In one keystroke-logging study, spontaneous sentence production under minimal planning is associated with gains in writers' self-rated understanding and with lower initial text quality¶

Claim · All claims
Evidence · 1 study · causal r2 · q3 · n=78 university students (Groningen, native Dutch speakers), randomly assigned to outline vs. synthetic planning
Currency · newest study 2018 · not yet checked against its sources · not reviewed by a person

When writers plan only a single overall goal before drafting (rather than a detailed outline), spontaneous, minimally-rehearsed sentence production predicts an increase in the writer's self-rated understanding of the topic — the opposite pattern from what occurs under detailed outline planning — while also predicting lower initial text quality; a separate process of revising the text's global structure independently predicts increased understanding regardless of planning condition, and the two together (text quality and understanding) are not simply positively related.

Subclaims¶

q3 i? Within the minimal-planning ("synthetic planning") condition, more spontaneous sentence production significantly predicted an increase in writers' self-rated topic understanding (b=-0.19, se=0.07, t(65)=2.80, p=.007). → Baaijen & Galbraith 2018

q3 i? Within the detailed outline-planning condition, the same measure of spontaneous sentence production instead significantly predicted a decrease in understanding (b=0.14, se=0.06, t(65)=2.28, p=.03) — the opposite direction from the minimal-planning condition. → Baaijen & Galbraith 2018

q3 i? Independent of planning condition or sentence-production style, revising the text's global structure (reordering sentences, revising away from the writing's leading edge) significantly predicted increased understanding (r=-.24, p=.04; β=-.27, p=.004 in a multiple regression). → Baaijen & Galbraith 2018

q3 i? In the minimal-planning, structurally-revised condition, more controlled (less spontaneous) sentence production significantly predicted higher text quality (b=1.65, se=0.47, t(68)=3.48, p=.009), indicating that the same spontaneous production that increased understanding reduced initial text quality. → Baaijen & Galbraith 2018

q3 i? Text quality and change in understanding were not simply positively related as the problem-solving account of writing predicts: the bivariate correlation was non-significant and slightly negative (r=-.14, ns), and in a multiple regression the relationship between text quality and understanding reversed sign depending on planning condition (synthetic planning: b=0.12, p=.01; outline planning: b=-0.10, p=.03). → Baaijen & Galbraith 2018

Evidence¶

Baaijen & Galbraith 2018¶

Baaijen, V. M., & Galbraith, D. (2018). Discovery Through Writing: Relationships with Writing Processes and Text Quality. Cognition and Instruction, 36(3), 199-223. https://doi.org/10.1080/07370008.2018.1456431

q3 · peer-reviewed experiment with random assignment to planning condition, keystroke-logged writing-process measures (Inputlog), validated composite scales (α=.79-.80) from a confirmatory principal-component analysis, and independently-rated text quality (interrater r=.84, p<.001); not pre-registered i? · printed figures are regression coefficients (b, β) and correlations (r), not a d-family effect size n=78 university students (Groningen, native Dutch speakers), randomly assigned to outline vs. synthetic planning · causal · r2

University students wrote a 30-minute argumentative essay after either detailed outline planning or minimal ("synthetic") goal-only planning. Writing processes were captured via keystroke logging and reduced to two validated composite scales — sentence-production style (controlled vs. spontaneous) and global-linearity (revision of overall text structure). Students rated their topic understanding before and after writing; two independent raters scored text quality. Regression models showed that spontaneous sentence production and structural revision predicted understanding through two independent pathways, and that the relationship between text quality and understanding depended on planning condition rather than being uniformly positive — directly testing and complicating the standard problem-solving account of writing. This informs Process-based Writing and Dual-Process Account of Discovery in Writing.

Discussion¶

The understanding measure is a single pre/post self-rating on a 7-point scale (arctan-transformed for skew), not an external test of learning — a genuine but subjective measure. The sample is drawn from one university writing task in one language and topic domain, and the effects are statistically significant (e.g., p=.007, p=.03) but no standardised effect size is reported. The claim that text quality and understanding are not positively related is the paper's most theoretically consequential finding and directly conflicts with the problem-solving account's core assumption; readers should note this is based on the same 78-participant sample as the other subclaims, not independently replicated elsewhere.

Learner Variables¶

• Writing Skills — outcome: instruction changes it

Related Claims¶

• Process Writing Improves Writing Quality — related
• Strategy Instruction Improves Writing Quality — related
• Extensive reading improves grammatical knowledge and writing style — related
• Peer response groups increase the number of idea units in English learners' writing but show no significant effect on composition quality or sentences written (Prater & Bermudez, 1993) — related
• Keystroke patterns during web-based DP exams suggested planning, revision, and sustained cognitive engagement — related
• SRL writing-stage clusters differed in production and revision but not in pausing — related
• Installing the synthetic position as promoter can raise self-narrative coherence while lowering authorship of it (authorship-coherence dissociation) — related
• A knowledge-constituting process synthesizes writing content from constraints within semantic memory — related