Writing by hand or digitally in first grade: Effects on rate of learning to compose text

In a natural experiment we compared development of writing composition skill in five Norwegian first-grade classes in which children (N = 90) learned to compose text by handwriting on paper and in five classes in which children (N = 91) learned by typing on a digital tablet using software that additionally provided read-back via text-to-speech synthesis. Children completed narrative composition probe tasks at five timepoints over eight months, writing in the modality in which they were learning. Students' narratives were evaluated in terms of a range of text features capturing both transcription accuracy (spelling, spacing, punctuation), and syntactic and compositional sophistication. Statistical analysis was by Bayesian modelling allowing for robust inference in the presence or absence of a modality effect. Children showed improvement in text length, syntactic accuracy and complexity, and narrative sophistication. However, rate of improvement was unaffected by modality. Spelling and spacing, which were directly supported by read-back functionality, improved just in the handwriting condition, with better performance but no improvement in the digital condition. Our findings provide evidence against claims that either learning to write by hand or learning to write digitally (typing supported by text-to-speech) are inherently better for students' learning of written composition, at least across their first year of writing instruction.