By Takanori Tamura, Voice Up Japan advisor.
The purpose of this essay is to show the rise of the trans exclusionary social movement on Twitter in Japan in December 2018 by quantitatively analyzing Twitter posts about “trans woman” and to counter the historical revisionism that argues as if it did not exist.
The transphobic movement, which became prominent on Twitter in December 2018, is a discrimination that afflicts many people(尾崎, 2019) (ゆな, 2020), and the discrimination is an enemy to this entire society. Various arguments have already been published on this matter, such as (Hori, 2020; アジア女性資料センター, 2019; 三橋, 2019; 藤高, 2019) , from the perspective of gender theory.
Leaving the theoretical issues to those discussions, in this essay, I will use text mining technology to analyze the content of Twitter posts. The data itself is the same as what I have already published on my Twitter account, but I have reanalyzed it and added an interpretation.
The reason for this analysis, needless to say, is that this transphobic movement began as a discourse on Twitter. While an analysis of the transphobic movement on Twitter requires an understanding of the Twitter posts they are the premise of the ideological argument, not everyone uses Twitter. Even if they do, it is difficult to share the entirety of the vast text beyond subjectivity.
So, by doing a quantitative analysis of the Twitter posts, I hope to help you see the actual situation surrounding the claim that the discussion is centered around “women’s public baths” and “women’s toilets” (Hori, 2020). I am going to organize the data, so if you are a regular Twitter user who have read various articles on this issue, you may not find anything new. In that case, I hope you will think, “Just as I expected.” I would like to explain the obvious in detail.
Another motivation for the reanalysis is in Yuki Senda’s discussion. Senda said :
“Among the people that I have met who are said to be trans-exclusionary, almost no one had discriminatory belief against transgender people.” (千田, 2020).
First of all, the phrase ” almost no one had discriminatory belief” is a common phrase used by “modern discriminators” , so it is questionable to use it without sharing the content and context of the interview. Because “modern discriminators” as (高 & 雨宮, 2013, p. 68) introduced are less likely to be aware of discrimination and often consider themselves victims, statements such as “I didn’t mean to discriminate” and “this is not discrimination, I am the victim” are often used to counteract the protest against discrimination.
Secondly, as mentioned above, the issue is the statement and the text and not the intention Even when the content and context are indicated, the meaning of a statement is not determined solely by the intent of the speaker. This is a recent philosophical argument (三木, 2019), and it is also in line with our lived experience of frequently clarifying ourselves with phrases such as, “I didn’t mean to do that.” The meaning of a statement is a mutual decision between the speaker and listener. As mentioned above, the “I didn’t mean to” defense is often used by the speaker to deny the meaning that has arisen beyond the speaker’s intent.
Thirdly, the effect of words cannot be judged by the narrator’s intention alone, as words often directly influence another person in addition to narrating facts and opinions(オースティン，Ｊ．Ｌ, 2019; Matsuda, 1993, p. 49,62; 梶原, 2007, pp. 58–63; 鈴木, 2014, pp. 213–214) . If you borrow money from a friend and promise to pay it back by the due date and then say, “I just read the sentence and don’t actually mean it,” you are going to be in trouble. A person living in a verbal destination is not a pebble or a sign as (郭, 2014) described.
Apart from this, I decided to analyze Twitter posts because I think the problem is the discrepancy in quantity understanding between the expression “TERF War” at the beginning of Senda’s essay and the description of “as far as I know”. It is unlikely that huge amount of twitter posts that can be likened to “war” can be understood with “as far as I know” observations. I think she is making the problem smaller. Including “discriminatory belief” issue, argument of (千田, 2020) is revisionist history. This is possible because the amount of the Twitter posts about transphobia is only vaguely grasped. However, as many of you know, there was an amount of text out there that “as far as I know” could not grasp.
Even so, simply saying, “there’s actually a lot of them,” i is not enough explanation to people who do not read Twitter. So, I am going to re-analyze the statistics of Twitter posts containing the word “trans woman” that I have collected previously. And with that, I hope to prove that there was the rise of the trans-exclusionary social movement on Twittersphere. Also, by doing so, I hope to counter historical revisionism we see in (千田, 2020).
2. Description, Overview of Subjects
- The number of Tweets including the word “trans women “spiked in December 2018, and the number of accounts that tweeted it also increased.
- Hypothesis (1) : Around December 20, 2018, transphobic speech and its counter speech probably increased and became a social movement.
