How Do Female and Male Characters Speak in the Japanese Translation of English Crime Novels?

The repertoire of linguistic expressions that index sociopragmatic meanings differs considerably from language to language. This difference becomes particularly noticeable when one language is translated into another. As an example, this study examines dialogs in the Japanese translations of two English crime novels to see how the translator deals with normatively gendered morphological forms in Japanese for which no corresponding forms exist in English. The analysis shows that although the same imperative, declarative, and interrogative forms are used for female and male characters in the English originals, in the translations, gendered forms are used not simply based on the gender of the characters but on the interaction of gender with other social variables, in particular class and age. The results and their theoretical implications are discussed, employing the notions of indirect indexing, double-voiced discourse, and cultural filter.


Introduction
In language and gender research, constructivist approaches have been widely adopted since the early 1990s, and it is commonly maintained that a speaker discursively constructs diverse forms of femininity or masculinity as an aspect of their social persona, or identity, by deploying linguistic and paralinguistic resources that index sociopragmatic meanings such as femininity and politeness. The repertoire of these resources (e.g. standard and non-standard language variant forms, morphological and lexical items, (in)direct speech acts, intonation patterns) differs widely from language to language, which is often illuminated when one language is translated into another. How do translators deal with this difference? What sociopragmatic meanings are lost or added through translation? How does the translator's choice affect the construction of characters' personae? This study addresses these questions by examining Japanese translations of dialogs in two English novels, focusing on genderrelated meanings. Japanese is considered a highly gendered language in which many morphological and lexical items (especially in standard Japanese) are normatively gendered as feminine or masculine variants. In contrast, there are no forms in English that correspond directly to these gendered forms. Specifically, I analyze how the translator assigns (normatively and stereotypically) feminine or masculine sentence-final morphological forms to different characters in the novels to construct their personae and why the translator assigned the variants in the way he did, which necessitates considerations of the indexical process that involves a 'cultural filter' concerning the norms of the target language (House 2018(House , 2019. Section 2 briefly describes the theoretical framework for this study; Section 3 presents an outline of the Japanese gendered linguistic forms; Section 4 describes the data and the methods of data analysis; Section 5 analyzes the data quantitatively; Section 6 examines the data qualitatively and considers how and why the translator chose gendered forms for different characters in the novels, and what implications the findings have for the relationship between linguistic forms and social identities related to gender; and Section 7 presents conclusions.

