Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

In this chapter, I examine one sociolinguistic niche that has been somewhat downplayed in mainstream work, but as I will show below, it is one that has interesting ramifications for an understanding of the complexity of language in social life and the progression of linguistic change. The focus here is on the elite sociolect of the UK, the generational successor to Received Pronunciation (RP), also known as Standard Southern British English (SSBE). Taking a viewpoint that social class (admittedly a complex concept, as the debates in Skeggs 2015 show) continues to manifest in sociolinguistic life in the UK, I examine here a selected set of sociophonetic changes that characterise the history of the elite sociolect. It is trivially true that all language varieties change; the point of interest in this chapter is the sociolinguistic ramifications of the continued existence of elite sociolects, and whether they continue to signal and construct social difference in the community. Our claim here is that, far from being entirely levelled to other social varieties in the south of England, for example, these voices are still distinct and sociolinguistically significant.

Theoretical Preliminaries

Fifty years of sociolinguistic research have shown how we can see language practice, language ideology, social fabric and social practice as intertwined, mutually constitutive semiotic processes ebbing and flowing in the course of history (Labov 1994, 2001; Eckert 2008). Social processes such as large-scale urbanisation in the nineteenth and twentieth centuries, the upheavals of the Second World War and de-industrialisation in the late twentieth century had large impacts upon the human landscape of Britain, with waves of de-dialectalisation, dialect levelling and regionalisation as some of the sociolinguistic consequences (Trudgill 1986; Britain 2016; Coupland 2014, 2016). Multi-ethnic immigration is also presently bringing about linguistic transformations of many kinds (this is especially well researched in the UK: see e.g. Cheshire et al. 2011; Rampton 2011; Kerswill 2013). These large-scale social movements have had consequences for the entire sociolinguistic landscape of the UK.

As one case in point, on a political level in the UK at the moment, there seems to be a striking contrast to social class discourses of the late 1990s, when I began researching RP sociolinguistically. At that time, one dominant political current was encapsulated in the critiqued concept of the ‘meritocratic society’ (Adonis and Pollard 1997). Accompanying this was widespread talk in the media and among language experts of Estuary English as a levelled local replacement for RP in younger generations of speakers in London and the Home Counties (Kerswill 2001; Przedlacka 2002; Altendorf 2003). Writer India Knight in 2001 reported anecdotally that ‘the only accent it is now actively all right to pillory is the so-called posh–the clear enunciation that comes from being privately educated or having upper-middle-class parents’.Footnote 1 At the time of writing this chapter, however, there has been something of a revival of interest in and a redefining of elite/establishment positioning in the public media sphere: television series such as Life is Toff, You can’t get the Staff and Posh People: Inside Tatler have all been broadcast in recent years.Footnote 2 This is perhaps not coincidental, given the political context of the neo-Thatcherite premierships of old Etonian David Cameron from 2010–2015, and under a Conservative Party majority from May 2015 to July 2016. Class and class inequality are in the news again. At the same time, an academic sociological interest in elites and in class structure (Savage et al. 2013; Skeggs 2015) has intensified recently in the UK, alongside the Brexit vote of 23 June and the political upheaval it has caused so far, leading to Theresa May’s premiership from July 2016.

What sociolinguistic implications and consequences can possibly be extrapolated from this history? What if, to paraphrase Coupland (2000: 264), where he writes: ‘(é)lites perpetuate élite society by being seen to be élites’ (and contra to the Estuary English discourse of 20 years ago), élites still perpetuate elite society by being heard to be élites? Can sociolinguistic research identify continuing or renewed accent-stylistic dividing lines: clusters of phonetic features which alone or in combination as posh styles are enregistered in the UK, styles whose linguistic makeup certainly have changed from what they were in the first half of the twentieth century (e.g. Fabricius 2017), but whose distinctions function, if not as the reported accent-bar of the 1950s–1960s (Abercrombie 1965), at least as some sort of class-framing for speakers and listeners, as enregistered construct resources (Fabricius and Mortensen 2013)? Does class still resonate in the UK, and can sociolinguistics contribute to a nuanced understanding of new constellations of class and language?

In this chapter, then, I explore some of the quantitative research hitherto on this particular sociolinguistic niche. My summary of these findings look at changing sociophonetic features such as word-final /t/, weak vowels, the short vowel system and pre-vocalic, syllable-onset /r/, diphthong smoothing and yod coalescence. These socially distributed phonetic variations are of interest per se to linguists as sociophonetic changes with different trajectories in historical linguistic terms; their implications for wider social practice and the enactment of class remain unexplored, for instance by ethnographically oriented work, and this is a gap in the literature that I see as needing to be filled. The chapter also briefly discusses evidence of lifespan change in RP speakers, as well as recent language-attitudinal work on RP. What remains to be implemented is a wider agenda of theoretical research into elite sociolects, which will need to employ distinctions between social change, linguistic change and sociolinguistic change, the latter encapsulating changes in the sociolinguistic status and social implications of language forms (Coupland 2014). This is because the historical trajectory of an elite sociolect or an elite accent style is best understood through a wide linguistic anthropological and ethnographic lens, using a range of methods. I cannot do this research horizon full justice here, but I can outline some findings so far and suggest paths for future scholars to follow.

