Two Ergativities and Their Cultural Correlates

Two Ergativities and Their Cultural Correlates proposes a distinction between two types of languages with ergative structures: a group which includes languages partially ergative from the Jê family, and another group of languages constituted mainly by languages also partially ergative from the Tupí-Guaraní family. We discuss the cultural correspondences of these two types of ergative structures, based on observations of various Amazonian indigenous groups: some Jê groups, on the one hand, and various Tupí-Guaraní groups, on the other hand, among which, the Kawahíwa, Kayabí and Araweté, whose cultures will be focused.


Introduction
Ergativity seems to me a trickster, playing with our notions of agency, and forcing us to question our assumptions about who is acting in solitary activities, or activities without an object -walking, thinking, loafing, wandering in the shadows. And it raises the question of the activity of the object in acts directed at others.
What is an ergative structure -or more precisely, an ergative-absolutive structure?
The linguistic definition is simple: it is a syntactical structure in which the subject of an intransitive verb takes the same form as the object of a transitive verb -they both take the "absolutive" case. And the subject of the transitive verb takes a distinct form, called the "ergative" case. The great expert on ergativity, R. M. W. Dixon, designates each of these three positions with a letter proper to it: the subject of a transitive verb is A, for "agent;" the object is "O"; and the subject of an intransitive verb is S. "Generally," Dixon informs us, "the case that includes S," the subject of the intransitive verb -here the absolutive case -"is not grammatically marked." In ergative structures, it is usually the ergative case, the subject of the transitive verb, that is marked with a distinctive form. This is a very economical structure. Only two forms really have to be distinguished from each other: the subject and the object of a transitive verb, to know who is doing what to whom. The intransitive verb only has one argument, so that doesn't need to be distinguished from any other. So one can then just mark, grammatically, either the object of the transitive verb, making it an accusative case in a nominative-accusative structure; or the subject of the transitive verb, which is then the ergative case. Thus we have the two structures, the ergative and nominative-accusative, distinguished primarily by the treatment of the subject and object of a transitive verb -the intransitive subject does not need to be marked. If the transitive subject alone is marked, it is in the ergative case; the unmarked object and unmarked intransitive subject are absolutive.
That is the way it is in Jê languages. It is not so simple in Kawahíwa, as we see below. The ergative structure appears in many languages of Native American cultures, including many Amazonian language families -in Karíb languages (Franchetto 1990, Payne 1990, Derbyshire 1999, Arawák, many Jê languages (Urban 1988, Rodrigues 1999:193-194, Cabral, Rodrigues and Costa 2003, and Tupí, among others (Aikhenvald and Dixon 1999:348, 366). In some Karíb languages, the ergative structure constitutes the principal verbal construction of the language; but it more commonly appears as an alternative to a nominative-accusative structure -a situation referred to as "partial ergativity," or "split ergativity." In the end, what difference does it make? What is the difference whether we mark the transitive subject, leaving the intransitive subject taking the same form as the object of a transitive verb, or if we distinguish the object of the transitive verb with a separate form, leaving the other two positions (subject of the intransitive verb and of the transitive verb) identical with each other, as in familiar European languages?
That the ergative form does convey a meaning was demonstrated by Dixon, when he dedicated his book "to Sasha, the only ergative woman." Still, he did not specify what that meaning was. (I only disagree that she is the only one; I dedicate this work to another ergative woman, Lúcia).
For us, it is quite logical that the subject of the verb should take the same form, whether the verb is transitive or intransitive: it should be nominative. When a subject does something -whether he hits someone else, writes a poem, thinks, or just walks around -this subject is the agent of what he does.
