Article body

Introduction

Various Japanese scholars have claimed that the Japanese history of writing started with translation, and this phenomenon is well documented (Furuta, 1963; Morioka, 1968, 1988, 1999; Yoshioka, 1973; among others). Yet, the field of Translation Studies is not as established in Japan as it is in the West (Itagaki, 1995). Books on translation theory are available, but most of the theories explained in these books are drawn from the works of Western theorists such as Dryden, Benjamin and Nida, among others (Furuno, 2002; Hatano, 1963; Hirako, 1999; Itagaki, 1995). Also, so far, translation scholars in Japan have not undertaken descriptive research into contemporary translation on a large scale. My goal in this paper is two‑fold. I will provide a background to translationese[1] in Japan and introduce a preliminary yet promising pilot study which includes descriptive analyses of translationese in literature. This is preliminary to the task of expanding our understanding of the field and improving the quality of Japanese translation in general, including translator education and translation publishing. I hope that by presenting this pilot study, which yielded promising results worthy of further study on translationese in literature, this paper will contribute to the study of translation.

Japanese Translationese: Background and Development

Japanese translationese, or hon’yakuchoo (hon’yakubuntai, hon’yakugo), has been established in Japan since well before the hypotheses of translation universals were consolidated in the West. It refers to a kind of language, or sometimes words and expressions, specifically used in translating foreign texts into Japanese (Yanabu, 1998). The features of translationese are said to consist of distinct linguistic structures that are not found in more natural Japanese. These features can vary from the word to the sentence level (Sato, 1972; Shibatani, 1990). Despite some scholars’ arguments against translationese as being bad Japanese, many argue that translationese has contributed to the development of the Japanese language throughout its history (Furuta, 1963; Morioka, 1968, 1988, 1999; Taniguchi, 2003; Yoshioka, 1973; among others).

In order to gain a better understanding of contemporary Japanese translationese, a history of the Japanese writing systems and translationese will be discussed here. Japanese readers seem to have a higher tolerance for the unnatural version of Japanese found in translated texts compared to readers in the Anglo‑American tradition in Canada or the US. One possible explanation for this tolerance may be found in the history of translation in Japan.

The Japanese did not initially possess their own means of writing and thus adopted the imported Chinese writing system in the fifth century. Subsequently, Japan’s contacts with China became more frequent, and new cultural elements and ideas started arriving in Japan in the form of writing (Mitani and Minemura, 1988). However, in order to access the content of these new materials, the Japanese had to learn the Chinese writing system. Japanese and Chinese belong to different language families, thus they are structurally very different (Crystal, 1987). For example, the Japanese word order is canonically Subject-Object-Verb, while the Chinese word order is Subject-Verb-Object. Thus, in reading and writing Chinese texts, the Japanese utilized markings and signs that enabled them to treat these texts as Japanese (Sato, 1972; Fujii, 1991; Wakabayashi, 1998, 2005). This practice of reading Chinese in Japanese was called kanbun kundoku, meaning “interpretive reading of Chinese” (Furuno, 2005) or “Chinese read in the Japanese manner” (Wakabayashi, 2005). These texts were far from natural Japanese; nonetheless, readers were expected to study the rules of this technique and to learn words and concepts associated with it. In a sense, this was the first form of translation in Japan, and it was also the first writing system. The ability to read and write Chinese texts as Japanese was viewed as the mark of an educated person by literate people who accepted this unnatural language (Shibatani, 1990). Thus, the fact that one could read and write Chinese texts was associated with a positive notion in spite of it being “unnatural” Japanese. Early translation in Japan was a type of foreignized translation as stated above. At the same time, the translation culture in Japan may have begun with “adequacy” of translation which valued translators’ adherence to the source norms (Toury, 1995, p. 57).[2]

