Jane Lynch and / s / : The Effect of Addressee Sexuality on Fricative Realization

Although there has been a sizeable amount of work on the speech of gay men (e.g., Podesva 2007), there has been little to no research on gay or bisexual women, whether interspeaker or intraspeaker. This dearth is possibly due to the lack of a stereotypical gay speech style for women. Most people will recognize the gay man speech style exemplified by characters such as Kurt Hummel on Fox’s Glee, but there seems to be no female equivalent. While there may be visual stereotypes of sexuality such as “butch” lesbians sporting baseball caps and Doc Martens (or think of Old Hollywood bisexual Marlene Dietrich’s controversial love of tuxedos), this does not come with a particular speech style. Studies such as Podesva and van Hofwegen’s (2014) analysis of lesbian, gay, bisexual, and transgender (LGBT) residents of Redding, California, have found differences in the realization of /s/ (as defined by spectral measurements such as center of gravity or the mean frequency over the spectrum) between straight and lesbian or bisexual (LB) women. This study examines American lesbian actress Jane Lynch’s realization of /s/ according to center of gravity measurements in two different interviews, with the aim of determining if her /s/ realization is affected by the sexuality of her interlocutor. Lynch’s speech was measured across two topic-controlled interviews, one with lesbian host Rachel Maddow and the other with two non-lesbian women hosts, Gayle King and Erica Hill. Results show that Lynch used lower /s/ realizations (i.e., a lower spectral mean) with the lesbian host than with the non-lesbian hosts. The analysis explores how she uses /s/ both responsively and actively to index a non-heteronormative identity and conceptually aligns herself with the lesbian host. This is mainly presented within the frameworks of Bell’s (1984) theory of audience design and indexicality. It is argued that /s/ may not be consciously salient, but it is perceptually salient on some level (e.g., Mack and Munson 2012). It may therefore, along with other possible features, contribute to an individualized group-marking style. In the absence of a well-known “lesbian accent”, it is argued that Lynch uses /s/ as a tool to create and control her self-presentation to a heteronormative society. Secondarily, some LGBT vs. non-LGBT topic effects within one of the interviews are discussed, with the finding that Lynch has a lower mean of /s/ while discussing LGBT topics, such as same-sex marriage, than unrelated topics, such as her acting career.


Introduction and Background
Jane Lynch is an American comedian and actress best known for her role as Sue Sylvester on the popular television show Glee.She is also known for being open about her identity as a lesbian and for speaking out about gay rights, though she insists that she is not an activist (Couric 2010).Lynch was born in 1960 and grew up in Dolton, Illinois, a suburb of Chicago (L/Studio 2014).She currently lives in Los Angeles.
To date, there is very little literature that examines the speech styles of lesbian or bisexual women.This study asks the question: "does the sexual orientation of her interlocutor affect Jane Lynch's realization of /s/?"The results show that her realizations of /s/ have a lower center of gravity when conversing with a lesbian interlocutor than with non-lesbian interlocutors.The analysis explores her realization of /s/ as influenced by the sexual orientation of her interlocutor and situates it within the frameworks of audience design and indexicality in relation to a heteronormative society.

The Sociolinguistic Marker /s/
The spectral features of /s/ have been investigated extensively in previous literature, where it has been argued that gender differences in /s/ are both biologically and socially determined (Fuchs and Toda 2010).To illustrate, studies of male and female realizations of /s/ have found a higher center of gravity, or mean frequency over the spectrum, for female /s/ than for male /s/ (Fuchs and Toda 2010).It has also been shown that /s/ is used as a sociolinguistic marker for GLB (gay/lesbian/bisexual) speech (Podesva andvan Hofwegen 2014, Mack andMunson 2012).Studies of gay men's speech have found a higher center of gravity for /s/ for gay men than for straight men, a phenomenon often called "gay lisp" (Mack and Munson 2012).Studies of listeners' judgments of sexual orientation from recorded speech have similarly shown that spectral differences in /s/ (e.g., center of gravity) contribute to a speaker being judged "gay-sounding" or "straight-sounding" (Mack andMunson 2012, Munson et al. 2005).All told, however, most work correlating /s/ with gender and/or sexuality has not targeted lesbian/bisexual (LB) women, a gap that this study attempts to bridge.

