“(r) You Saying Yes to the Dress?”

Abstract

This paper investigates variation in rhoticity on the reality television show Say Yes to the Dress. The study examines the speech of five bridal consultants working at Kleinfeld Bridal in Manhattan. Using the brides’ budgets as a proxy for social status, we ask whether variation in the consultants’ use of (r) correlates with the amount of money the bride states she is willing to spend on her dress, which ranges from $1,500 to unlimited. Mixed-effect logistic regression analysis shows significant differences across three budget categories, a finding that echoes Labov’s original department store study as well as later replications. We discuss our findings within the frame of an audience design approach to style-shifting and the reality television genre and explore how such a mediated data source can be a fruitful area for sociolinguistic research.

Keywords

rhoticity New York City English style-shifting reality television audience design social class

Introduction

Labov’s (1966) foundational study of English in New York City established the thoroughness with which social class stratification, reflected in linguistic variation, pervades the metropolitan area. He demonstrated this with a number of sociolinguistic variables, one of which was the variable presence of /r/ in the syllable coda (referred to hereafter as (r), following convention for sociolinguistic variables), in words such as start, more, and father. While non-rhoticity was once the prestige variant in New York, Boston, and New England, as it is in London, non-rhoticity over time lost its prestige in the United States and is often a stigmatized feature in those areas in which it remains (see Labov 1972; Preston & Niedzielski 2003). Accordingly, a number of studies have documented a change toward rhoticity in areas where it has been historically variable (Labov 1966; Feagin 1990; Labov, Ash & Boberg 2006; Nagy & Irwin 2010). In the most comprehensive study of rhoticity on New York’s Lower East Side since Labov’s groundbreaking work, Becker (2014a) confirms that some speakers, particularly Chinese, Jewish, and white groups, are showing progress toward rhoticity in apparent time, while African American and Puerto Rican speakers in the region are not. At the same time, non-rhotic pronunciations remain a powerful marker of local identity in some regions of the United States (Becker 2009; Schoux Casey 2013). In the current paper, we build on previous work on rhoticity to investigate the sensitivity of the variable (r) to social class differences in New York City, using the reality television show Say Yes to the Dress as the locus of our analysis.

In carrying out his survey of English in New York City, Labov devised the rapid and anonymous survey, which enabled him to collect data from speakers while they were unaware that their speech was being observed. In the well-known department store study, Labov (1972) elicited tokens of /r/ from employees in three stores of differing levels of prestige: Saks, Macy’s, and S. Klein. Having identified that the variable (r) was “extraordinarily sensitive to any measure of social or stylistic stratification,” Labov (1972:44) predicted that variability between rhoticity and non-rhoticity (referred to as [r-1] and [r-0], respectively) would correlate with the social standing of the store, as employees “borrowed” prestige from the store in which they worked. Labov (1972) notes that based on occupation alone, for example, the workers in these three stores would be classified as belonging to the same social class group (with Saks actually lower in social class than Macy’s, as the pay is lower in Saks); he instead demonstrates that employees are differentiated by the prestige of the store in which they work, even while they all belong to the same occupational group. He writes, “the working conditions of sales jobs in the three stores stratify them in the order: Saks, Macy’s, S. Klein; the prestige of the stores leads to a social evaluation of these jobs in the same order. Thus the two aspects of social stratification—differentiation and evaluation—are to be seen in the relations of the three stores and their employees” (Labov 1972:47–48). In total, Labov surveyed 264 speakers across the three stores: 68 in Saks, 125 in Macy’s, and 71 in S. Klein. The results of Labov’s study showed clear stratification of (r) across the three stores and its employees, confirming that the speech of the salespeople in each of these locations reflected the prestige of the store and its customers. Moreover, there was stratification within individual stores as well. In Macy’s, employees were differentiated by the position they held in the store (e.g., stock boys versus cashiers); in Saks, employees working on the upper floors, in which the most expensive merchandise (“high fashion”) was sold, exhibited higher rates of [r-1] than those who worked on the lower floors.

The department store study has been replicated at least two times since Labov’s initial publication (Fowler 1986; Mather 2012), with similar patterns emerging in each case. A comparison of the results from these three studies is seen in Figure 1, which shows the percentage of all [r-1] in the stores. As is evident there, the department store studies demonstrate that New Yorkers are moving toward rhoticity in real time, as each replication documents higher rates of [r-1] across the three department store categories. Becker (2009, 2014a) confirms this in her studies of rhoticity on the Lower East Side. In this paper, we do not focus on the progress toward rhoticity, either in real or apparent time. Instead, we are concerned with the finding of the department store studies that the variable (r) remains tightly connected to social class in New York City, as well as being a highly salient feature of New York City English (NYCE). As Becker (2009:644) says, “both New Yorkers and non–New Yorkers alike do identify non-rhoticity as a salient feature of NYCE, one that (in combination with other NYCE features or even alone) can index a New York persona.”

Figure 1.

Comparison of All [r-1] across the Three Department Store Studies.

Labov (1972) proposed that the differences in the use of (r) observed in the three department stores were a reflection of the social class stratification of the employees working in them—borrowed from the stores’ clientele. Bell (1984) uses the example of rhotic variation in the department store study to help motivate his theory of audience design. In his seminal paper, “Language Style as Audience Design,” which we discuss in more detail below, he argues that the department store study is a particularly good example of the “addressee taking over”; that is, the prestige of the customers in the three different stores is what motivates a difference in style with respect to (r), not the prestige of the employees themselves. Bell (1984:170) writes, “the indicators Labov used to rank the stores … are essentially reflectors of the relative status of the stores’ clienteles. The linguistic stratification of (r) across the three stores thus correlates with the social stratification of their average customers.”

Building on this treatment of the department store studies, the current paper presents an analysis of rhoticity in New York City. We use a body of data gathered from the reality television show Say Yes to the Dress, which focuses on brides-to-be shopping for wedding gowns at Kleinfeld Bridal Salon in Manhattan. While we use the department store studies as inspiration for the current work, there are several important differences between the two investigations. Whereas the previous department store studies used methodology that was rapid and anonymous, and elicited just four tokens from a great number of speakers in three different socially stratified locations, our study analyzes a great number of tokens of (r) that appear in the speech of just five bridal consultants, who are being video recorded for a television show, in a single location, with a number of interlocutors (brides) who differ in social class. Also unlike the previous department store studies, in which the employees designed their speech based on an ideal customer, in the current study, the customers that the speakers interact with have explicitly stated how much money they would be willing to spend. Thus, employees at Kleinfeld Bridal have somewhat more information on which to base a judgment about social status, beyond a customer’s mere physical presence in a store. Not only do consultants have the monetary figure of the budget in mind throughout the appointment, but they are interacting with the bride for an extended period of time—much longer than, for example, the time needed to answer a question about where an item is located in the store. In the current paper, we argue that the consultants working in the upscale bridal salon Kleinfeld and being filmed for the show Say Yes to the Dress are sensitive to the social prestige differences present among the clients with whom they work, based on the amount of money the customer is willing to spend on a wedding gown. We show that the consultants design their speech with their audience in mind, shifting toward more rhoticity/standard speech when their client is willing to spend more money. The reality show Say Yes to the Dress provides a rich source of data through which to discuss the social correlates of rhotic variation in New York City. In the following sections, we first discuss style-shifting in sociolinguistics with a particular focus on Audience Design. We then move on to the genre of reality television as a source of data for sociolinguistic research and then introduce the show Say Yes to the Dress, which served as the source of data for the study to follow. Our analysis leads to a discussion of the ways in which reality television can be utilized for addressing sociolinguistic questions.

