Does recognition of single words predict recognition of two?

Participants studied a list of words, presented 1 at a time. When they were tested on pairs of words and asked to indicate whether both had occurred in the list, they gave more positive responses than would be expected on the basis of results from participants tested 1 word at a time. When participants were asked whether either member of a test pair had been shown on the study list, they gave positive responses less often than expected. When participants were asked to indicate how many words in a test pair came from the study list, they answered "2" more often and "1" less often than would be expected on the basis of results from single-item recognition. Accuracy was not affected by testing items in pairs.