Quick viewing(Text Mode)

Towards a Sociolinguistics of Style Allan Bell

Towards a Sociolinguistics of Style Allan Bell

University of Pennsylvania Working Papers in Linguistics Volume 4 Article 2 Issue 1 A selection of papers from NWAVE 25

1-1-1997 Towards a of style Allan Bell

Gary Johnson

This paper is posted at ScholarlyCommons. http://repository.upenn.edu/pwpl/vol4/iss1/2 For more information, please contact [email protected]. Towards a sociolinguistics of style

This working paper is available in University of Pennsylvania Working Papers in Linguistics: http://repository.upenn.edu/pwpl/vol4/ iss1/2 U.PennWorkingPapersinLinguistics Volume4.1(1997)

TowardsaSouolinguisticsofStyle recognizesthatstyleoperatesonthefullrangeoflinguisticlev

els—inthepatternsofspeakingacrosswholediscoursesandcon

versationsaswellasinthephonologyorsyntax.Onthe'social' AllanBellandGaryJohnson

side,awiderangeoffactorsthatmayaffectthedifferentwaysan

individualtalksaretakenintoaccount—includingpurpose,topic,

1. Introduction genre,channelandaudience.

Thesecond,variationistapproachtostyleismuchmore

Thebasicprincipleoflanguagestyleisthatanindividualspeaker strictlydefinedonboththesocialandlinguisticdimensions.It

doesnotalwaystalkinthesamewayonalloccasions.Style waspioneeredbyLabovinhisNewYorkCitystudy(1966,1972)

meansthatspeakershavealternativesorchoices—a'thatway' andhasbeenfollowedanddevelopedincountlessstudiesinmany whichcouldhavebeenchoseninsteadofa 'thisway'.Speakers countriesinthesubsequent30-oddyears.Variationist sociolin talkindifferentwaysindifferentsituations,andthesedifferent guisticshasusuallyworkedwithmicroaspectsoflinguisticstruc

waysofspeakingcancarrydifferentsocialmeanings. ture—thealternationofspecificphonologicalvariants.Ithasalso

Thestudyofstylehashadachequeredcareerinsociolin- usuallyfollowedatightlydefinedapproachtothesocialdimen guisticsoverthepast20years,butisnowattractingmoreinterest sion,intermsofparticulardemographicparameterssuchasgender againfromvariationists.Thatrenewedinterestcanbedatedfrom orethnicity. theworkdonebyJohnRicicford& FayeMcNair-Knox,aspre So on theonehandwe havea verybroad-brush, sentedinaplenarypapertoNWAVE in1991.We concurwith 'maximalist'approachtobothlinguisticandsocialphenomena. theirassessmentinthepublisaedversion(1994:52): Andontheother,amuchmorerigorousattempttocontrolboth

thesocialandlinguisticvariables.Ourapproachinthispaperand Withrespecttotheorydevelopment,stylistic theworkitreportsonisinpartanattempttoblendthetwo,inpar variationseemstooffermorepotentialforthe ticularthequantitativerigourwiththequalitativebreadth. integrationofpastfindingsandtheestablishment

ofproductiveresearchagendasthanvirtuallyany 2. TheGistofAudienceDesign otherareainsociolinguistics.

Theworkthatwedescribebelowhasjustsuchagoal. Inapaperpublishedin1984,BelldevelopedtheAudienceDesign

Generalizinggrossly,wecandistinguishtwomainap frameworkwhichhashadsomecurrencysincethenasasociolin- proachestothestudyofstyleinsociolinguistics.Thefirst,ethno guisticapproachtostyle.AudienceDesignproposedthatstyle graphicapproach—associatedespeciallywithDellHymes(e.g. shiftoccursprimarilyinresponsetospeaker's the audience.Its

1974)—encompassesthemanywaysinwhichindividualspeakers maincontentionscanbesummarizedthus: canexpressthemselvesdifferentlyindifferentsituations.This

1. Styleiswhatanindividualspeakerdoeswithlanguage a

inrelationotherpeople. to We acknowledgethesupponoftheNewZealandFoundationforRe

2. Stylederivesitsmeaningfromtheassociationoflinguis search,Science& Technologyinfundingthestudyreportedbelowas part

ticfeatureswithparticularsocialgroups. oftheNewZealandEnglishProgrammeconductedintheDepartmentof

Linguistics,VictoriaUniversityofWellington.AllanBellisgrateful to 3. Speakersdesigntheirstyleprimarilyforandinresponse

theDepartmentforitshospitality(notalwaysfunded)overa periodof totheiraudience. years. 4. Audiencedesignappliestoallcodesandlevelsofalan

guagerepertoire,monolingualandmultilingual.

