In Study mode, hovering a bubble re-typesets the original Japanese vertically with the reading over each kanji run. Crucially, the reading sits over the kanji only, and okurigana (the trailing kana) stays as plain text, so わ分かった reads correctly instead of smearing the whole reading across the character.
The original Japanese is run through a morphological analyzer (the same dictionary family that powers Japanese input methods), which splits each word and returns its reading. Common ambiguous readings are pinned to the form a reader actually hears (明日 → あした, not あす). It is the readings that matter for learning, so accuracy is the whole point.
Free, 200 translations a week. Open any Japanese or Korean page, press T, then turn on Study mode in the popup to add furigana and native voice.
Install free for Chrome →Furigana is specific to Japanese. For Korean manhwa and Chinese manhua you still get faithful translation and native-voice playback, just no furigana, since those scripts do not use kana readings.
The reading is placed over the kanji run only and leaves okurigana as plain text, the correct furigana convention, and common ambiguous words are pinned to their everyday readings.
