An Introduction to Indic Scripts

Updated 20 October, 2022

This article provides an introduction to the major Indic scripts used on the Indian mainland. Those addressed in this paper include specifically Bengali, Devanagari, Gujarati, Gurmukhi, Kannada, Malayalam, Oriya, Tamil, and Telugu.

You can see a list of the characters represented in any coloured example by clicking on it. The information will pop up from the bottom right of your browser window. You'll see the example text, split into grapheme clusters, with a list of component characters below. You can try it now by clicking on the following: हिंदी

Although the Indic scripts are often described as similar there is a large amount of variation at the detailed implementation level. To provide a detailed account of how each Indic script implements particular features on a letter by letter basis would require too much time and space for the task at hand. Nevertheless, despite the variations at the detailed level, the basic mechanisms are to a large extent the same, and at the general level there is a great deal of similarity between these scripts. It is certainly possible to structure a discussion of the relevant features along the same lines for each of the scripts in the set. It is these common themes that this discussion will attempt to highlight, although we will also point out one or two of the differences along the way.

After some historical and phonetic background, we will tackle the subject in two parts:

  1. In the first part I will survey the visual characteristics of these scripts.

  2. In the second part I will make brief reference to some of the practical ways those characteristics are supported using Unicode.

Historical background

The similarity of features across all these scripts is not surprising if you consider their history. As shown in the illustration below, they all derive from a common ancestor. Note also that these scripts are used for two distinct major linguistic groups, Indo-European languages in the north, and Dravidian languages in the south.

An illustration of the derivation of the character NNA, showing how from a common source (Brahmi) all the different forms arose for the modern scripts.  The diagram shows an early divergence between north and south indian scripts.

The diagram above is an illustration of the derivation of the character NNA, showing how from a common source (Brahmi) all the different forms arose for the modern scripts. The diagram shows an early divergence between North and South Indian scripts. (Adapted from Daniels and Bright, The World's Writing Systems.)

Sounds of Indic languages

One of the defining aspects of a script is the repertoire of sounds it has to support. Because there is typically a letter for each of the phonemes in an Indic language, the alphabet tends to be quite large. The table below shows a superset of Indic consonant sounds in a traditional articulatory arrangement. It is meant to be illustrative rather than exhaustive, so as to give you an idea of the number of sounds most Indic scripts must support. It does not include all sounds, for example a number of Dravidian alveolar sounds are not shown. The table also provides an approximate idea of how Unicode character names map to actual sounds, though it has to be stressed that this is only very approximate. The IPA transcription is shown to the left, followed by the standard Unicode name for that sound. Note the following:

  Uvular Velar Palatal Retroflex Dental Labial
Plosives Voiceless Unaspirated q QA k KA c CA ʈ TTA t TA p PA
Aspirated KHA CHA ʈʰ TTHA THA PHA
Voiced Unaspirated g GA ɟ JA ɖ DDA d DA b BA
Aspirated GHA ɟʰ JHA ɖʰ DDHA DHA BHA
Nasals ŋ NGA ɲ NYA ɳ NNA n NA m MA
Fricatives Voiceless x KHHA ʃ SHA ʂ SSA s SA f FA
Voiced ɣ GHHA z ZA
Flapped & tapped sounds ɽ DDDHA ɾ RA
ɽʰ RHA
Aspirate, semi-vowels and liquid h HA j YA ɭ LLA l LA v VA

There are up to 18 Unicode code points dedicated to vowels in each script block, although fewer than this are actually needed on a per language basis. Nearly always these are simple vowel sounds, although occasionally a symbol may represent a diphthong (especially AI and AU). The Unicode names for the whole list are:

A, AA, I, II, U, UU, VOCALIC R, VOCALIC L, CANDRA E, SHORT E, E, AI, CANDRA O, SHORT O, O, AU, VOCALIC RR, VOCALIC LL

Indic languages are syllabic in nature, and the inherent vowel is an important concept (see below). Unless otherwise indicated, each consonant is typically followed by this vowel sound. The inherent vowel can vary in pronunciation from script to script, and examples include ə, ʌ, and ɔ.

Nasalisation of vowels is also an important phonetic feature that affects the written form of several South Asian languages. The effect is similar to the nasalisation of words like 'en' in French.

Characteristic script features

Direction and positioning of script

All Indic scripts run left to right, although some combining glyphs appear to the left of their base character for display (see the discussion of vowel signs below).

