IDEOLOGIES OF HONORIFIC LANGUAGE

Pragmatics 2:3.25l -262 InternationalPrasmaticsAssociation IDEOLOGIES OF HONORIFIC LANGUAGE Judith T. Irvine 1. Introductionr All sociolinguistic sy...
Author: Osborn Lawson
1 downloads 2 Views 796KB Size
Pragmatics 2:3.25l -262 InternationalPrasmaticsAssociation

IDEOLOGIES OF HONORIFIC LANGUAGE Judith T. Irvine

1. Introductionr All sociolinguistic systems,presumably,provide some meansof expressingrespect (or disrespect);but only some systems have grammaticalized honorifics. This paper comparesseverallanguages- Javanese,Wolof, and Zulu, plus a glance at ChiBemba - with regard to honorific expressionsand the social and cultural frameworks relevant thereto.2The main question to be explored is whether one can identiff any special cultural concomitants of linguistic systems in which the expression of respect is grammaticalized. Javanese"languagelevels"are a classicand well-describedexample of a system for the expressionof respect. In the sense in which I shall define "grammaticalized honorifics,"Javaneseprovidesan apt illustration.Wolof, on the other hand, does not. Of course,Javaneseis only one of several Asian languageswell known for honorific while Wolof, spokenin Senegal,comesfrom another part of the globe. constructions, But the presence or absence of honorifics is not an area characteristic of Asian languagesas opposed to African languages.As we shall see, Zulu has a system of lexicalalternatesbearing a certain typological resemblanceto the Javanesesystem. Moreover,many other Bantu languages(suchas ChiBemba) also have grammaticalized honorifics,but in the morphology rather than in the lexicon. Focusing on social structure instead of on geographical area, one might hypothesizethat grammaticalized honorifics occur where there are royal courts (Wenger1982)and in societieswhosetraditionsemphasizesocialrank and precedence. Honorificswould be a linguisticmeansof expressingconventionalizeddifferencesof rank.The languagesI shall compare will make it evident,however,that a hypothesis causallylinkinghonorificswith court life or with entrenchedclassdifferencescannot be

1 An earlierversionof this paperwas presentedat a sessionon "Languageldeology"at the 1991annualmeetingof the AmericanAnthropologicalAssociation.Thanksare due to Bambi Schieffelin, PaulKroskrity,Kathryn Woolard,and Debra Spitulnik for their helpful comments. 2 Th. discussion of Wolof in this paperdrawson my fieldworkin Senegal,mainly in the 1970's. I am gratefulfor the supportof the NationalInstituteof Mental Health,lhe National Scienc€ Foundation,and BrandeisUniversity.ChiBembacitationscome from elicitation sessions with ChiBembaspeakersat BrandeisUniversity.

252

JudirhT. Irvine

adequate.While some such link may hold true for Javanese,it does not for the other systems.The Wolof had royal courts until the French conquest a century ago, and retain (especiallyin rural areas)a socialsystemstructuredon inequalitiesof birth and family origin - inequalitiesso marked that the ethnographicliterature on the region usually describesthem as caste differences.Neither in preconquesttimes nor today, however, is there any indication of grammatical honorifics in the Wolof language, althoughthere were and are other waysto expressdeference.Zulu society,even at the height of the Zulu state,was somewhatlessstratifiedthan Wolof society(to saynothing of Java),and the Bemba polity was lessstronglycentralizedthan any of the other cases; yet both Zulu and ChiBemba have honorifics.Courts and social stratificationare not irrelevant to honorific language,but they do not predict honorifics'grammaticalization. In exploringthesesociolinguisticsystemsI do not believeany simple correlation between forms of "on-the-ground"social structure (such as the existenceof a royal court) and forms of talk (such as honorifics)is likely to be found. Instead, as I have argued elsewhere(Irvine 1985, 1989),the relationshipbetween the distributions of socialand linguisticforms is more productivelysoughtin cultural ideologiesof language - those complex systemsof ideasand intereststhroughwhich people interpret linguistic behaviors. In this paper, therefore, I shall pay special attention to the linguistic ideologiesthat link ideas about languagewith ideas about social rank, respect, and appropriate conduct- includingthe nativemetapragmaticterminologyand theoriesthat articulate and rationalize perceptionsof languagestructure and use (see Silverstein 7979). I draw on a concept of ideology,rather than merely a "culture of language," because "ideology,"whatever else it may mean, suggestsa connection with power relations and intereststhat are fairly central to a social order. Some such connection is surely relevant to honorific language. As Silverstein(1979)points out, linguisticideologymust be clearly distinguished from linguistic structuresand from the distribution of uses.It is this distinction that makes the present comparison possibleand, further, sheds light on these systems' historical dynamics,as I shall briefly observe.