U.PermWorkingPapersinLinguistics,Volume4.1,1997 Towards a Sociolinguistics ofStyle Bell & Johnson U. Penn Working Papers in Linguistics Volume 4.1 (1997)

5. on the style dimension within the speech of a Table 1 Grid for interviews with 4 informants each talking to 3 single speaker derives from and echoes the variation different interviewers which exists between speakers on the 'social' dimension. INTERVIEWERS 6. Speakers have a fine-grained ability to design their style for a range of different addressees, as well as for other MM MF PM PF audience members. Pine Pania Paul Jen 7. Style shifting according to topic or setting derives its meaning and direction of shift from the underlying asso ciation of topics or settings with typical audience mem INFORMANTS bers. MM Duncan 1st 2nd 3rd — 8. As well as the 'responsive' dimension of style, there is the

MF Kay 2nd 1st — 3rd 'initiative' dimension, where the style shift itself initiates

a change in the situation rather than resulting from such a PM Lee 3rd — 1st 2nd change. PF Sally — 3rd 2nd 1st 9. Initiative style shifts are in essence referee design, by which the linguistic features associated with a reference group can be used to express identification with that Ethnicity: Maori Gender: Female group. Pakeha (Anglo) Male

These nine points nave been enlarged upon elsewhere (Bell in press), and the last three in particular critiqued and re vised. The basic premise of audience design is that style is ori 3. Designing Research on Style ented to people rather than to mechanisms such as attention. Style focuses on the person. It is essentially a social thing. It marks We now turn to report on a study which was explicitly designed to interpersonal and intergroup relations. test out several of the Audience Design hypotheses. It is a three- In initiative style shift, the individual speaker makes year project (just completed) which was funded by the New Zea creative use of resources often from beyond the immedi land Foundation for Research, Science & Technology under the ate speech community, such as distant , or stretches those NZ English Programme at Victoria University of Wellington. The resources in novel directions. With Bakhtin we may call this di project examines and seeks to explain the ways speakers talk dif mension 'stylization' (1981;, and the responsive simply 'style'. ferently to different audiences, and how they present their own Initiative style shifts derive their force and their direction of shift identities through language. from their underlying association with types of persons or groups. The language sample consists of three interviews con Referees are third persons who are not physically present at an ducted with each of four speakers. A set of four informants aged interaction but who are so salient for a speaker that they influence in their twenties were interviewed in succession by a set of four style even in their absence. This is the area where we believe interviewers (Table 1). The informant and interviewer samples audience design to be in need of serious rethinking. And this— were each structured by gender and ethnicity, so that each of them along with an approach to blending the quantitative with the qualitative—is the second main goal of the project we are working on, and of this paper. Towards a Sociolinguistics of Style Bell & Johnson U. Penn Working Papers in Linguistics Volume 4.1 (1997) contained a Maori2 woman, Maori man, Pakeha woman and view—the cross-gender combination—was gender, focussing the Pakeha man. Thus for example, the Maori man was interviewed informants on their own gender identity and its contrast with the first by the Maori male interviewer, second by the Maori woman, interviewers'. Similarly, the primary topic of discussion in the and third by the Pakeha man. The fourth possible combination of third, cross-ethnic interview was the issue of ethnic relations and interviewers and informants was intentionally excluded (the prac identity in New Zealand. ticalities of a fourth successive interview with each informant were This was an ambitious research design, particularly in its prohibitive). repeated interviews involving the same set of informants and in While gender and ethnicity were varied, other speaker terviewers. Recording failure or speaker withdrawal could have characteristics were held as constant as possible: jeopardized the whole project, requiring location of fresh speakers and re-recording interviews in order to maintain the integrity of the • Age: all eight speakers were in their early to mid 20s. design. However, all 12 interviews were completed despite this • Social class: all were middle class, university educated. potential for disaster. The interviews averaged over an hour long • New Zealand origins: all were New Zealanders of several gen each. They have been transcribed in full, timed, and their content erations' standing. logged under topic headings. The sample amounts to over 13 • Degree of familiarity: all informants and interviewers were hours of taped interviews, about 650 pages of transcripts, and a strangers to each other. total count of some 140,000 words.

