Working-Class Heroes : Intraspeaker Variation in General

We examine “Liverpool lenition” in the speech of Len McCluskey, a speaker of “Scouse”. Scouse is a variety of Liverpool English associated with the working-class persona of the “Liverpudlian”. We hypothesised that McCluskey, General Secretary of the trade union Unite, would use Scouse relatively more often when speaking to an audience sympathetic to the Labour Party than to an audience that is not. We analysed his rate of lenition in two social settings: speeches to a Labour-sympathetic audience and interviews with a non-Laboursympathetic audience. We find that McCluskey’s spiranisation is more frequent in the speeches than in the interviews. We argue that lenition is a resource for the construction of a working-class persona that may be beneficial for political purposes.


Theories of Intraspeaker Variation
Intraspeaker variation has been accounted for by several approaches in sociolinguistics. Where Labov's (1966) attention-paid-to-speech has been criticised for giving a unidimensional approach to intraspeaker variation (Bucholtz 2009:1), Bell's (1984) audience design theory, though more concerned with context in intraspeaker variation due to its focus on audience, has been criticised for underestimating "the contributions of individual participants in speech-events" (Coupland 1985:153). Bell's (1984) audience design theory distinguishes between different audience types: the addressee (directly addressed listener), the auditor (indirectly addressed listener), the overhearer (non-ratified listeners of whom the speaker is aware), and the eavesdropper (nonratified listeners of whom the speaker is not aware). Audience design argues that intraspeaker variation is the result of a speaker adjusting their speech to their audience according to these different audience types. This theory is based on Giles et al.'s (1991) communication accommodation theory, which explains intraspeaker variation in terms of "convergence" and "divergence". Convergence can be characterised as the strategies through which interlocutors adjust their language in order to reduce what social differences between them are linguistically evident. Divergence, on the other hand, can be characterised as the strategies whereby the social differences between parties are emphasised through language. Since these are theories that focus on context as a trigger for intraspeaker variation, Schilling-Estes (1998:73) maintains that at least equally as important is the shift in style that the speaker undertakes without an apparent contextual trigger: "style-shifts undertaken in the service of identity projection are not necessarily or even primarily conditioned by changes in contextual factors external to the speaker". This calls for a theory that affords the speaker greater agency. Speaker design is such a theory, proposed in response to earlier theories such as Bell's (1984) audience design theory, and it places emphasis on speaker agency in intraspeaker variation.
Speaker design refers to a theoretical approach to intraspeaker variation that regards speakers as proactive in their style-shifting (Schilling-Estes 2008a). Further, speaker design focuses on how speakers use variation to shape themselves and their environment (Chambers and Schilling 2013). Speakers orient themselves with respect to complex indexical meanings that are inferred from a certain variety or its features. As such, language varieties and their features become the tools through which a speaker's identity is built (Schilling-Estes 2008a). Indexicality is related to the speaker's ability to use "the social meaning of linguistic forms" in order to "construct personas" (Bucholtz 2009:2), or identities. Kirkham and Moore (2016) explore this kind of identity construction through the linguistic variation of the UK Labour Party's former leader, Ed Miliband. They examined Miliband's use of glottal variants of /t/ and alveolar variants of /t/ in a speech to representatives of the Labour Party and a speech to a more politically neutral British audience, arguing that Miliband's increased use of the alveolar variants of /t/ rather than glottal variants functions to build his persona as an "articulate and reliable" politician to a Labour-Party-oriented audience (Kirkham and Moore 2016:108). Speaker design approaches define identity along "personal and interpersonal dimensions" created during social interaction (Schilling-Estes 2008a:388). Emphasising the interpersonal dimension, Schilling-Estes (2008a) highlights the social character of identity as created in relation to others. Coupland (2009:288) states that these personas, or identities, can be seen as a performance. Thus, speakers "perform" language by using features of a certain style that indexes a certain variety. Groups of people who speak a variety often have traits associated with them, which can be drawn upon by using the features of that variety, thereby creating an associated identity. In addition to creating identities by using linguistic features that have social meaning, speakers can also use certain features to distinguish and distance themselves from identities they do not want to be associated with (LePage and Tabouret-Keller 1985). We adopt speaker design as an approach in which the speaker has agency in intraspeaker variation and evokes social constructs.

