Ergativity, Case, and Transitivity in Eastern Pomo Author(s): Sally McLendon Source: International Journal of American Linguistics, Vol.

44, No. 1 (Jan., 1978), pp. 1-9 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/1264759 Accessed: 07/12/2010 05:09
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ucpress. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of American Linguistics.

http://www.jstor.org

InternationalJournal of AmericanLinguistics
VOLUME 44 January 1978 NUMBER 1 ERGATIVITY, CASE, AND TRANSITIVITY IN EASTERN POMO1
SALLYMCLENDON
HUNTER COLLEGEAND THE GRADUATE CENTER, CITY UNIVERSITY OF NEW YORK

Evidence that Eastern Pomo is ergative Pronouns Classes of nonverbs Classes of verbs Syntactic implications: plurals and switch-reference 5. Conclusion 0. 1. 2. 3. 4.

0. The classic definition of an ergative language-ergative languages (1) mark difxai s wa-du'kiya2. ferentially the subjects of transitive verbs Rattlesnake' went away2. as distinct from the subjects of intransitive Eastern Pomo thus looks quite ergative. verbs and the objects of transitive verbs This ergative type of distinction goes (which are marked the same way), and beyond the morphology of nouns, however. (2) lack passives-associates ergativity and A number of verbs show a variation in their 1 Fieldwork on Eastern Pomo which supplied stem shape to mark plurality which also data for this article was carriedout between 1959 follows an ergative pattern. That is, change and 1976 with the support of the Survey of in the shape of the stem marks plurality of California Indian Languages, the American the objects of transitive verbs such as kill, PhilosophicalSociety, the City Universityof New and marks plurality of the subjects of York Faculty Research Award Program Grant and the GuggenheimFoundation, to all intransitive verbs such as sit, but never #11369, of which I am most grateful. I am especially marks plurality of the subjects of transitive grateful to the Guggenheim Foundation for a verbs like kill. Thus: research fellowship in 1975-76 which freed me (2) du'di'ka from teaching responsibilities so that I could put on one thing pursue this line of researchand carry out further du'ya'wka fieldwork. Versions of this article were read to the Fourteenth and Fifteenth Conferences on put on several things American Indian Languages at the American (3) sa'k Anthropological Association meetings in San kill one Francisco, 1975, and in Washington, 1976, and du'ley at a colloquium of the Departmentof Linguistics, kill more than one of California,San Diego, in the spring University of 1976. I am grateful for all the stimulating (4) phu dipki discussionduringthese presentations,particularly hang up.one thing to Margaret Langdon, Sandra Chung, Alan phu'bepki Timberlake, and Arthur Schwartz-none of hang up more than one thing whom necessarilyagrees with me. (5) pha'di'l [IJAL, vol. 44, no. 1, January 1978, pp. 1-9] one leaf drifting ? 1978 by The University of Chicago. All rights reserved. 1

transitivity. Eastern Pomo, a Hokan language spoken in Northern California, lacks a passive and has noun suffixes, {-la'}, {-u'la'},which specificallymark the subjects of transitive verbs (but not of intransitive verbs). Compare: (1) xa'su'ld'l wi2 kokh6ya3. Rattlesnake' bit3 me2. 12 was bitten3by the rattlesnake1.

INTERNATIONAL JOURNAL AMERICAN OF LINGUISTICS TABLE 1

VOL.44

{ha'} lst-person singular........................ {wi} Ist-person singular oblique stem .............wi {ma} 2d-personsingular.........................ma {mi} 2d-personsingular oblique stem .............mi {wa'} Ist-person plural.......................... {mi-} 2d-personplural......................... {mi'-} 3d-personsingular + {-5} masculinestem /mi'p/ ..................mi'f + {-'} feminine stem /mi'f/ .................... {b6kh} 3d-personplural......................... {hi'} anaphoric 3d-personsingular.............. {h6ekh} anaphoric 3d-personplural............... {ki'ya-}interrogativeof person................... phayaiw

hi

wa ma.

wi'l ma'l mi'pal mi'ral b6ekal hi' hekal ki'ya'l

mit' b6ekh hekh ki'ya'

many leaves drifting (6) ka. one to sit/dwell na.ph6 several to sit/dwell 1. The pronouns, however, have previously been described by Kroeber2 and deAngulo and Freeland3 (as well as myself)4 as distinguishingsubjectsof transitive and intransitive verbs from the objects of transitive verbs, that is, as following an accusative rather than an ergative pattern. (The relevant inflections of pronouns are charted in table 1.) This apparent pattern of ergative inflection for nouns versus accusative inflection for pronouns is analogous to that described in a number of languages and has been called a split
2 Alfred L. Kroeber, The Languages of the Coast of California North of San Francisco, University of California Publications in American Archaeology and Ethnology 9 (1911): 320-47.