Previous LGBT Speech Studies
Unlike for gay men, there is no stereotypical "lesbian accent" in the public mind (at least in English), other than a usually hazy guess at more "masculine-sounding" characteristics.1 Several studies of the speech of lesbian and bisexual (LB) women have found vowel differences (Pierrehumbert et al. 2004, Munson et al. 2005) between LB women and straight women.Additionally, Podesva and van Hofwegen (2014) found socially conditioned differences in /s/ between LB and straight women in Redding, California.
It is important to note that the vowel differences found in previous studies were not a uniform adoption of male vowel spaces: in other words, LB women are not simply "imitating" men's speech.This idea of indexing masculinity will be explored in specific relation to Lynch's linguistic behavior.
Previous studies on gay, lesbian, and bisexual (GLB) speech have more often been concerned with documenting the existence of inter-group variation (i.e., GLB vs. straight) than with variation within GBT individuals (Munson and Babel 2007).That being said, intraspeaker variation in gay men has been studied in detail (e.g., Podesva 2006, Podesva 2007); what has not been done is a similar case-study style investigation of a lesbian woman.
The current study, like Podesva and van Hofwegen (2014), argues that lower /s/ in the speech of Jane Lynch, a lesbian woman, is socially conditioned rather than a categorically ever-present feature of her speech.In this case, the social factor examined is audience, or specifically the second person "addressee" (Bell 1984).Here, this role is fulfilled by television interviewers.
It has been observed that characteristics of speech associated with gay/lesbian/bisexual speakers are "more salient in groups than individuals" (Munson and Babel 2007:442).For example, pitch-related variation in gay men has been observed to occur more in conversation with other gay men than in other situations (Podesva 2011).This study adds to the literature by looking at a possibly related phenomenon with a gay woman.

Methods
The realization of /s/ was operationalized as the center of gravity measurement over the entire fricative.This is in line with previous studies on /s/ (e.g., Podesva and van Hofwegen 2014).Word-initial, medial, and final /s/ were all included in the envelope of variation, as well as clitics and morphologically derived /s/ (e.g., plurals).Instances where the /s/ was preceded or followed by another fricative, such as /z/, were excluded, as it was not possible to reliably locate the boundaries of the target fricative.
Data were taken from two interviews: one with the openly lesbian host Rachel Maddow on her political news and opinion program "The Rachel Maddow Show" on MSNBC on 17 May 2012 (NewsPoliticsInfo 2012), and the other with non-lesbian hosts Gayle King and Erica Hill on the morning news show "CBS This Morning" on 15 May 2012 (CBS 2012).(N.B.Both Gayle King and Erica Hill are married to men, though this does not, of course, rule out possible bi-or pan-sexuality.The most that can be said is that they are very probably not lesbian.)The CBS interview consisted of two topics that can be defined as "acting and family" and "LGBT" (lesbian, gay, bisexual, and transgender).The former covers Lynch's start in acting and her parents' initial objection to it, and the latter her reaction to Obama's statement in support of same-sex marriage.The Maddow interview exclusively discusses samesex marriage and the Obama statement.Made only a few days apart, Lynch's statements about Obama and the gay rights movement in the two interviews are extremely similar and even identical in places, for example: (1) Maddow interview: It [i.e., gay rights] is an issue I have not really taken personally so much (…) but when the president came out and said that he supported the dignity of our families, and ourour relationships, that really moved me, that really touched me.
(2) CBS interview: I don't take this, the debate about gay rights, personally sometimes, I kind of see it objectively, and I took this really, I really felt it, I really feltwhen he, the president came out and said um, you know, "I -I believe that gay people should be able to marry each other, I think it's a right they should have as well" and all of a sudden, I took that, I -I that really moved me, it touched me.
Although topic is a legitimate variable for possible investigation, it is largely controlled for in the two interviews in question.The CBS "acting and family" segment, then, acted as a type of "unmarked" baseline (unmarked in view of this study in that there was no LGBT content or direct audience).Ideally, a recording with the same topic varied by interviewer would be used as a control, but such a recording was unfortunately not available.The interviews were also controlled for interviewer gender, all three of the interviewers being female.
The recordings were digitized using Soundflower (Ingalls 2012) and sampled at 44.1 kHz.Measurements were done acoustically in Praat (Boersma and Weenink 2014) by first applying a low-pass filter and obtaining the center of gravity over the entire fricative.2All instances of /s/ in Lynch's speech were coded for, in line with the envelope of variation defined above.Rachel Maddow's speech was also coded for /s/.However, the speech of Gayle King and Erica Hill was not examined, due to time limitations and the complication of having two interviewers instead of one.Each token was also coded for syllable position (onset or coda) as in Podesva and van Hofwegen (2014).All data were coded by the author, but an inter-rater reliability test for a 30-second clip of the corpus resulted in a 94% correlation.The first 2 minutes and 35 seconds of the Maddow interview were excluded from collection, as Lynch was addressing the audience and giving a political report in imitation of Maddow during this time, and she was not therefore speaking directly to or with her.In total, 239 tokens were collected over 17 minutes and 16 seconds of recordings.For Lynch's speech, 70 were from the Maddow interview, 47 from the "LGBT" CBS segment, and 31 from the "acting and family" CBS segment.For Rachel Maddow's speech, 91 tokens were collected.All statistics and graphs were done in R.

