The unique orthographic complexities of Japanese, which utilizes multiple types of scripts (morphographic kanji and syllabic hiragana and katakana) for the same spoken language, place unique demands on early learners. Much research has centered on the average ability of Japanese readers, but given the varying challenges of these scripts, attention may need to be oriented towards the outer bounds of ability, specifically those who both deeply struggle (including dyslexia) and excel at reading. In this article, we provide a review of past research on Japanese literacy along with a pre‐registered analysis using quantile generalized additive models that tested two theories in cross‐language literacy research, the orthographic depth and breadth hypotheses. Our results confirmed that the relationships between cognitive skills (phonological awareness and morphological awareness) and reading outcomes differed by script (hiragana vs. kanji). However, their relationships varied in curvilinearity and associations seemed to change based on reading ability. These findings provide supporting evidence for the general scope of the orthographic depth and breadth hypotheses, but may ultimately question the uniformity of these theories.