Although linguistic behavior is in general guided by normative rules, it need not all be rule-guided. Also, rules of language differ in their degree of constitutivity, and hence, in their force of obligation. The so-called thematic rules are characteristically of a “weak” constitutivity, and often mere regularities of usage devoid of normativity. Their possible normativity is assessed in terms of contextual-pragmatic correctnees. The thematic organization of sentences illustrates particularly well the special relationship between rule and action, those areas of linguistic behaviour where normativity may shade off into spatiotemporality. From the methodological point of view, all this requires the investigation of thematic systems as mediated through action, and further, the combining of speakers' intuitive knowledge and of observation of actual linguistic behaviour.

Taggin is the proces of assigning morphological, sytactic and /or semantic information to linguistic units in a text. The most important problems in connection with tagging are, of course, those related to subject matter, but in computerized tagging systems, which this article is about, there are also several technical problems. They automatically announce thermselves, and have to be solved one way or the other. In this paper a fairly extensive description of a computer system (THE TAGGER) is given, which is an attempt to solve some of the problems diswcussed. This article is non-technical in the sense that it presupposes no actual Knowledge of programming, but some general experience of computers and computational linguistics is of course helpful.

The article discusses two types of discourse coherence, propositional coherence and interactional coherence. These two types are related to the mode of discourse as well as to the level of style in the discourse. It takes up the concept of frame as an organised subsystem of knowledge shared by human beings, and suggests ways in which this concept could be used in studies of coherence in everyday conversations, prose texts and lyrical poetry. The article also suggests a way of describing discourse coherence in terms of the discourse topic and considers the relationship of the two types in different kinds of texts.

The article contains a critical survey of Nordic umlaut research during the last thirty years. An attempt is made to evaluate the various contributions to this research which have been based, alternately, on structural generative phonological theory; in contrast to the earlier heavily datae-based research, recent work in this field has indeed been more exclusively theory- oriented. that the results of these ene deavors have on the whole not been commensurate with the effort; in spite of the evident optimism which, at certain times at any rate, has dominated this research, it apperars indeed that some of the traditional or classical problems of Nordic umlaut research tend to persist, or continually toreappear, in one guise or another, and thus to continue to defy a convincing solution.

A better understanding of problem of productivity of WERs is necessary not only to define the concept of a WEF, but also to find a “place” for the WF component in the model of generative grammar. The paper analyese one approach, in which the productivity is measured in terms of competence restrictions, on the basis of Nouns of Location in Polish. It is claimed that the productivity of a WFR is determined not only by relations of a base to an affix, but also by interrelations within the system of WFRs. A functional and/or formal conspiracy of WFRs may have an inhibitory and facilitatory effect on productivity. Thus, the concept of competence restrictions on the Productivity of WFRs developed so far is too narrow.

Recent grammars rule out uninverted tag questions (unnegated tag questions when the main clause is also unnegated and vice versa) as ungrammatical. It is argued that the inversion is merely a conversational phenomenon which is not always required. Linguists requiring inversion have been misled by their “intuitions”, which cast general doubts on their role in recent linguistic theorizing.

The present article aims to show that the prepositions in so-called “transitive verb groups” in Norwegian, like, e.g., på and til in passe på and ta hensyn til, respectively, are not to be conflated with the real verb particles in cases like drikke opp etc. The two groups of words have quite distinct semantico syntactic properties. In particular, the NP constituent in cases like drikke opp vinen is subject to the well-known non-specificity condition at work in Norwegian impersonal passive constructions with a purely formal subject det, whereas its counterpart in cases like passe på hunden is not. This difference is readily accounted for if it is assumed that prepositions are a rudimentary species of one-place predicates, where the one argument to be found cannot be classified as either a logical or syntactic subject or object.