Speaker Background
The speaker of this study is Len McCluskey, the General Secretary of the trade union Unite. McCluskey joined the Labour Party in the 1970s (Hernon 2009), and has been the General Secretary of Unite since 2011, having been re-elected in 2013. Unite has, as it is called on their website, "an historic link" with the Labour Party ("Unite and the Labour Party" n.d.). This link between Unite and the Labour Party is evident through the Trade Union and Labour Party Liaison Organisation, a Labour Party organisation which financially supports activities and campaigns by unions openly affiliated to the Party. As such, Unite receives benefits from this organisation. In addition, Unite has a formal role in the Labour party: Unite delegates attend local Labour Party meetings, have a place on the National Executive Committee, and play a part in policy commissions.
Before McCluskey became a full-time union official in the 1980s, he spent years working at the docks in Liverpool, his birth city. McCluskey, reflecting on this time, asserts that he built solidarity with the dockers: "I was politicised by the docks. They were great years. There was tremendous solidarity amongst us dockers" (Hernon 2009). After he moved to the union's London headquarters in 1990, his love of Liverpool city remained (Hernon 2009). McCluskey grew up in Liverpool, speaking "Scouse"-the variety of English spoken in Liverpool. McCluskey is a fascinating subject for study, since his speech shows features of Scouse, revealing his Liverpool background and indexing a working-class identity.

Scouse, Unite, and Working-class Identity
The enregisterment of Scouse, a process "whereby distinct forms of speech come to be socially recognized (or enregistered) as indexical of speaker attributes by a population of language users" (Agha 2005:38), resulted in the indexing of a working-class identity in Liverpool. According to Crowley (2012:107), due to newspaper publications in the early to mid-twentieth century, Scouse became the index "not simply of Liverpool identity, but of Liverpool working-class identity embodied in the figure of the dockworker and of the 'real' Liverpudlian, notably distinguished from the middle-class 'Liverpolitans'" (Crowley 2012:107, our italics). A number of TV series (such as The Liver Birds) from Liverpool from the 1960s and 1970s, in which working-class Liverpudlians were featured, played a crucial part in the enregisterment of Scouse. The working class can be defined as the class comprising "people employed for wages, especially in manual-labour occupation and skilled industrial work" (Oxford Living Dictionaries n.d.), although it is often used to also refer to a social class consisting of a large majority with little financial means. McCluskey's time on the Liverpool docks qualifies him as a "Liverpudlian" influenced by the community, culture, and spirit of the docklands of Liverpool.
McCluskey's long-time membership in the Labour Party provides an indirect link between the working class and Unite. In addition to McCluskey's reputation as a supporter of the working class, the Labour Party is traditionally rooted in the working class (Lilleker 2002). Unite confirms this link on the website: "Unite members are encouraged to take a leading role in the Labour party at every level so that all parts of our democracy reflect Unite's working class values" ("Unite and the Labour Party" n.d.). We propose that McCluskey's style-shifting is governed by the notions of enregisterment and indexicality, so that he creates the identity of a Liverpudlian, i.e., a member of the Liverpool working class, by using certain Scouse linguistic features which serve as indexes.