Much later in the history of Japanese translation, when a large number of Western texts reached Japan in the Meiji period (1868-1912),[3] they were also translated into an unnatural version of Japanese, which was called oobun chokuyakutai,[4] meaning “direct translation style of European texts” (Sato, 1972; Yanabu, 1982, 1998; Fujii, 1991; among others). Oobun chokuyakutai includes the following features: using loanwords (Yanabu, 1982, 1998, 2003), creating specific phrases to take the place of linguistic structures absent in Japanese (Hatano, 1963; Morioka, 1988, 1999; Sato, 1972), utilizing Sino-Japanese words to express concepts foreign to the Japanese people (Yanabu, 1982, 2003), and making explicit use of linguistic forms deviating from natural Japanese (Fujii, 1991; Morioka, 1988; Yanabu, 1998). These textual features were quite different from everyday spoken language and made it necessary for readers who wanted to read translated texts to raise their educational level by learning the structures and vocabulary used in the language of translation. This background may have promoted the readers’ high tolerance for the unnatural Japanese because of the high status associated with being educated and able to read translated texts. The techniques for reading these texts were essentially similar to those for reading Chinese. This translationese changed over time to incorporate some aspects of more natural Japanese. However, today the language used in translation still retains some of the features of oobun chokuyakutai. I call this “contemporary Japanese translationese.”

Two positions for translation exist in Japan as in the West. Some argue for domestication and others for adequacy. Tanizaki calls adequate translation, or even writings based on translationese, “bakemono (monster)” Japanese (1975 [1934], p. 70). Also, Mishima (1973 [1959]) claims that readers should refuse to read translations that are written in Japanese and are hard to understand even when they are faithful to the original. On the other hand, Kawamori Yoshizo, a French literature scholar and translator, argued in his article entitled “Hon’yakuron [Translation Theory]” in favor of using foreign expressions from the original texts because translation is a way to supplement what is lacking in the Japanese language (Kawamori, 1989 [1944]). Although these parties argue against each other, all agree that the language of translation had an influence on the Japanese language.

Many studies have been carried out to reveal the features of translationese in oobun chokuyakutai (e.g., Furuta, 1963; Morioka, 1968, 1988, 1999; Taniguchi, 2003; Yoshioka, 1973). These studies essentially focus on the types of Japanese that were formed in the process of translating European languages into Japanese during the period following the opening of the country. Most of them conclude that features of translationese have become more common in texts originally written in Japanese.[5]

Many of the structures that were representative of translationese in the Meiji period still sound like translation to contemporary Japanese readers. In publications on translation, the type of language that contains these features is often referred to as hon’yakuchoo, contemporary Japanese translationese (Furuno, 2005; Itagaki, 1995).

Japanese Sociocultural Situation of Translations

As seen above, translationese in the West and in Japan has a very different historical background. In the West, “translationese” has been thought of as a trait of bad translations (Venuti, 1995, p. 4; Steiner, 1998, p. 280). In Japan, however, the acceptance of the language used in translation played an important part in the development of the written language. In this section, sociocultural aspects of Japanese translation are briefly illustrated.

Translators in Japan seem to enjoy more visibility than in Anglo‑American society. In contrast to American norms (Venuti, 1995, 1998), translators’ names are almost always clearly placed on the front cover in the same size font, though usually underneath or next to authors’ names,[6] which suggests a different position in terms of the translators’ social status. However, if the translator is particularly popular, then his/her name appears more clearly on the obi or band, which is usually placed over the book cover,[7] along with exclamatory phrases such as “a long-awaited new translation by so-and-so!” Moreover, literary translation is seen as a desirable occupation. Furuno reports that in the 1970s numerous translator-training institutions were established, which “reflect[s] the growing demand and interest in Japanese society for translation” (2002, p. 325). Also, my preliminary search for translation schools in Japan using an Internet search engine revealed hundreds of different schools, programs and correspondence courses (online or by post). As well, there are a number of “how‑to” books on translation for sale. A monthly periodical called Hon’yaku no Sekai [The World of Translation] often includes articles on how to become a translator or how to translate bestselling books. According to Shuppan NEWS,[8] about 7.6% of the 77,074 books published in Japan in 2006 were translations. In contrast, in English‑speaking countries (US, UK, Canada, Australia and New Zealand), translations made up 3.8% of the 375,000 books published in 2004. In the US alone, translations were only about 2.5% of the 195,000 books published in the same year.[9] In other words, Japan produces a relatively large number of translations every year. In bookstores, there is often a “foreign literature” section containing translated literature. Translators would probably admit that their profession arouses envy. These are all indications of a strong interest in translation in Japan.

Such interest may have had an influence on the amount and degree of translationese in Japanese translations. Further investigation through attitude studies may reveal the underlying complexities of the situation. As mentioned above, Toury (2004 [1978], 1995, 1999) regards translation as a norm-governed activity in which participants in society determine norms in terms of historical, social and cultural aspects. Studies of translations should be able to provide insights into the norms that govern translational activities.

