Our Language — Linguistic Ideologies and Japanese Dialect Use in L 1 / L 2 Interaction 1

This study uses conversation data and ethnographic interviews to examine the role of meta-talk in speaker legitimacy for L2 Japanese speakers. Autoethnographic analysis of conversation data demonstrates how an L2 speaker is co-constructed (jointly positioned) as a (non)legitimate speaker of Japanese Dialect. The researcher, an L2 Japanese speaker, recorded Japanese conversations with L1 interlocutors, namely, her L1 Japanese spouse and in-laws. Two contrasting cases of L2 Japanese Dialect use are examined. In the first case, L1 interlocutors respond to the L2 speaker’s dialect with meta-talk about “our language,” co-constructing the L2 speaker as a non-legitimate dialect user. In the second case, the L2 speaker’s dialect use is affirmed when the L1 interlocutor uses similar dialect; no meta-talk occurs. The conversation data is supplemented with ethnographic interview data which underscores the prevalence of meta-talk. Meta-talk reveals speakers’ beliefs about legitimate speakerhood in which “our language” does not include L2 speakers. Conversely, the absence of meta-talk affirms the L2 speaker’s dialect use and depicts dialect as a shared form of “our language.” This study contributes to understanding linguistic ideologies, demonstrates how language ownership and speaker legitimacy manifest in Japanese interactions, and adds to research examining Japanese Dialect use by L2 speakers.


Introduction
First language (L1) speakers of Japanese respond to second language (L2) speakers in a variety of ways, sometimes in ways that question the legitimacy of L2 speakers. Researchers draw on notions of language ownership (e.g., Wee 2002) and legitimate speakerhood (e.g., Bourdieu 1991) to explore questions of who has the right to speak a language and in what ways. Language ownership refers to ideological ways that speakers can be positioned as having control over a language and its use; legitimate speakerhood refers more specifically to a speaker's right to speak and be heard. Recent work connects ideologies of ownership and legitimacy to examine what happens when ideologies result in us-them dichotomies that treat L2 speakers as non-legitimate speakers of the L2 (O'Rourke 2011).
Less attention has been paid to how L2 speaker legitimacy is negotiated in the context of regional dialects. Japanese Dialect remains an important and commonly used speech style despite wide-spread standardization (e.g., Carroll 2001b;Jinnouchi 2007) and its use presents challenges for L2 speakers living in smaller towns and rural areas of Japan. In addition, because of its association with specific regions, Japanese Dialect is an ideal candidate for analysis of language ownership and legitimate speakerhood with regard to Japanese L2 speakers. The use or non-use of Japanese Dialect raises important questions about L2 speaker legitimacy, which can impact interactions L2 speakers have with L1 Japanese speakers. For example, how are ideologies of legitimate speakerhood revealed in L1/L2 interactions in which Japanese Dialect is used? Can L2 speakers be legitimate speakers of Japanese Dialect? To consider these questions, I conducted an autoethnographic analysis of naturally occurring conversations in which an L2 speaker (this researcher) uses Japanese Dialect in L1/L2 interactions. 2 The conversation data show how reactions to L2 dialect use differ depending on the L1 interlocutor. My analysis demonstrates that L2 dialect use may trigger meta-talk, in other words, talk that focuses not on the topic at hand, but on languagerelated aspects, including the linguistic form, word choice, speech style, as well as the interlocutors themselves. Findings from the conversation data are supplemented with ethnographic interview data (Takeuchi 2015(Takeuchi , 2018(Takeuchi , 2019 in which L2 speakers reported experiences with meta-talk about dialect. Taken together, these two data sets demonstrate that metatalk plays a role in how an L2 speaker is co-constructed as a (non)legitimate speaker of Japanese Dialect. Meta-talk shifts the conversation away from the topic at hand and instead focuses on how language is used. Meta-talk about dialect also reveals speakers' beliefs about language ownership and notions of legitimate speakerhood in which ownership of Japanese Dialect is not extended to the L2 speaker. Conversely, the absence of meta-talk affirms the L2 speaker's dialect use and depicts dialect as a shared speech style, or "our language." Linguistic ideologies are "beliefs about language articulated by users as a rationalization or justification" (Silverstein 1979:193) for language use. 3 Blackledge and Pavlenko (2002:124) highlight the role of linguistic ideologies in "the production and reproduction of social difference" and they detail how linguistic ideologies perform a "gatekeeping" function in which linguistic ideologies "create, maintain and reinforce boundaries between people in a broad range of contexts" (Blackledge and Pavlenko 2002: 131). They also explain that "ideologies are constructed in discourse at micro and macro levels, and in institutional as well as everyday practices" (Blackledge and Pavlenko 2002:122). I would add that it is at micro levels, such as interactions between individual interlocutors, that linguistic ideologies are made visible and are recursively created and recreated through what speakers say and how they say it.
Linguistic ideologies are at the center of beliefs about which speech styles are more appropriate than others. An extension of appropriateness is the notion of speaker legitimacy, as introduced by Bourdieu (1991) and expanded upon by numerous researchers (e.g., Blommaert 1999;Liddicoat 2016;Norton 1997). At issue is the question of who is a legitimate speaker, in terms of both the right to speak and the right to be heard (Bourdieu 1991). The right to be heard goes beyond the ability to merely command a listener's attention and includes the right to have the content of one's speech attended to, as opposed to attention only to its linguistic form. Liddicoat (2016) addresses this in his discussion of legitimacy and authenticity, describing interactions in which L1 speakers comment on the form of an L2 speaker's utterance while ignoring the content of that utterance. In some cases, the L1 speaker's attempts to correct or rephrase the L2 speaker's utterance may impose a meaning based more on the L1 speaker's assumptions than on the L2 speaker's intended meaning. The result is that the L2 speaker is "not positioned as a valid creator of meanings but as an imperfect executor of the meanings the native speaker has inferred" (Liddicoat 2016:416). Such positioning demonstrates an asymmetry of power in which the right to be heard can be denied to L2 speakers.
Ideologies of legitimacy form the underpinnings of the ownership of language. Wee (2002) describes language ownership as the control speakers have over a language's use and development. Ownership of English has been studied extensively (e.g., Norton 1997, Wee 2002, Widdowson 1994, especially with regard to native and non-native teachers of English-as-a-second-language. Widdowson (1994) and others (e.g., Pennycook 2017, Wee 2002 argue that internationalization and the spread of English have expanded ownership of English. Wee (2002:285) argues that "English is now owned by all who use it internationally." However, these ideas are not well-known beyond academia and L2 speakers themselves may discount their right to have control and a sense of ownership over their L2 (e.g., Parmegiani 2014). The disconnect between scholarly advocacy for expanding language ownership, on the one hand, and the experiences and beliefs of L2 speakers "in the wild" on the other demonstrate the importance of continued work in this area.