In a number of scripts, characters commonly have a headstroke and a high baseline. Such characters typically hang from the line when written.

Consonants and inherent vowels

These scripts are often called abugidas or alpha-syllabaries.

In this type of script, consonant characters represent a consonant+vowel syllable. The consonant is associated with an inherent vowel that has to be overridden if it is not the required vowel sound for a particular spoken syllable. For example, the character in Hindi (Devanagari script) is pronounced rather than just k. The ə sound is the inherent vowel, and is usually transcribed as a.

Note that the inherent vowel is not always pronounced. For example in Hindi it is not usually pronounced at the end of a word, although a ghost echo may appear after a word-final cluster of consonants, eg. योग्य jogjᵉ, or राष्ट्र ɾəstɾᵉ. In addition Hindi has a general rule that when a word has three or more syllables and ends in a vowel other than the inherent a, the penultimate vowel is not pronounced, eg. समझ səməɟʰ but समझा səmɟʰaː; रहन rəhən but रहना rəhnaː. (For a number of reasons, however, this rule does not always hold.)

Nonetheless, on the whole, Indic scripts are close to phonemic transcriptions. The pronunciation of consonants is typically quite regular and predictable, although there is the occasional exception. The following are two examples of exceptions:

Example 1: voiced aspirated plosives and the non-initial letter HA in Gurmukhi are used to indicate tones rather than sounds. For example a voiced, aspirated plosive in word-initial position represents an unvoiced unaspirated plosive sound with a low tone on the syllable, eg. ਘੋੜਾ kòɽɑ (The primary use of all voiced aspirated plosives in Gurmukhi is to express tone information.)

Example 2: in Tamil, consonants such as are typically phonemic rather than phonetic. In practise, this consonant may represent any of kʌ, gʌ, xʌ, ɣʌ, hʌ.

Most scripts supplement a basic set of letters with additional letters used to represent the sounds of other languages, such as Sanskrit and English. These additional letters are commonly formed by adding a diacritic to an existing letter. This diacritic is called a nukta in the Unicode Standard, although the name used by speakers of different Indian languages may vary. Some scripts use this diacritic with several basic letters (eg. Devanagari), others not at all (eg. Kannada).

Examples:

Devanagari, क़ (cf. )

Gurmukhi, ਲ਼ ɭə (cf. )

Oriya, ଢ଼ ɽʰ (cf. ɖʰ)

It is possible to 'kill' the inherent vowel sound where it would normally be pronounced. This is achieved by attaching a small diacritic mark, called a virama in the Unicode Standard, to the consonant in question.

Examples:

Gujarati, ક્ k ( cf. )

Tamil, க் k ( cf. )

Telugu, క్ k ( cf. )

In the examples that follow, single consonants pronounced without the inherent vowel are depicted with a virama.

Vowel signs

Where a consonant is followed by a vowel other than the inherent vowel, the change is produced by adding a vowel sign (called a matra in Sanskrit) to the base consonant. A consonant can only support one vowel (and one vowel sign) at a time. A vowel sign may appear to the left or right, above or below the base consonant, and sometimes surrounds the base consonant on more than one side.

The following illustrates the use of vowel signs with the consonant in Hindi, and the resultant sounds:

की kiː   के ke   कू kuː

Vowel signs may also appear to the left of the base consonant they are related to. For example:

Gujarati
+ િ [U+0A95 GUJARATI LETTER KA + U+0ABF GUJARATI VOWEL SIGN I] -> કિ ki

Tamil
+ [U+0B95 TAMIL LETTER KA + U+0BC8 TAMIL VOWEL SIGN AI] -> கை kʌy

Oriya
+ [U+0B15 ORIYA LETTER KA + U+0B47 ORIYA VOWEL SIGN E] -> କେ ke

Occasionally a vowel sign may be composed of multiple parts. In some cases such a split vowel sign may have parts on both the left and right of the base character simultaneously, eg. in Tamil + [U+0B95 TAMIL LETTER KA + U+0BCA TAMIL VOWEL SIGN O] -> கொ ko. In Kannada there are no vowel signs that surround a base character on both left and right, but there are some that have multiple parts above and following the base character, eg. + [U+0C95 KANNADA LETTER KA + U+0CCB KANNADA VOWEL SIGN OO] -> ಕೋ koː. Another alternative is top and bottom, eg. Telugu + [U+0C15 TELUGU LETTER KA + U+0C48 TELUGU VOWEL SIGN AI] -> కై kaj. In some cases, the additional parts can be viewed as lengthening marks.

