ASSIDUE Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt Nr. 20 Hocak as an active/inactive language Why there is no passive/antipassive Johannes Helmbrecht February 2006 ISSN 1612-0612 Erfurt Seminar für Sprachwissenschaft der Universität Impressum: Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt Seminar für Sprachwissenschaft Philosophische Fakultät Universität D -99105 Erfurt Herausgeber: Prof. Dr. Christian Lehmann © bei den Autoren ISSN 1612-0612 Table of contents 1. Typological features of agentive languages_______________________________________ 4 1.1. The Hocak language and sources of data __________________________________________ 5 1.2. Some typological characteristics of Hocak _________________________________________ 5 2. Verb classification and the marking of Actor and Undergoer ________________________ 6 2.1. Intransitive Active verbs________________________________________________________ 7 2.2. Intransitive inactive verbs -full pronominal paradigm. ______________________________ 8 2.3. Intransitive inactive verbs -restricted pronominal paradigm _________________________ 9 2.4. Conclusions I _________________________________________________________________ 9 2.5. Transitive verbs______________________________________________________________ 10 2.6. Transitive experiencer verbs ___________________________________________________ 11 2.7. Conclusions II _______________________________________________________________ 12 3. The SBJ.3PL marker -ire as a passive marker? __________________________________ 12 3.1. Grammatical properties of a canonical passive construction _________________________ 12 3.2. Grammatical properties and use of 3PL -ire in Hocak ______________________________ 13 3.3. Conclusions III ______________________________________________________________ 15 4. The OBJ.3PL pronominal prefix wa-___________________________________________ 15 4.1. As a 3PL marker of P _________________________________________________________ 15 4.2. As an antipassive marker? _____________________________________________________ 16 4.3. As a nominalizer _____________________________________________________________ 17 4.4. Conclusions IV_______________________________________________________________ 17 5. References: _______________________________________________________________ 18 1. Typological features of agentive languages Languages which are classified as having a semantically-based alignment system have in common that the argument NPs in the intransitive and transitive clause are morphosyntactically coded predominantly according to semantic properties either of the event expressed by the clause, or of the participants of the event expressed by the NPs of the clause. Aspectual distinctions such as stative vs. dynamic play a role in the first case. Hence, these alignment systems are called active/ stative or stative-active, or the like. Agentivity and control properties of the participants involved in the event play a role in the second case. Therefore, they are called active/ inactive or simply agentive languages; the latter term is suggested by the editors of this volume and will be used throughout this paper. Agentive languages belong to the second group of semantically aligned participant encoding languages. The peculiarity of these languages is that they show a differential marking of the intransitive argument according to the agent/ control properties of the participant thus reflecting a division of intransitive verbs in active and inactive verbs. These semantic principles also determine the coding of the transitive A and the transitive P argument (following Comrie's abbreviations for these basic grammatical relations) in the clause. Since it is rather the semantic properties of A and P that determine their encoding, these languages also have been called role-dominated languages (cf. Foley & Van Valin 1984). Further, the semantic role of the arguments is also the crucial property in various syntactic processes in these languages. For instance, equi-NP deletions are rather controlled by the actor than by a more generalized grammatical relation subject. Instead of a grammatical relation subject, Foley & Van Valin (1984), and Van Valin & Lapolla (1997) speak of a Semantic Pivot. Another observation with regard to agentive languages is that we do not find passives in this type of languages. A passive clause is an alternative construction to an active clause which allows a) promoting a pragmatically important participant from the object function in the active clause to the syntactically prominent subject position in the passive clause, and b) de-ranking the original subject participant either to an oblique function or c) to suppress the subject participant completely. Such morphosyntactic operations are not available in agentive languages. If this were the case, there were no dominant semantic determination of the marking of the transitive and intransitive actor and undergoer. The existence of a well-established passive construction in a certain language is almost the same as to say that there are grammatical relations such as subject and direct object in this language. The functional background or motivation behind an active/passive distinction in a language with grammatical relations is operative in an agentive language such as Hocak, too. It is one of the goals of this paper to demonstrate this with data from Hocak, a North American Indian language of the Siouan family. First, some typological and grammatical properties of Hocak are presented in §1.2 through §2 which outline the specific agentive properties of Hocak. In §3 and §4, the semantics and the distribution of the third person plural markers are discussed demonstrating how they can be used to suppress the subject or object participant without being a passive construction. These third plural pronominals have indefinite and impersonal readings rendering the actor or undergoer of the clause invisible. In historical linguistics, it is well established that impersonal readings of third person plural pronouns may eventually develop to passive markers. It will be shown that this is not the case in Hocak. However, an incipient stage of such a development can be observed in closely related Lakota. 