Morphological case marking is one of several vulnerable domains in heritage languages and has been reported in a wide variety of heritage languages in various contact situations. As recently as 2015, however, research on heritage Icelandic has maintained that case marking is largely intact, a claim largely based on anecdotal evidence. The maintenance of case marking in heritage Icelandic is an outlier in an overall trend towards case loss in heritage languages:
what makes it different? The answer may lie in a proposed typological distinction based on event semantics (Ritter & Rosen, 2000), with roots in a formal comparison of case and subjecthood in German and Icelandic (Zaenen, Maling and Thrainsson, 1985). Ritter & Rosen propose that Icelandic is a so-called “initiation language,” which prioritizes the thematic roles of an event when assigning case. For instance, given the sentence “The key opened the door,” an
initiation language cannot assign nominative case to the noun phrase “the key,” because a key cannot function as the agent of opening, only as the instrument. In contrast, German and English are what Ritter & Rosen term “delimitation languages,” that primarily utilize structural case assignment, regardless of thematic roles. Many of the world’s languages, according to Ritter & Rosen, are delimitation languages.
I hypothesize that the initiation/delimitation typology results in the erosion case morphology when languages of the same type (“i” or “d”) come into contact, and the maintenance of case morphology when languages of differing types meet. In this talk, I present data from a systematic study of the gender and case paradigms in heritage German and Icelandic in contact with North American English. The results are largely consistent with previous findings. Case in heritage German is largely lost, and case in heritage Icelandic is largely maintained, however Icelandic case is not as robust against erosion as previously claimed. I then connect this data to a formal model of information sharing between argument structure, grammatical functions, morphology and syntax. The data raise the question of at what point in the language production process morphology is accessed: before or after grammatical functions have been assigned?