Studies of North American Indian Languages

Annu. Rev.Anthropot. 1gg0.)9:309-30 Copytisht @ 1990 b Annual Revie'')slnc. Au rishts rcsened STUDIES OF NORTH AMERICAN INDIAN LANGUAGES Marianne Mithun Department of Linguistics,Universityof Califomia,SantaBa$ara,Califomia93106 sociolinguistics contact,geneticrelalionships, language KEY WORDS: lhguistics,discourse, The study of North American Indian languageshas been shapedby several circumstances.Since its beginning, most researchhas been basedin fieldwork: Datahavecomefrom direct contactwith speakers,usuallyin thefuown cultural settings,ratherthan from secondarysources.Although the numberof to NorthAmericais large,severalhundred,thenumber languages indigenous of scholarsworking with most of them hasbeenrelatively small, often only oneor two individualsper language.A centerof scholarlyinteractionhasthus been the community of those studying languagesall over the Americas, languagesthat are quite diversegeneticallyand typologically. Thesefactors, the groundingin fieldwork andthe comPositionof the scholarlycommunity' have affectedthe kinds of work undertaken,the theoreticalissuesaddressed, the natureof the explanationssought,andthe applicationsmade.The field is in many ways highly anthropological:strongly contextualizedlinguistically, culturally,and typologically. The researcherwho works directly with speakersof a little-documented languageneeds basic proficiency in all areasof linguistic structure. It is seldom possible to achieve insights about syntax, fol example, without sufficient phonetic skill to hear accurately, sufficient understandingof phonology to construct a usabletranscription, and sufficient knowledge of mor?hologyto be awareof the grammaticalfunctionssignaledwithin wordsThe importanceof broad competencehasshapedthe kinds of descriptionand explanationthat havecomeout of the field. Structuresare usually considered in the context of other structuresin the languagerather than in isolation. 1015-0309$02.00 0084-6570/90/ 310 MITHUN Few linguists working with North American Indian languagescan rgnore their rich cultural settings or, in many cases, the rapid changes they ue undergoing. Good linguistic work in such settingshas seldombein limited to the elicitation of grammatical paradigms or sentences.The rich but often fragile cultural settings of North American Indian languages have prompted a tradition of collecting texts of all kinds: religious or political oratory; legends; historical accounts; reminiscences;children's stories; descriptions of ceremoniessuch as naming, marriage,burial. selecrionand installaiionof leaders, etc; various aspectsof daily life, such as hunting, fishing, cooking, medicine, basketmaking, games, songs, etc; and now, with the availability of tape recorders, conversation as well. Such documentationhas been valuable in itself, in many cases providing the only descriptions in the speakers,own words of earlier events and customs that are now fading from comrnon memory. It has also had an important effect on the study of languageand its use. The textual material provides linguistic and cultural context for the studv of grammar. If the investigation of grammatical structures were limited to ;ata consisting of isolated sentencestranslatedfrom a contact languagelike En_ glish, much would be missed. Translationscan easily distort G giammatical patternsof a languagein ways that obscuretheir actual functions: Too often, aspectsof the languageunder investigationare understoodas perfect counter_ parts to their translationswhen in fact they are parls ofvery different systems. Much can also be lost when sentencesare examinedin isolation: The use of many constructrons is govemed by factors beyond the limits of a sinsle sentence.Relianceon elicitation alone has a further drawback: If all anatyiis were based on elicited data, the analyst might never notice structuresthat he or she had not anticipated in advance. When grammatical shuctures are studied in the context of naturally occuning, connectedspeech,their precrse functions, and their differencesfrom similar structuresin other languages, can b€ detected in ways that are often not possible when data are-tiriiteO to isolated elicited translationsof sentences. Linguistic typology and American Indian languageshave long enjoyed a special relationship. Since von Humboldt first brought glimpses of New World languagesto Europeanscholars,the exotic charjcter of their structures has arousedinterest.The work of Boas, Sapir, Bloornfield, and othersassured the role of North American languagesin the developmentof linguistic theory in this century, a role that continues, partly becausethese languagescontain structures not often found in other parts of the world. At the same time, exposureto other languageshas played a significant role in work with North American languages.The more experienceone has with the structuresof a variety of languages,the more effectively one can perceive patterns in an undescribedlanguage. The major journal for those working with American AMERICAN INDIAN LINGUISTICS 311 Indian languages,the Internationql Joumal of AmericanLingaistics, arldthe primary scholarly meetings,the SSILA (Society for the Study of the Languagesof the Americas) sessionsat the annual meeting of the American AnthropologicalAssociationand elsewhere,consistof work on all of the languagesindigenousto the Americas.Any North Americanistwho reads IJAL 2[rd aftends one or two SSILA meetings per year, as most do, is constantlyexposedto languagesgeneticallyand typologically quite diverse. This exposurehas beena constantsourceof fresh ideas. Current work with North American Indian languagesaddressesa broad range of topics. Since few if any of the languages have been fully documented,and the opportunitiesfor recordingmany of them will soon dictionaries, and disappear, descriptioncontinuesto be a priority.Grammars, collections of texts are being.producedwith increasingsophistication.More areasof linguistic structurearebeing investigatedandunderstoodin termsof more factors, including discoursepattems, social and geographicvariation, anddiachmny.In the followingsections,a few of themaincunentsof recent work on the languagesnorth of Mexico will be sketched. STRUCTURALCONTEXTUALIZATION Much current work on grammar,both morphologyand syntax, demonstrates the importanceof linguistic contextbeyondthe boundsof the sentence.One areaof interestto linguists working within most theoreticalmodelsis reference.North AmericanIndian languageshavemuchto contributeto theoriesof reference becausetheir repertoires of referential devices are not always isomorphicwith thoseof more familiar languages. Refer ential Alt ernativ es Much generaltheoreticalwork on referencehas concentratedon coreference pronouns,etc)within sentences relationsamongnominals(full nounphrases, in a numberof North languages. Recentwork on reference in Indo-European AmericanIndian languages hasshownthat referentialsystemscan be much more varied and complex than might be supposedon the basis of IndoEuropeanalone. Furthermore,most can be fully understoodonly when their use is examinedin largerstretchesof naturalspeech. Watkins (92) showsthat the relative frequenciesof full noun phrasesand zero anaphorain Kiowa differ radically acrossdiffelent kinds of texts. In storiescenteredarounda singleprotagonist,this characteris typically named only onceearly in the narrative,then subsequendynot mentionedat all (zero anaphora).Suchtexts may containasmany as20 clauseswithout a singlefull noun phrase. In stories with severalcharactersof comparableimportance, 3t2 MITHUN who interact significantly, eachmay be identified by a full noun phrasein nearlyevery clause,emphasizing the opposition. Describing Central