4 Johannes Helmbrecht, Hocak as an active/inactive language 5 1.1. The Hocak language and sources of data The Hocak language belongs to the Mississippi-Valley group of the Siouan language family closely related to Chiwere (Otoe, Missouri, Iowa), the Dakotan (Lakhota, Assiniboine, Stoney), and Dhegiha (Omaha-Ponca, Osage, Kansa, Quapaw) languages. Its alternative name "Winnebago" was used by the neighboring tribes of the Algonquian family and taken over by linguist and anthropologists. Since Hocak is the self-denomination of the tribe, it will be used throughout this paper. Hocak is a highly endangered language still spoken in Wisconsin and the Winnebago Reservation in Nebraska by approximately 200 elders. None of them is younger than 50 years. The data for the present survey are taken from fieldnotes collected by the author over the last eight years during several trips to Wisconsin. An important additional source for the present survey is the quite extensive text corpus of Hocak and the Hocak lexicon data base created by the author and the research assistants of the DOBES project "Documentation of the Hocak Language" funded by the Volkswagen Foundation (Germany) since 2003 (cf. for instance Helmbrecht (in prep.), 1 Helmbrecht et al. (in prep.)). Other sources are Lipkind (1946), Susman (1943), White Eagle (1988), Miner (1992). 1.2. Some typological characteristics of Hocak The following typological characteristics are relevant for the coding of clausal participants in Hocak. First, there is no case marking on nouns, there are no adpositions, and word order does play only a marginal role in the disambiguation of the semantic functions of the participants. These properties are illustrated in E 1. 2 E 1 CO005 Johnga haasti k weehí ánaga ciinak eja wawirúwina /John-ga haastikweehí ánaga ciinakeja wa-wirúwi-na/ J.-PROP blueberries gather and town there 3PL.U-sell-DECL 'John picked blueberries and sold them in town.' The first part of the complex sentence in E 1 contains two NPs, John and haastik 'blueberries', which do not show any semantic function indicating morphological sign. Even the order of both NPs is not obligatory, although A-U-V is the preferred order in transitive clauses with two nominals. The transitive A and U are cross-referenced on the verb by means of pronominal prefixes unless they are third person singular participants that are always zero (compare the paradigms in E 4 and E 7 below). This is also the case here with regard to A John. The transitive U haastik 'blueberries' is plural, but indefinite. Therefore, no pronominal reflex can be found on the verb in the first clause, but an 3PL.U marker wa-in the second one. While there is almost no nominal morphology, Hocak like other North American Indian languages, shows a rich inventory of verbal morphology. There are, for instance, two series of pronominal affixes indicating actor and undergoer arguments (see below), which are the basis for 1 Further information on the documentation of the Hocak language and the DOBES project at the University of Erfurt can be obtained from the project website: http://www.uni-erfurt.de/sprachwissenschaft/Vgl_SW/ Hocank/index_ frames.html. 2 Abbreviations used for the grammatical glossing of the examples are the following: A = actor, U = undergoer, SBJ = subject, OBJ = direct object, SG = singular, PL = plural, DU = dual, 1, 2, 3 = first, second, third person, 1&2 = first person acts on second person, I = inclusive, E = exclusive, DECL = declarative, ST = part of stem, PROP= proper name marker, NP = noun phrase, INST = instrumental prefixes, APPL.INESS = locative applicative (inessive 'in'), PASS = passive, TOP = topic, FUT = future, A = transitive Agent, P = transitive Patient, S = intransitive Subject the active/inactive alignment and the corresponding classification of intransitive verbs. These two series of pronominal affixes allow cross-referencing up to tree arguments of the verb, one actor and two undergoes arguments such as patient and recipient; see E 2a-b. E2 a. wooniik'uuna /wa-ho-nii-k'u-na/ 3PL.U-ST-1&2-give-DECL 'I gave them to you' or 'I gave you to them' b. woínak'uuna /wa-ho-hi-ra-k'u-na/ 3PL.U-ST-1E.U-2SG.A-give-DECL 'You gave them to me' or 'you gave me to them' If there are two undergoer participants expressed by two pronouns of the undergoer series, the assignment of undergoer roles to these pronouns is context dependent. Hocak has a wealth of valence increasing derivations such as causativization, transitivization (by means of eight instrumental prefixes), two locative applicatives, an instrumental applicative, and a benefactive applicative. However, there are almost no valence decreasing devices except perhaps reflexivization. Hocak has no passive and antipassive construction as will be demonstrated in §3 and §4 below. In a way functionally similar to valence decreasing is the usage of third person pronominal affixes (wa-3PL.U/ -ire 3PL.A) as indefinite actor pronoun (3PL.A) and indefinite undergoer pronoun (3PL.U). Again, the function and usages of these forms will be discussed in §3 and §4. Hocak has two different ways of external possessor marking, one by the benefactive applicative and the other by a special reflexive possessor marker indicating that the transitive A possesses the transitive U. To conclude: Hocak is a language with strong head marking properties on the clause level. 