In addition, we tried to keep aspects of the setting constant. 4. The Discourse Features • Interviews were conducted in the informants' own homes. • No third parties were present. • Interviewers were asked to dress in a similar and 'neutral' fash The linguistic analysis we will report on covers a subset of the features often known as pragmatic markers—typically the sen ion (neither too formal nor too casual). tence-final tags such as / think and like that we scatter like discour- The attempt to hold factors constant extended to interview sal and interactive glue throughout our conversational encounters. design as well. The elicitation of maximally informal speech had Among these features there is a subset sometimes known as the to be sacrificed to the need to ensure comparability across the in 'addressee-oriented' pragmatic markers—you know, tag questions terviews, e.g. by topic—one example of the different methodology such as isn 't it, and so forth. They have been studied in New Zea needed for style research. land by Janet Holmes and Maria Stubbe (e.g. Stubbe & Holmes Three standardized questionnaires were designed, one for 1995) in particular. The chief function of these features seems to each of the three interviews conducted with each informant. Each be interactive, for the speaker to seek reassurance of the listener's interview consisted of four components: free conversation, set continuing attention to what is being said, or confirmation of topics, reading tasks and other tasks. A basic principle of the in shared experience or knowledge. terview design was to make aspects of the informant's identity The four features we shall look at here are: Y'KNOW, salient at particular times. So the set topic for the second inter- TAG questions, the discourse particle EH and High Rising Termi nal intonations (HRTs). While Y'KNOW and TAGs need little introduction, the other two invite more discussion, partly because Maori are the indigenous Polynesian inhabitants and now make up some they are characteristic of NZE, although not exclusive to it. 15 percent of the population. 'Pakeha' is the term for New Zealanders of The particle EH functions syntactically very much like mainly British origin who colonized the country from the 19th century Y'KNOW or TAGs. EH also occurs in other varieties of Eng- (some 80 percent of the population). Towards a Sociolinguistics ofStyle Bell & Johnson U. Penn Working Papers in Linguistics Volume 4.1 (1997)

lish—at least Canadian (e.g. Gibson 1976) and the of um YOU KNOW they've still got to keep up with that Guernsey in the Channel Islands (Ramisch 1989). The leading original kaupapa of making sure our children are getting study of EH in NZ English to date is Meyerhoff s (1994) analysis taught the best they can yeah of the Porirua social dialect survey (Holmes, Bell & Boyce 1991). EH also carries considerable social meaning, which we will come Transcript 2 Kay: clustering of High Rising Terminal tones to shortly. Transcript 1 comes from the interview between the two (| = HRT) Maori men (pseudonyms used), and gives a sense both of the data in general, and also in particular of EH and its usage, K: I remember oh I was about eleven or twelve and um we'd The High Rising Terminal (HRT) is not a pragmatic par been jumping off this bridge ticle but an intonation pattern, however its discourse function is I: yeah=/ very similar. This intonation is becoming familiar in English in K: /=into this um oh into the water below it and it was ternationally, both through usage, and because of research and a lagoon going out to the open SEA| and there was quite a publication. It is in common usage in New Zealand, where the strong current taking all the water out //and\ um I'd dive= leading study is by David Britain (1992), again on the Porirua data. I: /yeahW K: =bombed this GUY| and splashed him so he started racing Transcript 1 Duncan—EH clustering over to the road BRIDGE| and I was swimming back against the current to the other side of the lagoon and um ( ) unclear speach my toes had just touched the GROUND|=/ = continuation of turn or latching I: /=yeah=/ // \\ overlapping speech K: /=and he jumped off the road bridge and hit me on my SHOULDERS| and jarred my SPINE| and I was kohanga reo: language nest (preschool immersion class) PARALYSED! I couldn't move=/ kaupapa: philosophy, principles I: /=God=/ K: /=and all I th-1 D: first we did Heretaunga and then er one a few um ko just thought YOU KNOW all I can do is try and float try hanga reo from Poneke (yeah) and then a few from and float and just lie back and relax and try and float and Rangitane and we er got back to the to the um real kau um I was going help me help me and Dad came out and papa of what kohanga reo is all about because it's becom rescued me and blew up the kid [inhales] and um and ing a bit of a business now EH and they're losing the los //then he found out I was\ you know I couldn't move for ing what it's the rea! meaning of it //(YOU KNOW) it's\ about two days and then I was fine for our children YOU KNOW= I: /mm\\ Transcript 2 is a danger-of-death narrative from the interview be D: =although a lot of the people in there EH they work tween the two Maori women. blimmin hard man and they get stuff all for it and some Initially we will present quantitative findings on the dis times you don't blame them EH 'cause they're getting no tribution of these features. But then we want to move on to what rewards out of it //but um\ we're trying to (and then)= we consider to be a complementary approach, that is a more I: /mm\\ qualitative analysis of where the features occur on-line during D: =tha- that sort of thing EH er with kohanga reo there will speech and why, and also how the four features co-occur—or oth never be many rewards for the people working in it but erwise—with each other.