Often the pairing of base character and vowel sign produces a change in the basic shape of either base character or vowel sign or both. Tamil provides many such examples, especially with u and . For example, the following are a selection of the Tamil consonants, each followed by the same vowel sign, [U+0BC1 TAMIL VOWEL SIGN U]:

Independent vowels

Vowels that appear at the beginning of a word or after a preceding vowel with no intervening consonant are typically rendered using independent vowel letters. The following table illustrates the correspondence between the most common independent vowels and vowel signs in Telugu:

Note that there is no vowel sign for the sound associated with the inherent vowel A. Vowel signs are only needed to change the inherent vowel.

Because a consonant (or consonant cluster) can only support one vowel at a time, note the difference in Devanagari between की [U+0915 DEVANAGARI LETTER KA + U+0940 DEVANAGARI VOWEL SIGN II] pronounced kiː and कई [U+0915 DEVANAGARI LETTER KA + U+0908 DEVANAGARI LETTER II] pronounced kəiː. The first example follows the base consonant with a vowel sign; the second, with an independent vowel. In the second case only, the independent vowel sound is retained immediately after the initial k, and the sequence is pronounced with two distinct vowel sounds.

Gurmukhi is unusual in that, with the exception of [U+0A05 GURMUKHI LETTER A] when used for the sound ə, there are no independent vowels. Instead there are special 'vowel-bearer' glyphs (of which is one) that are used to support the vowel signs.

[U+0A05 GURMUKHI LETTER A] is used for ɑ , æ , ɔ ,

[U+0A72 GURMUKHI IRI] is used for ɪ , i , e ,

[U+0A73 GURMUKHI URA] is used for ʊ , u , o ,

Consonant clusters

Where consonants appear together without intervening vowels special steps need to be taken to indicate that the inherent vowels have disappeared. There are many ways in which this is achieved in Indic scripts, and the specifics of how each character behaves are too many to catalogue here in detail for each script. There are, however, two main approaches: either (a) change the shape of the consonants or merge them together in some way (a conjunct form), or (b) use a special diacritic to indicate the absence of intervening vowels.

A number of strategies are used to show consonant clusters by merging or changing shapes, and nearly all scripts employ more than one of these approaches. The following are a few examples:

Another approach is to simply show the virama we introduced earlier. This is really the standard approach for modern Tamil, eg. இந்த intʌ (the dot is the virama), but may also be used for any other script if the font being

There is actually a third approach, and that is to simply rely on the user to recognise contexts where the inherent vowel is dropped. This occurs in some specific situations such as at the end of a word or those examples described earlier. The only script that does this as a general rule is Gurmukhi. Very few letter combinations are handled as conjuncts in Gurmukhi, most of the time the reader just has to know where the inherent vowel is not pronounced, eg. ਉਤਸੁਕ utsuk.

A common feature of Indic scripts is the gemination or lengthening of consonants. For example, note the lengthening of the l sound in चिल्लहट cilːʌhʌʈ (Devanagari). Such consonant lengthening is typically handled just as a normal consonant cluster. Gurmukhi, again, is somewhat non-standard in that it uses a special diacritic called addak in this situation. To indicate a geminated consonant, the addak sits above the preceding syllable, eg. ਪੁੱਤਰ putːər.

The letter RA is an example of a letter that typically behaves quite idiosyncratically in consonant clusters, and typically quite differently depending on whether it appears at the beginning or end of the cluster. Its placement also often involves apparent reordering. This can be illustrated with the Devanagari RA . When in initial position in the cluster the letter is typically displayed as a small mark above the right shoulder of the last letter in the syllable, eg. शार्मा ʃaːrmaː. This is called a repha. A in final position in a conjunct cluster is displayed as a small diagonal mark, but precisely where it appears depends on the shape of the previous consonant, eg. प्र pra, त्र tra, ह्र hra. With TTA and DDA it needs a little supporting line, eg ट्र ʈra, ड्र ɖra.

Note that a cluster is not limited to two consonants, eg.

र् r + ग् g + dʰʌ -> र्ग्ध rgdʰʌ

This use of the repha, appearing as it does to the right of the whole cluster, demonstrates the syllabic nature of the Indic scripts. The following extension of the example shows even more clearly that the repha is actually positioned to the right of the syllable, rather than just the cluster, since it appears above the vowel sign.

