<<

What’s happening in RP? An empirical look at variation and change in Received Pronunciation

Bente Rebecca Hannisdal University of Bergen

Introduction Received Pronunciation, or RP, is probably the most thoroughly described accent of English. There has, however, been very little quantitative empirical research into the accent. The RP tradition is dominated by informal, impressionistic observations, or introspection. The RP accent has been somewhat neglected in the field of variation and change, where the focus tends to be on non-standard varieties. Also, the codification of RP carries with it the danger of perceiving the accent as static and uniform, and the phonetic properties of RP are often taken for granted. This paper presents some results from a recent study of variation and change in RP, which aims to fill some of the empirical gap by supplying quantitative corpus- based data on current RP usage.

Empirical data 30 hours of speech from 30 television newsreaders (15 males and 15 females) from BBC World, Sky News and ITV News.

Phonological variables 1. CURE lowering, which involves the increasing use of / N9 / instead of traditional / T? / in words such as sure , pure , tourist , etc.

2. Smoothing, which refers to the reduction of the sequences / `H?+ `T? /, as in fire and power , to [ `? ] or [ `9 ].

3. GOAT allophony, which refers to the realisation of / ?T / as [ PT ] before non-prevocalic /l/ in goal , hold , shoulder , etc.

4. R-sandhi, which comprises the use of linking and intrusive /r/ to avoid vowel hiatus, in phrases such as here /r/ and there , law /r/ and order , etc.

5. T-voicing, which involves the realisation of intervocalic /t/ as a voiced tap [3] rather than a fortis , in e.g. getting , British , a lot of , etc.

6. Yod coalescence, which involves the coalescent assimilation of / si+ ci / to / sR+ cY / in stressed environments, as in tune , studio , endure , etc.

1 Results

1. CURE lowering CURE lowering is a change that has been going on for quite a long time. The / T? /- has often been predicted to disappear altogether from the system of RP (see e.g. Gimson 1964, Brown 1990, Nolan 1999). The results reported here, however, revealed that all 30 speakers have / T? / as an active member of their phonological system. They actually prefer the diphthong in the majority of CURE words, and lowering is established only in a few items (see Table 1).

Table 1. The lexical distribution of /T? / and /N9 / in CURE

word /T? / /N9 / word /T? / /N9 / bureau 6 - sure 7 38 cure 4 - assure 6 11 during 171 - assurance 2 1 Europe 360 - ensure 6 11 furious 10 - insurance 10 4 jury 12 2 tour 20 31 lure 2 - touring 1 2 moor 1 3 tourism 12 3 poor - 28 tourist 49 17 purely 4 - tournament 8 28 secure 18 1 Uruguay 1 - security 239 - your/you’re - 129

Based on these findings, there is apparently still a long way to go before / T? / is replaced by /N9 /. A closer look at the distribution further shows that the two vowels seem to be preferred in different phonetic environments: / T? / is used after a consonant + /j/, as in during , secure , furious , etc., and before prevocalic /r/, as in Europe , tourist , while / N9 / is preferred in other contexts, in words like poor , sure , tour , etc. The variability between / T? / and / N9 / in CURE seems, then, to be largely determined by the phonetic context, though not so systematically that it can be reduced to an allophonic split.

2. Smoothing The reduction of / `H?+ `T? / to monosyllabic [ `?+ `9 ] is also a traditional RP feature, mentioned already in the first edition of Jones’ English Pronouncing Dictionary (1917). My analysis showed that reduction of / `H?+ `T? / remains a common process in RP, as almost half the tokens were realised with Smoothing (see Table 2). The phonetic outcome of Smoothing is in most cases a centring diphthong [ `? ]. It is occasionally reduced to a monophthong, but usually with a front quality, avoiding a merging with the phoneme / @9 /.

