ATEL aims to develop a model for determining children's early writing development. The model will be based on analyses of a large number of texts written by pupils in the Skriv og Læs learning tool. Together with the National Centre for Reading, DPU is developing a framework for detailed descriptions of texts written by pupils and will use this framework to analyse pupils' texts. The analyses will then be processed using the statistical Rasch model, making it possible to identify and describe development processes for children's writing. The results from these analyses will then be used as a basis to develop automated assessment of pupils' texts. This part is being developed by DTU using machine learning and a significant number of pupils' texts from the WriteReaders database. On this basis, DPU will contribute input on how teachers can support their pupils' literacy development based on knowledge about the pupils' level of development.

The strength and quality of the automated assessment will be determined by comparing with various language and reading tests commonly used in Danish primary schools, including language assessment tests, word comprehension tests, national testing (Danish literacy performance) and risk-testing for dyslexia.

The Skriv og Læs learning tool is based on principles inspired by the didactics of invented spelling, where children make first attempts at writing words and an adult then writes an 'adult version' (i.e. correct written language) of the child's text. Another part of the project is to develop autogenerated 'adult versions' of the pupils' texts to ease the teacher's workload when using Skriv og Læs in class. DTU is responsible for this part of the project.

Project period: 1 September 2018 → 31 March 2022