[yummyfajitas]

From what little I've seen of French, the French person just knows how Poisson should be pronounced and does it.

This is even how most of english works. And there is actually a much better phonetic correspondence between Devanagari and English phonemes than between roman letters/letter pairs and English phonemes.

Consider "The" (as in the 3 letter word) vs "Theranos". The roman string "the" represents a different sound in both words. In Devanagari the two sounds are represented by "द" and "ये" (the letter is य means "tha" and the े changes the "a" to "e" as in "egg"), respectively, so you'd transliterate to द and येरनौस respectively. (A native speaker please correct me if I'm getting this wrong.)

It's possible that in German and Italian, the correspondence between letters and sounds is a lot better than in English. But from what I can tell, devanagari would make a more phonetic English alphabet than Roman letters do.

[davidivadavid]

English is notoriously awful when it comes to phoneme/grapheme correspondence (French is pretty close).

Essentially what you're saying is that Devanagari is superior because there's something closer to a phoneme/grapheme bijection. Which may or may not be true, even though it wouldn't surprise me (I don't know it at all, but I assume some English sounds aren't as trivial to transliterate).

But that's entirely orthogonal to the alphabet. In fact, you could perfectly map Devanagari (or IPA) to unambiguous roman syllables, so that the "Th" sound in "The" is written differently than the "Th" sound in "Theranos." It's just a matter of coding.

[yummyfajitas]

Which may or may not be true, even though it wouldn't surprise me (I don't know it at all, but I assume some English sounds aren't as trivial to transliterate).

From what I can tell Hindi phonemes are a superset of English phonemes. I can't think of any that don't transliterate well, certainly not any common ones. I've been told by several native Hindi or Marathi speakers that they rarely run into a western phoneme that they can't already pronounce.

I agree that we could make a better roman-like alphabet. My only point is that Devanagari already seems to do a good job of this, and this may be why it seems a bit more complicated than Roman letters - it expresses things that Roman letters don't.

[vinay427]

> From what I can tell Hindi phonemes are a superset of English phonemes.

This isn't exactly correct. There are a variety of allophones that many Hindi-native speakers use in their English that aren't used by most native English speakers outside of India such as the retroflex t, trilled or flap r, and others. There are also others missing from Hindi entirely such as /ʒ/.

See phoneme tables for English and Hindi/Urdu (in IPA for instance) for more details.

I agree with your premise, in any case, that Devanagari describes far more than the English/Roman script.