Pomo narative and conversation,Mithun (69) shows that altemations among several referential devices are determinedby the status of the referent within the consciousnessof the speaker.Full noun pronounsderivedirom phrasesintroducenew referentsinto consciousness. demonstrativesbring referentsback into consciousness after any discontinuity in discussion. Referentsalreadywithin consciousness arenotovertlyreidenti_ fied at all (zero anaphora).A specialset of empatheticpronounsprovides anotherkind of distinction,encodingreferentswhosepoint of view is repre_ sented. A related language,Northem Pomo, contains regular third person pronouns,in additionto empathetic pronouns,demonstratives, andthe optionof zero anaphora.Masculineand feminine gender.and singularand plural numberare distinguished by the pronouns.O'Connor(73), examiningcon_ versation,demonstrates how the altemationbetweenthesepersonalpronouns and the demonstrativesreflects not only a human/nonhuman distinction, but alsothe socialcontextandthe speaker'sattitudetowardthird penonsunder discussion,and interactswith the indicationof evidencefor thi information presented. Kinkade(49) discusses the relationbetweendiscoursetopicalityandagency in a numberof Salishanlanguages. In all of thelanguages, transitiveverbs may be passivizedif theirpatientsaremoretopicworthywithin the discourse thantheiragents.At least6 ofthe 23 Salishanlanguages containan additional device;a specialtopicalobjectmarkersetsoff gammaticalobjectsof special importancewithin a stretchof discoursecontainingmultiplethird persons. Kinkadenotesthat the neighboringbut unrelatedSahaptinalso has sucha marker, which may be the sourceof the Salishansuffixes.Rude (79) describesa similarphenomenon in Klamath. Goddard(29) detailsthe functionof the proximate-obviative distinctionin Fox narrative:"The proximateis the unmarkedthird personcategory;ifthere is only onethird personreferentin a context,it can only be proximate. . . . In contextsthat havea third personanimate(noun or pronoun)and, in addition, anotherthird person,the higher-statusor more central third personis proximate and the lower-statusor lesscentralthird personis obviative." Goddard exploresthe points at which the proximatecategoryshifts to new referentsin discourse.While proximateor obviative statusis determinedin certain contextsby syntacticfactors,the altemationmostoftenfunctionsas a discoune device,"a highly significantaspectof the structuringandresuttingnarative textureof discourse." AII of these referential distinctions-full noun phrases/zero,pronouns/ zero, demonstratives/pronouns, empathetic/personal pronouns,topical ob- AMERICANINDIAN LINGIjISTICS 313 jects/zero, and proximate/obviative---+learlydependon factors beyond the scopeof the sentence.Most involve topicality or point of view in somesense, but they differ in important ways that are only beginningto be understood. The Forms of ReferentialDevicesand Their Structural Ramifications Of courseall languagescontainfull nounphrases,althoughthe samekinds of distinctions are not always encoded,such as definitenessor number. All languagesalso containpronouns,but their forms and usesvary in interesting ways from languageto language,ways that can have wider ramifications in the syntaxand discourse. Pronounsin some languagesare much like English pronouns: separate words that altematewith full noun phrasesto refer to identifiable persons, objects, etc. In many North American Indian languages,however, regular pronounsare verbal affixes. They appearwith every verb whether an additional nounphraseis presentin the sentenceor not. The presenceof a full set of pronorninalaffixes, in threepersons,is often conelatedwith certainother grammaticalcharacteristics(43, 64). The pronominalaffixes function as the primary argumentsof clauses,so any coreferentnoun phlasesin the clause typicallyhavea somewhatloosesyntacticconnectionwith theverb. Accordingly, the relativeorderof nounphrasesandverbsin manyof theselanguages may reflect not syntactic relations but rather the pragmatic status of the within thediscourse(65,76, 86). Informationthat informationtheyrepresent is new and/or important often occurseady in the clause,with more predictable and incidental information appearinglater. Noun phrasesthemselves may not have intemal constituencylike that in Europeanlanguages(66). Subordinationmay not be as strongly grammancized(44,62)' Languagesof this type have sometimesbeen referredto as "nonconfigurational"or "pronominalargumenf' languages. Discoursepragmaticfactorsin the word ordersof otherkinds of languages havealsoreceivedattention.Eastman(16), Eastman& Edwards(17), Edwards(18, 19), andEnrico(20)discusstherolesof focusandtopicin Haida, a basicallyverb-finallanguage.Chafrc(13)pointsout therole of idiomaticity of Subject-Verband Verb-Object complexesin word order in Caddo. TheRolesof Referents:Voice Another areain which a discourseperspectivehasprovenimportantis that of voice, As long as voice altemationsare comparedonly in pairs of sentences out of context, it may be possible to isolate their forms but difficult to determinetheir full functions. Several studies have demonstratedthe importanceof examining voice phenomenain discourse. Rude (80) investigatesthe discourse-functionalcontexts that trigger al- 314 MITIIUN temationsamongantipassive,ergative(transitive), andpassiveconstructrons in Nez Perce.He concludesthat the altemationis bestunderstoodin termsof the comparativetopicworthinessof agentsand patients. ,.In the antipassive, the agentfar ourweighsthe patientin topicality,in the passivethe agentis completelysuppressed,andin the ergativeconstructionthe patientis clearly a secondary;in terms of cataphoriccontinuity it equalsthe agent." Whisder (93) examines a construction sometimeslabeled ,bassive" in Nootkanlanguagesandconcludesthat the phenomenonis betterdescribedas inversemarking, whosefunction is best understoodin terms of the thematic organization of texts at the "paragraphlevel." ,.Constituentclausesof a paragraphare marked direct or inverse dependingon the ACTOR versus GOAL statusof a thematic participant." Thompson(89) showsthat the inverseconstructionin Koyukon Athabas_ kan (the famous yrTbl altemation) differs significantly from passive and impersonalconstructionsin that the agentis seldomsuppressed,and when it is, this is becauseit is anaphoric.The inverseshouldbe viewed functionally as "a marker of a deviation from the normal topicality relations.The researchdescribed here constitutes only a small portion of work currently being caried out on the structures of North American Indian languages.It is, however,representativeof the cunent interestin considerins grammaticalstructuresin their larger contextof naturalconnectedspeechof i variety of styles. SOCIOCULTURAL CONTEXTUALIZATION The fact that muchwork on North AmericanIndian languageshasbeenbased on texts (narative or conversation)recorded in their cultural settings has meantthatmoststudiesof theformsof thelanguages arenot isolatedfromthe communicativeand social functions they serve. Much work has taken into accountthe variety of registersavailable to speakers,as well as the social contextsin which they are spoken:geographical,social, and temporal. Registers Linguistic structuresvary accordingto the cultural functions they are called upon to serve. Somerecentwork on suchstylistic altemations,or registers, has examinedthe specialspeechforms usedwith children, in ritual. and in writing, among others. In many cultures,adultsspeakdifferently to youngchildrenandsometimes to petsthan to otheradults.The speechformsthemselves,aswell asspeakers' attitudes toward them, are interesting both cutturally and linguistically. In somecultures,they are treasuredasa way of