Social Setting
To measure McCluskey's variation with respect to social setting we operationalised setting in a binary way: interviews and speeches that differ with respect to audiences. The interviews were filmed as part of the BBCs Newsnight programme and have thus been broadcasted to several thousands of viewers (Glover 2014). These interviews were also uploaded to YouTube on BBC's channel after the live broadcasting, increasing the potential viewership. Apart from the interviewer as a direct audience, then, the interviews have been made widely available to the public, forming a national (or probably international) audience. The BBC is claimed to be independent from political or commercial influences and is required to remain impartial in their reporting. Although the BBC has been accused of political bias by several media sources (e.g., The Guardian, The Independent, The Telegraph), the accusations assert that there is bias in the direction of both left-and right-wing preferences, so that there is no general consensus concerning the direction of the BBC's alleged political bias. In addition, many of the publications accusing the BBC of bias have their own known political affiliations. Thus, we presume the BBC to be politically impartial, and the wide viewership of Newsnight to comprise a general audience not specifically leaning towards (or against) the Labour Party.
Interestingly, one of the interviewers, Jeremy Paxman, could be said to have leanings towards the conservative party. While he has "had to remain politically neutral during his journalism career" (Sweney 2014), it has long been suspected that he leans towards the conservative party (Neville 2014). During his last appearance on BBC Newsnight, aired on 18 June 2014, he dodged confirming the statement put to him directly by Boris Johnson that he was "the last 'one-nation Conservative' at the BBC" (Sweney 2014), but later admitted that this was true to a reporter, after his retirement from the programme. He even commented that Newsnight was "made by idealistic 13-year-olds who foolishly think they can change the world" (Johnston 2014). That Paxman (McCluskey's main addressee) holds conservative political leanings may have influenced McCluskey's approach to the interviews, depending on his awareness of Paxman's views clashing with his own. In either case, the interviews can be said to take place in a social setting in which allegiance to the Labour Party is not assumed, as compared to the second social setting.
It is clear that McCluskey's two speeches, on the other hand, are directed towards a more Labour-Partyallied audience than the interviews are, given that the speeches are directed to proponents of Unite, which is openly affiliated with the Labour Party. The speeches (the second social setting type in this study), had a bigger live audience than in the interview setting, as they were directed to a large crowd at rallies. As stated in the description of one of the YouTube clips (UniteTheUnion 2015), one speech was directed to "Unite officers and organisers". In addition, these speeches were uploaded to the Unite YouTube channel shortly after their occurrence.

Hypotheses
The connections between Scouse, the working class, the Labour Party, and Unite raise two questions with respect to McCluskey. First, does McCluskey style-shift between his use of Scouse in the context of the Labour Party and one that is politically neutral? Second, with what purpose or goal could McCluskey be style-shifting with his use of Scouse? We hypothesise that McCluskey will use Scouse features more frequently when his speech is directed to an audience that is sympathetic to the Labour Party than when it is directed to an audience that is not. Specifically, we propose that the identity McCluskey creates by incorporating Scouse features into his speech is one of the working class, in line with his own background, and easily reconciled with membership of the Labour Party. Our results indeed support this hypothesis.

Liverpool Lenition
Scouse has several prominent features, including TH-stopping, non-rhoticity, the absence of contrast between certain vowel groups, and Liverpool lenition, arguably the most characteristic of Scouse (Watson 2007:352), which we focus on here. In Liverpool lenition, there are two prominent characterisations of the type of plosive lenition seen in Scouse: affrication and spirantisation (grouped here under the term "fricativisation"). As such, Honeybone (2007:19) suggests that the processes involved "can best be understood as involving stages of phonological lenition". Honeybone and Watson (2013:13) suggest that Liverpool lenition is the most distinctive feature of Scouse: "arguably the largest difference between the consonantal system of Liverpool and those of elsewhere lies in the realisational potential of its plosives". Presently, the following consonants undergo Liverpool lenition: Thus, the present study employs Liverpool lenition as an indicator of Scouse to investigate patterns of intraspeaker variation in the speech of Len McCluskey. /t/, /k/, and /p/ are employed as dependent variables, and are studied in relation to two independent variables: social setting and word type. The social setting consists of two speech settings: interview and speech, while word type is operationalised as the classification of content and function words.

