Graph Theoretic Features of the Adult Mental lexicon Predict Language Production in Mandarin: Clustering Coefficient

Graph theory has recently been used to explore the mathematical structure of the mental lexicon. In this study we tested the influence of graph measures on Mandarin speech production. Thirty-six native Mandarin-speaking adults took part in a shadowing task containing 194 monosyllabic words, 94 of which consisted of 3 phonemes and were the items under analysis. Linear mixed effect modeling revealed that clustering coefficient (C) predicted spoken production of Mandarin monosyllabic words, while network degree, in this case its phonological neighborhood density (PND) failed to account for lexical processing. High C resulted in shorter reaction times, contrary to evidence in English. While these findings suggest that lexical processing is affected by the network structure of the mental lexicon, they also suggest that language specific traits lead to differing behavioral outcomes. While PND can be understood as the underlying lattice for which a similarity network is created, lexical selection is not affected by only a target word’s neighbors but instead the level of interconnectivity of words (C) within the network.