This research used a copying task to study spelling acquisition from a perception and action perspective. First to fifth graders copied words and pseudo-words on a digitiser. Simultaneously, a camera registered the children's gaze lifts. First and second graders copied the first syllable and then produced a gaze lift to obtain information on the spelling of the second syllable. Words required fewer gaze lifts than pseudo-words. Third, fourth, and fifth graders copied the items as whole orthographic units and did not produce any gaze lifts. The analysis of movement duration revealed that, for all the children, the gesture to produce the first syllable was programmed before movement initiation. A systematic duration peak at the first letter of the items' second syllable was then observed. This indicates that the gesture to write the second syllable was programmed during the execution of its first letter. Thus, the younger children used syllables to articulate visual parsing and motor programming. The older children used whole word visual units but still organised their graphic production according to the syllable structure. Syllables therefore appear to act as functional units at both the visual and motor stages of processing during spelling acquisition. (Contains 5 figures.)