2. Verb classification and the marking of Actor and Undergoer The Hocak plain, un-derived verbs can be divided at least into seven different classes according to the pronominal marking of their arguments. We can distinguish ditransitive, transitive, transitive experiencer, and intransitive verbs. The intransitive verbs can be further subdivided into active and inactive ones. The inactive intransitive verbs can be further subdivided into the ones that take the full pronominal paradigm and the ones that are restricted to the third person. Cf. the classification of Hocak verbs in Figure 1 6 Johannes Helmbrecht, Hocak as an active/inactive language Figure 1 Principal verb classes in Hocak Plain (non-derived) verbs Intransitive Transitive Ditransitive Inactive Active Experiencer Transitive Inactive: Inactive: 3rd person full only paradigm 2.1. Intransitive Active verbs The class of intransitive active verbs includes all verbs designating movements, body positions, actions, which have in common, that they designate an action controlled or instigated by an animate or human controller; cf. the examples in E 3. E3 híi 'to arrive, to get there' hiirá 'to bathe, to swim' naaži. 'to stand' hikšá 'to smile' hikorohó 'to get ready, to get dressed' and many more Morphologically, these verbs are distinguished from intransitive inactive verbs by requiring the actor set of person markers either of the first or of the second conjugation. The paradigm of the active intransitive pronominal prefixes of the first conjugation is given in E 4. Note that the paradigm consists of discontinuous pronominal affixes in the plural except the 3PL –ire form that is a suffix. E 4 Paradigm of the active intransitive verb šgáac 'to play' SG DU PL 1INCL hi-šgác hi-šgaj-wí 1EXCL ha-šgác ha-šgaj-wí 2 ra-šgác ra-šgaj-wí 3 Ø-šgáac3 Ø-šgaaj-íre However, there is a number of intransitive active verbs that do not really match the semantic distinction behind the active/ inactive verb classification; cf. the examples in E 5. The zero sign in the pronominal paradigms here is introduced for reasons of a better and more transparent exposition of the morphological structure of the personal inflection. The question whether there is indeed a zero or rather nothing has to be examined further. Arguments for a 3SG zero morpheme are a) that there is a corresponding non-zero 3PL form in both paradigms, and b) that the zero form always implies a 3SG anaphoric relation. However, I prefer to leave this question open here for further inspection. 4 E5 bookéwe (boákewe, boorákewe)'to stumble, to fall down' caap (hacap, racap) 'to have as kin' hokurujis (waakúrujis; horakúrujis) 'to be young, immature' naa (hana, rana) 'to sleep, to fall asleep' nihee (hanihee, ranišee) 'to be' rujís (tuujís, šurujís) 'to fall short, to fail' and a few others Up to now, we have about 250 intransitive active verbs in our database of the Hocak/ English- English/ Hocak Dictionary (cf. Helmbrecht et al. in prep.). However, this figure will probably change in the course of the future work on the dictionary. 2.2. Intransitive inactive verbs -full pronominal paradigm. The class of intransitive inactive verbs comprises all verbs designating properties such as colors, dimensions, values, and so on, as well as unintentional processes, and results of processes. Compare the examples in E 6. E6 šii'to be fat' pi i 'to be nice, to be good' naawó 'to swell' šišré 'to break, to be broken' hokižác 'to be puzzled' cóo 'to be blue/green' and many more Morphologically these verbs are distinguished from the active intransitives in that they take personal affixes of the undergoer series; cf. the paradigm in E 7. Note that there is no distinction between first and second conjugation in the undergoer series of pronominal prefixes (The Hocak/ English-English/Hocak Dictionary contains at least 200 intransitive inactive Verbs so far – this figure may rise in the future work on the dictionary). E 7 Paradigm of the inactive intransitive verb š'áak 'to be old' SG DU PL 1INCL waagá-š'ak waagá-š'ag-wi 1EXCL hi-š'ák hi-š'ag-wí 2 ni-š'ák ni-š'ag-wí 3 Ø-š'áak Ø-š'aag-íre The structure of the paradigm is exactly parallel to the one for the active intransitive verbs. There is an inclusive/ exclusive distinction with an isolated first dual inclusive (1D.I) form reminiscent to so-called minimal/ augment pronominal paradigms in Austronesian and Australian languages. We Since the pronominal prefixes are in fact infixes in many verbs, it is necessary to provide the first and second person inflected word forms in order to show where the pronouns appear in the word. For instance, boo_kéwe 'to stumble, to fall down' has the pronominal affixes between the submorphemic initial stem part boo-and the bound root –kewe. Since the internal slot for the person markers is not fully predictable, inflected word forms have to be given in the lexicon. 8 Johannes Helmbrecht, Hocak as an active/inactive language have discontinuous pronominal affixes in the plural except for the 3PL form –ire, and the 3SG is zero in both paradigms. 