The tenability of the regression hypothesis as a basic principle for language attrition within languages has not been really questioned until now. In order to find out to what extent attrition might be the reverse of the process of acquisition, phenomena of attrition in the use of the German case system were studied. Two hypotheses were tested: the Linguistic Hypothesis and the Cognitive Hypothesis. the Linguistic Hypothesis is based on the notion of regression. According to this notion, attrition is the reverse of the language acquisition process. The Cognitive Hypothesis is based on the assumption that in natural cases of language acquisition, there is a tendency to establish a one-to-one correspondence between cognitive function and morphological case assignment. The results from experiments on case marking in headline-type constructions showed differences between native (L1) speakers and second language (L2) learners. In L1 speakers the relation between case marking and underlying semantic functioning becomes more prominent, whereas in L2 learners the nominative is used as a default case, indicating that morphological differentiations become reduced. This means that for L1 speakers the Cognitive Hypothesis provides a more adequate explanatory framework, whereas for L2 learners it is the Linguistic Hypothesis that is more appropriate.