Standard Languages, Elite Sociolects and Language Change

Variationist sociolinguistics can be characterised as the linguistic study of variation and change in vernacular language varieties, understood as the systematic forms of language acquired by children as their first language(s) of socialisation. Pioneering survey-based studies such as Labov (1963, 2006), Trudgill (1972), and Macaulay (1977) concentrated on data from speakers in the middle of the social hierarchy. Upper-class speakers were often not included in survey samples, or were the subject of particularised studies (Kroch 1995), and as a result were regarded as much less interesting for mainstream variationist work for a long time, seen as being far from the locus of sociolinguistic change, conservative followers rather than first-movers. Similarly, Rampton (2009) discusses for instance a prevailing romanticisation of the working class in academic work of the time. Upper- and to some extent upper-middle-class speech also had a somewhat tenuous place within sociolinguistic thinking, because it was regarded as affected by conscious educational standardising processes that could modify the vernacular (understood here to mean the first language of socialisation).

This theoretical stance conflates two distinct sociolinguistic phenomena, however: the standard language and the elite/establishment sociolect. The tendency has been that upper-class groups’ sociolect and the standard language were often assumed to be identical, and these speakers have been excluded as being not vernacular (where vernacular could also be understood to mean non-standard) (Bex and Watts 1999; Milroy and Milroy 1999; Kerswill 2006). ‘The concept of the standard’ was therefore for a long time, as Milroy (2004: 162) has pointed out, ‘surprisingly underspecified and undertheorized’ in variationist sociolinguistics, and standard variants were presented as self-evident counterpoints to the non-standard vernacular variants which dominated the sociolinguistics literature. This ignores the fact that, firstly, upper-class and upper-middle-class speakers, of course, do acquire their own vernacular (meaning primary language of socialisation), and that successive generations of such speakers exhibit vernacular variation and change over time. These sociolinguistically embedded changes have always provided a challenge for linguists aiming at describing a standard codified variety, often for foreign language teaching purposes (Wells 1990, 1994, 1997; the title of the latter is, revealingly, Whatever Happened to Received Pronunciation?; Upton 2012: 55–58).

In Fabricius (2000), the distinction between elite vernacular sociolect and standard language construct was captured under the terms native-RP and construct-RP, given the need to separate the vernacular aspect ( phonetic features as part of a first language of socialisation), from the standard language, an abstract, explicitly codified and folk-linguistic model. Agha (2003, 2007) has since introduced the concept of enregisterment using RP as his canonical example. He claims that RP over time became enregistered as a folk concept, a recognised set of phonetic patterns, a certain type of voice in folk terms. The enregistered voice is indeed part of what construct-RP was intended to cover in Fabricius (2000, 2002a, b), but it also included codified manuals and dictionaries of the accent as explicit models, text-artifacts in Agha’s terms (2003) which themselves also function as vehicles of enregisterment processes in the chains of transmission in which they participate historically.

As Agha (2003) describes it, the process of systematisation/codification of the accent and its characterisation as received or authorised by an external authority is part of an anthropological mechanism that produces a standard accent ideal that is external to any one speaker . This sense of distance between ideal and reality eventually makes it easier for claims to be made that no one speaks RP any longer if RP is solely understood as a construct model that comes up short against the forces of variationist linguistic change and no longer matches the way people are hearing language being spoken in their everyday lives. If the term standard language/variety is reserved for such a socially-generated and historically-sustained mental ‘construct’, it can be kept distinct from the concept of an elite (or even establishment) sociolect. This latter term can then be reserved to refer to linguistic patterns evidenced in the first language of socialisation (i.e. the vernacular in that particular sense) of a social group occupying a particular socio-economic niche within a socially stratified society. The term is, of course, also an idealisation, since no group contains completely homogeneous or identical speakers , enabling the identification of one single sociolect shared by all. Until sociolinguists appreciated this c-RP and n-RP distinction, there could not be progress in the empirical, variationist study of elite sociolect pronunciations as part of the sociolinguistic makeup of society, not solely as a model accent for foreign language teaching purposes, which was one of the major reasons for its continuation in books such as successive editions of Gimson’s Pronunciation of English (in press continuously since 1962). This progress was needed: there is simply an acute empirical gap in understanding the sociolinguistic makeup of a class-stratified society if elite sociolect speakers are not represented. Debates within sociology about elites, class formation and social stratification in British society are particularly intense at the moment (Macionis and Plummer 2012; Lui 2015; Mills 2015; Savage 2015; Skeggs 2015; Wakeling and Savage 2015), and this renewed discussion of elite formation, the sociological perpetuation and reinforcement of class, and class disparities in Britain is prominently on the sociological agenda, and this should also inform British sociolinguistics in the future.

What’s in a Name?

