[LittleCloud]

I'm sure what I'm going to write are not new insights but, to add to your point, generally we expect to consume information much quicker through reading than through hearing. When reading, say English, we don't really sound through each word but recognize the shape as you say. More concretely, it's probably our brains' pattern matching of frequently occurring clusters/patterns of letters: "psy" in "psychology" for example, if you spelt that "phonetically" as "saikolojee" I doubt people would recognize that.

I don't know Korean, but I've been told that it has plenty of Chinese loanwords (or Chinese-derived character compound "words") like Japanese, yet because it is less homophonous than Japanese, Korean people nowadays are very able to recognize such words by the shape of them when written in Hangul.

As a native Chinese speaker, Cantonese to be precise, my opinion is probably biased but I find Japanese to be so homophonous (for Sino-Japanese words 漢語 but even for native words 和語 to a degree) that even if Japanese adopted a more "efficient" syllabary like Hangul, words would still be difficult to decipher, spaces included. Pitch accent only helps so much (and of course that isn't written down in kana), the typical examples being 紙 ("kamì": paper), 髪 ("kamì": hair) and 神 ("kàmi": god). I'm trying to indicate the pitch accent in an adhoc manner with the accented letters; note that according to my (native) Japanese dictionary, 紙 and 髪 have the same pitch accent. So I think, a more phonetic writing system for Japanese, that remains efficient for reading, would end up annotating words with semantic or etymological hints, like the idiosyncratic spellings of Latin and Greek root words in English.

Also, the texts we read are allowed to use more sophisticated words, more literary words, with more complicated sentence structures. So while phonetic spelling ought to work to represent ordinary speech, that's not necessarily the case for general written expression. In practice, I find 漢語 in Japanese speech to be surprisingly difficult to comprehend, in the sense that, words are often too indistinct for me to pick them up by ear if I have not heard them in speech before, even if I have seen them in text before and would know the characters already (including their readings 音読み in Japanese).

That's different for me as a Cantonese speaker where I can pick up new literary compound words if their constituent characters are ones I know from other words/compounds, even rather infrequent ones. I would say it's because the sound system of Cantonese has 6 tones, mapping nearly one-to-one with the tones + voiced/voiceless distinction from Middle Chinese, which in turn is a much more monosyllabic language where characters have quite distinct sound values compared to any of the modern Chinese languages.

Incidentally I have a harder time picking up new terms in Mandarin by ear where many characters that are distinct in Cantonese sound the same; and it's widely agreed that Mandarin Chinese has evolved more disyllabic words to compensate. Again, yes, even when I know the compounds already through writing, learning to pattern-match for them in hearing in real time, has been, unfortunately, for me at least, a different story.

(Edit: added below)

I believe also that both Chinese and Japanese got more homophonous because they could get away with it (i.e. people got lazy in pronouncing more intricate sounds) when there was 漢字 to distinguish characters/words when needed. So there's certainly a feedback loop in there. If for the 2 to 3 thousand years of history there was no writing in ideographic characters, there would have been evolutionary linguistic pressure against too many homophones in the languages.

[astrobe_]

> When reading, say English, we don't really sound through each word but recognize the shape as you say

Apparently, some people do and some people don't; and each category is surprised the other exists. I'd be curious to know if the "visual" category is more prevalent for ideogram-based writing systems.

> my opinion is probably biased but I find Japanese to be so homophonous

It's difficult to evaluate what "so" means here, but it seems to me that homophony is made a bigger problem in this thread than it really is.

For instance, in French, we have words that have many homophones. For instance there is "vert" (green) "ver" (worm) "verre" (glass) "vair" (a type of squirrel) "vers" (the "to" conjunction) "vers" (verse) "verrent","verre" (conjugations of a very rare verb that means "to pounce" - probably most French speakers don't even know it; if you use it in a sentence their speech recognition module will probably segfault).

There's also common homophones "père" (father) and "paire" (pair) and "pair" (peer as in P2P), "mère" (mother) and "mer" (sea), "serre" (greenhouse) and "serre" (talon) and "sert" (conjugation of serve), "je suis" ("I am") and "je suis" ("I follow").

Some of these homophones double as homographs, as you can see. They are all "strict" homophones BTW, as we don't have distinctions between short and long vowels like English has, for instance.

But both in written and spoken language, grammar and context usually disambiguate the meaning - if any. Based on that observation, it seems to me it wouldn't be more difficult to figure which "kami" it is than which "vers" it is (worms, conjunction or verse), unless the sentence is specifically designed for that purpose.

Amusingly, the homophones "vair" and "verre" led to a small quarrel in the 19th century about what were the shoes of Cinderella made of - fur or glass? People who want to show off sometimes bring that up, because "vair" is a rarely used word [1].

[1] https://en.wiktionary.org/wiki/vair

[LittleCloud]

Interesting examples from that wicked language, thank you! (Studied French before but unfortunately lost interest.)

You're right about the context and grammar being usually sufficient to disambiguate. I thought about the examples 話す ("hanâsu"; to talk) and 離す ("hanâsu"; to separate) I gave earlier, and I don't remember ever confusing the two in speech dialogue. But it's probably that there are enough non-homophonous near-synonyms for these words in Japanese that would get used in practice, if we imagine contexts where a word could conceivably be confused with another homophone, e.g. 言う【いう】, 喋る【しゃべる】, 語る【かたる】 in the case of 話す【はなす】.

The above words are "native" Japanese words 和語. Definitely the problem of homophones is way less serious for those words than for Sino-Japanese vocabulary, 漢語.

I think Japanese is untypical in that it's a language with a limited repertoire of syllables adopting words from a language with a much richer system of sounds, Chinese, and trying to map the (compound) words character by character. By the way, that probably relates to why most learners find pronunciation of any variety of Chinese to be difficult; the language needs to make the necessary aural distinctions, including tone (famously), which are apparently subtle to non-native speakers.

There are countless two-syllable compound words in Chinese, a good portion being used in common speech, but of the ones adopted into Japanese, many of them turn into essentially two "syllables" also. (Actually, some modern terms are back-borrowings from Japanese coinages through writing, just so that I'm being fair and historically accurate in this comment.)

Of these terms there is at least a few, that in Japanese, would be confused in speech, that in practice the pronunciation gets mangled to disambiguate. Here are some examples:

私立 ("privately established (institution, organization, etc.)") versus 市立 ("municipally established") has the following disambiguation in common speech:

  私立【しりつ】→【わたくしりつ】

  市立【しりつ】→【いちりつ】

科学 ("science") versus 化学 ("chemistry") has the disambiguation:

  化学【かがく】→【ばけがく】 (as if the word were 化け学 but it's never written that way)

Basically the native Japanese reading of a character 訓読み is substituted for the Sino-Japanese reading 音読み in speech, even though the latter is the proper, original reading for that character in the word in question it occurs in.

For the reader's curiosity, there's also one case of pronunciation mangling that I know in Mandarin Chinese:

炎 ("inflammation") versus 癌 ("cancer"):

  In Mandarin, we have 炎 yán, 癌 ái even though the sound value of 癌 ought to have been also yán, homophonous with 岩 according to the rules of sound change over the centuries.  Contrast with the following:

  Cantonese 炎 jim4, 癌 ngaam4, 岩 ngaam4
  Japanese 炎 en エン 癌 gan ガン, 岩 gan ガン

Basically it's the consequence of packing too much information onto individual syllables/characters while Western languages would simply devise longer words (with more syllables) for more sophisticated/technical concepts.

[astrobe_]

> Interesting examples from that wicked language, thank you! (Studied French before but unfortunately lost interest.)

I won't blame you, as this was my thought when I wrote it.