Japanese Dialect in contemporary Japan
The discussion below follows the common practice of discussing Japanese Dialect in the aggregate as a speech style in contrast to Standard Japanese. This facilitates a comparison of ideologies of standard versus dialect, although it does so at the expense of an in-depth examination of the rich dialectal variety found in Japan. 4 Recent findings demonstrate the complex role that Japanese dialects occupy in contemporary Japan, in particular, as dialects have evolved from being highly stigmatized to locally valorized. For example, researchers report changing dialect attitudes (e.g., Watanabe and Karasawa 2013) and examine dialect as a speech style resource (e.g., Okamoto 2008aOkamoto , 2008b. Other work examines the linguistic ideologies involved in how Standard Japanese and Japanese Dialect are depicted and understood (e.g., Carroll 2001b, Heinrich 2012. Intense promotion of Standard Japanese in the first half of the twentieth century included corresponding stigmatization of Japanese Dialect (e.g., Gottlieb 2005, Heinrich 2012). However, as Standard Japanese spread, rather than replacing dialect with standard, speakers instead adopted the practice of switching between standard and dialect (e.g., Carroll 2001, Long 1996, a practice that continues today. Indeed, in an opinion survey conducted by the Japanese government to examine views about dialect and standard, almost 80% of respondents said that it was acceptable to practice tsukaiwake, i.e., to make distinctions by switching between standard and dialect depending on interlocutors and context (Agency for Culture Affairs 2010).
The widespread acceptance of tsukaiwake is particularly notable given earlier efforts to eradicate dialect, for example through shaming and punishment when schoolchildren used dialects (e.g., Ramsey 2004). After World War II and in particular starting in the late 1970s, opinion of dialect has gradually become more positive. Evidence of this can be seen in surveys reporting positive views of dialect's emotional qualities and warmth (e.g., Carroll 2001a, Okamoto 2008b. For example, some dialect users report "affection for their local dialect" (Occhi 2008:108). Other studies have found that speakers describe dialect as being warmer and more friendly than Standard Japanese (e.g., F. Inoue 2000, Watanabe andKarasawa 2013).
Because dialects are used in specific regions, it may seem obvious to say that dialects are closely associated with regionality. However, this association can transcend actual residence, as when dialect is used to create and maintain a symbolic hometown. This is possible because of dialect's association with a "local community of people with a shared history; [dialects] provide a metaphor for a sense of community" (Carroll 2001b:14). Although linguists have found that dialect serves a variety of social and interactional functions, a common belief is that the purpose of dialect is to index its speakers as being from a particular region (e.g., Okumura 2016). Thus, dialect is connected to a sense of hometown, and the choice to use dialect may be popularly explained as being due to interlocutors' knowledge (or assumption) of shared origins (cf. Kobayashi 2007).
Given dialect's connection to hometown nostalgia, it is not surprising that researchers have found that dialect functions as a resource for solidarity and belonging (e.g., Ball 2004, Kobayashi 2004, SturtzSreetharan 2006, Sunaoshi 2004. For example, Ohuchi (2014) examined dialect used in radio programs after the Great East Japan Earthquake of 2011 and found that dialect "displayed a sense of belonging and feelings of hometown" (Ohuchi 2014:15), functions which were felt to be of great importance for people displaced by the earthquake and tsunami. Dialect's function as a symbolic representation of hometown is also seen in dialects used in television dramas and commercials (e.g., F. Inoue 2000). The visibility of dialects in media may be partly responsible for a "dialect boom" (e.g., Jinnouchi 2007, Rodriguez 2018. Researchers report trends in young people's use of dialect as a resource for identity construction (e.g., Kobayashi 2004Kobayashi , 2007Tanaka 2011). More recent research details how television and internet contribute to young people's exposure to dialect (Okumura 2016) and provide a venue for use and discussion of dialect (Rodriguez 2018). These findings demonstrate that dialect remains a relevant and highly salient aspect of spoken Japanese.
At the same time, dialect can also be a contested code choice, perhaps because stigmatization from earlier eras has not been completely eradicated (e.g., Kubota 2014). While some dialects have been found to have significant prestige, especially Kansai dialects (e.g., M. Inoue 2006, Shibamoto Smith andOcchi 2009), others, such as Tohoku dialects, continue to be stigmatized (e.g., Kumagai 2011, Miyake 1995, Sunaoshi 2004. Issues of prestige emerge in how speakers engage in so-called codeswitching between standard and dialect. For example, Occhi (2008) described how dialect users attended closely to switching between standard and dialect. One participant described the effort of switching as "stressful" (2008:101); others were described as using a kind of "passing" behavior in Tokyo when they avoided dialect use to conceal their regional origins (2008:102). A similar account is found in M. Inoue's (2006) depiction of a speaker from the Tohoku region who, after relocating to Tokyo, made conscious efforts to rid her speech of dialectal elements, adopting highly feminine, standardized Japanese. Inoue concludes that such efforts reflect a desire to emulate an "idealized image of the urban middle-class woman" (2006:270). Conversely, Inoue described speakers who moved to Tokyo from the Kansai region but continued using Kansai Dialect; Inoue argues that these different approaches represent the differing prestige of the two dialect groups.
Inoue's depictions reveal the tension between dialect use and ideological notions of what counts as feminine speech. The rejection of Tohoku Dialect is closely linked to an acceptance of linguistic norms for feminine speech, which are patterned on stereotypes of middle-class Tokyo housewives' speech. Similarly, Shibamoto Smith and Occhi (2009) examine fictional representations of female speech and find that in romantic stories, female leads are given standard, stereotypically feminine speech with limited dialect elements. Prestige dialects (e.g., from Kansai) are incorporated in limited ways in the speech of heroines, but full use of a low-prestige dialect is not included in a heroine's dialogue. These findings point to the complex interplay between gendered language norms and dialect and underscore the need for more research to consider the degree to which certain dialects may be "off-limits" for female speakers.
The above studies demonstrate that dialect remains an integral part of spoken Japanese. At the same time, its use is complex and linguistic ideologies of appropriateness and legitimacy must be included in the discussion of dialect. Further, it is likely that speakers, consciously or otherwise, regard dialect as a speech style with limited "deployability"in other words, it is widely recognized that one cannot simply use dialect any time or with everyone. Rather, dialect's use is limited to specific interactions; for example, with interlocutors from the same hometown. Additional factors, such as a dialect's relative prestige, or the speakers' gender, interact in complex ways to impact choices between Standard Japanese or Japanese Dialect. These studies offer important background for understanding dialect's complex role within the larger repertoire of Japanese speech styles. However, missing from the discussion is an examination of language ownership with regard to dialect speakers.

