Constructions and Frames - Volume 4, Issue 1, 2012

This paper discusses the classification of null instantiation phenomena in Construction Grammar and proposes a different treatment of so-called free null instantiation (FNI). Based on e.g. control data, different types of alleged FNI are shown to be more accurately classified as definite (adjunct control), generic (e.g. tough constructions), or unspecified for interpretation (e.g. passives). A striking pattern is that general constructions, such as infinitives and gerunds, license unspecified null instantiation (simply NI), whereas more specific control constructions are associated with a definite (DNI) or generic (GNI) interpretation. Hence, the paper proposes a null instantiation taxonomy that distinguishes (unspecified) NI and the specific subtypes definite (DNI), indefinite (INI), identity of sense (ISNA), and generic (GNI) null instantiation.

This paper argues for the discoursal motivation of constructions and the constructional motivation of a Modern Greek indefinite article, by focusing on generics and some other nominal constructions in Modern Greek containing the indefinite determiner enas/mia/ena (“a/an”, masc./fem./ neut. respectively). Moreover, it illustrates the cross-linguistic relevance of Construction Grammar. A family of indefinite nominal constructions is identified, including the Indefinite Generic Construction, the Indefinite Proper Noun Construction, the Indefinite Predicate Nominal Construction, and the Proverbial Indefinite Construction, which is further shown to be primed by specific pragmatic and discoursal features. On the basis of the proposed analysis, and consistently with established views on semantic change (Traugott 1989), I suggest that the expressivity and the discoursal characterization associated with this family of indefinite constructions in Modern Greek motivate the partial de-semanticization of the indefinite determiner, its synchronic variability (after Hoffmann & Trousdale 2011) and, ultimately, its function as an article in the language.

In many formal theories of grammar, pairs of expressions such as the active and the passive are treated as variants of each other — the passive typically being a secondary construction derived from the active by operations that change the syntactic structure. Recent accounts based on Cognitive Grammar and Construction Grammar have questioned the validity of such an analysis, arguing that these “variants” are actually independent constructions with their own usage conditions and meaning. An important piece of evidence comes from so-called split constituents, discussed by Croft (2001: 191), who argues that expressions like A guy who I hadn’t seen since high school came in vs. A guy came in who I hadn’t seen since high school differ in their grammatical structure and usage. In this paper we discuss the Finnish split genitive construction where the assumed genitive modifier is separated from its head by intervening material, typically the finite verb. In many respects, the split genitive resembles constructions of external possession, but its range of usage is relatively limited, and in the grammatical system of Finnish it can be seen in an intermediate position between adnominal genitive constructions, on the one hand, and productive external possessor constructions based on local cases, on the other hand. Traditionally, the split genitive has been taken to be a discontinuous variant of a contiguous NP where the genitive is positioned next to its head. However, this study shows that the two constructions differ in pragmatic, semantic and grammatical terms. The split genitive construction is more limited in its usage, and it serves more specific semantic functions such as the topicalization of the genitive-marked element that carries the role of an experiencer. As in many external possessor constructions cross-linguistically, these constraints restrict the types of genitive modifiers that are available in the split genitive construction.

In the constructionist view, the grammar of a language is represented by constructions organized in taxonomic networks. This paper addresses the question of how one should account for the differences and similarities in the organization of such networks in different varieties of a language. In particular, a corpus-driven quantitative methodology is developed that can provide evidence for modelling the relevant constructional networks. The method is based on hierarchical cluster analysis applied to multivariable data, and several statistical criteria of cluster stability. These points are demonstrated by means of a corpus study of the Netherlandic and Belgian variants of the Dutch causative construction with doen “do”. The paper describes the national differences in the structure of the constructicon and interprets them from a usage-based historical perspective.