Predicting idioms. The role of suprasegmental information in the visual recognition of chengyu (成语)

In language comprehension, predictions on the various linguistic levels are generated (Kutas & Hillyard, 1984; Federmeier & Kutas, 1999; Van Berkum et al., 2005). Comprehension is easier in high-cloze contexts, that is, informative contexts in which the final word is highly predictable for the majority of people. This work aims to provide an insight on the predictive processes behind the comprehension of Chinese four-character idioms, chengyu (成语). Chengyu have been mostly studied from a didactic perspective; however, they are the perfect tool to investigate prediction, for their following characteristics: they (i) exhibit a fixed morphosyntactic structure, usually composed of four characters; (ii) do not admit syntactic or semantic modification; (iii) are inherently highly predictable; (iv) are widely used in everyday language, both written and oral (Guo, 2017; Conti, 2017; 2020; Li, 2023). In this thesis, our goal is to exploit the unique properties of chengyu to test whether supresegmental phonology (tone) is accessed during reading, and whether it is a relevant cue for prediction. In the literature, with respect to syntax and semantics, it is indeed more controversial whether there is prediction of phonological and prosodic information (DeLong et al., 2005; Martin et al., 2013; Pickering and Garrod, 2013; Ito et al., 2020), especially in visual-word recognition (Perfetti and Zhang, 1991, 1995; Perfetti and al., 2005; Zhang et al., 2012). There are several recent studies, all using eye-tracking of printed-words, that have not found an effect of tone in prediction (Shen et al., 2021; Li and al., 2022; Yang and Chen, 2022; Zhao et al., 2023; Xu et al., 2025). Our goal is to contribute to the debate by proposing a different task to test the presence of suprasegmental cues in the literature. We therefore designed a visual-word Lexical Decision Task, with chengyu as our stimuli. We manipulated the final character of the idiom in two conditions: (i) by replacing it with a completely homophonous character (different orthography and meaning), and (ii) by replacing it with a character sharing the same segmental structure but with a different tone. All stimuli were balanced from the orthographic and semantic point of view, in order to isolate the effect of tone alone. By introducing an alternative task design, this study provides a novel method for investigating the processing of formulaic expressions, and aims to reveal differential effect of segmental and tonal cues on predictability.