Theoretical Framework
This study employs a constructivist approach, especially drawing on the notions of indexicality and language ideology (Ochs 1993;Eckert 2008) and double-voiced discourse (Bakhtin 1981;Bucholtz 1999). I also employ the notion of 'cultural filter' (House, 2018(House, , 2019, an ideological process that has important bearings on translation in general. The approach taken in this study thus also entails critical discourse analysis in that it considers the role of (hidden) ideology in the choice of sociolinguistic variant forms in translation. Eckert (2008: 453) argues that the sociopragmatic meanings of a linguistic form as a variant of a variable (e.g. a standard or non-standard form) are potentially multiple, imprecise, and variable depending on the social context: … the meanings of variables are not precise or fixed but rather constitute a field of potential meanings -an indexical field, or constellation of ideologically related meanings, any one of which can be activated in the situated use of the variable. The field is fluid, and each new activation has the potential to change the field by building on ideological connections.
To account for this flexible indexical relationship, Eckert (2008: 455) further contends that: the very fact that the same linguistic variables may stratify regularly with multiple categories -e.g. gender, ethnicity, and class -indicates that their meanings are not directly related to these categories but to something that is related to all of them. In other words, variables index demographic categories not directly but indirectly (Silverstein 1985) through their association with qualities and stances that enter into the construction of categories.
For example, the potential meanings in the indexical field of the standard velar variant [N] of the variable (ng) in English may include stances and qualities such as being formal, articulate, educated, and even pretentious. These stances and qualities may then be ideologically linked to middle-class people, women, mid-westerners, broadcasters and others. Honorifics offer another example. They are commonly associated with stances and qualities such as being respectful, polite, reserved, and refined (e.g. Okamoto 2021). These stances may then be ideologically linked to women, members of higher social echelons and others. Thus, the velar variant in English and Japanese honorifics both illustrate Eckert's point that the same linguistic form stratifies with multiple social categories/types through indirect indexing.
Another important notion for the analysis in this study is that of 'doublevoiced discourse' (Bakhtin 1981;Bucholtz 1999). Double-voiced discourse uses constructed, or reported, speech, such as the dialogs examined in this study, and "serves two speakers at the same time and expresses simultaneously two different intentions: the direct intention of the character who is speaking, and the refracted intention of the author" (Baktin 1981: 324). That is, one utterance simultaneously represents the voice of the character in the story and the voice Okamoto 10.1163/26660393-bja10031 | Contrastive PragmaticS (2021) 1-28 of the narrator or writer of the story. Bucholtz (1999: 448) gives an example in which Brand One, a male high school student, tells a story to the researcher about his experience of encountering an African American 'antagonist' in the following example: (1) 15. And then he walked up beside me right?
16. And there was like a wall {right there kinda you know?} (high pitch) 17. And then (I pushed him up against it) and he's like, 18. {"What you gonna do you little punk ass whi:te bi:tch"} (slower rate, lower pitch) The antagonist's utterance in line 18 is an instance of double-voiced discourse in that it represents the voice of the antagonist who has racialized the conflict using AAVE and referred to Brand One as "little punk ass whi:te bi:tch," which challenges Brand One's white masculinity. At the same time, this utterance represents the voice of the narrator (Brand One), who assesses the speech of the social type of the antagonist in the context in question and chooses to use stylized, or stereotypical, AAVE to construct the antagonist as hostile and threatening, juxtaposing it with Brand One's standard English in which he presents himself as a non-confrontational 'mainstream' white male. The present study examines the dialogs in novels translated into Japanese from English. I regard translated dialogs as instances of a complex form of double-voiced discourse. It is complex because the dialogs in the Japanese translation are double-voiced discourse transformed from the dialogs in the original novels in English, which are already double-voiced discourse. In the case of translated dialogs, at one level, we 'hear' the voice of a character speaking in Japanese, who takes a certain stance toward the given situation, especially toward the addressee; and at another level, we 'hear' the translator's voice, which assesses the speech style that he/she thinks is appropriate to construct the character of a certain social type interacting in a particular context. In this assessment, the translator is most likely to call upon Japanese linguacultural norms vis-a-via the linguacultural norms in American English. In the case of gender-related linguistic forms, cultural ideologies of language and gender become most relevant. In this respect, translation is a good example of a 'cultural filter' that operates as a form of intercultural communication, in which the translator "takes into account culture-specific target norms such as conventions of text production and communicative preferences in certain genres" (House 2020: 15). This study illustrates that examining the choice of expressions, or variants, in translations can illuminate the ideologically-mediated process of linking the stances (e.g. polite, forceful) and qualities (e.g. refined, It has been argued, however, that these gendered forms (in standard Japanese) represent 'traditional' linguistic gender norms and stereotypes (Inoue 2006;Nakamura 2007aNakamura , b, 2014Okamoto and Shibamoto Smith 2016). That is, they are constructed based on the ideology of gender in that women are expected to use feminine forms as they are associated with stances and qualities such as being gentle, polite, and refined, while men are expected to use masculine forms as they are associated with stances and qualities such as being forceful, assured, and coarse. Obviously, Japanese speakers are not automata and do not always conform to these linguistic gender norms in practice for a variety of reasons, as has been demonstrated by a number of empirical studies (e.g. Okamoto 1995Okamoto , 2016Shibamoto Smith 2004, 2016;Miyazaki 2004;Mizumoto 2006;Okada 2008;Sunaoshi 2004;Abe 2010).
Unlike the diverse use of normatively gendered forms in real social situations, however, stereotypically gendered forms are extensively used in the media, including novels, films, anime, and other genres of fiction. It has been reported that main characters in fiction tend to use standard-Japanese-based gendered language, while secondary and peripheral characters often use a non-standard or regional (or 'pseudo-) dialect (e.g. Kinsui 2003;Satake 2003;Shibamoto Smith 2008, 2016;Shibamoto Smith and Occhi 2009;Nakamura 2013). This stratification indicates that it is not simply gender but the intersection of gender and other social variables (e.g. class, age) that affects the media representations of speech styles of characters in certain social groups. Such stereotypical speech styles tend to be recycled in the media as a kind of mediatized language (Agha 2011).

