A constructicon may be roughly characterized as a dictionary for grammatical constructions. The objective of the project A Constructicon for Swedish was to develop such a resource, as a freely available online database (http://spraakbanken.gu.se/konstruktikon). The result after four years of work 2013–2016 is a Swedish constructicon comprising about 400 construction descriptions, as well as the beginnings of a support material for constructicon-based second language pedagogy, several language technology applications, and a basic structure for the development of a multilingual constructicon, the latter in collaboration with constructicon projects for other languages. As constructicons are a new kind of linguistic resource, the development has also led to theoretical and methodological advances in several areas, which was one of our aims from the onset. The project has to date generated seventeen scientific publications, a large number of conference presentations, and over a dozen student’s theses; another five publications are forthcoming (specifically, an anthology and four out of nine chapters therein; see below).
 
The selection of constructions covered has both a wide range, in order to develop a model capable of handling different kinds of constructions, and a special focus on two types of constructions. These are constructions of particular relevance to second language learning, on the one hand, and so-called partially schematic constructions, on the other. The latter type consists of constructions that combine lexical and grammatical properties and are therefore difficult to handle for either grammars or dictionaries. A constructicon, however, is designed to integrate grammar and lexicon, which puts such constructions in the center of attention instead of the periphery.

By its (partial) focus on partially schematic constructions, the construction offers coverage of precisely the kinds of patterns that tend to be neglected in grammars and dictionaries alike and that, partially for that reason, are problematic for both language education and language technology. Consequently, the focus on partially schematic constructions and the focus on L2 relevance overlap to some extent.

The Swedish Constructicon is designed as a multi-purpose resource, to benefit linguistics, language pedagogy, and language technology. Hence, it is not specifically tailored to suit the particular needs of either one, and different aspects of the description format cater more or less to different kinds of uses. To some extent this is adjusted for by a flexible interface; in addition, a complementary support material for language pedagogy is in development. Language technology applications are usually based directly on the database and therefore less dependent of the user interface; nevertheless, they have put certain demands on the description format, in particular as regards formalization.

On dimension that played a small role in the original proposal but has grown to be more central in the course of the project is international collaboration. To begin with, there was only a constructicon for English; then came corresponding projects for Swedish, Japanese, and Brazilian Portuguese, all around the same time; and later also Russian and German followed, the latter two adopting the description format of the Swedish constructicon. Throughout, there has been extensive collaboration between these groups, to the benefit of all, at first with a focus on methodology for single-language resources, but later also towards the creation of a multilingual constructicon. A common constructicon anthology is almost finished and will hopefully be released this year. This volume – Constructicography: Constructicon development across languages – may be seen both as a final report of this project and as a starting point for future international constructicon development.