Distributional semantics in aphasia: An exploratory analysis using word embeddings.

Authors

Keywords:

Distributional semantics, word embedding, aphasia, semantic vector.

Abstract

Distributional semantics theory brings a cognitive/linguistic theoretical background to natural language processing. Thus, the analysis of the semantic distance between words could involve the study of language impairments, such as in individuals with aphasia. The present study explored the “Cinderella” discourse of 80 neurotypical individuals, 108 individuals with non-fluent aphasia and 138 individuals with fluent aphasia. Data were recollected from the AphasiaBank database. Semantic vectors were obtained using the word2vec model, a word embedding model to explore the vectorial projections of words and their meaning construction based on the context. Individuals with aphasia showed larger word vectorial distances in comparison with neurotypical controls. Furthermore, individuals with fluent aphasia had the larger word vectorial distance within the aphasia group. This is in concordance with findings of semantic feature impairments in individuals with fluent aphasia. The influence of using a computational approach with theoretical linguistic background in clinical linguistics is further discussed.

Author Biography

Sebastian Bello Lepe, Universidad de Valparaíso

Escuela de Fonoaudiología

Downloads

Published

2022-12-31

How to Cite

Bello Lepe, S. (2022). Distributional semantics in aphasia: An exploratory analysis using word embeddings. Revista Signos. Estudios De Lingüística, 55(110). Retrieved from https://revistasignos.cl/index.php/signos/article/view/927

Issue

Section

Monographic Section Articles