Lexical network properties in aphasic discourse production

Evaluating discourse production provides an effective means of assessing the language competence of persons with aphasia (PWA) in naturalistic and ecologically valid communication contexts. Existing research on PWA discourse has focused on discrete linguistic measures such as sample length or type-token ratios. To provide a holistic view of the interconnections of linguistic entities in PWA discourse, this study compared the lexical networks of discourse produced by PWA with those by healthy controls. At the macro level, PWA networks showed a smaller size and diameter but higher density compared to those of healthy controls. In addition, networks of both groups showed small-world characteristics. However, the degree of small-worldness was lower for the networks of PWA compared to those of healthy controls. At the meso level, network modularity measures of healthy controls were larger than those of PWA in the story-narrative task. At the micro level, the degrees of nodes in the networks of both groups followed a power-law distribution, suggesting that their networks were scale-free. Our findings reveal that lexical networks of PWA discourse are constrained and less efficient but still retain some basic characteristics of typical language networks.