2. Linguistic structures To saythat honorificsare grammaticalizedin a particularlanguageis to saythat expressionsof deference,or of differential status-marking,are incorporated into the language'sgrammaticalrules (rules which include its lexicon). Thus a system of grammatical honorifics is a system of alternate linguistic expressionswhich are values,they differ only in their isosemantic:havingthe samereference-and-predication pragmatic values (expressing degrees of cleference,respect, or distance). That pragmaticvalue operatesas part of sentence-meaning, not utterance-meaning. That is, in honorifics,deferenceis incorporatedin the constructionof the sentenceper se, rather than dependingupon how the sentenceis deployedin its socialor discourse context.(Note that the possibilityof regular,sarcasticusesof honorificsdependson this condition.) In Javanese,respectful expressionoperates through a system of lexical

Ideologies of honoiftc

langtage

253

alternates.In the sentencein (1), taken from Errington (1988),each word has a set of alternates,whose combinationsdefine a systemof six "levels"of speechstyle:3 (1)

Javanese"languagelevels"(Errington 1988:90-91):

KMMA:

I. menapa 2. menapa

nandalem mundhut panjenengan mendhet

sekul semanten sekul semanten

MADYA:

3. napa 4. napa

sampqan sampqan

mendhet njupuk

sekul semonten sega semonten

NGOKO:

5. apa 6. apa

sliramu kowe

mundhut njupuk

sega semono sega semono

Gloss:

Question marker

'you'

'take'

'rice' 'that

much'

'Did you take that much rice?' Though rarely exhibiting the complexity and elaboration of the Javanese languagelevels,systemsof honorific lexical alternates- respectvocabularies- are also to be found in many other languages.Among such languagesare Zulu and its closest relatives,the other Nguni languages(Xhosa and Seswati). In Zulu and Xhosa the respectvocabularyis known as hloniphc. A few examplesare given in (2): (2)

Zulu hlonipha vocabulary (Doke & Vilakazi 1958):

'graze;weave' 'be dejected' 'affair' 'my father' 'hippopotamus' 'lion' 'house' 'our' 'thy'

ORDINARY

HLONIPHA

aluka jaba inda6 a Lt6a6 a imvu6 u imbu6 e indlu -itltu -kho

acuka gxaba injuJo utlat{a incu6 u inju6 e incumba -itlu -to

I Note that Errington (1988)doesnot call the languagelanguagelevels'honorifics,nbut insteadreserves that term for thoselexicalitemsexpressing respectfor a referentrather than an - unlikesomeother authorswho use "honorifics"for both. Although a distinctionbetween addressee reference andaddressforms is important for his analysis,I prefer the broaderusagefor the mmparativepurposesof the presentpaper.

I

I

254

Judith T. Irvine

= clicks; 6 : implosivebilabial stop) " Many Bantu languagesfound to the northeast of the Nguni group also have respect forms, but locate them in the morphology of the noun classsystemrather than in a set of alternate stems. In ChiBemba (a languageof Zambia), for example, there are no sound shapes exclusively reserved for honorific reference, but respect is expressedby the use of plural prefixes (or pronouns) for singular human referents, as in (3). Noun classes 1 and 2, the singular and plural classesused for most nouns referring to humans, are the main ones affected. Thus the class 2 prefix that marks plurality in (3c) marks honorific singularin (3b): (c,

(3)