3 Jaime deAngulo and L. S. Freeland, "The 'Clear Lake' Dialect of the Pomo Language in North-Central California" (manuscript in the Boas collection of the American Philosophical Society, Philadelphia;circa 1935). 4 Sally McLendon, A Grammar of the Eastern
Pomo Language, UCPL, vol. 75 (Berkeley and

ergative system.5 If one examines only Eastern Pomo translations of typical examples of transitive and intransitive sentences in English, such as (7), the accusative patterning of pronouns seems clear: (7) ha61wa-du'kiya2. I'ml going2. ha1. mi.pal2 sa-ka3. I1 killed3 him2. Nevertheless, if one considers something close to the entire repertory of verbs, the pronominal forms which at first seem to distinguish subjects and objects in this typically accusative manner turn up in various perplexingcombinations with what seem to be both transitive and intransitive verbs. Thus, although some apparently intransitiveverbs such as wa-du'ki'go away and kaM always select the same form of sit the pronoun to translate what is the subject in English, as do clearly transitive verbs such as kill (column I in table 1), other apparently equally intransitive verbs such as fall, be burned,be blistered, forget, bleed, split open, feel heat, dream, sneeze, belch, be stuck, be tired, feel good, think of, get crippled, become crazy, become wrinkled,
See, for example, Michael Silverstein,"Hierin archyof Featuresand Ergativity," Grammatical Categoriesin Australian Languages,ed. R. M. W. Dixon (in press).
5

Los Angeles: University of California Press, 1975).

NO. 1

ERGATIVITYIN EASTERNPOMO

become sick or choked, become full, be frightened, become frightened, and get cramped require the use of Eastern Pomo pronominal forms which are also used to translate the objects of transitive verbs (column II in table 1). Thus: (8) mi'pal1 xa'2 ba'ku'ma3. He1fell3 in the water2. wil ?eckiya2. I1 sneezed2.

Academic Press, 1976],p. 5) distinguisheda third type of language from ergative and accusative ones, of which he cites Dakota, Wichita, and a Caucasian language, Bats, as examples. He provides the following Bats examples which are strikingly parallel to the Eastern Pomo phenomenon described:as woie Ifall; Ifell [onpurpose]; so woze me fall; I fell [by accident]. Curiously enough, however, the verb to fall behaves differently in Eastern Pomo and Bats, since in Eastern Pomo, fall may only occur with a single patient unless further suffixed with the causative or reflexive. Thus Eastern Pomo and Bats seem to have similar systems, but verbs with the same semantic referentsare assigned to differentparts of the system in each language.

Those verbs which select for the pronominal forms that mark the objects of transitive verbs (column II in table 1) can occur with the pronominal forms that mark the subject of transitive verbs (column I in table 1), however, if suffixed with the reflexive {-k} or the causative {-qa}; compare (12) below with (9) above: (12) ha'l da-sul/ka2. IP can'tfind something2. mi.pl iuy- u.wi2 pha'bekhqaya3 (9) mi'pall pha-beka2. He' got burned2. He' burned3his eyelashes2. Or: wi' da'sula2. I1 misplaced (something)2. (13) mi'p1 x6he2 si'naqa3. Still other, apparently intransitive verbs He' put3 thefire2 out3. can occur with either the form associated versus: with the subjects of transitive verbs or that mi'pal1 x6he72 si'naya3. associated with the objects of, transitive Thefire2 went out3 on him'. Other verbs like di'kh6hit can occur with verbs, the change in the form of the pronoun being associated with a lexical either pronominal form, similar to verbs like ce'xel in (10) above, but with a nonchange in the English translation, as in: lexical change in their translation: (10) wi' ce'xelka2. I'ml slipping2. (14) ha1l di.kh6ya2. hal ce-.xelka2. I' hit (it)2. I'm sliding2. wil di'kh6ya2. I1 got hit2. (11) ha' ba-tecki' I got bumped(on purpose). Some verbs, such as to miss someone or to love/like someone can occur with not wi- ba'tecki I got bumped(accidentally).6 one but two pronominal forms marked as patient: 6 Stephen Anderson (in his article, "On the (15) be-kall wi2 phile-mka3. Notion of Subject in Ergative Languages," in 12 miss3 them'. Subject and Topic, ed. Charles Li [New York: mi'rall wi2 ma'ra'3.
I2 love3 herl.