The ergative structure puts all of this in question. Putting the subject of an intransitive verb together with the object of a transitive verb in the same form leads us to a paradox: It suggests that the subject of an intransitive verb acts in the same way as the object of a transitive verb. The subject of the intransitive verb with a form parallel to that of the object of a transitive one, seems (as Lacan says) subjected to the act of the verb: I don't think; somethingthe other -thinks me, or makes me walk around. Jacques Lacan could very well make use of an absolutive form when he speaks of the way that we are "spoken" by our unconscious: "ça me parle," "It speaks me." Or else, another way of thinking about this equivalence is that the object of a transitive verb participates actively in the action of which it is the object, just as the subject of an intransitive verb does. At the time I was beginning to write this work, a report came out in the Chicago newspaper on a court case that all the papers found disturbing. A man was tried for shooting another man in a bar brawl. The man he shot was left seriously wounded, hospitalized with the lower part of his body paralyzed. The shooter was accused of assault and battery, and condemned to many years in jail. After twenty years, he gets out of prison; and a few years later, the victim ends up dying. His doctors determine that his death was a result of the wounds received in the fight. In this situation, the aggressor is brought to trial once more, this time accused of a homicide. "Isn't this double jeopardy?" the newspapers ask. No, respond the lawyers, because now it is a different crime. The crime, which before was simple assault and battery, now with the death of the victim has become a case of murder. It is the condition of the victim -the object of the act -that determines what the act is, whether it is a simple assault and battery or murder. The object of the act participates determinatively in defining what the act is. And it participates in an intransitive way: he lives, or dies, or just suffers. This is a situation whose logic fits better with the logic underlying an ergative absolutive structure than with nominative-accusative logic.

Two Ergativities
There are important differences in the ways that ergative structures are composed in different Amazonian languages. For the most part, Jê languages conform with Dixon's dictum that it is generally the ergative case that is marked. In the Jê languages in which I have seen an ergative construction described, the ergative case is marked with a postpositional particle: tõ in Xokléng (Urban 1985:166), te in Timbíra and Maxakalí (Rodrigues 1999:193-4), je in Xikrín (Cabral, Rodrigues and Costa 2003:26-27). In one instance, in Kipeá of the Karirí family (Macro-Jê phylum, Rodrigues 1999:194), it is marked with the prepositional particle no.
The Kawahíwa ergative structure, on the other hand -as in other Tupí languages -does not conform to Dixon's dictum. In these languages, the formally marked case is the absolutive case, in both its manifestations: the subject of the intransitive verb ("S" in Dixon's algebra) and the object of the transitive verb ("O").
Kawahíwa, like the other Tupí-Guaraní languages that have an ergative construction, is partially ergative, or "split ergative." In Kawahíwa and other Tupí languages, there is not marking of cases by a particle placed before or after the subject or object. In Kawahíwa, the marking of the cases is purely positional.
Kawahíwa verbal constructions follow a pattern that is doubtless quite familiar to linguists working with Tupí languages -similar, for example, to the Kamaiurá verbal constructions described by Lucy Seki (1990). In Kawahíwa, as in Kamaiurá, there are three series of person markers that are placed before the verb-stem (Betts 1981:17, reorg. to conform with the charts in Seki 1990:369, tab.1). (The two "portmanteau" prefixes oro-and opo-also occurr, but will not be discussed here.)

Prefixes and person markers [personal clitics?]
There are three verbal constructions that make use of these three different series of person markers. In "verbal construction 1," which uses the prefixes of series 1 as its person markers for the subject of the verb, the subject and object are also designated as pronouns generally following the verb; the order is VSO. With an intransitive verb, the order is VS. Thus, both the personmarking prefix and the postverbal pronoun designating the subject occupy the same position with respect to the verb in both the transitive verb and the intransitive: -hó "to go" ahó ji "I go" -juká "to kill." ajuká ji ga "I kill him." Thus, both in terms of the prefix marking of the subject and the postverbal pronoun that reinforces the marking of the subject, the subject marking is the same in transitive and intransitive verbs so this verb construction is nominative-accusative.
