L2/04-???

Devanagari Eyelash Ra

Indic Working Group,
November 7, 2004

Eyelash ra refers to the written sign which represents some kind of /r/ sound.

All sources agree that when a /r/ sound follows another consonant (in pronunciation), it is written either as a stroke attached to that other consonant (e.g. ) or a circumflex-like sign below it (e.g. ). In the remaining cases, the /r/ sound is written in one of three ways: using , using a repha over the next consonant, or using an eyelash ra . The description of those cases differs across the sources.

A Marathi reference grammar, Maxine Bernsten and Jai Nimbkar, Philadelphia : South Asia Regional Studies, University of Pennsylvania, 1975 (PK2357.B39 1975).

The eyelash ra and repha forms are discussed on page 22:

Side note: the Note in the middle of the text above is not a problem for character encoding, as the two words have a different written form with different representation in Unicode: in the first word, RA and VA are in the same consonant cluster (i.e. the representation is <RA, VIRAMA, VA>), while they are in separate clusters in the second (i.e. the represention is <RA, VA>). The second word also illustrates the disconnect between the pronounced form (pure /r/ consonant, without a vowel sound) and the written form (full RA, without any indication that the inherent /a/ is not pronounced).

Learning Marathi, Kalyan Kale and Anjali Soman, Pune : Shri Vishakha Prakashan, 1986 (PK2355.K35 1986).

The eyelash ra and repha forms are discussed on page XXVI:

Note the contrast between the first example where a repha is used and the last example where an eyelash ra is used. While the two contexts are not equal, they are essentially equivalent for our discussion, i.e. these two written words show a contrast between repha and eyelash ra.

A descriptive grammar of Nepali and an analyzed corpus, Jayaraj Acharya, Washington, D.C.: Georgetown University Press, 1991 (ISBN 0878402829).

This author analyzes Nepali syllables as being of the form (C)(G)V(G)(C), where the first C is the onset consonant, the first G is a glide (/y/ or /w/) he calls the pre-peak satellite, V is the peak vowel, the second G is a glide he calls the post-peak satellite, and the final C is the coda consonant. He then describes the display of a RA:

Peri Bhashararao gives in L2/02-402 a minimal contrast in Marathi, आचार्यास “to the teacher” vs. आचार्‍यास “to the cook”; (from Naik, B.S., Typography of Devanagari-1, Bombay: Directorate of Languages, 1971).

Gautam Sengupta gives the following minimal contrast in Marathi: दर्या /darya/ “ocean” vs. दर्‍या /darya/ “valleys”.

The Dravidian languages have a retracted R sound, which is still written in modern Malayalam using the sign and in modern Tamil using the sign ; that sound is written in older Telugu using the sign , and in older Kannada using the sign . (older is very relative; e.g. the Telugu sign was still taught a few decades ago). Those signs have the usual inherent vowel, which is "removed' in writing by the usual devices (half-form, explicit virama, whatever).

Hindi does not have that sound, and the “core” Devanagari script does not contain a sign for it. In order to transcribe the Dravidian languages in Devanagari, the combination of a RA with a nukta is used as the target of the various RRA signs.

Unicode 2.0 prescribes the use of RA+VIRAMA+ZWJ to represent the eyelash-RA. This is captured in what was then rule R5 (which is now rule R5a, with the words “for compatibility with The Unicode Standard, Version 2.0” inserted).

ISCII 1991 encoded the RRA (at D0 208) and gave it the name “Consonant Hard RA (Southern Scripts)”. ISCII also described the eyelash-RA as the half-form of the RRA (on page 12, under a note).

Unicode 3.0/4.0 reflected the ISCII choice, in what is now rule R5: “In conformance with the ISCII standard, the half-consonant form RRAh is represented as eyelash-RA. This form of RA is commonly used in writing Marathi and Newari.”, and U+0931 ऱ DEVANAGARI LETTER RRA was annotated with:

The RA with a nukta is encoded as U+0931 ऱ DEVANAGARI LETTER RRA, and is annotated:

• for transcribing Dravidian alveolar r
• half form is represented as “Eyelash RA”

First, it is clear that ISCII treatement of the eyelash-RA as a half-form of RRA has no basis, as the two signs have different origins. At the most, eyelash-RA is an alternate form of RA and representing it by some decoration on RA rather than RRA (as Unicode 2.0 did) is much more sensible. That being said, the particular choice of RA vs RRA is only secondary to our discussion.

There is no question that representing an eyelash RA by the sequence RA VIRAMA ZWJ is sufficient to implement rendering systems; existing implementations are a proof of that. From that point of view, many other conventions would have worked just as well, starting with RRA VIRAMA ZWJ, and continuing with (why not?) NNNA VIRAMA ZWJ and LLLA VIRAMA ZWJ. In fact, any character not yet encoded in Unicode could be accomodated by using the convention that it is represented by, say, U+20AC ZWJ. Of course, such a representation would not make much sense; the point here is that the decision of character encoding is based on more than “can it be implemented that way”.

The real question is whether the eyelash RA is the rendering of an abstract character distinct from RA (or RRA, or NNNA or LLLA). If we looks at the constrastive pairs of section 1, the answer has to be positive. In other words, we need to encode a new character.

The next question is whether a fragment like र्‍या should be represented as a single written consonant cluster, i.e. whether there should be a VIRAMA between our new character and YA. Phonetic considerations alone are not enough to decide this question: there are many examples of single phonetic clusters realized as multiple written clusters (e.g. Hindi उपदेश which is pronounced without a vowel between /p/ and /d/). However, these examples involve words derived from other languages (Sanskrit in this case) where the vowel was pronunced. There is no evidence of such historical development for the eyelash ra, therefore we propose that the fragment should be represented as a single written cluster, i.e. without a VIRAMA. At this point, there are no known cases where an eyelash RA would be the single character in a cluster.

The proposal is to encode a new Devanagari character. The proposed name is DEVANAGARI EYELASH RA, with the same properties as the other Devanagari consonants, and no decomposition.

Rendering behaviour: This character is not known to ligate with other consonants in the same cluster. All known uses of this character are in written clusters involving other consonants; therefore, the its visual appearance as the single consonant of a cluster is not known.

Usage: all the known uses of this character in the orthographies of modern languages are as the first consonant of a cluster involving another consonant. For example, दर्‍या is represented by the sequence <DA, EYELASH RA, VIRAMA, YA, VOWEL SIGN AA>.

Document History

Author:  Indic Working Group