The saliency of deep-structure syntactic relations in the memory representation of sentences was investigated using two measures of saliency: accessibility as to-be-recalled information and ability to access other sentence information. A cued-recall task employed sentences in which the semantic, deep-structure syntactic, and surface-structure syntactic relations were varied orthogonally across target nouns. In one condition, the target noun served as the retrieval cue for the rest of the sentence, and in the second condition, all the words in the sentence excluding the target noun were used as a retrieval cue for the target noun. When the sentence served as the recall cue, the surface-structure object of the preposition was better recalled than the surface-structure subject. When the target noun served as the retrieval cue, the surface-structure subject was a better recall cue than the surface-structure object of the preposition. There was no significant difference between the deep-structure subject and deep-structure object of the preposition either in their retrievability given the sentence as a cue or in their ability to cue recall of the sentence. There was a small advantage of experiencer over goal under each of these cued-recall methods. The asymmetrical aspect of the saliency of the surface-structure grammatical relations was related to general methodological issues in the attempts to establish psychological differences between grammatical relations. The finding of no effect of syntactic deep structure was taken as evidence against the psychological saliency of this level of language.
Hiroshi IchikawaHakoda KeitaTaiichi HashimotoTakenobu Tokunaga
Raymond W. KulhavyNancy E. ThorntonThomas E. HancockJames M. Webb