8 Towards a Sociolinguistics of Style Bell & Johnson U. Penn Working Papers in Linguistics Volume 4.1 (1997)

By way of orientation, we present in Table 2 the raw 4. EH occurs overwhelmingly in the speech of the counts of the features, with no allowance for amount of talk or Maori man Duncan, although there are some tokens interview length. We can make some observations on the strength by Kay the Maori woman. of these figures for tokens: 5. HRTs are common except by Lee the Pakeha man. Note that there is a kind of complementary distribu 1. Y'KNOW is the feature of choice, especially for the tion of the two last features for the Pakeha man and Pakeha man Lee in expository mode. It appears to carry woman—Lee uses Y'KNOW and not HRTs, and little identity meaning, although research would tend to Sally HRTs and not Y'KNOW. associate it with women's style rather than men's. 2. But for the Pakeha woman Sally the default feature is the 5. Quantitative Analysis HRT, and she has remarkably few Y'KNOWs. (She also has some other individualistic preferences—e.g. always One of the main problems with discourse variables is deciding using KIND OF where the other three use SORT OF.) what to count. The main issue is what do we count as potential but 3. Tags are infrequent, but there is an indication that they not actual occurrences of pragmatic features such as HRTs or EH? are used more by Pakeha than Maori. Here we have quantified all four features over the amount of speech produced by the particular speaker, and amount of speech Table 2 Number of tokens of 4 addressee-oriented pragmatic fea is in terms of word count. This produces an index for the feature, tures in the speech of 4 informants talking to 3 different interview which consists simply of the number of occurrences of the feature ers divided by the number of words produced by the speaker, and then Number of tokens multiplied by 10,000. The multiplier of 10,000 yields indexes By To generally in double digits, so easy to grasp. And 10,000 words is Informant Interviewer Y'KNOW TAG EH HRT actually close to the average amount of informant speech per in terview, so it represents in some sense a normalized interview Maori MM Pine 133 0 48 32 length. man MF Pania 98 2 20 56 Duncan PM Paul 69 0 16 53 5.1. EH by Informants

Maori MF Pania 39 0 3 50 The pragmatic particle eh is one of the most high-profile sociolin- woman MM Pine 86 1 2 40 guistic markers of English within New Zealand. It is criticized by Kay PF Jen 29 0 0 16 prescriptivists, satirized by comedians, and utilized by advertising copywriters to create social caricatures (Bell 1992). Both the New Pakeha PM Paul 21 3 0 17 Zealand stereotype and the research findings associate the variable man PF Jen 106 4 1 7 EH with the speech of Maori rather than Pakeha, and to a lesser Lee MM Pine 210 8 0 9 extent with men rather than women. In Table 3 EH is used by Maori speakers, overwhelmingly by the Pakeha PF Jen 26 4 0 59 Maori man Duncan—84 tokens in all (see Table 2 for raw tokens). woman PM Paul 5 1 0 31 In fact his index while talking with the Maori male interviewer is Sally MF Pania 8 2 0 55 similar to the index for young Maori males in Porirua study

10 Towards a Sociolinguistics ofStyle Bell & Johnson U. Penn Working Papers in Linguistics Volume 4.1 (1997)

Table 3 EH Index in speech by Informants to Interviewers 5.2. HRTs by Informants

To Interviewers The High Rising Terminal involves an intonation pattern in the form of a high rise, questioning pattern, but used on a tone group Pine Pania Paul Jen which is a statement. One interpretation is that its use indicates MM MF PM PF hesitancy or doubt, but NZ researchers have interpreted it as a marker of interactive solidarity and affect (Britain 1992). This By Informants feature is stereotypically associated in New Zealand mainly with Duncan MM 46 26 19 — younger Pakeha women. The research partly confirms this. David Kay MF 2 4 — 0 Britain's findings were that HRTs are used mainly by younger Lee PM 0 — 0 1 0 speakers (i.e. our group of speakers), particularly by women, and Sally PF — 0 0 to a lesser extent by Maori. The research has also shown that HRTs are sensitive to the genre or text type of the speech in which (Holmes, Bell & Boyce 199 i). Kay the Maori woman also uses they occur, being particularly common in narratives. The analyses some EH, but at a much lower frequency -only 5 tokens. By con we present unfortunately still lack this sub-categorization. trast, the Pakeha speakers use virtually no EH. Sally uses abso In Table 4 Sally the Pakeha woman uses by far the high lutely none at all, and there Is only 1 token from Lee the Pakeha est level of HRTs. Lee the Pakeha man uses very considerably the man in nearly four hours of recorded talk. lowest, and the others are in between. So the identification of We can see thus how EH is functioning mainly as a HRTs with women, particularly Pakeha, seems confirmed, and marker of group identity primarily of ethnicity (Maori), and sec also possibly with Maori. ondarily of gender (Maori men). This pattern of usage fits the Who are HRTs used to? Tracking the shifts between in association of linguistic features with group usage which we out terviews, we can see that HRTs are used more to women than to lined in the summary of audience design above. It also accords men. So Duncan, the Maori man, uses most HRTs to Pania the both with our previous findings, and with popular stereotype. Maori woman, and fewer to the two men who also interviewed Turning to the shifts which informants make in different him. Sally the Pakeha woman uses fewest to Paul the Pakeha man, interviews, as hypothesized in audience design, the speakers use different amounts of EH with different interlocutors. In particular, Duncan the Maori man uses EH more often in interview with Pine Table 4 HRT Index in speech by Informants to Interviewers the Maori male, less with Pania the Maori woman, least with Paul To Interviewers the Pakeha man. At a very much lower level of frequency, this is parallelled by Kay the Maori woman informant. She uses some Pine Pania Paul Jen EH with her most like interlocutor (Pania the Maori woman), less MM MF PM PF with Pine the Maori man, and none with Jen the Pakeha woman

(despite the Pakeha female interviewer using one token of EH her By Informants self). Duncan MM 31 72 62 — These are the kinds of fine-grained shift which is the core Kay MF 34 72 — 38 principle of audience design as outlined above. It conforms with Lee PM 6 — 23 6 an interpretation of EH as a marker of Maori identity, particularly 80 Sally PF — 90 60 for men.

