The aim of the paper is to account for the facts of the prepositional passive or pseudopassive (P-passive) without assuming the morphological process of verb-preposition (V-P) reanalysis or preposition incorporation. Although this process has been a standard assumption in LFG since Bresnan 1982 as a requirement for passivization to apply to an intransitive verb with a prepositional complement, the evidence presented by Postal 1986, Baltin and Postal 1996, and Lødrup 1991 against it is so compelling that it is necessary to reject V-P reanalysis. An alternative is possible that assumes that P-passive is a raising construction in which the object of the preposition is structure-shared with the subject of the passive verb. This means P-passive comes within the scope of the theory of Structure-Sharing (Alsina 2008), designed to explain the facts of constructions involving structure-sharing: long-distance dependencies (LDD) and raising. Thus, the main features of P-passives follow from this theory without additional assumptions.
Parameters of variation are proposed to explain the observation that there are three types of languages with respect to the possibility of preposition stranding (P-stranding) and a fourth logical possibility is missing. There are languages where P-stranding comes as a result of both LDD and P-passive (e.g., English), languages without P-stranding of any kind (e.g., French and other Romance languages), and languages with P-stranding arising from LDD, but not from passivization (e.g., Icelandic and Swedish). The proposed parameters allow for these three types of languages, while excluding the fourth, a language with P-stranding in passives but not in LDD. Explaining this dependency between the two phenomena (LDD and P-passive) is possible only in a theory such as the present, where they are both subject to the theory of Structure-Sharing.