The goal of this talk is twofold. Using empirical data from Moksha, I first show that both locative cases and locative postpositions can represent the generalized category P. Second, I argue that the complement of the P head does not have to be a full DP; some P heads can only take bare NPs as complements. The language I investigate here (Moksha; Finno-Ugric, Uralic) has rich morphology. Moksha nouns inflect for number, possession, definiteness and case, and these categories interact in a non-trivial way. For instance, the affix order in Nominative, Genitive and Dative (the cases I refer to as grammatical (structural + inherent), as opposed to lexical) is ‑POSS‑CASE, whereas in other forms, such as locative cases, the order is ‑CASE‑POSS. Furthermore, number and definiteness can be expressed only in the grammatical cases. I account for this difference by analyzing lexical cases separate from grammatical cases, as instances of the category P. Similarities between locative cases and locative adpositions are well-known; however, PPs are always considered to be larger than DPs, as it is assumed that Ps take DPs as their complements. I show that this is not always true. In some languages, some Ps cannot take full DPs as their complements, and take smaller phrases instead. Thus, my account of the morphology of Moksha locative cases is rooted in syntax; I then compare this syntactic with possible morphological analyses. If my analysis is on the right track, we should expect cross-linguistic and intra-linguistic variation with respect to the size of a complement Ps can take.