Japanese Dialect and L2 speakers
Compared to dialect research focused on L1 speakers, fewer studies address the topic of Japanese Dialect and L2 speakers. However, it is clear that L2 speakers encounter dialect in Japan, need to negotiate its understanding, and must make decisions about whether to use it themselves (e.g., Itakura 2008, Mukai 2000, Siegal 1994). For example, Iino (2006) reports that American exchange students in Japan were exposed to dialect use during home stays. Mori (2012) argues that exchange students in Japan need to be able to understand a variety of speech styles, including dialect, based on her finding that host families in Japan mixed standard and dialect speech. Mori also described the efforts of one L2 participant to avoid adopting the dialect used by her Japanese co-workers. This author examined the dialect-related beliefs of L2 speakers in Ehime (Takeuchi 2015). All L2 speakers in my study reported encountering dialect and many reported making conscious efforts to understand it. Participants' beliefs about the importance of dialect use for L2 speakers varied widely and some participants chose to use dialect while others made a conscious effort to avoid it. These studies confirm that L2 speakers are exposed to dialect in Japan and demonstrate various ways that they negotiate dialect-related speech-style choices. However, previous studies do not consider issues of language ownership or legitimate speakerhood of Japanese Dialect in the context of L1/L2 interaction. The present study examines how language ownership of Japanese Dialect emerges in meta-talk about dialect and considers the impact of such metatalk with regard to L2 speaker legitimacy.

