Languages in Contrast - Volume 1, Issue 2, 1998

In this paper, I elaborate a 'lexicogrammatical approach to the locative alternation in Dutch and English, in which the semantics of the constructions is related directly to the semantic features of their verbs. First I take stock of the different verb classes associated with one or both variants in Dutch and English. Whereas both the spray and the load-classes alternate in English, only the spray-verbs do so in Dutch. However, the alternating spray-class is much more extended in Dutch, which I then argue is partly due to the use of the prefix be- in Dutch in one of the variants. Be-, I show, has an applicative meaning, and is used to bring out the correspondence between Patients designating a surface and the locative Process. When they are not prefixed by ht-, these verbs have a primarily dispersive meaning, which tends to collocate with the Patient designating the substance.

This paper is concerned with the relationships between the semantic role 'causée' and the morphosyntactic patterns used to express it in a range of Germanic and Romance languages. We will try to show that the causee — a hybrid semantic role as it is both a patient and an agent — has special relationships with object case marking. The evidence shows that Germanic languages such as German and English, and some Romance languages such as Spanish and Portuguese, resort to positional rules to preserve the distinction between causee and true object. Other Romance languages such as Italian and French, however, obtain the same result by morphological means, especially as regards the causee. We claim that such differences can be better understood in the light of a diachronic and typological study of causative constructions in these languages.

Among the theoretical frameworks which consider grammatical relations to be a possible but not necessary level of clausal organization, the approach to the object relation espoused in Simon Dik's (1978, 198g, 1997) Functional Grammar is the most restrictive. Unlike various other models of grammar, only one object relation is recognized and its presence in a language is conditional on the existence of a productive dative-shift opposition relating predications depicting the same states of affairs, as in the case of the English The teacher gave the picture to the child and The teacher gave the child a picture. Taking the existence of pairs of clauses such as these as a diagnostic of the object relation heavily reduces the number of languages manifesting an object relation, so the presence of an object relation emerges as a potentially interesting typological parameter. But do the languages manifesting an object relation, in this restrictive sense of the term, have any properties in common other than the object relation? Little attention has been given to this issue either by Functional Grammarians or other linguists who adopt a similar view of the object relation. The present paper seeks to redress this situation by exploring the cross-linguistic applicability of the object relation, as defined in FG, in an extensive sample of genetically and geographically stratified languages. It examines the typological characteristics of the languages with objects in the FG sense of the term and establishes the typological profile most likely leading to the presence of an object relation.

In this paper we show that the well known definition of affected object — as an object that is somehow altered or modified by the action expressed by the verb — is problematic with respect to middle formation, which has been claimed in the literature to be possible only with affected objects. The following puzzling facts are discussed: (i) in "plain" languages some predicates with unaffected objects may undergo middle formation whereas others may not; (ii) in "reflexive" languages some predicates with unaffected objects may undergo middle formation whereas others may not; (iii) "reflexive" languages may differ among themselves with respect to middle formation with unaffected objects. We argue that the notion of affected object has to be re-defined in terms of the aspectuality of the entire predicate.

This paper deals with non-agentive deverbal -er nominals in English and Dutch. It attempts to provide a grammatico-semantic explanation for the extension of agentive to non-agentive -er, and argues that the profile of the -er suffix, irrespective of whether it is agentive or non-agentive, is comparable to that of the subject-Junction of a clausal structure. More particularly, some clausal structures are discussed which show a high number of structural and semantic correspondences with non-agentive -er nominals. Whereas in English, the most important clausal agnate turns out to be the middle construction (e.g. this book reads easily), Dutch non-agentive -er nominals are shown to agnate with various structures, notably middle formation: especially the occurrence of intransitive or 'circumstantial' middles (e.g. asfalt fietst prettiger dan grind 'asphalt cycles better than gravel'), and the frequent use of 'let'-constructions in contexts where English would use a middle (e.g. dat boek laat zich makkelijk lezen [that book lets itself easily read], i.e. that book reads easily) offer evidence of the Dutch potential to construe non-agentive entities as subjects. Throughout the discussion, the clauses and -er nominals under scrutiny are illustrated by extracts from the COBUILD corpus and the Dutch INL corpus.

This paper examines resulting objects in English and Spanish. Fillmore's (1968) concept of effected object, e.g. He built the table has been extended to include certain resulting objects which are not sub categorized by their verbs: He kicked a hole in the door, She smiled her thanks or They pushed their way through the crowd. These constructions, we argue, are formed by fusing two semantic predicates into one: the object expresses a result while the verb designates the means by which the intended result is achieved. An analysis of corpus examples demonstrates that they are neither totally lexicalized nor regular productive patterns. The intermediate lexicalized position in question is due to a gradual metaphoric process found in this English construction, a process not allowed in Spanish. The paper concludes that the inability of Spanish to fuse two semantic predicates in this lexically unfilled resulting construction, which works independently of the particular lexical items that instantiate it, may be a consequence of the rigidity of its lexical items, which are strongly marked by morphology. The lack of inflectional information of most English words, on the contrary, confers more grammatical power to the construction.