Comment by Blikkentrekker | Hacker News Reader

Comment by Blikkentrekker | Hacker News Reader

Parent Full thread

It isn't entered phonetically; it's entered structurally.

There are actually three characters that are pronounced the same as three other characters, but those are entered in different ways for structural reasons, and Japanese characters are neatly arranged on a grid in a way that is sensible that does not necessarily correspond to pronunciation well but corresponds well to structure. The character entered with <ti> for instance in pronunciation corresponds more closely to the digraph entered with <tya>, but in structure corresponds more closely to the character entered with <ta>, and is therefore found on the same row.

When all digraphs and characters are summed's close to 90 characters I believe, which does not fit neatly on a keyboard, and then even enter the modern digraphs that are used in loan transcription which do not correspond to any native structure of Japanese which probably add another 20-30 characters which are also made with structural combinations that make sense. <thi> is a common combination to enter something that has nothing to do with the pronunciation <th> would suggest, but it structurally makes sense.

And then on top of that there are various nonsense accents and other modifications that officially have no place in the standard language but find their way in creative language.

If English had 80 characters in it's script, but the characters so neatly corresponded in a structural way that they could be arranged nicely on a grid, typing methods would probably also rely on combining simpler forms to produce them rather than have a keyboard that spans the full range as it's simply more efficient.

I don't think it's fair to say it isn't entered phonetically. Fundamentally typing in Japanese is grounded in providing the phonetics of a word which can then be converted to more precise characters.

As for Romaji based input, certainly there are some liberties taken based on the logical structure of kana that have been to applied to typing for convenience, but for the most part the Romaji has been chosen for phonetic reasons.

It's also worth noting that from an English perspective the sound you associate with <ti> might be different from the sound a Japanese person might associate with it. Just because the usage isn't strictly the English phonetics, doesn't mean the symbols aren't being used phonetically.

> I don't think it's fair to say it isn't entered phonetically. Fundamentally typing in Japanese is grounded in providing the phonetics of a word which can then be converted to more precise characters.

Yet <づ> and <ず> are phonetically indistinct but entered differently; the input scheme is really not based upon phonetics but on structure.

> It's also worth noting that from an English perspective the sound you associate with <ti> might be different from the sound a Japanese person might associate with it. Just because the usage isn't strictly the English phonetics, doesn't mean the symbols aren't being used phonetically.

I think you might confuse the words “phonetic” and “phonemic” here. One can argue that <づ> and ＜ず> are phonemically distinct in modern Japanese but certainly not phonetically, but <お> and <を> are certainly phonemically indistinct and <おう>, <オー> and <おお> are all phonemically and phonetically indistinct as well but entered differently.

More modern linguistic analyses also analyse the sound of <ち> as /ci/ in modern Japanese, due to the almost universal distinction made with <ティ> in modern Japanese, arguing that the historical /t/ has now split in /t/ and /c/, yet /ci/ is entered as <ti> and /ti/ as <thi>, simply because /ti/ is very rare and occurs only in recent borrowings so it makes more sense.

Of course <王> and <追う> sound quite different, but are both entered with <ou> and selecting the appropriate characters from the dropdown.

It really is neither phonetic nor phonemic.

> Yet <づ> and <ず> are phonetically indistinct but entered differently; the input scheme is really not based upon phonetics but on structure.

Not sure what what your point is here. ず＆づ may or may not be phonetically indistinct based on region.

> I think you might confuse the words “phonetic” and “phonemic” here.

I mean phonetic more in the layman sense e.g. relates to the sounds, not the most technical linguistic definition. To be blunt, the linguistic definition is silly anyways since literally no natural language will ever strictly meet it.

> It really is neither phonetic nor phonemic.

I guess my original point was that just because you can type certain kana in different ways that may not be strictly phonetic they make sense as simplifications based on the structures you mention, and they make typing easier.

This does not mean you cannot type using inputs that correspond to more strictly phonetic romanizations e.g. hepburn.

> Not sure what what your point is here. ず＆づ may or may not be phonetically indistinct based on region.

In Standard Japanese orthography, all occurrences of <づ> have been replaced with <ず> except where structural reasons remain; the etymology is irrelevant for the orthography and so is the pronunciation in different dialects. In dialects that maintain the distinction between them, many instances of <ず> are pronounced as <づ>, and dialects in general do not follow the standard structure and grammar.

It is absolutely not the case that some actual spoken dialect of Japanese exists somewhere that follows the distinction between them as written in standard Japanese.

> I guess my original point was that just because you can type certain kana in different ways that may not be strictly phonetic they make sense as simplifications based on the structures you mention, and they make typing easier.

They do, but the input method is not based on phonology but on structure and Japanese orthography which does not strictly follow the phonoogy of Japanese.

My point is that it is not entered how it is pronounced in Japanese, but that word in Japanese has an official hiragana speling, which may or may not correspond with how it actually is pronounced, and a system that neatly maps to that spelling is used to enter the characters, not the pronunciation of it.

The official hiragana spelling of such words came before the invention of i.m.e.'s.

> This does not mean you cannot type using inputs that correspond to more strictly phonetic romanizations e.g. hepburn.

I do not subscribe to the theory that Hepburn is more phonetic or phonemic than Kunrei myself; <hu> seems to be a closer approximation for modern Japanse than <fu> in particular than <tyu>, as in “tube” also corresponds better than <chu> as in “chew” to me.

Having said that, that's harder because Hepburn is lossy with regards to the original hiragana orthography and neutralizes <づ> and <ず> among some others.

Reply on news.ycombinator.com