As I have pointed out above, while the study and description of n-RP features has not been a major concern of sociolinguists until fairly recently, c-RP has very much been the focus of mainstream phonetic research since the earliest days of the phonetic sciences in Britain and Daniel Jones’ tenure at University College London . Cruttenden (2014: 77) writes that Jones’ efforts at description and codification of the accent were spurred by ‘increased interest in teaching English as a foreign language’. From the beginning, Jones’ descriptive publications, such as Jones (1914, 1917), were based on his own variety and that of close associates, which he at the time labelled (PSP). The term RP replaced Public School Pronunciation (PSP) by the time of Jones (1926), and Gimson’s Pronunciation of English from 1962 carried on this tradition. Under this name, the accent norm became well known in mainstream phonetics and in English-language teaching generally.

Wells (1990, 1994, 1997), acknowledging this tradition and heritage, discusses the implications for a pronunciation model of different ways of understanding the term RP: whether as a norm for foreign language teaching, as an ideal Platonic notion of correct speech or as a sociolinguistic concept, a function of a number of speaker characteristics centred on ‘ socioeconomic class, sex and age, perhaps with contextual style’ ( Wells 1994: 204). His clear preference was for a sociolinguistically-informed notion of RP, as this is the only one that could eventually incorporate changes (in many cases, originating from non-standard local varieties) in native-RP pronunciations, as they appear, disseminate and become established and standard, as t-glottalling has to some extent (Fabricius 2000). Wells (1990) recognised sociolinguistic continuity, as well as discontinuity, between speakers of RP in Daniel Jones’s day and the present time, implying that a sociolinguistically sensitive methodology would enable the description and codification of an evolving accent (n-RP) for explicit teaching purposes (c-RP).

The term RP has by no means remained unchallenged. As a result of the advent of the BBC in 1922 and its decision to apply a unified pronunciation norm for broadcasting purposes, BBC English came over time to be added to the terminological battery (for a comprehensive history of the BBC Advisory Committee on Spoken English, see Schwyter 2016).Footnote 3 The Oxford English Dictionary lists the term RP as first occurring in 1928, followed by three mentions during the 1930s. The term is still used to denote the accent in the Cambridge English Pronouncing Dictionary (Jones et al. 2011). The first Head of the BBC, John Reith wrote that ‘[t]he policy might be described as that of seeking a common denominator of educated speech’ (cited in Cruttenden 2014: 77).

Cruttenden (2014: 80) has recently argued in favour of Windsor Lewis’ (1972) appellation General British (GB), not as a different accent of British English from RP, but as ‘an evolved and evolving version of the same accent under a different name’. Cruttenden argues that the term RP is, in the public mind and for many linguists, a term used to refer to a particular sub-variety of RP: Wells ’ U-RP ( Wells 1982), Upton’s trad-RP (Upton 2012), the form spoken natively by an increasingly older population of aristocratic/upper-class origins, which Cruttenden himself labels Conspicuous GB.Footnote 4 Thus, he contends, the term RP is too restrictive and misleading (and age-based) to be of use for a more mainstream variety that is more widely socially based and more acceptable as a foreign language norm. In addition, the term General British parallels General American in the US, and Cruttenden argues that the two seem to serve somewhat similar social functions (although not everyone would agree with this point). Cruttenden (2014: 81) characterises Conspicuous GB (CGB) as ‘that type of GB which is commonly considered to be “ posh”, to be associated with upper-class families, with public schools and with professions which have traditionally recruited from such families, e.g. officers in the navy and in some army regiments’. Note that the definition of CGB as being found in certain families allows for the implication that it is still a vernacular for a young group of speakers (the generational successors). But Cruttenden clearly has an old-fashioned form of pronunciation in mind, as he states that it is mainly limited to older speakers. He provides a list of characteristic pronunciation features that can be said to typify CGB. An empirical sociolinguist should regard such a list as a set of empirical hypotheses that can be tested on language data from such groups of speakers. To take one instance, very open nurse vowels have recently been observed in a sociolinguistic interview of a female speaker born in 1990 and recorded in 2008 (Fabricius et al. 2012). This vowel quality is, therefore, not exclusively used by older speakers, although it is possibly more frequent there. The precise reach and quantitative distribution of such phonetic variants remain, however, subjects for future corpus-based research that is at present lacking.

These arguments for GB as a term for the codified norm notwithstanding, in this chapter, I regard the term modern RP as a more apt descriptive label for the evolving variable sociolinguistic phenomena represented by the speech of successive generations of sociologically identifiable speakers . This is precisely because the name makes an explicit link between present and previous distributions of accent features, the modern patterns being the historical, generational successors of the older patterns (in the same sense that modern speakers are descendants of older speakers). I leave aside the question as to whether these features form a consistently identifiable accent variety, since that question is perhaps more of an ideological one, a question of enregisterment (Agha 2003, 2007), and take the position here that I aim to simply describe elite sociolect features quantitatively and find out how they come together in sociolinguistic styles .

In addition, the issue of precise accent labels is something of a distraction, when the essential challenge of a sociolinguistic investigation could be said to be a secure sociological identification on criteria which keep it independent of linguistic form and thus avoid circularity. I argued in Fabricius (2000: 46–60) that socio-economic background and educational history would play a role here, but this is by no means uncontroversial, and I acknowledge it as a problem that needs to be tackled in more ethnographic depth. In operationalising this aspect for my original study, students at Cambridge University with independent school and upper-middle-class backgrounds (measured by examining parental occupations) made up the key set of speakers whose speech patterns (t-glottalling in particular) formed the basis of the quantitative investigation. Since these background factors were established independently, the task then became an empirical sociolinguistic one, and some of the findings that stem from this approach are detailed in the next section.

