A new study from Ibaibarriaga, Acha, and Perea (2025), published in the Journal of Experimental Child Psychology, tested something deceptively simple: does it matter whether a child writes a letter by hand or types it on a keyboard? The answer, backed by data across six different learning measures, is unambiguous. It matters enormously.
Ibaibarriaga et al. recruited 50 prereading kindergarteners (average age 5.4 years) and taught them nine unfamiliar letters drawn from the Armenian and Georgian alphabets, along with 16 pseudowords built from those letters. The unfamiliar scripts were deliberate: no child had prior exposure, so every gain was attributable to the training itself.
The children were split into four groups, each defined by two factors the researchers wanted to isolate. The first factor was graphomotor action: did the child form the letter by hand (copying or tracing), or did they press a key on a keyboard? The second was output variability: did the child see the letter in multiple fonts, or only one? This gave the study a clean 2×2 design. Hand-copying involved writing letters freehand after seeing them in varied fonts. Tracing meant following a single letter template. Typing with varied fonts meant pressing keys after viewing letters in different typefaces. Typing with a single font meant pressing keys after seeing one consistent form.
Three 30-minute training sessions later, the children were tested on letter naming, letter writing, letter identification, word naming, word writing, and word identification.
What the Data Shows About Handwriting vs Typing
The graphomotor hypothesis won decisively. Children who learned letters by hand, whether through copying or tracing, outperformed the typing groups on five of six measures. The single exception was letter identification, where all four groups scored around 92%. Every child could recognize the letters they had learned. But recognition was where the typing groups’ advantage ended.
In letter naming, handwriting groups averaged 92.3% accuracy compared to 70.9% for typing groups. In letter writing, the gap widened: 64.6% for handwriting versus 27.7% for typing. At the word level, the differences became stark. Handwriting groups named words at 71.9% accuracy; typing groups managed 37.9%. And in word writing, the most demanding task, the handwriting groups scored 69.2% while the typing groups reached just 7.8%.
The letter identification results are revealing precisely because they look like a success story for all four groups. At 92% accuracy, it would be tempting to conclude that every training method worked. Ibaibarriaga et al. are careful to note that visual identification of letters is necessary but not sufficient for reading. A child who can pick a letter out of a lineup has done something. A child who can name it, write it, and use it in a word has done something fundamentally different.
Ibaibarriaga et al. explain the mechanism through a perception-action lens. Handwriting integrates eye movement, hand movement, and visual form into a single coordinated act. The child sees the letter, tracks its shape, and reproduces it with a pencil, all in one continuous loop. Typing breaks that loop. The child sees a letter on screen, then presses a key that bears no physical resemblance to the letter’s shape. The motor action encodes nothing about the form itself.
This split between recognition and production echoes a pattern I have written about in the context of AI. Kosmyna et al. (2025) found that students using ChatGPT showed reduced neural engagement during writing tasks, and the students who thought independently before turning to AI produced stronger work. Fan et al. (2025) documented what they called metacognitive laziness: students using AI produced better essays, but their actual learning, measured by knowledge gain and transfer, showed no improvement. The product improved. The process did not.
The handwriting study maps onto the same fault line. Typing produced adequate recognition. Handwriting produced recognition, recall, naming, and the ability to assemble letters into words. The keyboard gave children a surface-level familiarity with letter shapes. The pencil gave them something deeper: a motor memory tied to visual form, sound, and meaning.
What This Means for Literacy in a Digital Classroom
Ibaibarriaga et al. are clear in their recommendation: keyboards should complement handwriting in literacy activities, not replace it. I agree, and I think their data is strong enough to say something broader.
The study controlled for variables that typically muddy literacy research. The letters were unfamiliar. The pseudowords had no prior associations. The Solomon four-group design ruled out testing effects. And the training was brief, just three sessions, which makes the size of the handwriting advantage all the more notable. If 90 minutes of handwriting practice produces a 60-percentage-point gap in word writing accuracy, the cognitive mechanism at work is not trivial.
Kalantzis and Cope (2025) have argued that literacy in the age of AI should be understood as design agency, the ability to assemble meaning from multiple modes and tools. I find that argument compelling. But design agency does not start in a vacuum. It starts with the foundational representations that allow a child to connect a visual shape to a sound to a written form. This study suggests those representations are built through the hand, not the keyboard.
Shaw and Nave (2026) coined the term cognitive surrender to describe how AI reshapes reasoning patterns when users stop doing the mental work themselves. The children in the typing groups did not surrender anything deliberately. They just never built the motor-cognitive link that handwriting creates. The outcome looks similar: surface competence without the deeper processing that makes knowledge usable.
None of this is an argument against technology in early education. It is an argument for understanding what each tool actually builds. A keyboard is excellent for fluency once letter knowledge is established. A pencil is better for establishing it in the first place.
The broader lesson for educators, whether they are teaching five-year-olds to write or college students to think critically with AI, remains the same. The tool that makes the task easier is not always the tool that produces the most learning. Sometimes the effort is the point.
References
- Fan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., Shen, Y., Li, X., & Gašević, D. (2025). Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance. British Journal of Educational Technology, 56(2), 489–530. https://doi.org/10.1111/bjet.13544
- Ibaibarriaga, G., Acha, J., & Perea, M. (2025). The impact of handwriting and typing practice in children’s letter and word learning: Implications for literacy development. Journal of Experimental Child Psychology, 253(6), 106195. https://www.sciencedirect.com/science/article/pii/S0022096525000013
- Kalantzis, M., & Cope, B. (2025). Literacy in the time of artificial intelligence. Reading Research Quarterly, 60, e591. https://doi.org/10.1002/rrq.591
- Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X.-H., Beresnitzky, A. V., Braunstein, I., & Maes, P. (2025). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing tasks. MIT Media Lab. https://www.media.mit.edu/publications/your-brain-on-chatgpt/