Novel syntax of NFCs in minor contexts

94 Jarawara Complement Clauses 243 Hiyama hiyama white-lipped peccary.m mee mee 3pl.s kabi kaba.nfin eat amosatee amosa-tee be good-hab mee mee 3pl.poss amake. ama-ke sec-decl+f ‘White-lipped peccaries are good to eat [lit., the eating of white-lipped peccaries is good].’ In 243, the S hiyama mee does not refer to a speciic group of peccaries, it means white-lipped peccaries in general, i.e. it is generic. When the Jarawaras want to refer to people in general, they use the inclusive plural pronominal ee, as in 244. 244 haaha haaha this +f ee ee 1in.poss taboro taboro place ee ee 1in.a awawamaraha awa-waha-ma-ra.+F see-change-back- neg tokomaki to-ka-makI. nfin away-gocome-following amosateemone amosa-tee-hamone be good- hab-rep+f amake ama-ke sec-decl+f neme neme sky. f yaa. yaa adjnct ‘If people come back to their village [after dying], they dont go to heaven well.’ This example also shows a quirk of the use of the pronominal ee, that sometimes it is not repeated in a sequence of clauses with the same subject. The S of the NFC tokomaki is ee, but it is phonologically null because it is understood from the preceding clause. This cannot be done with the other plural pronominals. 48