ChiBemba noun prefixes,classes1/2: (3a) not respectful:

isabi aleelya umo umukalamba waandi 9a 11 11 subject-tense-eat fish one older sibling my 'One of my older siblingsis eating fish' (3b) respectful: isabi

bamo abakalamba baandi

baleetya

222 one older sibling my

9a 2 subject-tense-eatfish

'One of my older siblingsis eating fish' (3c) ambiguous:

isabi baleelya babili abakalamba baandi 29a 222 subject-tense-eat fish older sibling my two 'Two of my older siblings are eating fish' Ip its noun classsystem,ChiBemba also provides for various pejorative usages, by shifting the classassignmentof a noun with human reference, as in (4): (4)

Chibemba honorific and pejorative noun prefixes (singular referent):

CLASS VALUE abakaJi umuka{i akakasi ilika{i ilikali

'(respectable)wife' 'wife' '(insignificant) wife' '(gross) wife' '(egregious [?]) wife'

2 1 12 7 5

honorific disrespectful insult insult "a little derogatory"

of honoific tanguage 255 Ideologies All these languagesthus have systemsof alternate expressionsdiffering only in pragmatic value. In all of these cases, the grammatical rules involved apply, fundamentally,to word formation (selection of stem or of prefix); the formation of sentencesis affected in consequence,through concord patterns and cooccurrence constraints. Except in (3c), the pragmaticvalue is unambiguousand undeniable. In contrast,Wolof does not have thesekinds of rules.Speakersexpressrespect in other ways.Although Wolof does have speechregistersconnoting social rank (as I havedescribedelsewhere,Irvine 1990),the registersare constructedquite differently from,for instance,the Javaneselanguagelevels.In Javanese,each word selectedfrom an isosemanticset has its particular pragmaticvalue. In Wolof, however, individual wordsor sentencesin the two registersare not strictly isosemantic,except insofar as utterancesmight rely exclusivelyon prosodiccontraststo create register differences.As outlined in (5), these prosodic patterns, though characteristic and striking, are and operatemore on the level of utterancemeaningthan of sentence non-segmentable meaning: (5)

Wolof prosodicpatterns:

pitch: volume: tempo: voice: contour: dynamicrange:

waxu g4€r 'Noble speech'

wctxttgewel 'Griot speech'

low soft slow breathy pitch nucleuslast narrow

high loud fast clear pitch nucleusfirst wide

Apart from their prosody,the Wolof registersdepend on semantic differencesand rhetoricalelaboration.Pragmaticvalueis built up over an expanseof discourse,for the registersembody contrastingrhetorical strategiesthat can rarely be displayedin an individualword or a brief, decontextualizedsentence.

3. Linguisticideologies lrt us turn to the linguistic ideologiespertaining to three of the languagesso far Wolof, and Zulu. How do speakersof these languagesperceive mentioned:Javanese, theseexpressions of deference,and how do they theorize about ihem? How do they with ideasabout respect,rank, and appropriate conduct? connectsuchexpressions For Javanese,I draw on Errington's (1984,1988)elegant and complex studies of the speechof the pruyayi,the traditional elite. I attempt no re-analysisof the material,other than my broader use of the term "honorifics." The Javanesesystem'scomplexityand subtletyare evidentlyrecognizedby the speakers themselvesas characterizingboth the systemand its highest-rankingusers.

