In this paper I will provide a new argument for post-syntactic morphology. The empirical evidence comes from so-called displaced morphology in German verb clusters, where the non-finite verb form selected by a given governor does not appear on the immediately dependent verb but rather on the linearly last verb of the selector’s complement. The placement of the morphology thus partly depends on linear notions and not exclusively on hierarchical relations. I will provide an analysis within Distributed Morphology (Halle and Marantz in The view from Building 20. Essays in linguistics in honour of Sylvain Bromberger, MIT Press, Cambridge, MA, pp 111–176, 1993), where exponents for non-finite morphology are inserted into separate functional heads which are linearized after their VP-complements. At a late stage of the PF-derivation, the exponents are associated with their verbal hosts by means of Local Dislocation, an operation that applies under adjacency (Embick and Noyer in Linguist Inq 32(4):555–595, 2001). As a consequence, the non-finite morphology always comes last in the selector’s complement. Displacement arises once the immediately dependent verb is not the last verbal element in the complement of its selector; this is generally the case in strictly ascending orders, while in the strictly descending 321 order the morphology is faithfully realized. The placement operation is thus always the same, displacement only emerges as a side effect of certain cluster orders. Further evidence for a post-syntactic approach to the placement of non-finite morphology and against a pre-syntactic perspective comes from the absence of semantic effects under displacement, the emergence of non-finite verb forms specified for more than one non-finite category under multiple displacement and the distribution of default forms.

在本文中，我将为后句法形态学提供一个新的论证。经验证据来自德语动词簇中所谓的移位形态，其中给定调节器选择的非限定动词形式不会出现在直接从属动词上，而是出现在选择器补语的线性最后一个动词上。因此，形态的位置部分取决于线性概念，而不仅仅取决于层次关系。我将在分布式形态学中提供分析（Halle 和 Marantz 在《The view from Building 20. Essays inlinguals in Honor of Sylvain Bromberger》，麻省理工学院出版社，剑桥，马萨诸塞州，第 111–176 页，1993 年），其中非有限形态学的指数被插入到单独的功能头中，这些头在其 VP 补码后被线性化。在 PF 推导的后期，指数通过局部错位与其语言宿主相关联，这是一种在邻接下应用的操作（Embick 和 Noyer in Linguist Inq 32(4):555–595, 2001）。因此，非有限形态总是在选择器的补集中排在最后。一旦直接从属动词不是其选择器补语中的最后一个动词元素，就会出现置换；在严格升序中通常是这种情况，而在严格降序321中则忠实地实现了形态。因此，放置操作始终相同，位移仅作为某些集群顺序的副作用出现。后句法方法放置非有限形态并反对前句法观点的进一步证据来自于位移下语义效果的缺失、为多个非有限类别指定的非有限动词形式的出现在多重位移和默认形式的分布下。