The Eastern Pomo pronominal system, then, can hardly be called accusative, but it does not conform to the classic definition of ergativity either, since it marks the subjects of some transitive verbs like kill in the same way as the subjects of some intransitive verbs like go, while it marks the subjects of still other intransitive verbs like fall in the same way that it marks the objects of transitive verbs, and still other intransitive verbs can take either or both types of pronominal forms, while some

INTERNATIONAL JOURNAL AMERICAN OF LINGUISTICS

VOL.44

apparently transitive verbs like love occur with two object forms, but no subjectforms. The problem seems to lie in part with the definition of ergativity in terms of a transitive-intransitive dichotomy, in part with the essentially syntactic notions of subject and object. The semantic notions of agent and patient seem more revealing, but again not in the context of a transitiveintransitive dichotomy. Rather, agents naturally occur with verbs that involve a significant degree of what I would like to call protagonist control (to avoid notions of causality and responsibility).7 Patients

naturally occur with verbs that presuppose a lack of protagonist control, verbs that presuppose that the protagonist(s) is/are significantly involved in the activity, but without a controlling role. The protagonist is caught up in the activity described, as when one is overcome with a fit of sneezing or suddenly faints. Verbs which naturally occur with patients and not agents can accept agents if a causative or reflexive suffix is added (as in [11] and [12]). Still other verbs can occur with either patients or agents depending on the speaker's perception of the presence or absence of 7 Larry Thompson (personal communication protagonist control, as in (10) above. of an extract from his paper presented at the 2. In Eastern Pomo, verbs do not conConference on American Indian Linguistics of trast merely with a class of nouns and one the LinguisticInstitute at SUNY, Oswego, 1976) has independently identified an analogous dis- of pronouns as seems usually to be tinction operating in Thompson, a Salishan assumed in descriptions of languages these language spoken considerably north of Eastern days, and as the discussion has so far Pomo: "The rough semantic coverage is an implied. Rather, verbs in Eastern Pomo opposition between those acts, events, or circumare distinguished morphologically and stances over which some individual entity exerts full or considerablecontrol and those over which syntactically from several classes of nonat best limited control is exerted. Probably in all verbs which can be hierarchically ordered Salishan languages a few roots are by nature with respect to the features of specificity, control roots-e.g. 'go', 'come'; other roots (the animateness, and humanness. vast majority) are non-control roots, and it is a Thus pronouns together with kinship natural result of this fact that predicates conof just simple roots are typically glossed terms and proper names8 are set off from sisting

Pomo, but is unrelated to Eastern Pomo. It is unfortunatelyvirtually extinct. 8 Although proper names exist (every Eastern Pomo individualformerly had at least one), they from the Fifth Regional Meeting of the Chicago are in principlealmost never used, and in fact are Linguistic Society [1969]: 348-56) has pointed out the necessity of recognizinga similar distinc- used quite restrictedly,their use being tabooed in tion between volitional, nonvolitional, and most contexts. The precise contexts in which naturalverbsin Cupeno, a Uto-Aztecan language personal names may be used are poorly underspoken quite far south of Eastern Pomo. Both stood at present (and may never be completely Cupeno and Thompson seem to employ different identified since Eastern Pomo is increasingly mechanismsfor indicatingthis sort of distinction rarely used these days by its few remaining from those described for Eastern Pomo. How- speakers).It seems clear, however, that one does ever, Kashaya Pomo, which is related to Eastern not use the name of an individualin his presence Pomo, has virtually the same system of marking or in the presenceof any of his kinsmen, since, as this distinction (Robert L. Oswalt, personal the Eastern Pomo put it, it makes them feel bad. communication), while Kroeber's 1911 descrip- Thus, although EasternPomo can readilyprovide a translationfor a sentencelike Johnkilled Mary, tion of Yuki (The Languages of the Coast of they do not in fact speak that way. Instead California North of San Francisco, pp. 371-72) suggests that it also existed in that language, sentences like He killed her or Her father killed which was spoken just to the north of Eastern his wife are usual.

by passives when a concrete subject is mentioned -or, alternatively phrased, the subjects of such predicates are patients." Jane Hill ("Volitional and Non-Volitional Verbs in Cupeno," Papers