In verbal construction 2, marked by an -i suffixed to the verb, the person marking is different. Instead of being marked by verb prefixes, the persons are marked by pronouns which occur just before the verb stem -pronouns of series 2 in the table of person markers.
When the verb is intransitive, the pronoun preceding the verb stem is the subject: kiro ji hoi "now I am going" (the most common form of farewell).
But when the verb is transitive, it is the object of the verb that precedes the verb-stem, and the subject follows the verb, a peripheral position: oro ga jukai ji "so then I killed him." If you want to clarify who is the third person referred to, that person's name may be inserted following the phrase: oro ga jukai ji, Itariano'ga.
So the subject of the intransitive verb immediately precedes the verb stem, as does the object of the transitive verb. The subject of the intransitive verb is treated in the same way, positionally, as the object of the transitive verbwhich is the definition of the absolutive case. Thus, it is the absolutive case that is (positionally) marked here, not the ergative case, which assumes a peripheral position following the verb.
Verbal construction 3 is the "descriptive," which was very well characterized by Lucy Seki (1990) in her article on the subject in Kamayurá. The descriptive is a construction that often functions as an adjective functions in Portuguese or English. These verbs, all intransitive, take the third series of person markers -which are the same pronouns as in the second series except for the third person, which uses the relational prefix i-. Thus: ikatú "he/she is good-looking, pretty" nde katú "you are good-looking, pretty" For clarification (for example, which of the four categories of third person is intended), or for emphasis, the subject is regularly specified immediately after the verb as a pronoun of the second series: ikatú ga "he is good-looking" ikatú hẽ "she is pretty" ikatú nga "they are good-looking" nde katú nde "you are pretty" ipají po ga "he probably has shamanic power" For pronouns, the post-verbal position is more peripheral than the preverbal. The post-verbal position is used for clarification and emphasis, hence it is less obligatory than the person-markers that occupy the pre-verbal position. In verb construction 2, the subject of the transitive verb falls in this peripheral position; it is clearly the absolutive case that is the positionally marked one.
Thus, the Tupí ergative-absolutive structure constitutes a type of ergativity distinct from that of the Jê, a type that goes contrary to Dixon's dictum that the formally marked case is generally the ergative case. Perhaps we can designate this structure the absolutive-ergative subtype of the ergativeabsolutive structure.

Cultural Correspondences
Are there any implicit attitudes embedded in an absolutive-ergative structure? And what are the implications of these two variants of the ergative structure?
Here, doubtless, someone will object: a language does not determine the attitude of the speaker. To the contrary, Edward Sapir said in his writings on language that language is an instrument of expression, equally capable (or equally incapable) of articulating any idea. "Language is an essentially perfect means of expression and communication among every known people" (Sapir 1949(Sapir [1933:7).
How can we speak of "implicit attitudes" in the basic structure of any language?
Begging pardon of Whorf, I agree that language does not determine thought -at least, not always. Sometimes, as Vygotsky shows, the structure of the language molds a thought like clay in the hands of the potter; but often it is the thought that a speaker is trying to express that determines the use of the language. Nonetheless, a language is the vehicle of the culture of any group of men (and women), and the means of social interaction among the members of a specific society. The structures evolved in the language of that society will necessarily be those that most easily express the concepts of relationship that conform to the presuppositions of life in that group, and with the values that govern their life.
It is also important to take into account that the majority of languages that have ergative structures, including the Tupí-Guaraní and Jê languages, are partially ergative; beside the ergative structure there is also a nominative-accusative structure. When a speaker in the language chooses one or the other structure to use to compose his discourse, he is saying something with that choice. We ask why many indigenous languages have this structure; one can equally ask why it is that Portuguese or English do not have the option to use an ergative structure? What are the communicative possibilities that we lose with this lack? 2 The differences between a nominative-accusative structure and an ergative-absolutive structure are multiple, and it is not easy to determine what are the influences they exert on the mode of thinking of speakers of a language utilizing this structure, nor to determine what is communicated with the choice of one of these two structures to express one's thoughts in. Furthermore, the differences between the Jê structure in which the ergative case is marked with a special particle, and the Tupí system where the absolutive case is positionally marked, are even more subtle.
