Deconstructing possession

The paper argues that clausal possession is to be decomposed into three distinct, independently attested, syntactic configurations, each associated with its own meaning. These include Location, represented as an ordinary small clause, the Part-Whole relation, which always has a complement structure within DP as its source, and an applicative structure ApplP, the source of (in)alienable possession, where humans are treated as special. The analysis we propose focuses on Palestinian Arabic and extends to English clausal possession and its realizations across have and be. Palestinian Arabic overtly distinguishes a number of ingredients which in other languages enter into possession less transparently: It marks Location and Part-Whole relations by distinct prepositions, it features a full-agreement/no-agreement distinction associated with scope, and, lacking have, it keeps separate P° and be, the ingredients often assumed to enter into its composition. The picture which emerges is partly familiar and partly new. We argue that the notion possession is never linguistically encoded as such, since none of the underlying representations proposed is associated exclusively with possession. We also argue that the subject in possessive clauses is a derived subject with both have and be. We attribute the differences between Palestinian Arabic and English to a difference in their agreement systems, which in conjunction with Economy, forces P° to extract from its PP, and leads to the formation of have. If we are correct, the cross-linguistic distribution of have and be may further reduce to parametric differences in agreement systems.