| dc.description.abstract | Semantic and phonological systems interact during word processing. However, the current approaches to studying these systems tend to examine them as separate entities with a focus on processes that occur in those systems. An alternative approach is to examine the underlying representations of these systems with the use of the computational and mathematical tools of Network Science. The analysis of language networks, where nodes represent words and edges represent relationships, have shown that network structure influences language processes. The present study analyzes a novel phonological network using collected phonological association data. 1,018 participants provided up to three phonological associates to a cue word. The cue and response words were used as nodes in the phonological association network, and edges were placed between cue and response pairs. The resulting phonological association network structure exhibited several characteristics, like small-world structure and assortative mixing by degree that were similar to the well-studied one-phoneme difference phonological network, but the phonological association network was also different in structure from the well-studied one-phoneme difference phonological network. In addition, three age-related phonological association networks were examined that represented young adulthood, early middle adulthood, and late middle adulthood. However, there was little phonological network structure change across these age-related networks. Lastly, cutting-edge research in Network Science that uses multiplex networks was employed to examine the semantic and phonological systems simultaneously. This multiplex consisted of two layers: semantic associations and phonological associations. Cue and response words were used as nodes and edges were placed between cue and response pairs in their respective layers. The two layers are distinctly different in their network structure as they represent different aspects of the mental lexicon. However, there was overlap between layers, or instances where a pair of words was connected in both the semantic and phonological layers. Regression analyses were conducted to further assess the influence of single-layer and multiplex network structure on behavioral performance. Specifically, the reaction time for visual lexical decision and naming were predicted using semantic degree, phonological degree, aggregated multiplex degree, multidegree, and the interaction between semantic and phonological degree. The results of a model building procedure indicated that all of the degree measures were needed in the regression analysis model, providing evidence that multiplex structure and the interaction between layers is important to word processing. In sum, the findings from this study provide evidence that phonological associations can be used to construct a representation of the phonological system, that phonological network structure does not significantly change with increasing age, and that the multiplex structure is important to language processing. | |