showingaffectionor considered an aid for children in leaming their first language;in others, they are dis- AMERICAN INDIAN LINGUISTICS 315 dained as an impedimentto children's proper acquisition of the language. Speakersgenerallyconsidersuchforms to be simpler, but ideasof sinrplicity are not necessarilyuniform from one group to the next. Many North American Indian languagesprovide an especially good basis for study of such notions of simplicity owing to their relatively complexphonetic, phonologi cal, or morphological structures. Earlier in this century, Sapir describedsuch specialforms in Nootka, a Wakashanlanguageof British Columbia(81, 82). More recendy,Kess & Kess (45) have reexaminedthis speechregisterin anotherNootkan dialect, Ahousaht. Ahousahthas a large inventory of consonants,including glottal(p', t', c', d', etc. andm', n', y', w', izedseriesof obstruents andresonants etc.), distinctivevelar(k, k', x), labiovelar(k*, k'*, x*), uvular(q, q', 4), ( f , h) andlarynCeal (e, h) series.As labio-uvular(q*, q'*, 1), pharyngeal might be expected,someof the articulatorily rnore complex soundsdo not q', q'*, {, 1-, soundsthat are in fact appearin baby talk (motherese): somewhatinfrequent in adult speechas well. Consonantclusters do not appear in baby talk, although they do in normal adult sFech. There is morphological simplification in the lack of imperative suffixes, but as in much baby talk, reduplicationand diminutive suffxes are comrnon.Thompson (90) has examinedthe ways speakersof Twana, a Salish languageof WestemWashington,addressyoung children. Characteristicsof this register are comparedto t}Ie Twana women's speechregister. A quiie different kind of register,highly developedin many North American Indianlanguages, is that of ritual language.Du Bois (15) hassurveyed the linguistic forms of suchspeechthat serveto conveyauthority, to makethe ritual utteranceappearself-evident. He points out that the ritual register is often characterizedby a marked voice quality and by stylized intonation pattems, often a much more restricted set than those found in colloquial language.They are sometimesdescribedas "singsong" or "chanted." Ritual speechis typically morefluent than colloquial speech,without the falsestarts, afterthoughts,or other disfluenciescharacteristicof everyday spoken language. The vocabularyoften containsmany archaic, borowed, or tabooed ierms, as well as euphemisticor metaphoricalcircun ocutory expressions. Certainstructuralpattemsare common.Many ritual languagesare characterized by coupletshucture-pairs of lines that are semanticallyand/orgrammatically parallel. Evidential markers, lexical and grammaticalmarkersof the sourceandreliability of the informationconveyed,often play a strongrole in ritual speech. Still anotherspecialregistermay be found in written language.Most of the languagesindigenousto North America do not have lengthy literary haditions, althoughsometraditions are well known, including thoseof the Massachusett,the Micmac, and the Cherokee. Massachusettliteracy, which 316 MITHUN spannedmore than a century during colonial times, is beautifully chronicled in Goddard & Bragdon's (30) two-volume work, which containsall known documentswritten by the Massachusett,with word-by-wordtranslationand extensivenotes.Micmac writing is tracedfrom aboriginalideographicorigins to a curent roman-basedsystemby Battiste(2). An entirevolume of articles on native wdting systemsin Canadahas been assembledby Burnaby (7). As moreNorth AmericanIndian languagesare written by their speakers,it is possibleto observethe effect of the written mediumon-theresister itself. Severalpapershaveinvestigatedthe specialcharacteristicsof the-newwntten medium. Mithun (63) discussedthe developmentof a written Mohawk style. When Mohawk speakerslrst beganto producewritten Mohawk texts, their work clearly showedthe influenceof English literary style, a style familiar to all of these writers. SpokenMohawk consistslargely of verbs, but early written texts containeda nour/verb ratio similar to that of Enslish. Word orderin spokenMohawkis purelypragmatically determined, butJarlywntten texts exhibit generalSVO order. An unusuallyhigh proponionoi particlesappeared,on the modelof Englishdefinite articlesandconiunctlons. "".trn Therewas little noun incorporation.As writersbecamemore experienced. however,theywereableto takeadvantage of the luxuryof time to'producea uniquely rich Mohawk style, characterizedby effective use of elaborate morphological constructions,including stylistic word order and noun m_ corporatron. Watkins (91) comparcdfour Kiowa styles: (d) informal spoken(soonta_ (r) plannedspoken(formal stories),(c) informa written neousanecdotes), (letters), and (d) formal written (written stories). Four shuctureswere ex_ amined: right dislocation, incorporation, relativization, and subordination. Shediscoveredthat right dislocationappearsonly in spokentexts, represent_ ing given material. The other thrce strucnles appearin both spoien and writtenstyles,but their frequencyof occurrence increases significantlyfronr type 4 to type d, with the time availablefor planning. Rhetorical Stucture An areathat has arousedspecialinterest in recentyearshas beenrhetorical structure. It has becomeclear that the presentationof oral texts in simnle paragraphform can obscuremuch of their inherentintemal structure.Two kinds of patteminghavebeenexploredin particular, that signaledby linguis_ tic form andthat markedby intonation.Hymes(3?), working with Chinookan andotherNorth AmericanIndian texts, manytranscribedbeforethe adventof taperecorders,hasarguedfor thepresentationof textualmaterialaccordrnsto lexical markers,syntacticstructure,and content.Chafe (10) and othersh-ave looked at the textual structureexpressedintonationally. The division of texts into intonation units separatedby pausesof various kinds, and the kinds of AMERICANINDIAN LINGUISTICS 3I7 prosodiccontours(involving pitch and rhythm) associatedwith theseintonation units, reveal much aboutthe systematicways information is presented. In somecases,the textualorganizationsignaledby lexical andgrammatical cuescoincideswith that expressedby intonationand prosody.In others, the two approachesrevealdifferent kinds of organization.Thesekinds of rhetorical structure and their relationships in North American texts have been discussedin a number of works. Woodbury (96) investigatesrhetorical structure in Central Alaskan Yupik discourse,"prosodically and intonally signaledphonologicalphrasingalong with whateverother significant formal featuresconsistentlypattem or interact with it." In addition to theil basic structural functions, he describesfour additional communicativefunctions associatedwith theserhetoricaldevices:organizationof information, expression of affective meaning, indexing of genre, and regulation of dialogic interaction. Other work on rhetorical structure in North American native discourseincludes,amongothers:Bright on Karok (4, 6), Chafeon Seneca (11), Goddardon Unami (25), Golla on Hupa (31), Kinkade on Upper Chehalis(47), Kroskrity(52)on Tewa,Mclendonon EasternPomo(58, 59), Roodon Wichita(78), Sherzeron Kuna (83, 84), Tedlockon Zuni (88), and especiallyWoodbury on CentralAlaskan Yupik Eskimo (95, 97). This work hashad far-reachingconsequences not only for the visual presentationof oral discoursebut also for our understandingof linguistic structure. Language Contact Investigationof the linguistic effects of contactamongspeakersof different languageshas always been a part of the study of North American Indian languages.North America contalns some notorious linguistic areas, geographic regions in which unrelatedlanguagesshare sets of traits, such as Northem Califomia. The distribution of such traits