One of the fundamental puzzles language learners must solve is the mapping of a string of words onto a particular (correct) syntactic structure. In this article, I examine the problem of how learners should resolve the ambiguity presented by a string that could have either a raising or a control structure. I provide both logical and empirical arguments against the view that children should be biased to assume that such a string has a control structure. Instead, I propose two families of cues, based on a psycholinguistic experiment with adults, which can be used in a probabilistic manner to parse an ambiguous string and to categorize raising and control verbs.