4. Language Patterns and Characterization
5.1 Overview
5.1.1 Summary of Chapters 1 to 4
The major goal of this thesis was to expand the general understanding of language in manga, which has hitherto been looked at as primarily a visual media. To do so, I constructed a corpus of popular manga, with an equal distribution of shoujo-manga and shounen-manga. As I outlined in Chapter 1, the corpus was structured so as to include all of the text seen in manga, organizing them into 8 different categories: Lines, Thoughts, Narration, Onomatopoeia, Background Text, Background Lines/Thoughts, Comments and Titles. As I argued, the unique structure of the corpus thus allows one to utilize it for any numerous types of queries, and is not limited to its ability to contribute to manga research alone.
I then began the data analysis in Chapter 2, where I examined how language in manga is characterized structurally. By looking at the distribution of the categories above, I showed that while Lines form the core of text in all of the series included in the corpus, individual series were distinguished by their use of the other, secondary categories. The data also showed that there was no significant differences between shoujo-manga and shounen-manga in regards to the number of Thoughts seen, thus not supporting arguments by Natsume (1997) and others that shoujo-manga are more “psychologically-oriented” than shounen-manga, which should result in more monologue-like, thought-like texts. Rather, I agreed with Ôtsuka’s analysis that shoujo-manga may no longer be described so clearly in this way, and instead focused on two categories, Background Lines/Thoughts and Comments, which proved to be more common in shoujo-manga than in shounen-manga.
I then took a step back to consider why these two text categories, both distinguished by their visual presentation, needed to be differentiated from Lines, Thoughts and Narration, with which they have very similar narrative roles. While suggesting that these texts are more prominent in shoujo-manga because of the emergence of a special authorial presence in shoujo-manga during the late 1960s and 1970s, I argued that these texts change the reading experience by visually encoding data for readers into more and less essential reading, allowing them to choose between different reading experiences. As handwritten text that speaks more strongly of a writer than type does, I also suggested that authorial comments are especially able to create a sense of community.
With these points in mind, Chapter 3 focused on the content of language in manga by looking at the orthographic styles and lexical characteristics of manga. In particular, I conducted three major studies, on (1) the orthographic characteristics of manga and irregular furigana, (2) the readability of manga and (3) a morpheme
164
analysis of manga’s lexical characteristics. In the first study, I looked at the distribution of the different scripts, demonstrating that one of the characteristics of manga is their use of symbols. While this differentiates manga from other genres, it also has certain similarities with other recent media, so that manga’s orthographic patterns helps create and locate a manga style of writing. In the second study, I used tools developed by Sato, Matsuyoshi
& Kondoh (2008) and the Readability Research Laboratory (2010) to examine how easy manga are to read.
While they graded fairly low, at a 5 to 6 grade level on average, they are only a few grade levels lower than Nishimura’s (2011) results for keitai-shousetsu and conventional novels.
I then followed these studies with a morpheme analysis of Lines. After reviewing the issues involved in conducting morpheme analyses in Japanese, I used the results from the previous section as a guideline for standardizing the text to make it more appropriate for analysis. I then summarized the data by the distribution of parts of speech seen; the usage of different words by their origin (Sino-Japanese kango, native Japanese wago, borrowed foreign words, etc.); and frequency charts for the top-10 nouns, verbs, adjectives and adverbs. The results will point to additional commonalities with spoken speech in the distribution of wago and interjections.
I also pointed out an interesting tendency for shounen-manga to use words related to their narrative themes, as opposed to shoujo-manga, whose frequency lists for nouns pointed instead towards the importance of interpersonal relationships.
Chapter 4 continued to move the focus one layer deeper by looking at specific language patterns seen and how they are used to create patterns; it particularly built up on the last chapter by utilizing the data produced by the morpheme analysis. Japanese language has traditionally been described as being gendered, but in recent years it is said to be going through a process of neutralization. Manga are often said to be influential in such changes, but reports differ on the speech patterns of characters in manga, with some describing highly stereotypical yakuwari-go patterns (Kinsui (2007), Takahashi (2009)), and others describing patterns similar to realistic speech (Ueno (2006), Aizawa (2003)). In response, I conducted three studies related to gender and characterization in manga: first, the distribution of speech between male and female characters, second, the use of personal pronouns, and third, the use of gendered sentence final expressions. While personal pronouns showed relatively conservative gendered patterns, sentence final expressions showed a major gap between shoujo-manga and shounen-manga. More specifically, female characters used many more strongly feminine expressions in shounen-manga. As became clear from the first study, shounen-manga is deeply skewed towards male characters, with very few central female characters, and in addition to more standard reasons such as author interference, I also argued that the peripherality of female characters in shounen-manga made them more susceptible to stereotyped speech patterns.
