Why do all romanization systems suck so much?

[chalimac]

Pinyin: 流水 is liúshuǐ?. MPS2 at least gets the vowels right: liou2shuei3.

MPS2: 西師 is shi1 shi1? Really? They don't sound the same to me.

Wade-Giles: 桌 is cho1? I think there's a "u" somewhere.

The ideal crossover system for me would be:

Pinyin but modified with:

- MPS2 vowels + Wade-Giles -ien for pinyin -ian.

- MPS2 ts (pinyin c) and tz (pinyin z).

- Now that c is free to use, use it instead of pinyin g.

- Some sign for the -i in shì or chí. Yale used the r. Something like sh:

Example:

Wǒ sh: Zhūngcuórén。tzàijièn!

Overall I think MPS2 is the strongest but for the confusion of q/ch into ch and sh/x into sh.

[Lu]

Taiwan taught me this: the only thing worse than a bad romanisation system is two romanisation systems.

[renzhe]

I think it's simply impossible to create a phonetic script without any controversy. If you look at all major languages, there are arguments ad nauseam about spelling reform. German has recently been through such a change, and back, and still there are plenty of people unhappy with it.

The things that bother me about pinyin:

- uei, uen and iou get shortened to ui, un and iu

- ui, un, uo, iu, iao, etc. change to wei, wen, wo, you, yao, etc, when alone

- the -i final is ambiguous

- the umlaut is sometimes written, and sometimes not (nv / yu). Also, it is cumbersome to type in many languages

Generally, these things bother people new to the language the most. All of them are compromises of some sort.

- The first one is there to reduce the number of keystrokes/characters

- The second one is to make it easier to write words together

- The last two are there to stick to the latin alphabet and avoid additional characters

At least, this motivation seems reasonable to me, and the decisions made with pinyin manage to achieve this, at the expense of consistency. If they had left things alone without making these changes, we'd have a more consistent script, but it would take 30% more space to write and be riddled with apostrophes and/or strange (non-latin) characters. I think that these decisions make more sense when one considers that pinyin was designed with Chinese speakers in mind, not foreigners learning Chinese.

Overall, I find that the issues with pinyin are minor, and can be quickly learned. I think that mapping initials to single characters is a great thing and preferable to character groups like "ts, hs, tz" and such. So I think that "x, q, c and z" are fantastic and should be left exactly the way they are.

I like pinyin far better than any other romanization system I've come across, despite the minor annoyances listed above.

[gerri]

If you look closely, you'll notice that all writing systems giving the sounds of speech suck like that... they never really give you the way it's actually said, even in standard speech (let alone the variety of accents, dialects, ...).

Who was it who said that "fish" could also be written "ghoti"... the "gh" being an "f" sound as in "enough," and so on... Just see it for what it is: a fickle attempt at avoiding 汉字

[Hofmann]

At first I thought Hanyu Pinyin was stupid, but upon further observation, I realized that it's quite neat and somewhat reflects historical phonology. There are other romanization systems, but they all suck in their own way. Even IPA sucks in that it's difficult to type.

[imron]

Pinyin definitely strikes me as being less sucky than the other alternatives, although the idea behind having different spellings for different tones in Gwoyeu Romatzyh has always intrigued me.

[Guest realmayo]

the idea behind having different spellings for different tones in Gwoyeu Romatzyh has always intrigued me.

Me too. But I guess once fully started on pinyin, it's a real hassle to switch.

[roddy]

I think there's a huge potential advantage for those from an English speaking (well, reading) background to have the tones in the letters somehow, which are after all the 'real' part of the word as far as we're concerned, rather than the diacritics, which are odd little decorations the French use, and as such easily overlooked. You might still not learn to say the tones early on, but I reckon you'd be a lot more likely to at least know what they are so that later on when your pronunciation catches up you don't need to go back and learn them.

That said, I don't think there's enough advantage between an 'ideal' system, and pinyin learnt well to make it worth changing.

[imron]

I don't think there's enough advantage between an 'ideal' system, and pinyin learnt well to make it worth changing.

Agree.

[chalimac]

Thanks for your informative post renzhe. But I don't agree with some of the arguments in favor of pinyin.