continues to be rnvestigated,becauseit will have much to tell us about the kinds of diffusion that are possible, and in many casesconstitutesan important piece of deep geneticpuzzles.Among sucharealstudiesare a discussionof the distribunon of sibilants in aboriginal Califomia by Bright (3) and of dental and alveolar apicalsin Californialanguages by Langdon& Silver (56). Becauseof the intensecontactsituationsall over North America, the effects of encroachingEuropeanlanguageson the indigenouslanguagescontinueto be documented.Effects of Frenchcan be seenin languagesof the Northeast, suchas Mohawk, andthe influenceof Russiancanbe seennot only in Alaska but as far southas Kashayain Califomia(74).'lle impactof Englishis of coursepervasive.The influence of Spanishcontinuesto be documentedfor languagesall over the Southwestand Califomia. [See, amongothers, Kroskrity & Reinhardt(54) and Gamble(22).1 Loanwordsindicate the natureof 318 MITHUN contact, such as namesof introducedcultural items including tools, domestic animals, foods, clothing, etc. The shapes of these words in the modem languages also indicate that the bonowing was not necessarily dircctly from Europeansthemselves,but often through other indigenous languages(5). When contact becomes particularly intense, and most speakersof the indigenous languagebecomebilingual, the original languagemay be used in fewer and fewer contexts until its sphere of use all but disappears.Some effects of such intensive contact have been describedespecially insightfully by Hill & Hill (36). They point out that bilingual situations can reflect tremendously creative management of the double resource of two linguistic systems, in this case a Uto-Aztecan language and Spanish. If speakersno longer use a languageon a regular basis, how accurately does their speechrepresentthe full original system?Cool (14) notes thar rwo of the most conspicuouscharacteristicsof dying languagesare (a) structural and stylistic simplifications and (r) dramatic increases in variability. He describesvariations resulting from simplifications, including syllable reduction, phonemic mergers, and reduction of allomor?hy, in two Athabaskan languages,Chipewyan and Sarcee.He proposesviewing thesechangesas the result of impeded languageacquisition, since none of the semispeakersever reachesthe highestlevel ofproficiency in the traditional grammaticalsystem. Moore (71) suggeststhat contemporary Wasco speakersview language primarily as a collection of words, whoseuse is primarily display. Investigating Gros Ventre (Atsina), a dialect of Arapaho (Algonquian), Taylor (87) discoveredthat as the languagewas replacedby English, analogicalleveling regularized singular and plural forms of some nouns and verbs. Comparing two dialectsofCayuga, one still spokenby a substantialcommunity, the other seldom used at all, Mithun (67) noted that in the little-useddialect, a complex phonological phenomenon involving the spreading of laryngeal features over odd-numberedsyllables had been simplified. As with any language, much vocabulary had been forgotten. The productivity of derivational morphology had also diminished, including the capacity to form new verbs with incorporaled noum- Overall morphological complexity was reduced, so that individual afixes might appear with particular words, but complex combinations of them were rare. Comparing the speechof speakersofCenrral Pomo with varying degreesof fluency, Mithun (68) discussesthe effect of languageobsolescenceon linguistic description. While those who have not used their first language for a number of years may forget considerablevocabulary and exhibit a reduced range of stylistic altematives,they can still be of importanceto investigators. They can set the stage for more fluent speakersto exhibit their own virtuosity and provide lexicalized expressions,which themselvesmay reflect complex derivational structures. AMERICAN INDIAN LINGTJISTICS 3I9 DIACHRONY Partly becauseof the large numberof languagesindigenousto North America, many only recentlydocumented,diachronicwork continuesto be of great interestto North AmericanIndian linguists. There is perhapsless separation here than in some languageareasbetweenthose doing diachronic and syncbronic work. Most scholarscurrently engagedin reconstruction,subgrouping, and the searchfor deepergenetic relations, have done extensivefield work themselveswith many or all of the languagesof at least one family, usually more. Work in eachdomaininfluencesthe other. Particulardiacbronic problemsarelesslikely to be consideredin a structuralvacuum;at the same time, a diachronic perspectivepervadesthe descriptionand explanationof synchronicpattems. Over a decadeago, scholarsworking with most of the major language families in North America met to sharenoteson the current stateof knowledge in their fields (9). Since that time, diachronic work of all kinds has continuedat an energeticrate. As more and better data havebecomeavailable, our understandingof the interrelationshipsamonglanguagesandof their earlier stateshas continuedto grow. A numberof cooperativeprojectsare in place, for families such as Siouan, Uto-Aztecan, and Yuman, in which specialists in related languagesare pooling theii data in large computer databases. Processes of Language Change and Reconstruction Study of languagechangeon all levels continues.Much of this work, like the synchronicwork describedearlier, is highly contextualized:The development of structuresis tracedin the contextof their relationshipswith other structures in the languagesand their social and communicativefunctions. One example of such work is Goddard's discussion of the Eastem Algonquian subordinative mode (24). Mithun (70) tracesthe sirnultaneousgrammaticization of conjunctionin sevenNorthem Iroquoianlanguagesunderthe influenceof English and French.Kroskrity (51) examinesthe discoursepragmaticfactors inJluencingsyntaxin his discussionof negationandsubordinationin Arizona Tewa. Kroeber(50) considersdiscourseandfunctionalfactorsin the development of ergativecasemarkingfrom passiveconstructionsin SouthemInterior Salish languages.Goddard (27) shows that apparentrapid changein Fox during this century can be tracedto the interactionof two styles: deliberate and casual. possible,sodoeslinguisticpaleonbecomes As morelexicalreconstruction tology, the reconstruction of wordsfor culturalelements.Fowler (21) suggests an original location of the Uto-Aztecan homeland shetching across modem Arizona, down into Mexico, and perhapsinto SouthernCalifomia. MITHUN This was followed by a north-southsplit. There occurreda gradual deeper penetrationof someof the southernlanguagesinto centralMexico, northward and eastwardexpansionof the Numic languages,and movementsinto adjacent desertsby Takic and Pimic groups. Mithun (61) reconstructsProtoIroquoian terms for flora, fauna, hunting, agricultural, and aquatic complexes,otheraspects of materialculture,andsomesocialtraditions.Kinkade (48) investigates theorigin of the Salishhomeland.The Salishhaveoccupied two markedly different ecological zones,the temperateNorth Pacific coast from northem Oregon to central British Columbia, and the arid interior plateau.Boashad suggestedthat thesepeopleoriginally inhabitedthe interior and then migrated to the coast; more recently, Suttles suggesteda coastal origin. Kinkade'sinvestigationof terms for flora and fauna supportsthe coastalorigin, exceptfor Bella Coola,whichcontainsevidenceof an interior origin. He pointsout that this conclusioncorresponds with that reachedby archaeologists,an agreementhe realized only after completing the work. Subgrouping progress As reconstruction hasprogressed, hasbeenmadein subgrouping as well, determininghow languages within familiesare