> You need to know where to read 1-3 letters ahead to read a word and interpret it. That's not really a thing in English
It happens in a English too, where you see a chunk of letters and mis-predict which word they represent in a way which affects its meaning [0], and sometimes that will also affect pronunciation. [1]
An example from the link:
> "The complex houses married and single soldiers and their families."
A reader linearly scanning along doesn't know whether "complex" is an adjective or a noun, and then whether "houses" is a noun or a verb. I'm pretty sure all human languages have similar problems where a certain amount of look-ahead or backtracking is necessary.
For another example to highlight pronunciation changes, consider the ambiguity of:
"I saw the rhino live in the zoo."
That could mean that the rhino was doing the verb of living, in which it rhymes with "give", or it could also mean that the speaker was seeing it in-person, in which case it rhymes with "drive".
Both rely on intonation (in addition to volume and pauses) for disambiguation, but the fun trick is that in the Chinese version the intonation is an integral part of the lexeme (i.e. it distinguishes between "words").
But I have to say, these kind of sentences (and full-fledged poems) are quite a different beast from simple cases of garden path sentences or syntactic ambiguity[1]. The poem lion-eating poet and the "buffalo buffalo buffalo..." sentence are both highly contrived and unlikely to be understood correctly on the first few goes even with the perfect prosody. They are cool "language hacks", but they do not occur in daily language and I personally believe (although I guess die-hard generative linguists would disagree) that they don't teach us very much about the language itself (except for what are the cool artistic possibilities it opens).
When this happens in English, teachers will label this as "bad English" and ask you to rewrite. That's how the formal language deals with this problem.
If anything, isn't that an informal solution? It relies on other people to complain that they dislike the sentence, without being able to point to any hard-and-fast rule.
The hard and fast rule is that repeating a word right next to itself is generally frowned up. It comes up with “that” a lot, like “he said that, that led to something else”. Sometimes people are doing something clever with the words, but it’s usually just poor English.
Yes, this happens in English too, but to find examples like this you have to go to Wikipedia, or wrack your brain and see if you remember one. In Japanese, almost every other word is like this.
As is usual for Japanese, this sentence contains a mix of Chinese(-origin) ("kanji", e.g. 袋 小 路 文 法 的) as well as Japanese phonetic ("kana", e.g. ふくろこうじぶん) characters. Usually, when in a multi-kanji word, kanji are pronounced with (a time-changed version of) Chinese pronunciation. For example, 文法 is "bun-pou", not "fumi-nori" or something else. However, the first character of the article title (fukurokoubunji), 袋, is "fukuro" here despite being in a four-kanji word. Further, 小 is "kou" here, which is nonstandard enough that its dictionary entry does not even list it as a possible pronunciation! [1] Then 路文 are both in Chinese pronunciation (ji-bun), but this does not necessarily make sense because the word is not split in two down the middle, but instead as 袋-小路-文 (bag-lane-sentence, where bag-lane is English cul-de-sac / blind alley). [2]
Now fukurokoubunji is a bit of a specialised word, so it might not be a great example. But in the rest of the sentence, we find 文, which is always pronounced "bun" (sentence) here, even when appearing separately, but could also (though more rarely) have been "fumi" (letter) — nothing but semantical context helps distinguish. Then we have 正しい "tada-shi-i", where 正 could have been "sei" as in 正確 "sei-kaku" (accurate) or "shou" as in 正直 "shou-jiki" (honest), but it isn't just because しい come after. Similarly, 生 in 生じやすい is "shou"(-ji-ya-su-i), which is conjugated from the base form 生じる "shou-ji-ru" and could have been "u" (生まれる "u-ma-re-ru") or "sei" (先生 "sen-sei") or "i" (生きる "i-ki-ru") or more (生 is somewhat infamous for having many readings). And I could go on: 書 could be "syo" (文書 "bun-syo") but is "ka" (書き出して "ka-ki-da-shi-te" conjugated from 書く "ka-ku").
This is a bit like the comments elsewhere here noting that the Chinese word for "sneeze" is a bad example because it happens to have so uncommon characters in it — and then people point to examples like "onomatopoeia" and "diarrhoea" as similar tricky examples in English. I can't comment on Chinese, but existence does not necessarily say much about frequency.
[2]: This analysis of 袋小路文 is not completely etymologically honest. By the etymology ( https://en.wiktionary.org/wiki/%E5%B0%8F%E8%B7%AF#Etymology_... ), we see that the "kouji" pronunciation of 小路 is really a corruption of ancient "ko-michi", which is a consistent Japanese-Japanese reading of the two characters. However, because "ji" is also an (uncommon) Chinese reading of 路, if you don't know the etymology of the word, the re-analysis is appropriate in the context of how hard it is to read the written language.
> However, because "ji" is also an (uncommon) Chinese reading of 路,
It's not a Chinese reading at all (as you can tell because it's ... wildly out of place with the the actual Chinese-derived readings ろ・る, onyomi are supposed to have semi-regular correspondences with each other and with Chinese Chinese readings). It's really just rendaku of ち, the basic root of fossilized compound みち (with still-salient prefix "honorific" み).
But most importantly, you never really see either 袋 or 小路 and expect them to have any other readings; maybe you'd expect しょうろ if you don't know the latter, but unless you're already literate in a Chinese or are blindly memorizing kanji tables, the other reading of 袋 (たい) probably isn't even salient, because it's one of those kanji that almost always takes its kunyomi even in compounds.
Side note, the line about u-onbin kind of buries the implication that this is a loanword from western Japanese, which is the culprit of several quasi-systematic but unevenly distributed divergences from regular sound changes.
Maybe because I've seen a similar example used before, but I immediately read it correctly the first time. Honestly these sort of 'problems' only ever seem to occur when specifically created to demonstrate this problem and almost never happen in regular writing.
It happens in a English too, where you see a chunk of letters and mis-predict which word they represent in a way which affects its meaning [0], and sometimes that will also affect pronunciation. [1]
An example from the link:
> "The complex houses married and single soldiers and their families."
A reader linearly scanning along doesn't know whether "complex" is an adjective or a noun, and then whether "houses" is a noun or a verb. I'm pretty sure all human languages have similar problems where a certain amount of look-ahead or backtracking is necessary.
For another example to highlight pronunciation changes, consider the ambiguity of:
"I saw the rhino live in the zoo."
That could mean that the rhino was doing the verb of living, in which it rhymes with "give", or it could also mean that the speaker was seeing it in-person, in which case it rhymes with "drive".
[0] https://en.wikipedia.org/wiki/Garden-path_sentence
[1] https://en.wikipedia.org/wiki/Heteronym_(linguistics)