Style-Shifting and Audience Design

The notion of vernacular speech has been a central concern in sociolinguistics, as researchers have traditionally sought to collect naturally occurring, spontaneous, un-self-conscious speech as the primary source of sociolinguistic data. Therefore, the sociolinguistic interview has been a standard way of collecting data, attempting to uncover how speakers normally talk in their natural environments (Labov 1972, 1984). Increasingly, however, sociolinguists have turned to other sources of data in analyses of variation and style. There has been growing acknowledgment that even such naturally occurring stretches of speech are also performative in nature (Rampton 1995; Bucholtz 1999, 2003; Podesva 2007) and that a focus on the self-conscious performance register itself need not be dismissed in variationist research and may in fact offer insight into the workings of style-shifting (Schilling-Estes 1998).

Style-shifting, or intraspeaker variation, has always been an important area of sociolinguistic research, in conjunction with investigations of interspeaker variation, or how language varies across different social groups. As such, several approaches to intraspeaker variation have been proposed over the years. In an early explanation of individual speaker variation, Labov (1966) proposed that speech styles become more careful as speakers move into more formal contexts. Shifting to more careful speech often entails a shift toward more use of prestige variants, such as [r-1], as opposed to [r-0]. To capture this, sociolinguistic interviews generally are designed to elicit a range of speech styles and often include experimental tasks such as discriminating between minimal pairs, during which maximal attention is paid to speech. At the other end of the spectrum, casual speech can be obtained during interviews when, for example, the interviewee is talking to a third party rather than the researcher. One of the limitations of the Attention to Speech model of style-shifting is that it is designed to account for speech styles exclusively within the context of the sociolinguistic interview (Labov 2001; also see Schilling-Estes 2002 for a useful critique of this and other models) and thus is not readily applicable to other situations that sociolinguists may be interested in pursuing.

The Audience Design approach (Bell 1984, 2001) departed from the idea that style varies according to the amount of attention paid to speech and instead maintains that style-shifting is a function of the speaker’s audience, based on Goffman’s (1981) taxonomy of audience members. This model draws heavily from Speech Accommodation Theory (Giles 1973), which proposes that speakers alter their speech in order to be more like their interlocutors. Audience Design assumes that speakers orient to different audience members and shape their speech accordingly. Rather than shifting speech styles due to the perceived formality of the context, the Audience Design approach claims that speakers shift in reaction to their audience. As Bell (2001:141–142) sees it, “style is oriented to people rather than to mechanisms or functions. Style focuses on the person. It is essentially a social thing.” The relationship between inter- and intraspeaker variation is also clearly articulated in Bell’s (1984) model, arguing that variation at the individual level derives from variation at the group level. “As is the habit of mirrors,” Bell (1984:153) writes, “the reflection is less distinct than the original: Style differentiation is less sharp than the social.” Thus, for linguistic variables that are stratified by social class, like (r), we would expect to find a narrower range of variation within individual speakers than across social class groups. We return to this below when we discuss our study in more detail.

A strong tradition of research has shown that speakers’ linguistic patterns are indeed affected by the person they are speaking to and also other third parties who, while not present, exert a substantial influence on the speaker (“referee design”; see Bell 1984, 1992). Most of this research has found that speakers accommodate to, or converge with, their audience or referee. Bell’s (1984) study of radio speech in New Zealand found that the same newscasters varied in their use of intervocalic /t/ depending on the station they were reading the news for: a local station, in which case the frequencies of a flap [ɾ] were greater, or a national station, in which case the frequencies for the standard stop [t] variant were more frequent. Rickford and McNair-Knox (1994) showed that “Foxy Boston,” a young African American woman, used higher rates of certain features of African American Vernacular English when the interviewer was also African American and lower rates of those same features when interacting with a white interviewer. In a study of referee-design in the media, Hay, Jannedy, and Mendoza-Denton (1999) documented Oprah Winfrey’s tendency to use higher rates of monophthongal /ai/ when the non-present guest she was introducing to the audience was African American. Again, while much of this research has shown accommodation to particular linguistic features of the audience/referee, disaccommodation, or divergence away from the audience, can certainly occur as well (see Giles, Coupland & Coupland 1991 for a thorough discussion). One objection to the Audience Design approach has been that it does not leave enough room for the agentive nature of style-shifting; that is, speakers can and do initiate style shifts in the absence of a change in audience (see Kiesling 1998). In Bell’s (2001) more recent reworking of the model, he does acknowledge that initiative is a greater part of the model than was originally conceived. Reality television may be a fruitful source of data for exploring the agentive element of audience design, perhaps in conjunction with speaker-design approaches to style shifting (Schilling-Estes 2002), and questions of performance and identity (e.g., Bakht 2010; Valentinsson 2010; Pardo 2013), though such a focus is beyond the scope of the current paper.

Reality TV as Sociolinguistic Data

Reality television is a genre that walks the line between fictional and nonfictional representations of life and events (Queen 2013) and, as such, does not provide the natural, spontaneous data that sociolinguists tend to rely on. However, some of the same methodological considerations are relevant in both of these contexts. The observer’s paradox acknowledges that sociolinguists wish to observe the speech of people not being observed, thereby collecting the most vernacular speech data possible. While researchers work to minimize the effect of this through methods such as rapid and anonymous surveys, sociolinguistic interviews, and participant observation, once participants are aware that they are being observed, their behavior, including their speech, may become less natural. The effects of observation may in fact be heightened on reality television. While likely not a testable hypothesis, being filmed for a broadcast that will be aired across the country and potentially around the world probably has a greater effect on participants’ awareness of their image and speech than does sitting in a quiet room with a single sociolinguistic researcher who is gathering data that, the participants have been assured, will not have any identifying information attached to it. Participants in sociolinguistic interviews are guaranteed confidentiality; participants on reality television shows have agreed to open portions of their lives up for the voyeuristic viewing pleasure of wide audiences. Thus, reality television fully acknowledges the observed nature of speech, with cameras, and a film crew, and the awareness of a future home-viewing audience. As such, intraspeaker variation within such a fully performative context is particularly interesting for the study of sociolinguistic variation: if speech in this context varies, then we can consider factors other than formality or the fact of observation for explanations of style-shifting.