Word Type
Word type-content versus function words-is also taken into account as an independent variable with the potential to condition the rate of stop lenition. Word type might affect the lenition of /t/ in Liverpool, because /t/ in pre-pausal position (i.e., preceding a pause) can be further debuccalised in "monosyllabic (pseudo) function words with short vowels (e.g., it [Ih], what [wAh], not [nAh], that [d1ah], lot [lAh])" (Watson 2007:353). The difference between function words and content words might also relate to effects of lexical frequency, since function words are generally more frequent. Hay et al. (2010:56) note that "reductive phenomena should be most prevalent in frequent words". While we do not test for word frequency directly, any difference between function and content words might suggest the need to do so in future work.

Linguistic Variables
The linguistic variables /t/, /p/, and /k/ are coded in a binary way: either fricative or non-fricative. Tokens are coded as fricative in the case that /t/ was realised as [θ̳ ], /k/ realised as [x], and /p/ realised as [ɸ]. In this study, we only included tokens in environments where spirantisation, the phonological process through which a plosive sound becomes a fricative, is most salient and least complex. Such linguistic environments consist of non-cluster, word-final positions (Honeybone 2007). Accordingly, we coded the data taking these factors into account. All tokens were coded manually by three different coders (one of whom is a native speaker of English), using Praat (Boersma and Weenink 2011). All three coded and compared the first minute of one speech and of one interview to test for inter-rater reliability. We found some tokens of affricated /t/ (realised as [tθ̳ ]), but we excluded them in this preliminary study for simplicity, because 1) the exact patterning of affrication and spirantisation is "rather complex" (Honeybone 2007:19), and 2) distinguishing an affricated /t/ and a released [t h ] in recordings with worse sound quality and background noises can be tricky. Therefore, we decided to focus on full spirantisations and only code tokens as lenited when there was no visible or audible burst as a fricative. In some cases, fricative realisations might appear different audibly and visually (for instance, one might sound like an affrication, but there is no visible indication of a burst on the spectrogram, like the one shown in Figure  1). In summary, only target tokens that meet the following criteria were coded: (1) non-cluster (2) word-final (3) no visible or audible plosive After the restrictions were agreed, a second pass showed inter-rater reliability to be 91%. Some tokens were coded as ambiguous and later excluded from the corpus. Situations include: (1) tokens preceded by sibilants (/s/ or /ʃ/) and not followed by a visible pause (2) tokens lying near the boundary of spirantisation and affrication (3) debuccalised /t/ tokens (realised as [h], and also the ones lying near the boundary of a glottal stop and fricative), as we were only focusing on spirantisation

Social Variables
The interviews listed in Table 1 were given by the broadcaster Jeremy Paxman and his successor Evan Davis, both speakers of (Modern) Received Pronunciation (RP), 2 in which /p/ and /k/ are mainly released in word-final position and /t/ can be pre-glottalised in coda position. We ensured that all speech data were captured after McCluskey's election in 2013, so that his political motivations remained relatively stable across recordings. In addition, using speech extracts from a narrow timeframe minimises variation as a result of lifespan change. Finally, we selected clips of roughly equal length in order to analyse the amount of tokens in relation to time. As a result, we had four clips of 5 to 7 minutes each. Table 1 shows the details of the data. Speech to 150,000 people participating in a rally against the Tories Finally, word type was operationalised as the distinction between function words and content words. Pronouns, conjunctions, prepositions, and some adverbs are coded as function words, while nouns, verbs, adjectives, and adverbs are coded as content words.

Results
We coded a total of 227 tokens that met the criteria outlined in Section 3.3: 173 for /t/, 39 for /k/, and 15 for /p/, of which 121 were from the interviews and 106 from the speeches. Table 2 shows the realisations across all plosives by social setting. The results in Table 2 do seem to support our hypothesis, indicating that more spirantisation occurs in the speech settings directed to union workers and Labour Party supporters than in the nationwide broadcasted interview settings. This finding is demonstrated in Figure 2, which indicates that 37% of the tokens were spirantised in the speech setting, compared to just 14% in the interview setting; this difference is significant (χ 2 = 14.53, df = 1, p < 0.001). Word type was also tested and showed a significant effect on McCluskey's plosive realisation (χ 2 = 31.18, df = 1, p < 0.0001). Word-final plosives are more likely to be spirantised in content words than in function words: 42.1% of the content word plosives were spirantised, compared to merely 9.2% of function words (Table  3 and Figure 3). This suggests that content words favour the fricative realisations of word-final voiceless stops more than function words do.