Judith T. Irvine

Indeed, ideas about subtletyand refinement,on the one hand, and violence and anger on the other, seem to be crucial to the prryayi conception of their language.The "higher" (basa,especiallykrama) levelsare consideredto be governedby an ethic of proper order, peace,and calm. In them one "does not express one's own feelings"(Wolff and Poedjosoedarmo7982: 41). The "lower" levels (ngoko) are "the languageone losesone's temper in" (Errington 1984:9). Yet, in some ways the point is not really what happensto one's own temper, but to one's addressee's.The languagelevelsare addressee-focused; they are thought of as means of guarding addressee'sequanimity, of avoiding angering the addressee,of expressingpoliteness by deferring to his/her wishesand effacing one's own (Errington 1988: 42-3). Polite conduct toward a respectedaddresseeis conduct that is stylized, depersonalized,and flat-affect.Still, the useof "high,"deferentialstylesalso implies the speaker'sown refinement, as shown by hisiher ability to efface emotion, sensitivityto the stability of others, and pragmaticdelicacy. Compare with this the linguistic ideology of rural, "traditional" Wolof. There are some quite similar ideas about rank, affectivity,and engagementwith the concrete:high rank impliesself-control,flat affect, the protection of others, and (especiallyfor the religious elite) disengagementfrom worldly involvements.Moreover, deferentialconduct toward others requiresa flow of flattering words. But unlike the Javanesecase,in Wolof a flow of words does not easily display any high rank or refinement on the part of their speaker. For some insightinto this difference,considerthe nativemetapragmatictheories regarding participant roles in speaking. As example (5) has shown, the Wolof metapragmaticterminologyfirmly identifiesthe two registers,'noble speech'and 'griot speech,'with the rank of speakers- nobles and griots being oppositely-rankedcastes. It is becausepersonsof these high and low ranks are ideologicallyaccorded certain temperamentalcharacteristics, suchas affectivityand excitability,that the registerstake the form they do. Thus 'noble speech'is flat-affect speech,while 'griot speech' is a high-affect,theatrical,hyperbolicstyle (see Irvine 1990). Despitethe terminology,however,the useof theseregistersis not limited to the socialranks they are named for. Both registersare usedby almost everyone.Still, their use always conveys a sense of contrasting ranks, even if only metaphorically. Normatively,'griot speech'is the way low-rankinggriots addresshigh-rankingnobles. Any personmay employ this registerto flatter an addressee;yet, in so doing, a speaker engagesin griot-like, hence low-ranking,conduct. The Wolof linguisticideologythus identifiesthe registersystemprimarily with Speaker (and Speaker's supposed temperament), only elevating Addressee by implication.The Javaneseideology,in contrast,identifiesthe stylesystemprimarily with Addressee, elevating Speaker only by implication (see Table 1.) Although the "implicated"participantrole informsparticipants'strategies, the fact that the ideological focus lies elsewhereconstrainswhat rhetorical effectsare achievable,and how.

Ideologiesof honoifrc language 257

Table 1: Participant role relationships

SPEAKER

ADDRESSEE

WOLOF'griot speech'

[,owering

(Elevating)

JAVANESE basa (high levels)

Lowering (Elevating)

Elevating

KEY:

No parentheses:primary, focal effect Parentheses:secondary,implicational effect

For example,Wolof nobles avoid the 'griot speech' register, and any other suggestionof low-casteconduct, in public, formal occasions.Instead, they hire a lower-rankingintermediary - usually a griot - to speak on their behalf. The comparisonof Wolof and Javaneseis further illuminated, I think, by a look at a third case,Zulu, which sharessome characteristicsof both. Zulu has two types of deferentialexpression,the native terminology suggests:hlonipha 'showingrespect', and 6 onga'praise'.These two types of expressionare ideologically linked with different socialcontexts(family and court) and different users (women and men), respectively. Zulu married women use hlonipha words, supposedly,in order to avoid uttering the nameof the husband'sfather - or any other word containingthe radical found in his name,or soundingsimilar.aIf Father-in-law'sname just happensto sound rather like 'hippopotamus', imvu6u the woman must call hipposincudu instead.(We might think of this pattern as a sort of anti-pun.) Hlonipha behavior applies to gesture and clothing as well as to words: to hloniphais to avoid eye contact, restrain one's affectivity, and cover one's body in the presence of respectedpersons.This apparentlyincludesusingconventionaleuphemisms for talk about bodily functions; it certainly includes covering over, or avoiding, expressing the soundsof respectedpersons'names.The substitute(respectful) term may derive from a descriptive or metaphorical construction, or it may derive from patternedphonologicalshifts. The phonological shifts have the effect of neutralizing consonantcontrasts,since the tendencyis for all consonantsto become [+Coronal] (especiallythe coronal affricatesrJ andT tdll) or clicks.Many hlonipha homonyms are createdin this way, as illustratedin (6):

* SeeIftige (1950 [1936]:30-31),Doke 1961,Bryant 1949.Publishedsourceson Zulu hlonipha, thoughincludingextensivelists of forms (especiallyDoke and Vilakazi 1958),describethe patternsof decades ago.It is not clear to me whetherthe Zulu respectvocabularyis still in use. Contemporary usageof hloniphaamongrural Xhosawomenis reported,however,in severalrecent papersby Finlayson(1978,1982,1984).