NO. 1

ERGATIVITYIN EASTERNPOMO TABLE 2

all other nonverbs by both their inability to occur with the specifying suffixes{-he?}and {-he)e} and their unique ability to occur with patient-marking suffixes ({-al} for kinship terms and pronouns, {-yiy} for personal names), thus constituting a super class which I call Animates.9 Pronouns, kinship terms, and proper names are inherently specific, hence their inability to occur with the specifying suffixes. They are also inherently human (there being no pronoun "it" and the Eastern Pomo having no named pets or domesticated animals prior to contact). Pronouns, kinship terms, and proper names are inflected for a full range of cases, while categories of location, direction, and instrument are distinguished with the majority of nonverbs which constitute a class called Common Nouns, all of which do occur with the specifying suffixes, but do not occur with the patient-marking suffixes {-al} or {-yiy}.'1 Pronouns and kinship terms may be directly marked for number, common nouns may not. The distinction of number sets off still another class of nonverbs-Personal Nouns, a small, closed set of nouns referring to age grades and statuses of people charted in table 2. Like pronouns, kinship terms, and proper names, but unlike common nouns, all personal nouns have both singular and plural forms, most of which are suppletively related. Like common nouns, but unlike pronouns, kinship terms, and proper names, personal nouns cannot
9 McLendon, p. 82.
10 John Lyons (in his Introduction to Theoretical

THE CLASS OF EASTERN POMO PERSONAL NOUNS Singular

boy.............. girl.............. man............. woman........... young man....... young woman..... old man.......... old woman.......

qa'wi' daxac ka.kh da se'la dayawal bufiike

daqa'ra

qa'wikh qa'ra'ya qa'wikh hi-baya 4a'ra'ya s6'la'ya ma'yawala buc'iya maiqathora

Linguistics [Cambridge: Cambridge University Press, 1968], p. 293) has pointed out a similar feature in Indo-European languages: a distinction between a nominative and an accusative form is never made for neuter (i.e., nonanimate) nouns in an Indo-European language. He goes on to add that this may be regarded as "a consequence of the greater importance of the distinction between subject and object with respect to animate

be directly suffixedwith the patient suffixes, but do occur with the specifying and agentmarking suffixes. Thus personal nouns are inherently animate and human, like pronouns, kinship terms, and proper names, but are not inherently specific (or natural agents, it would seem). Proper names, kinship terms, and pronouns are all prime candidates for agentive roles, but must be explicitly marked for the patient function, the unmarked form of all three classes being the agent. Common nouns are the most likely candidates for patients, and therefore agent function is explicitly marked by the affixation of an agent-marking suffix, the unmarked form being used for patients. Thus the unmarked order would have an Animate as agent and a Common Noun as patient: (16) ka-cil1 bu-raqal2sa-ka3. Ka'cil1 [a man'sproper name] killed3 [a] bear2. mi'pl bu-raqal2sai-a3. He' killed3 [a] bear2. But a reversal of these roles would require explicit marking: (17) bu-raqalla-1ka-ciliy2 sa-ka3. [A] bear-agent' killed3 Ka cilpatient2. bu'raqalla mi'pal saf'ka3. [A] bear-agent' killed3 him-patient2. 3. Thus, instead of a binary classification of verbs as either transitive or intransitive, at least a fivefold classification seems

nouns."

INTERNATIONAL JOURNAL OF AMERICANLINGUISTICS

VOL. 44

needed: (1) agent-taking verbs, which may also take patients such as sa'k kill or qa'ne' bite; (2) single patient-taking verbs, such as ba'kuifall, ?ecki sneeze, or da'sul misplace [something],which may only take an agent if suffixed with the causative suffix {-qa}, the reflexive {-k}, or the transitivizing secondary stem /-ki'ya-/; (3) two patienttaking verbs such as ma'ra' love, ma?as hate; (4) verbs which can occur with either a patient or an agent depending on the speaker's perception of the presence or absence of participant control, such as ce'xel slip/slide; (5) a small class of verbs of location and directed motion which take the agent form of pronouns, kinship terms, and proper names, but the patient forms of common nouns and personal nouns, and indicate protagonist number suppletively (this class includes verbs like ka'/na'ph6 sit, dwell, wa l/phi'lal go, walk, travel, k6phki/phu aphki stand up [from a nonstandingposition]). The first class of verbs could be called transitive, and the fifth class, intransitive, but classes 2 through 4 cannot be squeezed into this dichotomy without distortion. Moreover, both the first and the fifth classes of verbs have only a small number of members, albeit of high frequency of occurrence in actual speech, and are therefore not representative of the majority of verbs. At the same time, the members of these two classes of verbs cannot be determined on the basis of some universal definition of transitive-intransitive, such as "takes an object or not," but only by examining the case marking and number marking characteristically associated with each verb. If one were to talk meaningfully about a split system in Eastern Pomo, then, one would have to say that the system splits common nouns and personal nouns for which agent function must be marked, from pronouns, kinship terms, and proper

