Submission 103
Capturing Cumulative Semantic Interference in Language Production for Different Semantic Relations with Cosine Similarity
SymposiumTalk-02
Presented by: Marisha Herb
Pooling data from previous experiments, we explore the degree of semantic interference elicited by different types of semantic relations in the continuous naming paradigm (taxonomic/categorical, associative, thematic, unrelated). A distributional semantic model estimating semantic relatedness showed that (1) closer relations lead to larger interference; (2) event relations loosely related to a common theme can induce small but robust interference. We provide an estimate of the sensitivity and broad scope of semantic interference in the continuous naming paradigm and show that cosine similarity can be understood as a common measure of different types of semantic relation.