2.3. Intransitive inactive verbs -restricted pronominal paradigm There is a large group of intransitive verbs that should be classified as inactive intransitive verbs on semantic grounds. The peculiarity of these verbs is that they can be inflected only for either animate or inanimate third persons. Since the third person singular actor and undergoer is generally zero marked, and the third person plural suffix –ire does not distinguish between actor and undergoer, the active/ inactive distinction is morphologically neutralized in these verbs. In other words, the active/ inactive distinction in intransitive verbs exists only for verbs allowing a first and second person intransitive argument. Some examples for verbs that allow only third person subjects are given in E 8. The Hocak/English-English/Hocak Dictionary (cf. Helmbrecht et al. in prep.) contains about 50 verbs of this type. E 8 roocu. 'to become ripe (3rd person only)' túuc 'to be cooked (3rd person only)' xuuxré 'to break, to be broken (3rd person only)' and many more E 9 xuuxré 'to break, to be broken (3rd person only)' šišré 'to break, to be broken' (full paradigm) Interestingly, there are numerous lexical oppositions between intransitive inactive verbs that allowing a full personal inflection and intransitive inactive verbs that are inflectionally restricted to third persons. Such a lexical opposition is given in E 9. These lexical oppositions like šišré versus xuuxré 'to break, to be broken' (cf. E 9) reflect a classification of participants according to the empathy or animacy hierarchy. The third person participants of the inactive intransitive verbs with the reduced paradigm are mostly inanimate. 2.4. Conclusions I The split marking of the intransitive argument S among intransitive verbs is almost exclusively dependent on the actor properties of S such as agentivity and control. Aspect and aktionsarten, respectively such as dynamic versus stative, are not relevant in Hocak. The selection of pronominal affixes from the actor or undergoer series is lexicalized. There is practically no freedom for an alternative choice between actor or undergoer inflection with regard to one verb base. Such a freedom of choice has been observed occasionally in other agentive languages. An example like the one in E 10a. and E 10b is really the exception in Hocak. The former forms show the intransitive verb xgazí with pronominal infixes of the actor series, the latter the same verb with pronominal infixes of the undergoer series. The choice of pronominal affixes of the two different series results in a typical meaning difference between the two inflected words. The inflected forms in E 10a designate an action controlled or actively instigated by the participant. The inflected forms in E 10b, on the other hand, designate an experiencer situation, a feeling, which is not controlled by the participant. E 10 a. xgazí '(to) relax, to rest' xgaazí 'I relax' xganazí 'you relax' b. xgazí '(to) be relaxed, to be rested' xgaizí 'I am relaxed' xganizí 'you are relaxed' The function of passives is to promote direct objects in subject position and to demote former subjects to adjunct status. Mostly, subjects are dropped altogether. The passive construction is chosen by the speakers, if the former direct object is unusually high on the animacy/ empathy hierarchy – direct object are usually low on this scale of participant types – or if the direct object is particularly salient in the current stretch of discourse. One result of a passive operation is an intransitive verbal predicate. Since Hocak has such a large and varied group of intransitive inactive and active verbs, they can be considered to compensate for the lack of passive constructions and the lack of passive participles that are so pervasive in English. In addition, the group of inactive verbs is further subdivided in the ones taking the full paradigm and the ones taking only third persons (mostly inanimate participants). It is in particular the latter group that is often translated in English with a passive participle. Since they allow only for inanimate third persons as subject, they are equivalent to passive participles in other languages with a passive. There is a significant group of inactive verbs that look like passivized verbs. They are all ending in -re. Compare the forms in E 11a-b. –gaas is a bound lexical morpheme that does not occur independently. E 11 a. gigás 'to tear sth. (like paper) by striking' gi-gáas INST-tear b. gaasré 'to be torn, to be ripped' gaas-ré This is not a productive process, and not all inactive verbs ending in -re have a passive-like meaning. However, it can be hypothesized that this –re form once was a derivational morpheme to the effect that transitive verbs were intransitivized very much like passive participles. There are no diachronic data supporting this speculation. 2.5. Transitive verbs The vast majority of transitive verbs show some kind of morphological derivation. Plain un-derived transitive verbs are less frequent. Transitive verbs designate actions that are instigated and conducted by an intentional and controlling actor on a more or less affected or effected undergoer; cf. some examples in E 12. Morphologically, these verbs require a pronominal prefix of the actor series and one of the undergoer series, unless none of the arguments is 3SG. Since third person singular actor and undergoer is marked zero, the verbs in E 12 could also be translated as honi 'he/she hunts it', ru'a. 'he/she carries it' and so on5. The plain uninflected verbs express a complete proposition, but do not constitute a complete utterance. A declarative suffix such as -na or -šana is required usually. E 12 honi. 'to hunt sth., to look for sth.' ru'a. 