11 12 TowardsaSociolinguisticsojStyle Bell& Johnson U.PermWorkingPapersinLinguistics Volume4.1(1997)

moretothetwowomenwhointerviewedher.Thesamepattern mographicallydistanttheinterlocutor—inthespeechofPaulthe holdsforKaytheMaoriwoman,althoughherfrequencytoJenthe Pakehamaleinterviewer,quitestrikinglyso.Inthebaselinein

Pakehawomaniscloseto thatPine totheMaoriman.Soagain, terviewwithLeethePakehamaleinformant,PauluseszeroEH—

wehavesomeconfirmationthattheinformantsareshiftingtheir demographicallyappropriate.HehasanEHindexof14toSally

styleaccordingtotheiraudienceforHRTs. thePakehawoman,whichintermsofherownlinguisticbehaviour

Butquestionsremain:whydoesthePakehamaleinfor isinappropriateaccommodationassheherselfusesnoneatall. mantLeeproducehisonlyappreciablelevelofHRTstoPaul,the Mostnotably,hisindextotheMaoriman—thehighEHuser—is

Pakehamaleinterviewer,andnotinparticulartothePakeha 29. woman?Andwhy—ifHR.Tsareparticularlyidentifiedwith Thiscanreasonablybeinterpretedashyper-accommoda

Pakehawomen—aren'tmoreusedtothePakehafemaleinter tion.PaulinfacthasagooddealhigherlevelofEHintheinter viewerJen? viewthanhisinformantdoes.Theinterpretationofhyper-

accommodationreceivessupportfromotherfacetsofthisinter

5.3. EH byInterviewers view.Paulwasclearlynervousinconductingit,presumablybe

causeofthe ethnicdistinctionbetweenhimandtheinformant.

Sofar,sotidy(moreorless).Letusturnnowtotheinterviewers' Thiswasmarkedinvariousways,butespeciallythroughagood

usageofEH andHRTinthesesameinterviews.Hereitneedsto dealofprolongednervousgigglingatquiteinappropriatepointsof berememberedthatthesewerenotordinaryconversationswith theinterview. bothparticipantsclaimingequalrightstospeakingtime.These Sowecanobservethatthereismutualaccommodationin wereinterviews,andtheinterviewersprovidedmuchlessofthe Paul'sinterviewwithDuncantheMaoriman,withtheMaoriman talk.Thekindoftalktheyprovidedalsonecessarilymilitated shiftingtoalowerEH level,andthePakehashiftingfromazero

againstusageofsomeofthesepragmaticfeatures.Inparticular, basetoaquitehighlevelofusage.Lookingatthenumbers,we bothY'KNOWandHRT byandlargetendtooccurinaflowof couldinfactsaythattheyareshiftinghalfwaytomeeteachother. talkofa kindwhichinterviewersarenotusuallyproducing. TheMaorimaleinformantDuncanreceivesmoreEH

Ontheotherhandtheirroleasinterviewersistoestablish thananyoneelse.ButalongsidethatwehavetonotethatSallythe thekindofrapportwiththeinformantsthat willencouragethemto Pakehawomanreceivesalmostasmuchwithindexesof35,14

relaxandtalk.Thatis,thepressuresontheinterviewertoaccom modatetotheinformantareprobablygreaterthanviceversa,de

Table5 EH IndexinspeechbyInterviewerstoInformants spitethecomparativelylittlespeakingtimetheinterviewerwill havetodisplaythislinguistically. By Interviewers Table5showsthattheinterviewersusemoreEH thanthe

informants(cfTable3)—withtheoneexceptionofDuncanthe Pine PaniaPaul Jen

MaorimantalkingtoPineiheMaorimaleinterviewer.Why is MM MF PM PF this?Ingeneraltermswecanreferittotheirroleintheinterview

andtheonusthatison themtointeractinapositiveandsolidary ToInformants waywiththeinformants.Itappearsthatinordertodothis,they makeuseofthefeaturewhichisavailableintheNZE speech Duncan MM 6 28 29

communityforthatfunction,tosomeextentwithoutregardforthe Kay MF 10 25 5

socialmeaningsitbringswithit. Lee PM 0 0 3

ThereisanindicationthatmoreEHisusedthemorede- Sally PF 35 14 9

13 14 TowardsaSociolinguisticsojStyle Bell& Johnson U.PennWorkingPapersinLinguistics Volume4.1(1997)

and9. Why?!—well,we thinkthisisaudiencerelated,butnot Table6 HRTIndexinspeechbyInterviewerstoInformants

demographic.Sallyisaslow,hesitantspeaker—notuncoopera

tive,justreticent,theleasttalkativeofthefourinformants.And ByInterviewers

ourinterpretationisthattheinterviewersworkedparticularlyhard

Pine PaniaPaul Jen toencouragehertalk,andthisexplainsthehighusageofEH to MM MF PM PF her,asitwereagainstthedemographicassociationsofthefeature.