Results
Figure 1 shows the relationship between addressee and /s/ realization."CBS (acting/family)" is the non-LGBT topic segment with non-lesbian interviewers."CBS (LGBT)" and "Maddow" are the two interviews with LGBT topics, varied by interviewer sexuality.Means are indicated by a dot with the value below it.Horizontal lines represent medians.The two dots outside of the middle box represent outliers.
Figure 1 shows that the mean center of gravity for the acting/family segment was higher than in the other two contexts (M=7158.56).The mean for the non-lesbian interviewer context was higher than that of the lesbian interviewer context (M=6955.81 and 6733.09,respectively).This supports the hypothesis that Lynch uses a more retracted /s/ when speaking to a lesbian addressee than when speaking to non-lesbian addressees.
Figure 2 shows Rachel Maddow's /s/ realizations in comparison with Lynch's.Maddow realized /s/ with consistently higher Center of Gravity (CoG) measures than Lynch (M=7501.4 and 6733.09,respectively).
The linguistic factor of syllable position showed a small difference between coda and onset position tokens (see Figure 3), in keeping with Podesva and van Hofwegen's (2014) findings.in their study, onset fricatives have a slightly higher mean CoG than fricatives in codas.In all positions, however, these data continue to show the same pattern found above (i.e., highest mean CoG in the acting/family segment and lowest in the Rachel Maddow interview).

