Detection of a nonlinguistic stimulus is poorest at the end of a clause.

Subjects detected a brief near-threshold tone while encoding two-clause sentences for later report. The objective tone locations were at the end of the first clause, at the beginning of the second clause, or in the clause boundary. The effects of intensity variations of the speech signal were assessed by having subjects detect the tones in the same speech stimuli played backward. Tones at the end of a clause are relatively harder to detect than in other positions, comparing forward and backward speech. This supports the view that listeners are preoccupied with internal processes at the end of a clause.