The distinction between shamsii ("sun-related") and qamarii ("moon-related") letters comes down to a phonetic one, and is useful only in pronouncing Arabic phrases involving two nouns joined by the particle al . This particle, which is usually pronounced "ul" in Urdu, is sometimes a definite article, but very often in Urdu means something like "of the." The phrase thus typically takes the form "Noun of the Noun." The al construction is not productive in Urdu (that is, you don't use it to invent new phrases yourself), but it is enshrined in certain names and other fixed phrases.
Spelling is not a problem: it is spelled alif + laam , just as it should be, and appears at the beginning of the second word in the phrase (so that the second word connects right into it). The problem comes in pronunciation, and here is where the bifurcation takes place.
For the shamsii letters-- which are the fourteen letters te, ;se, daal, ;zaal, re, ze, siin, shiin, .svaad, .zvaad, :to))e, :zo))e, laam, nuun -- one pronounces the phrase by cancelling out the laam in the ul construction and replacing it by doubling the first consonant in the second of the pair of nouns. Thus: ni:zaam ud-diin , muniib ur-ra;hm;aan , malik ush-shu((araa , ((a:ziim ush-shaan . Note that these changes are in pronunciation only, never in spelling. They may or may not be reflected in transliteration, depending on the bias (toward spoken or written forms) of the person doing the transliterating.
For the qamarii letters-- which are the remaining fourteen letters of the Arabic alphabet, alif , be , jiim , ba;Rii ;he , ;xe , ((ain , ;Gain , fe , qaaf , kaaf , miim , vaa))o , chho;Tii he , chho;Tii ye --no such pronunciation change is made. One simply pronounces the ul just as would be expected. Thus: ;ziyaa ul-;haq , daar ul-islaam , shams ul-((ulamaa , mu;hsin ul-mulk .
As for letters that don't appear in the Arabic alphabet, they're not a problem, because they naturally never occur in al constructions. If you ever see such a construction with a non-Arabic word in it, you know it's an error or neologism of some kind.
A quick and easy phonetic way to tell which category a given consonant falls into, is to try treating it as a shamsii letter, by doubling it and pronouncing the phrase that way. If the letter can be so treated, it is usually a shamsii letter. If it sounds awful or is totally unpronounceable as a shamsii letter, then it's probably a qamarii letter. Even better, if you can think of a name you've seen before involving that letter, you're home free. Or if you're really a perfectionist, you can memorize the lists given above.
The case of the shamsii and qamarii letters points up some of the larger difficulties of transliteration: does one want to capture the sound of the word as it is spoken (thus ud-diin ), or the spelling of the word in the script (thus al-diin )? In either case, there are various complexities involved.
Phonetic transliteration: Practical-minded transliterators usually want merely to allude to the word, assuming the reader will recognize it at once. A great trove of examples can be found on billboards advertising Hindi films, on listservs where people quote Urdu in roman script, etc. As long as the intended audience consists of other Urdu-knowers, the exact form of transliteration doesn't much matter (jast az, iph ewe noe inglish wel, ewe kan reed dis).
But to represent or convey the actual sounds of the language in a systematic way takes more effort. At a minimum, it's necessary to indicate: 1) long versus short vowels; 2) dental versus retroflex consonants; and 3) nasalization. All these have to be shown, because they're crucial elements of the phonetic structure of the language. There are many possible ways to go about doing showing them. Very effective, though of course odd-looking, transliteration systems can be created using only a standard typewriter keyboard (doubling the vowel to show a long vowel, capitalizing to show a retroflex consonant, etc. etc.). Other systems involve special diacritics: most commonly, a bar or macron over a long vowel, an underdot beneath a retroflex consonant, etc., with many possible variations and choices involved.
The advantage of this approach is that you don't need to worry about subtleties of spelling, or even which script is used. Spoken Urdu and Hindi are mostly almost identical in their phonetics; the differences are due mostly to regional pronunciations inflected by other languages, rather than to any systematic distinctions. You can usually (though not always) forget about scripts, and just concentrating on capturing spoken sounds.