NoticethatthelevelofEH fromtheinterviewersin ToInformants creasesasthedemographicdistanceincreases,sothatJenthe

Pakehawomaninterviewerusesleast(9)Sally, to PaulthePakeha Duncan MM 0 5 10

mannext(14),andPaniatheMaoriwomanmost(35).Thisseems Kay MF 5 12 0 5 tobecountertowhatonewouldexpectfromaudiencedesign:the Lee PM 0 0

interviewersareusingafeaturewhichdoesnotappearinherown Sally PF 22 7 9

speech.Itisindeedcountertothedemographicassociationsof

EH,butwethinktheinterpretationjustofferedisnotjustanad- andinterviewersdonotdomuchofthat.Neverthelessthepattern

hocattempttorescuetheframeworkinthe faceofcontraryevi ofPaulthePakeha maleinterviewerforHRT isexactlythesame

dence. asforEH—againusingmoreofthisfeaturethemoredistantthe

Italsoraisesanimportantpoint.Whilewehavepre demographicsoftheinformant.HeuseszeroHRTstoLeethe

sentedouranalysisherelargelyintermsofdemographiccharacter Pakeha man,anindexof7toSallythePakehawoman,and10to

istics,accommodationtoone'saudienceisinfactmuchwiderthan DuncantheMaoriman.Sothispatternmayreflectsomething

that.Itincludesspeakersmakingactiveuseoftheresourcesof genuineaboutthisinterviewer,hisinformantsandtheirinteraction

theirspeechcommunityinordertoaccomplishtheirconversa intheseinterviews.

tionalpurposes,inthiscaseasuccessfulinterview.Youwillsee Relativeusagetoinformantsisinlinewiththeinfor

thatwearenowdoingwhatRickford& McNair-Knox(1994) mants'ownproductionofHRTsinTable4—byfarthemostHRTs

eventuallyfoundtheyhadtodoastheyexploredthestylepatterns areusedtothePakehawoman,andleasttothePakehaman.In

oftheirspeaker—interpretingasbestwecanwhatappearsinthe addition,relativeusagetointerviewers(Table4)hasexactlythe

samestructureasusagebyinterviewers(Table6). quantification,eventhoughitdoesnotfitourhypothesesvery

well.Thiscanbe regardedintwolights:eitherascommonsense

flexibilitytointerpretthemeaningofthepatternsthatoccur,using 6. QualitativeAnalysis whateverexplanationseemsmostappropriate.Oralternatively,it

canbeseenaspost-hocrationalizingofwhateverhappenstoturn We believethequantificationjustoutlinedtellsusa gooddeal upevenifitconflictswithyourowntheorizing.Thisseemstobe aboutthestyleofspeakerswithintheseinterviews.Butsuch oneofthemainissuesforanyattemptataframeworkforregular

quantificationdoesnottelluseverythingthereistoknowabouta izingstyleshift. speaker'sstyleinacertainstretchoflanguage.Linguisticvari

ablesdonotjustoccurasfeaturestobecountedwithinanundif- 5.4. HRTsbyInterviewers ferentiatedchunkofspeech.Theyoccuron-lineintheflow of

speech.Whenandwheretheyoccurinthecourseofaconversa Toshowthatourinterpretationsarenotpurelyadhoc,lookatTa tionmaytellussomethingaboutthespeaker'sstyle,abouttheir ble6,whichpresentstheinterviewers'usageofHRTs.Nowrecall patternsofidentityexpressionoraudiencedesign.Individualto- thatHRTsgenerallyoccurinnarrativesorat leastaflowoftalk,

15 16 TowardsaSociolinguisticsojStyle Bell& Johnson U.PennWorkingPapersinLinguistics Volume4.1(1997)

kensofthevariablemayhaveheightenedsignificanceintheflow lengthandingreatdetail—foroverhour an onethnicrelationsin

oftheinteraction,ortheymayclustertogetherwith eachotheror NewZealand.HeproducesfrequentclusteringsofY'KNOW,with withothervariablesinawaywhichissignificant. 5-10tokensinarun,andeventheoccasionalpairof2Y'KNOWs

