Abstract The serialization patterns of result predicates correlate in aspecific manner with headedness: in VO languages, resultpredicates are found potentially in two different positions, a V-adjacent one and a position at the end of the VP. In OVlanguages, the position of result predicates is invariably V-adjacent. The serialization of depictive predicates with respectto result predicates is a function of both a contiguityrequirement between the depictive predicate and its target, andthe identification parameter for argument positions. Theimplications of the analysis of the factors that determine theserialization of predicates provide crucial evidence for atheoretical issue. It is argued that headedness (head-initial vs.head-final) is best accounted for in terms of head chains ratherthan in terms of evacuated head-initial projections.