This special issue of the Nordic Journal of Linguistics is devoted to Comparative Nordic Semantics. Whereas much research has been carried out on comparative syntax, morphology and phonology in the Nordic languages, much less work has been done on the comparative semantics of these languages. But the fact that some of the Nordic languages, namely the Scandinavian ones, Danish, Faroese, Icelandic, Norwegian and Swedish, are historically, lexically and structurally very similar means that they provide an interesting target for semantic research. Are there systematic semantic differences between these languages? If so, are the formal semantic analytic tools that have been developed mainly for English and German sufficiently fine-grained to account for the differences among the Scandinavian languages? These were some of the questions asked in the research project Comparative Semantics for Nordic Languages (NORDSEM), which was funded by the Joint Committee of the Nordic Research Councils for the Humanities in 1998–2001 and which involved researchers at the Copenhagen Business School, Göteborg University and the University of Oslo. Two of the papers in this issue (by Carl Vikner and Kjell Johan Sæbø) derive directly from the NORDSEM project whereas the third paper, by Erich Round, pursues some issues investigated during the project by Joakim Nivre and published in Nordic Journal of Linguistics 25:1 (2002).

This paper deals with Scandinavian sentences where the predicative adjective apparently disagrees with its subject, such as pannekaker er godt ‘pancakes-pl. is good-neut.sg.’, vodka er sunt ‘vodka-masc. is healthy-neut.sg.’. In this paper, the use of neuter in such sentences is seen as a case of semantic agreement. Thereby, pancake sentences comply with Corbett's (1979) agreement hierarchy. The controllers in pancake sentences are low on the individuation scale. This is the reason why they are neuters. The controllers in pancake sentences are mirror-images of the controllers in more well-known examples of semantic agreement. Not only morphosyntax but also semantics, pragmatics and discourse play a role in agreement.

Synchronic alternations involving similar types of consonant/consonant metathesis are examined in Faroese and Lithuanian from both formal and functional perspectives. It is argued that by taking into account speech perception-based phonotactic constraints, we are able to gain insight into why metathesis occurs. The proposed formal account of the process is of particular theoretical interest given the apparent opacity involving the interaction of metathesis and place assimilation in Faroese. While straightforward in a serial-based account, this type of opacity has posed a challenge to Optimality Theory requiring the inclusion of additional formal devices into the theory. As we show, however, by modifying one assumption concerning the realization of the sequence, this additional formal machinery is rendered unnecessary. Opacity involving Faroese metathesis turns out to be an artifact of a particular phonological analysis.

The system of temporal connectives in Scandinavian exhibits an interesting variation in that Danish, like e.g. German, is a two-‘when’ language, i.e. it has two temporal connectives that have divided between them the semantic area covered in English by the single connective when. One of the two Danish connectives (da) is restricted to past episodic clauses, while the other one (når) may be used in past and present habitual clauses and in future clauses. Swedish, on the other hand, like e.g. English, is a one-‘when’ language: it has only one temporal connective corresponding to the two Danish ones, whereas Norwegian presents an intermediate situation, possibly a stage in the development from a two-‘when’ to a one-‘when’ system. This paper proposes a semantic analysis of the two ‘when’s in Danish: On the one hand, the semantics of da-clauses is similar to the semantics of definite DPs in that a da-clause presupposes that, in the current discourse situation, there is one and only one eventuality corresponding to the description it conveys. This makes it possible for a da-clause to have a reference-setting function with respect to its superordinate clause. On the other hand, når-clauses are similar to indefinite DPs in that they contribute propositions with an unbound eventuality argument, and therefore they yield descriptions of eventualities that never get referentially bound, but always occur in the scope of a non-existential quantifier. This restricts the use of når-clauses to habitual sentences and future sentences. This analysis involves the elaboration of a novel and more adequate formal semantic description of habitual sentences.

The purpose of this article is to give a description of the syntax of clausal complementation in Norwegian within the framework of Lexical-Functional Grammar (LFG). The data are mostly authentic sentences from text corpora and web pages. The description given is rather different from traditional descriptions. An important point to be made is that Norwegian (finite and non-finite) clausal complements are not homogeneous with respect to their syntactic properties. Even if most clausal complements behave like objects, Norwegian also has clausal complements that do not behave like objects. This phenomenon is well known from other languages, but has never been observed in the literature on Norwegian. The structure and function of non-object clausal complements are discussed, and the traditional idea that they are PP obliques is rejected.

English some and Swedish någon have a number of ‘subidentificational’ meanings, as in English some general (or other) has been shot; it was some shooting (or something). This paper reviews those meanings and attempts to determine how many context-invariant meanings are needed in order to account for their full range of meanings in context. By explicitly setting out processes of inference generation within a Gricean framework, it is found that for a large number, only one underlying (i.e. coded) meaning is required, which in context generates inferences such as lack of speaker knowledge, recall or interest in the NP referent, regarding either its type or which entity it is. Nevertheless, not all meanings can be handled in this way and additional coded meanings are required. Two diachronic pathways are discussed via which one coded meaning might extend to another. Some conclusions relevant for future work on indefinites are drawn.

This paper addresses the methodological issue of what data can safely be disregarded in the quest for a good analysis of a given semantic phenomenon. I argue that as there is no objective measure for this, making the temptation to adjust the terrain to the map hard to resist, it is strongly advisable to make use of text corpora in formal semantic research. I illustrate this argument by considering in some detail the recent history of research on Free Choice items.

This paper deals with the SIMPLE-DK lexicon, a computational lexicon for Danish developed at the Centre for Language Technology in Copenhagen within the European Union project SIMPLE. The general SIMPLE model, on which the Danish lexicon is based, is presented, and the way in which several specific aspects of Danish, such as nominal compounds and time expressions, are accommodated in this model is then described. Phrasal verbs – in particular phrasal motion verbs – are shown to be a challenging phenomenon since they are difficult to place in the SIMPLE event ontology, and pose problems regarding the interpretation of the directional particle they combine with. The encoding strategy that is proposed here accounts for compositional and non-compositional types of phrasal verb, and captures the relation between act-denoting and transition-denoting senses of the same verb in terms of regular polysemy. The final part of the paper deals with the exploitation of SIMPLE-DK as an ontological and lexical source in the Danish project on content-based querying OntoQuery. In the OntoQuery ontology, the structured concepts in SIMPLE-DK are combined with nutrition concepts, and the resulting ontology is used for matching evaluation. It is also discussed how selectional restrictions and qualia roles from SIMPLE-DK can be included in a conceptual grammar to be used for query and text analysis.