Soforthe12interviews,wehavegraphedtheoccurrence followingeachotherdirectly.Itstriking is thatwheretheother on-linethroughouttheinterviewofthefourfeatures,partlywitha speakerswoulduseHRTsinnarrativesornarrative-liketextsorin viewtoseeingwhereindividualtokensofthefeatureoccur,more sensitive-topiccontexts,LeeusesarunofY'KNOW. particularlytoseeiftokensclusterorscatter.Secondly,inorderto Forexample,whendiscussingtheissueoftheirown seeifthereareanynoticeablepatternsofco-occurrencebetween competenceintheMaorilanguage(inInterview3,withthesame- thedifferentfeatures,sothatdifferentfeaturestendto eitheroccur genderMaoriinterviewer),Leeusesa clusterof12Y'KNOW. togetherorbeincomplementarydistribution. WhenthesamequestioncameupinSally'sinterview,shepro

Figure1 trackstheoccurrenceoftokensofthefour ducedaseriesofHRTs.Exactlythesamehappenedinbothinter

pragmaticfeaturesthroughoutinterview.ThisisInterview1be viewswhenthequestionofsensitivitytoMaoricustomsinrelation

tweenthetwoMaorimen,DuncanandPine.It86 isminuteslong, tofoodandhygienearose—astringofY'KNOWfromLee,andof taking nearly3 sidesoftape.Thedifferentsectionsoftheinter HRTfromSally. viewaremarkedinthefigure. Thirdly,likeY'KNOW,HRTstendtocluster,butmorein

We canseewhatwealreadyknewfromthequantitative pairsthanmultiples—althoughmultiplesdooccur.Tothosefamil analysis—therearezeroTAGsinthisinterview.Theotherthree iarwiththefeature,thiswillbenosurprise.EH ofcoursecan featuresdooccur,however.EH makesaslowstart,butincreases reallyonlybeexaminedinthespeechofDuncantheMaoriman.

especiallyinthelastquarterjftheinterview,andthereisacertain Itnoticeable is thatDuncantendstoholdbackuseofthisethnic

amountofclusteringoftokens.ThepatternforY'KNOWissimi identitymarkeruntilsomewayintoeachinterview.Evenin

lar—alatestarter,withsomeclustering,butitoccursmorefre speakingtoPinetheMaorimaleinterviewer,hescattersjust 3 to quently.WealsohavetowaitforHRTstobeginoccurring—first kensacrossthefirst20minutesoftheinterview,beforesettling tokenat9minutes,secondandsubsequenttokensfrom17min intoamuchmoreregularlevelofthevariable.Itisasifheis utes—andthetokensaremorescattered. testinghisinterlocutorout beforecommittinghimselftouseofthis

Theoccurrenceoftokenson-linehasbeengraphedina ethnicallymarkedparticle.WithPinetheMaoriman,itisadis similarfashionforallinterviews,enablingtomake us somegen cussionofhisgrandmother'stangi(funeral)thattriggersa runof eralizationsaboutthequalitativeoccurrenceofthefeatures.First, EH tokens.TherearenotmanyrunsEH, ofbut theyoccurin wecanseeataglancethatTAGsarerareandscatteredthrough variablywith'Maori'topics—family,thereo(Maorilanguage), out,withnoobviouson-linepatterning. Maoriculture,etc.Theyoftenco-occurwithuseofMaorilexical

Secondly,tokensofY'KNOWarescatteredthroughout itemsborrowedintotheEnglishdiscourse. mostinterviews.Itclearly is thedefaultfeature,asthequantifica Maoriissuesarediscussedfrequentlyinthisinterview, tiontendedtoindicate.Itsrelativelackofsocialassociationsis andusuallytriggeratokenortwoofEH.Whilespeakingtothe

showninitbeingthemostevenlydistributedthroughoutthein Pakehaman,however,aboutMaoriissuesitisverynoticeableon terviewsofthefourfeatures. theon-linegraphthatDuncanproduceslittleEH,butlot a ofHRT

HoweverthereareparticularconcentrationsofY'KNOW andY'KNOW.ThisreinforcesourviewofEH asaningroup intheSetTopicsectionsofseveralinterviews.Itisclearthatfor identitymarker.DuncandoesnotuseittoclaimMaorinesstoa thePakehamanLeeespeciallyY'KNOWisaprimaryexpository non-Maori,butto establishsolidaritywithotherMaori. particle.Inthediscussionof genderandethnicissues,whichwere thesettopicsininterviews2 and3,heexpoundshisviewsat

17 18 TowardsaSociolinguisticsofStyle Bell& Johnson U.PermWorkingPapersinLinguistics Volume4.1(1997)

7. Conclusion A

■is

.253 Inconclusion,wewouldarguetwothings.First,weneedcom ^ .

; plementaryqualitativeandquantitativeanalyses.Atveryleast,the

)? qualitativeenrichesourinterpretations.Itmayevenchangethe 3= ft interpretationswewouldreachbasedonlyonquantification.It

showsupthingsthatarenotevidentinquantification,suchasthe **>

"J differentusagesbydifferentinformantsintheseinterviewsonthe

sametopic.Talkisanon-linephenomenon.Whenwecount,we

a necessarilylumpthingstogether,andthatisa needfulpartof

analysis.Butthereisalsoatimetokeepthingsseparate,andex s aminehowindividualtokensareoperatingon-lineintheflowofa A* —' conversation.

