2014 Volume 145 Pages 27-62
This paper argues that Japanese benefactive constructions with (-te) kureru/yaru have the five synchronic variants shown in (1):
(1) Type A: X-ga Y-ni Z-o kureru/yaru (MV)
Type B1α: [X-ga Y-nii [Y-nii Z-o V] te-kureru/te-yaru] (AuxV)
Type B1β: [X-ga Y-ni [Z-o V] te-kureru/te-yaru] (AuxV)
Type B2: [X-ga […V ] te-kureru/te-yaru] (AuxV)
Type B3: [[ (X ga) …V ] te-kureru/*te-yaru] (AuxV)
These variants can be explained as products of a grammaticalization process (Type A→Type B1α→Type B1β→Type B2→Type B3) based on four semantic/syntactic criteria. This model in (1) can also be applied to the analysis of the benefactive constructions with GIVE verbs in languages like Korean (cwuta), Chinese (gěi), and Marathi (deNe). I propose the following implicational hierarchy to predict the occurrence of possible benefactive construction types in a given language:
(2) Type A < Type B1α < Type B1β < Type B2 < Type B3