by history writers when it comes to evaluating the appropriate textile jobs that embody an exemplary female.
Fig. 4. Trends of spinngin 紡, weaving 織, sewing 紉, and needlework 針黹.
In today’s piece, I have examined the frequencies of four terms that describe female textile making activities in official Chinese histories. The Trends function in Voyant Tools reveals the emergence of sewing and embroidering in the Ming and Qing periods, whereas spinning and weaving appear almost constantly throughout the history of imperial China. My next contribution will use Collocates and Terms Berry to investigate the relationships between various terms for a comprehensive understanding of female textile making roles in official Chinese histories.
(This post is a revision ofThe Changing Role of Textile Making: Text Analysis of Digitized “Lienü Zhuan” which I presented at the 2021 AAS Conference on March 21-26, 2021.)
Bray, Francesca. Technology and Gender: Fabrics of Power in Late Imperial China (Berkeley, CA: University of California Press, 1997).
Rockwell, Geoffrey, and Stéfan Sinclair. Hermeneutica: Computer-Assisted Interpretation in the Humanities (Cambridge, Massachusetts: The MIT Press, 2017).