Professional Documents
Culture Documents
Introduction:
The -kar particle primarily features as the main verb in Complex Predicate constructions, which
is a productive syntactic phenomenon in Hindi. It can form predicates with nouns, adjectives,
and with other verbs. There are two distinct types of V-V compound relationships the particle
establishes, one where it retains its original semantics as the main verb, the other where it adopts
‘converbial’ interpretation.
As is evident from the translation, (4) is to be distinguished from (1), (2) and (3). (4) depicts a
series of events in succession, with one subject being shared across different clauses; it exhibits a
macro-event, (a morning ritual), with each clause describing one event in the chain. Moreover,
the utterance has the additional semantics of ‘correlational coherence’ brought to light by the
usage of -kar, not as a light verb but as a clause chaining device instead. This ‘clause-chaining’
function will be discussed at length later in the paper. (5) shows an instance of -kar having two
(5)
Other functions of -kar include (6), where (as noted by Davison 1985), the -kar clause lends a
conditional interpretation to the utterance, especially if the main verb carries future tense.
(9) Yah Baat Sun kar Bhi usei Krodh Nahi Aya
This Matter Hear Also He.DAT Anger Not come.perf
‘Even though he heard this matter, he did not get angry.’
(Davison 1981:112)
Inclusive particles (like also) when used alongside the -kar particle, lend a concessive
This section intends to focus on the clause chaining function of the -kar particle. The
phenomenon of clause chaining is most commonly associated with West African Languages (eg.
Osam 1994) and has recently been worked on in Indo Aryan Languages like Oriya (Beermann &
Hellan 2002). A clause chain involves a series of (VP) clauses bearing a sequential relationship
with one another. All adjoined clauses barring one are non-finite, and the tense feature of the one
fully inflected verb determines the tense feature of the utterance as a whole. (Payne 1997: 321;
Kroeger 2004:242-250) (see Longacre (1985: 263-83) for more on chaining). Similar to (8) other
(10)
(11)
(12)
(13)
least in Hindi), object sharing however, is not. The PRO subject of the -kar clause is obligatorily
controlled by the matrix subject. In addition, as illustrated in (11), and (12), each verb in the
sequence has its own direct object and/or adverbial modifier, i.e, their own independent object
domain.
(14). a.
The above sentences have all been taken from Bhatia (1978), who claims that (14b) and (15b)
are ill-formed while (16b) is not. He claims that this difference is based on world knowledge
rather than grammatical knowledge. However, we can see that (16b) is only considered
acceptable because of its concessive/instead of nature. Negation is not possible with the non
finite -kar clauses. Additionally, negating the main verb leads to ambiguity as illustrated in (17).
(17)
According to Subbarao 1996, (17) is interpreted as (i) when the matrix verb is in the scope of the
negative, and as (ii) when the negative has scope only over the non-finite -kar clause.
As for interaction of interrogatives with VP chains, the position of the question particle
(18)
In (18) it the washing of the mangoes which is being questioned, rather than the eating. (19) on
(19)
The construction takes on two forms which are identical in their semantics- ‘kar/ke’,
‘Padhkar/Padhke’ both are translated as ‘Having read’. Grierson (ed. 1885: 65) provides an
…‘Conjunctive Particle is formed by adding ‘ke’ to the 1st verbal noun. Thus dekhke, deke,
payke. The word ke commonly called the conjunctive part of ‘kar’ is in reality no such thing. It is
derived independently from the Vaidik ‘karya’ (Skr. Kritva), hence Pr. ‘kria’ (Hem Cha. IV,
271) and Bihari (with elision of r) ke. Hence dekhke, means ‘having done the action of seeing’…
As for the grammatical category, other diachronic research (Hook 1991, Tikkanen 1987,
Hendriksen 1944, Chatterji 1926) lead us to believe that the modern V-V complex predicate in
Hindi-Urdu had its origins in the Sanskrit “gerund” or “absolutive” suffixes as in -tv¯a(ya), or -
Whitney 1889:345–360) more recently, termed as conjunctive participle (CP). Davison (1986)
claims that syntactically, -kar functions as both an affix which carries perfective aspect and as a
clause connective, whereas ‘the semantic content of -kar is perfective, very general aspectual
meaning, rather than some collection of aspectual, temporal, causal, conditional or manner
adverbial meaning’
The conundrum, here, lies in the fact that there is no consensus about how to classify such
constructions. While some have categorized the -kar clause as an affix carrying perfective aspect
like (Davison 1979,1981) others (Abbi 1984, Davison 1981, Kachru 1981, Kellogg 1938) have
In recent years, ‘converbs’, (“verbal adverbs” (Hasplemath 1995: 4)) have been recognized as a
cross linguistically valid grammatical category. Properties of the particle (which usually occurs
as a verbal affix) include an absence of tense and agreement features, dependency on a finite
verb in the same utterance, clause linking, and sequencing of events (Bickel 1998, Genetti 2005,
Hasplemath and König 1995, Masica 1991). Bickel 1998 attributes an additional
‘narrative/clause chaining’ function to converbs found in some Asian languages, which their
European counterparts lack. All the aforementioned properties can be attributed to -kar which
subordinating device, like that in Davison (1979, 1981). Clause-chains are neither subordinate
nor coordinate in nature. Unlike coordination, VP chains involve an asymmetry in the verbs
involved, with only one verb being inflected, while the others remain ‘frozen’ in past tense,
coordinates is possible which would not have been the case if it was a true instance of
(20)
Yah Vo Machli Hai Jisko mene dhokar Chaval banakar saath khaya
This 3ps.REL Fish is.PRES which.REL I.NOM wash Rice Cook together Ate.PERF
‘It is this fish that I washed, then after making rice, ate (the rice and fish) together.’
