Comment by kazinator

Firstly, the native syllabary uses conventions that give rise to ambiguities. For instance えい (ei) can encode a long e sound like in "sensei" (teacher) or a separate e and i, like in "deiriguchi" (combined exit and entrance).

Secondly, there are many homonyms: words written with completely different kanji that sound exactly the same, very similar (e.g. same modulo pitch accent) or use the same spelling in kana.

Here are two words that don't sound the same at all. Let's use Hepburn, because it distinguishes them: kõri and kouri. One has a long o, the other has separate o and u. In hiragana, both are written こうり: exactly the same. kōri might be 公理 (axiom, self-evident truth) or 高利 (high interest rate). kouri is 小売 (retail, lit. "little selling").

If you're having a conversation with someone and don't know which "kõri" they are talking about, high interest rate or axiom, you can ask. But you cannot just ask a written text. You have to work it out from context.

Kanji eliminate the ambiguity, vastly simplifying reading. Reading Japanese that has been normalized to kana is murder. Especially if spaces are not introduced to mark word divisions. That is only done in hiragana books for small children!

Sometimes this happens in writing. A sentence happens to start with a couple of words that happen to be usually written in hiragana, with some particles after them, and you're left with a puzzle just working out where the word boundaries are. This requires backtracking in the general case; there is no principle like longest match.