Linguistische Berichte Heft 184

There are some linguistic expressions in German which help speaker and hearer to organize comprehension. One of these expressions is ach so. The article intends to give an analysis of ach so in dialogues and so-called monologues. The thesis is that ach so is used in order to manage the (sudden) change from divergence to convergence in discourse. Analysed separately the interjection ach and the adverb so can show how this sudden change works in detail. The analysis leads to some more general assumptions about how knowledge is processed in discourse.

We provide an overview of research on the syntactic comprehension deficits of English and German speaking agrammatic Broca's aphasic patients. Following a brief introduction of early work on this topic during the first half of the century, we concentrate, mainly, on studies carried out within linguistically informed frameworks. We argue that the trace deletion hypothesis ( e.g., Grodzinsky 1995b) and derivatives of this interpretation of agrammatic comprehension ( e.g., Hickok and Avrutin 1995) cannot account for the variety of performance patterns agrammatic Broca's aphasics display. Other theories such as the hypothesis in terms of impairments ofCase (Druks and Marshall 1995) however are able to account for a wider range of deficits.

This paper is concemed with the so called unmarked (in the sense of Höhle 1982) or basic word order in German. I want to argue that neither is it the case that the unrnarked linear order of the arguments in a German sentence depends on the individual verb ( contra Haider 1992, Haider 1993) nor is the unmarked word order derived by S-structure movement from a fixed base generated order (contra Müller 1999a). Instead I would like to propose that the unmarked word order is the result of a process of optimisation at D-structure. The approach is formulated within the framework of Optimality Theory (cf. Prince & Smolensky 1993, McCarthy & Prince 1993) and is based on the analysis ofthe dative as a Jexical case that has been proposed by Vogel & Steinbach (1998).

In this paper I discuss some syntactic and selectional problems conceming dependent clauses with verb-second structure in German. These clauses occur as propositional arguments after certain classes of matrix verbs or adjectives. They alternate systematically with verb-final clauses introduced by a complementizer (daß or wenn). In generative accounts of these data, both realizations of the propositional argument have usually been treated as instances of syntactic complementation, i.e. both types are generated in a structural complement position. But dependent verb-second clauses display syntactic properties that are at odds with such an analysis. There is strong empirical evidence against the complement status. Another perspective on the same problem is the question whether dependent verb-second clauses are selected by means of lexical information. This question arises, because syntactic configurations are supposed to represent the argument structure of lexical items in an unambiguous way. I will provide evidence that there are in fact restrictions for the realization of verbsecond clauses that cannot be part of the information of a lexical entry under common assumptions about the lexicon. Dependent verb-second clauses are therefore neither syntactic complements nor lexically selected objects and present serious diffculties for an account within a generative framework.

Part-of-speech categories are generally expected to fulfil two functions: In the lexicon they are expected to determine which elements belong to a given part of speech and in the grammar they are expected to determine which positions elements of a given part of speech may occupy in complex linguistic expressions. This paper argues that part-of-speech categories are unable to fulfil both functions and that two completely different kinds of linguistic categorization need to be distinguished.