र् r + ग् g + dʰʌ + -> र्ग्धी rgdʰiː

(Note that syllable boundaries in spoken text do not equate to those in written text. For example, 'Hindi' is spoken as 'hin-di', but written as 'hi-ndi'.)

Vowel signs used with consonant clusters

Where the vowel following a consonant cluster is rendered with a vowel sign, the placement of the vowel sign may need attention. As with the examples of the repha at the end of the last section, the syllabic nature of the script becomes apparent with the use of reordrant vowel signs attached to consonant clusters.

In Devanagari, where a vowel sign that is is normally rendered to the left of a character is pronounced immediately after the cluster, it will be rendered to the left of the whole cluster, eg. in मुश्किल muʃkil, the ि is pronounced after the .

Vowel signs in a script like Kannada are visually attached to the first consonant in the cluster. Note how the vowel sign appears over the k in ಕ್ರಿ kri, since the r is rendered as a reduced appendage at the bottom right of the first consonant in the cluster.

Nasalisation and alternative nasal letter representations

Three diacritics are associated with the nasalisation of vowels or the alternative representation of nasal consonants as part of a consonant cluster. Which diacritic is used for which purpose varies from script to script. The following are a few examples of usage. (As usual, although these diacritics have their own names in the various languages represented by the scripts, I will refer to them using the generic names used in the Unicode Standard):

The visarga

The visarga is commonly required for transcribing Sanskrit, but occasionally has more specific uses too. It is not used at all in Gurmukhi.

The pronunciation of the visarga may vary. In Kannada it is commonly pronounced ha, eg. ಪುನಃ punəha. In Gujarati it is typically silent.

In Tamil the visarga is known as 'aytham' and is used to create additional fricative sounds. Before PA it creates f, and before JA it creates z, eg. ஃபீசு fiːsɯ, ஃஜிராக்ஸ் ziɾoks.

Glyphs for the visarga should not have a dotted circle before them. This is feature of some older fonts, inherited from the fact that incorrect semantics were applied to this Unicode character prior to Unicode version 3.2.

Numbers

All scripts have their own number shapes. While some scripts, such as Tamil, tend to favour European numerals over their own in modern text, other scripts, such as Hindi, still make heavy use of their native shapes.

The following table shows the number symbols:

European 0 1 2 3 4 5 6 7 8 9
Devanagari
Bengali
Gurmukhi
Gujarati
Oriya
Tamil
Telugu
Kannada
Malayalam

Tamil number shapes are not based on the decimal system, and so there is traditionally no zero. There are however additional symbols to represent 10 , 100 , and 1000 . Modern Tamil typically uses European numerals.

Text units & punctuation

Sub-sentence units (words) are separated with spaces.

Modern text commonly uses western punctuation, but some scripts sometimes use traditional punctuation. For example, the DANDA may still be used in Devanagari to mark the end of a sentence, or the DOUBLE DANDA in Telugu for certain abbreviations.

Unicode implementation notes

An important distinction

In this section we need to make a clear distinction between characters, the basic codepoint units provided by the Unicode Standard for representation of the script in memory, and glyphs, the visual representation of one or more underlying characters when displayed or printed. As with other complex scripts, and unlike text in English, it is common to find situations where there is not a one-to-one mapping of characters to glyphs.

I will try to use the terms character and glyph carefully in the following text to clarify whether we are talking about the representation of the text in memory or as rendered on-screen or in print.

Character sets

The characters in an Indic Unicode script block are a superset of the ISCII (Indian Standard Code for Information Interchange) character sets. The ISCII standard includes separate encodings for each of the scripts discussed here, using escape sequences to shift between them. The Unicode blocks were originally based on the 1988 version of ISCII encodings. ISCII published a new version of the standard in 1991 with a few changes to order and repertoire of characters. Unicode, nevertheless, remains a superset of all the ISCII codes, with the exception of a few Vedic extension characters.

When first encoded in Unicode, the first 85 characters in each Unicode block were placed in the same order and position, on a script by script basis, as the 1988 ISCII characters for the respective script. Every script block orders analogous characters in the ISCII range in the same relative locations to the start of the block across all 9 scripts under consideration in this paper. The next 27 characters are additional Unicode characters, where each analogous character across the scripts is also assigned to the same code point relative to the beginning of the block. The final column is reserved for script specific characters. There is no special ordering here.