Methodology: Autoethnographic Conversation Data and Ethnographic Interview Data
This study makes use of two data sets: conversation data from recordings of naturally occurring interactions and ethnographic interview data. Because I am a participant in the conversation data, I use "analytical autoethnography" (Anderson 2006) to examine Japanese Dialect in conversations in which I am a participant. I then consider those findings in conjunction with ethnographic interview data to gain insight into linguistic ideologies about L2 speakers.
Analytic autoethnography refers to "research in which the researcher is (1) a full member in the research group or setting, (2) visible as such a member in published texts, and (3) committed to developing theoretical understandings of broader social phenomena" (Anderson, 2006:373). One benefit of autoethnographic data is that the researcher's status as a member makes it possible to draw on greater knowledge of the social context and offers access to data that might otherwise be difficult to obtain (Anderson 2006). Because the central focus of my research is L2 speakers of Japanese, and I am an L2 speaker of Japanese, using myself as a unit of analysis is a natural choice. At the same time, all methodologies have strengths and limitations, and autoethnography is no exception. One limitation is the researcher's subjectivity; another limitation is the small amount of data because the researcher is the focal participant. I sought to mitigate these concerns in the following ways. First, unlike some forms of autoethnography, I do not rely on my own recall of language use. Instead, I use audio recordings and transcriptions of the conversations, allowing me to analyze what was actually said. Second, I incorporate findings from analysis of interview data to supplement the autoethnographic conversation data. Interview data, described below, come from an ethnographic interview study I conducted with L2 Japanese speakers who are long-term residents of Japan (Takeuchi 2015(Takeuchi , 2018(Takeuchi , 2019. Although generalization is not a primary goal of qualitative research, the interview data provide triangulation (cf. Duff 2008) and reinforce the findings from the autoethnographic data.

Participants and Settings
Participants in the conversation data set are my husband, his parents, and myself. My husband is an L1 Japanese speaker born and raised in Ehime, Japan. My father-in-law and mother-in-law are also L1 Japanese speakers who were born and have lived their entire lives in Ehime. 5 I am an L2 Japanese speaker and L1 English speaker. I lived in Ehime for over ten years.
The ethnographic interview data set comes from a larger study conducted during fieldwork in Ehime, Japan. Participants in this data set were twenty-five Japanese L2 speakers and twenty-seven Japanese L1 speakers (Takeuchi 2015(Takeuchi , 2018(Takeuchi , 2019. For this article, I focus on findings from interviews with L2 participants who lived in Ehime. The L2 participants are all L1 English speakers living and working in Japan longterm, some for ten years or more. All have intermediate or advanced Japanese ability (based on self-reports, experience with the Japanese Language Proficiency Test, and my observations). One-on-one qualitative, semi-structured interviews were conducted with L2 participants in English and transcripts of the interviews were used for analysis.
Ehime is often viewed as a rural prefecture because of its location and population. While Matsuyama, Ehime's capital, has all the services, facilities and amenities of a regional city (chihō-toshi), the rest of the prefecture is moderately to very rural. The participants included in the present study lived in smaller cities or towns, areas where dialect use remains common (Takeuchi 2015).

Data and Method of Analysis
Autoethnographic data in this study consist of recorded conversations from two separate interactions. The first recording was made in Japan during dinnertime; Husband, Father-in-law, Mother-in-law, and I are present in the recording. The recording is approximately one hour long. The second recording was made in the United State, also during dinnertime; Husband and I are the participants. The recording is approximately forty-five minutes long. It should be noted that participants were not prompted prior to recording. Language use of all participants (myself included) was naturally occurring and unscripted. Excerpts with meta-talk and dialect were selected for examination.
Analysis of autoethnographic data was based on interactional sociolinguistics (e.g., Gordon and Kraut 2017, Gumperz 2015, Rampton 2010, which combines microanalysis of conversation (Erickson 1992) with ethnographic observation (Kasper and Omori 2010). While numerous approaches to qualitative analysis of conversation exist, interactional sociolinguistics focuses both on communicative practices and on the content of talk, including lexical items as well as prosodic and paralinguistic features. The original recordings were used along with transcriptions to attend to these features within the context of the conversational focus.
Analysis of interview data made use of content and thematic analysis (Saldaña 2013). Interviews lasted one to three hours and focused on participants' daily life, work-related experiences, beliefs about Japanese language, and experiences with and opinions about learning and using Japanese. The interview protocol was designed to examine participants' views about Japanese speech styles in general, and about dialect in particular. As I was not originally focused on meta-talk, there were no questions that asked directly about meta-talk. However, in the course of reporting their language-related experiences, participants often described meta-talk exchanges. For this article, I examined meta-talk episodes that were specifically dialect related.

Dialect in the Data Sets
Dialect found in both data sets include lexical items, sentence-final particles, dialectal verb conjugations and verb suffixes. Although there are dialectal variations within Ehime, for convenience sake, this discussion will refer to Ehime Dialect in the aggregate. Table 1 presents some examples of dialect that appear in the data. 6 It is notable that the most common examples of dialect use occurred in sentence-final expressions and verb inflections. During fieldwork, I found that it was not uncommon to hear features from Standard Japanese and dialect together in the same utterance, recalling research on dialect and standard being used together as a mixed code (cf. Okamoto 2008a). In what may be a byproduct of increasing standardization, the use of lexical dialect items was less common, while use of dialect in phonological features, verb inflections, and sentence-final particles was more common.
I also noted that these sentence-final and inflectional dialect tokens were often treated by participants as salient, for example, when speech that included only one or two dialect tokens was nevertheless characterized as "dialect" despite being primarily standard. Drawing from my interview findings with L1 speakers and from studying Ehime's linguistic landscape during fieldwork, I concluded that Ehime Dialect is viewed positively by its residents, who tend to see the local dialect as a marker for friendliness and hometown nostalgia (Takeuchi 2015). Most L1 interviewees reported positive associations with the local dialect, describing it as a resource for expressing warmth and familiarity. Some also described dialect's importance as local cultural heritage and used dialect to demonstrate their connection to the region. Conversely, L1 participants also discussed times when dialect was not appropriate; some described discomfort that might arise for those unfamiliar with the dialect, and a few discussed the complex relationship between dialect and Standard Japanese. Comments that dialect needs to be preserved also suggest the degree to which standardization is increasing. In short, local perception of Ehime Dialect was largely positive and although there was attention to the complexities involved in choosing between dialect or standard, the strong stigma that has been found with regard to other dialects was not observed during my fieldwork.

