Phonological Variation and Change: A Case Study of Intervocalic /t/-glottaling in Cambridge RP

1

Phonological Variation and Change: A Case Study of

Intervocalic /t/-glottaling in Cambridge RP

Chen Shen1March 20141. IntroductionResearch in linguistic variation and change has been carried out bysociolinguists extensively during the past few decades. Under the influence ofLabovian sociolinguistic approach to language change, many studies were conductedon a phonological level (Kerswill, 2011). Although the majority of the analyses ofphonological variation were based on English data, more general principles andproblems could be revealed (Foulkes, 2006). The aim of this study is, accordingly, toconduct a sociolinguistic investigation on one phonological variable in a linguisticvariety in English, and to try to illustrate how social characteristics influencespeakers’ phonological realisation. The main focus here is intervocalic /t/-glottallingin two different speaking styles (formal and casual) in Cambridge RP. Other socialparameters will also be briefly discussed.2. Received Pronunciation (RP) and /t/-glottallingWells (1982) defines RP as a relatively regionless accent originating from south-east England. Since it is widely used in English radio and TV, and is the accent taughtto students worldwide who are studying British English, it can be heard not onlythroughout the UK but also around the world. However, its actual speakers onlyaccount for 3-5% of the population of England, and it is a social rather than regionalaccent which associated with the upper-middle and upper class (Trudgill, 2001;Roach, 2004). Although RP is widely considered as a more static linguistic variety, alllanguage varieties are subject to variation and change and RP is no exception. Recentstudies like Harrington, Palethorpe and Watson (2000) on certain pronunciation ofvowels in the Queen’s Christmas messages during 1950s and 1980s, Trudgill (2001)on innovations of RP, Fabricius (2000, 2002) on ongoing changes in modern RP, and1 Taught postgraduate student in MSc Forensic Speech Science, University of York. Email: [email protected]

2

Hannisdal (2006) on six phonological variables of broadcasting RP havedemonstrated the changing trendAmong many phonological variables that have been studied, plosives, especiallythe realisation of /t/, have been extensively discussed and investigated. Ladefogedand Maddieson (1996), Johnson (1997), and Foulkes et al. (2010) all mention thatstops are highly complex sounds with complicated articulatory and acousticproperties. Foulkes and Docherty’s (1999) study illustrates that among all stopvariants, glottallised stops are also highly variable. Watt and Milroy (1999) reportthat there are two distinct types of glottal variants in Newcastle English. Dochertyand Foulkes’ (2005) study on glottallised variants of /t/ in Tyneside Englishgenerates a thorough image of various realisations of glottalised /t/. Moreover,Collins and Mees (1996) illustrate the chronological spreading of glottalisation in RPEnglish. Altendorf’s (1999) study of Estuary English and RP highlights that /t/-glottalling, as a phonetic variable, varies in accordance with different social factors.Fabricius’s (2000) and (2002) studies explain the destigmatisation of /t/-glottallingin Modern RP by conducting sociolinguistic examinations on this variable undervarious contexts. This study aims to explore the trend of /t/-glottalisation inCambridge RP produced by a group of speakers under a certain linguisticenvironment, and to correlate it with some extralinguistic social variables.3. MethodologyThe studies mentioned above point out that ‘intervocalic’ is the least favouredand the most stigmatised context for /t/-glottalling, but Fabricius (2002, p.119)reveals that /t/-glottalling “remains an as-yet-incomplete change in progress”.Therefore, this study will take this stigmatised phonological context as the mainpoint of analysis, aiming to find out its correlations with other social factors andpossible ongoing changes.Foulkes (2006) and Foulkes and Docherty (2007) illustrate several socialparameters that are frequently used in phonological variation studies, namelygeography, social class and social network, age, sex and gender, race and ethnicity,and communicative context. These dimensions share equal sociolinguisticimportance, but due to the nature of the corpus and the space available in this study,a comprehensive descriptive survey of all parameters is not achievable. Speaking