As for evidence against true subordination, in (10) for example, there is no instance of
embedding at all simply because neither clause fulfills any argument role for the other.
Nonetheless, there exists a dependency relationship between the VP chains because the tense
interpretation of the non-finite verbs is determined by the final finite verb, which does not
constitute as complete subordination. Foley and Van Valin (1984: 240-263), following Olson
(1981) put forth an interesting distinction between embedded dependent clauses (subordinate
clauses) and non-dependent embedded clauses which they term ‘cosubordination’. The
relationship between the linked clauses in the VP chains is classified under the same. (Foley and
Van Valin’s 1984, more recently Good 2003). A detailed discussion on cosubordination is
Another distinct line of analysis is that of the ‘-iccu’ particle in Dravidian Malyalam as a Serial
Verb Construction. This is similar to Beermann and Hellan 2002 for fellow Indo-Aryan language
Oriya, which features a clause chaining construction with ‘-i’ (glossed as ‘dependent marker’)
comparable to -kar in Hindi. The authors classify the constriction as a subtype of Serial Verb
Beermann and Hellan 2002. Osam 1994 and Kroeger 2001 provide more details concerning the
This analysis is also deemed plausible for the -kar construction, based on the characteristics of
“…(1) the compound verb expresses perfective aspect and is the marked member of a privative
opposition perfective-imperfective (2) one of the functions of the compound verb is to express
the completion of one action relative to the completion of another (3) if a verb is stative then it is
Butt’s ideas of complex predicates involve a grammatical complex structure (with two or more
semantic heads) and the grammatical function structure as that of a simple predicate (single
subject and no embedding). Again, all the features outlined above can be attributed to the -kar
clausal construction.
5. Analysis
I intend to follow a similar line of analysis for the -kar construction in Hindi, with it being
classified as a non finite VP functioning as an adjunct to the matrix V’ which is similar to the
(21)
Seema daud-kar aai.
Subj.ERG Run Come.PERF
Seema came running.
Figure 2.
We can extend the same structure to constructions like (11) (repeated below), with a slight
modification. Since VP-chains are independent predicates, which function as a series of adjuncts
that modify a single finite verbal head, we will posit the existence of a INFL node as well, one
that is non-finite, lacking tense and agreement (following Jayseelan (2017) and Beermann and
Hellan (2002)).
(11)
Mohan Kitab padh-kar Lekh Likhega
Subj Book Read Essay wrote.Fut.Masc.
‘After reading the book, Mohan will write an essay.’
Figure 3.
As for sentences like (22) which exhibit the phenomenon of ‘object sharing’, we assume that the
SVC construction has an empty category pro (similar to the ‘argument sharing’ in the analysis of
Ewe by Collins (1997) since Hindi has both null subjects and null objects.
(22)
Figure 4.
6. References
• Beermann, D., K. Sahoo, and L. Hellan . 2001. What is ‘Argument Sharing’? A case
pragmatic analysis for the Hindi conjunctive participle. In: Cole Peter (ed), 1981. Radical
• Dwarikesh, Dwarika P. 1971. The historical syntax of the conjunctive participial phrase
10.1075/sl.27.1.05goo.
• Hook, Peter Edwin. 1974. The Compound Verb in Hindi. Center for South and Southeast
• Kachru, Yamuna. 2006. Hindi. (London Oriental and African language library, 12.)
Philadelphia: Benjamins