Devanāgarī (देवनागरी), sometimes called Nagari for short, is a writing system of about 52 primary letters which combine to form syllables. Devanagari was designed for the Prakrit language c. 13th century CE, an intermediate language between Sanskrit and Hindi, and later elaborated for Sanskrit, Hindi, Marathi, Nepali, and other languages.
Devanagari writing is often likened to a washing line: a line is drawn above the words, and the letters are hung out to dry below the line. A break in the line indicates a break between words.
Devanagari is classified as an abugida, which means that each character represents a syllable, not a single letter as in English. If the character is a consonant, the implicit vowel following it is assumed to be a, unless modified by special vowel signs added above, below, after or even before the character.
Each vowel has two forms: an "isolated" form when beginning a word or following another vowel; and another used within a word by use of diacritics called मात्रा mātra. As an example, the forms used with consonants are placed with the letter त्. Note that if there is no vowel sign, the vowel is assumed to be a.
|अ||a||as in about||त (implicit)|
|आ||ā||as in father||ता|
|इ||i||as in sit||ति|
|ई||ī||as in elite||ती|
|उ||u||as in put||तु|
|ऊ||ū||as in flute||तू|
|ऋ||ṛ||as in Scottish heard, trip.||तृ|
|ए||e||long e as in German "zehn". It is not a diphthong; the tone does not fall.||ते|
|ऐ||ai||as in Mail, sometimes a longer ए. In Eastern dialects as in bright (IPA ıj).||तै|
|ओ||o||as in German Kohle, not a diphthong; tone does not fall.||तो|
|औ||au||as in oxford. In Eastern dialects as in German lauft, or English town.||तौ|
|क||k||as in skip.|
|ख||kh||as in sinkhole.|
|ग||g||as in go.|
|घ||gh||as in doghouse.|
|ङ||ṅ||as in sing. Used only in Sanskrit loan words, does not occur independently.|
|च||c||as in church.|
|छ||ch||as in pinchhit.|
|ज||j||as in jump.|
|झ||jh||as in dodge her.|
|ञ||ñ||as in canyon. Used only in Sanskrit loan words, does not occur independently.|
|ट||ṭ||as in tick. Retroflex, but still a "hard" t sound similar to English.|
|ठ||ṭ||as in lighthouse. Retroflex|
|ड||ḍ||as in doom. Retroflex|
|ढ||ḍ||as in mudhut. Retroflex|
|ण||ṇ||retroflex n. Used only in Sanskrit loan words.|
|त||t||does not exist in English. more dental t, with a bit of a th sound. Softer than an English t.|
|थ||th||aspirated version of the previous letter, not as in thanks or the, but like pathetic|
|ध||dh||aspirated version of the above.|
|न||n||as in none.|
|प||p||as in spin.|
|फ||ph||as in uphill.|
|ब||b||as in be.|
|भ||bh||as in abhor.|
|म||m||as in mere.|
|य||y||as in yet.|
|र||r||as in Spanish pero, a tongue trip. Don't roll as in Spanish rr, German or Scottish English.|
|ल||l||as in lean.|
|व||v||as in Spanish vaca, between English v and w, but without the lip rounding of an English w. (IPA: ʋ).|
|श||ś||as in shoot.|
|ष||ṣ||almost indistinguishable retroflex of the above. slightly more aspirated. Used only in Sanskrit loan words.|
|स||s||as in see.|
|ह||h||as in him.|
|क्ष||ksh||as in worksheet.|
त is used here for demonstrative purposes:
|तँ||tan, or 'tã'||candrabindu (lit. moon-dot)||nasalizes the vowel as in French sans. Sometimes shortened to a bindu, in which it can be mistaken for the anusvāra|
|तं||taṃ, tan, tam||anusvāra (lit. after-sound)||Makes the preceding vowel nasal, as in "count" or "Sam". In writing it can substitute for the appropriate nasal consonant when the nasal consonant comes just before one of the first 25 consonants. For ex. in पंजाब (Punjab) the appropriate nasal consonant ञ, instead of being written in full, is represented by the dot (anusvāra) above the प. Thus the anusvāra automatically makes the n sound that comes before the j.|
|तः||taḥ||visarga||produces a "puff" of air after the consonant, and makes the inherent vowel shift towards "e" as in jet. Used in Sanskrit loan words like शान्तिः- peace, छः - six.|
|त्||t||virama||removes the vowel attached to a consonant.|
|तॅ, तॉ||tă (there is no standard transliteration)||cand||This is a modern invention which shortens or modifies the Devanagari vowel, and is used to write foreign; particularly English, loan words, e.g. टॉर्च torch; फ़ट बॉल football.|
One of the things which appears daunting to most beginners are the over 100 conjunct characters. These happen when two or more consonants are joined together (with no vowel between). Upon seeing all these, the new learner might gasp, thinking that they will have to memorize each one as if they were Chinese ideograms. The good news is that most of these are quite simple and merely involve dropping the inherent 'a' stem. e.g.:
- त् + म = त्म
- न् + द = न्द
- स् + क = स्क
However there are a few special constructions. For many of these, you may also use the previous method though. e.g.