'to carry sth., to lift sth.' honac 'to borrow sth.' and many more Note that there is no infinitive category in the Hocak verb. 10 Johannes Helmbrecht, Hocak as an active/inactive language A transitive verb is conjugated by a combination of pronominal affixes of the actor series and the undergoer series. This is illustrated in E 13a-c. E 13 xée ‘to bury someone’ a. hinaxé ‘you (SG) bury me’ /hi-ra-xée/ 1SG.U-2SG.A-bury b. hixáire 'they bury me' /hi-xée-ire/ 1SG.U-bury-3PL.A c. waaxé ‘I bury them’ /wa-ha-xée/ 3PL.U-1SG.A-bury There are two peculiarities in the transitive paradigm. 1) The 3PL suffix -ire which marks the 3PL subject in active and inactive intransitive verbs is used exclusively for the Actor in transitive verbs; cf. E 13b. 2) There is special pronominal prefix wa-which marks the 3PL.U in transitive verbs. This form does not occur in the conjugation of intransitive verbs at all; cf. E 13c. 3) The pattern that arises with regard to third person plural participants is of the accusative type: S/A = -ire, opposed to P= wa-. 2.6. Transitive experiencer verbs There is a small group of transitive experiencer verbs, which exhibit very specific restrictions with regard to argument encoding compared to regular transitive verbs. An experiencer situation is constituted by an human/ animate experiencer who feels or perceives something. The stimulus is the cause or source of the feeling or perception. The restrictions that can be observed with transitive experiencer verbs in Hocak deal with the stimulus participant. The stimulus participant, i.e. the entity that causes the perception or feeling or is the source thereof, can only be a lexical NP without any pronominal trace in the verb, while the experiencer is always rendered by a pronominal affix of the undergoer series. The result is that these verbs superficially look like intransitive inactive verbs, but in fact, they are transitive always requiring a stimulus NP. This type of argument frame can be illustrated with the verb téek 'to hurt'. This verb is transitive requiring a nominal argument designating the source of the pain, plus the experiencer argument that is expressed by means of an undergoer pronominal affix: 'body part hurts U'; cf. the examples in E 14a-b. The particular verb in E 14 requires a body part noun as stimulus, but no animate or human stimulus/ agent is allowed as is demonstrated in E 14b. The same is true for verbs like šuwú 'to itch' with an argument frame 'body part itches U' (cf. E 15) and naat'is 'to swell, swollen' with an argument frame 'body part is swollen to U' (cf. E 16). E 14 a. siirá hitékšana 'The foot hurts me' /sii-rá hi-téek-šana/ foot-DEF 1SG.U-hurt-DECL b. *huucrá hitékšana 'The bear hurt me' /huuc-rá hi-téek-šana/ bear-DEF 1SG.U-hurt-DECL E 15 naasúra hišúwu 'My head itches' /naasú-ra hi-šúwu/ head-DEF 1SG.U-itch E 16 huurá nait’isšana. 'My leg is swollen' /huu-rá naa-hi-t’is-šana/ leg-DEF ST-1SG.U-swell-DECL A transitive experiencer verb that is not restricted to body part nouns as stimulus is haréc 'to lack sth.' with an argument frame 'U lacks sth.' (cf. E 17a-b). E 17 a. žuurá hiirécšana 'I lack money' /žuurá ha-hi-réc-šana/ money ST-1SG.U-lack-DECL b. *žuurá šuucrá wiirecšana 'I lack them (the pennies/coins)' /žuurá šuuc-rá wa-ha-hi-rec-šana/ money red-DEF 3PL.U-ST-1SG.U-lack-DECL There is no way to get a pronominal affix cross-referencing the nominal stimulus even if this NP is pluralized as illustrated in E 17b. There is a blockage for a pronominal marking of the source/ stimulus. 2.7. Conclusions II The marking pattern for the core syntactic relations is controlled by the semantic role of the participants. Transitive actors (A) are marked by a pronominal affix of the actor series, and transitive undergoers (P) are marked by a pronominal affix of the undergoer series. This holds at least for the first and second persons. For third persons, we find an accusative marking pattern. The 3PL.A -ire marks the intransitive subject S (no matter whether we have an active or inactive verb, cf. the paradigms in E 4 and E 7 above) and the transitive actor A, while the 3PL.U marker waexclusively marks the transitive undergoer P. We can conclude that the active/ inactive alignment holds only for first and second persons. This semantic alignment type is opposed to the marking patterns for the third persons which is accusative. This alignment type split is in line with the observation made by Dahlstrom (1983) that split marking of intransitive subjects follows the person hierarchy [1<2<3], i.e. we find an active/ inactive split marking first in the first person and the second person, and then in the third person, but not vice versa. 3. The SBJ.3PL marker -ire as a passive marker? 3.1. Grammatical properties of a canonical passive construction The grammatical properties of a canonical passive construction can be summarized as follows: • Direct object is promoted to subject • Direct object (P) receives case marking of subjects • The original subject is demoted, either completely or to adjunct status • The original subject receives (if not dropped completely) oblique case marking • The transitive verb is detransitivized • Passive is overtly morphologically marked on the verb – in European languages we have passive participles used for this purpose 12 Johannes Helmbrecht, Hocak as an active/inactive language 3.2. Grammatical properties and use of 3PL -ire in Hocak The SBJ.3PL –ire can be used anaphorically to refer to a group of already mentioned individuals, but it can also be used for an unspecific and indefinite group of actors. In this case, the number distinction is neutralized. It does not matter whether the indefinite actor is a single participant or a group of