First, attitudes toward translationese can be studied through an examination of “how-to” books published for translators and aspiring translators. The attitudes toward translationese found in such books are quite negative.[10] These books emphasize the importance of domestication, precisely what Venuti (1995, 1998) argues against. Using the results of her investigation into translators’ attitudes towards translation, Furuno (2002) argues that tolerance for translationese in Japan started shifting in the 1970s and is now moving toward domestication, a tendency similar to that found in Anglo-American translation. A sociolinguistic research technique can be applied also to the investigation of attitudes towards transaltionese. So far, Furuno’s 2005 study is the only one that has attempted to investigate attitudes towards Japanese translationese. Furuno (2005) collected survey responses from students in a translation school. However, this type of selection can skew results because those who are studying how to translate may have already been primed to have a certain attitude towards translationese. I believe that in order to arrive at generalizations about the attitudes of Japanese people towards translationese, one would need to collect survey data from a larger range of general readers who are consumers of books and who have no preconceived notions regarding translations.

Descriptive Studies of Contemporary Japanese Translationese: Research Prolegomena

There are almost no descriptive research studies of contemporary Japanese translationese. As noted earlier, scholars who study translationese tend to focus on the Meiji, Taishō and early Shōwa periods and their influence on the contemporary Japanese language. Since there was a great deal of translation activity during the Meiji period owing to the need to translate Western materials, scholarly interest in translationese from that time is understandable. I believe, however, that the study of the current situation of translationese will contribute to the understanding of translation in contemporary Japan. In my research, I investigate translationese in literary translations. I chose to focus on one genre simply because the variety of genres and associated styles makes it difficult to group them together for an effective study. In addition, literature potentially draws on all types of discourse in its narrative strategies and representation of speech genres.

Although it seems everyone agrees that translationese exists in Japanese, features of contemporary translationese should be identified within the domain of a descriptive study. So far, there has been no systematic study done to identify the features of contemporary Japanese translationese, but suggestions laid out by Baker (1993, 1995, 1996, 1999, 2004) for the corpus-based approach may help in discovering them. First, translated texts need to be compared with non-translations (texts that were originally written in Japanese). For this, comparable corpora are useful. Comparable corpora include a set of translated texts and a set of non‑translations. Also, parallel corpora of original texts and translated texts can be helpful when a researcher needs to investigate the original text as well. In order to identify and substantiate specific characteristics of translationese in English‑Japanese translation, linguistic features deemed to be characteristic of translationese should be checked in both the translation corpus and the non‑translation corpus (Baker, 1993; Laviosa‑Braithwaite, 1995).

For instance, if one wishes to test the hypothesis that the use of third person pronouns is more explicit, i.e., overused or used where they are not needed in Japanese, in translationese, one needs to draw on both the translation corpus and the non-translation corpus in order to extract the frequency of the personal pronouns in question. This will test whether the pronouns being investigated occur more in translated texts than in non‑translated texts as has been hypothesized. The investigation would be easier if corpora of translated literature and non-translated literature were already available, as in similar projects in English (Baker, 1996, 1999, 2004) and in Finnish (Tirkkonen‑Condit, 2002; Puurtinen, 2003a, 2003b). However, in Japanese, there are currently no ready-made corpora that suit the needs of this type of research. As a result, such sets must be developed. There is technology available that would make this task easier. A set of corpora can be created by scanning printed books using an OCR (Optical Character Recognition) equipped scanner to digitize the texts. The resulting corpora may be too small for definitive generalizations when compared to English or Finnish counterparts, which were created within institutions with a number of researchers. However, creating a set of corpora, no matter how small, is a necessary first step in identifying the features of translationese.