In the following charts I will use colour coding to indicate the different ranges as follows:

ISCII-derived coordinated Unicode extension additional script specific characters

The zones described above are illustrated here using the Devanagari script block.

Picture of devanagari block showing zoned areas.

The fact that ISCII and Unicode attempt to use the same code point relative to the start of the code block for analogous characters across all nine Indic scripts theoretically allows for easy transliteration between the various scripts, however in practice there are quite a few exceptions in transliteration, so specific tables have to be developed anyway.

Each script block has a different number and distribution of characters. The following table contrasts the allocation of characters for Devanagari, Bengali and Tamil.

Devanagari Bengali Tamil

Other notes:

  1. In older fonts the Tamil visarga, or aytham, was in some cases incorrectly rendered with a dotted circle before it in word-initial position (eg. in a word like ஃஜிராக்ஸ்). This is because the Unicode Standard initially classified the Tamil visarga as a combining character. An erratum issued in September 2001 corrected this, changing the General Category from "Mc" (Mark, combining) to "Lo" (Letter, other).

  2. The Kannada character [U+0CDE KANNADA LETTER FA] was incorrectly named. A more appropriate name would be LLLA, rather than FA. Because of the rules for Unicode naming, the current name cannot, however, be changed. Fortunately this letter has not been actively used in Kannada since the end of the 10th century.

Combining characters

It is important to correctly order Indic characters in memory where combining characters are involved. The Unicode Standard requires that all combining characters be stored in memory after the base character they are combined with. This is a fundamentally important concept. It means that, even if a combining glyph appears to the left of its base character, it is stored after the base character in memory. (This is often referred to as logical ordering.)

For example, the characters in the word हिंदी hindiː are stored in memory as:

h + ि i + n + d +

Lets see a slightly more complicated example from Kannada. The visual sequence ಕ್ರಿ is pronounced kri; the RA is rendered as a subscript to the bottom right and the vowel is rendered as a diacritic above the symbol for KA. The order of characters in memory (ignoring the virama, which is introduced in the next section) is:

k + r + ಿ i

It is common that multiple combining characters are asssociated with a single base character. Examples of such combinations from Devanagari include an anusvara with a vowel sign, eg. हिंदी hindiː, or a visarga with a vowel sign, eg. दुःख duhkʰ. Where there are multiple combining characters, the Unicode Standard provides rules about relative ordering in memory that should be observed. If there is a nukta it must immediately follow the base character. Next comes any virama or any vowel sign, then any bindus. Observing these rules improves operability and simplifies operations such as searching, sorting, character indexing, and the like.

The treatment of combining characters in Indic scripts also necessitates the use of context-based rules in the font to ensure the correct positioning and behaviour of displayed glyphs (a glyph being the visual representation of an underlying character). The position of a glyph for a combining character will commonly vary according to the shape and position of its base character, and any other combining characters associated with that base. In a number of cases, the combination of base character and combining character produces a fused shape that must be rendered by use of a ligature glyph or other special context-sensitive glyph forms (see also the next section for use of the virama).

Note also that reordering is not limited to displaying certain vowel signs to the left of the immediately preceding base consonant. In a consonant cluster a vowel sign that appears to the left may need to be displayed to the left of the whole consonant cluster, not just the preceding character. Similarly, the symbol for the consonant RA may be rendered as a diacritic at the far right of a syllable involving a consonant cluster that it logically begins.

In addition, because the base character is typed first during normal keyboarding, the base character will typically need to be 'moved' slightly to the right to accommodate the combining character glyph that joins to the left. In practice, the entire word is redrawn with every Indic letter. This is typically done in an off-screen buffer and blitted to the screen. The effect is that characters appear to move and change shape considerably while one is typing.

As mentioned earlier, a number of scripts (Bengali, Oriya, Tamil, Telugu, Kannada, Malayalam) have vowel signs that are composed, visually, of more than one part. Such multi-part vowel signs can normally be represented using a single character. For example, Tamil கௌ kʌʋ can be represented using the characters [U+0B95 TAMIL LETTER KA] and [U+0BCC TAMIL VOWEL SIGN AU]. Kannada ಕೋ koː can be composed of the characters [U+0C95 KANNADA LETTER KA] and [U+0CCB KANNADA VOWEL SIGN OO]. The Unicode charts typically do also provide separate characters that can be used to represent multi-part vowel signs. If these parts are not already available as simple vowel sign characters, they are provided as special 'length mark' characters such as [U+0BD7 TAMIL AU LENGTH MARK] and [U+0CD5 KANNADA LENGTH MARK]. Note also that if two characters are used to represent a split vowel sign both combining characters must follow the base character in memory (eg. + + for the Tamil example கௌ above).