Findings
In this section, I present findings that show how meta-talk impacts interaction and, conversely, what happens when meta-talk does not occur.
I then introduce interview accounts of meta-talk to examine its prevalence and salience. These examples demonstrate the impact that meta-talk has on L1/L2 interaction.

Meta-Talk about Dialect
In the first excerpt, the four participants are having dinner at the home of Mother-in-law and Father-in-law. Prior to the excerpt, Husband (H), Mother-in-law (M) and Father-in-law (F) used a mix of dialect and Standard Japanese. I (the researcher, R) used only Standard Japanese up until this excerpt, which begins about 25 minutes into the recording. In response to Husband's comment about upcoming plans, I made a teasing comment using a dialect ending (line 3 below). My use of dialect was followed by meta-talk discussion (starting in line 5) about my dialect use and about my knowledge and use of dialect in general. It is important to note that because meta-talk is not related to the original topic, it takes attention away from the topic at hand. In addition, the use of meta-talk facilitates observation of participants' assumptions about dialect use and dialect users. Through meta-talk, participants display, first, their assumptions about who is expected to use dialect, and second, their understandings about when, how, and with whom dialect should be used.

H: Aikidō no (1.5) anō
there's an Aikido, um In lines1, 2, and 4, Husband tells Mother-in-law about an upcoming day on which he will not need dinner. This was relevant because we were staying with my in-laws while visiting Japan, and early in the visit, Mother-in-law said she wanted to know our schedule to plan meals accordingly. In line 3, I say daibu saki yan (that's really a ways off) ending with the dialect sentence-final particle yan. Mother-in-law does not respond to Husband's comment, but instead addresses me in line 5, repeating what I said. This is followed by Mother-in-law's and my laughter, and then, in line 7, Mother-in-law says kotchi no kotoba ga deta (our side's language came out). Her use of the expression kotchi no kotoba is notable: kotchi refers to "this side," or in this context, her side rather than mine. Her comment labels my use of the dialect token yan as "this side's language," or "our language." Further, although Mother-in-law does not say something like "you shouldn't use dialect," by saying deta (came out, intransitive verb) rather than repeating the word "say," Mother-in-law implies that my dialect use was unintentional, as well as unexpected or perhaps even inappropriate. This is followed in line 9 by me saying barechatta, "busted" or "you caught me," which functions as a way of aligning with Mother-in-law's use of deta and her characterization of my dialect use. Mother-in-law's meta-talk caused the topic to shift, focusing on my dialect use and questions such as what dialect tokens I understand and whether or how much I use dialect. This shift produced an extended metatalk sequence, presented below, in which numerous dialect tokens were discussed.
(2) Data excerpt 1-2: Do you say gaina? In line 10, Husband confirms that I use dialect, even when we are in the United States. The verb form he uses, deyoru, is the dialect form similar to the standard deteiru and is used to express present-progressive tense or habitual actions. By using this form, Husband stresses that I use dialect regularly or habitually. This can be seen as responding to Mother-in-law's surprise at my dialect use, in particular because he says "yappa," "as expected," thereby depicting my using dialect as something that happens as a matter of course. In line 11, I confirm Husband's claim about my dialect use, using the same verb conjugation and adding the sentence-final particle yo for emphasis, stressing that my dialect use is habitual. This is followed by a question from Father-in-law in lines 12 and 14, confirming Father-in-law's belief that I am accustomed to hearing dialect. His utterance reveals several implicit assumptions. First, he asks about iwakan, feeling of strangeness, displaying his assumption that dialect sounds strange to someone unfamiliar with it. This idea of iwakan hints at the privileged status of Standard Japanese (cf. Kubota 2014) but does not go so far as to depict the dialect in more negative ways. Next, in saying mō (by now, anymore) and iwakan ga nai natta (standard: iwakan ga naku natta "become no longer sounding strange"), Father-in-law displays his assumption that in the past, I must have felt dialect sounded strange, though I probably no longer do, because, as he says in line 14, I am used to hearing it. This comment also displays his assumption that I hear dialect at least often enough to become used to it-in other words, that dialect is commonplace. Next in line 16, Husband offers a common example of dialect, oru ("be" verb) and explains that I understand it; this can be heard as Husband's assertion of my familiarity with dialect. At this point, Fatherin-law starts to introduce representational examples of dialect, asking if I use them. The first example appears in line 18, when Father-in-law asks if I use the word gaina (really), to which I reply that I do not. The conversation continues on this topic, with Father-in-law presenting a dialect word and Husband and I discussing whether or not I use it (because of similarity, I omit the rest of the conversation for length considerations). This exchange co-constructs Father-in-law and Husband as dialect-experts who possess the requisite knowledge to introduce examples of dialect. Similarly, it co-constructs me as a dialect-novice, as someone less expected to use or know about dialect than the other members of the interaction. (See Hosoda 2006 for in-depth discussion of how expert/novice identities are co-constructed in L1/L2 interaction.) As seen in line 20, for most of the lexical examples presented, I responded that I do not use those items, further constructing myself as "not really a dialect user." Returning to the beginning of the excerpt, we can consider possible reasons why Mother-in-law commented about my dialect use. First, my utterance daibu saki yan can be seen as teasing Husband, and it is possible Mother-in-law wanted to shift the focus away from teasing to avoid potential conflict. Second, Mother-in-law may view dialect as less than appropriate for me because of gendered norms about how female nonnative speakers "should" speak Japanese (cf. M. Inoue 2006). Third, while the dialect token I used, yan, might be seen as a "weak" dialect token (especially compared with lexical dialect tokens), it nevertheless contrasted with the rest of my speech, which was primarily Standard Japanese. Thus, it is likely Mother-in-law had expectations about how I speak Japanese. Based on those expectations and on how I had been speaking up until that point, she reacted to what she saw as an unexpected difference when I used dialect. These factors likely combined to play a role in Mother-in-law's reaction. Regardless of the reason, my use of dialect triggered her response in the form of meta-talk about dialect, thereby altering the course of the conversation, as reviewed above.
It is also notable that Mother-in-law says kotchi no kotoba, rather than (city-name)-ben, a common way to refer to the local dialect and one that Father-in-law uses in line 12. Labeling the dialect as kotchi no kotoba demonstrates a sense of ownership (cf. Wee 2002). While we might speculate about the motivation for Mother-in-law's remarking on my dialect use, it was dialect that was available as a candidate for her commentary, which took the form of a claim of ownership. Hence, what stands out here is Mother-in-law's description of dialect as kotchi no kotoba, "our side's language" or "our language." Finally, my response, barechatta "busted," provided uptake of Mother-in-law's comment and, crucially, can be heard as admission that my dialect use was somehow unexpected and possibly inappropriate.
The meta-talk that unfolded after Mother-in-law's comment depicted my dialect use as unexpected and positioned me as a dialect novice in comparison to the expert status assumed by Husband and Father-in-law. Examination of the meta-talk excerpt reveals four factors that contribute to the co-construction of me as someone not expected to use dialect: (1) Mother-in-law comments explicitly on my dialect use, (2) she labels the words I used as "our language" (this side/her side, i.e., not my side, not my language), (3) I ratify her claim by saying "busted," and (4) Father-inlaw and Husband engage in an account-taking of my knowledge and use of dialect in an extended meta-talk sequence. What is at issue here is how utterances come to be heard or not heard. Meta-talk has the effect of ignoring the content of the utterance and focusing on its linguistic form (cf. Liddicoat 2016). As a result, the utterance fails to be heard, denying the speaker's legitimacy (cf. Bourdieu 1991). In the above example, both Husband's announcement of upcoming plans and my comment about the announcement were lost in the meta-talk discussion. Meta-talk allows L1 speakers to exert their ownership of the language (in this case, of the dialect), making them qualified to comment on the linguistic form of an L2 speaker's utterance. As this discussion demonstrates, meta-talk is one example of what can happen when an L2 speaker uses dialect. However, L2 dialect use does not always trigger meta-talk, as can be seen in the data presented below.