Since December 2018, a number of posts including the word “trans women” were made on Twitter. I collected Twitter posts that included “trans women” from March 2018 to January, 2019. The posts were collected from the 1st to the 10th of each month; for December 2018, we collected them twice.
I have divided them into two groups : 1,728 Tweets from March 2018 to November 2018 (hereinafter referred to as for convenience) and 8,588 Tweets from December 2018 to January, 2019 (hereinafter referred to as ).
According to Table 1, the number of Tweets suddenly increased in December 2018, as did the number of accounts that tweeted. Obviously, there has been a big change.
The total number of posts in was 1,728 in 90 days, but in , there were 8,588 tweets in 30 days. The average number of Tweets per day in the term 20_0726_174540_had increased from 19 to 279. It became 491 on January 2019. Accordingly, “the number of unduplicated accounts ” increased.
The number of Tweets per account did not increase as much as the overall Tweet did. You can see that the number of accounts participating has increased, rather than a small number of accounts tweeting in large numbers. Thus, we can see that there was a major development point around December 20, 2018. In January 2019, 1,112 accounts had joined in 10 days and 4,913 tweets were made.
The statistics above cannot entirely be understood by an “as far as I know” observation. We can build a hypothesis (1) that it was at this stage that the transphobic discourse and the discourse against it increased and became a social movement.
Table 1 Change in the number of Tweets containing “trans women,” March 2018 to January 2019
In order to validate the Hypothesis (1) formulated in the previous section, we will conduct a “content analysis” using text mining.
Various methods have been devised to address the question of how the text should be interpreted (フリック, 2002) Of course, there is the usual method of reading and summarizing. Another way to analyze content is to classify the texts into categories and then compare the amount of texts in each of those categories. It is called content analysis.
Content analysis is “a technique used for the scientific study of media messages on the basis of statistical research”（日吉, 2004:5）. Traditionally, the content analysis involves tasks such as assigning numbers to content or classifying it into categories (coding) based on the analyst’s (coder’s) judgment (千葉, 2019:30). The text mining uses a computer for this task. It decomposes sentences into words and examines the frequency of their occurrence. It also finds out the words that appear at the same time (co-occurrences). Based on that, we analyze the text.
Text mining can be used for a variety of purposes but in the social sciences, it is often used to summarize the entirety of a text.
You may be wondering whether looking up the words in the texts will give you a summary. I think so too. To begin with, summarization is generally about interpreting the original sentence to create a new sentence. This process is actually quite difficult, and it is also difficult to explain why we summarized it that way. So, as mentioned above, a variety of methods were tried and tested.
Words are not entire sentences, but we pay attention to words because it is one of those things that is shared between the writer and the reader before the sentence begins, and then shared anew after the sentence has been read. Without this sharing, the writing does not make sense to the reader. Text mining is what explores this assumptions of interpretation of documents (秋庭 & 川端, 2004, pp. 268–271). Another advantage of counting words is that it allows you to share the process of summarization with others and to compare multiple sentences. However, it is a method of interpretation, so even if you use a computer, you cannot get the correct answer. there is no superiority or inferiority between the summaries created by text mining and the ones created by reading with your eyes, because the ideas behind the two methods are different like an apple and an orange.
I used the text mining tool KH Coder 3b01 for analysis. The target of the analysis was narrowed down to nouns which are easy to interpret. KH Coder decompose sentences into words and calculated their frequencies and co-occurrences. We used KH Coder’s dictionary as a criterion for breaking sentences down into words, and we manually added important compound words that were not included in the dictionary, such as “trans woman”.
- Word frequency and a new hypothesis (2)
- As for the hypothesis (1) in Chapter2 : “Around December 20, 2018, transphobic speech and its counter speech probably increased and became a social movement,” we can say that this is supported.
- I built a hypothesis (2) : “Twitter discourse around ‘trans women’ quantitatively expand instead of qualitatively shift after December 2018.”
Table 2 Number of occurrences of the word and the number of Tweets containing the word
I myself was aware that there were discourses on women’s college admissions until November 2018 which then turned into a controversy over toilets and women’s baths. Actually , if we look at the word frequency table like this, we can see that the topic of women’s college was often mentioned in , while it disappeared from the best 30 in .On the other hand, when we look at the frequency table, the words “トイレtoilet”, “加害assault”, “被害damage”, and “恐怖fear” that became a problem in were already present in , and their rankings were not much different from .
