3.4 Formal properties of DOM
3.4.7 DOM in complex predicates
Complex predicates in Nepali can be defined parallel to Chintang (cf. section 2.6.5.3): they consist of a noun (“N”) coding a state of affairs combined with a light verb (mostly gʌr- ‘do’), which together can be viewed as a single semantic predicate. Most abstract notions in Nepali can only be expressed as complex predicates, so this construction is very frequent, especially in the wrien language. e inventory of complex predicates is likely to be several hundred times larger than in Chintang. Below is a first example.
(68) Murkhʌ
idiotic manche-laiperson-DATkʌs-ʌi-lewho-FOC-ERGpʌnialso adʌrrespectgʌr-dʌin-ʌn.do-NEG.NPST-3p
‘Nobody respects an idiot.’ (NNC:book-popularlore-balsukti-2061:864) Morphosyntactically adʌr ‘respect’ clearly is a noun, as shown in (69), where it is modified by an adjective and marked by ERG:
(69) Mʌ-lai
1s-DATbhʌemisritmixed.with.fearadʌr-lerespect-ERGher-nelook-IPFV.PTCPgʌr-th-e.do-PST.HAB-3p
‘ey used to look at me with respect mixed with fear.’ (NNC:book-autobiography-mero- aviral-jivangit-2060.2495)
Still, adʌr does not behave like an argument in (68). Instead, adʌr gʌr- looks like a single predicate with two arguments, an A (kʌsʌile) and a P (manchelai).
e questions regarding the relation between complex predicates and DOM are similar as for S/A detransitivisation. e possibility of O-AGR corresponds to the possibility of marking N by DAT. e question of whether an additional argument besides N is allowed is the same. An addi- tional question is whether A gets marked by ERG or not. In Chintang this factor is tied up with O-AGR, but in Nepali A and O marking are independent of each other (section 3.4.1), so this point is worth looking at. Further, since the question is relevant for Nepali whether some O may be considered as incorporated (see section 3.4.5), the same may also be asked for complex predicates. So far I have not come across any DAT-marked N in natural spoken Nepali. Corpus searches over the NNC for about a dozen highly frequent N with the V gʌr- ‘do’ also did not yield any matches. N-DAT is sometimes marginally possible in elicitation but never preferred over N-NOM.
is makes it very different from O-AGR with N in Chintang, which likewise was not the default but well possible with many complex predicates, and justifies treating complex predicates as a formal rather than a functional factor in Nepali DOM. One of the rare cases where an N-DAT could be elicited is shown in (70), where DAT is marginally acceptable on yatra ‘journey’:
(70) Us-le
DIST-ERGbistwentybʌrsʌ-ʌgaɖiyear-ANTE bihemarriagegʌr-e-pʌchido-NMLZ-TMP.POSTgʌr-ekodo-PRFV.PTCPyatra(?-lai)journey-DATpheriagain gʌr-y-o.
do-PST-3s
‘He made the (same) journey again that he had made aer marrying.’
(elicitation BP/KP 2012) e possibility of DAT seems to depend on whether N can be construed as an independent referent of which several instances can be easily separated and identified with each other. But this alone is not sufficient for DAT. (71) shows an example of a referential N which has properties that are typical of O-DAT (highly specific, marked by one of the focal demonstratives that are otherwise frequently used with DOM – cf. 3.5.11) but which still can only be marked by NOM.
(71) Us-le
DIST-ERGtehiMED.FOCkam(*-lai)work-DAT gʌr-y-o.do-PST-3s
‘He did that very same work.’ (elicitation BP/KP 2012)
Also note that the reason why kam cannot be marked in (71) is not that it codes a process. e very same noun can be marked by DAT in sentences such as (72), confirming once more that the disfavouring of N-DAT in complex predicates is a formal factor:
(72) Hʌusʌla,
encouragmentprer˜ʌɖamotivationrʌandramrogood kam-laiwork-DATprʌs˜ʌsapraise gʌr-nedo-IPFV.PTCPgʌr-nudo-INF1
pʌr-ch-ʌ.
be.necessar-NPST-3s
‘One should encourage and motivate them and praise good work.’
(NNC:sadhana-psychology-2061-10.105) Interestingly, in spite of the near-ungrammaticalness of N-DAT, the A of a complex predicate must always be marked by ERG in perfective tenses, no maer whether there is an O-like argument besides N or whether N itself is P. is is even the case with N that are minimally referential and have no chance of ever becoming the head of an expanded NP, such as hʌar ‘hurry’ in (73): (73) Us*(-le)
DIST-ERGhʌarhurry gʌr-y-o.do-PST-3s
‘He hurried.’ (elicitation BP/KP 2012)
is shows that A and O case marking react to very different criteria. e fact that complex pred- icates like hʌar gʌr- require A-ERG is a strong argument for analysing N as a special kind of argument (with role = P) even when it is marked by NOM, at least as long as there is no addi- tional object-like argument. Where such an argument is present as in (74), N-DAT is never even marginally grammatical, although there are no restrictions other than the usual ones on the case of the additional P:
(74) Us-le tyo misin(-lai) nas(*-lai) gʌr-y-o.
