Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Aging by naming performance: Semantic network structure differences in the young and older adults
Poster C25 in Poster Session C, Friday, October 25, 4:30 - 6:00 pm, Great Hall 4
Jinyi Hung1, Yuya Huang1; 1MacKay Medical College, New Taipei City, Taiwan
The semantic system is crucial in supporting our naming ability. Since it is a dynamic network that changes across the lifespan, we investigated whether and how differences in the properties of the semantic network contribute to older adults’ naming performance. We hypothesized that older adults’ naming performance is related to the structure of the semantic network regardless of its degradation. Sixteen healthy older adults (age range 55-72 years, mean age = 63.7 years, mean education = 12 years) completed the Boston naming task (BNT) and a semantic relatedness judgment task. Participants were subgrouped into three (low, mid, high) based on their naming performance in the BNT. The semantic relatedness task was constructed based on four different semantic categories, within each 3 most common and 3 least common exemplars were selected (Banks & Connell, 2022). Participants made semantic relatedness judgement of all possible pairings between two different exemplars, resulting in a total of 276 individual judgments, on a Likert scale ranging from 0 (unrelated) to 10 (very much related). These 24 x 24 semantic relatedness judgments were studied as an adjacency matrix of a weighted, undirected network in which nodes represented the exemplars, and edges represented the averaged relation between two exemplars. Network properties and sub-communities were constructed and compared among three naming sub-groups. Their results were also compared with that of a group of college undergraduate students (N=7, mean age = 21 years) for discussion. All analyses were performed using the igraph R package. Overall, older adults’ semantic network had lower cluster coefficient than young adults’ (young = 0.74, old = 0.65). Network random walk results showed that both young and older adults had three major network communities, corresponding to animal/fruit, tool, and musical instrument categories (modularity young = 0.52, old = 0.51). The animal/fruit community in the older adults’ semantic network can further parse out a smaller community constituted mainly by low frequency exemplars, such as ‘butterfly’ and ‘ant’, whereas the communities of the young adults’ remained the same. Subgroup network analyses based on older adult’s naming performance showed that the Low performance group had relatively low connectivity among nodes and cluster coefficient (0.60). Network random walk results showed that both High and Mid performance groups had the same three major community structure as the young adults (modularity high = 0.34, mid = 0.50); however, the Low performance group had four major community structure corresponding to category membership (modularity low = 0.44), except that one low frequency exemplar (i.e., fish) was not included in the animal category. The preliminary findings showed that the network analysis captures semantic representation in the young and older adults, and the constructed network structures are in general consistent with the typical category boundary. Moreover, older adults with different naming performance could have different network structures. A semantic network with sparser connectivity and lower cluster coefficient is observed in older adults with poorer naming performance.
Topic Areas: Meaning: Lexical Semantics,