Given all of this, what we are looking for is not something in Tupí-Guaraní thought that is determined by the absolutive-ergative structure, but rather, what ideas or assumptions are facilitated in their expression by such a structure? Or what is communicated by the choice of that way of constructing what one is saying, when one uses that structure rather than framing your thought in a nominative-accusative phrase? Let's hazard a few observations. To begin with, we note that the ergative structure isolates the subject of the transitive verb as an "ergative" case, and so highlights the activity of someone who acts on someone else or on something -an act which affects another person, a concrete object, or even an idea. When the ergative case is grammatically marked, that heightens the emphasis on the person who acts on someone or something else, who dominates. This person is marked as an agent.
This privileging of the act of dominating can perhaps be seen in the intense political competition which is described in the ethnographies of the Shavante or the Kayapó. Some who live and work with them have noted a certain prizing of an assertive attitude on the part of Shavante and Kayapó men. This privileging of dominating may be one thing that makes the Shavante and Kayapó so effective in indigenous politics, occupying a central role in the leadership of protest against being dominated and subjugated by the white political structure, and taken in by the perfidy of the white man.
There is another side of ergativity, the inverse side. The absolutive case is the opposite of the ergative -the subject of the transitive as agent; but it is also contradictory in itself. In what resists the governing dominance, as Greg Urban (1991) shows in his analysis of the image of the rock in the Xokléng myth -the rock that does not let itself be penetrated by the beak of the woodpecker, nor by the arrow of the warrior. Still, this case embodies a certain contradiction in itself, since it brings together the object of the transitive verb -that which is acted on -with the subject of an intrasitive one. And in Tupí-Guaraní ergativity -in Kawahíwa ergativity -what is marked is this contradictory side, this combination of the object of some act upon it with the subject of an action that has no object. And the subject of the descriptive verb is marked in almost the same manner, with the same pronouns preceding the verb. 3 The absolutive case combines the subject of an act that does not do anything to anyone, or of a simple state of being, with the object which has something imposed on it by an active agent -the object of someone else's act. It is this contradiction to which its syntactical marking calls attention. In thus bringing together two different positions, each position influences the connotation of the other: it suggests that the subject of an intransitive act or state may be having something imposed on it, something that one suffers being done to one -as one says "it occurs to me" when a thought suddenly comes into one's mind, as if one was not the agent of the thought: I didn't think that, it occurred to me.
Or, on the other hand, the object of the action of a transitive verb pulls toward a sense of participating in the act of which it is the object, in the way that the subject of an intransitive verb participates in the act he is engaged in by himself.
In coming together in the same syntactical place, the meaning of S, the subject of an intransitive verb, and of O, the object of a transitive act, come closer together. The victim of the warrior-killer, the one who is killed by the warrior, participates in the act by dying -or by not dying, if he escapes. Ka'gwýri-pe ji hoi "I go into the forest," oro ji jukái nde "then me killest thou" -the "I" of "I go to the forest" occupies the same syntactical absolutive slot, before the verb, as the "me" of "me killest thou." Two Tupí-Guaraní groups quite distant from each other -the Kayabí who come from the Telles-Pires River, a tributary of the upper Tapajós, and the Araweté of the lower Xingú -both share a kind of song which men sing on certain ceremonial occasions: a song, called "Jawosí" among the Kayabí, that glorifies an act of, typically, the killing of an enemy. The song may be inherited from the singer's ancestor, and may commemorate a war exploit of the ancestor, but sung as if it were the singer's act and referring to a moment in the life of the singer. But what is important to note is that the song is always sung from the point of view of the victim, as if it were the victim singing it. The words are the words of the enemy, telling about his encounter with (the ancestor of) the singer: he recounts the fear he feels of the great warrior, and his grief anticipating his imminent death. Among the Araweté, the singer says the song was taught to him by the spirit of the dead enemy, who is the true composer (Viveiros de Castro 1986:585-597, 1992:242-245, Oakdale 2005. So, it is the warrior's victim who determines the significance of his act of killing, who defines it as a triumph. In the end, it is the object of the action that confirms the glory of the executioner in his victory: "the enemy's spirit," says Viveiros de Castro, "which is always 'with' or 'in' his killer -hopǐ'hã nehe -goes up to the sky with him when he dies" (Viveiros de Castro 1986:595). He joins his killer, or becomes an integral part of the killer, and goes up to the sky world joined to him in the afterlife, the two "fused in a single figure, the killer-dead enemy, immune to being eaten by the Maǐ, the gods" (ibid.: 578).
