The compositional nature of the passive
It is claimed that in languages without a synthetic passive verbal morphology, the passive diathesis is compositional derived by implication or radical underspecification rather than generated in a direct way. As finitizing auxiliaries, equivalents of <i>BE/ESSERE </i>or <i>HAVE/HAB¯ERE </i>are employed together with the participial ANTERIOR (‘past’) morpheme. A distinction needs to be made between ‘passive sense’ and ‘passive denotation’. The claimis that languages such as German, along with numerous other languages, may be seen to provide a ‘passive’ by derivation from the complex ‘AUX+past participle’. The main body of the paper consists of evidence for the assumption that no direct passive meaning is provided by this complex ‘AUX+Anterior participle’, foremost from German. The past participle form has more than one function, an active and a passive one. The claim is that the past participle is void of any diathetic/ voice denotation except for the categorial status of adjectival. The latter allows the inference of a number of syntactic properties, i.e., most prominently a none-agentive external argument which, in syntactic collocation with the monovalent copulas <i>sein </i>and <i>werden</i>, contributes to the passive sense. A further distinction will be made between two different past participle lexicals: one on the basis of aspectual perfectivity implying an ‘approach phase’, antecedent to a result phase and presupposing an agentive external argument; and another, imperfective PP, which is amenable to the presupposition of such an ‘approach phase’ by way of pragmatic implicature. This distinction will be spelled in some detail in terms of formal semantics. Evidence for the formal and empirical correctness of this approach will be drawn from various languages similar to German in some, but not all, respects: Russian, Swedish, Dutch, and Latin. Turning to passive syntax it will be shown that the mere assumption of a passive morpho-syntax to be represented directly in UG-terms will not do justice to the evidence provided by the data and their distributional properties. Rather, it will be held and shown in some detail that passives are to be distinguished in accordance with their derivational basis as perfectives or imperfectives, not only as lexicals but also in terms of true clausal aspectual phrases (e.g., as regards directional vs. non-directional adverbs). Ergatives, commonly held to be a verbal class in their own right, will turn out as belonging to the perfective type of voice. <br /> In essence, this paper wants to demonstrate that there are principally two pathways toward a solution for the undecided semantics of the past participle forms: the ‘pragmatic’ solution presented in Section 4.5., where the fundamental semantics of the participle form is taken to be past without any voice reading. In Section 10, on the other hand, it is assumed that the participle formis a categorially underspecified ‘root’ in terms of Distributed Morphology reaching a specific reading only in an extended morphosyntactic context.