258

Judith T. Imine

(6)

Zulu respect homonyms(Doke and Vilakazi 1958):

swing' annoy'

ORDINARY

HLONIPHA

lenga nenSa

cenSa cenSa

(c : dental click) (Uku-)6 onga 'to praise',on the other hand,is to 'expressgratitude' or 'worship', but in a very different way. In particular, the term identifies an exuberant, poetic style of male public oratory, usuallyaddressedto important political figures (clan leaders, chiefs,kings,visiting dignitaries) at public gatherings.Although anyonecan use this style to shout out praisesand encouragement,(say) to a man engagedin a fight, or to an important man who appears on the scene, the model performers are the male professional praise-poets (izim6 ongz,singular im6 ongt) supported by chiefs or kings. A praise-poetmay begin his training by memorizingtraditional praisesassociatedwith clansand historicalpersonages,and some of this memorizedmaterial may be included in his performance. Successfulperformance does not depend on exact recall of fixed text, however. On the contrary: vivid, detailed imagery, and a sense of spontaneous enthusiasm,are crucial. Interviewswith prominent poets (see Opland 1983) indicate that effectivepraise-performancemust be spontaneous, inspired,and visionary.Several poets told Opland (1983: 64 tt.) that the impulse to becomea praise-performercame them originally in a dream; that their experiencesare like those of a diviner who is summonedby ancestralspirits;and that the wordscome to them suddenly,unrehearsed and unprepared,the product of inspirationand emotional outpouring.s Zulu praise-utterance has its poeticisms, but it does not seem to have grammatical honorifics. Thus the Zulu expressive system includes an ecstatic, high-affect, engaged style without honorifics, and a flat-affect, disengaged,avoidance style with honorifics. Now, notice that the Zulu praise-oratorystyle resembles,in some ways, the 'griot Wolof speech'register:both focus on praisingthe addressee,through dramatic, heightened-affect, semanticallyelaborateddiscourse.The resemblanceis not accidental. Praise-oratory of this type is widespread in Africa; among the Wolof, it is the griots' specialty.But Wolof and Zulu differ in that the Zulu linguistic ideology does not connect praise-performancewith low caste,or with particularlylow rank of any sort. Speaking in 6 onga style does not compromise the speaker's status. Notice too a typological parallel between the Zulu hlonipha styleand the higher levels (termed basa) of Javanese. Both involve lowered affect, euphemism, neutralization of certain contrasts,and conspicuousconventionality.The Zulu system,then, incorporatesboth

5 Although Opland'sbook concernsXhosa praise-poetryprimarily,he considersZulu sourcesas well. Zulu and Xhosa praise-poetryseemvirtually alike as genres,as indeedthe languages are. Both systemsalso havehloniphavocabulariesas well, as I havealreadynoted.

Ideologies of honoific language 259 patternsof deferential talk describedearlier for Wolof and Javanese. Comparingall three languages,Table 2 summarizesthese relationshipsbetween aspectsof linguisticstructure and aspectsof linguisticideology,as these concern verbal conductconsideredappropriate for elevating others. Table 2: Ideology and Structure of Other-Elevating Expressions PATTERN A:

PATTERN B:

Ideology:

l-owered-affect; Conventionality

Heightened-affect; Spontaneity

Structure:

Semanticor sound neutralization; Grammaticalhonorifics hlonipha 'respect'

Semanticor sound elaboration; No grammatical honorifics

ZULUIXHOSA: WOLOF: JAVANESE:

6 onga'praise' waxugewel'griot speech'

basa ('high'levels)

