The final article is available via its DOI: https://doi.org/10.1016/j.jml.2020.104170 Abstract The Korean writing system, Hangul, is an alphabetic syllabary which permits a unique opportunity to investigate how both the syllabic and letter similarity structure of the Korean mental lexicon influences lexical processing. We capitalized on the availability of lexical decision behavioral data for thousands of Korean words, and the tools of network science to quantify the similarity structure of Korean words, to investigate the relative and interacting influences of syllable and letter similarity effects on Korean word recognition. Syllable and letter similarity measures were computed at both local and global levels of the lexicon, by considering the immediate neighbors of the target word (i.e., local) and the centrality of the target word with respect to the entire lexicon (i.e., global). Regression analyses showed that both local and global syllable and letter similarity influenced lexical decision performance, revealing a complex pattern of similarity effects where greater local similarity inhibited word recognition, and greater global distinctiveness facilitated word recognition. Furthermore, there were significant interaction effects between syllable and letter levels, suggesting that both syllabic and letter information are processed in an interactive fashion by Korean readers. Our findings demonstrate the importance of adopting computational methods that go beyond standard neighborhood measures and that can quantify the similarity relationships between words at multiple levels. Keywords: syllable similarity, letter similarity, visual word recognition, Korean Hangul, lexical decision, megastudy, language networks