Reality television shows also offer an interesting vantage point from which to study the effects of the audience on speaker style, in large part because of the structure of such programs. There are two main reasons for this. Thornborrow and Morris (2004:248) argue that participants on reality TV are orienting to a “double frame”: the immediate and local audience of the other show participants and the “mediated context of a reality TV show,” which provides a national viewing audience. In the language of audience design, those in face-to-face interactions during the taping of the show are “addressees” (known participants, who are ratified and addressed), and “auditors” (those who are known and ratified, but not directly addressed); the audience at home, on the other hand, comprises an unknown number of “overhearers” (those who are known but not ratified in the interaction). While addressees and auditors are the main focus of the interaction, there is an awareness of the overhearers at all times as well. After all, participants are connected to microphones, and camera crews follow them around while they engage in various activities, acting (like the recording equipment during a sociolinguistic interview) as constant reminders that they are being observed. Second, reality TV shows are generally composed of two parts: those in which participants interact with others during the “action” of the show (e.g., selling a designer wedding dress, baking an elaborate cake, competing in a strenuous physical challenge), and testimonials (one-on-one interviews), during which participants talk to a person or group of people (unseen and unheard by the home audience) to reflect on people and events that occurred during the taping of the show. This alternation between individuals talking to various interlocutors face to face (as part of their audience), and then talking about those same people when they are not present (as referees), all while being filmed for home viewers, thus offers an interesting avenue for examining style-shifting, particularly with respect to audience design. While participants may orient more toward the local audience while filming the events of the show and more toward the home audience during testimonials, it is possible that both audiences exert some effect on the speech of participants at all times. Previous work on reality television shows has demonstrated the linguistic consequences of simultaneous orientation to these two different audiences in terms of the speech and discourse of show participants. Thornborrow and Morris (2004), for example, examine the functioning of gossip on the reality show Big Brother. They show that gossip is a strategy used by contestants to simultaneously present themselves favorably to those watching on television (who can vote them out of the house and off the show), and to forge relationships and alliances within the house with the other contestants. While participants know that not all of the things they say will be broadcast on national television, there is the potential for any of the things that they say to be used in the final product released to audiences. Contestants on Big Brother explicitly comment on the fact that the things they do and say will be seen by the viewing audience.¹

In the current paper, we add to research on reality television by examining variation in rhoticity on TLC’s Say Yes to the Dress. Through investigation of a well-known variable with a relatively new type of data source, we revisit questions posed in the department store studies and further explore the medium of reality television as a source of data for the sociolinguistic study of style. Instead of examining social-class-based variation in (r) among speakers in different stores, we examine how the perceived prestige of clients within a single store, Kleinfeld Bridal, correlates with the variable among employees who work there. Based on previous work on (r) in New York City, as well as Bell’s model of audience design, we predict that the consultants’ use of (r) will vary in accordance with the prestige of their clients; that is, we expect consultants’ rhoticity to correlate with the amount of money their clients intend to spend on their wedding gowns.

Say Yes to the Dress

The data source for this paper was the reality television “docusoap” Say Yes to the Dress, which began airing on TLC in 2007. The show is set in the New York City bridal store Kleinfeld, located between Sixth and Seventh Avenues in Manhattan, and centers on the quest of brides to find their “dream” wedding gown. Each episode of the show features 3 to 4 brides who come into the salon to shop for their dresses or to have a fitting in the alterations department. To select their dresses, the brides often bring with them a number of close friends and family members (most often all female), to assist in the process. Each bride has a bridal consultant assigned to her, whose job it is to find her “perfect dress” among the thousands of gowns that the store sells. The managers of the store are responsible for matching up brides and consultants, an important role in setting the stage for a successful sale to occur. On occasion, the managers make it a point to tell a consultant that she has been specifically requested by the bride she is going to work with. Bride-consultant pairings may also be based on factors such as previous sales records and personality, and perhaps the higher-end clients are routinely placed with particular consultants. Because they are participating in a reality television show, it is likely that the preferences of the producers are an influential factor in these matches as well. Unfortunately, we have no information about how these matches are made that would shed light on the present study.

Once the bride and the consultant have introduced themselves, the consultant attempts to get to know the bride, gathering information that presumably will assist her in selecting the kind of dresses that the bride would be interested in purchasing. Below is an example of such an exchange between Dianne and her new client, Magdalonie.

While it is likely that the focus on the budget is primarily for dramatic effect, it is a key focus of the show and each bride’s story. Kleinfeld specializes in designer dresses, and wedding gowns range from about $1,500 to upwards of $40,000, clearly making it an upscale retail store (much closer to Saks than to Macy’s or S. Klein in the original department store study). The importance of determining the budget of the bride is emphasized repeatedly on the show, by the managers, consultants, and the narrator. Failing to find out how much a bride is willing to spend on her wedding dress creates the risk that the consultant may be showing dresses to the bride that she will not end up purchasing because of cost. Indeed, this “rookie mistake” was one of the reasons given for the poor sales performance of consultant Claudia, who eventually left the store’s employ. In all likelihood, the consultant knows how much the bride’s budget is before the taping of the show begins, as it is part of the application to appear on the show, and so this heavy focus on the budget is mostly for the sake of the TV program.

Excerpt 1.

Magdalonie’s Introduction (Season 4, Episode 9).

Dianne:	OK come on in and make yourself comfortable.
	What were you thinking for your
Magdalonie:	I like the ballgown.
Dianne:	Traditional?
Magdalonie:	Yes. Definitely traditional.
Magdalonie ((to camera))²	I’m going for the Cinderella, tiara, long train, ’cause I plan on having a horse and carriage.
	The ceremony will be at the Brooklyn Tabernacle.
	We have about 200 guests.
Dianne:	Do you like beading?
Magdalonie:	I love sequins.
Dianne:	You love beading. Okay.
	Something sparkly?
	Strapless?
Magdalonie:	Yes.
Dianne:	OK. I’m gonna bring some in.
Magdalonie:	But I’m looking for two dresses
	’cause I’m gonna switch for the reception.
Dianne:	OK you are gonna switch.
Magdalonie:	Yeah. So I’ma stay in one for the ceremony.
	The reception maybe
Dianne:	Sexy?
Magdalonie:	Yes.
	Sexy.
	I like to dance, so
Dianne:	Okay.
Magdalonie:	Something that’s gonna allow me to flow.
Dianne:	Okay.
Magdalonie ((to camera)):	I wanna keep the expectations high and definitely
	wow ((laughs)).
Dianne:	Is there a price range that you wanna keep it somewhere? Cap it at something?
Magdalonie:	The ballgown, I’m not going over five thousand.
Dianne:	Okay.
Magdalonie:	The second dress
	about twenty five [hundred].
Dianne:	Okay. I’ll be right back.
Narrator:	Dianne must find two dresses for her bride to live the dream.