3

style has been an enduring focus in sociolinguistics (Foulkes, 2006), and manystudies have proven the existing influence of contextual styles on speech. Forinstance, early variationist work done by Labov (1966, 1972) demonstrates thespeech variability caused by style. Giles (1973) points out contextual constrainsinfluence pronunciation patterns of speakers. More recent work such as that byAltendorf (1999) and Fabricius (2000, 2002) also generate similar conclusions. Thisstudy will hence select speaking style as a sociolinguistic parameter to explore itscorrelation with the speakers’ phonological variation in one linguistic variety.The data of 12 Cambridge RP speakers’ speech samples were chosen from theIViE (Intonational Variation in English) corpus.2 Since the corpus was designed toinvestigate cross-varietal and stylistic variation in English intonation, as well as toconstruct a statistical computational model of intonation that takes account ofvariation due to dialect, speaking style, gender and individual speaker habits (IViEhomepage 2008), the sociolinguistic properties of the corpus were restricted. Allspeakers were of similar age (young) and social background (middle class), andwere all born and grown locally (Grabe, 2004). The recordings contain five tasks:three individual tasks including 22 context-free sentences reading (as there was noelement of /t/ in those sentences, none was used in this study), a passage readingtask of the fairy tale Cinderella, and a retelling task of the same passage. Other twointeractive tasks consist of a map task and a conversation about ‘smoking’ done by 6single-sex pairs (Grabe, 2004).As most of the studies make a distinction between formal and casual stylesbased on either reading or conversation tasks, how to group the retelling task in thisdata becomes problematic. It could be argued that speakers are more egocentric inretelling than in reading task, they are, nonetheless, not ‘free’ enough to expresstheir own thoughts due to the relatively restricted content. Additionally, Grabe(2004) separates the five tasks into individual and interactive sub-groups (seeabove), retelling tasks are, consequently, different from the other two interactivetasks due to the lack of any interlocutor. It is also clearly shown on the graph (Figure1.) that retelling style is more similar to passage reading in terms of intervocalic /t/realisation. Therefore, it was firstly grouped into formal style and then into casual2 Economic and Social Research Council award to F. Nolan (University of Cambridge) and E. Grabe (University ofCambridge and Oxford). Research Associates: K. Farrar and B. Post. Funding period: 1997 - 2002.

4

style when conducting binary style comparison tasks later on (results see section 4below).

Figure 1. Percentage of intervocalic /t/ realisation in four tasks.Foulkes et al. (2010) state that auditory analysis of consonants, including stops,is often used in sociolinguistic studies of phonetic variants in certain phonologicalentities. Sociolinguists generally generate patterns of statistical correlationsbetween variant forms and social categories based on quantitative data. However, asa single research approach, it has both pros and cons. Milroy and Gordon (2003)point out that subjectivity is the main concern of auditory technique, as it reliesheavily on analysts’ perceptual categories and their own consistent interpretation.But on the other hand, analysts’ perceptual system could help filtering complex datasuch as natural speech, and normalising tokens that vary in subtle ways intomeaningful categories based on the main interest of the study (Foulkes et al., 2010).Acoustic analysis, as another most commonly used technique, is able to providemore reliable and detailed information of variants as long as the settings andinterpretation are consistent (Ladefoged, 2003). Acoustic features could be used asan assistant of checking the auditorily disputed words. Therefore, in this study,auditory analysis is complemented by referring to waveforms and spectrograms.

5

20.8197833 20.9348473P5CMC

Time (s)20.6 21.160

5000

Freque

ncy (Hz

)

0

1000

2000

3000

4000

5000

20.6 20.71 20.82 20.93 21.04 21.16

Figure 2. Waveform and spectrogram of a token daughter produced by a speaker(MC) showing an aspirated alveolar stop.39.4701699 39.5412111

R3CMC

Time (s)39.31 39.770

5000

Freque

ncy (H

z)

0

1000

2000

3000

4000

5000

39.31 39.4 39.49 39.59 39.68 39.77

Figure 3. Waveform and spectrogram of a token beautiful produced by the samespeaker (MC) showing a glottallised stop.Table 1. Features of two tokens from two categoriesToken beautiful daughterAuditory realisation glottal [ʔ] stop [th]Auditory property creaky phonation stop phonationTransient burst absent presentAcoustic feature strident glottal pulses high frequency energy

