Two randomised control experiments examined spelling outcomes in a repeated measures design (pre-test, post-tests; 1-day, 1-month follow-up, 5-month follow-up), where students learned Spanish irregular words through (1) immediate feedback using self-correction, (2) visual imagery where children imagine and represent words using movement, and (3) copying words. The two control groups engaged in reading and dictation. We expected that methods with kinesthetic components and segmentation of stimuli would yield better orthographic representations in memory. Results showed that orthographic errors on Spanish irregular words diminish when children receive self-correction instruction. Imagery and copy instruction also reduce these errors, but less consistently. Five months later, self-correction group maintained their advantage in spelling compared with students who read or performed dictation. Experiment 2 replicated the spelling instruction effects and showed large effects on children who had hardly received any spelling instruction in Spanish. Results indicate that research-based practices can significantly improve spelling outcomes in primary children.