Once the consultant has ideas for dresses to select from the storeroom, she goes to find dresses for the bride to try on. The remainder of the episode shows the bride trying on a number of dresses, listening to her friends’ and family members’ opinions, and then finally deciding on a purchase (a “Yes” to the question “Is this your dress?”) or leaving the store empty-handed. Although the focus in each case is on the bride, the consultant has a pivotal role to play: she guides the bride during the appointment, asking questions, offering advice, and helping the bride make the best decision.

Each episode also contains segments in which the bridal consultant, the bride, and sometimes members of the bride’s party speak separately to the camera, after the appointment has ended, responding to questions in an interview situation, though the questions and the interviewer remain unknown and unseen to the audience. As discussed above, these “testimonials” are a hallmark feature of reality television shows (Bignell 2005). In this case, participants reflect on, explain, or otherwise offer commentary on the events that unfolded during the bridal gown appointment. Consultants in these situations often vent about a member of the bridal party or complain about some circumstance or event that may have negatively affected the sale. Using data from both of these segments of the show, we tested the hypothesis that the bridal consultants on Say Yes to the Dress would vary in their use of (r) depending on the budget the bride stated at the outset of her appointment. We now present the design and results of this study, which will be discussed in relation to the issues surrounding reality television outlined above.

Data and Methods

Seasons 1 through 5 (2007 to 2010) of Say Yes to the Dress were analyzed, for a total of seventy-eight episodes, roughly twenty minutes in length each. The analysis presented here includes five consultants featured on the show: Audrey, Camille, Debbie, Dianne, and Keasha, chosen because they are the consultants who are variably rhotic.³ These consultants had been working at Kleinfeld for between eight and fifteen years during the taping of the show, though specific dates were not available. And while the actual ages of these consultants are unknown, all appear to be between about forty and fifty-five years old.⁴ The analysis includes interactions with a total of 126 brides. Keasha worked with the most brides (41), followed by Audrey (28), Dianne (24), Camille (19), and Debbie (14). Figure 2 shows the number of brides each consultant interacted with in the three budget categories, which are described below. Not only does Keasha work with the most brides overall, but she is also assigned more brides in each of the three budget categories, and almost twice as many in the High category (7) as the others, who all have 4 brides in the High category, except Dianne, who has only 1. As mentioned above, the pairing decisions are not random, though we have no information about how the decisions are made (besides the occasional comment by a manager that a bride has requested to work with a particular consultant). Some consultants, like Keasha, may be regularly assigned more high-budget brides for class-based reasons, which may be intertwined with language. To draw a further parallel to the department store studies, the hiring decisions made at Saks, Macy’s, and S. Klein may be made at least in part on the basis of the speech of the prospective employee, so that those who are already more rhotic are more likely to be hired at Saks, and, maybe within that store, those who are most rhotic are more likely to be assigned a post on the upper floors. This is just not information that we are privy to, though it is interesting to consider in the context of these studies.

Figure 2.

Number of Brides per Consultant, Separated by Budget Category.

Internal Constraints

In non-rhotic dialects, /r/ is variably present when it occurs in the coda of the syllable. When /r/ is in an onset, whether underlyingly (military) or as a result of resyllabification (flattering), it is always pronounced; thus, such words were not included. “Linking r,” in which /r/ crosses a word boundary and the following word is vowel-initial (butter up), was included in the analysis, as recent studies have found variation in this context (Becker 2009; Nagy & Irwin 2010). The preceding vowel has been consistently found to constrain rhoticity. Vowel categories from Wells (1982) were used: Start, North, Square, Near, Cure, and Fur. Fur is separated into stressed and unstressed levels, following recent studies (Nagy & Irwin 2010; Becker 2014a). Tokens in the Cure class were very few (n = 4) and thus were removed from the analysis. Tokens were also coded for syllabic stress and the following context (vowel, consonant, or pause) as well as whether the words belonged to lexical (open class) or functional (closed class) categories, all of which have been shown to have significant effects on rhotic variation.

External Constraints

Driving the research question for this paper was whether the budget of the bride would correlate with the bridal consultants’ use of (r). The budget was divided into three categories: High, Medium, and Low, using the pricing categories on the store’s website as a guide (www.kleinfeldbridal.com, accessed July 18, 2012).⁵ Budgets that were Low were $3,000 and under. The Middle category ranged from $3,500 to $7,500, and budgets considered High were above $8,000. If the bride gave a range for her budget (e.g., $3,000–$5,000), the upper limit was used. We predicted that consultants would use the bride’s budget as an indicator of social class and therefore favor rhoticity when the budget was High, analogous to the favoring of [r-1] in Saks.

Based on previous work on audience design and accommodation theory, we predicted that consultants would exhibit lower rates of [r-1] when there was someone in the audience who also was variably non-rhotic. In order to identify such possible effects, we included a factor group to control for whether anyone in the bridal party—bride, friends, or family—also exhibited non-rhotic pronunciations. In such cases, each token from the bridal appointment during which an audience member was non-rhotic was identified as being produced within an interaction in which [r-0] was present among the consultant’s interlocutors.⁶

As already mentioned, an important feature of reality television is the use of the post-show interview to provide commentary about events being broadcast to the audience. On Say Yes to the Dress, this includes commentary from the bride, members of her family or friends, the bridal consultant, and other employees from the salon who might have been involved in the sales interaction. In order to differentiate between moments in which a consultant was speaking to her client and those in which she was talking about her client, we included a speech design factor group. Tokens coded as “audience” were those in which the bride, friends, and family (i.e., any person who could be paying for the dress and/or influencing a decision about the purchase) were a ratified participant in the interaction. Tokens coded as “referee,” on the other hand, were those in which the consultant was talking about the bride or friends and family in attendance, to someone else, and neither the bride nor any member of her party was present for the interaction. As discussed above, Bell (1984:186) defines referees as “third persons not physically present at an interaction, but possessing such salience for a speaker that they influence speech even in their absence.” The vast majority of referee-designed speech occurred during interviews after the show had been taped, and thus the consultant was speaking to the camera (and whoever was conducting the interview) about the bridal appointment and its participants.⁷ Based on the fact that during the testimonials, consultants are more directly orienting to the wider audience at home, we expected to find a favoring of [r-1] for referee-designed speech.

Logistic regression was conducted in Rbrul, an add-on package for the statistical program R (Johnson 2009). Rbrul allows for mixed-effect modeling, which treats designated factor groups as random effects and others as fixed. Here, “Word” was included as a random effect, because this approach controls for the possibility that some words will favor or disfavor a particular variant beyond what is predicted by the linguistic and social factors in the model (Johnson 2009). The same is often true for speakers; however, since we are interested in variation at the level of these five individual speakers, we included Speaker as a fixed effect.⁸

Tokens of /r/ were coded perceptually as rhotic or non-rhotic by one analyst. A total of 2,496 tokens were included. A random sample of 200 tokens was coded separately by a second analyst and then compared to the original scores. This cross-check showed 88 percent agreement between the two coders. The categorization of the remaining 12 percent of the tokens was discussed by the two analysts to come to an agreement as to their classification.

