The Signwrite corpus
Signwrite is a corpus consisting of texts written by Swedish children with different degrees of hearing and sign language background.
Photo: Pexels
The writers include deaf, hard-of-hearing, hearing children of deaf adults (CODA) with Swedish Sign Language (SSL) proficiency, and a control group of hearing children without knowledge of SSL. The corpus is partly longitudinal, with an age span of 8 to 21 years.
A text in correction annotation mode (detail).
Apart from the final text, we also provide keylogging data documenting the writing process, a version of the final text normalized (corrected) to standard orthography, and annotations for each edit with the error type.
The intended target groups for the corpus are researchers of writing development, especially among deaf and hard-of-hearing children, as well as researchers in Natural Language Processing (NLP) building and evaluating tools for writing support and writing assessment targeted at this group.
The Signwrite data originates from the dissertation work of Moa Gärdenfors, assistant lecturer at the Department of Linguistics, Stockholm University.
- Språkbanken CLARIN
has co-funded further work on annotating the Signwrite data and making it available. - The research project Språkbanken
at the Department of Linguistics, Stockholm University, is a node within CLARIN.
Last updated: 2026-11-06
Source: Department of Linguistics