participants. The impersonal use of –ire is similar to the impersonal use of English they; compare the examples in E 18a-b. E 18 a. They stopped him before he entered the construction site with his car b. John was stopped before he entered the construction site with his car In E 18a-b, two strategies for the backgrounding of the actor are shown. The a. sentence uses the 3PL pronoun as an indefinite pronoun, the b. sentence a passive construction. In both cases, it is irrelevant who had stopped John and how many they were, i.e. one or a group of individuals. It is the backgrounding of the actor that impersonal subjects and passive clauses have in common. A Hocak clause like the one in E 19 has a definite and an impersonal reading of –ire. Two translations in English are possible, one employing the impersonal usage of they, the other employing the passive construction. E 19 Johnga hojii neena /John-ga hoji-íre-na/ J.-PROP beat.up-3PL.A-DECL 'They beat John up.' Or 'John got beaten up.' Or 'Someone beat John up'. It is common knowledge in historical linguistics that passive constructions may develop out of impersonal passives, often based on an original third person plural marking. Since there is no grammaticalized passive construction in Hocak, it is worth looking whether the impersonal pronoun –ire could be analyzed as an incipient passive marker. This idea was advanced but not published by Kenneth Miner (1992). There are some examples that suggest such an analysis. A similar discussion arose with regard to Lakhota 3PL –pi (cf. Van Valin 1977:106ff), but this Lakhota form is not cognate with Hocak -ire, but rather with -wi (PL) which has a different function in Hocak. The verb cu u 'to have many, to give birth' is inflected for the 3PL.A –ire in the expression designating 'birthday', i.e. literally 'the time when someone is born'; cf. E 20a. The undergoer pronoun refers to the person who is born; hence, the actor suffix must refer to the person who gave birth which cannot be a 3PL group. The same problem of interpretation is met in the definition hocuinera 'birthday' in White Eagle's dictionary given in E 20b. Compare also the examples in E 21a-b with the verb giwé 'to sting (of bees)' which are translated by means of a passive in English. E 20 a.  honicuinera /ho-ni-cuu-ire-ra/ APPL.INESS-2SG.U-give.birth-3PL/PASS?-DEF 'your birthday (lit. 'time when you were born')' b. Haaprá hiža ejá cuuíregi /haap-rá hiža ejá cuu-íre-gi/ day-DEF INDEF there give.birth-3PL/PASS?-TOP 'The day that he/she was born' (White Eagle 1988:46) c. (Maryga) nikjágra jóop wacuuna /(Mary-ga) nikják-ra jóop wa-cuu-na/ M.-PROP child-DEF four 3PL.U-give.birth-DECL '(Mary) she has four children. ' E 21 a. heezik giwáireena /heezik giwé-ire-na/ bee sting-3PL/PASS?-DECL 'One of the bees stung him' (White Eagle 1988:14) b. ke hirarakikaranigiži heezik nigiwéirekjareena /ke hira-ra-kikara-ni-giži heezik NEG ST-2SG.A-take.of.self-NEG-if bee ni-giwé-ire-kjane-na/ 2SG.U-sting-3PL/PASS?-FUT-DECL 'If you aren't careful, you will get stung by bees.' (White Eagle 1988:14) Examples such as the ones given in E 20 and E 21 lead Miner (cf. Miner 1992) to the conclusion that -ire has a passivizing function. This is too far fetched, though. First, as said above, -ire can be used impersonally which implies that number distinctions are neutralized. If an actor is irrelevant and unimportant to the expression of an event, its number is it too. In the case of the 'birth' and 'birthday' expressions, it is the person born who is relevant, not the identity of the mother. The impersonal usage of -ire simply means that someone gave birth to the undergoer, it is not necessary to refer to a passive interpretation of -ire. Since cu u is a normal transitive verb, the actor (mother) can be brought in in the regular way; cf. E 20c. The same holds for the verb giwé 'to sting'. Again, the nature of the actor is strongly implied in this verb (i.e. there are quite specific selection restrictions for the semantics of the actor in cuu 'to give birth' and in giwé 'to sting'). And again, it is the undergoer of stinging who is more relevant to the speakers than the insects which are usually not individualized and counted. The impersonal reading of –ire seems therefore appropriate, and no passive interpretation is necessary. I assume that it is a special feature of the semantics and pragmatics of both verbs discussed here that they force a backgrounding of the actor. The Hocak way to impersonalize the actor and neutralize the number category of the actor is to use -ire. There is more evidence that –ire should not be analyzed as an incipient passive marker. It is not possible to elicit pairs of clauses, which exhibit a passivization relation. Consider the examples in E 22a-b. The first one (E 22a) has a definite and an indefinite/ impersonal reading such as 'they (already mentioned) borrowed the book', 'someone borrowed the book' and 'the book was borrowed'. E 22 a. waagáxra honacíreena /waagáx-ra honac-íre-na/ book-DEF borrow-3PL/PASS?-DECL 'They borrowed the book/ someone borrowed the book/ the book was borrowed' b. *Peterga waagáxra honacíreena '*the book was borrowed by Peter' /Peter-ga waagáx-ra honac-íre-na/ P.-PROP book-DEF borrow-3PL/PASS?