Results

The overall rate of [r-1] in the sample was 49 percent (1,215/2,496). Table 1 provides the main effects for the internal factors selected in Rbrul, with [r-1] as the application value. As with other studies of rhoticity, these internal factors are strong, with preceding vowel, following environment (vowel/consonant/pause), syllabic stress (stressed/unstressed), and word type (lexical/functional) all being significant predictors of [r-1].

Table 1.

Internal Constraints on (r), Main Effects.

Best run, stepping down
Deviance2741.82	Degrees of freedom18	Intercept−0.584	Grand Mean0.487
Random effect of Word (SD = 1.35)
Preceding Vowel (p < .0001)	N	% [r-1]	Factor Weight
FUR	786/1,420	55	.86
NORTH	192/489	39	.47
SQUARE	102/252	40	.42
NEAR	61/145	42	.33
START	74/190	38	.31
Syllabic Stress (p < .0001)	N	% [r-1]	Factor Weight
Stressed	908/1,682	53	.79
Unstressed	307/814	37	.20
Following Environment (p < .0001)	N	% [r-1]	Factor Weight
Vowel	250/353	70	.74
Pause	225/387	58	.48
Consonant	740/1,756	42	.26
Word Type (p = .001)	N	% [r-1]	Factor Weight
Lexical	741/1,474	50	.63
Functional	474/1,022	46	.36

Nagy and Irwin (2010) note that a universal finding for (r) is that [r-1] is strongly favored for stressed schwa, whereas unstressed schwa disfavors [r-1]. This is borne out as well in the current study. Consistent with Becker’s (2014a) large-scale study in New York City, both stressed and unstressed schwa tokens are contained within the Fur factor group, and syllabic stress was coded separately. Our findings parallel those of both Becker (2014a) and Nagy and Irwin (2010) in that Fur and syllabic stress both favor [r-1]. The other vowels disfavor [r-1] and are ranked in the following order, from highest to lowest factor weight: North > Square > Near > Start. The following environment has a significant effect, with a following vowel strongly favoring rhoticity and a following consonant strongly disfavoring the [r-1] variant. A following pause slightly disfavors [r-1]. Finally, the word type is significant, with lexical words favoring [r-1] and function words disfavoring [r-1]. In addition to these linguistic constraints, there were several external factors determined to be significant in the model, discussed below.

External Constraints

The fixed external factors selected as predictors of [r-1] were Budget (Low, Medium, High), Speech Design (Audience or Referee), Non-Rhoticity in Bridal Party (whether or not anyone in the bridal party exhibited some [r-0]), and Speaker (the five bridal consultants). Table 2 presents the main effects for these factors, which we discuss in turn.

Table 2.

External Constraints on (r), Main Effects.

Best run, stepping down
Deviance2741.82	Degrees of freedom18	Intercept−0.584	Grand Mean0.487
Random effect of Word (SD = 1.35)
Budget (p = .01)	N	% [r-1]	Factor Weight
High	204/356	57	.55
Medium	558/1,095	50	.50
Low	453/1,045	43	.44
Rhoticity in Bridal Party (p = .01)	N	% [r-1]	Factor Weight
some [r-0]	347/645	53	.53
all [r-1]	868/1,851	46	.46
Speech Design (p = .009)	N	% [r-1]	Factor Weight
Audience	546/1,277	42	.46
Referee	669/1,219	54	.53
Speaker (p < .001)	N	% [r-1]	Factor Weight
Keasha	415/746	55	.61
Dianne	307/583	52	.55
Debbie	146/305	47	.52
Camille	138/352	39	.40
Audrey	209/510	40	.40

Budget

The current study was driven by the question of whether the brides’ budgets on this reality television show would act as a marker of social class and thus have an effect on consultants’ use of (r), in a similar manner to a shopper’s presence in a particular department store in the department store studies. When consultants are working with a bride in the High budget category (over $8,000), [r-1] is the favored variant, with a factor weight of .550. The Medium category ($3,500 to $7,500) marginally favors [r-1], with a factor weight of .504. Finally, the Low category (less than $3,000) disfavors [r-1], showing the lowest probability for the rhotic variant at .448. Such a result indicates that the consultants do respond to the budget of their bridal customers in a similar manner to the way in which employees in the New York City department stores designed their speech according to the imagined social status of their ideal customers. That is, just as employees at Saks were more likely than those at Macy’s and S. Klein to produce [r-1] in the phrase fourth floor, employees at Kleinfeld are more likely to use the prestige variant when their customers express an intention or ability to spend larger amounts of money. Thus, Kleinfeld consultants are operating on the assumption that a bride’s budget works as a subtle indicator of her social standing, and this in turn exerts an effect on the consultants’ use of this linguistic variable. While the differences in percentages of [r-1] are small between the three budget categories (ranging from 43 percent to 57 percent), they are significant nonetheless; indeed, we would expect intraspeaker differences to be small even if interclass differences are great, given Bell’s (1984) assertion that variation in style for most linguistic variables will inhabit a smaller range than social variation since the former derives from the latter (see also Schilling-Estes 2002). Furthermore, this closely parallels the finding from the original department store study that within the single high-end store Saks, clerks who worked on the upper floors exhibited higher rates of [r-1] than employees in the same store who worked on the lower levels. We should also take into account that the budget of the bride may motivate style shifts among the consultants for very practical reasons. As the consultants work on commission, the sale of a dress translates into more money for them, as well as increased job security (consultants can be fired if they do not meet their monthly sales goals); additionally, the more money a client spends on a dress, the higher the consultant’s commission is. This monetary incentive may indeed be part of the motivation to design their speech toward a more standard, more prestigious form, the higher the budget of the bride. As is the case in the department store studies, it is certainly possible that a customer at Kleinfeld does not actually have the money required to shop at the store or to buy the desired dress. On Say Yes to the Dress, sometimes the consultants identify clients who appear unable to pay for a gown that they are asking to try on or who are not serious about making a purchase, and who, to the consultants’ great annoyance, are just “playing dress up.” Labov (1972:57) makes a similar comment about the upper floors of Saks, in which employees “are stationed at strategic points to screen out the casual spectators from the serious buyers.” In other words, in both cases, there are people who enter these upscale stores who may not be in a position financially to actually purchase anything within them. Despite the existence of these cases in the department stores and in Kleinfeld Bridal, the level of the customers’ prestige presented on the surface acts as an influence on the employees’ linguistic performance.

