The force of lexical case: German and Icelandic compared

In this paper, I compare some constructions of German and Icelandic in which lexically marked verbs behave differently, and which therefore have attracted the interest of linguists for many years. Particularly challenging are the Dative-Nominative (experiencer) verbs, in which the higher argument is marked by Dative and the lower receives Nominative. In German, it is the Nominative argument that can be controlled and must be raised to object (thereby inducing multiple raising), whereas in Icelandic it is the Dative argument that can be controlled and must be raised to object (thereby leaving the Nominative argument as part of the infinitive clause). The classical account claims that only subjects can be controlled and raised to object. I take this as a generalization that implies too strong a notion of subject. There are some aspects of the constructions involved (for instance, the fact that German allows multiple raising, and that Icelandic allows a Nominative argument not to agree with the finite verb) that call for a more differentiated approach. In dispensing with the notion of subject, I will follow the idea that it is rather a set of ordered constraints that determines the behavior of lexical items. The lexical differences between German and Icelandic (including the possible case patterns associated with verbs) turn out to be marginal. I will propose that instead, all major differences between the two languages come about by different rankings of only a few syntactic