Quantitative Empirical Studies of Phonetic Variation in RP Speech

T-Glottalling

Fabricius (2000) provides an empirical study of variation in word-final /t/ in a single-age-cohort of speakers recorded in 1997 and 1998. As described above, these speakers had been educated at public and independent schools and were students at Cambridge University at the time of recording, aged between 18 and 30. Socio-economically, these speakers came from upper-middle-class family backgrounds: their parents (in the majority of cases, their fathers) had occupations at the upper levels of the Cambridge Scale for Occupations (Prandy, 1992), being for example barristers, solicitors, accountants, medical specialists and high-ranking civil servants (Fabricius 2000: 77–78, 163–164). Interview and reading passage data obtained from 12 male and 12 female speakers yielded 9888 tokens of word-final /t/, analysed auditorily. Analysis of variances (ANOVAs) explored social and linguistic factors that determined variation in rates of t-glottalling across groups of speakers. The results showed that the speakers in the 1997–1998 corpus used t-glottalling at a uniformly high rate pre-consonantally within interview style (60–70%). The utterance-final position (in the study, designated pre-pausal) showed greater variation between speakers , and this variation was shown to be regionally determined, with higher rates evident in speakers who had grown up closer to and in London. High rates of t-glottalling in the pre-vocalic environment in interview style were likewise restricted to speakers with London origins. In addition, pre-pausal and pre-vocalic (but not pre-consonantal) t-glottalling was widely avoided in reading passage style. The study concluded that there was support for the idea that the pre-pausal environment would stand to become the next widely acceptable environment for t-glottalling, perhaps within the next generation or two. In the study, however, pre-pausal and pre-vocalic t-glottalling were not in evidence in reading style. Moreover, since usage of pre-pausal t-glottalling in interview style did not show the same consistently high rates as the pre-consonantal environment, it was considered premature to accord pre-pausal t-glottalling the same sociolinguistic status as pre-consonantal t-glottalling.

Weak Vowel Variation

Fabricius (2002a) focused on the status of two ongoing processes of historical change within weak syllables that have been previously recognised in the literature on RP. The first was a change in the phonetic quality of the weak high front vowel within word-final V# syllables from [ɪ] towards [i], known as happy-tensing (Wells 1982). The second change, sometimes referred to as the drift from [ɪ] (also known as the kit vowel in the British tradition following Wells 1982) to [ə] in closed weak syllables ( Wells 1997: 18) is a change that has progressively affected a heterogeneous group of lexical words such as item and civil as well as certain derivational and inflectional affixes such as -less, -ness, -ily, -ed, and -es. Both of these changes have been reported as being generationally-based (e.g. Cruttenden 2001: 107). The data set consisted of four male and four female speakers, a subset of the interview corpus analysed in Fabricius (2000). Happy vowels in -y and -ly contexts, and variation between kit and schwa in plural -es and past -ed were measured acoustically. First and second formants were measured at the midpoint of duration of the vowel for kit, happy and schwa tokens and at the F1 maximum for fleece (following Labov et al. 1972; Labov 1994), which was used as a reference vowel for visual and statistical purposes. Comparisons were made in the data between values for F2-F1, as a reflection of the fact that more peripheral vowels have a higher F2 and lower F1, while more central vowels have a higher F1 and lower F2. F2-F1 thus gave a generalised measure of peripherality.

All speakers showed a tendency towards an intermediate or fronted value for the final vowel in -y words, either midway between the kit and fleece ellipses or with the majority of tokens within the fleece area. Some tokens, however, appeared more conservative, being within the range of kit. One male speaker’s data, shown below, gives an example of this range: three tokens of #V are within the kit area (indicated by + symbols), but the majority of tokens are either within the intermediate area, close to but slightly lower than fleece, or located within the hand-drawn ellipse for fleece (indicated by x). Similarly, the means and standard deviations data for this speaker show the #C environment as having the most advanced mean value for F2-F1, while the highest standard deviation and thus the greatest spread in these values is associated with the #V environment (Fig. 3.1; Table 3.1).

Fig. 3.1
figure 1

Locations of kit, fleece and happy tokens, one male speaker (Reproduced from Fabricius 2002a: 228)

Full size image
Table 3.1 Means and standard deviations in F2-F1, male speaker (Data from Fabricius 2002a: 228)
Full size table

In the case of kit→schwa drift, the data analysis demonstrated that there was no general evidence of drift towards schwa in the production of -es and -ed suffixes.