relatedto eachother. Recentwork on subgrouping includesamongothersthaton Yokuts(23, 94), on Uto-Aztecan(60), on Muskogean(46, 72), on Siouan(75), and on Algonquian(77). Deeper Genetic Relations Demonstrablegenetlc relationships among families on the level of IndoEuropeanhaveessentiallybeenestablishedin North America, althoughwork continuesto refine reconstructionand subgrouping.The samescholarsinvolved in synchronicandfamilyJevel diacbronicwork continueto investigate deepergeneticrelationshipsamongfamilies. Severalof the groupingsoriginally hypothesizedby Kroeberand Sapiron the basisof superficialstructural similarities continueto be exploredas more data becomeavailablefor com(39) assess parison.Langdon(55) andJacobsen variousaspectsof the Hokan hypothesis.More recently, ambitiousunpublishedwork by T. Kaufman provides mounting lexical evidence of relationships among some of the languagefamiliesgroupedas Hokan,but suggests that others,suchas Chu(85)assesses mash,do not belongwith thisgroup.Silverstein thestatusof the Penutianhypothesis.Investigationcontinuesin this areaas well in largely unpublished workby S. Delancey,V. Golla,andK. Whistler,amongothers. In 1987,J. Greenbergpublisheda book, Languagein the Americas,n which he proposedthat all of the indigenouslanguagesof the Americasfall into only threegeneticgroups:Eskimo-Aleut,Na-Dene,and what he calls "Amerind." His work hasnot generallymetwith success amongspecialists in AMERICAN INDIAN LINGUISTICS 321 the field; seediscussions by Adelaar(l), Campbell(8), Chafe(12), Goddard (26, 28), Golla (32), Jacobsen(4O-42),Matisoff (57), anda reply by Greenberg (35). Both his methodology and its application have been severely criticized. The hitial stepin comparativework hasalwaysbeena surveyof languages for obvious similarities. The better one knows the languages,of course,tlle more similarities one can perceive. When superficial resernblancesare detected among a set of languages,one can begin work to determinewhether tlese aredueto chance,to borowing, or to commoninheritance.Greenberg's methodologyis essentiallythe first step without the second.His technique consistsof what he terms "multilateral comparison":"looking at the basic vocabularyandconcretegrammaticalmarkersof a largenumberof languages (34:648). simultaneously" The methodologypresupposesits conclusion: Some genetic relationship falls out of it automaticallyin any case.Thoselanguagesthat sharethe most superficial resemblancesare assumedto be the most closely related. In his first chapter, Greenbergstatesthat "basically, the wrong questionhas been asked,namely, when are languagesgeneticallyrelated?. . . What shouldbe askedis, how arelanguages to be classifiedgenetically?"(33:3).He clearly expectsthat all languagesin the world will ultimately be demonstratedto be genetically related by his method. "The ultimate goal is a comprehensive classificationof what is very likely a singlelanguagefamily" (33:337). Greenbergfeels that the geniusof the methodlies in the large numbersof languagesconsideredsimultaneously.It is certainlytrue that more similarities will be found amongten languagesconsideredsimultaneously,for example, than between any two. When the number is raised to 2000, all the more sirnilaritieswill becomeobvious.He seemsto havefailed to takeinto account the mounting role of chance. The more languagesconsidered,the more chanceresemblanceswill appear.Mohawk, for example, contains 9 consonants:/, k, s, n, w, r, y, h, 7, The verb root for 'eat' is -/r-. One would probably not have to look too far to find another languagewith a verb containing /r, or perhapsa sound somewhatlike it such as g or r, whose meaning is somethinglike 'consume,' 'eat,' 'bite,' 'chew,' etc. Unfortunately,the methodologystopsshortof separatingthe role of chancefrom that of genetic relationship in producing resemblances. It might be hopedthat the datapresentedin supportof the proposalscould at least be utilized by other scholars to determine their validity. Unfortunately, as most reviewershavepointed out, the dataare so riddled with seriousproblemsthat they could not provide a reliable basisfor any further work. The problemsrangefrom lowJevel clerical errorsto the criteria usedin determining sound and meaningcorrespondences. Basic identificationsof languagesarenotoriouslyunreliable.In many cases 322 MITHUN it was not noticed that different word lists actually representa single lan_ guage. Kashaya and Southwestpomo are in fact the same langua=ge, for example,as are Clear Lake andEasternpomo. Theseare simply w:ell_town altemativenamesfor the samespeechcommunities,but they arepresentedas distinctlanguages. Campbell(8) lists caseswherenamesof to*n, o. people_are confused with languages.particular languagesare not always "u"n given the sameaffiliation. Catawba,for example,is sometimesgrouped with the Iroquoianfamily (33:179),sometimeswith the Siouanfu.ily ?::,fZS, which is correct). Greenbergnotes that work as ambitious as his had not been attemDted previouslyowing to thepaucity andinferior quality of documentationof many of the languagesof the New World. Inferior datapresentno problemwith his method. "The fact is, the methodof multilateral comparisonis so powerful that it will give reliable resultseven with the poorestof materials,,(33:29). The materialsarc indeedpoor, but they neednot havebeen.His dataareoften drawn from brief early notesmadeby explorerspassingthrough an areafor the frst time, rather than the rich, technically excelient diJtionaries and gnmmars now available. In an attemptto increasethe compatabilityof the lists, he hasretranscribed them into his own system,apparentlywithout knowledgeof the actualsound systemsof the languages.Numerouserrorshaveaccordinglybeeninhoduced at every tum. The retranscriptionadditionallyrendersit impossibleto recover the original sourcesof the material, noneof which are citJd because,.listins all tlese sourcesin a generalbibliography would have addedgreatly to th! lengthand cost of the work" (33:xv). Identification of comparableforms is a seriousproblem. Many Amencan languagesare polysynthetic: Words arc often long, composedof multiple meaningfulparts. Accuratecomparisondependson the correct identificatron of the parts, becauseit is tiese that will be comparableacrossdiachronically relatedlanguages.Compare,for example,the two wordsbelow. Mohawk Tuscarom naienenhstoientw:ko, lahni2nt? 2 .that com will be ha ested, .he sat there' Theseare in fact related.They are basedon cognateverb roots that havenor evenchangedmuch in soundor meaningfrom their original forms, although the spelling conventionsarc slightty different. Mohawk -lezr_and Tuscarora -yt- are both traceableto a Proto-NorthernIroquoianroot *_y€r.seVlay.,The Mohawk word means literally,that-would-one_comJay_caor"_r;l,"r."