Proceedings of the Annual Meeting of the Cognitive Science Society

Most Chinese words are compounds formed through the combination of meaningful characters. Yet, due to compositional complexity, it is poorly understood how this combinatorial process affects the access to the whole-word meaning. In the present study, we turned to the recent development in compositional distributional semantics (Marelli et al., 2017), and employed a deep neural network to learn the less-than-systematic relationship between the constituent characters and the compound words. Based on the compositional representations derived from the computational model, we quantified compositionality as the degree of overlap between the compositional and the lexicalized representations as well as the degree of distinctness of the compositional representation. We observed that these two compositional attributes can affect compound recognition over and above the effects of constituent character features and compound features. Moreover, we found that this effect was increasingly stronger when holistic access to the compound meaning became more challenging. These findings therefore, from a computational perspective, provided new evidence for the combinatorial process involved in Chinese word recognition, which also shed light on the universal process of compound comprehension.