The data for one female speaker are presented in Table 3.2 and Fig. 3.2. Comparison of the means and standard deviations shows the suffixes to be centred on the same F2-F1 values as kit, in a pattern that corresponds to the six of the eight speakers in the corpus data, where the past and present/plural suffixes resemble each other. Two speakers, however, diverged slightly from this pattern and in a similar way. Both showed lower average F2-F1 values and greater standard deviations for the present/plural suffix than for the past suffix. The overall impression from the data was of a long-lasting process of change from kit to schwa that had to some extent stalled or at least become sluggish as far as -ed and -es suffixes were concerned. Since kit as opposed to schwa in such suffixes is a local British phenomenon that separates it from North American, Australian and New Zealand pronunciations, from a dialect contact point of view, maintenance of kit in these suffixes may be playing a role as a salient British feature.

Fig. 3.2
figure 2

Vowel plot, female speaker. Location of -es and -ed suffix vowels vis-à-vis kit +, and schwa x (Reproduced from Fabricius 2002a: 222)

Full size image
Table 3.2 Means and standard deviations for kit/schwa, female speaker (Data from Fabricius 2002a: 222)
Full size table

Changes in the RP Short Vowel System

Fabricius (2007) assembled a comparative corpus of formant measurements designed to give a real-time view of the changing short vowel system of modern RP during the twentieth century. The data derived from instrumental acoustic measurements of vowel formants in speech were obtained from the following sources:

  1. (a)

    Radio broadcasts by two male RP speakers from The Machine Readable Spoken English Corpus (MARSEC) corpus (Roach et al. 1993) analysed in Deterding (1997);

  2. (b)

    Elicited citation forms spoken by a homogenous set of 25 male RP speakers born before 1945—representing average values for the 25 speakers in the corpus of Wells (1962);

  3. (c)

    Elicited citation forms spoken by 20 male RP speakers in four age groups—representing individual values for citation forms (Hawkins and Midgley 2005);

  4. (d)

    Broadcast speech tokens from Queen Elizabeth II’s Christmas broadcasts over three decades: the 1950s, 1960s and 1980s (converted from Bark values given in Harrington et al. 2000);

  5. (e)

    Sociolinguistic interview speech tokens from four male speakers of modern RP from the Cambridge corpus collected in 1997 and 1998 (discussed above and in Fabricius 2000).

Data were obtained as either Hertz values or Bark-transformed data (Harrington et al. 2000); the latter were transformed to Hertz using a conversion table between Hertz values and Bark values based on Zwicker (1961).

The chapter presented the first use of an innovative analytical procedure using angle and distance calculations to represent the geometric relationship between two vowel positions. This was a combinatory method which made the statistical comparisons typical of mainstream phonetics accessible to the two-dimensional vowel plots approach of Labovian sociolinguistics (taken further by Nycz and Hall-Lew 2014). The visual comparison method has been widely used for examining vowel variation (Labov et al. 1972; Labov 1994; Watt and Tillotson 2001; Fabricius 2002a; Torgersen and Kerswill 2004). The benefits of visual comparisons for understanding change in vowel systems are considerably enhanced by methodologies that also allow replicable statistical comparisons.

To turn to the results of the study, the combined data showed that the short vowel space of these speakers fitting the RP sociolinguistic profile had undergone a change from an early configuration (with the earliest speakers born before 1920) with strut as the lowest point, through a phase in the mid-twentieth century with trap and strut on a similar low level, to a late configuration, with trap lowest and strut centralised characteristic of speakers born late in the twentieth century. trap/strut rotation was proposed as a label for this trend as a shorthand term for the lowering and backing of trap and the backing and subsequent raising/centring of strut .

Smoothing and Yod Coalescence

Hannisdal (2006) is a quantitative analysis of British newsreaders’ speech in broadcast situations, consisting of data from 30 speakers employed by three TV channels ( BBC, ITV, Sky). Six phonetic variables were investigated in all; the present brief discussion limits itself to two: smoothing or monophthongisation of diphthongs, and yod coalescence.

Smoothing (discussed in Wells 1982: 238–242; as cited in Cruttenden (2014: 160), the term first used in Sweet (1888: 22)), is the phenomenon whereby the diphthongs price and mouth, in combination with schwa, as in the words fire and power, respectively, undergo various degrees of monophthongisation to either a centring diphthong or monophthongal start /ɑː/. Hannisdal’s quantitative analysis of 1339 potential smoothing items showed that 46.4% were realised overall as smoothed variants (Hannisdal 2006: 200–203), and moreover, fire-type words were slightly more likely to be smoothed (48.7%) than power-type items (43.5 %).

Hannisdal (2006: 200) also noted a certain amount of individual variation in rates of smoothing. The most striking of these results was the male dominance of rates of smoothing: male speakers used smoothing at an average rate of 59% and female speakers at an average rate of 33.5%, consistently across TV channels. Hannisdal argues that the sharply gender-differentiated pattern reflects a stance towards articulatory explicitness, with women preferring full forms rather than smoothed forms as part of a ‘clarity of diction’ ideology. At present, it is difficult to know how these speakers (aged as they were between 30 and 60 at the time of Hannisdal’s data collection) fit into a larger diachronic picture of the fate of smoothing in modern RP, but the gender division does at least suggest that some sort of sociolinguistic dynamic is present.