Of course, you'll then have to decide what to do about the various regional pronunciations: to reflect them (at the cost of shifting sound effects for different speakers), or to standardize them (at the cost of losing the rhythms of actual local speech). What about the varying pronunciations of vuh ? What if somebody says jabaan for zabaan , or gajal for ;Gazal ? What if somebody says Kishan for Krishna, or jogii for yogii? How to render the "phantom vowels" of "eh" and "oh" (see Section 8), which have their own distinct sounds? Many such problems loom, and will have to be resolved.
Script-based transliteration: The project of showing the exact spelling of a word in the original script is more complex, and requires a larger and more elaborate set of differentiations. For example, the four "z"-sounding letters in Urdu must be distinguished from each other, and the three "s"-sounding ones, and so on. And you may want to work out special renderings for, say, the (basically) unpronounced vaa))o in words like ;xvush . (I show it with a "v," while otherwise, if it were pronounced, I would write ;xuush ). The "phantom vowels" of "eh" and "oh" (see Section 8) will need to be crammed awkwardly into short vowels ("eh" as zer or zabar , "oh" as pesh ). Many other such cases appear, and will need to be resolved beforehand.
Special problems to be aware of: The biggest problem most transliterators face is the great difficulty in maintaining consistency. This difficulty takes two forms: practical, and theoretical.
In practical terms, inconsistent or sporadically executed transliteration is almost always worse than none at all. For example, if you saw mahabharat, you might notice that it didn't have any special hints, and conclude that no information about vowel length was being given to you. But if you saw mahaabharat, you would probably think it only had one long vowel; how would you know the transliterator had simply neglected to put in the second, to make it mahaabhaarat? Instead of being uninformed, the reader would then be badly misled. Yet it's hard to train oneself to type the very odd clusters of keystrokes that make for accurate transliteration! I don't know any quick fix for this problem; one just has to keep working on it, and be alert for errors.
In theoretical terms, consistency requires careful and thoughtful advance planning. People too often adopt pre-existing transliteration systems without checking them for overlaps or other sources of confusion.
For example, in the Library of Congress's Sanskrit transliteration system, an "r" with an underdot is the semivowel "ri" (as in Krishna). And in its Hindi transliteration system, an "r" with an underdot is retroflex "r" (as in kap;Raa ). So if you need to use both Sanskrit and Hindi words in your work, you need to change one or the other of these diacritics; or else to mark every word as either Sanskrit or Hindi.
Similarly, in the Library of Congress's Hindi transliteration system, an "s" with an underdot is retroflex "sh"; in its Urdu transliteration system, an "s" with an underdot is the letter .svaad . So if you need to use both Hindi and Urdu words in your work, you need to adjust your diacritic system to eliminate this problem.
And there's an Indo-Persian poet who, in one Persian transliteration system, is called "Bidel"; in the common Urdu transliteration, his name would be "Bedil." Such Persian/Urdu vowel shifts reflect pronunciation to some degree, but often simply rely on conventions of transliteration for the two languages. Such difficulties can definitely be overcome, but only through forethought and care.
A systematic north-south difference: In the north, dental consonants are commonly transliterated plainly, and an extra "h" is reserved for aspiration. In the south, dental consonants are commonly followed by an "h" in order to mark them as dental rather than retroflex. Also, in the south, unlike the north, people don't seem very concerned to mark "sh" as distinct from "s." This kind of regional difference in the south-- which is partly based, I've been told, on certain kinds of carry-over from Dravidian languages-- naturally gives rise to many confusions. For example: the feminine name shaa;Ntaa is commonly transliterated as "Shanta" in the north, and "Santha" in the south. But remember that most people transliterate very casually and unsystematically anyway: they just pick up whatever they see around them. So it's hard to get a complete systematic explanation of such differences, and nobody but a scholar would even think to want one. Just shrug your shoulders, and let it go. If you need to make a transliteration system for scholarly purposes, you'll need to start from scratch anyway..
== on to Section 15 == Urdu script index page == fwp's main page