(punctual)': 'that one would unplantcom., The Tuscaroraword means literally 'there-he-self-set-(punctual)': .there he set himself'. Accurate segmentationis obviously not an easytask with suchlanguagesif one is not AMERICAN INDIAN LINGUISTICS familiar with their grammars.Unfortunately, in Greenberg'swork, wrong cuts are consistentlymade and the wrong parts of words compared. Specialists in most languagefamilies have been perplexed to find in Greenberg'swork spuriousfoms that do not resembleany known forms from the languagethey are saidto represent,evenwhenmiscuftingandmiscopying are taken into account. This is unfortunate, in that it contributes to the problem of an uffeliable data base. One scholarremarkedthat not a single form attributedby Greenbergto a languagehe works with is actuallyfrom that language. Greenberg'sgroupingsof the forms into "etymologies" are disappointing for severalreasons.It should be noted that in contrastwith usual usage,in which "etymology" refers to the demonstratedhistorical sourceof a form, Greenberg's"etymologies" are simply sets of forms that appearto him on inspection to be similar in form and meaning. His notion of similarity is considerablylooserthan most. One setof "morphemes"groupedinto a single 'Amerind etymology," for example (33:233-j4), are variously glossed 'mouth,''ear,' 'listen,''cave,''hear,''smell,''blownose,'and'sniff.'The form they shareis somekind of sibilant,variouslys, .1,i', f, f', d, x, plus nasal,m, or n, althoughnot all forms containa sibilant, or a nasal,or either. The vowel betlveenthe sibilant and nasalis variously i, e, , q, o, u, or tlo vowel at all. Someof the forms contain materialbeforethis sequence,some after, and someboth beforeandafter. While shifts in meaningbetweensome of the glossesare not impossible,andshifts betweensomepairsof soundsare also not impossible,it mustbe admittedthat ifone wereto examineall words for'mouth,' 'ear,' 'listen,' 'cave,' 'hear,' 'smell,' 'blow nose,'and'sniffl in 2000 unrelatedlanguages,the probability of finding severalforms containing a sibilant and/or nasal is not small. When such wide rangesof sound and meaningare classifiedas equivalent,it simply becomesdifficult to rule out the role of chalce. Greenberg's distribution of forms into "etynnologies"shows numerous inconsistencies.Often words from a singlelanguagefamily, known not to be cognate,are listed aspart of the sameetymology.Under 'sweet,' for example (33:178),are listedSenecaazz- andMohawkaniere,supposedly relatedto *?an'e:m 'be tasty.' The Senecaform (actually-ez) is the Proto-Keresan descendantof Proto-Northem-Iroquoian*-ran-. The Mohawk word -nielrte' (root *-nyeht- with neuterprefix o-) means'snow'. (Entirely different words areusedfor 'sweet'and'sugar'.)At the sametime, formsknownto be close cognateswithin families arc listed underdifferent etymologies.Threedifferent "Amerindian etymologies" are given for words meaning'man,' for example(33:'24243). Under Manl is given, amongothers, Central Pomoiai (actruallyia:i'), related to Achomawi is, Chimariko iCi, iti, Shasta?is, Tequistlatecaians, Yara ftisi, Wintun siw-y, Plains Miwok sau'we,Chiti 324 MITHTJN macha ?qsi, Tunica.ii, Atakapa,i4, li, Huave na-iej, and Quiche Mayan air.J. Under Man2 is Eastern Pomo ka:/, related to Alakaluf freknje, Aymara hake, Iaqans haqi, Barbareflo Chumash axor,r, Ynezei:,o uryiy, Coahuilteco .ragr, and Karankawa qhaks. ln fact, the Central Pomo iq:i' and Eastem Pomo ta.'// are well-known, systematically related cognates. Of course tlrere is an easy response. If all languagesof the world are assumedto be ultimately related anyway, the different "etymologies" for'man'may themselvesbe cognate. Unfortunately, too many of the etymologies Greenbergprovides as evidencefor wider relationshipsbetweenstocksactually consistonly of cognates from languageswhose genetic relationshipshave already been established. One of the subgroupsproposedunder "Amerind" is "Almosan-Keresiouan," consisting of the Caddoan, Iroquoian, Keresan, Siouan-Yuchi, Wakashan, Chemakuan, and Salish families. The relationship of Caddoan, Iroquoian, and Siouan was establishedlong ago and discussedin severalworks referred to by Greenberg. Yet a large proportion of the etymologies cited as proof of "Almosan-Keresiouan"contain forms from thesefamilies alone. Thesesetsof forms actually constitute no more than hypothesizedevidence for what is already known. In many cases,the number of forms under a single etymology is also small. Jacobsen(41) points out, for example, that of 281 compafative sets in Greenberg'sAmerind Etymological Dictionary, 107 contain only two members. The averagenumber is 3.5. Abundant use of the transitivity argumentfurther complicatesthe weighing of evidence. It is assumedthat if a word in languageA is related to one in languageB, and a word in languageB is related to one in languageC, then languageA is relatedto languageB, B is relatedto C, andA is relatedto C. In principle, such reasoningis valid. Transitivity chains can only be as strong as their weakestlinks, however. If one relationshipfails, the whole chain falls. In Greenberg'swork, whole families and stocks are in many casesjudged to be relatedto other stockson the basisofa few similarities among a few words in a few languages. In too many cases the particular forms chosen for comparison are disappointing. It is clear from his notebooks,on file at StanfordUniversity, that Greenberghad accessto publications containing full cognatesetsfrom many families, complete with reconstructionsand detailed listing of the sound changes undergone by each language. Many of his forms were drawn from these sources, although they are not cited in his publication. Yet in many cases he chose idiosyncratic forms that are known to be the result of recent innovations, because they were superficially more similar to other forms outside of the family. Of course it would be a phenomenalachievementto deal with the number of languages discussed in this work without some inaccuracies in detail. AMERICAN INDIAN LINCUISTICS 325 Greenbergfeels that suchdetails are unimportant,becauseof the sheermass of the data. Unfortunately,mostspecialistswho haveexaminedthe work feel that the massof the material falls on the wrong side of accuracy.Goddard examinedthe 142 word setsthat include Algonquian forms and discovered that"errorsin theAlgonquiandataaloneinvalidate93 oftheseequations.. . . There remain 49 word sets in which the shape and meaning Greenberg assumesfor the Algonquianforms are approximatelycorrect ., An evaluationof theseproposedequationsshowsthat they do not demonstrateor for Algonquianandmustbe considered evensuggestgeneticrelationships to be the result of chance,if they require any notice at all" (26:656). Most Amedcanistsarethemselvesextremelyinterestedin uncoveringdeeper geneticrelationsamongthe languagefamilies of the New World, and hypothesesarean importantfirst stepin the process.UnfortunatelyGreenberg hasherepresented hypotheses asfact. It mightbe suggested thattheproposals simplyneedto be disproven.Unfortunately, thereis no directway to demonstratethe lack of a geneticrelationshipamonglanguages. At best,one can note that there is insufficientevidenceto posit a particularrelationship. Althoughspecialists arein a positionto evaluatetheevidencepresented here, thosein otherdisciplines,to whom suchproposalsmay be of considerable importance,may not be. Most linguistscurently workingwith North American languageswould ratherspendtheir time creatively,exploringactual relationships, thantryingto disentangle mistakes.Only too soon,directstudy of theselanguages will no longerbe possible,andthenumberof scholarsin a positionto evaluatethe evidenceat all will havediminisheddrasticallyas well. RESPONSIBILITIES Most linguists working in Native Americancommunitieshavegrappledwith the issueof their appropriate rolesandresponsibilities in thesecommunities. Thereis greatconcemin mostcommunities aboutthepossibledisappearance of their languages. Whena languagedisappears, someof the mostimportant parts of a culture disappearaiong with it. Ways of thinking can be lost along with waysof speaking.Oftenmuchof the cultureis not handeddown in the new language.Sometimes traditionallegendsand storiesare not passedon, becausethey lose so much when renderedin a new language.Traditional etiquetteandstylesof humormaybe lost in translation.Youngergenerations can be left without the sameawarenessof or feeling for their cultural heritage thattheirancestors had.Sucha lossmaynot haveobviousimmediatematerial ramifications,but it canhavea profoundeffecton anindividual'ssenseof self and community. Decisions about what should be done as languagesare