The second of Hannisdal’s phonological variables I will look at here is yod coalescence, which in her analysis encompasses the phonological variables (tj) (dj),Footnote 5 in words such as tune and dune, with their possible variants [tj] [tʃ] and [dj] [dʒ]. Hannisdal’s corpus of 617 tokens showed coalescence occurring in 46.4% of these. Lexical considerations played a role, as the word during contributed almost half of the items showing coalescence, with coalescence in 83% of tokens. Hannisdal concludes that there is ‘no doubt that yod coalescence in stressed syllables is becoming established in RP speech’ (Hannisdal 2006: 213), and, indeed, she notes that yod coalescence was marked as non-RP in the second edition of Wells’ Longman Pronunciation Dictionary, except for the case of the word during ( Wells 2000).

/r/ Variation in RP

Fabricius (2017) reports on a quantitative study using BBC recordings from 14 English upper-class and upper-middle-class speakers born between 1880 and 1920. These recordings provided a corpus of tokens of syllable-initial /r/ (N = 2511). The results of an auditory analysis showed that two variants, in particular, tapped /r/ (including a small number of cases of trilled /r/) in medial and r-sandhi positions, and labialised /r/, mostly in word-initial position, had significantly different social and linguistic profiles. Tapped /r/s are now very rare in modern RP speech, while labialised /r/ now seems to be on the increase in many parts of British society (see Foulkes and Docherty 2000).

As Fig. 3.3 shows, cross-tabulations of the data showed decreasing tapped /r/ usage across the decades of the recordings. This decrease was found most starkly across medial (intervocalic) and linking /r/ contexts. The trend was independent of the speakers’ dates of birth and suggested a changing style of the time whereby taps and trills became increasingly rarer in these BBC recordings.

Fig. 3.3
figure 3

Trends in rates of tapped and trilled /r/ by word position according to decade of recording (Reproduced from Fabricius 2017)

Full size image

The data were then modelled according to word position and decade of recording, with year of birth as a continuous factor and speaker as a random factor. This model was highly significant, while factors such as gender and speech context (interview versus monologue) were not. Table 3.3 shows the results for 13 speakers and 2289 tokens (omitting OM4, as noted above) for tapped and trilled /r/, recoded and categorised together as taps. The three independent factors, in order from least to most significant effect, were position in the word (p = 6.26e-132), decade of recording (p = 157e-07) and year of birth, examined as a continuous variable (p = 0.000325).

Table 3.3 Mixed methods logistic regression modelling for tapped and trilled /r/, N = 2289 (excluding OM4; data from Fabricius 2017: 56)
Full size table

While in the 1950s data the tapped/ trilled /r/’s factor weight favoured taps and trills at 0.635, there is already a slight disfavouring of the feature in the 1960s and a further decrease in the 1970s. The significance of Year of birth as a continuous variable also showed that taps and trills steadily decreased with generation, following the decade of recording trend, while both are independently significant. Word-medial, intervocalic positions most highly favoured taps and trills (e.g. the common stereotypical pronunciation of very).

As Fig. 3.4 shows, labial variants (which included labiodentals and labialised alveolars) in the corpus were predominantly produced by three individuals. OF2 (Baroness Stokes), OM4 (Lord Halifax) and YF3 (Daphne du Maurier) were the only three speakers whose production of labials was above the average for all speakers in the corpus of around 10%. OM4, Lord Halifax (with an aristocratic background) is by far the most prolific user of labials for /r/ at 52.7%. While he is the only speaker of this type in the corpus assembled for this study, the BBC archive potentially holds other examples of comparable recordings which could provide a firmer basis for future conclusions.

Fig. 3.4
figure 4

Labial and non-labial /r/ in the corpus, by percentage (Reproduced from Fabricius 2017: 57)

Full size image

The logistic model showed that labials were most strongly favoured in initial position and medially. Decade of recording is strongly favoured only in the case of the 1930s, which isolates the single recording of OM4 referred to above. Other decades do not favour labial production, but as Table 3.4 shows, this is a result which is strongly affected by the dominance of a single speaker in this limited corpus. The result for speech type shows a strong factor weight favouring ‘Interview’ as speech context, which may seem anomalous, given that OM4’s recording is a monologue, but Fig. 3.4 also shows that a large number of labial tokens also occur in the interviews recorded with YF3 (Daphne du Maurier) and OF3 (Baroness Stokes). Although this small-scale study could not tell the definitive story of labial r in RP, there are indications that labial variants could be better characterised as idiosyncratic features in this corpus rather than a general sociolinguistic feature of the group, as tapped and trilled /r/s seem to be.

Table 3.4 Mixed methods logistic regression modelling for labiodental and labialised /r/, N = 2511 (including OM4; data from Fabricius 2017: 57)
Full size table

Variation and Change in RP Over the Lifespan

Studies of RP speakers have also contributed to an understanding of accent change over the lifespan post-adolescence. Harrington et al. (2000), Harrington (2006) and MacKenzie (2014) have employed close phonetic analyses of broadcast recordings to explore this, using Her Majesty (HM) Queen Elizabeth II’s Christmas speeches and Sir David Attenborough’s natural history programmes for the BBC.