Data and Method of Analysis
The data consists of dialogs in two English crime novels The Burning Room (2014) and The Late Show (2017) by Michael Connelly, the best-selling author, and their Japanese translations by Furusawa Yoshimichi (2018 and 2020, respectively). In The Burning Room (BR), the protagonist is Harry Bosch, a 60-year old male detective in the Open-Unsolved Unit in the Los Angeles Police Department (LAPD). His partner is a Hispanic female detective, Lucia Soto. She is 28 years old and has worked in the LAPD for five years. They are assigned as partners to a 10-year old unsolved case, which turns out to be a murder case involving the city's powerful businessmen and a former mayor. They have never worked as partners before this case.
In The Late Show (LS), the protagonist is Renée Ballard, a female detective in the LAPD. She is in her mid-30s and single. Although she is a competent detective, she is ordered to work the night shift, a less prestigious post, because she lost the sexual harassment case she brought up against her superior due to the false testimony of the key witness. In this novel, she investigates two casesa brutal beating and torturing case of a transgender prostitute and the murders of five people in a nightclub. Ballard's partner is John Jenkins, a veteran of 25 years in the LAPD, who now works the night shift due to his wife's illness.
The two main reasons for choosing these novels are (a) they both include characters from diverse social backgrounds, such as detectives, their supervisors, suspects, witnesses, and family members, allowing an examination of what variant forms the translator chose for characters in different social groups; and (b) the protagonist in The Burning Room is a male detective and that in The Late Show is a female detective, which may be relevant as the primary focus of this study is gender and speech. For example, in the English original the female protagonist occasionally uses 'rough' language, including expletives, and I was curious to see how they are translated into Japanese.
To analyze the dialogs, I chose three types of sentences in the English originals: imperatives, declaratives, and interrogatives. I examined the Japanese translation of each sentence, focusing on the sentence-final form. The gender of each form -i.e. feminine, masculine, or neutral -was identified based on the normative gender classification of variant forms given in previous studies (Masuoka and Takubo 1992;Okamoto and Sato 1992;Ide and Yoshida 1999;Okamoto and Shibamoto Smith 2008). Furthermore, only the sentences in Japanese that were in the so-called plain, or informal, forms were examined because formal utterances, which usually end with addressee honorifics, rarely indicate gender differences.
I first examined the sentences uttered by the two protagonists, Bosch and Ballard, and then those uttered by other characters (i.e. secondary and peripheral characters). For each type of sentence used by each protagonist, a maximum of 20 sentences addressed to each of the other characters were examined from the beginning of the novel. For each type of sentence used by each of the other characters, a maximum of 20 sentences addressed to the protagonist were examined. When there were fewer than 20 sentences uttered by a protagonist or one of the other characters, all of them were examined. The data were first analyzed quantitatively to see the distribution patterns of feminine and masculine forms with characters of different social types. They were also analyzed qualitatively to see how certain affective meanings associated with different variant forms are linked to characters in different social categories.

Distribution of Gendered Forms among the Characters of the Novels
This section presents the findings concerning the distribution of gendered sentence-final forms in the Japanese translations of imperatives (5.1), declaratives (5.2), and interrogatives (5.3) with regard to their sentence-final forms.