Reciprocal Dialect Use
Unlike the excerpt above, the next excerpt presents an example of L2 dialect use without meta-talk. This segment is from a recording made during dinnertime at our home in the U.S.; Husband and I are the only participants. There is a mix of Standard Japanese and dialect throughout the recording, then for approximately four minutes prior to the segment below, both Husband and I used Standard Japanese. At the point where the transcript begins, I use dialect and my dialect use is followed by similar dialect use by Husband. After this exchange, the conversation continued without a significant shift in topic.
Prior to this segment, Husband and I discussed seeing each other on a crowded bus. Because Husband was at the back of the bus when I got on, we could not sit together and did not speak. I waved to Husband before getting off the bus. The excerpt is our discussion about that interaction.
(2) Excerpt 2: What is she doing? waving at," with dialect tokens at the end of his utterance. Husband's utterance aligns with mine in two ways: First, he matches my humorous tone in imagining the reaction of people around us. Second, Husband's dialect in line 5 is similar to mine in line 4 because both use dialect verb inflection and dialect sentence-final particles. After this exchange, we laugh, and the talk continues and gradually shifts to other topics. Unlike the segment introduced in Excerpt 1, in this exchange, dialect is not explicitly attended to. Crucially, my dialect use is not problematized as in the previous excerpt. As a result, my dialect use did not trigger metatalk, and instead, was followed by similar dialect use by Husband. His dialect use is an implicit affirmation of my dialect use, and depicts my dialect use as ordinary and not noteworthy. We might speculate about the reasons for the lack of problematization: in particular, it is likely that Husband is accustomed to my using dialect, especially in light of his comments in Excerpt 1 about my habitual dialect use. Regardless, Husband's lack of reaction to my dialect use, along with his reciprocal dialect use, has the effect of affirming my dialect use and depicting dialect as a shared repertoire.