4. Distribution of deferential styles As Silverstein(1979)observes,though ideologyaffectsspeakers'strategiesof language use, linguistic ideology is not the same thing as linguistic structure, and native perceptionsof use are not the same as the (comparative)distribution of uses,as an comparison,such as outsideobservermight perceivethem. Attempts at cross-linguistic the presentcomparisonof deferentialstyles,are vitiated if ideologyand distributionare Were one to suppose,for Javanese,Wolof, and Zulu, that the not clearlydistinguished. ideologicalrationale for the deferential styleswere also their description,the three 'noble speech'and 'griot systems might appearnot to be comparableat all. For Wolof, speech'might appearto be socialdialects,not registers(seeIwine 1990for discussion); and the ZluJuhlonipha vocabularymight appear to be merely a series of idiosyncratic, varyingfrom one woman to another,rather than a systematic ephemeralconstructions linguisticresource. That is to say, the father-in-law name-avoidancerationale for Zulu hlonipha would suggestthat each set of daughters-in-lawwould have a different respect vocabulary, dependingupon the father-in-law'sname;that a particularspeaker'srespect terms might not be very numerous; and that a term would disappear upon the death.Buthloniplla usageis (or was) actuallymuch more widespread, daughter-in-law's involvingmale speakersas well as female, court contexts as well as domestic, and variouskindsof respectedbeings.From Krige (7936:31) we learn, for example,that hloniphaformsare usedby men to avoid usingthe name of the mother-in-law,though the custom"is not so strict"for men as it is for women.Furthermore,"the whole tribe"

260

JudithT. Irvine

must hlonipha the name of the king or chief, while those resident at the royal court must hlonipha the names of the king's father and grandfather as well (Krige 7936: 37, 233; seealso Bryant 1949). In an example conspicuouslynot involving a father-in:law, Bryant (1949: 220) notes,too, that "for a Zulu woman to call a porcupine by its proper name, iNgungumbane,were but to provoke it to increaseddepredation in her fields; 'politely'as 'the-little-woman',or umFazazana...". The thereforeit must be referred to Doke and Vilakazi (1958)dictionary,which citeshundredsof hlonipha words, includes respectforms for kin terms, chiefly and royal titles, and possessivepronouns. All these forms, apparently longstandingitems of Zulu vocabulary,are unlikely to be merely avoidances of a father-in-law's name. The existence of a widely-known respect vocabularyseemsto be a f.actof the distributionof uses,not inherent in its rationale. The distribution of Zulu speech styles thus does not conform strictly to the linguisticideologythat links them to gender relationsand domesticvs. public arenas. Nevertheless,the linguisticideologydoes suggestthat the primary focus and principal set of connotationsfor the hlonipha vocabularylie in domesticrelations, and enter the court only by extension.It is likely, in fact, that Zulu hlonipha usageantedatesthe rise of a strongly centralized Zulu state in the late eighteenth century. The existence of hlonipha among the Swazi and the less traditionally-centralized Xhosa implies its relative antiquity and tends to confirm the idea that the respectvocabularyarose,not in connection with the state, but in the power dynamics of Nguni family and affinal relations. Comparing the linguistic ideologywith the distribution of uses - a distinction essentialfor cross-linguisticstudies- thus affords a glimpseof the historical dynamics of sociolinguisticsystems.As Silverstein (1979) argues, the relationships among languageideology,structure,and use form a dialecticalprocesswhich, in conjunction with local contingencies,induceschange.For Zulu honorifics,it is worth noting in this regard that the kinds of sound changesinvolved in hlonipha words resemblesome of the sound shifts that differentiate the Nguni languagefamily from its Southeastern Bantu relatives, such as the shifts 6 > tS, mb > nj, and the acquisition of click consonants.Were it possibleto delve more deeply into Zulu or Xhosa ideologiesof languageone might be able to illuminate this processby investigatingnotions of sound symbolism,and attitudes toward the Khoisan languagesfrom which the clicks were acquired. Conceivably,then, the constructionof Nguni honorific avoidanceforms included the importation of "foreign"words and sounds,some of which may later have lost their specificallyhonorific value, being replacedby new avoidanceforms. If so, this would not be the only case where honorifics have behaved like a currency in inflationary conditions.The other casesof grammaticalhonorificswe haveexaminedhere,Javanese and ChiBemba, both evidence processesof pragmatic devaluation.(Wolof, lacking grammatical honorifics, is not directly comparable.) Thus Errington (1988: 115) commentson Javaneseterms that "haveundergonepragmaticdevaluationas the result of recurring patterns of strategicother-exalting,self-abasingspeech style use." For ChiBemba, Richardson (1967) reports a wide extensionof honorific usage in urban settings, so that some honorific address forms have tended to lose any specially

Ideologies of honoificlanguage 267 honorificimplication. Differences in honorific usagebenveenmy Bemba consultantsof rural and urban origin confirm this trend and suggestthat urban speakersmust resort to additionalmeansif they wish to underscorerespectfulness. To explore further the contemporarydynamicsof any of these honorific systems wouldrequirerecognizing,however,that languageideologiesare alsosubjectto change. Increasingly participatingin a globalpoliticaleconomyof language,thesesociolinguistic systemsand their ideologiesare being reconfigured.