DIST-ERG MED machine-DAT destruction-DAT do-PST-3s
‘He destroyed that machine.’ (elicitation NP 2012)
Both complex predicates with an additional P and where N itself is P do not show characteristics of noun incorporation. Words can intervene between N and V (75a), the order of N and V can be reversed (75b), and V can be gapped when it has just been mentioned (75c). Examples where N is the head of a complex NP were already shown above, e.g. in (70).
3.4. FORMAL PROPERTIES OF DOM (75) a. Chʌlphʌl
discussionmatrʌionly gʌr-ch-ʌn.do-NPST.3p ‘ey only discuss.’
b. Kaʈhmanɖu-ma Kathmandu-LOC gʌr-ch-ʌ, do-NPST-3s kam. work ‘He works in Kathmandu.’ c. Tʌrʌ
but ʌɖkʌlestimationrʌandnapmeasuregʌr-nedo-IPFV.PTCPkamworkphʌrʌkdifferentch-ʌ.be-NPST-3s
‘But estimating and measuring are different.’ (elicitation SAR 2011) A special subgroup of complex predicates is constituted by etymologically related N-V combina- tions. In these predicates N is truly semantically empty because exactly the same meaning is also coded by V. Nevertheless, the behaviour of predicates with figura etymologica is identical to that of other complex predicates: DAT is mostly ungrammatical as in (76) but rarely possible with highly referential N as in (77). ERG is obligatory on A, as shown in (76).
(76) Hami*(-le)
1p-ERG tehiMED.FOCkhel(*-lai)game-DATkhel-thy-ʌũ.play-PST.HAB-1p
‘We used to play that very same game.’ (elicitation BP/KP 2012) (77) Prithbi
Pṛthvī NaraẽɖNārāyaṇSa-koŚāha-GENjibʌn(?-lai)life-DAT ʌrkoothermanche-leperson-ERGjiu-nʌlive-INF2
sʌk-dʌinʌ-thy-o. be.able-NEG-PST.HAB-3s ‘Another person couldn’t have lived the life of Pṛthvī Nārāyaṇ Śāha.’ (elicitation BP/KP 2012)
Many N take cannot only take gʌr- ‘do’ as their V but also the copula hu-. e effect is a passive: A is removed completely so that it cannot be re-introduced by -dwara ‘by’, and verbal agreement is re-linked to O. N does not change its shape in this process. Below is an example for nikasi ‘export’ (nikasi gʌr- ‘export’, nikasi hu- ‘be exported’).
(78) a. Sʌhʌkari-le
cooperative-ERGjʌɖibuʈiherbs khʌridpurchasegʌr-ido-CVB1
Bharʌt
India nikasiexportgʌr-ch-ʌ.do-NPST-3s ‘e cooperative purchases herbs and exports them to India.’
(NNC:kantipur-business-2061-12-20.27) b. Nepali
Nepalesegai,cowbhʌı̃si-kobuffalo-GENposilonutritiousdudhmilk BharʌtIndia nikasiexportbhʌ-y-o.happen-PST-3s ‘e nutritious milk of Nepalese cows and buffalos was exported to India.’
(NNC:himalkhabarpatrika-2061-01-01.696) Differently from the morphological passive discussed in section 3.4.6, the light verb passive cancels the possibility of DOM. is can be aributed to the fact that while the morphological passive keeps A in the valency, the light verb passive completely removes it so that O truly becomes S. (79) shows contrasting examples for a sentence in the active and in the two passives with the complex predicate khʌtʌm gʌr- ‘ruin’ / khʌtʌm hu- ‘be ruined’:
(79) a. Rasʈrʌ(-lai)
state-DAT khʌtʌmend gʌr-ch-ʌn.do-NPST-3p ‘ey ruin the state.’
b. Rasʈrʌ(-lai)
state-DAT khʌtʌmend gʌr-i-nch-ʌ.do-PASS-NPST-3s ‘e state is (being) ruined.’
c. Rasʈrʌ(*-lai)
state-DAT khʌtʌmend hu-nch-ʌ.become-NPST-3s
‘e state gets ruined.’ (elicitation SAR 2011)
Another peculiarity of complex predicates is that when there is an O besides N, this O may fre- quently not only be marked by NOM or DAT but also by GEN (i.e. as the possessor of N). An
example of NOM alternating with GEN is shown in (80) below. Since O-GEN are never possible with simplex predicates, they will be ignored in the remainder of this work.
(80) Nepal-ma
Nepal-LOCciya(-ko)tea-GEN utpadʌnproductiondherʌimuch gʌr-i-nch-ʌ.do-PASS-NPST-3s
‘A lot of tea is produced in Nepal.’ (elicitation SAR 2011) In summary, complex predicates in Nepali work similarly to those in Chintang (see section 2.6.5.3). In the case of Nepali, the main argument for treating N as P in the absence of other non-S/A arguments is the obligatoriness of ERG on A. Independently of that, the threshold for marking N with DAT is much higher than the threshold for linking N to O-AGR, to the extent that N-DAT is so far unaested in corpus data. Complex predicates with gʌr- ‘do’ as their V can frequently also use hu- ‘be’ instead. An O besides N becomes S in this process and accordingly cannot be marked by DAT any longer.