Interview Accounts of Meta-Talk
The above excerpts present two examples of what may happen when an L2 speaker uses Japanese Dialect in conversation with an L1 speaker. In this section, I introduce three accounts of meta-talk exchanges from ethnographic interview data. When the autoethnographic data introduced above are considered in conjunction with these accounts, it is clear that meta-talk exchanges are not only a common experience for many L2 speakers, but also can have negative impacts on them. As explained above, I conducted interviews with L2 speakers who were long-term residents of Japan. In their interviews, almost all of the L2 participants reported having meta-talk discussions with L1 speakers; these discussions tended to focus on L2 participants' Japanese language use and abilities. For the purposes of this study, I examined only accounts of meta-talk which were specifically dialect related. Meta-talk exchanges were sometimes viewed positively, especially when they presented opportunities to learn more about Japanese. However, meta-talk exchanges were unwelcomed when they were seen as derailing the conversation.
One participant, Grace, had been living in Japan for ten years at the time of the study. 8 Grace spoke Japanese with coworkers, in her community, and also at home with her L1 Japanese spouse. Grace explained that she began learning Japanese only after moving to Ehime, and, as a result, she used numerous dialect features in her everyday speech. Grace said that sometimes when she used dialect, L1 speakers laughed at her and commented on her dialect use. She described this as "annoying" because it detracted from the conversation. On the other hand, when the L1 interlocutor did not respond to her dialect use with meta-talk, Grace found dialect to be useful for expressing and creating friendliness. Grace explained that this was why she continued to use dialect despite it sometimes resulting in unwelcomed meta-talk.
Melissa had been living in Ehime for three years, where she worked as an assistant English teacher. During her interview, she shared two experiences of using dialect expressions that she learned by hearing them used in her workplace. Similar to Grace's experiences, Melissa said that she received contradictory reactions from different L1 speakers. In one instance, when she used a dialect word, an L1 coworker encouraged her to continue using it because it would make the schoolchildren she taught feel closer to her. In a separate exchange with a different L1 speaker, after Melissa used a dialect word, the L1 speaker told her that she should use the Standard Japanese version instead, because standard was "more cute" for her. Reflecting on these experiences, Melissa noted that although she was interested in using dialect, she had decided to avoid it because L1 speakers often reacted to her dialect use rather than to the actual content of her words, which caused the conversation to "branch off." Liam had also been living in Ehime for three years. Liam described experiences with meta-talk but unlike Grace or Melissa, he was not troubled by these exchanges. Liam said he made a conscious effort to use dialect, partly because that was how the L1 speakers around him spoke. He reported that L1 speakers sometimes reacted with surprise or laughter to what they perceived as a disconnect between his status as a foreigner and his use of dialect. However, unlike other participants, Liam seemed to relish this type of reaction, interpreting it as positive attention.
The above examples demonstrate some ways that L2 speakers encounter meta-talk in response to dialect use. Although there are no recorded data of exchanges participants reported, their experiences were salient enough for them to be included in responses to questions about language-related experiences. Further, both Grace and Melissa consciously made decisions about their language use as a result of metatalk encounters. These accounts also offer insight into how language ownership and speaker legitimacy emerge discursively in conversation. In Melissa's depictions, we see L1 speakers giving her unsolicited advice about whether or not to use dialect, which conveys that Melissa is someone not fully in control of her linguistic choices. Liam's depiction presents the most explicit example of meta-talk functioning as a display of language ownership, in that the L1 speaker treated Liam's dialect use as an opportunity to highlight Liam's status as a foreigner. In each of these accounts, L1 speakers reveal their ownership of dialect.
These depictions of meta-talk exchanges come from self-reports and are based on recall rather than on recordings of actual interaction. Nevertheless, it is clear that meta-talk exchanges about Japanese Dialect use were common for these participants. More important than a quantitative account of the frequency of meta-talk exchanges, however, is the salience that such exchanges held for participants. For Grace, Melissa, Liam, and me, meta-talk shifted the focus away from the topic at hand and onto the language use of the speaker (cf. Liddicoat 2016).

Discussion
With the above findings in mind, I return to the questions asked at the beginning this paper. The first question considered how language ownership and speaker legitimacy are revealed in L1/L2 interaction. One example can be seen in Mother-in-law's meta-talk in Excerpt 1. First, she explicitly remarks on my dialect use, repeating my words and marking them as noteworthy, which implicitly calls into question the legitimacy of my dialect use. Second, she claims ownership of the dialect by labeling the words I used as kotchi no kotoba or her side's language. Next, my response, barechatta (busted), ratifies Mother-in-law's claim of language ownership of dialect, thereby completing co-construction. Further, the meta-talk which followed marks dialect as distinct from Standard Japanese and delineates it as a candidate for ownership. We see a similar process in the interview data. Namely, L1 speakers commented on L2 speakers' dialect use, gave advice about whether or how L2 speakers should use dialect, and, in the case of Liam, treated dialect use as an opportunity to highlight the L2 speaker's foreignness. These actions allow L1 speakers to claim ownership of dialect and give advice about its use. At the same time, by ignoring the content of speech and focusing only on dialect use, meta-talk denies L2 speaker legitimacy, with the result that L2 speakers lose opportunities to have the content of their speech attended to. In comparison, when there is an absence of meta-talk and two speakers use dialect in similar ways, dialect is depicted as a shared form of "our language." This was seen in the reciprocal use of dialect in the second autoethnographic data excerpt and in Grace's example of using dialect as a resource for friendliness.
The second question asked whether an L2 speaker can be a legitimate speaker of dialect. The autoethnographic data show that co-construction by interlocutors can result in the creation of the L2 speaker as either a legitimate or non-legitimate dialect speaker. When dialect use is explicitly attended to and becomes the topic of discussion, the resulting meta-talk foregrounds the speaker's dialect use as unusual, unexpected, and worthyof-note (cf. Liddicoat 2016). Uptake from the L2 speaker contributes to co-construction, such as that seen in Excerpt 1. Together these actions prevent the L2 speaker from being positioned as a legitimate speaker of dialect. While we cannot do the same kind of analysis with the interview data, it seems likely that a similar process occurred based on participants' accounts. In contrast, when meta-talk about L2 speakers' dialect use is absent, as seen in the second excerpt, and there is reciprocal use of dialect from the L1 interlocutor, the L2 speaker's use of dialect is affirmed. Thus, reciprocal use of dialect facilities the L2 speaker's co-construction as a legitimate dialect user.

