The position of adverbials in German clause structure: on the interaction of global and specific constraints

最近的研究认为，德语状语（PP，ADV）的顺序限制可以在特定的条件下为不同的状语类制定。句法论元仅作为锚点相关，因此可以识别状语位置。虽然有些研究者假设了实际的基本位置，但也有人认为顺序限制只是相对的，即在不同类别的状语之间。我们得出的结论是，状语的顺序制约应该被定义为状语特有的制约之间的相互作用（例如，事件内部状语的回指条件），以及全局约束，即对论元和非论元同样适用的约束（例如，[+animate]先于[-animate]的条件，这些研究还为事件外部状语和过程状语提供了新的研究领域。关于前者，项目第一阶段的调查对状语类的概念本身提出了质疑，因为各个成员表现出不同的语序偏好，并且在主语和宾语的前后出现各种各样的位置。一个可能的原因可能是所选择的类成员（时间PP和伴随情况的PP）不能被归类为原型，另一个原因可能与所使用的实验设计有关。因此，我们提出了一个双重扩展的分析：一方面是扩展的状语类成员，另一方面是引入一个实验设计，允许确定偏好方面的主题和对象在同一设置。对过程状语的考察表明，对过程状语的不同实现形式（PP、副词）进行综合分析是必要的。有趣的是，研究传统集中在副词上，而我们研究的是PP。因此，在本研究的第二阶段，我们提出了一个与过程相关的PP和副词的综合处理。通过这种方式，我们也能够解决一个理论问题，即使用的操作，产生不同的分析链相同的结构。这样的操作阻碍了经验验证策略，因为它可能会提出“总是另一个结构”在这里是相关的。这一问题不仅存在于过程状语的分析中，也存在于事件状语的分析中。结合不同类别的诊断，我们提出了一个实验设计，不仅要跨越链相同操作的证伪差距，但也产生了一个统一的分析的基础上的顺序约束的状语类的整体和具体的限制。

Recent research assumes that order constraints for adverbials (PP, ADV) in German can be formulated in specific conditions for different adverbial classes. Syntactic arguments are only relevant as anchors, so that adverbial positions can be identified. While some researchers assume actual base positions, others propose that order constraints are only imposed relatively, i.e. between adverbials of different classes.Through the theoretical, corpus-based and experimental investigations in the first stage of this project, we have come to the conclusion that order constraints for adverbials should be defined as interactions between constraints specific for adverbials (as e.g. an anaphoricity condition on event-internal adverbials), and global constraints, i.e. constraints that hold equally well for arguments and adjuncts (as e.g. the condition that [+animate] precedes [–animate], which would hold for the nominal complement of an adverbial preposition).The investigations carried out also suggest new research venues for event-external and process-related adverbials. With regard to the former, the investigations of the first project stage cast doubt on the very concept of the adverbial class, since individual members show different word order preferences, and also appear in a variety of positions in front of and behind the subject and object. A possible reason might be that the class members chosen (temporal PPs and PPs of concomittant circumstance) cannot be classified as prototypes, another reason might have to do with the experimental design used. We thus propose a twofold extension of the analysis: on the one hand an extension of the adverbial class members, and on the other the introduction of an experimental design which allows the determination of preference with regard to the subject and object in the same setting. The investigation of process-related adverbials makes it clear that the different realisations (PPs, adverbs) must be included in a comprehensive analysis. Interestingly, the research tradition has focused on adverbs, while we have looked at PPs. For the second stage of this project, we thus propose an integrated treatment of process-related PPs and adverbs. In this way, we are also able to address a theoretical problem, viz. the use of operations, which yield different analyses for chain-identical structures. Such operations block empirical falisification strategies, because it might be possible to propose that it is 'always the other structure' that is relevant here. This problem is present not only in the analysis of process-related but also of event-external adverbials. Combining the diagnostics developed for the different classes, we are proposing an experimental design which should not only get across the falisification gap of chain-identical operations, but also yields a unified analysis of the order constraints for the adverbial classes on the basis of global and specific constraints.