Effects of Training Method and Word Order on Adults’ Acquisition of Miniature Linguistic Systems

Abstract

Three experiments were conducted to evaluate the effects of training method, English vs. non-English syntax, and number of training items on the learning of two-word miniature linguistic systems. Language matrices were constructed consisting of nonsense syllables representing modifier and object referents. Adult subjects were exposed to nonoverlap (diagonal) or overlap (edgewise or stepwise) training conditions. In spite of the different number of items trained, there were no significant differences in the initial rate of learning among the three conditions. Overlap training, however, provided more extensive generalization to untrained items than did non-overlap training. The addition of overlapping training items was sufficient for subjects originally in the nonoverlap training condition to demonstrate recombinative generalization. Although both modifier-object and object-modifier word order rules were employed, the use of English word order did not facilitate learning nor generalization. Equating the number of training items in the two overlap training conditions served to attenuate differences in maintenance and generalization effects. An analysis of errors revealed that object words were produced correctly more often than modifiers, regardless of syntax and training condition. It is suggested that careful selection of training items and frequent monitoring of generalization can enhance the efficiency of intervention programs focusing on lexical and syntactic development.