Harrington et al. (2000) presented an analysis of vowel tokens gleaned from Queen Elizabeth II’s Christmas broadcasts in three periods (the 1950s, the late 1960s/early 1970s, the 1980s). They demonstrated an expanded vowel space on the F1 dimension in the 1960s and 1980s data compared to data from the 1950s. These results were compared to the vowel positions of a set of 1980s Standard Southern British speakers from the MARSEC corpus, reported in Deterding (1997). The authors concluded that the Queen’s vowels in the Christmas broadcasts had individually shifted in the direction of more mainstream forms of RP. However, Fabricius (2007) provides a comparison of the Queen’s short vowel system with a set of vowel systems from contemporary and younger RP speakers, which shows that the Queen’s short vowel system, even in the later recordings, if understood holistically as a geometrical configuration, is most closely aligned with speakers born in the same decade and consistently different from that of speakers born in subsequent decades.

In another study, MacKenzie (2014) conducted an analysis of Sir David Attenborough’s broadcast speech, examining, in particular, the realisation of /r/ as taps intervocalically within lexical items and in cases of linking /r/. Data from two nature documentaries, Zoo Quest (1959) and Planet Earth (2006) provided tokens of /r/ for auditory coding. The results were modelled using logistic regression, and it was found that while overall rates of tapped r were not significantly different, the speaker had gone against community change over time in one particular detail. Sir David was found to use higher rates of tapped /r/ in linking /r/ contexts, and in ‘voiceover’ mode (which to some extent is equivalent to ‘careful’ speech in sociolinguistic work) but not in his onscreen, more ‘casual’ mode in the more recent recordings.

Attitudinal Studies of RP: Dialect-in-Discourse

Alongside a large body of literature on attitudinal studies of the reception of RP carried out in the 1970s (Giles et al. 1990), there is also a small amount of more recent literature on attitudinal studies of RP. The ‘dialect-in-discourse’ model of attitudinal analysis (Garrett et al. 2003) inspired a small-scale study of attitudes to modern RP in York in 2002 (published as Fabricius 2005, 2006). The study examined school students’ attitudinal responses to recordings of the speech of upper-middle-class speakers born in the 1970s, with the recordings used being sourced from the set of sociolinguistic interviews conducted for Fabricius (2000). Sequences from sociolinguistic interviews were used to elicit responses. A total of 161 adolescents, aged 14 on average, from one independent and two comprehensive high schools in York took part in the study. Results comparing three of the six speakers made direct comparisons between responses to a male regional and a male non-regional speaker (Fabricius 2005), and to two non-regional speakers (one female, one male; Fabricius 2006).

The results showed that modern RP as a perceptual concept is to some extent hinged on gender: responses to the female modern RP speaker tended to be more positive than to the male modern RP speaker. Dynamism (encompassing independence and straightforwardness, Kristiansen 2001) distinguished the male regional speaker and the male non-regional, modern RP speaker, who was regarded as less dynamic. These two male speakers scored equally well on traits reflecting academic success (well educated, intelligent), differing somewhat from earlier evaluations of RP versus non-RP voices (Giles et al. 1990), where RP speakers were rated significantly higher than non-RP speakers. More in line with earlier results, the regional male speaker was graded more highly on interpersonal or social competence traits such as friendly, pleasant or trustworthy. The non-regional male speaker scored particularly badly relative to the female speaker in terms of what can be termed social dynamism (Kristiansen 2001), being enthusiastic and interesting. Fabricius (2006) ventured the conclusion that the male non-regional speaker’s speech performance was particularly resonant with echoes of an older style of RP speech and an elitist discursive stance, and conjured up overt class differentiation, being redolent of the male public school voice.

To sum up then, with a long-term diachronic view of the transformation of RP into modern RP , it seems clear that there is evidence of generational change in vocalic realisations, in both stressed checked and free vowels, as well as weak vowels, alongside the persistence to some extent of diphthong smoothing. In the consonantal system, we can see changes in /t/, /r/ and /dj, tj/ realisations. In the attitudinal responses to gendered modern RP voices, we see indications that these sociolinguistic styles make coherent sense to listeners, and that the ‘male public school voice’ continues to conjure up overt class difference and distinction. The picture then is one of vigorous variation and change in native-RP alongside other elements of stability , as seen in the persistence of perceptions of construct-RP.

Sociolinguistic Change and Modern RP

In this concluding discussion, I take a step back from language variation and change, and bring issues of sociolinguistic change to the fore. Coupland (2014: 69) suggests that ‘in its canonical form, variationism is not motivated to discover socially significant change, and it has no apparatus for gauging social impacts of change’, although recent ethnographically inspired work (e.g. Snell 2010; Moore 2010; Kirkham 2015) and studies of sociolinguistic perception (e.g., Campbell-Kibler 2010; Drager 2011; Levon and Holmes-Elliot 2013; Pharao et al. 2014) have ameliorated this to a large extent. No matter the variationist details of language use that can be identified in one or another set of corpus recordings, and important as these patterns can be for understanding sound change trajectories as purely historical linguistic processes, we cannot evade the issue that the changing social context in which speech takes place over time remains to be addressed theoretically within sociolinguistics, and this is precisely where Coupland’s important notion of sociolinguistic change is needed. Studies that embrace close ethnographic as well as close discursive analytical perspectives on this area are also needed.

