This publication is in the INCUBATOR section.
Roland Schäfer (in preparation). The limits of predictability: sequences of oblique attributive adjectives in German. To be submitted in Q3 or Q4 of 2018.
Full dataset and scripts on GitHub. (To be released after paper has been accepted.)
In this paper, I discuss the inflection of sequences of oblique attributive adjectives in German, most notably in the masculine/neuter dative singular, where affixes alternate when there is no determiner.
- mit heiß-em (KONJ) lecker-en Kaffee
- mit heiß-em (KONJ) lecker-em Kaffee
For KONJ, writers use either a comma, the conjunction und (‘and’), or the conjunction or (‘or’).
Under the perspective of alternation research, I test factors influencing the choices predicted by existing theories in a large corpus study and a sequence of experiments. Based on the problematic results of all of these studies, I argue that the idea that all alternation phenomena should be strongly predictable by taking into account a (potentially large) set of influencing variables (be they contextual, morpho-syntactic, or lexical) is absurd. Rather, individual speaker-hearers and/or writer-readers (at least in the case at hand) use available cues to mark/decode syntactic structures. If the marking/decoding is not necessary to arrive at a sufficiently precise syntactic representation, diffuse alternations result, which cannot be strongly predicted.
See also my talk Parallel- und Wechselflexion in obliken Adjektivreihen – Eine empirische Vervollständigung at IDS Mannheim on 31 May 2017. It represents an earlier stage of my view of the phenomenon.