[_emacsomancer_]

I understand that. But I don't know what the linguistic equivalence of 'lossy compression' is.

All natural languages use a good bit of 'lossy compression' in a sense: we use a lot of shared world knowledge and context to 'get' meaning of things people say/write.

[Zarel]

I would agree that English is more redundant than Japanese. Knowing both, a mispronunciation is a much bigger deal in Japanese than in English.

Japanese has significant vowel length, which English doesn't. In Japanese, "un" means "yes", and "uun" means "no". Imagine if "yes" and "yesssss" meant opposite things in English.

Vowel sounds in general are also more approximate in English, and differences are often chalked up to different accents.

Words commonly misinterpreted due to mispronunciation by non-native speakers include "kawaii" (cute) and "kowai" (scary), as well as "sawatte" (touch it) and "suwatte" (sit down).

It's _much_ more common for people to stare at you in confusion when you slightly mess up a pronunciation in Japan than in the US.

[_emacsomancer_]

> Japanese has significant vowel length, which English doesn't. In Japanese, "un" means "yes", and "uun" means "no". Imagine if "yes" and "yesssss" meant opposite things in English.

That's sort of true, but you also have to realise that English underwent a sound change that changed the purely quantitative (vowel length) distinction into a qualitative one, changing Middle English long vowels into diphthongs. [And for things that sound very similar but where one means 'yes' and one means 'no', think about English 'un-huh' ( https://en.wiktionary.org/wiki/uh-huh#English ) and 'unh-uh' ( https://en.wiktionary.org/wiki/uh-uh#English ). See the links for both audio files and IPA transcription.]

That just means in modern English vowel length isn't distinctive, but it is in Japanese. English on the other hand has a distinction between /r/ and /l/ that Japanese doesn't ( https://en.wikipedia.org/wiki/Perception_of_English_/r/_and_... ).

In other languages (like Tamil), voicing distinctions in stops aren't phonemic (so 'log' and 'lock' would be non-distinct). And so on.

If as a non-native English speaker you produced 'rog' when your intended meaning was "lock", people would stare at you in confusion too.

[zerb]

The language isn't using lossy compression, but rather it's robust to loss. Sort of the opposite of compression.