Depending on what previous researches have shown (三橋, 2019; 藤高, 2019; 堀, 2019; 小宮, 2020), these “toilets,” “assault,” “harm,” and “fear” also “women’s bath” in may be indicators of whether or not there is trans-exclusionary speech. Therefore, because the number of relevant tweets, accounts and the number of indicator words have been increased around December 20, 2018 , as for the hypothesis (1) in Chapter2 : “Around December 20, 2018, transphobic speech and its counter speech probably increased and became a social movement,” we can say that this is supported.
At the same time, it is likely that Twitter discourse did not qualitatively shift around December, but rather quantitatively expanded. Then I built a hypothesis (2) : “Twitter discourse around ‘trans women’ quantitatively expand instead of qualitatively shift after December 2018.”
- Visualization of word frequency and co-occurrence relationships.
- Through the multidimensional scale method, I found that in both cases, and , the words were broadly divided into a group of words related to trans exclusion and a group of words related to gender theory. This result supports H1 : “Twitter discourse around ‘trans women’ quantitatively expand instead of qualitatively shift after December 2018.”
- The two figures also show, besides the debate on transphobia, the persistence and rise of discussions on gender and feminism. Thus, summing up the whole thing as a “TERF war” (Senda, 2020) is just one side of the story.. It may not represent the whole of the phenomenon.
In order to capture the aforementioned word frequencies and co-occurrence relationships as a whole, I used the multidimensional scale method to create the figure. Here, we consider the high co-occurrence rate of the two words in terms of distance. If the same two words are co-occurring multiple times in a single Twitter post, we consider the distance between the two words to be close. Then, I calculated the distances between the many words to each other. This is a bit like making a map. Kyoto and Osaka are close by, and Tokyo and Kanagawa are also close by. However, Tokyo and Osaka are far away, and Hokkaido and Kyushu are far from all of them. This kind of interrelationship can be represented in a map of Japan. In this way, we create a word map.
This is how I diagrammed the words that appear on Twitter. Frequent words are shown in large circles. Circles close to each other are words that come up together more often than not. Cluster analysis is performed at the same time as the multidimensional scale method. Those that can be divided into groups by cluster analysis have the same color of the circle.
On Figure 1: Analysis of the results on
At the center of Figure 1 is an island of “trans women” and words that co-occurs with it. For the convenience of interpretation, this is set as  and other islands are also numbered from  through .
The word group  is lined with “トランス女性trans women,” “差別discrimination,” and “シスcis”. The word groups  (“女湯women’s bath”, “加害assault”, “犯罪crime”, “暴力violence”, “被害者victimization”) and  (“身体body”, “恐怖fear”, “排除exclusion”) have separate clusters, but are close and related to each other.
The same is true for  (“女子women”, “トイレtoilet”, “大学University”, ” 自認self-identification”) and  (“女子大women’s u”, “入学entrance”, “権利rights”).  seems to be a theoretical discussion of feminism and gender.
Figure 1 Word frequency and co-occurrence using a multidimensional scaling .
Figure2 Analysis of
For your convenience, I have organized the clusters with the symbols from [A] to [G]. The central island [A] is dominated by “トランス女性trans women” and is made up of “差別discrimination”, “排除exclusion” and “シスcis”. To the left of this is the cluster group of [F] [G] [H] [B] . Clusters on “トイレtoilets,” “女湯women’s public baths,” “犯罪crime,” and “加害assault”. To the right of the central island is [C][D][E]. I assume that these are related to theoretical discussions about gender theory and feminism.
Figure 2: Word frequency and co-occurrence by multidimensional scaling construction method .
commonalities and differences
Comparing the two figures for and , we can see that  for women’s colleges in is absent in . At the same time, the rest of the words    and [F] [G] [B] [H], also,  and [C] [D] [E] are made up of similar or related words. Broadly speaking, in , words on trans exclusion, such as  , and words on gender theory, such as , are divided into two groups. In , a similar construction is shown in the form of increased word type and quantity.
The result of this shows:
- We can support hypothesis H1 : “Twitter discourse around ‘trans women’ quantitatively expand instead of qualitatively shift after December 2018.”
- The two figures also show, besides the debate on transphobia, the persistence and rise of discussions on gender and feminism. Thus, summing up the whole thing as a “TERF war” (Senda, 2020) is just one side of the story.. It may not represent the whole of the phenomenon.