4. Conclusions The comparison of these four languages suggeststhat grammatical honorifics accompanylinguistic ideologies that specify flattened-affect, conventionality, and avoidance of engagementwith the concreteor the sensory,as a way to expressrespect for others(rather than as a way to expressone's own rank). Put another way, honorifics are embeddedin an ideology in which a low-affect style can be other-elevating.They are connectedwith the managementof affectivity and conventionality, and with the waysthese relate to rank and power. What kinds of rank and power are concerned varies from one system to another. Grammatical honorifics have no necessary connectionwith royal courts, or with class stratification. Even where courts exist, domesticpower relations may be the honorifics' primary arena. It would not have been possible,I believe, to reach this conclusionwithout clearlydistinguishingthe distribution of forms of talk from the linguistic ideolory that interpretsand rationalizesthem. A focus on ideologyhas been crucial in accountingfor similarities and differencesamongthe four casesconsideredin this paper; but ideology it interprets,would not alone,withoutconsiderationof the behaviorsand circumstances be sufficient.To considerall of these- linguisticstructures,ideologies,and distributions - facilitatesthe comparisonof casesand helps illuminate the dynamicsof historical change. In short,linguisticideologymediatesbetweenforms of speakingand conditions of sociallife in a complex way.

References Bryant, AlfredT. (1949)TheZulupeople. Pietermaritzburg: ShuterandShooter. (l%1) Textbook Doke,Clement of Zulugrammar (6thed.).Johannesburg: lnngmans. andB. W. Vilakazi(1958)Zulu-English (2nded.).London:Oxford. Doke,Clement, dictionary Errington, J.Joseph (1984)Language andsocialchange in Java.Athens,OH: OhioUniversityCenter Studies. forInternational

262

ludith T. Irvine

Press. Philadelphia:Universityof Pennsylvania Errington,J. Joseph(1938)Stntctureand stylein Javanese. 24 48-63. Finlayson,Rosemary(1978)"A preliminarysurveyof hloniphaamongthe Xhosa'"Taalfusette Finlayson,Rosemary(1932)"Hlonipha- the women'slanguageof avoidanceamongthe Xhosa.' South Afican Joumal of Afican Languagesl(l)' Finlayson, Rosemary(1934)

'The changingnature of.isihloniphosabafazL'African Sndies 43: 137-46.

Irvine, Judith T. (1985)"Statusand style in language.'AnnualReviewof Anthropologt14:557-81. Ameican Ethnologist Irvine, Judith T.(1939)"When talk isn't cheap:languageand political economy.n 16:248-67. 'Registeringaffect:Heteroglossia in the linguisticexpressionof emotion'" In C. Irvine,Judith T. (1990) Lutz and L. Abu-Lughod(eds.),Languageand the potiticsof emotign.Cambridge:CambridgeUniversity Press,L26-61. Shuterand Shooter. of the Zulus.Pietermaritzburg: Krige, Eileen (1950) [1936] Thesocial system Opland, Jeff (1983)Xhosa oral poetry.Cambridge:CambridgeUniversityPress. Richardson,Irvine (1967\'Linguistic evolutionand Bantu noun classsystems."In G. Manessyand M. Paris: CNRS, 373'88. Houis (eds.),La classificationnominaledans les languesnCgro-africaines. 'l,anguagestructureand linguisticideologt.' In P. Clyne,W. F. Hanks,and Silverstein,Michael (lg7g\ C. L. Hofbauer (eds.), The elements:A parasessionon lingubtic units and levels. Chicago: Chicago Linguistic Society, t93-241. Wenger,JamesR. (1982) Someuniversalsof honoific langtagewith specialreferenceto Japanese.Ph.D. dissertation,Universityof Arizona. Wolfl J. U. and S. Poedjosoedarmo(1982) Communicativecodesin centralJava. Cotnell University Departmentof Asian Studies,SoutheastAsia Program,Data paper 116.

Suggest Documents