2 Table 2. Smoothing: total scores

N % [`?+ `9 ] 621 46.4 [`H?+ `T? ] 718 53.6

Smoothing seems to represent a case of stable variation rather than change in progress. It can be viewed as an alternative pronunciation of FIRE and POWER words, which some RP speakers have as part of their phonetic inventory. It can also be viewed as a feature of phonological reduction and simplification. In this perspective, Smoothing would be expected to vary according to the level of speech attention and the nature of the linguistic surroundings.

3. GOAT allophony The back rounded [ PT ] variant in pre-/l/ GOAT items is common in English, where the realisational variability has resulted in a phoneme split. It has also been described as typical of (e.g. Wells 1994). GOAT allophony is generally not recognised as a feature of RP, and is not included in the traditional descriptions. Based on my findings, however, GOAT allophony is becoming firmly established in modern RP. As Table 2 shows, 24 of the 30 speakers use this categorically, while only two speakers lack the allophone. The extensive use of GOAT allophony is a clear indication that this feature should be considered a part of modern RP, and that it extends beyond Estuary English or London accents.

Table 3. The use of GOAT allophony

Categorical Variable No allophony allophony allophony Number of speakers 24 4 2

4. R-sandhi R-sandhi is one of the most common connected speech phenomena in RP and other non-rhotic accents. Linking and intrusive /r/ represent the same phonetic process, but they have different evaluative status. The use of linking /r/ is generally considered correct and desirable, and is, according to Cruttenden (2001: 289) “generally present” in RP speech, whereas stigmatisation has arisen for intrusive /r/. As expected, then, linking /r/ was found to be considerably more frequent than intrusive /r/ (see Table 4).

Table 4. R-sandhi: total scores

Linking /r/ Intrusive /r/ N % N % /r/ 3612 59.8 182 32.6 Ø 2433 40.2 376 67.4

3 Although /r/ is the dominant variant, a total percentage score of 59.8 is lower than expected, and suggests that linking /r/ may not be so categorical in RP as is often claimed, and that its status as “correct” RP usage is of less relevance. At the same time, it is possible that the low usage levels for sandhi /r/ may be a direct result of the nature of the speech situation. The newsreaders’ focus on clarity may be one reason why so many sandhi r’s are omitted: there may be a general conception that separating the words by leaving out the r-link leads to greater articulatory distinctness. The analysis further revealed a strong conditioning by certain linguistic environments on the use of sandhi /r/, motivated mainly by prosodic, semantic and articulatory considerations. Linking and intrusive /r/ are significantly disfavoured - before a stressed for !hours, were !injured, Malaga !airport - before proper nouns Mr Annan, Doctor Austen, Katya Adler - in the vicinity of another /r/ terror attacks, extra information, area around

5. T-voicing T-voicing is one of the many phenomena that are known to affect /t/ in English accents. Within Britain, t-glottalling is the feature that has been most extensively discussed and investigated, also with reference to RP (cf. Fabricius 2000). Voicing of /t/ has not been subject to the same scholarly attention. To the extent that it is mentioned, it is said to be characteristic of casual, colloquial style or vernacular varieties. RP is generally not thought of as having t-voicing, and RP /t/ is said to be a voiceless plosive in all environments (cf. e.g. Wells 1982, Chambers 2002). This study investigated t-voicing in intervocalic environments following a short vowel. The findings showed an overall usage level of 35.2% for [ 3], which is strikingly high, especially considering the relative formality of the newscast setting. It clearly demonstrates that the tap is a common realisational variant for /t/ in RP, and that it is not limited to vernacular speech or casual style.

Table 5.1. T-voicing: total scores

N % [3] 1976 35.2 [s] 3639 64.8

A morphological analysis further showed that voicing of intervocalic /t/ is primarily a feature of connected speech: the percentage score for the voiced variant across word boundaries (what about, a lot of ) was a staggering 66.9%, which establishes t-voicing as one of the most common connected speech processes in modern RP. T-voicing was largely resisted word-internally, except in certain frequent items, such as British , getting .

