Interference with the binaural intelligibility of spondee words and the binaural detection of a sinusoid of continuously varying frequency (250 to 4000 Hz) produced by two continuous independent random noise signals, N¹ and N², was measured in 10 normal-hearing adults at three masker levels under five conditions of listening:

1. homophasic (N¹₀N²₀S₀);

2. parallel antiphasic (N¹_π N²_π S₀);

3. opposed antiphasic (N¹_π N²_−π S₀);

4. parallel time delay (N¹_0.8N²_0.8S₀);

5. opposed time delay (N¹_0.8N²_−0.8S₀).

The two antiphasic conditions produced equivalent masking-level differences (MLDs) for spondee words (5.3 dB re homophasic level). The MLDs for the two time-delay conditions (t = 0.8 msec) were also approximately equal, but were appreciably smaller (3.8 dB). For pure-tone signals, averaged MLDs were also less for time-delayed than for antiphasic listening, but the four nonhomophasic conditions yielded functions relating MLD to frequency that had a common pattern. MLDs increased with frequency between 250 and 400 Hz, were maximal in the range between 400 and 800 Hz, and then declined to very small but fairly uniform values above 1200 Hz. A comparison of the MLDs yielded for pure tones with those for spondees supports the conclusion that MLDs for spondees are mediated by the release from masking that occurs for the spectral components involved in spondee reception at threshold level.