"The IPA is seriously deficient..."

[alice]

"... and thus useless for any kind of serious linguistic work.This is a universal phonetic alphabet which fixes all of its problems and which everyone should start using without delay."

Not a direct quote, but it captures the essence of the website. Any thoughts?

(Try replacing "The IPA" with "English" and "a universal phonetic alphabet" with "an IAL"...)

[zompist]

Seems completely barmy, sorry. Extremely weird symbol choices (q for [ə]??); cumbersome multi graphs (o// for œ); questionable analyses (e.g. mistaking English ɪ/i as distinguished by length); shaky grasp of the uses of phonetic and phonemic transcriptions.

A lot of his concerns are expressed here:

For example, an English speaker, when reading Hindi/Urdu in the universal script, should be able to immediately comprehend that the phone [v] can also be pronounced as a [w], although when reading English in the same universal script, [v] and [w] are pronounced quite differently. Similarly, a Hindi/Urdu speaker should immediately be able to comprehend, when reading English in the same universal script, that [p] and [ph] have the same value in English, unlike the case in his/her own language. As another example, an English speaker, when reading Arabic in the same script, should immediately be able to understand that [p] and［b］are not separate phonemes in Arabic, and such bilabial sounds are usually, but not always, pronounced as ［b］, the [p] being absent in nearly all (but not all) Arabic dialects. Etc. etc.. The universal script must contain and be able to convey this information. That is to say, a truly universal script must convey information on phonemic idiosyncrasy.

"Phonemic idiosyncrasy" is his term for allophony. What he wants is a "universal script" which somehow communicates allophony; he doesn't seem to realize that he can't do that and be a phonetic alphabet. As an example he gives b_∞ for Mandarin /b/, which is supposed to mean "voiced + unvoiced, i.e. that this can be uttered as a b or a p sound". Which has two problems:

1. It confuses allophony with free variation. In fact ［b］and [p] are not in free variation in Mandarin.
2. It loses the phonetic information. The example he gives, 不, is an unaspirated [p], it's not a［b］and definitely not a quantum superposition of p and b.

What he seems to want is a system that can be used both as an actual writing system and as a phonetic transcription; he doesn't seem to realize that the goals conflict, and that a "universal script" is going to produce an awkward and difficult writing system for a particular language.

[----]

At first I was skeptical, since this system is completely unreadable, but all my suspicions were put to rest when I noticed that there are a huge number of transcription mistakes just on this one page. I'm for this 100%!

[alice]

Is this better moved to the quackery thread, then?

Here's something rather less egregious; the basic ideas seem to have merit, even if the execution leaves something to be desired.

[Mâq Lar]

Is this better moved to the quackery thread, then?

Here's something rather less egregious; the basic ideas seem to have merit, even if the execution leaves something to be desired.

The symbols are far too similar to each to be practical. I'm getting a headache just looking at it.

[Ryan of Tinellb]

So, paraphrasing: "I just want a phonetic script where I can look at the word 'cat' and immediately hear every single production of that sound in the history of ever. Is that so difficult?"

[jmcd]

Is this better moved to the quackery thread, then?

Here's something rather less egregious; the basic ideas seem to have merit, even if the execution leaves something to be desired.

He has a good point in giving his reason for making the UPA but the problem could be solved more simply by including symbols from other alphabets e.g. Cyrillic, Arabic.

[finlay]

Is this better moved to the quackery thread, then?

Here's something rather less egregious; the basic ideas seem to have merit, even if the execution leaves something to be desired.

Less? This is worse - it's completely unreadable.

I couldn't work out what any of the symbols on the navlipi page mean. Using so many different languages as examples is such a bad idea... and they're ridiculous symbols too...

[HoskhMatriarch]

The first thing is useless since it doesn't tell you the environments for each phoneme, aside from being poorly designed with the subscript 0 vs. O. The second script is impossible to read because all the symbols are too similar. Also, epiglottal fricatives and trills are the same thing and don't need seperate symbols.

[Hydroeccentricity]

To be fair, Thai glyphs look identical to someone who doesn't know the script. At first glance, everyone is incredulous that any of the symbols can be distinguished; but it works just fine once you get some practice. Presumably anyone who was familiar with this system would be able to read it just fine.

[jmcd]

That's a good point. I'm sure symbols look similar in other scripts as well. One could claim that Cyrillic is harder to read because it doesn't have descenders but I doubt it's actually more difficult.

[zompist]

Unfortunately, it seems like they thought about aesthetics for more than usability. Lots of piddly little differences that would be easy to confuse in handwriting or poor reproduction. Mirror images with no structural function-- not so nice for dylexics. And then actually re-using symbols upside down for vowels. Though I suppose it would be fun to make messages with hidden readings if you flip the page over...

[Richard W]

To be fair, Thai glyphs look identical to someone who doesn't know the script. At first glance, everyone is incredulous that any of the symbols can be distinguished; but it works just fine once you get some practice.