names for which patient function must be marked, and that while common nouns are unmarked with verbs of class 5, following an ergative pattern, pronouns, kinship terms, and proper names are also unmarked with members of this class, which does not conform to the ergative model. (Sandra Chung [personal communication] has suggestedthat pronouns, kinship terms, and proper names in Eastern Pomo conform to what would appear to be a stativeactive pattern [following Stephen Anderson's terminology], while common nouns follow an ergative pattern.) 4. It has been hypothesized that ergative languages (presumably whether split or not) are only morphologically ergative, but can be seen to be syntactically accusative.1l Eastern Pomo seems to provide a counterexample to this claim. Eastern Pomo distinguishes singular and plural in pronouns, kinship terms, and personal nouns. Plurality is cross-referenced in the verb in at least two ways. The plural suffixes, {-yaki-} indicating that more than one individual is involved and {-ma-} the extentive plural suffix indicating that more than one individual is involved, each acting individually in an extended action or state, may be affixed, or, as was mentioned earlier, suppletive stems are paired to distinguish plurality of patients for a number of verbs of high frequency in actual speech. Although the two plural suffixes have previously been described as distinguishingplural subjects,12they in fact reflect only plural agents and only occur affixed to agent-taking verbs of class 1 or class 4 (when these are taking agents): (17) mi'pl mi'pal2 duyka3. He' instructed/commanded3 him2.
be-khl mi-pal2 duykakiya3.
11
12

See Anderson (n. 6 above). Kroeber, p. 340; and McLendon, pp. 82 and

85.

NO. 1

ERGATIVITYIN EASTERNPOMO

him2. They' instructed/commanded3 sa'ka3. mi'pl mi'pal2 Hel killed3 him2. be-khl mi-pal2 sakakiya3. Theyl killed3 him2. Many of these verbs, such as kill, also cross-reference the plurality of the patient suppletively: (18) be'khl be-kal2 du-leyakiya3. They' killed3 them2. A few single patient-taking verbs crossreference the plurality of the patient suppletively, but most patient-taking verbs do not cross-reference the plurality of the patient in the verb at all. Thus: (19) mi'pall qa-lal-maya2. He' got sick2. be'kall qa'lal-leya2. Theyl got sick2. wil sakh62so'yo6ya3. II got cramp3[in] leg2. be-kall sa.kh62soy6oya3. They' got cramp3[in] leg2. wi1 siphikh-maya2. I1 choked2. be'kall siphukh maya2. They' choked2. Verbs which can occur with either patients or agents, depending on the speaker's perception of the presence or absence of protagonist control, crossreference the plurality of the agent with the appropriate one of the two plural suffixes, but do not cross-reference the plurality of the patient. Thus:
(20)
be'khl ce'xelkhma
2

and personal nouns. Members of this class all have suppletive stems which indicate that plural protagonists are involved. A typically Animate argument associated with such a verb, therefore, is both "in control" and "undergoing the activity." (21) mi'pi kaluhuya2. He' went home2.
be-khl kailphi l?ya2.

They' went home2. mi-pl ka2. He's sitting2.

They' are sliding2. be-kall ce-xelka2. They' slipped2. Finally, the small class of verbs of location and direct motion (which includes verbs such as go, come, run, dwell, and be located) always take the unmarked agent form of Animates (pronouns, kinship terms, and proper names), but also the unmarked patient form of common nouns

