Constructions and Frames - Volume 6, Issue 1, 2014

This article addresses the possibility of linking constructicon resources for different languages, in particular English and Swedish. The entries in Berkeley’s English constructicon have been compared to Swedish, with a focus on potential correspondences in a Swedish constructicon. In most cases, approximately corresponding Swedish constructions could be established, although typically with minor differences, often concerning grammatical markers. The closest equivalents are, typically, relatively general grammatical constructions, whereas constructions containing specific lexical elements tend to differ more. In order to link all corresponding constructions between the two resources, a combination of strategies seems to be required. Constructions with a referential meaning may be linked via FrameNet frames, while those with a more abstract grammatical function may be related in terms of their grammatical properties.

This paper proposes three policies for the annotation of constructions in FrameNet Brasil, and, potentially, in other FrameNets. Annotation policies are defined so as to both avoid uncontrolled redundancy in the database and respect the theoretical and methodological foundations of Frame Semantics and Construction Grammar. The first policy is concerned with the task of deciding whether a given piece of language should be analyzed as an instance of a construction, or as a valence pattern of a lexical unit; the second specifies criteria for the definition of Construct Elements; finally, the third policy regulates the interconnections between constructions and frames in the database.

This paper presents work on developing Swedish FrameNet (SweFN) as a resource analogous to the original Berkeley-based FrameNet. We describe the theoretical and practical basics of FrameNet, and articulate some multilingual issues that arise in expanding a linguistic resource from one language to another. SweFN uses FrameNet as a starting point in order to save time and effort, and to make it compatible with other FrameNet-based resources. The lexical units are from the pivot lexicon SALDO, making SweFN compatible with other resources of the larger project SweFN++. It is a corpus-based resource, meant to support tasks within natural language processing relying on semantic data.

This paper reports on the development of a domain-specific multilingual electronic dictionary covering the domains of soccer, tourism and the World Cup: the Copa 2014 FrameNet Brasil project. Specifically, we discuss three points: (i) the definition of the tourism frames and their deployment as interlingual representations; (ii) the implementation of the trilingual annotation software used for populating the Copa 2014 database; and (iii) the creation of the Translation relation, a computational solution for the deployment of frames as interlingual representations.

We evaluate several lexicon-based and corpus-based methods to automatically induce new lexical units for the Swedish FrameNet, and we see that the best-performing setup uses a combination of both types of methods. A particular challenge for Swedish is the absence of a lexical resource such as WordNet; however, we show that the semantic network SALDO, which is organized according to lexicographical principles quite different from those of WordNet, is very useful for our purposes.

We present an experiment where natural language processing tools are used to automatically identify potential constructions in a corpus. The experiment was conducted as part of the ongoing efforts to develop a Swedish constructicon. Using an automatic method to suggest constructions has advantages not only for efficiency but also methodologically: it forces the analyst to look more objectively at the constructions actually occurring in corpora, as opposed to focusing on “interesting” constructions only. As a heuristic for identifying potential constructions, the method has proved successful, yielding about 200 (out of 1,200) highly relevant construction candidates.