- Classification of articles by coding rules
- In order to understand the transitions described in the previous section in terms of change in quantity of words, we categorized words according to coding rules and charted the difference in their proportions. Compared to , is similar in some respects, but there was a noticeable increase in the *犯罪crime category* and *恐怖fear category* in the same composition.
Here, based on previous discussions, we used words as an indicator to form a coding rule, which is an article classification criterion, and applied the same coding rule to and for comparison. In other words, it is an attempt to determine the scale, measure the two texts, and compare them. The words for the coding rules were mainly taken from clusters of (Figure 3). Hereafter, an asterisk (*) indicates that it is a category name.
The way the coding rule works is that, for example, Tweets containing the words “心mind”, “体body”, or “気持ちfeelings” are tagged as *心と体mind and body. Of course, there are many words in a Tweet, so it is possibe for a tweet to have multiple tags. “＊当事者 Tojisha” in Japanese is a concept that cannot be translated into English, so I use it as it is. 当事者 Tojish is translated into English as person concerned, person who has the experience, positionality, persons at the center, people who are affected individual autonomism etc. Tojisha means all of them and more. For the intricacies of the situation, see (McLelland, 2009).
Figure 3 Coding rules.
According to the coding rules mentioned above, we categorized and and calculated the percentage of the occurrence of each code.
Table 3 Classification results by coding rules
Then, I expressed this table in a radar chart, except for the ones I could not categorize.
Figure 4 Comparison by percentage of occurrence for each category in and .
Three of the eight items, * 自認gender identity, *心と体mind and body, and *ジエンダー gender, look similar. However, in , there was an increase in *犯罪crime (+14.20%), *恐怖 fear (+5.00%), and * 女装cross-dressing (+3.94%) compared to . In other words, we can see that the types of categories of two data are somewhat similar for all except for *女子大women’s university>, and the number of Twitter posts increased significantly from to , and the number of Twitter posts on crime (+14.20%) and fear (+5.00%) increased significantly.
As I mentioned at the in the introduction, the purpose of this essay is to illuminate the quantitative aspects of the debate about “trans women” in Twitter. Repeating the results of our analysis, we found the following.
Hypothesis (1) : “Around December 20, 2018, transphobic speech and its counter speech probably increased and became a social movement” is supported.
From analysis of word frequency, I established a hypothesis (2): Twitter discourse around “trans women” qualitatively shift after December 2018, but did something similar quantitatively expand.
We found the following from visualizing the frequency and co-occurrence relationships of words. The words were broadly divided into two groups, a group of words related to trans exclusion and a group of words related to gender theory. This result support the hypothesis (2). It also shows that summing up the whole thing as a “TERF war” is just one side of the story. (Senda, 2020). It may not represent the whole of the phenomenon.
From categorizing according to coding rules, we found that was similar to in some respects, but it had a noticeable increase in the *犯罪Crime category* and *恐怖Fear category* in the same composition. These categories contain words as indicators of trans-exclusionary speech, as mentioned above. It means that there was the rise of the trans-exclusionary social movement.
To sum up, I found that the discourse on Twitter around “trans women” did not change significantly in content after December 2018, but rather increased the type and quantity of words and a significant increase in participation in a similar construct. In Twitter posts, especially since December 2018, there has been an increase in discourse about crime and fear, which has caused a lot of people to suffer.
That could be a mundane conclusion for heavy Twitter users. However, as I mentioned at the beginning, I think that we have proved that the quantity of Twitter posts cannot be grasped by observation in the “as far as I know” range. I also proved that there was exactly the rise of the trans-exclusionary social movement on Japanese Twittersphere in December, 2018.
I merely put numbers, but I think it may have helped to counter the “there’s actually no such thing as trans exclusion” revisionism, for example, we see in (千田, 2020).
Azumi Tamura, who threw herself into the social movement after 3.11 earthquake examined the movement wrote the below.
On the streets after 3.11, participants in the activities had to acknowledge that they were imperfect and forgetful, but still attempted to seek ethical behavior. Their ethic is quite simple. It is to be ‘open’ to others who compel you to remember. (田村, 2020, p. 137)
I am also imperfect and forgetful, but I wanted to leave a clue for myself to be “open to others”.
I would like to thank the editor-in-chief Johann Fleuri (Director, Voice Up Japan) for allowing me to publish it and Mai Miura (Assistant Chief-editor, Voice Up Japan Media Team) for revising the English text and made pertinent points. The poor quality of the English text is entirely my fault, however.