-DECL If the clause E 22a. would be a passive clause, it should be possible to add an oblique actor phrase to the clause. This is not possible, there is no way to attach a noun phrase expressing the actor, cf. E 22b. There is no possibility to interpret Peter-ga as actor; the actor is always the 'third plural' or the 14 Johannes Helmbrecht, Hocak as an active/inactive language 'someone' of the pronominal suffix -ire. Further evidence against the interpretation of -ire as a passive marker: The original undergoer does not receive the coding properties of the transitive subject, i.e. it is not expressed by a pronominal affix of the actor series. So, if the borrowed book in E 22a were pluralized, the 3PL.U prefix wa-would appear indicating that the book is still a participant in P position and not promoted to A. 3.3. Conclusions III There is no passive construction in Hocak. There is no construction that allows promoting a transitive undergoer (direct object) to actor (subject). The actor of a transitive verb cannot be demoted. The actor has to be expressed obligatorily by a pronominal affix. The third person pronominal affix -ire, however, has an impersonal reading resulting in the backgrounding of the actor participant. The 3PL -ire suffix is not an incipient passive marker, since there is no way to add an oblique actor to the clause. The transitive verb remains transitive. -ire fills the actor argument slot being definite or indefinite. 4. The OBJ.3PL pronominal prefix wa4.1. As a 3PL marker of P The prefix wa-is a 3PL.U pronominal affix indicating the person value of the transitive P. Its reference is not restricted with regard to animacy, i.e. it may refer to 3PL inanimate, animate, and human referents. Its obligatory occurrence, however, depends on the definiteness of the antecedent noun phrase. If the lexical P noun phrase is definite and plural, wa-is obligatorily used. If the noun phrase is not definite, wa-needs not to occur. This rule is illustrated in E 23a-c. The first clause (E 23a.) shows that wa-is used together with the definite noun phrase wažatírera 'the cars'. Note that wa-is the only marker of plurality in this clause. Nouns are generally not marked number in Hocak. If the P noun phrase is not definite, wa-is usually dropped (cf. E 23b.). wa-is obligatory, if the P noun phrase is again marked as definite with the definite article –ra (cf. E 23c.). E 23 a. wažatírera waajáana 'He sees the cars' /wažatíre-ra wa-haja-na/ car-DEF 3PL.U-see-DECL b. wažatíre jóop hajáana 'He sees four cars' /wažatíre jóop haja-na/ car four see-DECL c. wažatírejobra waajáana 'He sees the four cars' /wažatíre-joop-ra wa-haja-na/ car-four-DEF 3PL.U-see-DECL If wa-is used with plain or derived transitive verbs, it always fills the undergoer argument slot of the verb and may in every case be interpreted as agreeing with an optional lexical P noun phrase. Many transitive verbs with wa-have two readings, a nominal one and a verbal one; cf. the examples in E 24a-d. E 24 a. wa-hoku. v.tr. 'he/she ministers them, he/she preaches to them' n. 'minister' < hooku v.tr. 'to preach to someone' b. wa-gigus v.tr. 'he/she teaches them' n. 'teacher' < gigus v.tr. 'to teach someone' c. wa-gina v.tr. 'he/she twists/ spinns them' n. 'rope, string' < gina v.tr. 'to twist sth./ to spin sth.' d. wa-k'i. v.tr. 'he/she carries them on the back' n. 'backpack' < k'ii v.tr. 'to carry sth. on the back' All four words can be used without morphosyntactic modification either as verbal predicate or as head of a noun phrase. If wahoku. and wagigusare used as nouns, i.e. with the nominal meaning, then the reference of wa-is indefinite which could be translated as 'the one who teaches them', the one who ministers them'. This means that wa-may also have an indefinite usage, which is found in particular in lexicalized words beginning with wa-and having a nominal meaning. Words like wahoku. and wagigusbrought Lipkind (1945) and others to the conclusion that wa-has also a nominalizing function (see §4.3 below). 4.2. As an antipassive marker? The fact that many Hocak words expressing nominal concepts in the English translation begin with wa-was taken as evidence for the detransitivizing function of this prefix (cf. Lipkind 1945:17). This conclusion is justified, if there is a set of (plain or derived) transitive verbs, which can be turned to intransitive verbs by means of wa-. I will argue that this is almost never the case. The few exceptions will be reviewed in the subsequent paragraphs. The transitive verbal root ruuc 'to eat sth.' (cf. E 25a) may receive the wa-prefix (cf. E 25b), deriving an intransitive active verb. That warúc is not a transitive verb with a 3PL.U can be seen from the following examples in E 25c-d. No lexical noun phrases expressing the transitive patient ('meat', 'apples') are possible here. There is no agreement relation between wa-and the P noun phrases in these cases. E 25 a. kšée ruuc-šana 'He eats an apple' apple eat-DECL b. warucna kšana 'He is eating (sitting)' /wa-ruuc-na k-šana/ 3PL.U-eat-be.sitting-DECL c. *wani warucnakšana '*he is eating meat' /wani ruuc-na. wa-k-šana/ meat 3PL.U-eat-be.sitting-DECL d. *kšee-rá warucna kšana '*he is eating apples' There are a few other cases in the Hocak lexicon which are analog to the warúc example, e.g. wa'u. 'to be' derived from 'uu'to do, to make sth.'. Another clear case is woohí (v.act.) 'to win' which is derived from hohí (v.tr.) 