Collecting Semantic Similarity Ratings to Connect Concepts in Assistive Communication Tools

To compensate for the common inability of people with lexical production impairments to access and express intended concepts, we make use of models of human semantic memory that build on the notion of semantic similarity and relatedness. Such models, constructed on evidence gained from psycholinguistic experiments, form the basis of a large lexical database, WORDNET. We augment WORDNET with many additional links among words and concepts that are semantically related. Making this densely connected semantic network available to people with anomic aphasia through assistive technologies should enable them to navigate among related words and concepts and retrieve the words that they intend to express. 1 Background and Motivation In this section, we briefly review a debilitating language disorder known as anomic aphasia: the inability to access, retrieve, and produce words. Technology can aid people suffering from the failure to generate the words they wish to express. Our work is motivated by the belief that the effectiveness of such tools can be enhanced by our knowledge of human semantic memory.