Cognitive processes in writing-to-learn

Abstract

Writing-to-learn is the idea that writing is not only an effective tool for communication, but it also has the capacity to shape learning and facilitate the development of knowledge (Emig, 1977; Klein, 1999). Specifically, it is writing that facilitates learning, and not another medium such as thinking of speaking (Galbraith et al., 2025). This thesis explored writing-to-learn in the context of closed-book source-based writing, under the lens of the dual-process model of writing (Galbraith & Baaijen, 2018). It also aimed to compare different measures of learning from writing.

In Paper 1, 78 higher-education students read a source text passage about solar activity (Arnold et al., 2017), created either an outline or synthetic plan, and wrote an essay or wrote down everything they could remember from the text (free recall). Two days later, participants returned to answer text comprehension questions to assess their knowledge of the source material. They also rated their subjective knowledge of the source text before writing, after writing, and two-days later. In support of a previous study by Arnold et al. (2017), there were no differences in text comprehension scores across the planning or writing conditions, because all conditions involved retrieval practice. However, participants subjective knowledge decreased from pre- to post-writing, regardless of condition. Writing about the source helped participants to identify knowledge gaps, consistent with the experience of the illusion of explanatory depth (Rozenblit & Keil, 2002). The knowledge-telling process was re-interpreted to suggest it can have a role in learning through metacognitive monitoring. Importantly, participants subjective knowledge ratings positively correlated with their text comprehension scores, supporting the validity of the measures.

Paper 2 investigated the effects of written and non-written (mental) source-based activities on text comprehension and participants’ subjective knowledge. Experiment 1 was a methodological check of a new source text passage. In the main experiment, Experiment 2, 116 participants read one of two source texts, and then wrote or thought about writing either an essay or free recall. Two-days later, participants completed text comprehension questions, and rated their subjective knowledge before writing, after writing, and two-days later. Results showed that writing about the source text led to better performance on short-answer questions than thinking. Participants in the writing condition also experienced a greater decrease in their subjective knowledge than participants in the thinking conditions. These results highlighted that writing, and not another medium, has a unique value for enhancing text comprehension and metacognitive monitoring. Subjective knowledge ratings were positively correlated with short-answer question scores and factual multiple-choice question scores, replicating the results from Paper 1.

Paper 3 aimed to test the reliability and validity of a new methodological tool for assessing learning from writing in a replication and extension of Silva and Limongi (2019). Experiment 1 was a direct replication of Silva and Limongi’s study, which compared the effects of writing versus speaking summaries of short texts on participants’ reaction time on an episodic memory word-recognition task. Experiment 2 adapted the procedure to control for a potential confounding factor (modality effect). Across both experiments, results replicated Silva and Limongi’s findings. Drift-diffusion modelling of reaction time distributions revealed that after 3 writing summaries, participants had faster reaction times than after speaking summaries, represented by the non-decision parameter (encoding and/or motor processes).

Finally, Paper 4 aimed to investigate the cognitive processes underlying writing-to-learn in a closed-book, multiple source-based writing task. A second aim was to assess how the three measures of learning established in Papers 1 to 3 were related to one another. 40 university students read two source texts before writing either a synthesis essay or writing down everything they could remember from the two texts. After writing, they completed a word-recognition reaction time task, and a text comprehension test. Participants also rated their subjective
knowledge before reading, before writing, and after writing. The most important finding was that whilst a subjective knowledge decrease was observed for participants in the free recall condition, participants in the synthesis condition did not experience such a decrease. This suggests the synthesis task enabled participants to carry out knowledge-transforming, whereas the free recall participants were essentially told do carry out knowledge-telling. Crucially, subjective knowledge ratings were significantly positively correlated with short-answer question scores, a finding that has replicated across these papers.

More information

Identifiers

ORCID for Amy Peters: orcid.org/0000-0001-9833-7709

ORCID for David Galbraith: orcid.org/0000-0003-4195-6386

Catalogue record

Export record