Complement clauses in Megrelian

zyxwv zyx zyxwvuts zyxw SlUdrO Llflgulstrco 45( I /2) Karina Vamling and Revaz Tchantouria I99 I COMPLEMENT CLAUSES IN MEGRELIAN Complement clause types include sentence-like subordinate clauses with indicative and subjunctive verbs as well as verbal noun clauses. Megrelian lacks infinitives. The division between indicative and subjunctive complement clauses is shown to correlate with the distinction between Truth and Action modality complements. The study focuses on the subordinating element ni, that is found in clause final position in subject, object and temporal clauses. As predicted from the fact that Megrelian is Head-final and Specifier-first, it places the complementizer, the enclitic element ni, in clause final position and wh-phrases and relative pronouns initially. 1. Introduction Our paper addresses some basic aspects of complement clauses in Megrelian with the focus on ni complement clauses. By complement clause we mean a sentence-like clause occurring in an object or subject position. (1) (a) mus what.DAT ap’irenk? vap’irenk diSk’a 2.3.intend.PRES? 1.3.intend.PRES wood-NOM dop’c’k’ire(n) 1.3.cut.OPT.C What do you intend? I intend to cut the wood. (b) Seuglebeli impossible re tena gavak ’ete(ni) is it.NOM 1.3.do.OPT.(C) It is impossible that I will do it. The outline of the paper is as follows. After some introductory remarks on the Megrelian language (2.1) and its grammatical structure (2.2)’ the complement types in Megrelian are presented (3). The following sections are devoted to general aspects o f ni clauses (4.1), with special 71 z zyxwv zyxwvut z zyxwvutsr Karina Vamling and Revaz Tchantouria attention to constituent order (4.2) and restrictions on the choice of the subordinate verb form (4.3). The observations presented in this paper are based on data collected during field work with Megrelian informants from Senak’i, mostly elderly people (special thanks to Ksenia Grigolia). Georgian was used as the mediator language in translating the examples (some reference to Georgian will be made throughout the paper). 2.1. The Megrelian language Megrelian is a South-Caucasian, or Kartvelian, language spoken in the western parts of Georgia. It is related to Georgian, Svan and Chan. Some treat Megrelian and Chan (or Laz, now spoken mainly in Turkey) together as dialects of the Zan language (cf. Jorbenadze 1991:15 - 17). Megrelian and Chan/Laz are fairly closely related and allow for some mutual intelligibility. The relationship to both Georgian and Svan is more remote, and speakers of the three languages do not understand each other without prior knowledge of the other languages. Megrelian exists primarily as a spoken language, as the literary language of most Megrelians is Georgian. From the point of view of nationality, the Megrelians consider themselves Georgian. 2.2. Preliminaries The unmarked word order is SOV (2a). The general ordering of elements is head-final, correlating with postpositions (b) and N final in NP:s (c-d). (2) (a) manana Manana.NOM givis iCinens Givi.DAT 3.3.know.PRES ‘Manana Knows Givi.’ (b) boSi(S) c’k’uma boy.GEN with with the boy (c) Ek’imi dudi my head my head (d) didi daExiri big fire big fire 72 Complement clauses in Megrelian Megrelian is not a language with rigid word order, SVO is a fairly common ordering. Alternative patterns exist in the NP as well. Postponed adjectives occur as stilistically marked as well as postponed genitive attributes, in particular with nouns denoting relatives. Specifiers are p.roposed, as in (3a-b): (3) (a) te this didi daExiri big fire zy this big fire (b) 3alam jgiro very well very well Grammatical relations are marked both by noun case and by cross-reference markers in the verb. The marking of grammatical relations in Megrelian is rather similar to the Georgian system. In Megrelian, as in Georgian, it is dependent on the choice of tense/aspect as well as verb class (cf. Harris 1985 for an extensive