Coupland presents an interesting and relevant (albeit, he demurs, tendentious) list of social changes in Britain between 1960 and 2009, ‘some more material, some more ideological’ (2009: 29–30). Those which seem at first glance most relevant to the topic at hand are: ‘the decline of the EstablishmentFootnote 6…Failing trust in professional authority…The growth of the middle class but the accentuation of the rich/poor divide’, although others, such as ‘massively increasing geographical mobility’ and ‘reframing and rescaling of local-global relationships’ (Coupland 2009: 29–30) may also have a role to play. Further into the chapter, Coupland hypothesises the potential consequence for the standard language hegemony of a ‘social change towards relative classlessness or towards more omnivorous cultural consumption’ (Coupland 2009: 35):

…even where patterns of linguistic variation persist across class-indexed groups (as of course they do, despite degrees of linguistic levelling), we would expect the sociolinguistic indexicality of class – the value associations of ‘standard’ and ‘non-standard’ speech – to be weaker and less significant.

To re-phrase Coupland’s quote, I would rather refer to ‘the value associations of higher class and lower class speech’ rather than ‘standard and non-standard speech’ in order to avoid the complicating factors of standardness and standardisation, as was discussed in the section on standardisation above.Footnote 7 Nonetheless, the distinction between n-RP and c-RP I have made there seems to enable us to specify the kind of difference Coupland is referring to: a weakening of the standard language hegemony would correspond to changes in c-RP whereby the import and authoritative position of an accent would be eroded, whereas (again, to quote Coupland 2009: 35) ‘patterns of linguistic variation persisting [or changing – my addition] across class-indexed groups’ (which I would define in terms of professional status, educational background, income and other sociological factors) would be matters of the description of changing patterns of sociolinguistically identifiable phonetic features, and thus changes in n-RP. The two types of change are closely related but also potentially independent sociolinguistic processes of considerable complexity.

Coupland further urges the consideration of a third type of change, sociolinguistic change, which is a complex of notions I will exemplify briefly with one case study below. In Coupland (2014: 69), sociolinguistic change is described as ‘a broad set of language-implicating changes’ within a society, and the study of sociolinguistic change involves ‘discovering changing relationships between language and society and their instantiation at the level of practice’ (Coupland 2014: 70). Coupland makes the case for an integrative sociolinguistic approach that can encompass a triad of types of change (social, linguistic, sociolinguistic), equally theoretically relevant and potentially empirically tractable in various ways.

One example of an integrative approach that illustrates how sociolinguistic changes over generations could be identified is in studies of metalinguistic discourses, as part of the study of ‘discursive practices’ in Coupland’s exemplificatory Figure 1 (2014: 74). One such study is found in Fabricius and Mortensen (2013), where the authors examine the construct resources (understood as dynamic, socially situated and identifiable ideological elements of c-RP) in a metalinguistic discourse surrounding the concept of accent in the UK. The chapter examines a brief stretch of talk extracted from an interview recorded in 2008 with a student at Cambridge University, as a response to the question Do you think that accents matter? In this brief stretch of talk, the interviewee invokes a number of construct resources, such as the location of posh accents in the south of England, as well as other resources, such as the idea that accent prejudice is off the record. The latter compares with Abercrombie’s assertion (1965, originally published in 1951) that the question of which side of the RP/non-RP accent-bar any speaker was placed on was never formulated explicitly at that time. The interviewee also performs hyper-stylisations of the accent (in the word posh itself, for example) to get her message across. The authors argue that these findings give access to an emic perspective on the role of language variation in the community, and as such are a valuable supplement to more etic experimental approaches to language attitudes surrounding RP (such as Fabricius 2005, 2006 and their antecedents, discussed above). It is clear that we will need a battery of methods and approaches for the study of elite sociolects as dynamic and changing sociolinguistic phenomena in the future.

Conclusion

Our focus in this chapter has been to set out the state of the art for studies of the sociolinguistics and sociophonetics of the elite sociolects of the UK, known as they are under various labels that have been part of the linguistics and phonetics literature for a century. As I have shown, these forms of speech and their sociolinguistic profile within the setting of the south of England and beyond are a fruitful area of investigation. In order to tackle these phenomena, we need an arsenal of approaches encompassing many of the tools from all three waves of sociolinguistics (to use the label from Eckert 2012). I have set out to show historical trajectories and variations in consonantal and vocalic production, as well as attitudinal and language-ideological aspects of the elite sociolect.

In taking stock of the research that so far has been produced relating to the sociolinguistic study of elite sociolects in the UK, I have shown that this is a fruitful area that touches on many aspects of sociolinguistic theory, ranging from the more strictly linguistic understanding of variation and change to the more qualitative study of the attitudinal, ideological and meta-discursive space of an elite sociolect in the sociolinguistic landscape of Britain over time. Examination of what constitutes sociolinguistic change in this case can also be particularly rewarding, positioned, as these accent features and accent tropes are, at the nexus of a particular set of social, linguistic and historical circumstances, however, we may choose to name the variety or varieties, and whatever the future of its accent features may turn out to be.