threatenedwith 326 MITHUN disappearancebelong with the communitiesthemselves,the speakersand their descendants.Opinions vary considerably, *ithin communides, concemingrvhatactionsshouldbe takento record"u"n the languagesandio teach them to youngergenerations.Somepeopleseelittle pointin jocumentrng otO ways in a modem world, while othersare intently aiwareof tfre greai toss of knowledgethat occurseverytime a speakerpassesaway. S"-" ii"t tfr"t a.tp"t an.Indianlanguageis economicallyus"l"rr,'-- *o..", ""t u :9111y nanolcaprn_an English_speaking world. Many of thesepeopleare themselves successfulbilinguals, but they rememberthe misery of arriving at school without English-andof beingridiculed andpunishedfor speaking tiJ motfre. tongue.Othersfeel stronglythat they want to equip youngergenerations with the sbnseof self that can come from knowing uUo"t ,i.i."fr".ioe". Sonr" believe that documentationof the languageis worthwhile Uut uti"_po ,o teachthe languages wiD be fudle. Education programs have tle best successwhen the impetus for their creationcomesfrom within. Many communitiesare now manuging tar,grrage pJogramsbut calling on linguistsfor technicalsupport in certainaris. one of theseis^orthographydevelopment:. devisinga workablepractiJ spetting ryrT-.fo1l gartrlularlanguage.Anotheris the development of language curricula. Mohawk lessonsbasedon the model of Frencir will never make sense,but a linguist shouldbe able-to show teachershow to devisepn grams that will presentthe languagein its own terms. A third i. tfr" t li".g Natt:/: te-ach-erl. Being able to speaka languageis essentiala U"i*-uUf",o"f teach it, but it is not sufficient. Most languagestructuresare **i"iou.,o native speakers,especiallyspeakersof languageswithout literary traditions. Although languageteachersneednot teachgrammaras a subjeci, un u**"_ ness of the structuresof their languagecan allow them to presentmore cohetent classes-.Finally, linguists can provide tretp in' Jeuetoping documentationof the language:dictionaries, texts, and p"rfrrp, grua^u.r. uTu.ut meetitrgsof specialistsin most of'thJ tanguage ,^]:l-" T: T*- of ramlltes, At most these,Native languageteachersand linguistscom€ togetherto discussthe issuesinvolved in the maintenance of d;u; programs_as well asvariousaspects of the languages themselv.r.to ,.i"niy"_r, the Athabaskanconference,the Salishconierence,the meeting of the Friends of Uto-Aztecan,the Siouanconference,the Iroquois confer"n'c", ,t Afgon" undoubtedtyothershave all rraOsesrions'spectricaUy ::.T^"j:1.:Tl oevoredto tanguage -d teaching, and will undoubtedlycontinue to do so. CONCLUSION The languagesof North Americaoffer a wealthof examples ofthe varietiesof linguistic form. This variety is delightful in itself, ani l i, *u "-"iuiio AMERICAN INDIAN LINGUISTICS 327 general understandingof the phenomenonof language. North Amencan languagescontain someof the most elaboratephonological,morphological, and syntactic structuresin the world. Most study of these languagesis securelygroundedin data, usually gathereddirectly by the researcherin the field; but it is not necessarilyatheoretical,althoughthere is little interestin model building for its own sake.Becauseof the structural, social, typologi cal, and diachrcnic contextualizationof most research,more kinds of explanationsarc called into play. Describingsuchwork, Kroskrity (52:325) aptly commentsthat "this moreholistic approachemulatesthe anthropologicalideal of understandingcultural phenomenanot only in terms of contemporary structureand function, but also as the productsof a uniquehistory involving both continuation of cultural tradition and diffusion from neishborins groups." Literqture Cited 1. Adelaar, W. F. H. 1989. Review of Gre€nbery.Lingua 781249-55 2. Battiste.M. A. 1984.An historicalinvestiqation of the social and cuhural consequencesof Micnac literacr. Pl:tD thesis. Stanford Univ. 212 pp. 3. Bright, W. 1978.Sibilantsand naturalnessin aboriginalCalifomi^.J. Calif. Anthropol.Pap. Lineuist. li39--64 4. Bright, W. 1979. A Karok myth in "measuredverse":the translationof a performance.J. Calif. Great Basin Anthropol. l.ll'l 23 5. Bright, W. 1979.Noteson Hispanisms. Int. J. Am. Linguist. 451267-8E 6. Bright,w. 1980.Coyote'sjou$ey.Arr. Indian C lt. Res.J. 4(l 2)2148 7. Bumaby,B., ed. 1985.Promotih|Natiw Writing Systemsin Canada.Torontor Ontario Inst. Stud. Educ. 222 rD. 8. Campbell,L. 1988.Review articleon Cneenberg 198'1.Language(*591-615 9. Campbefl.L.. Mithun. M. 1919.The LanSuagesof Nattue America: Historical and Comparatiy€Assessment.A\\stinr Univ. Texas Press. 1034 pp. 10. Chafe.W.L. 1985.Texrforna$ing asa theory of languaee producrion. tuesentedat Annu. Meet. Am. Anthropol. ' Assoc.,84th, Washinglon,DC I l. Chafe,W. L. 1985.Informationflow in Senecaaod English.Proc. Ilth Annu. Meet.BerkeleyLinguist.Soc.pp. 14-24 12. Chafe, w- L. i98t. Reviewbi creenberg 1987. Curr. Anthropol. 28:652-53 13. Chafe.W. L. 1989.Hota)theorde neoI verbsand no n phrasesis detcnined by infon&tion flow in Caddo. Prcsentedat Arnu. Meet. Am. Anthrcpol. Assoc., 88th, Washington,DC 14. Cook, E. D- 1989.ls phonologygoing haywire in dying languages'!Liiglsoi 18:2:23155 15. DuBois, J. 1986. Self-evidence and itual speech.ln Efident tlity: The Linguistic Codine of Epistemology,ed. W. L. Chafe,J. Nichols,pp. 3t1 36. Norwood, NJ: Ablex. 346 pp. 16. Eastnan,C. M. 1979.Word otder in Haida. AL 45t14148 17. Eastrnan, C. M., Edwards,E. A. 1979. New information and conhastivenessrn Haida. Pap. LinBuist. 12'.39-56 18. Edwads, E. A. 19'19.TopicnarkinBin Haida. IJAL 45t149'56 19. Edwards,E. A. 1983.Focusand constituentorder in Haida. Can. J. Liheuist. 28:.149-57 20. Emico, J. 1986.Wotd order, focus, and toDic id Haida. IJAL 52t91 123 21. Fdwler,C. S. 1983.Somelexicalclues to Uto-Aztecanprehistory./"141,49: 224-57 22. camble, c. 1989.Spanishloansin Wikcharnni. In General and ArErindian Ethnolinguistics: In Remembranceof StanleyNewnan,ed. M. R. Key, H. M. Hoenigswald.Berlin: Mouton. 496 pp. 23. Gamble, G. 1989. Palewyami: lhe yokuts arclror. Presented'at Annu. Meet. Am. Anthropol. Assoc.. 88th. Washingron, DC 24. Goddad, I. 1983.The EastemAlgonquian subordinative mode and the-imtortance of morphology. IJAL 49,351E7 25. coddard, L 1985.Patternsof Enru..rlc, discourse, a d deliyery features in a hi+h-reqistcr Unami (Algonquian) text. Presented at Annu. Meet. Am. An- 328 MITHUN thropol.Assoc.,841h,Washinqlon.DC 26. Goddard.L 1987. Review oT creertf"rs,. Curr. AnthroDol. 28.656 57 27. Coddard.f. 1988.-srylisticdialecrsirt Fox linguisric change. lt\ Historical Dial?ctoloey,ed. J. Fisiak,pp. 191-209. llerlrn: Mouton 28. Goddard.L l99O.Reviewofr JoseDh H. GreenbeE:Innguage in the Anelricas. ZirrS4r1rrr'.ar. In press 29. Goddard,l. 1990.Aspectsof the topic sEucturcof Fox narrltives: prorimbrr shifts and the use of overt and ir! flectional NPs. 1JA-L56:3. In Dress 30. Coddard.I., Bragdon.K. l98i!. Na,iy. Writingsin Massachusen. Philadelphia. Am. Philos.Soc. 791 pp. 31. Goffa, V. 1985. Meaiured verse in Hupa traditiotal naftatiy?. Presentedat Annu. Meel. Am. Anrhlopol. Assoc. 84th, Washington,DC 32. GoUa.V. 1987.Reviewof: JosephH. Grcenbetg:Language ih thc AmeTicas. Cuft . Anthropol. 28i651-59 33. Greenbelg,J. H. 1981.Languaeein the Aneflcdr. Stanford: Slanford Urd!. Press.438 DD. 34. Greenberg, J. H. 1987.Language in the Americas:author's precis. Cuft. Anthtopol. 