They're' sitting2. Thus plurality of protagonists is marked ergatively, with affixes indicating plurality of agents, but suppletion or no change indicating plurality of patients (which includes the nonverb argument associated with class-5 verbs of directed motion and location). Finally, Eastern Pomo, like many Hokan languages, has an elaborate system of switch-referencing suffixes (in fact only Kashaya Pomo has more as far as I know). There are four pairs of suffixes, given in table 3, which simultaneously (1) mark the clause suffixed as embedded in a matrix clause, (2) distinguish categories of sequentiality, prior necessity, or simultaneity, and (3) indicate whether coreferentiality exists between a nonverb argument in each clause. The examples in (26) and (27) illustrate all three functions. The suffix /-y/ in (26) below signals that the nonverb arguments of these two one-place verbs are coreferential, and that hence the nonverb argument of the second, matrix clause is routinely deleted. (26) hai' kaluhuy2, si'ma' merqaki'hi3. I1 went home and then2went to bed3. (27) ha"1kaluhuqan2,mi p3 merqaki-hi4. I1 went home and then2 he3 went to bed4. This coreferentiality has usually been described in terms of the notion "subject,"

JOURNAL OF AMERICANLINGUISTICS INTERNATIONAL TABLE 3

VOL.44

Action of verb suffixed precedes in time that of main verb................................. Action of suffixedverb (1) explains,justifiesthat of main verb, (2) is simultaneouswith that of main verb..................................... Action of suffixedverbis priorto and a prerequisite for the realizationof the action expressedby the main verb................................ Action of main verbcontinuesover same period or begins with time specifiedby suffixedverb.....
1 Only meaning (1) applies.

{-iy} {-in} { i

{-qan} {-sa}1 {-phila} {-iday}

{-bya}

but sentences such as those in (28) below occur in which a patient-takingverb occurs in one clause and an agent-taking verb occurs in the other, and the switchreference, noncoreferential suffix from column II is used, even though the pronoun occurring in both clauses actually refers to the same individual in reality.13 (28) haPl xai qakkiqan2,wi3 qa-lal tala4. I1 took a bath, so2 13 got sick4. ha3 khuyhi4 wil q4alalmaqan2, q6yuhu'5. I1 got sick, that's why2 P3 didn't4 come5. Example (29) illustrates how the system of switch-referencingsuffixesinteracts with the classification of verbs, to keep track of

who is doing what to whom in a complex sentence that at first glance strikes most English speakers as quite opaque. One must keep in mind that the second and third verbs, mu'tut.ki curl up and mu'dal die, are single patient-taking verbs of class 2. (29) mi'pall ki2 k6xqan3mu'titkiy4 him he shot SR curl up COR mu'dala4. die Hel shot3 (+switch-reference {-qan}) him' [and he2] curled up4 (+coreferential{-iy}) [and he2] died5.

Example (29) presumably should be derived from underlying (30), with routine equi-deletion of the second and third 13 Such sentences normally occur in the most occurrenceof the coreferentialthird-person spontaneous and natural settings, such as narra- masculine singular control pronominal tives and conversations.In elicitingfrom English, patient mi'pal,: however, Eastern Pomo speakers sometimes use in such (30) mi'pal, khi2k6xqan mi'pal, the coreferential suffix of each pair sentences, presumablyunder the bilingual presmu'titkiy mi'pall mu'dala.
sure of translating the English pronouns which are coreferential. In such cases, Eastern Pomo speakers will also accept the sentence with the switch-referencesuffix of the pair substituted. This variation is not possible when the noun argumentsare not coreferentialin reality (as in: When I got sick, he ran away) or when they are clearlycoreferentialwith respectto case as well as reality, that is, when both verbs are agent taking or both verbs are single patient taking (as in: When he gave her the book, he kissed her).

5. A definition of ergativity tied to the notion of transitivity and associated with the distinction of subjects and objects does not seem, then, to account simply for the Eastern Pomo facts. Recognition of an agent-patient distinction (differentially marked in the inherently specific and human classes of pronouns, kinship terms,

NO. 1

ERGATIVITY EASTERNPOMO IN

and proper names versus the noninherently specific and human class of common nouns), which interacts with a feature of protagonist control characterizing verbs (and cuts across traditional categorizations of verbs based on transitivity) seems to account simply for a range of morphological and syntactic phenomena. One thing at least seems clear: Eastern Pomo is not an accusative language. Neither is it a uniquely aberrant variety of speech as I once feared. Rather, I would like to suggest that Eastern Pomo is typical of a variety of languages, different from both

classically ergative languages like Basque and accusative languages like English, which are probably a great deal more common than current descriptions of languages might lead one to suspect, and which could be productively studied in much greater detail. In addition, Eastern Pomo provides still another example of the remarkable degree to which semantic properties which frequently are only indirectly inferable for languages like English are directly reflected in both the surface morphology and syntax of Native American languages.