The three kana that change when they are particles

Almost the only exception in kana

The great virtue of hiragana and katakana is that you read what is written. There is nothing like English "though / through / tough".

But three characters are read differently when they act as grammatical particles. Clear this hurdle and the rest really is phonetic.

を is pronounced the same as お, so it is less an exception than a character reserved for one job. Modern Japanese barely uses を for anything else.

In sentences

• わたしは がくせい です。 → "watashi wa gakusei desu" (I am a student). That は is "wa".
• はな は きれい です。 → "hana wa kirei desu" (flowers are pretty). The first は is "ha", the second is "wa".
• にほん へ いきます。 → "nihon e ikimasu" (I am going to Japan). へ is "e".
• みず を のみます。 → "mizu o nomimasu" (I drink water). を is "o".

How it happened

は really was read "ha". Then, over the Heian period, は-row kana in the middle and at the end of words shifted to わ-row sounds — a change called hagyō tenko (ハ行転呼).

It did not convert every は; it happened word-internally. かは (river) became かわ, こひ (love) became こい.

The 1946 spelling reform then respelled those words to match how they sounded. But the particles は and へ were judged too frequent to change without causing chaos, so they kept the old spelling. That decision is why they are still irregular.

How to keep them straight

The test is simple: is the character part of a word, or is it standing between words doing a grammatical job?

Inside a word it is plain "ha" or "he" — はな (flower), へや (room). Sitting alone after a word, it is a particle, so "wa" or "e".

This sorts itself out with reading practice. Japanese particle slots are predictable, so a lone は after a noun cannot be anything else.