1993 Volume 11 Issue 2 Pages 103-111
Fillmore (1968)'s case grammar suggested that verbs should be classified according to the logical structures of their case frame and the semantic constraints of each case relation in the case frame. An original binary tree model based on case grammar was designed to explain the memory structure of verbs in semantic memory. Two experiments were carried out to examine the prediction derived from the model, i.e. whether the case frame and the semantic constraints of each case relation that were determined by processing preceded noun phrases serve as cues for retrieving verbs from semantic memory. In both experiments, subjects were required to classify letter strings as words or nonwords under the following conditions: (a) when the target was preceded by a string of two noun phrases (the noun phrase was "a noun+a post-particle"; the post-particle is Japanese equivalent of a surface case marker) (b) when the target was preceded by a string of eight blank circles. Target words were verbs or nonsense syllables. Results showed the presentation of the string of noun phrases facilitated the classification of the appropriate verbs whereas it interfered with the classification of inappropriate verbs. These results were consistent with the prediction.