The things that bother me about pinyin:

- uei, uen and iou get shortened to ui, un and iu

That's quite a big deal. Someone needs to romanize the romanization.

Actually, comparisons to inconsistent spelling in languages are unfair. Romanization is a crutch to help learners to read chinese. If it drops vowels it doesn't fulfill it's purpose. It forces you to learn how to read the romanization instead. Besides, there are languages that are pretty consistent in spelling-pronunciation like spanish or romanian.

As for the text size argument, nobody reads large enough chunks of pinyin text to make the argument worth.

Edited April 16, 2009 at 09:17 AM by chalimac

[imron]

Romanization is a crutch to help learners to read chinese. If it drops vowels it doesn't fulfill it's purpose

It's also used to teach standard Mandarin pronunciation to native speakers and for dictionary classification by most major chinese-chinese dictionaries (although these also use radical based lookups, pinyin lookup is significantly faster assuming you know the pronunciation). You're also looking at pinyin from an English speaker's viewpoint. In reality, it doesn't really drop vowels because from a native speaker's point of view there are no vowels, there are only initials and finals. ui, un and iu should be thought of as a single unit with a given sound rather than a collection of vowel sounds.

It forces you to learn how to read the romanization instead.

Which is a good thing, because not everyone will pronounce these vowels the same way. Even forgetting speakers of languages other than English, there are huge differences in the way a native English speaker will pronounce these vowels (compare say the way a Scot will pronounce them compared to a Texan). Given that this is the case, people are going to need to learn the correct pronunciation anyway rather than just using the vowel sounds that they are familiar with.

[chalimac]

I think that mapping initials to single characters is a great thing and preferable to character groups like "ts, hs, tz" and such. So I think that "x, q, c and z" are fantastic

But we already have multiple characters initials like sh/zh/ch so why not ts or tz? At least it will stop people from saying names bad like Mao Zzedong or Kao Kao.

[leeyah]

It's also used to teach standard Mandarin pronunciation to native speakers

Yes, since it's invention, Chinese school kids are taught pinyin first, then hanzi

from a native speaker's point of view there are no vowels, there are only initials and finals

Right, that's how it's perceived. Try using the alphabet separately in explaining the pronunciation of, for example, a foreign name which may not exactly fit into already familiar romanised names, to an average native speaker and see where it gets you - right back to the strict pattern of pinyin combinations of initials and finals, and further into "pinyinisation" or better say a distorted pronunciation of the original name

But we already multiple characters initials like sh/zh/ch so why not ts or tz

But "we" already have c for tz/ts: so why write cuo as tzuo/tsuo ?

[chalimac]

But "we" already have c for tz/ts: so why write cuo as tzuo/tsuo ?

Because it is more intuitive for laymen. If cao cao were a street a westerner read in a guide, he will be stuck forever asking for kao kao.

I think many of you hit the root of the problems. What suits chinese children is not necessarily optimal for second language learners. Maybe there should have been a more consistent romanization for learners and occasional readers. But, alas, this is historical-fiction.

[imron]

Because it is more intuitive for laymen.

And here lies the key - the system was never designed for laymen and nor should it be. Even language learners who might find it a bit confusing at first will soon move away from being laymen. Also not all laymen will be native English speakers and therefore although ts tz or hs seems correct to you, it will seem incorrect for speakers of other (latin alphabet-based) languages - which group of laymen to you cater to? The best solution is that you don't.

But we already have multiple characters initials like sh/zh/ch so why not ts or tz?

For consistency and also to avoid proliferation of multiple character groups. There weren't enough letters to go round so there has to be doubling up somewhere, but it's best kept to a minimum. For consistency, sh, zh and ch were good candidates because there is a direct and consistent mapping from sh/zh/ch to s/z/c, with the former simply being the retroflex version of the latter.

The same can not be said for ts tz hs or other multiple groupings.