6

Tokens were first assessed auditorily into a binary categorisation under onephonological context: ‘t’ versus ‘glottal’ intervocalically. Considering the complexityand variability of not only the realisation of /t/ but also the glottal forms, onlyauditorily and acoustically perceived plain glottal stops [ʔ] without anyaccompanying [t]-onset were categorised as ‘glottal’ (although no complete glottalstop closure is present on spectrogram, see Figure 3 and Table 1). Furtherexplanations see Docherty and Foulkes (1999, 2005). Other variants such as glottallyreinforced double articulations [ʔt] or [tʔ], ejective [t’], aspirated [th], affricated [ts],tapped [ɾ] or voiced [t], and plain alveolar stop [t] were transcribed but allcategorised as ‘t’. The intervocalic context includes both word internal [V_V] (eg.daughter) and word boundary [V_#V] (eg. sort of) environments.4. Results and AnalysisA total of 519 tokens of variants of /t/ were selected from 12 speakers in thedata. Glottalised /t/, at least in intervocalic position, was found to be far morewidespread in casual speech than in more formal ones (Figures 4 and 5). Resultsfrom the chi-square tests presented in Table 2 strongly suggested the styledifference to be significant (p<.0001). An even stronger statistical effect (χ2=78.56)was found when retelling being grouped into formal style. Considering the nature ofthe retelling task mentioned above and for the purpose of consistency, it was thengrouped into formal style for the rest of the analyses.

Table 2. Chi-square tests showing significant style differencesχ2 df pstyle (retelling as formal ) 78.56 1 < .0001style (retelling as casual) 36.28 1 < .0001

7

Figure 4. Percentage of intervocalic /t/ realisation divided by style (with retellingtask grouped into formal style)

Figure 5. Percentage of intervocalic /t/ realisation divided by style (with retellingtask grouped into casual style)Labov (1972a, b) (as cited in Di Paolo and Yaeger-Dror, 2010, p.11) highlightedthat speakers are more self-conscious about their speech when producing formalstyle speech such as word-list or passage reading. Other more recent studies likeYaeger-Dror (2001) and Llamas et al. (2009) have confirmed Labov’s findings by

8

showing interviewees’ accommodation of prestige standard or interviewer whenreading word-lists as well as passages./t/-glottaling has been strongly suggested to be an innovated vernacular thathas spread to regional varieties in many parts of Britain including RP (see studiesmentioned in section 2). The results of current study displayed the similar trend.Although being the least favoured environment, glottalised intervocalic /t/ ispresent among middle class young Cambridge RP speakers, with a larger proportionin casual than in formal speech. Possible interpretations of this could be, firstly,speakers used lesser non-standard forms in more formal tasks as they were moreself-conscious about their language usage, and thus accommodated to more standardvariety. Secondly, since both the interviewer (map director) and the interviewee(interlocutor) in this data were close in age, social class, background, all six pairswere of same gender and all were participants of the two interactive tasks,henceforth, more vernacular forms were produced in more casual speech.It is worth mentioning that phonetic environment also has a significant role toplay. As shown in Figure 6, word boundary appeared to be the most favouredenvironment for intervocalic /t/-glottalling. Interestingly, although speaking stylehas an overall significant influence on /t/ realisation, more glottal tokens wereproduced in formal style word boundary (37.5%) than in casual style word internal(30.6%) environment. It is also noted that most of the word boundary glottalisedtokens were frequently used short grammatical phrases, such as sort of, out of, andlot of. Contrarily, [t] tokens were favoured by stressed content words like daughter,

dirty and advertising. More strikingly, in word internal position, one of the mostfrequently appearing glottalised tokens was whatever, a function rather than contentword. Just as Ogden (1999) points out, function words and auxiliaries undergodifferent reduction processes from content words. We can, thus infer thatgrammatical category also influences speakers’ degree of phonological realisation.

9

Figure 6. Percentage of intervocalic /t/ realisation divided by style in two phoneticenvironmentsFoulkes (2006) claims that sex-correlated differences can be found in almost allsociolinguistic studies, a clear-cut gender difference was also found here (see Figure7). A chi-square test showed the gender difference to be significant: χ2= 39.24, df = 1,p < .0001. Classical studies such as Labov (1966) and Trudgill (1974) prove thatspeakers, especially females, tend to use more standard speech in more formal styles.Cheshire (2008) also points out women are more closely associated with standard orprestigious language varieties. The results presented here support their findings bydemonstrating that women produced lesser non-standard glottalised /t/ in moreformal speech.