Discussion
We have demonstrated that the Scouse feature examined in the present study is more frequent in Len McCluskey's speech when it is directed to an audience sympathetic to the Labour Party than when it is directed to an audience that is unlikely to have this leaning. One reason that may help to explain the increased use of Scouse in the speech setting is that content words may be more likely to contain spirantisation than function words, and the proportion of content words coded in the speech setting is much greater than that in the interview setting. The spirantisations could reflect McCluskey's greater emphasis on content words, since a fricative can last much longer than a plosive or a glottal stop. The following extract illustrates McCluskey's spirantisation in content words in his speech, in which 1 indicates a spirantisation and 0 indicates not spirantised realisations:

Extract 1
Our union is over the next fourteen weeks facing maybe the biggest fight of our lives as Unite. 1 1 It's a fight not just for our union or its members; it's a fight for the future of our society.
1 0 1 From the extract, we can see that McCluskey uses the fricative variant of /t/ throughout the extract but glottalises the final /t/ in "not". This suggests that spirantisation might occur more in words that carry more meaning, especially when the speaker emphasises them to convey a political message, thus influencing the increased use of Scouse in the speech settings.  Hay et al. (2010) argue that frequent words are more prone to monophthongisation due to a combination of ease of processing and repeated opportunity of presentation, resulting in lexical items stored as single units. This leads to function words, which tend to be very frequent, being monopthongised, i.e., reduced. Similarly, the frequent glottalisation of coda /t/ in function words (cf., Extract 1) in our data can be caused by the easy processing of frequently used words rather than a content/function dichotomy, and further analysis on this might be conducted in the future.
We can explain McCluskey's decreased use of Scouse in the interviews by first drawing on Giles et al.'s (1991) accommodation theory and Bell's (1984) audience design theory. McCluskey's decreased Scouse use in the interviews indicates his convergence towards the RP-speaking interviewers. McCluskey here has a direct audience that is not predisposed to the Labour Party, and might even be politically biased towards the Tories, at least in the case of the interview with Jeremy Paxman. Therefore, the need to evoke a working-class image to his audience as a strategy to oppose the Tories is not politically beneficial and potentially to McCluskey's detriment. Finally, the BBC is not a regional station but a national one, suggesting that the viewers are also not particularly aligned with a political party, thus invoking the need to sound more "standard" (or "BBC-like"). The convergence can be in response to the auditors and the addressees, according to Bell's (1984) audience design theory. The purpose of McCluskey's convergence to the auditors could be to communicate his political message to the general audience, which is not directly present. Additionally, McCluskey's decreased use of Scouse in the interviews could be a response to the addressees, i.e., the RP-speaking interviewers.
McCluskey's increased use of Scouse in the case of his speeches can also be seen as a response to a working-class audience, in accordance with Giles et al.'s (1991) accommodation theory andBell's (1984) audience design theory. His construction of a working-class persona in the speeches occurs in response to the Labour-Party-oriented audiences, most likely consisting of working people, which is in line with the emphasis placed on the influence of those listening to the speaker in audience design (Bell 1984). As in the interviews, McCluskey's orientation to the audience can be characterised as convergence to that audience. Yet, although we recognise that the audience plays a crucial role in McCluskey's phonetic variation, since the working-class addressees are a crucial factor in McCluskey's use of Scouse in the speeches, we advocate the speaker agency that speaker design affords the speaker.
Speaker design provides a comprehensive account of McCluskey's intraspeaker variation in terms of his construction of a working-class identity. Through speaker design, one can "position oneself with respect to wider social groups or societal values or norms" (Schilling-Estes 2008b:975), create identities, personas, and one's environment. In both speeches, we propose that, as General Secretary of Unite and a member of the Labour Party, McCluskey positions himself as a proponent and advocator of working-class rights, such as fairness and decent wages for workers. As the results show, he does so all the while incorporating characteristics of Scouse, thereby indexing a working-class identity into his speech. Furthermore, in one of the speeches, McCluskey claims to fight "for the poor and the vulnerable", for people who "have little", and for "decent wages", emphasising the importance of the rights of the working class. As such, McCluskey creates a persona that prioritises values important to the working class.
Speaker design additionally considers why the speaker engages in style-shifting, that is, with what purpose or goal (Schilling-Estes 2008a). We propose that the goal of McCluskey's construction of a working-class identity through style-shifting can be seen as a political strategy to distance himself from the Tories. Speaker design theory can account for style-shifting by the speaker positioning him/herself against a certain identity or social group (LePage and Tabouret-Keller 1985). First, McCluskey characterises the Tories as a minority, juxtaposing them with a "movement of millions". This coincides with our earlier definition of the working class in Section 1.3, in which the working class work for wages, are the large majority, and have little financial means. In one speech, he describes the Tories as "plotting a reduction" of the role of the state that can only occur by dispensing with the trade unions in their current state. Therefore, McCluskey is not (merely) polarising the Labour Party and the Tories, but, more importantly, Unite and the Tories. Further, McCluskey's rhetoric constructs an "Us/Them" ideology (Van Dijk 2006) by emphasising his disagreement with the "elite", "the rich", and "corporate greed". His increased use of Scouse in the speeches, which indexes a working-class identity, can be seen as a political strategy that strengthens this opposition to the Tories.