- त् + त = त्त
- ष् + ट = ष्ट
- क् + ल = क्ल
Most often odd forms arise, in consonants without a stem. e.g.
- द् + भ = द्भ
- ह् + ल = ह्ल
- ट् + ठ = ट्ठ
Do not worry to much about conjuncts though, you may always suppress the inherent 'a' with a halant.
Another thing which causes problems for new learners is the use of र, which is treated as a vowel as in Hindi it is a "semi-vowel." There are three forms for conjuncting र, and one for ऋ:
1. After a consonant with a stem add a slash from the lower half of the stem (top-down, right-left). e.g.:
- प् + र = प्र
- क् + र = क्र
- ग् + र = ग्र
- श+ र = श्र
- त् + र = त्र.
2. After a vowel and before a consonant र is written as a small hook (a good mnemonic trick is to picture a stylized lower case r). This conjunct cannot occur alone, nor begin a word. Therefore, an example shall be given within the context of words:
- गर्म hot
- सिर्फ़ only
- कर्म karma (In Sanskrit, the last inherent vowel is not written long as it is in Hindi)
If followed by ā, ī, e, o, or ai the "hook" is moved one letter to the right, e.g. the name Marco would be written: मॉर्को.
3. In most letters without stems, the र is joined to the consonant by placing a circumflex-like diacritic below the letter, e.g.:
- द् + र = द्र
- ट् + र + ट्र
- ड् + र = ड्र
4. ऋ when preceded by a consonant is written as a small hook resembling the Polish ogonek attached to the stem. Only occurs in Sanskrit loan words, most notably the word Sanskrit" itself: संस्कृत.
Finally, र has two special forms when followed by u, and ū respectively:
- रु ru
- रू rū
Punctuation is the same as in English, except for the period, or full stop called the विराम virām: "।". When a question is used with a question marker like क्या kya, meaning what; no question mark is needed. In speech when no question marker is used, there is a rise in intonation towards the end of the sentence. Example, is he a good boy?:
क्या वह अच्छा लड़का है? — kya voh accha laṛka hai?
क्या वह अच्छा लड़का है। — kya voh accha laṛka hai?
वह अच्छा लड़का है? — voh accha laṛka hai?
Devanagari is quite regular, but there are a few pronunciation quirks to watch out for when using it to read Hindi.
"-a" though usually pronounced short, is always written long at the end of a masculine word (the exception are Sanskrit loan words) as a visible mas. marker -ā. The feminine "-ī" marker is pronounced as written.
When ह follows an inherent vowel as in ताज महल (tāj mahal), the 'a' preceding the 'h' becomes an 'e', as in यह (yeh = this), thus pronounced tāj mehal. Thus the transliteration in such cases is deliberate and not a typo! Another noteworthy aberration is वह (voh = that). Fortunately these are a few of the only words that aren't phonetically pronounced in Hindi. There is also a diphthong -आय which is pronounced as the 'i' in 'high', e.g. चाय (cāy) = tea'. And a double consonant isn't just there to look pretty, hold that consonant's sound a little longer. Finally, the final -ā is purposefully written without the macron, as this is misleading as to the pronunciation, which is more like a schwa sound. If this were Sanskrit, it would be practical, but not here. Just remember the inherent 'a' is always written at the end of a mas. word in Hindi.
The semi-vowel "ऋ" is normally transliterated in Roman as an "r" with a diacritical ring below. This semi-vowel is pronounced like "ri", but slightly trilled as in rip. Unfortunately, the proper Roman diacritic doesn't appear to be supported yet by unicode. It can be found in श्री कृष्ण (śrī kṛṣṇa) - "Lord Krishna"). For now the diacritical bindi (dot) will have to suffice for both of the flapped r's. Ambiguity shouldn't cause too much problems, as the trilled r in कृष्ण (kṛṣṇa) or ऋषि (ṛiṣi) occurs only in Sanskrit loan words, and is very rare in Hindi. In addition; if you are familiar with Devanagari, that should resolve any remaining confusion.