Conclusion
Meta-talk can deny speaker legitimacy. On the other hand, the absence of meta-talk, combined with reciprocal dialect use, can extend ownership to both L1 and L2 speakers in an interaction.
Together, the autoethnographic data excerpts and the interview data demonstrate the significance of meta-talk and suggest additional questions related to L2 speakers and Japanese Dialect. A key question concerns whether Japanese Dialect is more available as a candidate for ownership than Standard Japanese. This seems likely given the regionality of dialect, and findings presented here offer additional support. However, further examination of Japanese language use in L1/L2 interaction is needed, in particular, a comparison of L2 use of standard and dialect use would offer insight into how language ownership may arise in L2 use of Standard Japanese. In addition, studies are needed comparing prestige or stigma across different dialects to determine whether the likelihood of claims of ownership differs from one dialect to another. More research is also needed to better understand how gendered norms about dialect may interact with stereotypes about how L2 speakers "should" speak Japanese.
A final question is whether "non-nativeness" is always vulnerable to meta-talk and the resulting foregrounding of a speaker's non-legitimacy (cf. Kramsch and Whiteside 2008). Future studies should examine how foregrounding occurs and consider the role of foregrounding in perpetuating ideologies of language ownership. Further, we must also consider how foregrounding impacts L2 speakers in Japan who may conclude that their Japanese use is always open to comment, assessment, or censure. A challenge for teachers of Japanese as a foreign language is how best to promote appreciation for Japanese linguistic diversity while avoiding discouragement in the face of potentially contentious linguistic ideologies.
This study used autoethnographic conversation data and ethnographic interview data to contribute to inquiry into questions of language ownership and speaker legitimacy for L2 Japanese speakers. I have mitigated the limitations of autoethnography by relying on recordings of conversation rather than on memory. In addition, I supplemented conversation data with interview data from multiple participants to reinforce the findings by presenting additional examples of similar experiences. Based on these findings, I argue that the meta-talk found in the conversation data should not be viewed as an isolated case and may, in fact, be a representative example. However, it should be stressed that the goal here is not a quantitative account. It is not the number of metatalk episodes that matters, but rather the salience that any individual episode has for the interlocutors.
Any remaining limitations notwithstanding, this study underscores the importance of considering the implications of language ownership for L2 Japanese speakers. Unlike studies of ownership in L2 English, studies of ownership in L2 Japanese remain rare. Given the increase in foreign residents in Japan, the number of L2 speakers living in dialect-using regions will surely increase as well. Therefore, we may argue that ownership of non-standard varieties of Japanese should not be limited to only those speakers originally from that dialect-region. But if ownership of dialect were thus expanded, who then should count as a legitimate speaker of Japanese Dialect? A related question is whether assertion of language ownership by L1 speakers in L1/L2 interaction is detrimental for L2 speakers. One answer can be found in the data above, where the claim of "our language" caused the topic to shift and highlighted not the content but the form of the utterance. We might call this a kind of "othering," however unintentional. Thus, one-sided language ownership has complex impacts on L2 speakers' identity negotiations. On the other hand, when language ownership is shared, concerns about legitimacy are resolved and identity negotiations, because they are not in dispute, remain in the background and interaction moves forward freely.

NOTES
1 An earlier version of this research was presented at the 2018 spring conference of the American Association of Teachers of Japanese. 2 Autoethnography refers to qualitative research in which the researcher is also a participant and is present in the data (Hughes and Pennington 2017).
3 Scholars also use the term "language ideologies" and some use these terms interchangeably. I use "linguistic ideologies" to focus on ideologies about language, its use, appropriateness, and variation, as opposed to ideologies about specific languages. 6 A detailed discussion of the linguistic features of Ehime's dialects is beyond the scope of this article. For additional information about Ehime's dialects, readers are directed to Doinaka 2005 andNHK 2005. 7 Because recording occurred during dinnertime, sounds of eating are often heard during pauses. Thus, I did not analyze pauses and believe that they do not have a significant impact on the interaction. Note that "h" signifies laughter, and brackets add extra information in the English translation. In addition, the English translations are meant to convey the tone of the original and as such do not represent a word-for-word gloss. 8 All names are pseudonyms.