Non-Rhoticity in Audience

We predicted that if anyone in the bridal party showed non-rhotic pronunciations, [r-0] would be the favored variant rather than [r-1] because the consultants would show convergent behavior with their audience. Of the total of 126 brides whose appointments were included in our analysis, there were twenty-seven appointments during which some member of the bridal party (bride, family member, friend) was variably non-rhotic. This was a significant effect, but in the unexpected direction: the consultants used somewhat more [r-1] when there was someone in the group who exhibited variability in rhoticity. The mean [r-1] within this subset of tokens was 54 percent (347/645), which is slightly higher than for the overall data set. We propose that this is in part a result of the mediated data source we are working with, which may affect some speakers more strongly than others. We return to this argument below, when we discuss the five speakers in more detail.

Speech Design

Whether the token was produced as audience design or referee design was also a significant predictor of rhoticity. [r-1] was favored when consultants were speaking about the bride (referee) versus when they were speaking to the bride or someone in attendance with her at the salon (audience). Tokens that were coded as referee design were overwhelmingly produced as the consultant spoke directly to the camera (1 147/1 219), with the small remainder produced when the interaction was with another Kleinfeld employee, such as a store manager or the fashion director. As discussed above, while the consultants were likely aware that they were being filmed all the time during these appointments, the finding that [r-1] is favored in referee-designed speech may be due to the effect of the heightened awareness of being observed that is elicited by the one-on-one interview format. Presumably, there is a camera within eyeshot of the consultant during these interviews, and participants are no longer engaged in the events being filmed for the show, which may draw their attention to their actions and away from their speech. In the testimonials, as the audience multiplies exponentially (potentially millions of viewers at home), consultants produce higher rates of [r-1] than during direct interactions with their clients, shifting to a more prestigious style of speaking. Moreover, it is possible that the consultants operate under the assumption that the home audience is rhotic and so shift to greater rhoticity in order to accommodate to that norm.⁹ Awareness is a notoriously difficult concept to pin down in sociolinguistic research; however, this speculation seems plausible, given that we know of the different effects the viewing audience may have on the speech of reality show participants (e.g., Thornborrow & Morris 2004).

Speaker

Finally, we turn our discussion to the individual bridal consultants. As Table 2 above shows, Keasha has the highest probability for [r-1], followed by Dianne and Debbie, who also favor this variant. Camille and Audrey disfavor [r-1], showing the lowest probability for [r-1]. To further examine these patterns, we present the mean [r-1] for each consultant (Speaker), separated by budget category (see Figure 3), though this interaction was not statistically significant in the regression model. Above each column is the number of brides the consultant worked with in that budget category.

Figure 3.

Mean [r-1] for Speaker According to Bride’s Budget Category (with Number of Brides in Each Category).

As shown here, most of the consultants follow the overall pattern observed in the data, that higher rates of [r-1] occurred with higher budgets. The two deviations from this general trend are Debbie and Camille, who do not show a steady progression in the increase in [r-1] from Low to Medium to High budgets: Debbie shows slightly more [r-1] in the Low category than the Medium category, and Camille shows more [r-1] in Medium than in High. However, both speakers show more [r-1] in the High budget category as compared to the Low category, so that their general trend follows that of the larger set of consultants.

We noted above the somewhat unexpected result that [r-1] is favored when there is a person in the bridal party who exhibits non-rhoticity, when we predicted that consultants would accommodate to non-rhoticity in these situations, thereby favoring [r-0] instead. Figure 4 shows the interaction between non-rhoticity in the audience and the individual speaker, which was statistically significant (p < .01). This helps illuminate the earlier finding of what seemed to be disaccommodation, when we had predicted that accommodation would occur.

Figure 4.

Interaction between Consultant and Presence of Non-rhoticity in Audience (p < .01).

The differentiation among the individual consultants here shows interesting connections to the main effects discussed above. Audrey, Debbie, and Camille behave as accommodation theories would predict; that is, they disfavor [r-1] when there is someone in the interaction who is also non-rhotic, showing convergence with their interlocutors. Likewise, when all members of the audience are fully rhotic, these same three consultants favor [r-1]. Interestingly, Debbie had only three bridal appointments during which there was non-rhoticity in the audience, and all three of those brides were in the High budget category (of a total of four High budget brides that Debbie worked with over the course of the five seasons analyzed). This seems to suggest that accommodating to non-rhoticity exerts a stronger effect on Debbie’s variation in (r) than does the perceived social class of the bride. If Debbie were orienting more strongly to the bride’s social class, then the fact that the brides were in the High budget category—each willing to spend $10,000 on a dress—would trump accommodation to the non-prestige variant. Instead, we find that for Debbie, the effects of accommodation are stronger than the effects of the perceived social class of the bride. Dianne and Keasha produce unexpected results, favoring [r-1] when there is someone in the interaction who shows variable rhoticity; that is, they diverge from their interlocutors with respect to this variable or show disaccommodation (Giles, Coupland & Coupland 1991; see also Kiesling 2005). The finding that Keasha and Dianne, who have the highest rates of [r-1] overall, use more [r-1] when interacting with someone who is variably non-rhotic is intriguing. We might speculate that hearing the salient non-rhotic variant brings it to the forefront of speakers’ minds; in this case, Dianne and Keasha may be made more aware of the stigma that variant carries with it in the moment of the interaction and in turn increase their use of the prestige variant [r-1]. We would need more and different data to confirm such a hypothesis, but it does offer a potential explanation for the patterns we find among these two consultants and one that fits as well with the main finding that these speakers are more rhotic overall than the other consultants.

Although ethnicity is not a focus of the current paper, a final note should be made about Keasha, the only African American consultant featured on the show.¹⁰ Based on previous research on (r), we would expect Keasha to have a lower probability for [r-1] than the other speakers, even while her rates of (r) are within the range of the other consultants. Non-rhoticity is a well-documented feature of African American speech in New York and elsewhere (e.g., Labov 1966; Myhill 1988; Rickford 1999; Hinton & Pollock 2000; Green 2002). Becker (2014a:155) shows that African Americans in New York disfavor [r-1], showing a substantially lower rate of its use than the other ethnic groups sampled, and they are not moving toward rhoticity in apparent time. However, Becker (2014a) also finds that African Americans in the lower middle class do favor [r-1]. Similarly, Blake and Shousterman (2010) demonstrated that “Black ethnics” (which includes African Americans as well as second-generation Caribbean/West Indians) view the rhotic pronunciation as indexical of higher social class and prestige. In previous department store studies as well, African American employees used lower rates of all [r-1] than white employees. Labov (1972:54) found 50 percent of African Americans (n = 2) in Saks producing all [r-1] tokens, and Mather (2012:348) found 55 percent of African American employees (n = 16) in the same store producing all [r-1]. While Labov (1972) noted that there was a more pronounced difference between white and African American employees in S. Klein as opposed to Macy’s (again indicating that in the more prestigious stores there is more accommodation to prestige forms), in all cases, rates for African American employees are lower than the rates of all [r-1] for whites in that store. All of this leads to the question, why does Keasha, an African American consultant at Kleinfeld, use more [r-1] than her white counterparts?