秋庭裕, & 川端亮. (2004). 霊能のリアリティへ: 社会学、真如苑に入る. 新曜社. http://ci.nii.ac.jp/ncid/BA67723926
アジア女性資料センター. (2019). 女たちの21世紀: 特集:フェミニズムとトランス排除. 夜光社.
オースティン，Ｊ．Ｌ．. (2019). 言語と行為 いかにして言葉でものごとを行うか. 講談社
千葉涼. (2019). 内容分析研究の現状と今後の展望. マス・コミュニケーション研究, 95, 27–40. https://doi.org/10.24460/mscom.95.0_27
藤高和輝. (2019, 8月 15). 後回しにされる「差別」：トランスジェンダーを加害者扱いする「想像的逆転」に抗して. https://wan.or.jp/article/show/8527
フリック・ウヴェ. (2002). 『質的研究入門―「人間の科学」のための方法論』. 春秋社.
フランク, アーサー. (2002). 傷ついた物語の語り手: 身体・病い・倫理 (鈴木智之, 翻訳者). ゆみる出版.
Hori, A. (2020). Excluding Whom for What? A Look at Transphobia in the Japanese Twittersphere. https://wan.or.jp/article/show/8644
日吉昭彦. (2004). 内容分析研究の展開. マス・コミュニケーション研究, 64, 5–24.
堀あきこ. (2019). 分断された性差別― 「フェミニスト」によるトランス排除. 女たちの21世紀 = Women’s Asia 21, 98, 6–10.
郭基煥. (2014). 在日コリアンに対するヘイトスピーチとイデオロギーへの呼びかけ: ジュディス・バトラーによる「主体化」論を手引きに. 現代社会学理論研究, 8, 39–54.
梶原健佑. (2007). ヘイト・スピーチと「表現」の境界. 九大法学, 94, 49–115.
小宮友根. (2020). フェミニズムの中のトランス排除. 早稲田文学 ２０１９年冬号 〈シリーズ特集第１回〉ポストフェミニズムからはじめる
Matsuda, M. J. (1993). Words That Wound: Critical Race Theory, Assaultive Speech, And The First Amendment (1版). Routledge.
McLelland, M. (2009). The Role of the ‘tōjisha’ in Current Debates about Sexual Minority Rights in Japan. Faculty of Arts – Papers, 29. https://doi.org/10.1080/10371390903026933
三橋順子. (2019). 日本のフェミニズムが問われるもの――トランスフォビアの克服とトランス女性との連帯. 女たちの21世紀 = Women’s Asia 21, 98, 17–21.
尾崎日菜子. (2019). エイリアンの着ぐるみ (特集 フェミニズムとトランス排除). 女たちの21世紀 = Women’s Asia 21, 98, 11–16.
リクールポール. (2004). 時間と物語〈1〉物語と時間性の循環/歴史と物語 (新装版). 新曜社.
鈴木紫野. (2014). いかにしてヘイトスピーチに立ち向かうべきか: 内野正幸『差別的表現』を読む. Quadrante : Areas, cultures and positions = 四分儀 : 地域・文化・位置のための総合雑誌 : クァドランテ, 16, 237–246.
Tamura, A. (2018). Post-Fukushima Activism: Politics and Knowledge in the Age of Precarity (1版). Routledge.
高史明, & 雨宮有里. (2013). 在日コリアンに対する古典的/現代的レイシズムについての基礎的検討. 社会心理学研究, 28(2), 67–76.
盛山和夫, 金明秀, 佐藤哲彦, 難波功士, 浅野智彦, 村田泰子, 筒井淳也, 渡邊勉, 長松奈美江, 新雅史, 本郷正武, 山北輝裕, 赤枝香奈子, 進藤雄三, 立石裕二, 関嘉寛, 白波瀬達也, 高原基彰, & 鈴木謙介. (2017). 社会学入門. ミネルヴァ書房. https://ci.nii.ac.jp/ncid/BB23494749
千田有紀. (2020). 「女」の境界線を引きなおす: 「ターフ」をめぐる対立を超えて (総特集 フェミニズムの現在). 現代思想, 48(4), 246–256.
鈴木紫野. (2014). いかにしてヘイトスピーチに立ち向かうべきか: 内野正幸『差別的表現』を読む. Quadrante: Areas, cultures and positions = 四分儀: 地域・文化・位置のための総合雑誌: クァドランテ, 16, 237–246.
