The Chinese writing system was imported to Japan from Baekje around the start of the fifth century, alongside Buddhism.The earliest texts were written in Classical Chinese, although some of these were likely intended to be read as Japanese using the kanbun method, and show influences of Japanese grammar such as Japanese word order.The earliest text, the Kojiki, dates to the early eighth century, and was written entirely in Chinese characters, which are used to represent, at different times, Chinese, kanbun, and Old Japanese.As in other texts from this period, the Old Japanese sections are written in Man'yōgana, which uses kanji for their phonetic as well as semantic values.

Based on the Man'yōgana system, Old Japanese can be reconstructed as having 88 distinct syllables. Texts written with Man'yōgana use two different sets of kanji for each of the syllables now pronounced き (ki), ひ (hi), み (mi), け (ke), へ (he), め (me), こ (ko), そ (so), と (to), の (no), も (mo), よ (yo) and ろ (ro).[7] (The Kojiki has 88, but all later texts have 87. The distinction between mo1 and mo2 apparently was lost immediately following its composition.) This set of syllables shrank to 67 in Early Middle Japanese, though some were added through Chinese influence. Man'yōgana also has a symbol for /je/, which merges with /e/ before the end of the period.

Several fossilizations of Old Japanese grammatical elements remain in the modern language – the genitive particle tsu (superseded by modern no) is preserved in words such as matsuge ("eyelash", lit. "hair of the eye"); modern mieru ("to be visible") and kikoeru ("to be audible") retain a mediopassive suffix -yu(ru) (kikoyu → kikoyuru (the attributive form, which slowly replaced the plain form starting in the late Heian period) → kikoeru (all verbs with the shimo-nidan conjugation pattern underwent this same shift in Early Modern Japanese)); and the genitive particle ga remains in intentionally archaic speech.