Table 5.2. T-voicing according to word position

Word-final Word-medial N % N % [3] 1344 66.9 632 17.5 [t] 664 33.1 2975 82.5

4 6. Yod coalescence Yod coalescence is also traditionally perceived as non-standard or non-RP (cf. Wells 1994, Taylor 1998, Wells 2000). But in line with the general spread of the feature, pronunciations such as / sRt9a / tube and / cYt9m / dune have become increasingly accepted as part of modern RP (see e.g. Cruttenden 2001, Upton 2004). The results from the present study show that yod coalescence has penetrated the boundaries of RP. Almost half the tokens (46.4%) are realised with (see Table 6). This is a very high figure, in view of the status of stressed yod coalescence in RP and the formality of the speech situation, and is a clear indication that yod coalescence in stressed is becoming an established feature in RP and should be included in updated descriptions of the accent.

Table 6. Yod coalescence: total scores

N % /sR+ cY / 286 46.4 /si+ ci / 331 53.6

Defining RP T-voicing, yod coalescence and GOAT allophony are non-traditional features that represent new developments in RP. The question then becomes: how can we justify their inclusion in RP? Whether or not a feature should be accepted as part of RP depends on how the accent is defined. In spite of the large number of descriptions of RP, there exists no universal definition of the accent. Modern RP is particularly problematic to define, as the use of RP is not confined to one specific region or one identifiable social group. There is general consensus that the phonological core of RP is identical with the segmental system found in the traditional descriptions. The problems arise when observers encounter variation and potential change. As an object of study, RP has to be delimited. The choice of definition will affect the decisions regarding which features fall within or outside RP. It is therefore important that the defining criteria can be operationalised. Linguists writing about RP have operated with several different criteria for defining the accent.

An abstract norm One possible definition of RP views it as simply the codified version of English pronunciation, and reduces the accent to a theoretical construction – a standardised norm which is described in pronunciation dictionaries and textbooks, but which nobody really speaks (see e.g. Smith 1996). This view of RP as an abstract and unattainable entity, means that there are no RP speakers, and it becomes impossible to talk about changes in the accent. The majority of phoneticians, however, agree that we can identify a living speech variety that corresponds to the textbook descriptions of RP, and that some people have this variety as their native accent, others as a norm towards which they modify their speech. But RP speakers cannot simply be identified as those who speak the codified RP, or the whole concept becomes circular. We need some other criterion to select potential informants.

5 The BBC accent Because of the close association of RP with broadcast speech, and with newsreaders in particular, BBC English has often been used as a synonym for the accent, and several linguists define RP by reference to the form generally used by newsreaders of the BBC (e.g. Roach et al. 2003). RP is still the accent typically used by BBC newsreaders, but it is problematic to define RP as the pronunciation of BBC newscasters. Several of the BBC newsreaders now have accents that clearly fall outside RP, and RP can therefore not be equated with newscast English.

Prestige RP has traditionally been associated with prestige and status. One possibility is therefore to define RP as the most prestigious accent. Prestige is however difficult to define and measure, and it is no longer the case that RP will necessarily or consistently be rated as the highest prestige variety. The accent has typically been found to score highly for traits such as ambition, intelligence, education and wealth, but to be downgraded in terms of social attractiveness and solidarity. Moreover, the rise of various ‘regional standards’ complicates the picture even further (see e.g. Garrett et al. 2003).

Social criteria One approach to defining RP could be to relate the accent to a social group. The earliest accounts of RP described the accent of the educated members of the upper classes. Up until the middle of the 20 th century, RP remained the accent of one particular social group. Since then, however, there have been radical changes in the structure of British society and in the attitudes towards accents. Several of the criteria previously used to delimit RP are irrelevant today, as there is no longer a straightforward relationship between social class and education or profession. Educated people today speak a wide variety of accents, and the upper classes no longer have a uniform style of pronunciation. It is therefore problematic to use the speakers’ social background as the sole criterion for defining the RP accent.