5.1
Japanese Translations of English Imperative Forms Imperative sentences in the English originals are either affirmative (e.g. stand up) or negative (e.g. don't tell a lie). Tables 1 and 2 show the frequency of use of gendered sentence-final forms in the Japanese translations that correspond to English (affirmative and negative) imperatives used by Bosch and Ballard and by other characters (see Appendix for all the Japanese forms that corresponded to English imperatives, declaratives, and interrogatives that were found in the translations).  According to Tables 1 and 2, the forms used by Bosch and Ballard and other male and female characters in the Japanese translations are normatively gendered with only a few 'exceptions,' despite the fact that the forms in the English originals are all direct imperative forms and not gendered (see Section 6.1 for examples).

5.2
Japanese Translations of English Declarative Forms English declarative sentences include utterances such as You did a good job and That was a long time ago, which may be interpreted as an assertion, evaluation, and others depending on the context. Tables 3 and 4 show the frequency of use of gendered sentence-final forms in the Japanese translations of English declaratives used by Bosch and Ballard and by other characters.  As seen in Table 3, Bosch and Ballard used normatively gendered forms except for a few cases in which Ballard used masculine forms. In Table 4, however, this pattern of normative distribution is radically disrupted by the female other characters' use of masculine forms, which accounted for 93% of the total instances (see Section 6.2 for examples). Tables 5 and 6 show the frequency of use of gendered sentence-final forms in the Japanese translations of English interrogatives used by Bosch and Ballard and by other characters. Table 5 again shows a normative pattern of use by Bosch and Ballard. Table 6 shows that while the male other characters used only masculine forms, the female other characters used non-normative, or masculine, forms 27% of the time.

Intersection of Gender, Class, and Age
The analyses presented above show that although the distribution of gendered forms in the Japanese translations is normative to a large extent, there are a considerable amount of non-normative uses, especially by some of the female characters.2 Why did the translator assign masculine forms to some of the female characters? To consider this question, let us look at the social backgrounds of the female characters (Table 7). They are divided into Group A, consisting of female characters who rarely used masculine forms, and Group B, consisting of female characters who used masculine forms extensively. (Note that both groups also frequently used neutral forms.) Table 7 Female characters' social backgrounds and use of gendered forms  Table 7 shows that the female characters in Group A are all professional women in the middle class. In contrast, the female characters in Group B vary considerably in their social backgrounds. Among them, Maddie is a young woman in the middle class; Leslie, Tutu, and Beatrice are older women in the working class; and Alicia is a young woman in the working class.3 This suggests that it is not simply the gender identities of characters that affected but not directly determined the translator's choice of variant forms, but rather the interaction of gender, social class, and age. We discuss this issue further in the following section.

Indexicality and the Construction of Social Identities in Translation
This section considers why gendered forms in the Japanese translations were distributed to different kinds of characters in the way there were. In particular, I consider the ideological linking of the affective stances and qualities (e.g. politeness, forcefulness) indexed by variant forms to the identities and interpersonal relationships involving different characters. Ballard: Don't worry, I will, Shinpai-shinaide, hanasu kara.