10

Figure 7. Percentage of intervocalic /t/ realisation divided by gender in twospeaking stylesOne difficulty encountered while carrying out the study was token selection insuch naturalistic speech. Giegerich (1992, p.135) states that “connected speech issubject to adjustments in syllabification”. Foulkes (2006) explains prosody andstress pattern may also influence the articulation of utterances. Consequently,several word boundary intervocalic tokens in this data were excluded due to a pausein between. Besides, in free speech, although speakers are at a higher usage ofvernacular, the speech length and content of each individual are hard to control,hence speech of such kind may not contain as many phonologically interestingvariables (see speakers HB and MF in Figure 8). The asymmetrical distributionpresented in Figure 8 also reflects another interesting aspect of potential study: thephonological variability of individual speakers, and its correlation with certain socialvariables.

11

Figure 8. Percentage of intervocalic /t/ realisation by 12 speakers in two speakingstylesAnother interesting factor being found while performing auditory analysis wasthat alveolar tap was detected to be a relatively common realisation of intervocalic/t/ in female casual speech. Although t-voicing is mostly regarded as a feature ofNorth American English, some studies such as Watt and Milroy’s (1999) aboutNewcastle and Mees and Collins’s (1999) about Cardiff have proved the existence oft-voicing in accents of the British Isles, and Hannisdal (2006) describes t-voicing, asan existing phenomenon, is becoming more common in broadcasting RP. Thus thisvariant in varieties of British English might also be a potential interesting aspect towork on.

5. Conclusion and OutlookThe findings reported above show that stylistic variation parallels phonologicalvariation in Cambridge RP. Quantitative analysis suggests this correlation to bestrong. It is also seen that intervocalic /t/-glottalling as a stigmatised phonologicalvariation emerges systematically concerning style and gender. One salient findingwas this variation to be mostly realised under intervocalic word boundary context incasual speech.This study also illustrates the importance of specifying contextual style whendescribing phonological variations in certain linguistic variants. However, Schilling-

12

Estes (2008) points out the limitations of separating casual from formal speech insociolinguistic interviews. She argues that due to the complexity of different speechinterrelates with different settings, speakers are not simply accommodating to givennorms, they are, on the other hand, taking part in shaping and re-shaping thesenorms. The dichotomy between formal and casual style, like most dichotomies, isdisputed. Some styles, retelling for example, fall somewhere in between the formaland the casual extremes. Therefore, a fuller account of stylistic variation is neededfor further study.ReferencesAltendorf, U. (1999). Approaching the notion of ‘Estuary English’: /t/-glottalling and/l/ vocalisation by the Thames Estuary. In C. Paradis (ed.) Recent Trends in thePronunciation of English. Stockholm: Almqvist Wiksell. pp. 15-31.Cheshire, J. (2008) Sex and Gender in Variationist Research. In Chambers J. K.,Trudgill P. and Schilling-Estes N. (eds.) The Handbook of Language Variation and

Change Oxford: Blackwell.Collins, B. & Mees, I. (1996) Spreading everywhere? How recent a phenomenon isglottalisation in received pronunciation? English World-Wide 17: 175-187.Di Paolo, M., Yaeger-Dror, M. & Beckford Wassink, A. (2010) Analysing vowels. In DiPaolo, M. & Yaeger-Dror, M. (eds.) Sociophonetics: a Student’s Guide. London:Routledge. pp. 87-106.Di Paolo, M. & Yaeger-Dror, M. (eds.) Sociophonetics: a Student’s Guide. London:Routledge.Docherty, G & Foulkes, P. (1999) 'Newcastle upon Tyne and Derby: instrumentalphonetics and variationist studies'. in P Foulkes & G Docherty (eds), Urban Voices:

Accent Studies in the British Isles. Edward Arnold, London, pp. 47-71.Docherty, G & Foulkes, P. (2005) 'Glottal variants of (t) in the Tyneside variety ofEnglish: an acoustic profiling study'. in W Hardcastle & J Beck (eds), A Figure of

Speech: A Festschrift for John Laver. Lawrence Erlbaum, London, pp. 173-199.Fabricius, A. H. (2000). T-glottalling between stigma and prestige: a sociolinguistic

study of modern RP. Unpublished: Copenhagen Business School. PhD thesis.

