Constructions and Frames - Volume 1, Issue 2, 2009

Based on an extensive corpus study, this paper presents an overview of control patterns in Swedish infinitives and sketches a CxG account of the data. To capture the variety of control relations encountered, the approach combines elements of traditional CxG, Frame Semantics, and Sign-Based Construction Grammar. Three basic mechanisms are distinguished: control by selection, where the controlled element is coinstantiated with an argument of the selecting head; control by feature percolation, where the interpretation is determined by the syntactic and pragmatic context; and arbitrary “control”, which is treated as non-control, where the understood subject argument is specified for generic or arbitrary reference and, hence, needs no controller. More specific control patterns, including such issues as control shift and pragmatic control, are treated as specific variants of these three basic types.

Just like its English counterpart (cf. Goldberg 1995), the Dutch double object construction is a prime example of a highly polysemous argument structure construction, with a basic ‘X causes Y to receive Z’ sense and several extended meanings which depart from the prototype in various respects and to varying degrees. This paper provides a corpus-based overview of the semantic structure of this construction, following the multidimensional approach to constructional semantics advocated in Geeraerts (1998). On the basis of Stefanowitsch & Gries’s (2003) “collexeme analysis” method, we will identify the verbs which most typically realize the investigated construction in a one-million-word newspaper corpus. These verbs will be shown to instantiate extensions along various dimensions of semantic variation. Several of these semantic extensions are paralleled in English, while others are not.

The development of sentential complements in Greek, a language with systematic use of two complementizers, is studied in a corpus of spontaneous conversations by one girl and her family in the age range of 1;8 to 4 years. Formal, semantic and pragmatic analyses of child and child-directed constructions indicate parallels and divergences from previous acquisition research. Early formal marking of complements suggests ambient language effects. Yet, emergence of semantically/pragmatically more prototypical ones is protracted, favoring usage-based and semantically motivated explanations. The data also contribute to the theoretical description of sentential complement constructions, by charting some of their varieties and also suggesting explanations for the relations holding among these varieties.

Through a discourse-grounded internal reconstruction that aims at capturing the emergence of grammatical structure, the study examines the development of the subjective epistemic particle jestli ‘[in-my-opinion-] maybe’ in conversational Czech. Through internal reconstruction, the change (syntactic complementizer > speaker-centered epistemic contextualizer > subjective epistemic particle) is presented as a metonymy-based conventionalization of a pragmatic meaning implied by certain tokens of indirect Y/N questions into a new modal meaning. Taking a Construction Grammar approach, so far largely untested on diachronic data, the point of the analysis is to show that we can engage in a systematic treatment of the gradualness of change, by (i) combining the ‘holistic’ (constructional) dimension with the internal, feature-based and discourse-motivated mechanisms of complex grammatical shifts, and (ii) appealing to the explanatory potential of general cognitive and communicative principles as they manifest themselves in natural discourse. I also propose a formalism for representing the transitional nature of intermediate patterns.