THE EFFECT OF MASKER TYPE AND WORD POSITION ON IMMEDIATE SENTENCE RECALL
Noise maskers primarily result in energetic masking, whereas speech maskers create additional interference due to linguistic and acoustic similarities to the target (informational masking). Factors that facilitate stream segregation can greatly reducethe extent of informational masking. However, stream segregation often takes time to develop. In Experiment 1, nonsense sentences with 3 keywords were presented against a background of speech-spectrum noise or two-talker nonsense speech. With the speech masker, accuracy increased with word position. With the noise masker, accuracy did not vary systematically with word position. In Experiment 2, we noise-vocoded the speech masker using three bands to preserve envelope information while disrupting fine structure cues and minimizing semantic content. Here, performance was similar to that found with the noise masker. The results suggest that the ability to track a target sentence in conditions of informational masking improves as the target utterance unfolds over time.