Teachers and parents intuitively judge the 'level' of the child and the 'level' of the text and try to match them; they know that overestimation or underestimation of either will be met with restlessness or boredom. In this way, they have an empirical understanding of Vygotsky's ZPD--the zone of proximal development he envisioned as measuring the maturity of still-dependent functions in children. But this conception of the ZPD conflates two very different kinds of level: that of the child and that of the child's social milieu. We propose instead an analysis of child speech using Halliday's work on child development and his systemic-functional linguistics. We measure the emergence of modality, abstraction, and explicit clause Theme in Korean. We analyze four conversations with a single child over three years--from age four to seven. We find that there is a strong shift in the proportion of her clauses devoted to modal probability and towards explicit, textual themes. Counterintuitively, however, the nominal groups in her vocabulary become more concrete rather than more abstract. We conclude that Vygotsky's ZPD can be accurately gauged and traced through speech development, but for that very reason it cannot be reliably explained by either the child or the environment.