This paper contributes to an ongoing debate on the syntactic status of oblique subject-like NPs in the ‘impersonal’ construction (of the type me-thinks) in Old Germanic. The debate is caused by the lack of canonical subject case marking in such NPs. It has been argued that these NPs are syntactic objects, but we provide evidence for their subject status, as in Modern Icelandic and Faroese. Thus, we argue that the syntactic status of the oblique subject-like NPs has not changed at all from object status to subject status, contra standard claims in the literature. Our evidence stems from Old Icelandic, but the analysis has implications for the other old Germanic languages as well. However, a change from non-canonical to canonical subject case marking (‘Nominative Sickness’) has affected all the Germanic languages to a varying degree.

The Insular Scandinavian (IS) languages, Icelandic and Faroese, exhibit variation in subject case marking, primarily pertaining to an older idiosyncratic case and an innovative case that instantiates a regular pattern. This variation occurs to a considerable degree at the level of the individual speaker. We argue that this intra-speaker variation in IS involves optionality within one grammar and not competition between two different grammars or dialects in the sense of Kroch (1989Kroch ( , 1994. Mainly on the basis of data from the socalled Dative Substitution (DS) in Icelandic, we show that this variation does not involve any kind of parameter, and that it is not the result of dialect contact, as the grammar competition analysis would entail.

Syntactic reconstruction has long been virtually outlawed in historical-comparative research, more or less ever since Watkins's influential works on the problems of reconstructing word order for Proto-Indo-European. Recently, through the emergence of Construction Grammar, where complex syntactic structures are regarded as formfunction pairings, a resurgence of syntactic reconstruction is made possible, as complex syntactic structures become a legitimate object of the Comparative Method. Given the legitimacy of syntactic reconstruction, and hence the possible reconstruction of argument-structure constructions, a major question arises as to whether grammatical relations are also reconstructable for earlier undocumented language periods. We argue that if the constructions singling out grammatical relations can be reconstructed for a proto-branch, the grammatical relations following from these are also reconstructable for that proto-branch. In order to illustrate our methodology, we show how a reconstruction of the subject function in Proto-Germanic may be carried out, more specifically of oblique-subject predicates like 'hunger', 'thirst' and 'lust', based on the subject properties found in the earliest Germanic daughter languages.

No abstract

No abstract