Well done! I find I have to calibrate to an unfamiliar typeface - จ cho chan (/c/), อ o ang (/ʔ/ or /ɔː/) and ง ngo ngu (/ŋ/) are the usual problems. Until I find a ซ so so (/s/), I am rarely sure that a ช cho chang (/cʰ/) is not actually a so so.

On a screen typeface, ด do dek (/d/) and ต to tao (/t/) can have what feels like a one pixel difference and be incredibly difficult to tell apart, and ฎ do chada (/d/) v. ฎ to patak (/t/) has the problem that both are rare. The tone mark อ่ mai ek can be incredibly hard to make out when it runs into one of the อิ sara i family of vowel marks above e.g. อิ่ v. อี v. อี่ and อื v. อื่. The problem is that the marks above get squeezed to harmonise with Roman text with the same vertical spacing.

Handwriting has its own problems. ค kho khwai (/kʰ/) can range from ก ko kai (/k/) to ด do dek (/d/), and mai han akat (/a/) can be indistinguishable from the tone mark mai tho. (Some words are distinguished by one having mai han akat and the other having mai tho.) I had to rely on word recognition for reading my wife's letters while we were courting. (She had to put up with my bad grammar once I'd cobbled together a spell-checker to fix my lousy spelling.)

[Richard W]

That's a good point. I'm sure symbols look similar in other scripts as well. One could claim that Cyrillic is harder to read because it doesn't have descenders but I doubt it's actually more difficult.

Interestingly, fluent English readers largely look at the tops of letters when identifying letters. Perhaps descenders help with recognition by word shape.

[Salmoneus]

To be fair, Thai glyphs look identical to someone who doesn't know the script. At first glance, everyone is incredulous that any of the symbols can be distinguished; but it works just fine once you get some practice.

Well done! I find I have to calibrate to an unfamiliar typeface - จ cho chan (/c/), อ o ang (/ʔ/ or /ɔː/) and ง ngo ngu (/ŋ/) are the usual problems. Until I find a ซ so so (/s/), I am rarely sure that a ช cho chang (/cʰ/) is not actually a so so.

On a screen typeface, ด do dek (/d/) and ต to tao (/t/) can have what feels like a one pixel difference and be incredibly difficult to tell apart, and ฎ do chada (/d/) v. ฎ to patak (/t/) has the problem that both are rare. The tone mark อ่ mai ek can be incredibly hard to make out when it runs into one of the อิ sara i family of vowel marks above e.g. อิ่ v. อี v. อี่ and อื v. อื่. The problem is that the marks above get squeezed to harmonise with Roman text with the same vertical spacing.

Handwriting has its own problems. ค kho khwai (/kʰ/) can range from ก ko kai (/k/) to ด do dek (/d/), and mai han akat (/a/) can be indistinguishable from the tone mark mai tho. (Some words are distinguished by one having mai han akat and the other having mai tho.) I had to rely on word recognition for reading my wife's letters while we were courting. (She had to put up with my bad grammar once I'd cobbled together a spell-checker to fix my lousy spelling.)

Dear lord. Several of those pairs I can't tell apart even now - I've had to zoom in really far to spot the difference, and i STILL can't see the difference between ฎฎฎฎฎฎฎ... are you sure you didn't type the same letter twice?

The big difference, though, between that and the system in question is that while Thai may have many indistinguishable pairs, it also has many distinguishable pairs - letters can be recognised and any ambiguities can be deduced from that context. The problem with systems like the one in question isn't that any pairs are indistinguishable, as they aren't, but that there's a low degree of immediate reconisability for all the characters, so it would be hard to recognise word shape etc.

[M Mira]

Dear lord. Several of those pairs I can't tell apart even now - I've had to zoom in really far to spot the difference, and i STILL can't see the difference between ฎฎฎฎฎฎฎ... are you sure you didn't type the same letter twice?

To patak should be ฏ, so yes a typo.

[Richard W]

Dear lord. Several of those pairs I can't tell apart even now - I've had to zoom in really far to spot the difference, and i STILL can't see the difference between ฎฎฎฎฎฎฎ... are you sure you didn't type the same letter twice?

As Mira pointed out, I did accidentally type the same letter twice - they're on adjacent keys. I apologise. To see them side by side, here they are in the order do to : ฎฏ. And again, so you can see the difference: ฎฏ. The difference is the barely visible squiggle at the bottom right - all one can easily see is the horizontal extension to make room for it. The squiggle, or rather, dent, is the diacritic deriving ฏ to patak /t/ from the ฎ do chada /d/, from the other, just as ต to tao /t/ is derived from ด do dek /d/ and ซ so so /s/ from ช cho chang /c/. The two obsolete Thai letters are similarly derived from others by means of a dent. The bottom left of ฎ do chada /d/ and ฏ to patak /t/ should project under the preceding spacing glyph, but I've a feeling OpenType never supports special start of line forms, and typewriters certainly don't.