As we showed above, Keasha is assigned more brides than any of the other consultants, including more in the High budget category. Thus, there may be a relationship between Keasha’s more standard “establishment” language (Rahman 2008) and the budget level of the brides she is assigned by Kleinfeld management, which helps explain her higher rates of [r-1] in comparison to the other consultants. We also must consider who Keasha is, in the context of the store in which she works and the national show on which she appears. Several scholars have noted the particular position of African American women with regard to language practices, and the intersection of their race and gender, two marginalized categories in “a racialized, genderized, sexualized, and classed world … [in which] the Black female develops creative strategies to overcome her situation, to ‘make a way outa no way’” (Richardson 2003:77). Empirical evidence from Rahman (2008) indicates middle-class African Americans’ sensitivities to the appropriateness of stigmatized and nonstandard linguistic forms in various settings and interactions with various audiences (see also Hoover 1978). We would not want to claim here that Keasha’s patterns are indicative of linguistic insecurity (of the type Labov 1966 has shown for the lower-middle class in New York City with respect to non-rhotic pronunciations). At the same time, we might surmise that as an African American woman employee in a high-end store, which serves many white middle- and upper-middle-class brides, Keasha is higher in “stigma consciousness” (Pinel 1999) than some of her colleagues, who do not navigate structural and perhaps interpersonal racism in their lives. As a result, it may be that Keasha practices more avoidance of this stereotyped linguistic form, particularly in a situation where she imagines herself to be under increased scrutiny (i.e., being filmed for a national television broadcast). Finally, we must keep in mind that Keasha is a single individual, and thus the fact that she deviates from norms for African Americans more broadly may not be that puzzling. In fact, in the original department store study, Labov surveyed two African American employees in Saks. One of these speakers produced all four tokens as [r-1]; the other produced all four tokens with [r-0]. Neither of these speakers produces the expected results, as one deviates from ethnic group norms (we would not expect all [r-1] for an African American speaker based on the larger sample), and the other deviates from the overall pattern of the high-prestige store (we would expect at least some [r-1] in Saks). A great deal of recent work in intraspeaker variation has focused on explanations of identity and stance in elucidating individual linguistic styles (e.g., Eckert 2000; Kiesling 2009; Bucholtz 2011), and while such questions are outside the scope of the current paper, they might be well worth considering in pursuing more nuanced explanations for Keasha’s (and others’) linguistic performance on this reality TV show.

Summary and Discussion

This paper set out to provide a description of variation in rhoticity on the reality television show Say Yes to the Dress. The analyses presented here demonstrate that (r) continues to function in New York as a linguistic resource to mark prestige, particularly in relation to the audience. In this way, the quantitative results of (r) variation on Say Yes to the Dress parallel the three New York City department store studies (Labov 1966; Fowler 1986; Mather 2012). We have shown that the perceived prestige of clients exerts an effect on bridal consultants to the extent that they favor the prestige variant when they conceive of their client as being of a higher social class, thus highlighting the benefits of an audience design approach to intraspeaker variation. Even while taking additional factors into consideration, such as accommodation effects for non-rhoticity in the audience, this result is still borne out. And while there are some aspects of the bridal world that set it apart from other interactions within retail (Otnes & Pleck 2003; Engstrom 2012), our findings regarding dress budget here are similar to the effect of the prestige of the department store in earlier research and mirror quite closely the effect of different departments within the prestigious store of Saks Fifth Avenue.

We have argued that reality television is a rich resource for culling data that are well suited for addressing sociolinguistic questions. Reality TV provides a format that typically contains two different speech contexts (the show’s “action” and the one-on-one testimonials). These contexts ultimately offer sociolinguistic researchers the opportunity to study the linguistic consequences of different audience types, both real and imagined, as well as speech designed for audiences and referees and also the projection of identities as show participants perform mediated “reality” on camera. Additionally, further work on accommodation in the context of reality TV seems to be a promising area, particularly connected with speaker identities and ideologies, which was outside the scope of the current work. Within this particular reality TV show, it would be interesting to examine consultants’ stylistic variation with other linguistic features, such as the stable variable (ING) or the raising of BOUGHT, another iconic feature of NYCE (see Becker 2014b), to see if similar patterns obtain.

Our approach in this paper has brought variationist analysis to reality television, a relatively new data source in the field. We see this genre as having the potential to answer sociolinguistic questions in new ways and providing a lens through which to examine established sociolinguistic concerns, such as performativity and identity, and speech design and style-shifting. Future research on reality television will be able to further delineate the boundaries of what the genre can offer and to discover new ways in which such data are able to contribute to the sociolinguistic enterprise.

Footnotes

Acknowledgements

We would like to thank Julie Roberts and Christina Schoux Casey for constructive feedback on an earlier version of this article as well as two anonymous reviewers and the journal editors for helping to refine the arguments presented here. We also thank Alan Howard, who provided consultation on the statistical modeling. We are responsible for any errors or oversights that remain.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Maeve Eberhardt is an assistant professor of linguistics at the University of Vermont. Her research focuses on language and projection of identities surrounding race, gender, and sexuality as well as on the appropriation of linguistic forms, particularly African American English.

Corinne Downs is a recent graduate of the University of Vermont in anthropology. Her interests lie in linguistic anthropology, semiotics, in-group practices within web-based communities, and applied anthropology within the marketing and advertising field.

References

Bakht

Maryam

. 2010. Here’s “The Situation”: Style and identity on the “Jersey Shore.” Paper presented at New Ways in Analyzing Variation 39, San Antonio, TX, November 4–6.

Becker

Kara

. 2009. /r/ and the construction of place identity on New York City’s Lower East Side. Journal of Sociolinguistics 13(5). 634–658.

Becker

Kara

. 2014a. (r) we there yet? The change to rhoticity in New York City English. Language Variation and Change 26(2). 141–168.

Becker

Kara

. 2014b. The social motivations of reversal: Raised BOUGHT in New York City English. Language in Society 43(4). 395–420.

Bell

Allan

. 1984. Language style as audience design. Language in Society 13(2). 145–204.

Bell

Allan

. 1992. Hit and miss: Referee design in the dialects of New Zealand television advertisements. Language and Communication 12(3/4). 327–340.

Bell

Allan

. 2001. Back in style: Reworking audience design. In Eckert

Penelope

Rickford

John

(eds.), Style and sociolinguistic variation, 139–169. Cambridge, UK: Cambridge University Press.

Bignell

Jonathan

. 2005. Big Brother: Reality TV in the twenty-first century. New York: Palgrave Macmillan.

Blake

Renee

Shousterman

Cara

. 2010. Second generation West Indian Americans and English in New York City. English Today 26(3). 35–46.

10.

Bucholtz

Mary

. 1999. “You da man”: Narrating the racial other in the linguistic production of white masculinity. Journal of Sociolinguistics 3(4). 443–460.

11.

Bucholtz

Mary

. 2003. Sociolinguistic nostalgia and the authentication of identity. Journal of Sociolinguistics 7(3). 398–416.

12.