comparison of the system of grammatical relations in the Kartvelian languages). For reference, case morphemes and cross-reference markers are given here (those not relevant for the constructions treated below have been left out). As can be seen, first and second person pronouns lack case distinctions in the nominative, ergative and dative cases. In a sense this is compensated for by the fact that first and second person are marked by a richer set-up of cross-reference markers in the verb. zy zyxwvu zyxwv ( 4) Case markers in Megrelian Noun Nominative Ergative Dative Genitive -0 -k -S -5 Pronouns 1 ma ma ma Ek’im (singular) 2 3 si tina si tik si tis sk’an tiS Note that the traditional case labels used here do not fully correspond to the usual functions of these cases. In particular, the dative case 73 Karina Vamling and Revaz Tchantouna zyxwvu marks both direct and indirect objects (and subjects in some cases) and the ergative case is the general subject case in one TAM-series, not only the case of subjects in transitive clauses (see Klimov 1967 for a discussion of the status of the ergative in Megrelian). The verb includes cross-reference markers of both subject and objects. The markers occur both as prefixes and suffixes (the glosses give the meaning of these markers in the verb forms in the order subject, direct object, indirect object). zyxw zyxwvut zy zyxwv zy (5 ) Cross-reference markers 1sg 2sg 3sg lpl 2pl 3pl Subject markers (object 3rd person) Object markers (subject 3rd person) ‘I know him.’ etc ‘He knows me.’ etc. v -iEinenk 0-iEinenk iEinen-s v-iEinen-t 0-iEinen-t iEinen-a rn -iEinens g -iEinens iEinens rn -iEinena g -iEinena iEinena Megrelian has a rich system of tense, aspect and mood forms. As in Georgian, these forms are divided into series that share some morphological features and correlate with different patterns of marking grammatical relations. The scheme on the following page is taken from Kiziria’s sketch of the Zan language (Kiziria 1967:70). 74 zy zyxwvu zyxwvut Complement clauses in Megrelian (6 ) TAM series in Megrelian Marking with transitive verbs Subject Object Case & Case & Cross-ref. Crossref. I I1 Present Imperfect Subjunctive I Conditional I E‘aruns E’arundu E’arundas E-arunduko(n) Aorist Optative Conditional 11 (do)E’aru Erg. (do) E’aras V(do)E’aruk’o(n) Nom. uE’aru Dative ue’arudu mue’arudas uE’aruduk’o(n) Nom. 111 Resultative 1 Resultative 2 Subjunctive 111 Conditional I11 IV Nom. V- Dative m- m- zyxw z zyx Resultative 3 Resultative 4 Subjunctive IV Conditional IV noE’arue Nom. noE‘aruedu VnoE’aruedas noe’arueduk’on V- Dative m- In the table has been included the marking of grammatical relations in a sentence with a transitive verb: the subject and object cases and the series of cross-reference affixes (0-stands for the subject series and m for the object series). Below are given examples of sentences with transitive verbs in the different TAM-series as well as examples of other verb types. In TAM-I, subjects are marked by the nominative and objects by the dative, as illustrated below (7a-b). A group of verbs with experiencer subjects marks the subject ( =experiencer) by the dative case and the object (=source) by the dative case (c). 75 zyxwvuts zyxwv zyxw Karina Varnling and Revaz Tchantouna (7) (a) tina he.NOM suratis xant’uns picture.DAT 3.3.paint.PRES ‘He paints a picture.’ (b) tina sxap’uns he.NOM 3.dance.PRES ‘He dances.’ (c) (ma) mi?ors (I) tina 1.3.love.PRES he.NOM ‘I love him. - - tis - he.DAT (I) (ma) vu?ork 3.1 .love.PRES He loves me.’ In TAM-11, Megrelian has generalized the -k marker for subjects: (8) (a) tik he.ERG surati doxant’u picture.NOM 3.3.paint.AOR ‘He painted a picture.’ (b) tik he.ERG sxap’u 3.dance.AOR ‘He danced.’ (c) k’oEik man.ERG otaxuSa minilu roomhto 3.enter.AOR ‘The man entered the room.’ For the emotion verbs, however, the subject ( =experiencer) is marked by the dative case and the object ( =source) by the ergative case. (9) ciras mec’onu te boSik girl.DAT 3.3.like.AOR this boy.ERG ‘The girl liked the boy.’ In TAM-111, subjects are marked by the dative, direct objects by the nominative case and indirect objects are realized as postpositional phrases. ( 10) k’os 3ima-Sa c’erili meuE’aru man.DAT brother.for 1etter.NOM 3.3.write.PERF ‘The man has written a letter to his brother.’ 76 Complement clauses in Megrelian There is a group of typically low-activity verbs that take nominative marked subjects in TAM-111. (11) nodari doSkvid(el)e Nodar.NOM 3.die.PERF zy ‘Nodar has died.’ (Harris 1985:294) TAM-IV, examplified in (12)’ patterns like TAM-I. (12) Sima Pya c’erils noE’arue brother.NOM yesterday 1etter.DAT 3.3.write.RES3 ‘(Apparently) my brother wrote a letter yesterday.’ zyxw 3. Complement types in Megrelian and Georgian Even though Georgian and Megrelian are related, they exhibit important differences in their complementation constructions. One feature that immediately sets Megrelian apart from Georgian is the widespread marking of subordination by the clitic element -ni, that is cliticized to the verb (in the glosses the subordinating element ni is marked by C , complementizer). Georgian uses a construction with the initial complementizer rom (13b). (13) (a) bjers 1.3.believe.PRES guk’uk kayarda doC’aru-ni Dzuku.ERG 1etter.NOM 3.3.write.AOR-C ‘I believe that Dzuku wrote the letter.’ (b) mjera I.3.believe.PRES rom 3uk’um c’erili dac’era that Dzuku.ERG 1etter.NOM 3.3.write.AOR ‘I believe that Dzuku wrote the letter.’ This is the most common type complement type. Apart from this type we find, as in Georgian, and verbal noun construction and a rather rare type with the complementizer namda. In the verbal noun clause, the former direct object, diik’a ‘wood’, of the subordinated verb is marked by the genitive. The verbal noun itself 77 zyxw zyxwvu zyxw z Karina Vamling and Revaz Tchantouria is case marked as the object of the matrix verb (for the difference in case marking of the verbal noun in (( 14b-c), cf. section 2.2). ( 14) (a) (ma) (i1) diSk’as p’c’k’irunk wood.DAT 1.3.cut.PRES ‘I cut the wood.’ (b) vap’irenk I .3.intend.PRES diSk’a-S(i) c’k’irua-s wood-GEN cutting-DAT ‘I intend to cut wood.’ (c) dovap’iri 1.3.intend.AOR diSk’a-S(i) c’k’irua wood-GEN cutting-NOM ‘I intended to cut wood.’ None of the informants used the construction with the complementizer narnda spontaneously. It appears to have a more common use in complement clauses with negations. (15) muma pikrens, namda amdya va father.NOM 3.3.think.PRES that today marte not 3.corne.FUT ‘Father thinks, that he will not come today.’ A further limitation on namda clauses is that they do not occur in subject position. Here, only the ni clause (16b) and the verbal noun clause (c) are used. ( 16) (a) saE’iro necessary re c’erili dobE’aren(i) is 1etter.NOM 1.3.write.OPT.C ‘It is necessary to write a letter.’ (b) saE’iro necessary re c’eriliS E’arua is 1etter.GEN writing.NOM ‘It is necessary to write a letter.’ (c) ?saE’iro necessary re namda is that zyxw c’erili dobf ’are 1etter.NOM 1.3.write.OPT ‘It is necessary to write a letter.’ Apart from its function as a general marker of subordination in complement clauses the element ni also occurs as a marker of temporal 78 zy zyx Complement clauses in Megrelian zyxw subordinate clauses, corresponding to Georgian roca, ‘when’. The sentence in (17a) shows ni in its temporal function with an immediate perception verb. The simple sentence is given for comparison in (17b). (17) (a) kobgiri 1.3.see.AOR tina he.NOM [ otaxuSa minilu-nil roominto 3.enter.AOR-TEMP ‘I saw him when he entered the room.’ (b) jimak brother.ERG otaxuSa minilu room.into 3.enter.AOR ‘The brother entered the room.’ This might be compared with an object clause in relation with another immediate perception verb, where ni has the general subordinating meaning. (18) SevniSni [tik otaxuSa minilu-nil 1.3.notice.AOR he.ERG room.into 3.enter.AOR.C ‘I noticed that he entered the room.’ In what follows, we will focus on the ni complement type. z 4.1. The subordinating element -ni In order to recognize the marker ni, it is important to know that it occurs in different shapes; as ni and reduced to i or n, as was noted already by KipSidze ( 1914:141). For example: kmortasi < k m o r t asa+ni (in the examples in the paper, the parenthesis around ni indicates that it is reduced in pronunciation). The element ni occurs with different verb forms, for example a present tense form in (19a) and an optative in (19b). ( 19) (a) vpikrenk 1.3.think.PRES 3uk’u c’erils E’arunsi(n) Dzuku.NOM 1etter.DAT 3.3.write.PRES.C ‘I think that Dzuku writes a letter.’ (b) mok’o 1.3.want.PRES 3uk’u-k c’erili doE‘arasi( n) Dzuku.ERG 1etter.NOM 3.3.write.OF’T.C ‘I want Dzuku to write a letter.’ 79 Karina Vamling and Revaz Tchantouria zyxwvu Looking at the corresponding simple sentence, the difference emerges between the simple present and the present with the subordinate marker. The forms with and without the element have distinct distributions; forms with the subordinate marker may occur only in subordinate clause positions. (20) 3uk’u c’erils E’arunsl *E’arunsi Dzuku.NOM 1etter.DAT 3.3.write.PRES/ *3.3.write.PRES.C ‘Dzuku writes a letter.’ The same relationship holds between the simple optative and the optative with the subordinate marker. As seen when comparing (20a) and (b), the choice between minilasi and minilas is not positionally conditioned, as they both occur in final position. (21) (a) mok’o 1.3.want.PRES tik otaxuSa minilasi he.ERG room.into 3.enter.OPT.C ‘I want him to enter the room.’ (b) otaxuSa roomkto minilas/ *minilasi! 3.enter.OPT/ *3.enter.OPT.C ‘(May) he enter the room!’ It thus appears that the element -ni behaves as a complementizer. Its main function is to mark the clause in question as subordinate. The status of the element -ni as an affix or a clitic is not quite clear. As Megrelian does not have any standardized writing system, this does not give any indication of the traditional treatment. KipSidze (1914:141) characterizes ni as an enclitic particle, and translates it as “Kozda, Ymo, urno6bz, ecilu” ( =when, to, for. . . to, if. KV). He points out that ni merges and with the preceding word enclitically and thereby attracts the stress to the end of the word, and sometimes completely loses its consonantal element. As word order Megrelian is verb final, it is difficult to find examples with word order other than with the verb as the final element in order to check the placement of ni in such cases. In the material elicited from informants, there are some examples of ni cliticizing to a final noun. Below in (22), (a) was the first translation given, (b) was accepted as a possible variant when specifically asked zyxwvu zyxwvut zyxwv 80 Complement clauses in Megrelian about its acceptability, and (c) was considered to be the best one (the sentence in (a) is closer than (c) to the Georgian construction that served as source for the translation). (22) (a) bk’itxi 1.3.3.ask.AOR muE’o Vvali gaak’etuni how cheese.NOM 3.3.do.AOR.C zy ‘I asked how he had made the cheese.’ (b) bk’itxi mu20 eSeyu 1.3.3.ask.AOR how %ali// Vvalini 3.3.make.AOR cheese.NOM// cheese.NOM.C ‘I asked how he had made (taken out) the cheese.’ (c) bk’itxi ‘vali muE’o eSeyuni cheese.NOM how zyxwv 1.3.3.ask.AOR 3.3.make.AOR.C ‘I asked how he had made (taken out) the cheese.’ Thus, these examples indicate that ni occurs as the final element. It also shows that ni has greater freedom than a bound affix. Here, I will follow KipSidze and treat ni as a final clitic element, enclitic to the subordinate verb in the vast majority of cases. The following complex sentence in (23) with two instances of ni shows that it occurs in clause final position: (23) vecdebuki (tis) kurEue-ni c’erili 1.3.3.try.AOR (he.DAT) 1.3.3.advice.OPT-C 1etter.NOM midajyonasi-ni 3.3.send.OPT-C ‘I try to advise him to send the letter.’ It is interesting to note that ni is not the only element in Magrelian that occurs in clause final position. Yes/no-questions are formed by adding the element -0 to the verb form. Compare the declarative clause (tina) ragadans ‘He talks.’ and the question (tina) ragadans-o ‘Does he talk?’, and (tik) (tis) dut’aru, ‘He wrote to him.’ to (tik) (tis) dut’arun-o ‘Did he write to him?’. The addition of -0 has the effect of attracting the stress to the preceding syllable: ‘ragadans, raga‘dans-o, ( KipSidze (1914:92). (24) gives an example of a question marker in a matrix zy z 81 zyxwvu Karina Vamling and Revaz Tchantouria clause: (24) rjers-o pk’uk kayarda 2.3.believe.PRES-Q Dzuku.ERG letter.NOM zy zyx zy doE’aruni? 3.3.write.AOR.C ‘Do you believe that Dzuku wrote the letter?’ An indirect yes/no-question does not include both ni and 0, that otherwise would have indicated the order among them. (25a) shows the direct question and (b) the corresponding indirect question. (25) (a) bk’itxi 1.3.3.ask.AOR ninos: datok mortu-o? Nino.DAT Dato.ERG 3.come.AOR-Q ‘I asked Nino: Has Data come? (b) bk’itxi 1.3.3.ask.AOR ninos komortu tu vari datok Nino.DAT 3.come.AOR if knot Dato.ERG ‘I asked Nino if Dato had come’ Another element behaving in a parallel way to ni and o is da ‘iP. It is attracted to the preceding word that undergoes morphophonological changes similar to those in interrogative clauses (examples from KipSidze 1914:141). (26) gok’ona da, kamorti (cf.gok’o) 2.3.want.PRES if, cf.2.3.want.PRES) 2.come.FUT ‘If you want to, come to me.’ 4.2. Constituent order zyx In the section above we have seen that Megrelian has a final position for elements that mark the status of the clause - simple versus subordinate (ni), declarative versus interrogative (o), and neutral declarative versus conditional (da). It has been suggested that languages with clause final complementizers do not exhibit wh-movement and leftward relative clause formation (Bresnan 1970:319). However, as noted in Vamling & Tchantouria ( 1991:79), initial placement of wh-phrases does occur in Megrelian: 82 Complement clauses in Megrelian (27) [muner mankanas] mirEenk ip’ide-ni? [which 1.3.buy.AOR-C car.DATl 2.3.1 .advise.PRES zy ‘Which car do you advise me to buy?’ Relative clause formation reveals another exception. Relative clauses are formed by a head noun followed by the clause out of which it has been relativized: (28) viEinenk maxant’als, namutuk te surati 1.3.know.PRES artist.DAT who.ERG this picture.NOM doxant’u-ni zyxwv 3.3.paint.AOR.C ‘I know the artist, who painted this picture.’ The relative pronoun narnu- ‘which’ is case-marked according to its grammatical function in the subordinate clause and placed initially. Above in (28) the subject has been relativized and below in (29a) the direct object and (29b) a possessive expression. (29) (a) bsri 1.3.see.AOR ti surati, namutu megobark that picture.NOM, that.NOM friend.ERG maEuku-ni 3.3.l.gave.AOR-C ‘I saw the picture that my friend gave to me.’ (b) Sebxvadi megobars namuSi naxant’uti 1.3.meet.AOR fnend.DAT which.GEN picture.NOM gamopenas bgiri-ni exhibition.DAT 1.3.see.AOR-C z ‘I met the artist, whose painting I saw at the exhibition.’ Megrelian thus appears to exhibit a mixed system. The relative pronoun is placed initially but at the same time the complementizer ni is found in final position. In (30a) we have the usual case with ni cliticized to the verb and in (b) it has cliticized to the final NP. 83 Karina Vamling and Revaz Tchantouria (30) (a) kemebrti ti zyxwvu k’oEiSa, namuSeti p’asuxis 1 .come.AOR that man.to, who.from veludi-ni reply.DAT 1.3.wait.IMP-C ‘I came to that man, from whom a expected a reply.’ (b) kemebrti 1 .come.AOR ti k’oEiSa, namuSeti veludi that man.to, who.from 1.3.wait.IMP p’asuxisi-ni reply.D AT-C ‘I came to that man, from whom a expected a reply.’ Cooccurring relative pronouns and complementizers is not an unusual situation. It is for instance reported in Old and Middle English, Canadian French and Dutch (Radford 1988:486 and references therein). What is unusual about the Megrelian structure is the asymmetry of the placement of the relative pronoun and the complementizer: zyxwv zyxwv zyx A S C Actually, following current ideas in the GB-framework, this asymmetric placement of the complementizer and the wh-phrase and relative pronoun is to be expected on the basis of other typological parameters (we refer to Radford 1988 and references therein for an overview of GB). As noted above, Megrelian is Head-final and Specifier-first. In attempting to generalize the X-bar projection for the entire sentence structure, S’ has been analyzed as a projection (C’) of the complementizer head (C). The maximal projection of C ( C ” ) includes an optional Specifier position (XP) as a sister node of C’. Taking into account the information that Megrelian is Head-final and Specifier-first gives the structure below: 84 Complement clauses in Megrelian zy zyxwvu zyxw zyxwvuts zyxwv The specifier position is taken to be the landing-site of various constituents that are moved out of S: wh-phrases, topicalized elements and relative pronouns. This thus gives us the structures in (33a-c) for a complement clause, wh-question and relative clause, repeated from above (( 18, (29a) and (27)). (33) (a) SevniSni C”[ C’[S[ tik 1.3.notice.AOR he.ERG otaxuSa minilu] C[nil]] room.into 3.enter.AOR-C ‘I noticed that he entered the room.’ (b) bgiri ti surati C”[namutu C’[ S[megobark 1.3.see.AOR that picture.NOM, that.NOM friend.ERG -maEuku] C[ nil]] 3.3.1.give.AOR-C ‘I saw the picture that my friend gave to me.’ (c) C”[NP[muner mankanas] which car.DAT C’[S[ mirEenk-ip’ide]C[ nil]]? 2.3.1.advise.PRE.9 1.3.buy.AOR-C ‘Which car do you advise me to buy?’ 4.3. Restrictions on the choice of the subordinate verb form We have seen that ni clauses occur extensively, with a matrix verb of any type and in both subject and object position. In this section we will look at a subdivision of ni clauses. When we analyse complement clauses we find that some restrictions on the forms of the subordinated verbs correlate with types of matrix verbs. First compare the two examples: 85 zyxwvut z Karina Vamling and Revaz Tchantouria (34) (a) b3ers guk’u 1.3.believe.PRES Dzuku.NOM c’erils jgiro doE’arunsi(n) / 1etter.DAT well 3.3.write.FUT.C/ *doE’arasi(n) *3.3.write.OPT.C ‘I believe that Dzuku will write the letter well.’ (b) btxink I.3.ask.PRES jimas c’erili doE’arasi(n) / brother.DAT letter.NOM 3.3.write.OPT.C/ zyxwvu *doE’arunsi(n) *3.3.write.FUT.C ‘I asked (my) brother to write a letter.’ The verb in (34a) bjers ‘I believe. . .’, belongs to a type of verbs that select complements with truth modality meaning. Other verbs of this type are: (35) miEku vgebulenk vpikrenk vnanenk I I I I know. . . understand. . , think. . . regret. . . The matrix verb in (34b) belongs to a group of verbs that are characterized by Action modality complements, as in: (36) zyxwv zyxwvut btxink vap’irenk vocaduk vaSerenk vuzoSunk mok’o I I I I I I ask.. . intend . . . try. . . persuade. . . order.. . want. . . The categories Truth and Action modality originate from a study by Ransom 1986, where she proposes a rather elaborate system of tests in order to single out the verbs in the two groups. Here, we will only point to one test that sets the two groups apart and that illustrates the basic difference between them. Matrix verbs in (35) are compatible with that it is true, that. , . (Megrelian: martali re tina . . .) but incom86 zy zyxwvu zyxw zyxw Complement clauses in Megrelian patible with to perform the action of. . . (Megrelian: tik te sakme gak’etasi). The compatibility is the reverse when one selects matrix verbs from (36). In Action modality complements we observe the following restrictions. When the matrix verb occurs in a non-past tense, then the subordinate verb selects the optative. (37) btxink Simas diSk’a doc’k’irasi(n) 1.3.ask.PRES brother.DAT wood.NOM 3.3.cut.OPT.C ‘I ask (my) brother to cut the wood.’ When the matrix verb is a past tense form, the third conditional is chosen: (38) btxi smas disk’a duc’k’iruduk’on 1.3.ask.AOR brother.DAT wood.NOM 3.3.cut.CONDIII.C ‘I asked (my) brother to cut the wood.’ Similar restrictions are found in Georgian (Vamling 1989), with the exception that the pluperfect, not the conditional 111, is chosen in complements to past tense forms. (39) (a) vtxov 1.3.ask.PRES mas, rom es gaak’etos he.DAT that 3.3.do.OPT it.NOM ‘I ask him to do it.’ (b) vtxove 1.3.ask.AOR mas, rom es gaek’etebina he.DAT that 3.3.do.PLUP it.NOM ‘I asked him to do it.’ Such restrictions are not at hand when the complement is of the truth modality type, neither in Megrelian (40) nor in Georgian. 87 zyxwvut zyxwv Karina Vamling and Revaz Tchantouria (40) (a) miEku 1.3.know.PRES E’uman jima diSk’as tomorrow brother.NOM wood.DAT doc’k’irunsi(n) 3.3.cut.FUT.C ‘I know that (my) brother will cut the wood tomorrow.’ (b) miEku 1.3.know.PRES jima digk’as brother.NOM wood.DAT zyx c’k’irunsi(n) 3.3.cut.PRES.C ‘I know that (my) brother cuts the wood.’ miEku jimak diSk’a 1.3.know.PRE.S brother.ERG wood.NOM ‘I know that (my) brother cut the wood.’ doc’k’iru(n) 3.3.cut.AOR.C 5. Summary Megrelian has a set of complement types consisting of (a) finite clauses with the subordinating marker ni, (b) verbal noun clauses, and, more marginally, (c) finite clauses with the complementizer namda. Megrelian lacks infinitive clauses. The subordinating element ni is analyzed as a general complementizer in subject and object clauses as well as temporal clauses. ni is usually cliticized to the indicative or subjunctive verb form. The complementizer ni stands in clause final position, as expected from the fact that Megrelian is head-final. A parallel behaviour is noted for the interrogative particle o and da, ‘if’. As predicted from the typological parameters Head-final and Specifier-first, Megrelian places the complementizer in finaI position and relative pronouns and wh-phrases initially. On the basis of a division of matrix predicates taking Truth and Action modality complements respectively, we noted restrictions in the selection of the form of the complement verb. In Truth modality complements there are no restrictions on the complement verb, whereas there is a choice between the optative and the conditional I11 in Action modality complements. The choice between the latter two is determined by the selection of tense of the matrix clause. References BRESNAN, J. 1970. On Complementizers: Toward a syntactic theory of complement types. Foundations of Language 6 , 297-321. 88 zyxwvutsrq zyxwvu zyx Complement clauses in Megrelian A. C. 1985. Diachronic Syntax: The Kartvelian Case. Syntax and Semantics, 18. New York: Academic Press. JORBENADZE, B. 1991. The Kartvelian languages and dialects. Tbilisi: Mecniereba. KIPSIDZEI . 1914. Grammatika mingrel’skago (iverskago) jazyka. [ = Grammar of Megrelian]. Sankt-Peterburg: Imperatorskaja Akademija Nauk. KIZIRIA, A. 1967. Zanskij jazyk. [ =The Zan language]. Jazyki narodov SSSR. Vol. 4, 62-76. KLIMOV,G. D. 1967. K ergativnoj konstrukcii predloienija v zanskom jazyke. [ =On the ergative construction in Zan]. Ergativnaja konstrukcija predloienija u jazykax razlitnyx tipov, ed. V. M . Zinnunskij, 149-155. Leningrad: Nauka. RADFORD,A. 1988. Transformational grammar. Cambridge: Cambridge University Press. RANSOM,E. 1986. Complementation: its meaning and forms. Amsterdam/ Philadelphia: John Benjamins. VAMLING,K. 1989. Complementation in Georgian. Lund: Lund University Press. VAMLING, K. & TCHANTOURIA, R. 1991. The Caucasian languages. Complement structures in the languages of Europe - some preliminary surveys, eds. K . Borjars & N. Vincent. Eurotyp Working Papers, 111: 1, 71-87. Manchester University. HARRIS, zyxwvu Revaz Tchantouria Karina Vamling Department of Linguistics and Phonetics Lund Unicersity Helgonabacken 12 223 62 Lund Sweden 89