Another issue for descriptive studies of translationese is the type of texts to investigate. The genre of the texts being investigated may have an influence on the outcome. Furuno (2005) investigated translators’ attitudes towards translation using non‑fiction texts. However, her findings may provide only a partial description, since there may be differences across genres in the use of translationese. According to Sato (1972), a type of translationese that originated in translations of European languages is now used as a writing style in philosophy, the sciences and the social sciences. Another type of translationese that originated from kanbun kundoku, or “Chinese read in the Japanese manner,” is used for technical or scholarly writing (Morioka, 1968; Sato, 1972). Additionally, institutionalized writing (e.g., newspaper or magazine articles) uses a specific type of language, and Finnish readers found it difficult to identify such writing as translations or non‑translations (Tirkkonen‑Condit, 2002). This may be true of Japanese as well, because a certain style has to be followed in news reporting, as observed in books on newspaper translations (Negishi, 1997; Negishi, 1999). Therefore, if one were to investigate Japanese readers’ attitudes toward translationese, I would argue that examining translated general literature, such as fiction and other popular literature aimed at a wider range of readers, would provide more accurate information, since the styles used for these genres are not prescribed as in the genres mentioned above. Moreover, other genres are also represented in literature. For example, within works of fiction, conversation plays an important role in plot in addition to narration, and the representation of spoken Japanese is quite different from the narrative form. Translators of literature, then, cannot use the styles reserved for non‑fiction writing. Literary representation of speech may reveal differences in translations and non‑translations. I also argue that a study should be carried out using popular texts, such as bestsellers. Most of the translated books on the bestseller lists between 1970 and 2005 were fiction, though there were a few non-fiction works (e.g., how‑to books on business and self‑help books), and most of these were translated from English. An exploratory look at these translated books shows that translationese appears to have some particular linguistic features.

In sum, descriptive studies in contemporary Japanese translationese is an almost untouched area for Translation Studies. There is much to be done that may be important for translators’ practices, translator education and the publication of translated books. The research findings may also contribute to theoretical discussions. For example, the findings may provide more information for Toury’s notion of translation as a norm-governed activity (1978/2004, 1995, 1999) and for Polysystem Theory (Even-Zohar, 1979, 1978/2004;[11] Dimic and Garstin, 1988).

Preliminary Examinations of Features of Translationese

As demonstrated in the previous section, translation scholars in Japan have yet to conduct descriptive research on contemporary Japanese translationese. I intend to fill this gap in the course of my research. As a first step, I will introduce a pilot study in which translationese in literature is analyzed descriptively. Although attitude studies have to be conducted with a large number of participants, a preliminary examination of features found in translationese can be conducted with a small corpus. In this section, I will provide findings from an investigation of small comparable corpora. I emphasize that this is merely a preliminary examination. However, as can be seen in the exploratory analyses below, this pilot study stresses the need to develop more detailed investigations of translationese in Japanese.

The first book used for this preliminary inquiry is the translation of The Bridges of Madison County by Kiyoshi Matsumura and published as Madison‑gun no hashi in 1993. This translation was on the bestseller list in 1993 and 1994. The second book is a two-volume novel called Shitsurakuen [Paradise That Was Lost] written by Jun’ichi Watanabe. It was published in 1997 and was also on the bestseller list of the same year. These books have many similarities. The main characters in each are a middle‑aged man and woman who engage in an extramarital relationship.The stories involve much description of feelings and narrations of what a man and a woman in love experience. By comparing specific features in both translation and non‑translation, one can learn whether or not these features are specific to translated texts.

The comparative corpora were digitized, using a Japanese OCR‑equipped scanner. From the translation, Madison-gun no Hashi, pages 50 to 61 were digitized. From the non-translation, Shitsurakuen, pages 54 to 63. While the number of pages varies, the number of lines in the corpora was controlled to be almost equal.[12] The concordances were conducted using a program called ConcApp, and other analyses were done manually.

Scholars have argued that the following features are some of the characteristics of translationese: (1) use of overt personal pronouns (Yanase, 2000; Miyawaki, 2000; Nakamura, 2001); (2) more frequent use of loanwords (Yanabu, 1982, 1998; Yoshioka, 1973); (3) use of abstract nouns as grammatical subjects of transitive verbs (Morioka, 1988, 1997, 1999; Yoshioka, 1973); and (4) longer paragraphs (Miyawaki, 2000). For the examinations of features of translationese in this paper, the features mentioned above are considered. The hypothesis is that the comparative corpora can reveal preliminary evidence for these features of translationese.

With regard to the use of overt personal pronouns, Yanase (2000), Miyawaki (2000) and Nakamura (2001) claim that third person pronouns such as kare “he” and kanojo “she” are used more often in translation. The examples of translation and non-translation below show a passage from each. In the passage from the non-translation, there are no third person pronouns used in reference to a person. The passage is about a person whose name was brought up in the previous paragraph. In my own English translation, on the other hand, five personal pronouns needed to be supplied.

The passage from the translation, Madison‑gun no Hashi, on the other hand, shows four explicit personal pronouns, kare “he” and kanojo “she,” which are underlined.