13

Fabricius, A. H. (2002). Ongoing change in modern RP: evidence for the disappearingstigma of t-glottalling. English World-wide 23: 115-136.Foulkes, P. and Docherty, G.J. (1999). Instrumental phonetics and phonologicalvariation: case studies from Newcastle upon Tyne and Derby. In Foulkes, P. andDocherty, G.J. (eds), Urban Voices: Accent Studies in the British Isles. London: Arnold.pp. 47-71.Foulkes, P. and Docherty, G.J. (eds) (1999). Urban Voices: Accent Studies in the British

Isles. London: Arnold, 47-71.Foulkes, P., Docherty, G.J. & Watt, D. (2005) 'Phonological variation in child-directedspeech' Language, 81(1), pp. 177-206.Foulkes, P. (2006) Phonological variation: a global perspective. In B. Aarts & A.McMahon (eds). The handbook of English Linguistics. Oxford: Blackwell. pp. 625-669.Foulkes, P., Docherty, G. (2007) Phonological variation in England. In D. Britain (ed.)Language in the British Isles. Cambridge: Cambridge University Press. pp. 52-74.Foulkes, P, Docherty, G.J. & Jones, M. (2010) Analysing stops. in M Di Paolo & MYaeger-Dror (eds), Sociophonetics: A Student’s Guide. Routledge, London.Grabe, E. (2004) Intonational variation in urban dialects of English spoken in theBritish Isles. In Gilles, P. & Peters, J. (eds.) Regional Variation in Intonation.Linguistische Arbeiten. Tubingen: Niemeyer. pp. 9-31.Giegerich, H.J. (1992) English Phonology: An Introduction. Cambridge: CambridgeUniversity Press.Giles, H. (1973) Accent Mobility: A Model and Some Data. Anthropological Linguistics,15(2), pp.87-105.Hannisdal, B. R. (2006). Variability and change in Received Pronunciation: a study

of six phonological variables in the speech of television newsreaders. Unpublished:University of Bergen. PhD thesis.Harrington, J., Palethorpe, S. & Watson, C.I. (2000) Does the Queen speak the Queen’sEnglish? Nature, 408(6815), pp.927–928.Kerswill, P. (2011) Sociolinguistic approaches to language change: Phonology. In RWodak, B. Johnstone & P. Kerswill (eds.) The Sage Handbook of Sociolinguistics.London: Sage Publications, pp. 219-235.Labov, W. (1966). The social stratification of English in New York City. Washington,DC: Center for Applied Linguistics.

14

Labov, W. (1972). The social motivation of a sound change. In Labov, W.Sociolinguistic Patterns. pp. 1-42.Ladefoged, P. & Maddieson, I. (1996). The sounds of the world’s languages. Oxford:Blackwell.Ladefoged, P. (2003) Phonetic Data Analysis: an Introduction to Fieldwork and

Instrumental Techniques. Oxford: Blackwell. Chapter 6: Acoustic analysis ofconsonants, pp. 138-168.Llamas, C., Watt, D. & Johnson, D.E. (2009). Linguistic accommodation and thesalience of national identity markers in a border town. Journal of Language and

Social Psychology 28(4): 381-407.Milroy, J. Milroy, L. Hartley, S., & Walshaw, D. (1994). Glottal stops and Tynesideglottalization: Competing patterns of variation and change in British English.Language Variation and Change, 6(03), pp.327–357.Milroy L., and Gordon M., (Eds.). (2003). Sociolinguistics: Method and Interpretation.Oxford: Blackwell.Ogden, R. (1999). A declarative account of strong and weak auxiliaries in English.Phonology, 16, pp. 55-92.Roach, P. (2004). British English: Received Pronunciation. Journal of the

International Phonetic Association, 34(02), pp.239-245.Schilling-Estes, N. (2008) Investigating Stylistic Variation. In J. K. Chambers, P.Trudgill and N. Schilling-Estes (eds.), The Handbook of Language Variation and

Change, Oxford: Blackwell, pp. 375-401.Trudgill, P. (1974). The Social Differentiation of English in Norwich. Cambridge:Cambridge University Press.Trudgill, P. (2001). Received Pronunciation: Sociolinguistic aspects. Studia Anglica

Posnaniensia, 36, pp. 3-13.Watt, D. & Milroy, L. (1999) Patterns of variation an change in three Tyneside vowels:is this dialect levelling? In P. Foulkes and G.J. Docherty (eds) (1999). Urban Voices:

Accent Studies in the British Isles. London: Arnold, pp. 25-46.Wells, J. C. (1982). Accents of English (3 vols). Cambridge: Cambridge UniversityPress.

15

Yaeger-Dror, M. (2001). Primitives of a system for style and register. In Eckert P. andRickford J. (eds.), Style and Sociolinguistic Variation, Cambridge: CambridgeUniversity Press, pp. 170-185.