'to defeat someone' (woohí < wa-+hohí; the /h/ drops and the vowels assimilate). That hohí 'to defeat someone' is transitive can be demonstrated with e.g. huunahí 'you defeat me', which is not possible with woohi 'to win' which is intransitive. This list of verbs that allow an intransitivization by means of wa-is exhaustive. There are – to the best of my knowledge – no other instances of this usage of wa-. It is therefore not a productive process. If this grammatical operation were a productive process in Hocak, we had a kind of antipassive construction in Hocak. An antipassive construction in ergative languages mirrors the passive construction in accusative languages. The important properties of antipassives are: The ergative marked A of a transitive clause is promoted to an absolutive marked intransitive S, while the former absolutive marked P is demoted to adjunct status or completely suppressed. If the former P is not dropped completely, it will receive an oblique case marking. Antipassive clauses are intransitive like passive clauses. One semantic side effect of an antipassive is that the resulting construction receives a continuative or imperfective reading. It can safely be concluded that Hocak has no 16 Johannes Helmbrecht, Hocak as an active/inactive language established antipassive construction, although the few examples of an intransitivization with the 3PL marker wa-look like such a construction. 4.3. As a nominalizer It has been claimed in the literature (cf. Lipkind 1945) that wa-has a nominalizing function. To confirm and verify such a statement, it would be necessary to have a set of verbs (transitive or 6 intransitive) which may be turned into nouns, i.e. words designating nominal conceptsby attaching wa-to them. In order to qualify as nouns, the words derived by wa-should fulfill the following criteria for noun hood. a) They should not to be able to receive verbal inflection (person inflection) and hence, b) They should not be able to be used as clausal predicate without any auxiliary support. These criteria are not fulfilled in the examples in E 24a-d, e.g. by wagigus 'teacher' and wahoku. 'minister' mentioned above in E 24a-d. However, there are many words on wa-which could be analyzed as nouns fulfilling the criteria above. One of these is warocu. 'anything ripe (corn, pumpkins, vegetables)' derived from rocu. 'to be ripe, to ripen', cf. also E 26. E 26 warocu. 'anything ripe (corn, pumpkins, vegetables)' < rocu. 'to be ripe, to ripen' wacék 'young person' < céek 'to be young' etc. The deriving wa-could be glossed in this case as 'thing, something' leading to a literal translation 'the thing/something ripened'. The number of nouns which are derived in this way are not too numerous in the dictionary. It is not clear to me, whether this type of derivation is productive and to 7 what degree. However, the majority of words beginning with wa-are lexicalized to some degree no longer showing a derivational relation between the base form, i.e. the verbal root and the derived form. 4.4. Conclusions IV Besides the function of wa-as a 3PL.U marker for transitive P, there are other usages such as a detranisitivizer and a nominalizer, where wa-has an indefinite, unspecific meaning. These usages resemble to some degree the impersonal usage of -ire in that a morphosyntactically required participant is backgrounded. The function of the impersonal usage of -ire was to background the transitive actor; the function of the impersonal/ indefinite usage of wa-is to background the transitive undergoer. An antipassive construction, however, does not exist in Hocak. 6 For the problem of identifying nouns in Hocak, cf. Helmbrecht (2002b). 7 I am not talking about the derivations with the instrumental prefix wa-'by pressure, by pushing' that can easily be distinguished from the 3PL.U/ INDEF.U pronoun wa-with regard to the pronominal inflection type. Words with the instrumental prefix wa-always take the second conjugation. 5. References: Dahlstrom, Amy 1983. "Agent-patient languages and split case marking systems". In: Proceedings of the 9th Annual Meeting of the Berkeley Linguistic Society. Berkeley: BLS 9:37-46. Foley, William A. & Van Valin, Robert D. 1984. Functional syntax and universal grammar. Cambridge: Cambridge University Press (Cambridge Studies in Linguistics, 38). Helmbrecht, Johannes 2001. "Are there Adjectives in Hocak (Winnebago)?". In: Lehmann, Christian (ed.) Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt (ASSidUE) Nr. 14. Erfurt: University of Erfurt. (Available as a pdf file: http://www.unierfurt. de/sprachwissenschaft/ASSidUE/ASSidUE14.pdf ). To appear in: Lois, Ximena & Valentina Vapnarsky (eds.) 2005 Root Classes and Lexical Categories in Amerindian Languages. Bern: Lang Helmbrecht, Johannes 2002a "Argument structure of the Hocak (Winnebago) clause". In: Lehmann, Christian (ed.) Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt (ASSidUE) Nr. 4. Erfurt: University of Erfurt. To appear in International Journal of American Linguistics IJAL. (Available as a pdf file: http://www.dbthueringen. de/servlets/DerivateServlet/Derivate-1768/ASSidUE04.pdf) Helmbrecht, Johannes 2002b. "Nouns and verbs in Hocak". In: International Journal of American Linguistics 68,1:1-27. Helmbrecht, Johannes et al. in prep. Hocak/ English and English/ Hocak Dictionary. University of Erfurt. Helmbrecht, Johannes in prep. A Grammar of Hocak (Winnebago). University of Erfurt. Lipkind, William 1945. Winnebago grammar; New York; King’s Crown Press. Miner, Kenneth L. 1992. A grammar of Hocak. Unpublished ms. Lawrence: University of Kansas. Rankin, Robert L 2004. "A Diachronic Perspective on Active/Stative Alignment in Siouan." Ms. University of Kansas and Research Centre for Linguistic Typology. Susman, Amelia 1943. The accentual system of Winnebago. New York: Columbia University [Ph.D. dissertation]. Van Valin, Robert D. Jr. 1977. Aspects of Lakhota Syntax – A study of Lakhota (Teton Dakota) syntax and its implications for universal grammar. Unpubl. PhD Dissertation. University of California, Berkeley. White Eagle, Josephine P. 1988. "A lexical study of Winnebago". Cambridge: Center for Cognitive Science, MIT (Lexicon Project Working Papers, 26). 18