28:647-52 35. Greenberg. J. H. 1989.Classilication of AmericanIndiao laneuaqes: a reDlr tu Campbell.Language-65ii07-14 36. Hill, J- H., Hill, K. C. 1986.Speaking M?xicano:Dtnamicsof Slncrcti. I^1nguagein CenhalMexico.Tucson:ljniv. Arizona Prcss 37. Hymes, D. l98l. "In vainI tried to tell you" : Essa}sin Nativ?AmericanEthnopoetlcs. Philadelphia: Univ. PeIlIl Press.401 DD. 38. Deleted in iioof 39. Jacobse!, W. H. Jr. 1979.HolaninteF bmnch comparisons.See Ref.9. DD. 545-9l z!0. Jacobsen, W. H. Jr. 1988.Greehb?ry or' Hokon. Prcser,lf.dal 1988 HolanPenutianWorkshop. Univ. Ore., Eugene 41. Jacobsen,W. H. Jt. 1988,A look a, Creenbergs Alnosan-Kercsiouan hJpot ?sir. Presetrted at 2lrd Inr. Cont.on Salish^and Languages. ^Neighboring untv. ure., Eugene 42. Jacobseo.W. H. lr. 1989.ThePacifit orientation of westem North Ameriaan languages.Ptesented al Circum-Pacifi c Prehist.Conf., Seanle,Washinston 43. Jelinek. E. 1984. Empry caGgories, case,andconfisuralionllt|. Nat, Lane. and Linguist.Theory2(1):.39-76 44. Jelinek, E. I987. 'Headless'relarives and prcnominalarguments:a !''pological perspective. In Native American LanSuagesand GrummaticalTtpolosy, ed. P. D. Kroeber, R. E. MooG,DD.-f3G 48. Chicago:lndialla Udv. -Linquis Club 45. Kess,J., Kess,F., Kess.A. C. 1986. On Nootkababytalk. tAL 52:201-11 46. Kimbalf, C. 1989.Anotherproposalfot subgrcupingthe MuskoReaildnsuaies. Prcsent€dat Soc. Stud. Lans. Arn. Meet., Tucson 47. Kinkade,M. D. 1985.ThelineinlJDDer Chehalis nanatfue. Presentedat Aitiu. Meet. Am. A hropol. Assoc.. 84th. Washington,DC 48. Kinkade. M. D. 1989. Cotnparative linguistic evidenceabout Satishpreiistor). Presentedat Annu. Meet. Am. Anthropol. Assoc., 8Eth. Washinston. DC 49. Kinlade. M. D. 1990.Soningout thim persons in Salishatrdiscoufse.,JAZ 56(3): In press 50. Kroeber, P. D. 1988. Discourseano funclionailactorsin thedevelopmenr of Southemlnlerior Salish ergativecax markiLg. Proc. I4th Annu. Meet.Berkeley Linguist.roc., pp. 100 23. Berkeley: BerkeleyLingulst.Soc. 51. IGoskrity, P. V. 1984. Nesarionatrd subordinationin Arizona Tewa: discouse pragmatics influencins svnt&{. Int. J. Am. Lineuist.50:9,Lld4 " 52. Ituoskity, P. V. 1985.crowinq with stories:Iine, verse,and senrein tn Arizona Tewa Iext. J. inthroDot. Res 4l(2\:la3-2n0 53. Ituoskrity,P. V. 1985.A holistic understanding of ArizonaTewa passives. Laneuaqe 6l:306-28 54. Kroskdty, P. V., Reintrardt,c. 1984. SpanishaIId English loanwordsin Westem Mono. J. Calif. Great Bosin Anthropol.:Pap. Linpuist.4.107-38 55. Laoidon, M. 1979.Soqe thouqhrson Hokan with particular rcferenctto pomoaoandYuman.SeeRef. 9, pp. 592649 56. kngdon, M., Silver,S. 1984.Californi^ t/t. J . Calif. creat BasinAnthropol.: Pap. Linguist. 4t139-65 57. Matisoff,J. A. 1990.On mesalo-comparison:a discussjonoote. -l,anRuape 66:L ln press 58. Mctandon, S. 1982.Meanins.fietorical structure,anddiscourse or-sanization io myth. In Aulyzine Discoirse: Te aM Talk, ed. D. TanneD,pp. 28+305. Washington,DC: GemgetownUniv. rrcs6. J6Z DD. 59. Mclendon, S. 1985.Easten pomo oratory. Prcsentedat Annu. Meet. Am. Anthropol. Assoc., 84th, Washingto[, DC AMERICAN INDIAN LINGUISTICS 60. Miller, W. R. 1984.The classification basedolof the Uto-Arlecanlanguages lexical evidence.IJAL 50tl-'24 61. Mithun,M. 1984.TheProto-Iroquoians: cultural reconsFuclion from lexical materials.fn E tcndingthe Rallets:In' te is.iplinary Apprca.hes to lroquoian Sradies,ed. M. K. Fosler.J. Campisi M. Milhun.pp.259 8l Albany:Sttl'{Y Press.422 pp. 62- Mirhun. M. 1984. How lo avoid subordination.Pro.. lqth Annu MPet Be' rkleJ Linquist.Soc.,pp.49!523 8erkelev:BerkelevLinguist.Soc wrjte 63. Mithirn.M. 19{5. wien speakers Proc. llth Annu. Mcct. BerkelevLihpa|Jr. .Soc., pp. 25q 72. Berkeley: Soc. EerkelevLinqiilt. 64. Mirhun. M. 1q86. when zero isn l there.Prc.. l2th Annu M?ct Berkcl"y Linsuist.Soc.,pp. 195 21I 65. Miiruo, M. 1987.ls basicword order uni\e$ t? b Cohercnceand Grcundinq in Discourse,ed. R. S. Tomlin' PP. 281-328.Amsterdam:JohnBenjamins Jll Do, 329 ?5. Pa*s, D. R., DeMallie, R. J. 1989. Sio r-Assiniboine dialects: a clossification. Presentedat Amu. Meet. Am AnDC thrcpol.Assoc.,88th,WashitrgtoD, 76. PaGe. D. L. 1987.Informatioostructuring in Papago namtive discouse LanBuane63:181904' ??. Rhodes;R. A. 1989.TheCreeconnet,ion. Presentedat Annu. Meet Am. AnDC tfuopol. Assoc.,8E$, Washington. 78. Rooa, D. S. 1985.wichita tet( strucrurz- Presentedat Annu. Meet. Am. AnDC tlEopol.Assoc.,84dr,Washington, N. 1986.Semanticandpragmatic 79. Rud-e, obiectsin Klamalh. ln /n Honot of Mary pp.651-74BeF H;ds,ed.W. Shipley. lin: Mouton. 826 PD. 80. Rude, N. 1989.E gative,passive,and andpassivein Nez Perce:a discourse ln Passiveand Voice,ed. oersoective, M. Shibatani.pp. 547 60. Amslerdam JohnBeniamins.706 pp. 81. Sapn, E. 1915. AbnormaltyPesof sp+ch in Noo*a. Can. DePt Mines. Geol. Sun., tlem.62, Anthropol Ser 82. Sapir,E. I929. Noo*a babywords./nt 66. Mithiuir, M. 198?. The gammatical naJ. Am. Linauist.s:llFlg ture atld discoursepower of demonsraJ. 1q83.Kuna wals o[ Speak' 83. SherTer, tives. Proc. 13th Ann. Meet. Berkeley inq: An Ethnographi.PerspcctiveAvsLinpuist.Soc., pp. 184-s4. Berkeley: tin: Univ. Texas Press Berlelev LinsuiaaSoc. 6?. Mithun;M. 1989.The incrpientobsoles- 84. Sherzer,J. 1987.Poeticstructuringof Kuna discourse: the line ID Ndtfue Ca)'!8ain Ontario cenceofpolsynlhesis: Americandiscours?,ed. J. Sherzer,A and Oklahoma. In lnvestiSatine C. Woodbury, pP. 103-39 Cambridge: ed. N. C. Dorian, PP. Obsolescence, CambddseUniv. Press 243-57. CafilJ'jdge: Cambridge Univ. 85. Silve$teln, M. 1979. Penutian: an hess. ,145Dp. assessment.SeeRef. 9, PP 650-91 68. Mithun, M. 1990. lnryuage obsoles86. Sta*s, D. 1987. Word ordering: morc cenceand gpmmatical description.L/AZ than ordering subjecls. objects and 56tl-26 verbs.In Na,iveAmericanLanguages 69. Mithun, M. 1990. Third peNon referand Granmottcal TtPolog, ed. P. D. ence and lhe function of Pronounsin Krceber,R. E. Moore,PP 2l5-3l ChrCentral Pomonarunl speeci. IJAL 56t3 caso: 229 pp. In press 87. Tavlor. A.-R. 1989.Problemsin obsu70. Miihun, M. 1992. The extemal tdggerlesience research:the Gros Ventres of ing of internallyguided syntacdcdeMontana. In InvestiSating Obsolesc' velooment.ln Intenul and E ernol erce, ed. N. C. Dorian, PP 16?-79. Faciors in Syntactic chanSe, ed. M. Cambridge:CambddgeUniv 445 PP. Cenitson. D. Stein. Berlin: Mouton ln 88. Tedlock,D. 19'12.Findingthe Center: DIESS Nafiative Poefiy of the Zuni Indians. 71. Moore, R. 1988.LexicalizationveNus New York Dial lexical lossin Wasco-Wishamlanguage 89. ThomDson,C. L. 1989.Pronounsand obsolescence.IJAL 54].4544 voice in Kolukon AthaPaskan:a textII pre72. Muffo, P- 1987.TheM skogean basedstudv.IJAL 55tl-U their inplications for classtiand fites -calion. Presented 90. ThomDson:N. R. 1985.rwana babY at KenruckyForcign talk ind its theoretical implications. Lang. Conf., t€xinglon PhD lhesis. Univ. Wash. 307 PP. ?3. O'Cbnnor, M. C. 1990.Third person reference in Northem Pomo convefsa_ 91. Watkins, L. J. 1985.Spokenand written svle in Kiowa. Presenled at Annu tion: the indexingof discou$e geffe and Meet. Am. AnthroPol.Assoc.,84th, social rcIations. IJAI 56:3 Washinstotr. Dc 74. Oswalt.R. L. 1958.Russianloanwods 92. watkins-,L. J. 1990.Noun phrasever' Pomo.llAL 2+245 47 in sourhwestern 330 MITHUN suszeroin Kiowadilcourse.lnt. J. Am z,inatisl. 56:3. ln Dress 93. Whistler.K. W. t985. Focus,perspective, andpeFonmarkinsin Nootkad.It Grummai lnside and Ourside thc Clause, ed. J. Nichols, A. C. Woodb!ry, pp. 227 65. Cambridee:CambridgeUniv. Press.419 DD. 94. Whisller,K. W., Colla,V. 1986.protoYokuts reconsidered.IJAL 52:31'7-58 95. Woodbury, A. C. 1984. Cev'armiut qanemciit quhrait-llu: Naftatives and Talesfron Ctuvak,Alaskrr.Fairt)^r*s. Alaska Native Lang. Cent., Univ. Alaska. 88 pp. 96. Woodbury,A. C. 1985. Furctionsof rhetorical structure: a study of Central AlaskanYupi} Eskimodiscourse. farB. Soc. 14:150-93 97. Woodbuy, A. C. 1987. Rhetorical structurein a CentmlAlaskanYupik Es_Nariv, kimo traditional na[ative. In AmericanDiscourse,ed. J. Sherzer,A. C. Woodbury, pp. t76-239. Cambridge:CambridgiiJniv. Press.256pp. i' 1 I l