Secondly,wewouldargueforacomplementarityofaudi

enceandrefereedesign,ofresponseandinitiative,oftherelational

andtheidentityfunctionsoflanguage.Inthissensewewould

ft nowwanttoconsidermodifyingtheoriginalaudiencedesignpro

posalssothataudienceandrefereedesignareregardedasoperat s

— inginparallel,ratherthanrefereedesignbeinganoccasionalfac

■t" £ <» tor.Whatwecanobservefromourinterpretationsaboveisthat

whenaudiencedesignseemsnottohold,ourpost-hocexplanation

5« ofwhatisgoingonisstilllargelyeitherintermsoftheaudience S (forexample,whymoreEH isusedtoSally)orofanidentity,

refereefunction(e.g.intheclusteringofEH).Approachestostyle

SI areincreasinglytakingaccountofthesefacts.Certainly,webe --c lieveasociolinguisticsofstylewillbefoundinthefusionofthe O audienceandthereferee,andthequantitativewiththequalitative.

£

References

Bakhtin,M.M.(1981).TheDialogicImagination.Austin:Universityof v>

TexasPress.

Bell,Allan(1984)."Languagestyleasaudiencedesign."Languagein

Society13:145-204.

Bell,Allan(1992)."Hitandmiss:refereedesigninthedialectsofNew

Figure1.On-lineoccurrenceoftokensoffour addressee-ori Zealandtelevisionadvertisements."Language& Communica entedpragmaticfeaturesinspeechofMaorimaleinformant tion12:327-340.

(Duncan)toMaorimaleinterviewer,Pine. Bell,Allan(inpress)."Backinstyle:Re-workingAudienceDesign,"to

19 20 Towards a Sociolinguistics oj Style Bell & Johnson

appear in John Rickford & Penelope Eckert, eds., Style and The University of Pennsylvania Working Papers in Linguistics Variation (Cambridge: Cambridge University Press). (PWPL) is an occasional series produced by the Penn Linguistics Britain, David (1992). "Linguistic change in intonation: The use of high Club, the graduate student organization of the Linguistics rising terminals in New Zealand English." Language Variation Department of the University of Pennsylvania. and Change A: 77-104. Gibson, Deborah (1976). "A thesis on EH." Unpublished MA disserta Publication in this volume does not preclude submission of papers tion. Vancouver: University of British Columbia. elsewhere; all copyright is retained by the authors of the individual

Holmes, Janet, Allan Bell & Mary Boyce (1991). "Variation and Change papers. in New Zealand English: A Social Dialect Investigation" (Proj ect report to the Foundation for Research, Science & Technol Volumes of the Working Papers are available for $12, prepaid. ogy). Wellington: Victoria University, Linguistics Department. Please see our web page for additional information. Hymes, Dell (1974). Foundations in Sociolinguistics: An Ethnographic Approach. Philadelphia: University of Pennsylvania Press.

Labov, William (1966). The Social Stratification of English in New York The PWPL Series Editors City. Washington, D.C.: Center for . Labov, William (1972). Sociohnguistic Patterns. Philadelphia: Uni Alexis Dimitriadis versity of Pennsylvania Press. Laura Siegel Meyerhoff, Miriam (1994). "Sounds pretty ethnic, eh? - a pragmatic par Clarissa Surek-Clark ticle in New Zealand English." Language in Society 23: Alexander Williams 367-388. Ramisch, Heinrich (1989). Variation of English in Guernsey/Channel Islands. Frankfurt am Main: Verlag Peter Lang. Editors for this Volume Rickford, John R & Faye McNair-Knox (1994). "Addressee- and topic- influenced style shift: a quantitative sociolinguistic study," in Charles Boberg Douglas Biber & Edward Finegan, eds., Sociolinguistic Per Miriam Meyerhoff spectives on Register (New York/Oxford: Oxford University Stephanie Strassel Press). 235-276. and the PWPL series editors Stubbe, Maria & Janet Holmes (1995). "You know, eh and other 'exasperating expressions': An analysis of social and stylistic variation in the use of pragmatic devices in a sample of New Zealand English." Language & Communication 15: 63-88. How to reach the PWPL

U. Penn Working Papers in Linguistics Allan Bell (Dr.) Gary Johnson Department of Linguistics 619 Williams Hall PO Box 24-495 (Postal) 1 Grove Road University of Pennsylvania Royal Oak Kelburn Philadelphia, PA 19104-6305 Auckland 3, New Zealand Wellington 5, New Zealand

702 Manukau Road (Courier) grjohnson@sol. otago. ac. nz working-papers@ling. upenn. edu Royal Oak http://www. ling, upenn. edu/papers/pwpl. html Auckland 3, New Zealand allan@bell. ak.planet.gen. nz

21