In this paper, we present data from three Norwegian dialect types, NOR-1, NOR-2 and NOR-3, which differ with respect to the verb second (V2) requirement in wh-questions: NOR-1 (represented by Standard Norwegian) requires V2 in all main clauses, NOR-3 (represented by the Nordmøre dialect) lacks this requirement in all wh-questions, while NOR-2 (represented by the Tromsø dialect) lacks the requirement in questions with a short wh-word. Focusing on NOR-2, we will show that the choice of word order (V2 or V3) is dependent on the information status of the subject. We will argue that this can be related to the position subjects have in the IP domain, more specifically that given subjects occur in Spec-AgrSP and new information subjects occur in Spec-TP. Furthermore, based on the split-CP analysis we will account for the word order differences in the three Norwegian dialect types by postulating a parameterized requirement for filling different C heads. The analysis proposed will also account for why the complementizer som is inserted in NOR-2 and NOR-3 whenever the wh-constituent is the subject of the clause. In addition, on the basis of a comparison with Norwegian we will provide an analysis of the English subject/oblique asymmetry in wh-questions, i.e., of the fact that there is no auxiliary inversion/do-support in main clause subject wh-questions.

The absence of V2 in mainclause wh-questions is geographically speaking a very widespread phenomenon in Norwegian, absent only in central eastern and southeastern Norwegian dialects. The characteristics of the phenomenon are nevertheless not uniform across the dialects -three descriptive variables can be discerned: (i) whether or not V2 is required, (ii) whether or not V2 is optional alongside non-V2, and (iii) whether or not non-V2 is allowed with short wh-elements only. In addition these variables are relative to a ±subjecthood distinction on the wh-constituent. The phenomenon has received considerable attention by both dialectologists and theoretical syntacticians, and the main purpose of this paper is to systematize the information available in previous works. On the basis of the resulting overview a microparametric account of the variation will be proposed whereby the variation hinges on three microparameters: (i) whether or not interrogative C must be lexicalized, (ii) whether or not short wh-elements are heads that may lexicalize C, and (iii) whether or not the element som that appears in subject wh-questions is a head or not. Diachronic speculations concerning the development of the variation will also be raised.

The shape of manner how varies to considerable extent across Scandinavian dialects, and many of them quite clearly have a complex internal structure. This paper investigates the internal structure of these many forms, and it will be argued that all forms of complex manner how can be derived from just two abstract templates. Furthermore, it will be shown that the complex forms of manner how unlike the simplex ones interact with other wh-items morpologically speaking: the wh-part will always be identical to the degree wh-item in the various varieties, which in turn will be identical in form to either where or what. The few simplex forms that exist are identical to degree how, but do not share internal structure (apart from wh-) with other wh-items. Moreover, many of the complex forms of manner how are also used as determiners, and a comparison with complex wh-determiners will reveal that at a certain level of abstraction such wh-items share important parts of structure.

Some languages have articles, others have not. This article argues that semantically equivalent noun phrases have the same underlying phrase structure across languages, no matter whether articles exist or not. The core idea is that functional categories are abstract syntactic heads correlated with specific referential categories. Presence vs. absence of a functional category determines whether or not its particular referential properties are found on the noun phrase. The functional heads are found universally and are a part of Universal Grammar, but since they are abstract, they must be`identified': Identification is accomplished when an element with the relevant morphological features is merged within the functional head projection. When the numeration contains several candidates for identification, the one with the highest number of relevant morphological features (with particular qualifications) is chosen. Crucially, articles are on principle available as potential identifiers, and languages that have articles must therefore use them if they are more appropriate identifiers than any of the elements in a given numeration. Having no semantic properties of their own, articles become specialized identifiers for particular functional categories. This theory, Identification Theory, captures important traits of the article systems of various languages, notably those of English, Finnish, French, Icelandic, Mainland Scandinavian, and Northern and Northeastern Swedish`O È sterbotten' dialects.* This paper is a revised and substantially reduced version of part I of Vangsnes (1999).

The aim of this study was to find out how people process the dialectal variation encountered in the daily linguistic input. We conducted an eye tracking study (Visual Word Paradigm) that targeted the online processing of grammatical gender markers. Three different groups of Norwegian speakers took part in the experiment: one group of students from the capital Oslo, and two groups of dialect speakers of the Sogn dialect of Western Norway. One Sogn group was defined as “stable dialect speakers,” and one as “unstable dialect speakers,” based on a background questionnaire. The students participated in two eye tracking experiments each, one conducted in the their own dialect, and one in the other dialect (i.e., Sogn dialect for the Oslo students, and Oslo dialect for the Sogn students). The gender systems in the two dialects differ: the Sogn dialect makes an obligatory three-gender split (Masculine, Feminine, and Neuter) whereas the Oslo dialect only obligatorily makes a two gender distinction. The research question was whether speakers could use gender markers to predict the upcoming target noun in both local and non-local dialect mode, and furthermore, if they correctly could adjust their expectations based on dialect mode. The results showed that the Sogn speakers could predict upcoming linguistic material both in the local and Oslo dialect, but only the stable group were able to adjust their predictions based on the dialect mode. The unstable group applied a more general Oslo-compatible parsing to both the local and the non-local dialect. The Oslo speakers on the other hand were able to use gender markers as predictors only in their own dialect. We argue that the stable Sogn group should be treated as a bilingual group, as they show native-like skills in both varieties, while the unstable Sogn group can be seen as accommodated monolinguals, in that they treat the two varieties as sharing an underspecified grammar. The Oslo group on the other hand lacks sufficient competence in the other dialect to make use of grammatical markers to make predictions.