3.8 Novel syntax of NFCs in minor contexts

In view of the fact that some of the syntactic contexts discussed in this section were not included in table 2, I have included table 12 summarizing these novel contexts. I have not included data on NFCs with nima or those with forima na in the table, because it is not possible to characterize these cases as a relation of an NFC to an MC. Nor have I included the cases of postposed NFCs, since most of these could be included in table 2, and the rest do not have a regular syntactic relation to the MC. Table 12 Minor types of NFCs Type of NFC Function of NFC in MC Type of MC Raising, Control, or Neither Examples copular clause object transitive Ac control A-S 198 48 This phenomenon involving the pronominal ee occurs independently of the presence of NFCs, as is shown in i. In this sentence both clauses are inite, and ee is non-overt but understood as the A of the second clause, makari akawatee amake. i Ee ee 1in.s kama ka-ma+f gocome-back makari makari clothes.f akawatee aka-waha-tee wear-change-hab amake. ama-ke sec-decl+f ‘When we come back [from taking a bath in the stream], we change clothes.’ Distinguishing NFCs from other phenomena 95 intransitive copular complement copular clause with ama neither 199 intransitive copular sub- ject copular clause with ama neither 200, 201 intransitive adjunct intransitive control S-S 221, 222, 223, 224 intransitive adjunct transitive Ac control A-S 225 intransitive adjunct intransitive neither 226 intransitive adjunct transitive Ac neither 227 intransitive adjunct transitive Oc neither 228 In the next section I present several kinds of phenomena that need to be distinguished from NFCs. 4 Distinguishing NFCs from other phenomena In this last major section, I discuss several kinds of phenomena that are similar to NFCs in Jarawara but need to be distinguished from them. The most important of these are nominalized clauses, which may possibly be considered another kind of non-inite clause. Distinguishing them from NFCs is sometimes a subtle matter. There also seem to be a num- ber of “fossilized” forms that probably originally were NFCs but synchronically should be analyzed as inalienably possessed nouns. The habitual suix -tee has a nominalized form -ti, but these forms do not seem to be the same as the forms used in NFCs. Finally, there is the mysterious case of phrases with kihi. 4.1 Nominalized clauses Dixon 2004 devotes the whole of one chapter ch. 19 and signiicant parts of two other chapters chs. 20 and 21 to what he calls nominalized clauses. In the following discussion I assume, as he does, that this is a single phenomenon, but this may not necessarily be the case. My concern here, though, is not to go deeply into the analysis of these structures, but to distinguish them from NFCs. Dixon distinguishes the two, but ends up analyzing a number of what I consider NFCs as nominalized clauses. I believe the distinction can be made clearer. The most common syntactic context in which nominalized clauses are used is in clausal adjuncts marked with yaa, as in 245 and 246. 245 owa owa 1sg.o tiwasimi ti-wasi-ma.nom+f 2sg.a-ind-back yaa yaa adjnct oneba one-ba other+f-fut taa taa give tinamakehabanake ti-na-makI-habana-ke 2sg.a-aux-following-fut+f-decl+f owa owa 1sg ni ni to yaa. yaa adjnct ‘When you see me again you can sell me some more [bags of salt].’ 96 Jarawara Complement Clauses 246 Yawita yawita peach palm.m mete mete rear+m hate hata.nom+m be ripe yaa yaa adjnct faha fowe faha fowe lood.f kisate ka-risa-te gocome-down-hab amake. ama-ke sec-decl+f ‘When the bottoms of the peach palm nuts get yellow, the seasonal looding starts.’ There are two important diferences with NFCs to note here. First, there is gender agreement in nominalized clauses. In 245 tiwasimi is feminine to agree with the second person singular subject ti-, and in 246 hate is masculine to agree with the masculine subject yawita mete ‘bottoms of peach palm nuts’. There is no gender agreement in NFCs; as we have seen above, a non-inite form ends with i no matter what the gender of any argument in the sentence is. 49 A second diference with NFCs is that nominalized 49 To be more precise, a non-inite form ends in i for verb stems that end in a. This is the vast majority of verbs. There is, however, a signiicant minority of verb stems that end in one of the other three vowels, i.e. e, i, or o, or with the morphophoneme I. For those ending in i or o there is no vowel change when they are used in an NFC. Examples i and ii show that for a verb stem ending with i such as sawi ‘be located’, the non-inite form will end in i, whether the S is feminine, as in i, or masculine, as in ii. i oko o-kaa 1sg.poss-poss sawi sawi.nfin be at yaboharo yabo-haro be far-rp.e+f ama ama sec oke o-ke 1sg.poss-decl+f esibitao esibitao hospital.m yaa. yaa adjnct ‘I stayed a long time in the hospital.’ ii sawi sawi.nfin be at toyabo to-yabo.nom ch-be far yaa yaa adjnct naimise nai-misa+m be big-up ‘He stayed there a long time, and grew up.’ Similarly, examples iii and iv show that the non-inite form of verb stems ending with o always end in o. In iii, the NFC wato has a feminine subject, and in iv the non-inite kamo has a masculine subject. iii otaa otaa 1ex.poss ati ati voice wato wato.nfin know naira nai-ra+f be a lot-neg ‘She didnt know our language well.’ iv Okobi o-kaabi 1sg.poss-poss-father.m kamo kamo.nfin bury hawa hawa be inished toe to-ha.nom+m ch-aux yaa yaa adjnct ‘When the burying of my father was done.’ Verbs that end in e actually have a syllable ha added underlyingly, so the non-inite form ends in hi, as in the form ketehi in v. v Kobati kobati compadre.m tika ti-kaa 2sg.poss-poss ketehi ka-teha.nfin comit-spread something on onofa o-nofa 1sg.a-want okare. o-ka-re 1sg.a-decl-neg+f ‘Compadre, I dont want to rub [medicine on] you.’ Nominalized clauses 97 clauses have person preixes just as regular inite verbs do, cf. tiwasimi in 245. This is not true for NFCs, cf. for example tika tai in 27 above. In both these respects, in having gender agreement and in having person preixes, nom- inalized clauses are more like regular inite verbs than NFCs. These two characteristics are also present in another common syntactic context for nominalized clauses, i.e. when they are used as the S of a sentence with the copular verb ama, as in 247 and 248. 247 Owaka o-to-ka-ka.+F 1sg.s-away-comit-gocome moto moto loop okowami o-to-ka-na-waha-ma.nom+f 1sg.s-away-comit-aux-change-back ama ama be ahi. ahi then ‘I went away, and looped around.’ 248 Kana kana run ni na.nfin aux toke to-ka.nom+m away-gocome amake ama-ke be-decl ahi. ahi here ‘Here is where he ran.’ In 247 moto okowami has feminine agreement to agree with the irst person singular subject, and in 248 toke has masculine agreement to agree with the null subject, a man. For verb stems ending with the morphophoneme I, recall that the I will be realized as i if the preceding number of moras in the word is even, and e if the preceding number of moras is odd. However, the non-inite form of stems that end with I will always end with i, no matter what the number of preceding moras, and also no matter whether the subject is feminine or masculine. In the following examples, there is an even number of preceding moras in kamaki in vi and vii, and an odd number of preceding moras in fori in viii and tokomaki in ix. The subjects of the NFCs are feminine in vi and viii, and masculine in vii and ix. vi mee mee 3pl.s kamaki ka-makI.nfin gocome-following wawato wa-wato dup-know toaremetemoneke. to-ha-ra-hemete-mone-ke ch-aux-neg-fp.n+f-rep+f-decl+f ‘They never came back.’ vii Makoni makoni for this reason kamaki ka-makI.nfin gocome-following nofare nofa-ra.+M want-neg amane. ama-ne sec-bkg+m ‘Thats why he doesnt want to come.’ viii fori forI.nfin lie on top yaboha yabo+f be long ‘She lay there a long time.’ ix Tokomaki to-ka-makI.nfin away-gocome-following tee tee 2pl.a awini? awa-ni see-bkg+f ‘Did you see him go?’ The important point is that in none of these cases is there ever gender agreement. 98 Jarawara Complement Clauses Also, the exponent of the subject in 247 is the preix o- in moto okowami, and not oko as it would be if it were an NFC. 50 Dixon 2004:483 recognizes both of these diferences. Regarding the irst diference, he says: “Gender marking on the last element of the predicate nucleus is retained. A inal a becomes e for m agreement, and is raised to i for f agreement.” And regarding the second diference, he says this: “The predicate of a nominalized clause...is unlike a complement clause [i.e. NFC] in that it may also include 1sg o-...2sg ti- recall that in a complement clause these are replaced by oko and tika, respectively, in clause-initial position, and Oc preix hi-...” As we have seen in the introduction above, Dixon does not accept that an NFC can have a tense-modal attached to it; he analyzes all such cases as nominalized clauses. While it is true that nominalized clauses can have tense-modals attached to them, it is clear that NFCs have this capability as well. According to Dixons view, 20 above would have to be analyzed as a nominalized clause, but then there should be masculine gender agreement, and there is not. In summary, although it is easy to confuse NFCs with nominalized clauses, mainly be- cause nominalized clauses have an i ending for feminine agreement and as it happens, feminine nominalized clauses are more common by far than masculine ones, it is possible to distinguish them by the criteria of gender agreement and the presence nominalized clauses or absence NFCs of person preixes. Although it is probably the case that nomi- nalized clauses should be analyzed as lacking initeness like NFCs, in these two ways they are more like inite clauses than NFCs. 51

4.2 Fossilized forms