Compound meanings predicted better by analogy than global rules
Compound interpretation is based on analogy
Computation and Language
Summary
Figuring out what a compound word means based on its parts is tricky for computers. The authors propose a new method that combines the meanings of the parts with patterns found in similar compound words, like seeing how those parts have been used before. This method, called the Compound Analogy Model (CAM), does better than an earlier approach on Chinese compounds, especially for common two-character compounds. It also better predicts how quickly people recognize these compounds, suggesting it’s closer to how our brain understands word combinations.
What this means in practice
- •For natural language processing teams: Improve compound word meaning prediction in language models using analogy-based embedding methods tailored to Mandarin Chinese.
- •For language technology developers: Develop more accurate text analysis tools to handle compound words by incorporating analogical semantic structures.
Authors
Tian Shen, Harald Baayen
Abstract
How compound meanings are best predicted from constituent meanings remains a central question in computational models of lexical semantics. Comparing different computational models provides a way to evaluate alternative accounts of how semantic information is combined during compound comprehension. We propose a new model, the Compound Analogy Model (CAM), that predicts a compound's embedding by adding its constituent embeddings together with the average shift vectors of the two constituents' compound families. The resulting model is parameter-free and exploits local analogical structure in the semantic space. We evaluated CAM against the CAOSS model on Mandarin Chinese compounds. CAM consistently achieved higher prediction accuracy than CAOSS on both training and held-out data, with the exception of three-character compounds, for which analogical generalization is constrained by both small constituent families and a pronounced imbalance in family size between the two constituents. The advantage of CAM remained when evaluation was based on frequency-defined train-test splits that better approximate generalization from familiar to novel compounds. To assess the cognitive plausibility of the two models, we further examined whether model-derived semantic measures predict visual lexical decision latencies for two-character compounds. Predictors derived from CAM provided improved prediction for response latencies compared to predictors derived from the CAOSS model. These findings indicate that compound meaning is better characterized as local analogical generalization than as the application of a learned global linear transformation, and demonstrate that analogical semantic structure provides a cognitively plausible basis for compound comprehension.