This report reviews experimental techniques that have been used to assess sentence comprehension by preschool children and describes a new technique, word monitoring, which permits an assessment of the momentary demands made upon sentence-processing mechanisms as a child listens to a sentence. (The time it takes the child to push a button in response to a target word is the main variable in this technique.) Four experiments using the technique are described. The first explored the word-monitoring task to see if it would yield sensible data in a "clear case" situation. In succeeding experiments, various hypotheses concerning the processing of sentences that contain relative clauses were examined. In particular, Slobin's Interruption Hypothesis and Sheldon's Parallel Function Hypothesis were tested. Evidence in favor of the Interruption Hypothesis was obtained: children at this stage appear to enter into a processing subroutine when the cue word "that" appears, signaling a relative clause. Further work showed that children do not overgeneralize; they do not take the cue word "that" to be a signal for a relative clause when it is inappropriate for them to do so (e.g., when "that" is a demonstrative). (Author/AA)