Anderson et al. (1976) found that specific terms, which had not appeared in but fit the context of sentences, were better sentence recall cues than the general terms appearing in the sentences. They interpreted this as indicating that general terms in context were encoded on the basis of appropriate instantiations. Experiment 1, using part of the same material, demonstrated that the advantage of specific cues increased when the general terms were omitted from the to-be-remembered materials, and that other specific terms were also effective cues for the same material. Experiment 2 demonstrated that terms that could not be possible intantiations of any words in the sentences, but that shared common meaning with the sentences as a whole, were as effective retrieval cues as the specific terms. The advantage of specific cues was interpreted as attributable to feature overlap between the specific terms and the to-be-remembered materials rather than to instantiation of the general terms.