Questions asked in five play sessions by each of eight children aged 1.6 to 3.0 were analyzed for usage of the wh-interrogatives. About 93% of utterances using "what" and "where" were formulaic constructions (e.g., "Where's NP?" and "What's that thing?"). In order to determine whether children were cognitively segmenting these formulas and learning individual wh-words, the following observations were taken into consideration: (1) that repairs were observed in interrogatives with false starts; (2) that paradigmatic substitutions within formulaic interrogatives showed appropriate alternations (e.g., "What's that?" with "What're those?"); and (3) that wh-interrogatives confined to a single formula type in the first session at times were used in a variety of contexts by the same child in subsequent sessions. This third observation seems in part contradicted by a further observation that the child had not yet developed a what-where distinction by the third session. The following explanation is offered: A frame such as "Whusis?" ("What's this?") will be learned by a child as a unit whose wh-word is not perceived as conveying independent meaning; thus the wh-phoneme will represent a collocation of various wh-words, depending on the child's individual experience. The child becomes aware of the independence of the wh-interrogative and makes appropriate use of it in different speech acts, but the power of the original frame to govern a confusion between wh-words does not immediately diminish. (JB)