The Swedish subproject of ScanDiaSyn

Introduction
The Swedish part of ScanDiaSyn is called SweDiaSyn. This project, which is located at the department of Scandinavian Languages, Lund University, is funded by Vetenskapsrådet (the Swedish Reseach Council). It started in February 2005 and will run until December 2007.

Personnel
The people involved in the project are Lars-Olof Delsing (project leader), Christer Platzack and Henrik Rosenkvist (researcher). Six assistants are furthermore employed as transcribers, and in the near future, we hope to be able to recruit a PhD-student as well as assistants from the northern parts of Sweden.

Main goal
The main goal of SweDiaSyn, as specified in the project application, is to provide detailed and systematic knowledge of the syntactic variation in Swedish dialects, under the ScanDiaSyn umbrella. We intend to achieve this by transcribing recordings which were made by the project SweDia 2000 and, with these transcriptions as a base, set up a database with tagged transcriptions. Simultanously, we will study the interesting syntactic constructions that appear in the recordings.

The recordings
In SweDia 2000, recordings were made at 97 locations in Sweden and 10 in Finland. At each location, 12 informants were interviewed informally for about 30 minutes (and they were also asked to read lists of words, but that material is not of relevance for our project). The informants were both male and female, and both young and old. This means that from each location, there are recordings of three old men, three old women, three young men and three young women. All together, the recordings from Sweden comprise about 600 hours of speech. We will probably not be able to process all of the recordings, but will concentrate on older speakers in areas where we expect to encounter interesting syntactic phenomena.

Transcription
We use the program ClanX for the transcriptions – it was developed for language acquisition research and is available on the Internet (http://childes.psy.cmu.edu/).
In Sweden, some dialects are quite close to standard Swedish, while others are genuine dialects which speakers of standard Swedish do not comprehend. Dialects which belong to the former type are glossed directly, i.e. they are written in standard Swedish, with the exception that the gloss must reflect the word order and the exact number of words. Genuine dialects are however both transcribed, i.e. rendered in (relatively coarse) phonetic writing, and glossed. With this method, it will be possible to analyse all transcriptions automatically (since the glosses are written with standard Swedish ortography), and morphosyntactic features of the genuine dialects will be accessible both via the transcriptions and via the recordings.