This paper argues that approximative numerals in German such as an die zwanzig ‘up to twenty’ are not numerals contained in a PP, as suggested by Corver & Zwarts (2006). Such numerals are compared with regular PPs and shown to never pattern with PPs. Rather, they pattern with numerals modified by an adverb like ungefähr zwanzig ‘approximately twenty.’ The words appearing before the numeral in approximative numerals such as an die ‘up to’ are hence adverbs, contrary to surface appearance. This result is further supported by the observation that in expressions like an die ‘up to’ neither the alleged preposition behaves like a preposition nor does the alleged definite determiner behave like a definite determiner.

No abstract

No abstract

This paper offers an argument for the necessity of adopting grammatical relations as primitives in syntactic theory. The argument is based on the phenomenon of wh-copying in German. Wh-copying in German poses a problem for approaches based on phrase structure (PS) representations because the construction is governed by two generalizations which a PS approach fails to capture. As soon as a relational perspective on syntactic structures is adopted, however, the generalizations can be captured. I will present an analysis for whcopying in German within the Arc Pair Grammar framework, which does adopt such a relational view. It will be shown that the operation Replace in interaction with other principles of that framework successfully captures the two generalizations of wh-copying in German, and that it eventually even allows one to reduce the two generalizations to a single one.

This introduction proposes to investigate mismatches and indeterminacies in languages much more than has hitherto been done. Such seemingly unruly aspects of language(s), it is argued, are interesting since they may help shed light on the internal make-up of grammatical systems. The question of the internal make-up of grammar(s), it is argued, cannot be addressed by the normal modus operandi of linguistic research, which is to find matches (rather than mismatches) between the observable (sound and meaning) interface systems, and to find how the interface representations map unto each other deterministically: It is only in the “lo-fi” aspects of mappings that the internal mechanisms of the overall grammatical architecture may reveal themselves.The introduction also points out that our concern is independent of the various theoretical orientations linguists may choose for their work, since the problem presents itself in all approaches to language research currently available, it seems – if in slightly different ways.We propose, in sum, that mismatches and indeterminacies are an extremely worthwhile field for future linguistic research, and one that should be on the agenda (or minimally, within the field of view) for linguists of all theoretical convictions.