Bucholtz

Mary

. 2011. White kids: Language, race, and styles of youth identity. New York: Cambridge University Press.

13.

Eckert

Penelope

. 2000. Linguistic variation as social practice: The linguistic construction of identity in Belten High. Malden, MA: Blackwell.

14.

Engstrom

Erika

. 2012. The bride factory: Mass media portrayals of women and weddings. New York: Peter Lang.

15.

Feagin

Crawford

. 1990. Dynamics of sound change in Southern States English: From r-less to r-full in three generations. In Edmonson

Jerold A.

Feagin

Crawford

Mülhaüsler

Peter

(eds.), Development and diversity: Language variation across time and space: A festschrift for Charles-James N. Bailey, 129–146. Arlington: SIL/University of Texas at Arlington.

16.

Fowler

Joy.

1986. The social stratification of /r/ in New York City department stores, 24 years after Labov. Unpublished manuscript.

17.

Giles

Howard

. 1973. Accent mobility: A model and some data. Anthropological Linguistics 15(2). 87–105.

18.

Giles

Howard

Coupland

Nikolas

Coupland

Justine

. 1991. Accommodation theory: Communication, context, and consequence. In Giles

Howard

Coupland

Justine

Coupland

Nikolas

(eds.), Contexts of accommodation: Developments in applied sociolinguistics, 1–68. Cambridge, UK: Cambridge University Press.

19.

Goffman

Erving

. 1981. Forms of talk. Philadelphia: University of Pennsylvania Press.

20.

Green

Lisa

. 2002. African American English: A linguistic introduction. Cambridge, UK: Cambridge University Press.

21.

Hay

Jennifer

Jannedy

Stephanie

Mendoza-Denton

Norma

. 1999. Oprah and /ay/: Lexical frequency, referee design, and style. Proceedings of the 14th International Congress of Phonetic Sciences, San Francisco, CA, 1389–1392.

22.

Hinton

Linette

Pollock

Karen

2000. Regional variation in the phonological characteristics of African American Vernacular English. World Englishes 19(1). 59–71.

23.

Hoover

Mary

. 1978. Community attitudes towards Black English. Language in Society 7(1). 65–87.

24.

Johnson

Daniel Ezra

. 2009. Getting off the GoldVarb standard: Introducing Rbrul for mixed-effects variable rule analysis. Language and Linguistics Compass 3(1). 359–383.

25.

Kiesling

Scott

. 1998. Men’s identities and sociolinguistic variation: The case of fraternity men. Journal of Sociolinguistics 2(1). 69–99.

26.

Kiesling

Scott

. 2005. Variation, stance and style: Word final -er, high-rising tone, and ethnicity in Australian English. English World-Wide 26(1). 1–42.

27.

Kiesling

Scott

. 2009. Style as stance: Stance as the explanation for patterns of sociolinguistic variation. In Jaffe

Alexandra

(ed.), Stance: Sociolinguistic perspectives, 171–194. New York: Oxford University Press.

28.

Labov

William

. 1966. The social stratification of English in New York City. Washington, DC: Center for Applied Linguistics.

29.

Labov

William

. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.

30.

Labov

William

. 1984. Field methods of the project in linguistic change and variation. In Baugh

John

Sherzer

Joel

(eds.), Language in use: Readings in sociolinguistics, 28–53. Upper Saddle River, NJ: Prentice Hall.

31.

Labov

William

. 2001. The anatomy of style-shifting. In Eckert

Penelope

Rickford

John

(eds.), Style and sociolinguistic variation, 85–108. Cambridge, UK: Cambridge University Press.

32.

Labov

William

2006. The atlas of North American English. New York: Mouton de Gruyter.

33.

Mather

Patrick-André

. 2012. The social stratification of /r/ in New York City: Labov’s department store study revisited. Journal of English Linguistics 40(4). 338–356.

34.

Myhill

John

. 1988. Postvocalic /r/ as an index of integration into the BEV speech community. American Speech 63(3). 203–213.

35.

Nagy

Naomi

Irwin

Patricia

. 2010. Boston (r): Neighbo(r)s nea(r) and fa(r). Language Variation and Change 22(2). 241–278.

36.

Otnes

Cele

Pleck

Elizabeth

. 2003. Cinderella dreams: The allure of the lavish wedding. Berkeley: University of California Press.

37.

Pardo

Rebecca

. 2013. Reality television and the metapragmatics of racism. Journal of Linguistic Anthropology 23(1). 65–81.

38.

Pinel

Elizabeth

. 1999. Stigma consciousness: The psychological legacy of social stereotypes. Journal of Personality and Social Psychology 76(1). 114–128.

39.

Podesva

Rob

. 2007. Phonation type as a stylistic variable: The use of falsetto in constructing a persona. Journal of Sociolinguistics 11(4). 478–504.

40.

Preston

Dennis

Niedzielski

Nancy

. 2003. Folk linguistics. Berlin: Mouton de Gruyter.

41.

Queen

Robin

. 2013. Working with performed language: Movies, television, and music. In Mallinson

Christine

Childs

Becky

Herk

Gerard Van

(eds.), Data collection in sociolinguistics: Methods and applications, 217–227. New York: Routledge.

42.

Rahman

Jacquelyn

. 2008. Middle-class African Americans: Reactions and attitudes toward African American English. American Speech 83(2). 141–176.

43.

Rampton

Ben

. 1995. Crossing. London: Longman.

44.

Richardson

Elaine

. 2003. African American literacies. London: Routledge.

45.

Rickford

John

. 1999. African American Vernacular English. Malden, MA: Blackwell.

46.

Rickford

John

McNair-Knox

Faye

. 1994. Addressee- and topic-influenced style shift. In Biber

Douglas

Finegan

Edward

(eds.), Sociolinguistic perspectives on register, 235–276. Oxford, UK: Oxford University Press.

47.

Schilling-Estes

Natalie

. 1998. Investigating “self-conscious” speech: The performance of register in Ocracoke English. Language in Society 27(1). 53–83.

48.

Schilling-Estes

Natalie

. 2002. Investigating stylistic variation. In Chambers

J. K.

Trudgill

Peter

Schilling-Estes

Natalie

(eds.), The handbook of language variation and change, 375–401. Malden, MA: Blackwell.

49.

Schoux Casey

Christina

. 2013. Postvocalic (r) in post-Katrina New Orleans English. Unpublished manuscript.

50.

Thornborrow

Joanna

Morris

Debora

. 2004. Gossip as strategy: The management of talk about others on reality TV show “Big Brother.” Journal of Sociolinguistics 8(2). 246–271.

51.

Valentinsson

Mary-Caitlyn

. 2010. “The tits are coming out tonight”: Indexing gender on the “Jersey Shore.” Paper presented at New Ways in Analyzing Variation 39, San Antonio, TX, November 4–6.

52.

Wells

John C.

1982. Accents of English, vol. 3. Cambridge, UK: Cambridge University Press.