[Comments-devanagari-gurmukhi-gujarati-scripts-lgr-27jul18] A quick review of the Devanagari proposal

[梁海 Liang Hai]

- §2, “Latin transliteration of native script name: dévanâgarî”: Use a consistent transliteration scheme throughout the document.

- §3.3.1, footnote 5: “/a/ would be misunderstood” only because the authors don’t try to use consistent transliterations.

- §3.3.2, “However, the notion of maximum number of consonants joining to form one akshar is empirical”: Good. Such sensible statements are rarely seen.

- §3.3.3, Table 5: The vowel set seems sketchy. It doesn’t make sense to include letter and sign of vocalic rr but exlude vocalic l and ll. It doesn’t make sense to include letters and signs of oe, ooe, aw, ue, and uue (presumbly all for Kashmiri), but exlude short e and short o (which are also required by Kashmiri).

- §3.3.4: A typical confusion between the grapheme bindi and the phoneme anusvara (note the grapheme bindu/anusvara often represents a phonetic nasalization/anunasika in Hindi, but is encoded as bindu) when trying to introduce seemingly-well-understood orthography but not understanding the context of discussing text encoding. Over-emphasis of certain languages and writing systems’ orthography features. In this document’s concern, bindu/anusvara is just a sign representing certain nasal feature.

- §3.3.6, “… to represent sounds found only in words borrowed from Perso-Arabic”: Not true. Nukta is used for sounds (including languages’ native sounds, including loanword sounds from Perso-Arabic, English, etc, origins) that can’t be represented by the original set of graphemes in Devanagari. If the authors can’t figure out a good summary for a section at the beginning, the section should start with an introductory sentence “Something has following functions:” then.

- §3.3.6, “बढ़ /bədh/”: Use a decent transliteration or phonetic transcription.

- §3.3.8, “Earlier the ZWJ was recommended … However, with the new recommendations in place, this usage of ZWJ is now not encouraged.”: Unclear where this observation comes from. The Unicode Standard Core Specification currently doesn’t state a preference between the two encodings.

- §4.1.2.4: Make §3.3.3, Table 5 consistent with this consideration and §5.2. Authors seem to have a hard time figuring out how to deal with the duplicated information between §3.3 and §4/§5. I suggest §3.3 should only include encoding-ignorant information.

- §5.2, Table 6: Should note the “Indic syllabic category” column is not about the Unicode character property of the same name.

- §5.2, Table 6, row 67: Wrong glyph and name.

- §5.5, “… in the form of variables”: These are not variables but notation.

- §6, “There are no characters/character sequences in Devanagari which can be created by using the characters permitted as per the [MSR] and that look exactly alike.”: Not true. First, WLE is also required to prevent confusables (eg, vowel letter aa vs <vowel letter a, vowel sign aa>). Also, even with the WLE, the case of anusvara following a candra shape (part of vowel letters candra e, candra a, and candra o, as well as vowel signs cadra e and cadra o) should be examined, eg, Marathi बँक (bank) and Hindi हाँग काँग (Hong Kong) can be encoded with either candrabindu or <vowel sign candra e / vowel sign candra o, anusvara> and rendered the same in major fonts (and actually the latter encoding might be semantically preferred by many users, thus might even lead to a “allocatable” disposition).

- §6.1, Table 16: Glyphs should be manually drawn to better illustrate the proper rendering.

- §6.4: Just a feeling, the disposition of “blocked” might be too restrictive.

- §6.5, Table 19: Variants between Devanagari and Bengali don’t seem even close to being as complete as the Gurmukhi ones. Where is Bengali candrabindu, nukta, vowel sign aa, vowel sign ii, vowel sign u, virama, and certain consonant letters?

- §7: A comprehensible pattern for other reviewers’ reference: `C[N][M[N]][B|D|X] | V[N][B|D|X] | C[N]H`

- §7, Case of Eyelash Reph: Unclear what the reason 2 means.

- §7, Case of V preceded by H: This is too restrictive.

Best,
梁海 Liang Hai
https://lianghai.github.io

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mm.icann.org/pipermail/comments-devanagari-gurmukhi-gujarati-scripts-lgr-27jul18/attachments/20181008/1bbb5881/attachment.html>