Two markedness hypotheses in current language acquisition theory are examined. One view of markedness, the developmental hypothesis, states that the unmarked case is the child's initial hypothesis, i.e., the hypothesis that is set in advance of linguistic data. The developmental hypothesis further predicts that children will proceed in a fixed progression from unmarked to marked structures. The learnability conception of markedness states that the child acquires the unmarked form in the absence of evidence to the contrary, and that there is not necessarily a fixed sequence from unmarked to marked; it is neutral with respect to real-time acquisition. An experiment on the acquisition of pied-piping (generally considered unmarked) and preposition stranding (considered marked) that tests these hypotheses had as subjects 33 preschool children and used comprehension and production tasks. The results indicate that the developmental claim can not be supported because the children seem to have both structures in their grammars, with stranding firmly established, suggesting that markedness may not necessarily be indicative of the order in which certain constructions emerge. It also suggests that the learnability hypothesis can account for the findings. (MSE)