The investigation of the occurrences of third person pronouns in the corpora reveals a similar trend. Table 1 shows the frequency of kare “he” and kanojo “she” in the corpora. The difference between the two corpora is obvious: the third person pronouns appear much more frequently in the translation than in the non-translation.

Table 1

Third person pronoun occurrences

Third person pronoun occurrences

-> See the list of tables

The translationese feature claiming a more frequent use of loanwords can also be seen. In the sample passages above, loanwords are indicated in italics. Examination of the corpora shows that loanwords are indeed used much more frequently in the translation than in the non‑translation (Table 2). The “token” refers to actual occurrences of loanwords, and the “type” refers to the numbers of different loanwords used in each corpus. In other words, if the same word appears three times, then the token count is three and the type count is one. In the translation, in addition to 106 tokens of loanword occurrences, proper nouns as loanwords appeared 17 times, while character names appeared 26 times. The proper nouns included location names (e.g., Madison, Washington, etc.), company names (Samsonite, Zippo, etc.), and the name of a magazine. In the non-translation, there were no instances of proper nouns in loanwords.

Table 2

Loanword token frequency and the number of types

Loanword token frequency and the number of types

-> See the list of tables

As for the use of abstract nouns as grammatical subjects/objects (or semantic role of agent), there were no instances in the pages that were digitized. An explanation for this may be that this particular translator avoids using abstract nouns in this way. In other words, this may well be a feature of translationese, but much larger comparable corpora will be necessary to find enough cases of abstract nouns as grammatical subjects/objects before any definite conclusion can be drawn.

The feature of longer paragraphs in translation is also supported by these results. The paragraph lengths were measured using a “word count” tool in a word processing program. The counts were based on the number of characters in a paragraph (Table 3). As can be seen, the difference is clear. Since the non‑translation contained more dialogues than the translation, the results without dialogues are also shown. The paragraphs in the translation are longer than in the non-translation.

Table 3

Average paragraph length: number of characters per paragraph

Average paragraph length: number of characters per paragraph

-> See the list of tables

Additionally, the bottom row of the table shows the length of each turn of dialogue or conversation, which reveals that even the dialogues or conversations are longer in translation than in non-translation. A “turn” is a conversational convention defined as “a single contribution of a speaker to a conversation” (Crystal, 1987); in other words, the alternating participation of each speaker in the conversation (Levinson, 1983). Conversational turns have been studied mostly in the linguistic fields of pragmatics and discourse analysis, and it has been argued that turn‑taking patterns can differ depending on the language of the conversation (Tanaka, 2000). However, in the sample studied here, the non-translation contains more turns of dialogues (38 turns) than the translation (3 turns). This may reflect the differences in turn-taking patterns between Japanese and English. Also, as seen in example number 4 below, dialogues are often embedded in paragraphs in translation (shown underlined), rather than beginning a new line for each conversational turn. In other words, these variations may point to differences in speech representations in literature between translations and non-translations. This may have caused the differences in the length of the paragraphs, which is another aspect that needs further investigation.

Although this pilot study may not be conclusive at this stage because of the use of small comparative corpora, it nonetheless reveals obvious differences in the language used between translation and non-translation. The results found here could be due to the writing styles or habits of the particular translator and writer. However, the future research project for which this pilot is intended will produce more conclusive results from far larger corpora. The findings in this pilot project are very suggestive, and they indicate the possibility that the features of translationese investigated in this study are in fact used by translators, pointing to the establishment of translationese descriptions.

To conclude, as demonstrated above, it is possible to substantiate the features of translationese. Since empirical evidence for what constitutes translationese has been limited so far, this type of study can make a significant contribution to descriptive Translation Studies. Once the features are described systematically on a larger scale, further research into readers’ attitudes toward translationese can be studied in order to locate translationese within Japanese society.

In the West, translationese has traditionally been regarded as a sign of bad translation; however, a more neutral view has appeared in the works of Baker and Toury. On the other hand, Japanese translationese has followed a different path. Japan’s long history of documented written materials provides valuable data in understanding how different types of translationese over the centuries have influenced the Japanese language at various times. It is only recently that we hear more about the notion of more fluent or domesticated translation. Furuno (2002, 2005) has made a first step toward further understanding by analyzing the changing attitudes of the Japanese towards translationese in non-fiction writings, while Yanabu (1982, 2003) made his contribution by proposing a translation theory based on the phenomena of words in translationese. More investigation needs to be carried out in order to gain further understanding of the phenomenon of translationese in Japan.