Translating numerals into number words is a tacit task requiring linguistic and mathematical knowledge. This project expanded on previous number production models by examining developmental differences in children's number naming errors. Ninety-six children from grades one, three, five, and seven translated a random set of numerals into number words. Their errors were coded according to location and type. The data were analysed by grade using ANOVAs followed by Tukey post hoc tests. Results suggest that prerequisite linguistic conventions are needed to start and sustain the number production process, and may have implications for curricular recommendations concerning number naming. (Contains 2 tables.)