Conclusion
This paper has examined intraspeaker variation in the rate of use of Liverpool lenition, ultimately through the lens of speaker design theory, while also drawing on Bell's (1984) audience design and Giles et al.'s (1991) accommodation theory. Our subject, Len McCluskey, style-shifts with respect to this feature of Scouse according to different social settings. We hypothesised that McCluskey would use Scouse features more frequently when his speech was directed to an audience sympathetic to the Labour Party than to an audience that is not. Our data supports this hypothesis in that more instances of /t/, /p/, and /k/ are realised as fricatives in the speech setting (37%) than in the interview setting (14%). This significant difference between the social contexts might be, at least partially, explained by the different word types that were included in the analysis. First, spirantisation occurs more frequently in content words, influencing the increased use of Scouse in the speech setting in which certain content words, such as "fight", are particularly common. Second, the particularly frequent use of function words (where word-final /t/ is most commonly realised as [ʔ]) in an interview setting, in which speech is perhaps more improvised and spontaneous, could contribute to the higher percentage of Liverpool lenition than in the speech setting. The current study does not properly model and account for other processes of Liverpool lenition on the lenition continuum (affrication and debuccalisation), which should be included in future studies. Finally, it must be noted that, according to Eckert (2008:471), plosive release occurs regularly to create emphasis. As such, we acknowledge that the noticeably increased vocal effort in the speeches could affect McCluskey's plosive realisation, making it a potential confounding factor, while it also depends on whether this type of mapping could be said to manifest similarly with spirantised releases.
We believe that McCluskey style-shifts first and foremost to orient to particular social meanings associated with group identity (Chambers and Schilling 2013). We characterise this group identity as a working-class identity. When McCluskey directs his speeches to Labour-Party-friendly audiences, his increased use of Scouse forms his identity as a "Liverpudlian" trade union leader. In addition, the themes he addresses, such as fair wages, emphasise working-class values. We believe that he does so ultimately as a political strategy to distance himself from the Tories. In contrast to the speeches, the audience in the interviews is more politically neutral. This is likely to account for the fact that McCluskey exhibits less frequent Liverpool lenition in the interviews compared to the speeches. By exhibiting less Liverpool lenition, McCluskey sounds more "general" and can appeal to a more general audience, which is politically favourable. Future studies should further investigate how the construction of an identity might affect speakers' speaking styles by analysing more linguistic variables such as word frequency and taking into account different levels of vocal effort.