THE EFFECT OF SEMANTIC NETWORK STRUCTURE ON PAIRED-ASSOCIATES LEARNING PERFORMANCE

Abstract

Past studies have established links between closeness centrality and word degree with associative learning, but little is known about the effect of both measures taken together, as well as the presence of possible interaction effects. This study thus aims to investigate the combined effects of closeness centrality and word degree on associative learning, as well as to explore and detect possible interaction effects. Using a classic paired-associates learning task, two separate experimental designs are carried out, either with the use of words or nonwords as the target response item. Within-subjects analysis of variance (ANOVA) revealed that higher closeness centrality leads to improved associative learning. Interaction effects in the word study also hint that higher word degree improves associative learning, but the effects of word degree may be sensitive towards levels of closeness centrality. Additionally, differences in experimental yields suggest that processing costs are incurred when novel nonwords are introduced and adopted in place of words as response items. Overall, an advantage of higher closeness centrality on associative learning in large-scale semantic networks is evidenced, as closeness centrality is more predictive of associative learning than word degree, and the effect of word degree is dependent on the level of closeness centrality.