A study investigated the hypothesis that children are sensitive to functors in language and only omit them due to factors specific to speech production and after having analyzed them as separate morphemes. This hypothesis was tested as an alternative to two existing hypotheses concerning children's selective listening for content words and for stress. The subjects were three groups of toddlers. Their responses on an imitation task were analyzed for omission of functors for English and nonsense words. For one group, the experimenter gave content words more stress than functors. For the second group, the strings were generated on a speech synthesizer to give the same intonation contour to each. The third group was divided into two subgroups, one hearing recorded human speech and one hearing synthesizer speech, but in all cases, the English and nonsense strings to be imitated contained schwa. The results of the three studies contradict both the content attention and stress hypotheses and support the hypothesis under consideration: That children omit functors because they have analyzed them as separate morphemes. Further research on young children's speech production processes is recommended. (MSE)