Japanese Translation of English Imperatives
In (1) and (2), both Bosch and Ballard say Don't worry to their interlocutors, but in the Japanese translations, Bosch says Shinpai-suru na 'Don't worry,' a negative direct imperative form, and Ballard Shinpai-shinaide 'Don't worry, please' , using the V-te form, which is a continuous form of a verb and can be extended to either V-te kudasai 'please V' or V-te chōdai 'please V.' Thus, the voice of Bosch indicates that he is taking more direct, forceful, and less polite stances toward the addressee than Ballard.
Examples (3) and (4)  The form V -te kure in tanonde kure 'ask (her) for me' that Bosch uses in (3) is grammatically a direct and forceful imperative form of kureru 'to give (a favor),' even though Bosch is asking Soto to do him a favor. In (4), Ballard is extremely angered by Detective Rogers Carr to the extent that she even used two expletives (half-assed and fucking) in English, but in Japanese she uses no expletives and the request form V-te chōdai 'please do V,' which is said to be usually used by children or women and is gentler and normatively politer in stance than the direct V-te kure form that Bosch used. To give one more example, in The Late Show, Ballard is in a life-threatening situation, confronting the extremely dangerous prime suspect. Yet, her English commands Wait and Leave her out of this are translated into the V-te request forms matte 'wait, please' and kakawaranaide 'do not involve her, please' rather than direct imperatives. Thus, the Japanese translations of English imperative forms add sociopragmatic meanings that are not present in the English originals. Bosch's voice in (3) is direct, demanding, and therefore, masculine, while Ballard's voice in (4) and in the immediately preceding example is quite weak and undemanding even though the situation is extremely hostile. The choices the translator made in (1)-(4) suggest his voice, or the result of cultural filter, upholding the Japanese ideology of normative masculinity and femininity. To put it differently, the Japanese translations in effect erase the gender-neutral meanings of direct imperative forms in English.
The use of directives by characters other than Bosch and Ballard is also gendered except for the four utterances made by male characters. These 'exceptions' were all in the form of V-te/de 'V, please' or V-naide 'don't V, please,' which was also used by Bosch once. These directives were uttered by speakers who were in a weaker or powerless position (e.g. a suspect) or were attempting to show friendliness with the addressee, as illustrated by Example (5) Beaupre used the imperative form V-te kure (i.e. yondo4 kure), which is normatively masculine, rather than the gentler V-te or V-te kudasai/chōdai request form (i.e. yonde; yonde kudasai/chōdai). Her stance is informal and forceful; some may find it rude for an utterance addressed to a police detective that she has never met before. This choice suggests the translator's voice that regards a direct or plain speech style as suitable for a working-class, middle-aged woman who works in a pornographic film business and is bothered by the police officer's visit.

6.2
Declaratives Examples (7) and (8)  The English sentences bolded in these examples have the same copulative structure 'NP1 is NP2': it's walk-in Wednesday in (6) and It's a vampire case in (7). But Bosch's statement in Japanese ends with the auxiliary verb form of direct assertion da, while Ballard's statement ends with the particle yo, which when preceded by a noun makes the statement a milder assertion than the da-ending. Thus, Bosch's voice is forceful and hence ideologically masculine and Ballard's gentle and hence feminine. Similarly, in BR, when Bosch tells an uncooperative suspect, You now have a big decision, Rodney, the translated sentence Ōkina ketsudan o kudasanaito naranai zo ends with the prototypical masculine particle zo. In contrast, when Ballard angrily tells a hostile prime suspect, who went to her grandmother's house, (the grandmother) doesn't give up information to strangers, the translated sentence mishiranu ningen ni jōhō o ataewa shinai wa ends with the prototypical feminine particle wa, which indexes a milder or weaker stance toward the hostile addressee than the particle zo. Again, these differentiations seem to reflect the translator's voice, or expectations, about culturally normative femininity and masculinity.
As we saw earlier, other male characters than Bosch also used gendernormative declarative forms just as Bosch did without 'exceptions.' Other female characters in Group A (Table 7) showed the same gender-normative pattern of use of declarative forms as Ballard's. However, the women in Group B frequently used masculine forms. In Example (6) we saw one example of this. Example (9) illustrates another female character's use of masculine declarative forms.  (9) both end with a normatively masculine form without an addressee honorific. The use of these forms may sound blunt and impolite for utterances addressed to a detective in their first meeting, especially when Ballard is using addressee honorifics toward Lantana. But as in the case of Beaupre in (6), the forms used by Leslie in (9) suggests the translator's voice in assigning the language of an older and working-class woman -an ideological assumption about women in such a social group. Maddie in Example (10)  The three bold forms in (10) are all stereotypically masculine forms, making direct and forceful assertions, as compared to the corresponding feminine forms (i.e. Doyōbi no yoru yo 'It's Saturday night'; suru ki wa nai wa 'We're not going to'; tanoshimitai no 'want to have some fun') that the translator could have assigned to her. As mentioned in Section 2, it has been found that (standard-Japanese speaking) women, in particular young women, in contemporary Japan have been increasingly using masculine sentence-final forms (e.g. Mizumoto 2006;Okamoto 1995;Okamoto and Sato 1992). Maddie's use of direct forms seems to suggest that the translator considered the current speech styles of young women.