Discussion
The results above support the hypothesis that Lynch has lower /s/ realizations when talking to a lesbian woman than to non-lesbian on average, Lynch used a lower /s/ in the Maddow interview than in the CBS interview.Consistent with findings that GLB characteristics occur more in groups than individually (Podesva 2011, Munson andBabel 2007), the results can be interpreted on several axes.Bell's (1984) theory of audience design proves useful as the framework for one layer of analysis.In audience design, intraspeaker variation responds to interspeaker variation (Bell 1984).Here, the interviewers are the direct addressees, known and "ratified" by Lynch.In other words, although Lynch is being broadcast to thousands of viewers nationwide, these viewers are not present, nor do they converse with her.Here, Lynch appears to be "assess[ing] the personal characteristics of [her] addressees" (Bell 1984:167), i.e., their sexual orientation, and adjusting her speech accordingly, rather than responding to their use of /s/, specifically.Their sexual orientation conditions her speech, not through attention to specific linguistic variables (at least not in the case of /s/) but through their identity as an abstract concept of "lesbian" or "straight" (or more accurately, "not lesbian/not out as LGBT"), regardless of whether they employ the relevant variants or not.Linguistic data from Rachel Maddow's speech disfavors the "responding to particular /s/ frequency" interpretation: Maddow's /s/ realizations were consistently much higher than Lynch's.Instead, it appears that the concept of "lesbian" carries a number of social meanings for Lynch that she responds to, resulting in her own phonetic interpretation of a "lesbian style" in the absence of a socially marked "lesbian accent" (such as "gayspeak" in men).In this way, Lynch abstractly and conceptually, rather than phonetically, aligns herself with her lesbian interlocutor.This is underscored by the fact that in the interview with Rachel Maddow, Lynch shows strong agreement with Maddow's statements about gay rights and Obama's support for equal marriage: (3) Maddow: We are in a moment when the politics of gay rights are rapidly rapidly changing.
Lynch: I know.Fast, yeah, really fast.
(4) Maddow: It was one of those things, covering it as a gay person, thinking about it as a gay person, and covering it as a person in the news, I felt like half of me had to have a talk with the other half, because like -Lynch: Absolutely.
(5) (Lynch in parentheses) Maddow: The way people are still talking about it in pundit world, is how many people agree with the president (right), it's going to help him with them (right), and how many people disagree with him, it's going to hurt him with them, and the gay part of me, the sort of human part of me (yeah) wants to explain, you know what, the big issue here, the big question, is not who already agrees or disagrees with him, but who is still changing their mind (yeah), who is persuadable (yeah), and is he going to persuade anybody, does this change history (yeah).
Although Lynch shows "positive" alignment towards Maddow as a "fellow lesbian", the opposite cannot necessarily be said for the non-lesbian interviewers.Data on the two non-lesbian interviewers' speech was not collected, but based on Lynch's speech in the acting and family segment, she does not appear to "negatively" respond (i.e., diverge).Although it cannot be said that she has any one truly "unmarked" style, Lynch's lack of divergence can be taken as a reflection of heterosexuality as the societal norm: a lesbian interlocutor is more marked than a straight interlocutor in current society and therefore more likely to cue the "marked" variant (Podesva 2011).In other words, Lynch responds to Maddow, but she does not appear to show hostility or motivation to diverge from King and Hill.Indeed, in the Maddow interview, Lynch expresses a positive and hopeful view towards straight people in the LGBT rights' debate: (6) Lynch: I think a lot of it -I think a lot of people change where they're in the presence of one of us.They get to know us, and they get to know our family, and um, I think that's you know being -that's why I think being out, for me anyway, is important, and you know, uh, allowing myself to, you know, appear in public with my family not hiding about it, and I think that's really important.
As King's and Hill's speech was not analyzed, however, it would be premature to point to any one conclusion as the correct one.This framework sees Lynch's behavior as largely responsive rather than active.However, Lynch also uses /s/ as one tool to assert a non-heteronormative identity.In conjunction with addressee design, this creates a mutual sense of identity and membership, which exclusivizes herself, Maddow, and the wider LGBT community in opposition to the heteronormative society that Munson (2011) describes as "low-grade homophobic".This is distinguished from a constructed "persona" (Podesva 2007) in that it is not a "calculated" performative role, such as the diva persona in Podesva's (2007) study on a gay man's use of falsetto.Lynch rather uses /s/ to orient herself and identify with the LGBT end of the spectrum, as opposed to the "L" alone.The concept of indexicality also plays a role here.Lower /s/ broadly indexes masculinity, as reported in perception studies (Zimman 2013).This, in Lynch's case, could also more narrowly index a lesbian identity, or at least a non-heteronormative one.This application makes sense in light of what Lynch says about gender in the CBS interview: (7) King: But growing up you said "I wish I was a boy" because you were more comfortable -Lynch: I did, yeah, I wanted to be a boy, I used to sneak into my dad's closetoh how metaphoricaland put on all of his clothes … and I would take his cocktail glass and put water in it and ice, and hold a cigarette and look at myself in the mirror and go "now I feel like myself".
(8) Lynch: I think I do walk that male-female line a lot, I -I have a lot about me that you could categorize as masculine energy, and then I also have uhyou know, I'm my own little hodgepodge, as everybody is.
This exchange implies that Lynch sees gender and sexuality as related factors in her divergence from a heteronormative paradigm.Pierrehumbert et al. (2004) observe that lesbian speech is not wholesale shifted towards male norms (though they did not investigate /s/).However, based on this interrelationship, it seems reasonable to conclude that /s/ has multiple indexical meanings for Lynch that cannot be easily separated, but that work together as a variable to mark solidarity in sexuality (regardless of the interlocutor's, or at least Maddow's, linguistic behavior).Such an indexical meaning (of non-normative masculinity) has been found for /s/ in lesbian women in a previous study (Podesva and van Hofwegen 2014).
The CBS segment focusing on acting and family showed a higher mean /s/ CoG than in both other contexts.Though not the focus of this paper, it is interesting to note the effect of topic on the CoG values between two segments controlled for virtually every other factor (e.g., news/talk show audience, time, and gender).This segment is the most unmarked of the three in terms of heteronormativity: non-lesbian interviewers discussing non-LGBT topics.The results from this segment show that topic affected Lynch's realization of /s/ in the LGBT CBS segment, an effect which must also have taken place in the Maddow interview and been subsequently compounded by addressee effects.
There is also the question of salience.Podesva (2006) says that a variable must be salient to have social meaning.In the case of Jane Lynch, lower /s/ realizations are not overtly obvious to the listener and float beneath the level of consciousness.Does this mean they carry no social meaning?The realization of /s/ has been shown (as cited previously) to affect listener judgments of gender and men's sexual orientation.This suggests that it is salient or at least perceivable on some level, even if speakers and listeners are not directly aware of the variation themselves.In Munson et al. (2005), /s/ was not a significant predictor of listener judgments for women's orientation, suggesting that for some reason, whether acoustic or social, /s/ is not salient in the context of women's speech.However, /s/ in combination with other linguistic variables, such as vowels, may indicate a salient style.Since Lynch does vary as expected, it can be posited that /s/ realization carries some social meaning, such as "not-straight".