田村あずみ. (2020). 不安の時代の抵抗論:災厄後の社会を生きる想像力. 花伝社.
ゆな. (2020, 2月 20). 千田有紀「『女』の境界線を引きなおす：『ターフ』をめぐる対立を超えて」（『現代思想３月臨時増刊号 総特集フェミニズムの現在』）を読んで. ゆなの視点. https://snartasa.hatenablog.com/entry/2020/02/20/034820
- It makes sense that in a collection of essays, this work is a requiem in the form of a voice letter, a narrative, while describing facts and analysis. This is not only because the author is a novelist, but also because there is something else to be sensed, besides understanding.
It is very difficult to understand others who do not share experiences. In such cases, participation in the social situation that needs to be corrected is often chosen in terms of social justice. This is because the elimination of discrimination and equality of rights are universal. At the same time, we already live in the same society, even if we do not understand them. So, we may not understand them, but we may be able to feel them. Stories have that kind of power to cure others as (フランク, 2002, p. 4) claims and move others’ emotions to have them to go forward（田村, 2020:134）.
2. Discrimination is “the disproportionate distribution of social resources, not just based on ethnicity, but is based on identity groups that are considered to share a type of difference, such as religion, gender, sexuality, and mental or physical disability,” and “such a bias often arouses demands for correction because it is contrary to the principles of modern civil society, which is based on freedom and equality” (Moriyama et al., 2017).
Our society is based on the principles of modern civil society. Therefore, discrimination against the principles of modern civil society is a denial and destruction of modern civil society itself. In other words, it is the enemy to the whole society.
3. McConahay’s argument, as introduced by Fumiaki Taka, is as follows.
McConahay says that racism against blacks in America underwent a transformation in the mid-twentieth century. According to his classification, the prejudice that was once mainstream is called classical racism. This was an open prejudice based on the belief that blacks were inferior. However, as the idea that discrimination is bad spread, what has been named modern racism emerged. Modern racism often allows for racial equality, however, according to McConahay, the idea of modern discrimination is as follows.
- Prejudice and discrimination do not already exist and (2) the economic disparities that currently exist are not due to inequality, but to a lack of effort on the part of blacks but (3) blacks are seeking too much government favoritism and (4) receiving an unfair economic benefit.
Modern racism is a more subtle form of prejudice than classical racism, but it is associated with discriminatory behavior. However, modern racism is difficult for the person holding it to realize that it is prejudice, even for the person holding it. This kind of thinking will often lead the discriminator to believe that they are the victim.
This is an example of racial discrimination in the United States, but it can also be applied to discriminatory comments in Japan. Rather, it is better to say that modern racists widely use this rhetoric.
4. There was a debate in the United States on the question of why hate speech is just a word and yet it hurts people. One of the theories adopted at that time was the theory of verbal action. The philosopher, Austin believed that the function of words could be an act, other than something that simply states facts or writes opinions. A typical one is a “promise”. A promise is just an exchange of words, but making a promise is an act. Promises have strong influence on the people who make them.
5. Relying on Judith Butler’s theory of “subjectification,” (郭, 2014) argues that hate speech against Koreans in Japan is as follows.
Their subjectivity as a Japanese is subordinated to the nation Japan as the Subject with a capital letter. That is, “in the process of subjectification/reflective proof in which they aim to prove the legitimacy of their subjectivity, the term ‘Koreans’ loses its context, turning into a mere symbolic target of hatred. (郭, 2014:50).” Applying this theory to trans exclusion requires a different discussion, but the living individual, whoever he or she is, is not a signifier for the subject formation of others.
6. “The number of unduplicated accounts ” is a count of how many accounts actually tweeted, even if there were a lot of tweets.
7. This is a morphological analysis, but in this paper, it can be understood as a “word”.
8. This is an important book that introduces the use of text mining in earnest to Japanese sociology. The authors used text mining to analyze the life histories of new religious believers applying “threefold mimesis cycle” of Ricoeur (リクール, 2004). important groups of words found in text mining were positioned as mimesis1 (mimesis 1, préfiguration) in threefold mimesis. With this theory, the authors analyzed the life histories and argued that the denominational terminology in the remarks increased in depth as the progression became more proficient (mimesis 3, refiguration).
10. There are no [ ] symbols in the diagram, but they are added for description purposes