6.3
Interrogatives Examples (11) and (12)  Although Bosch and Ballard use the same English sentence structure you know NP?, in the Japanese translations, Bosch's question ends with the explicit question marker ka, which is regarded as masculine when the preceding verb is in a plain form. Ballard's question, on the other hand, ends with the particle kashira, meaning 'I wonder if …,' which is more indirect or equivocal as a question than Bosch's. Throughout the novels, Bosch's interrogative sentences end mostly with the masculine particles ka or dai ( In (13) and (14), Tutu and Beaupre use the normatively masculine particles kai and dai. (Tutu also used omae, a stereotypically masculine second-person pronoun.) As in the earlier examples, such assignments of masculine forms to Tutu and Beaupre seem to indicate the translator's ideologically-mediated assumption about the language of working-class older women, whose voices are presented as more direct and forceful than the use of -kashira by Ballard in (12). Given these findings, one might wonder if the translator would assign masculine forms to all older female characters or only to working-class women and not to middle-class women. The data examined in this study do not offer the answer to this question as both novels included no older middle-class female characters. It would be interesting to examine other novels regarding this issue. Earlier, I pointed out that according to the present data, it is not simply speakers' social categories (e.g. gender and class) or relationships (e.g. friendship), but also other contextual features, such as the speaker's feelings of friendliness or indebtedness, that may influence the translator's choice of forms. This is illustrated by Beaupre's shift of speech style when she left a phone message to Ballard, thanking Ballard for having saved her life. Except for the reference aitsu 'that guy' for the culprit, Beaupre did not use any masculine forms and instead used feminine forms three times (i.e. kurikaeru no yo 'they repeated'; sō itta no 'said so'; and shitteru desho 'you know it, don't you?'), which make her speech sound much gentler and politer, indexing the stance of gratitude toward Ballard.

Construction of Contrastive Personae of Female Characters
According to the foregoing analysis, the translator's choice of variants indicates that it was culturally filtered, or mediated by cultural ideologies of gender in the target language. However, this mediation took place in a skewed way in that normatively gendered forms were not straightforwardly applied to all male or all female characters. In particular, the female characters were divided into two: those who used normative forms and those who used non-normative forms, as shown in Table 7. To see this differentiation further, I present two contrastive examples in (15) and (16) and consider not only the use of sentencefinal forms of three types of sentences, but also the use of other forms, such as expletives and personal pronouns.  Ōmajime yo 19 Get the fuck out. 20 Totto to orite.
In the English original in (15), Ballard uses expletives five times (i.e. fucking, ass, shit, two instances of fuck). Table 8 shows the English sentences in which the five expletives were used and their Japanese translations. None of the Japanese translations in Table 8 includes an expletive. Moreover, Ballard's utterances in Japanese are much less forceful, compared to the corresponding English utterances. For example, her challenging question Are you fucking kidding me? in line 1 is translated as Fuzakete iru no? 'Are you joking?' (line 3) without an expletive and with the feminine final particle no, which In (16), except the first-person reference term atashi 'I' (line 14), Beaupre uses masculine forms consistently: i.e. the question particles dai (line 2) and kai (line 23), the auxiliary verb n da for a statement (line 15), the auxiliary verb da followed by the particle yo for a declarative sentence (line 24), the thirdperson reference aitsu 'that guy' (line 24), and the expletive kuso 'shit' (line 10). Interestingly, Beaupre's Fuck is translated into Kuso 'Shit' , while Ballard's Are you fucking kidding me in line 1 in (14) is translated into Fuzakete iru no? 'Are you kidding?' without an expletive and ending with the feminine particle no. Ballard's speech, on the other hand, is quite polite and formal, using referent (itadak in line 19) and addressee honorifics (desu in lines 6 and 12 and mas in lines 8 and 19) The contrast in the choice of gendered variants in (14) and (15) suggests the translator's belief about the language use of women who differ socially in regard to class and age -a belief about Japanese linguacultural stereotypes. Note also that in the translated version of LS, Ballard did use a few expletives, including three instances in which she was alone and talking to herself and three instances uttered in Japanese English. For example, when Ballard is confronting Trent, a dangerous suspect, she uses, Fakku yū, Torento, a transliterated English phrase, which the translator may have thought less direct than Kuso, which he assigned to Beaupre. But Ballard's Fakku yū co-occurs with sentences with feminine and gentle forms (i.e. omotteru no? 'do you think?'; Kore wa oshimai yo 'This is the end').
The contrast of (15) and (16) suggests that the translator's choice of variant forms is mediated by linguacultural ideologies, or ideologically based stereotypical views, about women in different and stratified social groups. The differential representations of female characters' language use seen in this section in effect contribute to reinforcing sociolinguistic stereotypes.