Conclusion
This study used center of gravity measurements for /s/ to show that Jane Lynch used /s/ realizations with relatively lower spectral means when speaking to another lesbian woman than to non-lesbian women.The syllable position of the fricative was shown to have a small effect on CoG measurements, but the overarching effect of addressee held true even with this distinction in place.The results were examined through several interconnected lenses, combining Bell's audience design, indexicality, and identity-marking to provide a cohesive explanation of Lynch's variation.Topic effects were also briefly discussed between two segments of the same interview.It was argued that Lynch considers gender and sexual orientation along the same axis and uses /s/ as one tool of possibly many to construct a unique lesbian or non-heteronormative identity.The question of salience was also considered, as was the consequence of /s/ realization being generally below the level of consciousness for its potential social meaning.It was concluded that /s/ can and does carry some social meaning, possibly in combination with other variables in a "style", given that it has been found to be relevant in some perceptual studies (e.g., Mack and Munson 2012).

Future Directions
There are many more aspects to be explored than the scope of this study has allowed.In future, it would be useful to look at the non-lesbian interviewers' linguistic behavior and whether they diverged or converged between topics.In order to provide more robust findings, it would also be useful to measure other spectral moments, such as skewness, which has been shown to be relevant in perceptual studies of /s/ (e.g., Mack and Munson 2012).Repeating this study with other lesbian TV hosts, such as Ellen DeGeneres, while in interview with either LB or straight women would also build on this research.
A fuller analysis of style would further require measuring of vowels, especially the back vowels previously found to correlate with LB identity or perception of LB identity (Pierrehumbert et al. 2004, Munson et al. 2005).This topic could also benefit from a third-wave analysis of style if more variables were analyzed.Sociolinguistic studies such as this could also help inform new directions in laboratory phonology; and it has been claimed that sociophonetics is the next challenge in such studies (Pierrehumbert and Clopper forthcoming, cited in Fuchs and Toda 2010:282).In general, literature on the speech of LB women in laboratory or social contexts is needed to develop this under-studied research area further.

Figure 1 :
Figure 1: Center of gravity measures for /s/ in each context, with means.

Figure 2 :
Figure 2: Center of gravity measures for /s/ for Jane Lynch and Rachel Maddow in the Maddow interview.

Figure 3 :
Figure 3: Center of gravity measure for /s/ according to syllable position for each context.