I've been thinking about how Japanese people can read a text (out loud, say) when the readings of the kanji can be so variable depending on okurigana, suffixes, or prefixes that change the meaning and/or expected part of speech, all of which can impact the reading of a kanji. This implies that you have to "read ahead" before you can be expected to pronounce the phrase correctly. In English, you are mostly guaranteed to have all the information you need to pronounce a word correctly after seeing all the letters of the word, with a few exceptions like "read" [red/ri:d] depending on part of speech, but in Japanese this distinction is muddier because word boundaries depend on correct word recognition in the first place. What are the worst examples you can see of this phenomenon in Japanese?
Although this view is in part informed by many examples which I cannot fully recall, the ones I can think of at the moment are:
- 一昨日【おととい】 / 昨日【きのう】 / 日【ひ】
- 一本気【いっぽんぎ】 / 本気【ほんき】
- 今日【こんにち】は、先生！ / 今日【きょう】は、先生が来てくれます。
There are of course also many words in Japanese whose readings are not distinguishable despite any amount of context, because the readings are interchangeable (family names often fall in this category), but I am interested in readings which are distinguishable but require a lot of context before they can be placed correctly.
By the way, what do you usually do when a kanji has multiple readings with different meanings but context can't distinguish them (because they are the same part of speech or grammatical role)? I had this issue recently with 開く【ひらく】 / 開く【あく】, and when I asked about it I was told that あく is more commonly transitive than ひらく. It still seems like a mess, though, even for a native.