The aftermath of compounds: Investigating Compounds and their Semantic Representations
This work addresses the problem of aligning computational models with human semantic processing in psycholinguistics, offering incremental insights into embedding-based modeling.
The study investigated how well computational embeddings align with human semantic judgments for English compound words, finding that BERT embeddings better capture compositional semantics than GloVe and that predictability ratings strongly predict semantic transparency in both human and model data.
This study investigates how well computational embeddings align with human semantic judgments in the processing of English compound words. We compare static word vectors (GloVe) and contextualized embeddings (BERT) against human ratings of lexeme meaning dominance (LMD) and semantic transparency (ST) drawn from a psycholinguistic dataset. Using measures of association strength (Edinburgh Associative Thesaurus), frequency (BNC), and predictability (LaDEC), we compute embedding-derived LMD and ST metrics and assess their relationships with human judgments via Spearmans correlation and regression analyses. Our results show that BERT embeddings better capture compositional semantics than GloVe, and that predictability ratings are strong predictors of semantic transparency in both human and model data. These findings advance computational psycholinguistics by clarifying the factors that drive compound word processing and offering insights into embedding-based semantic modeling.