Non-localisability RP has its origins in the southeast of England, as the pronunciation of educated speakers in and around the capital, and shares its main phonological basis with southeastern English. Today, however, it is non-regional, or supra-regional, in that it extends over the whole of Britain (or at least England), although only a small minority of people speak it. This spread of RP came about as a result of the nationwide diffusion of the educational system in the 19 th century. If we delimit RP as a non-localisable accent, the most important criterion for the inclusion of a new feature in RP must be whether the feature is localisable or not. One way of verifying that a feature is non-regional is if it is observed with speakers from different parts of the country. This criterion was used in the present study when investigating potential change. GOAT allophony, t-voicing and yod coalescence were included as potentially new RP features, as they can all be said to be non-localisable. GOAT allophony and T-voicing have been reported in a number of accents from all parts of the British Isles (see Foulkes and Docherty 1999), which shows that they are supra-regional features. And an extensive poll of British pronunciation preferences, referenced in Wells (1999), reveals that yod coalescence is common with speakers from many parts of the country. Non-localisability provides a criterion which is flexible in that it allows for changes in RP, and explains many of the ongoing tendencies in the accent. It secures a continuation of the RP tradition, and it can be tested objectively.

6 Conclusion The findings reported here may be seen to have implications for the phonological description of current RP. Several of the tendencies observed represent a deviation from the standard descriptions of RP, and suggest that the accent has undergone changes. The most important finding in this perspective is the prevalence of GOAT allophony, t-voicing and yod coalescence, which conventionally have been excluded from RP descriptions. The study further illustrates the importance of keeping the definition of RP updated and in line with social changes, as well as the importance of specifying which definition is the basis for descriptions of RP or for discussions of its status.

References Brown, G. 1990. Listening to spoken English . 2 nd ed. London: Longman. Chambers, J. K. 2002. Sociolinguistic theory . 2 nd ed. Oxford: Blackwell. Cruttenden, A. 2001. Gimson’s Pronunciation of English . 6 th ed. London: Arnold. Fabricius, A. 2000. T-glottalling between stigma and prestige: a sociolinguistic study of modern RP. Unpublished PhD thesis, Copenhagen Business School. Foulkes, P. and Docherty, G. (eds.) 1999. Urban voices. Accent studies in the British Isles . London: Arnold. Garrett, P., Coupland, N. and Williams, A. 2003. Investigating language attitudes: social meanings of dialect, ethnicity and performance . Cardiff: University of Wales Press. Gimson, A.C. 1964. Phonetic change and the RP vowel system. In D. Abercrombie et al. (eds.), In honour of Daniel Jones : papers contributed on the occasion of his eightieth birthday, 12 September 1961 . London: Longman. 131-136. Jones, D. 1917. An English pronouncing dictionary . London: J. M. Dent & Sons. Nolan, F. 1999. The shifting sands of English pronunciation. In P. Rajamäe (ed.), New Britain: The heritage of the past and the challenge of the future. Proceedings of the 2nd International Tartu Conference on British Studies, University of Tartu, August 24-25 1998. 76-93. Roach, P., Hartman, J. and Setter, J. (eds.) 2003. English pronouncing dictionary . 16 th ed. Cambridge: Cambridge University Press. Smith, J. 1996. An historical study of English: function, form and change . London: Routledge. Taylor, J. 1998. What, pray, is happening to dear old RP? La Linguistique 34: 2, 141-151. Upton, C. 2004. Received Pronunciation. In E. Schneider et al. (eds.), A handbook of varieties of English. Volume 1: . Berlin: Mouton de Gruyter. 217-230. Wells, J.C. 1982. Accents of English (3 vols.). Cambridge: Cambridge University Press. Wells, J.C. 1994. The Cockneyfication of R.P.? In G. Melchers and N.-L. Johannesson (eds.), Non-standard varieties of language . Stockholm Studies in English 84. Stockholm: Almqvist & Wiksell. 198-205. Wells, J.C. 1999. pronunciation preferences: a changing scene. Journal of the International Phonetic Association 29: 1, 33-50. Wells, J.C. 2000. Longman Pronunciation Dictionary . New edition. London: Longman.

7