Conclusion
This study illustrates the fact that the repertoires of linguistic forms that index sociopragmatic meanings (e.g. politeness, femininity) differ widely from language to language. It examined Japanese translations of dialogs in English novels, focusing on the use of gendered sentence-final morphological forms in Japanese that are non-existent in English. The results of the analyses and their theoretical implications are summarized as follows.
-While the assignment of gendered sentence-final forms in the Japanese translations to characters in the two novels was normatively gendered to a large extent, there were substantial non-normative assignments, especially to some of the female characters. In other words, although those variants are regarded (in Japanese sociolinguistics) as 'feminine' and 'masculine' forms, their distribution in the translations indicates that it is not simply gender but the interaction of gender with other social aspects of characters, in particular social class and age, that affected the translator's choice of variants. That is, a woman is not just a woman but always some kind of woman (e.g. young, working class). And the fact that 'masculine' forms were assigned to women in certain social groups implies that the same linguistic form as a variant (e.g. masculine forms) may be linked to, or may stratify, with multiple social categories or social types, as claimed by Eckert (2008). -As noted earlier, language use in real social life often diverges from linguacultural norms such that working-class women may use feminine forms, while middle-class women may use masculine forms. In other words, the gender and other social categories associated with a speaker jointly affect variant choice but do not determine it. This may be due to speakers' ideological differences, or the different attitudes or stances they hold about language use that mediate their variant choice, as well as to the differences in details of contextual features (see Okamoto and Shibamoto Smith 2016 for further discussion of this point). The translations examined here occasionally included such diverse uses (e.g. use of feminine variants by male characters to indicate friendliness), but media representations tend to efface the complex diversity that exists in real language practice and instead rely on sociolinguistic stereotypes as 'mediatized language' (Agha 2011). The translations examined here are no exception. The distribution of gendered variants we saw was largely normative, which suggests the effects of a cultural filter pertaining to the Japanese norms of gender and language use, which informs the translator affecting his mind-set. Furthermore, the differential assignment of gendered forms to female characters in different social groups also suggests the effect of a cultural filter pertaining to the stereotypical language use of women in different social categories, which informs the translator. -Thus, the cultural stereotypes presented in the translations are highly ideological. Linguistic forms as variants indirectly index certain social types/categories through the process of linking them to certain stances and qualities associated with those linguistic variants. For example, sentence-final forms such as Aitsu da yo 'He is the one' and Yondo kure 'Call me' index a direct and forceful stance, which is ideologically assigned to a working-class older female character (Beaupre), while forms such as Ōmajime yo 'I'm really serious' and Totto to orite 'Get off quickly' take a mild and gentle stance, which is in turn ideologically assigned to a middle-class professional female Okamoto 10.1163/26660393-bja10031 | Contrastive PragmaticS (2021) 1-28 character (Ballard). Analyzing translations in this way can help understand the ideological basis of indirect indexing. The two English novels examined in this study were both translated by Furusawa Yoshimichi, a male native speaker of Japanese, who specializes in the translation of mystery novels. BS and LS were translated when he was 60 and 62 years old, respectively. If the translator's voice is largely based on his ideologically skewed beliefs about the language use of different social groups, it is likely that different translators may translate the same expressions in the original language differently. Although previous studies of Japanese translations of novels and other forms of fiction have found that sociolinguistic stereotypes are recycled time and again, it would be interesting to examine different translators' work to see if there are other ideologically diverging choices of variant forms.