​The research reported here concerns the nature and structure of constructions called complex predicates widely found in South Asian languages. The strategy chosen is to explore domain delimitation questions for compound verbs (CVs) or aspectual complex predicates (ACP) and the conjunct verb/ ‘N-V’ constructions in Hindi-Urdu, esp. in ordinary speech and writing and in published prose. The study focuses on the V-V/ compound verb constructions in Hindi-Urdu and compares the behaviour of the light verbs found in them with that of the light verbs found in N-V constructions. The asymmetries observed in the behaviour of light verbs/ vectors in V-V constructions over against those in N-V constructions with respect to (echo-)reduplication and indirect causativization give us both empirical and theoretical grounds to claim an alternative analysis of the structure of the complex predicates. Our investigation of the ‘polar+vector’- type of V-V constructions reveals that they are in fact generated in the lexicon by morphological mechanisms, and that the widely accepted assumption that operations in the syntax alone can underwrite a comprehensive account of these constructions is inadequate. The N-V constructions may be seen as a case of an interaction between a nominal and a light verb, where the latter can be treated as an occurrence of an unreduced, regular, lexical verb, albeit with distinctive semantic properties in keeping with the context. This study argues that the nature and status of light verbs in the N-V constructions may be explained by invoking polysemy. These true light verbs, it is claimed, sharply differ from vectors, the so-called ‘light verbs’ that occur in Hindi-Urdu V-V constructions. Given these sharp differences, it is recommended that we reserve the term vectors for the ‘light verbs’ in the V-V constructions of the ‘polar-vector’-type and use the term light verb only for the verbal predicate in the conjunct verb/ N-V constructions. This investigation makes use of empirical data from 18 books written by 15 eminent writers of Hindi-Urdu to corroborate the claims made here and to counter the claims made in certain other studies. In addition, the study makes use of questionnaire-elicited speaker judgements and Google search results to fortify its claims. This study also flags a construction, labelled ‘bare conjunctive’, which shares a few features with ‘conjoined main verbs’ but shows distinctive syntactic and pragmatic properties. After arguing for drawing a sharp boundary between the light verbs in N-V constructions and what have been called ‘light verbs’ in V-V constructions, this study offers a morphological analysis of the ‘polar-vector’-type of V-V constructions in terms of the WWM (Whole Word Morphology) model, the machinery of analysis that the substantivist framework in linguistics uses for its morphological component. Finally, the study suggests repositioning the light verb category and the vector category on the cline of grammaticalization, given that entities of the two kinds behave differently with respect to morphology and syntax and call for separate methods of analysis. While a vector is deemed devoid of individually attributable semantics, a light verb does have predicational capacities, which play a role in shaping the semantic make-up of the clause, combining its own contribution with that of its nominal partner.