165
5.1.2 Genre, manga and language
Aside from the more general goal of establishing the characteristics of language in manga, another major goal of this thesis was to consider how language could be used to reconsider genre in manga. Several points within each of the studies suggested that linguistic characteristics were one effective way of looking at genre.
Summarizing the results, one could say that in general, shoujo-manga were typified by the diversity of the text types it used; its use of vocabulary suggesting interpersonal relations; their focus on a core group of characters, with balance between male and female characters; and more realistic speech patterns. On the other hand, shounen-manga were typified by using more kanji and being correspondingly more difficult to read; more diversity in the vocabulary used, with more technical or topical vocabulary; featuring more characters, with less balance between the genders; and stereotypical speech patterns which could be considered yakuwari-go, particularly for female characters.
While I argued that many of these issues could be explained by differences in plot, it is also possible to state these differences in a different fashion. Specifically, whereas shoujo-manga appears to use structural methods to their advantage—here, typified by their diverse use of text types, but also typical of their special use of frames (Takeuchi O. , 2005)—shounen-manga utilize linguistic methods to its advantage, actively calling upon linguistic stereotypes to create economic characters as well as to actively move forward their narratives.
While this means that shounen-manga has shown itself to be somewhat more difficult to read, and, with their use of stereotyped speech patterns, an interesting subject of research, that does not mean that shoujo-manga is inferior. Indeed, as Ôtsuka (1994) has argued, shoujo-manga have actually been criticized for being difficult to read: As opposed to difficulty in reading in shounen-manga, however, this is not so much a function of the actual language being seen, but rather how it is presented within the text. While the particulars of how this is balanced out cannot be given full attention in this particular venue, the corpus data has thus shown that language can be used as an effective marker for redefining genre in manga.
166
5.1.3 Future issues
One of the advantages of a corpus is the ability to reuse the data for additional studies depending on the topics of interest: Now that the data has been collected, research topics are only a matter of one’s imagination. However, in reviewing the results here, there are still certain issues that would make the corpus vastly more informative.
In particular, there are two points that must be left to be considered in a different forum in the future: first, the validity of the statistical analyses described here, and second, the actual reading patterns, habits and impact of manga.
One of the main goals of this thesis was to offer a quantitative analysis of language in manga to complement and expand upon the primarily qualitative analyses that have been conducted hitherto now. Creating a corpus was a major undertaking, and it has clearly offered many interesting points to consider and a great potential for continued future research beyond the studies described here. However, while I have conducted statistical analyses wherever possible, it should be noted that while the corpus comes to a large amount of data, it is still in some ways limited in its statistical power. That is to say, while the amount of text overall is considerable, 10 series, or 5 per each genre, leads to the minimum amount available to conduct statistical analyses. It is unclear how the results might change given more series, but it is not unreasonable to expect that there would be differences. As I discussed in Chapter 3 on the morpheme analysis and lexical items, the results for shounen-manga seem to be largely a result of their peculiar plots, and it may be that adding additional series that are less restricted in plot for both shoujo-manga and shounen-manga would change the results—or, perhaps, confirm that these are valid differences.
Another issue that I was not able to approach here was how manga are actually read, and how readers interact with them. These are points that came up along the way when thinking about the impact of the linguistic structures and patterns seen in manga. One of the themes of Chapter 2, on how different categories of text can potentially create different reading experiences, begs one to think about how manga are really read, and how such hypotheses could be tested experimentally. The discussion in Chapter 4 on characterization through language patterns was started partially with the question of manga’s popular linguistic impact in mind, and it, too, would benefit from more input on the reading and linguistic habits of young people. While this was outside the scope of this thesis, it is a point which would be well worth considering in future studies using the corpus data. Given the amount of data collection they would require on top of the already labor-intensive basic corpus construction and morpheme analysis, neither of these points could be properly assessed in this thesis, but they are issues that I plan to approach in future research.
167