Not only after death, but also during his life the killer remains identified with the enemy he killed, often (among the Tupínambá as well as the Araweté) assuming the name of the enemy he had killed.

Identification with the object of the act
Among the Parintintín, speakers of Kawahíwa, the pajés (shamans) in trance sing the arrival of the spirits invoked in the curing rite: in fact it is the spirits who are singing, announcing their arrival through the voice of the pajé. Similarly, the Araweté pajés sing the songs of the recently deceased, and the songs of the Maĭ (the gods) themselves. It is the object that enters into the pajé; it is the other who controls him -and who exists through him.
"Pajé," the Tupí word for "shaman," comes from the Tupínambá descriptive ipajé, which in Tupínambá, as in Kawahíwa, means "possessed of shamanic power." I translate this absolutive expression in the passive voice because the Parintintín pajé is literally "possessed" by shamanic power. He is born as an avatar of the spirit that will serve as his rupgwára, his familiar spirit -and more than this, as his alter-ego.
When a new pajé is about to be born, a spirit -the spirit of an animal, of a bird, or of one of the Celestials, the yvága'nga, "Sky People" -appears in the dreams of an old pajé and asks to be born. The pajé, in the dream indicates to the spirit an appropriately expectant woman, a woman a few months from giving birth, for the spirit to enter into. When the child is born, this child will be the new pajé, the jihúva'ga, "born one," of the pajé whose dream directed the spirit to that woman. And the spirit who entered into the woman, who was born in the child (or who was born as the child) will be the new pajé's rupigwára -when the child grows up and goes through apprenticeship to become the new pajé.
The spirit who entered the woman, then, is the child who was born, and is the potential for that child to become the pajé, and is at the same time his rupigwára, the other who will appear in his dreams to tell him the future, who will help the pajé in his acts of pajelança, his acts of shamanism. And it is this spirit, this rupigwára, that is the pajé's soul that travels to the sky to visit the spirits of the cosmos and call them to the tokáia to help in the cure of the pajé's patients. Thus, this rupigwára enters into the pajé as his shamanic power.
These two observations, from three Tupí-Guaraní cultures, support Eduardo Viveiros de Castro's conclusion that the Tupí-Guaraní peoples have a fundamental inclination to identification with the other -or, as he says, "become the other." Furthermore, some of the complexes that I have just described were the observations that were the foundation for his characterization of Tupí culture, as having this Tupí-Guaraní propensity for "becoming the other."

Concluding remarks
In my experience, the facility with which the Parintintín adopt the manners and practices of the whites, and integrate themselves into regional society, well confirm this generalization of Eduardo Viveiros de Castro. Paradoxically, because in thus "becoming Brasilian" -"becoming the other" -they show themselves still fundamentally Tupí, fundamentally Kawahíwa. Or "Parintintín," the name given to them by their ancient Mundurukú enemies, but which they have now adopted as their own.
And this paradoxical identity reflects the paradoxes of this ergativeabsolutive form -above all when the absolutive is the highlighted case.