Take a look at the sound groups for different mandarin initials

b p m f

d t n l

g k h

j q x

z c s

zh ch sh

If you were going to add ts, tz, hs etc, then instead of the nice neat grouping where the only doubled letters belonged to the retroflex group with a direct mapping to their equivalent non-retroflex sound, you'd not only get inconsistent double letters in a single sound group - tz ts s - you'd then also lose the consistent mapping to the retroflex version unless you wanted to then have tzh tsh sh and so on. It just makes things messy compared to the comparatively clean solution used by pinyin.

[chalimac]

imron, your explanation is philologically sound (I'm a philologist myself). But the problem is that although...

the system was never designed for laymen.

pinyin is all over the place in road signs, maps, guides, conversation books, encyclopedias, etc. whose natural target audience are laymen.

pinyin, as any romanization, is a compromise between two extremes: on one side IPA with full mapping of sound to symbol, on the other side things like this funny pre-Wade-Giles by Giles:

Poo yow raw shooey (Bù yào rè shuǐ)

http://www.pinyin.info/romanization/wadegiles/giles_samples.html

Pinyin tends too much to the symbolic side at the cost of sacrificing intuitiveness.

[renzhe]

Romanization is a crutch to help learners to read chinese.

It's more specific than that. It is a crutch to help Chinese people learn to read Putonghua (standard Mandarin). That has always been its purpose and more than 99% of all users of pinyin (my estimate pulled out of thin air) are Chinese.

And Chinese people learning Putonghua in a Chinese setting are less likely to pronounce "shui" as "shoo-ey". Just like "st" in German, which is ambiguous, but most learners pick up when to read it as "sht".

Besides, there are languages that are pretty consistent in spelling-pronunciation like spanish or romanian.

Even they are not perfect. Modern Spanish doesn't distinguish between "b" and "v" ("b de burro y v de vaca"), and between "ll" and "y". So you write "mayor", but "Mallorca", although the origin of the word is the same.

Slavic languages are generally very good about writing phonetically, but even they deteriorate with time. Modern Croatian, for example, is not read exactly the way it is written due to developments in spoken language.

As for the text size argument, nobody reads large enough chunks of pinyin text to make the argument worth.

People regularly type large chunks of pinyin text nowadays, though, as pinyin is the most common input method for Chinese.

Overall, pinyin is not perfect, but I find that all its shortcomings are minor annoyances that one gets used to rather quickly.

[roddy]

pinyin is all over the place in road signs, maps, guides, conversation books, encyclopedias, etc. whose natural target audience are laymen.

If we can just figure out where these laymen are from . . .

[renzhe]

But we already have multiple characters initials like sh/zh/ch so why not ts or tz? At least it will stop people from saying names bad like Mao Zzedong or Kao Kao.

sh, zh and ch are the only exceptions, and they are exceptions for a very good reason -- they highlight the connection of sh, zh and ch to s, z and c. In many southern dialects of Mandarin (like Sichuan), these are merged into each other. Generally, the amount of "tongue-curling" depends on the region, and thus ch, zh and sh move closer to c, z, and s the more south you go.

And I think that using z, c, q and the like is acceptable because these are sounds specific to Chinese which have no English equivalent. Someone who doesn't speak Chinese will not pronounce Chinese names properly, and it's not really important whether they read it as "Mao Zzedong" or "Mao Tazidong", it's still wrong.

English speakers pronounce German names incorrectly too. "Einstein" is pronounced wrong by most English native speakers, and same can be said for many others. Yet Germans and others are not rushing to change their writing system to accommodate this, because the writing system is suitable for writing the German language, and you need basic familiarity with German writing and pronunciation to pronounce them properly.

Same goes for pinyin. It is a Chinese writing system for writing Chinese. "c" is a sound that doesn't exist in English, but it exists in Slavic languages, where it is written "c". so I have no problem with using it.

[renzhe]

Because it is more intuitive for laymen. If cao cao were a street a westerner read in a guide, he will be stuck forever asking for kao kao.

Not really.

A person from a Slavic country would pronounce it perfectly. The same person would have problems with "tsao". Germans pronounce "Tsingtao" beer as "Tuh-sing-tao".

Pinyin was not meant for English speakers, nor is it possible to devise a romanisation system that people all over the world will intuitively understand. If you created a system based on English spelling, then many other groups (think Russians and Vietnamese) would be struggling far more than now.