Although Unicode typically provides single characters for letters formed by the addition of a nukta (eg. [U+0958 DEVANAGARI LETTER QA], [U+0A33 GURMUKHI LETTER LLA], and [U+0B5C ORIYA LETTER RRA]), these almost all have canonical decompositions to base character plus nukta diacritic (ie. क़ [U+0915 DEVANAGARI LETTER KA + U+093C DEVANAGARI SIGN NUKTA], ਲ਼ [U+0A32 GURMUKHI LETTER LA + U+0A3C GURMUKHI SIGN NUKTA], and ଡ଼ [U+0B21 ORIYA LETTER DDA + U+0B3C ORIYA SIGN NUKTA.

Since there are alternative ways of representing multi-part vowel signs and consonants created using the nukta, the question arises, "Which approach should be used when entering Indic text?" The W3C recommends the use of NFC (Unicode Normalization Form C) for web content. NFC represents all multi-part vowels as single characters, but all combinations of consonant plus nukta as two separate characters (apart from the following exceptions in Devanagari: [U+0931 DEVANAGARI LETTER RRA], [U+0934 DEVANAGARI LETTER LLLA], and [U+0929 DEVANAGARI LETTER NNNA]).

It was mentioned above that Gurmukhi is somewhat unusual in that vowel signs are carried by special 'vowel bearer' letters to create independent vowels. While Unicode does provide character codes for these vowel bearers, [U+0A72 GURMUKHI IRI] and [U+0A73 GURMUKHI URA] (plus, of course, [U+0A05 GURMUKHI LETTER A]), their use isn't recommended. Instead Unicode provides precomposed characters for all the independent vowel sounds needed, eg. [U+0A06 GURMUKHI LETTER AA], [U+0A07 GURMUKHI LETTER I], [U+0A09 GURMUKHI LETTER U], etc.

Note however that, for the general case, whereas some other encoding systems for Indic represent an II SIGN, for example, by VIRAMA + VOWEL II, Unicode does not do that. It considers these two sequences to not be equivalent, and not have the same rendering.

Variant glyph forms

Unicode follows the rule of 'encode characters not glyphs'. This is a fundamentally important concept relating to the support of Indic scripts in Unicode. Even though there are many potential shapes for a character when displayed (half-form, conjunct, ligature, diacritic, etc.), the rule means that there is only one codepoint to represent that character. This is a major advantage for conducting operations on the text such as string comparison, collation, etc. It also allows for a much simpler keyboard, and simpler correspondence between the keyboard input and the stored text. The task of producing the right shape for printing or display of a character according to its context falls to the rendering algorithms of the font, application or system.

We have already mentioned that combining character glyphs may sometimes adopt different shapes or merge with and alter the shape of the base consonant. Another key area where intelligent glyph shaping is required is the display of consonant clusters.

Consonant clusters are invariably indicated in a sequence of Unicode characters by the presence of a VIRAMA character, whether or not the glyph for the virama will be visible on display. Thus the virama is the trigger for any complex glyph shaping that may be applied to a conjunct by the font or rendering algorithms.

The outcome of a consonant+virama+consonant sequence will vary according to the characters, scripts, and fonts involved. Some possibilities are:

In Unicode it should be possible to force a consonant + virama sequence to display the virama (rather than convert the consonant to a half-form or ligature) by adding the character U+200C ZERO WIDTH NON-JOINER immediately after the virama of the dead consonant. For example, this produces क्‌क rather than क्क.

To force a dead consonant to assume a half-form rather than combine as part of a ligature, place the character U+200D ZERO WIDTH JOINER immediately after the virama. For example, this produces क्‍ष rather than क्ष. The zero width joiner can also be used to produce an example of a half-form on its own for illustration purposes, eg. क्‍. You can also create half-forms of combining ligatures, eg. त्त्‍.

Other practical considerations

Where scripts use glyphs that hang from the baseline, rather than sitting on the baseline, it is important to ensure that any glyphs from another intermixed script (eg. Latin script letters) are correctly aligned with the Indic script. It is also important to ensure that the glyphs are aligned as expected with other elements, such as table cells, graphic elements, and the like. For a detailed treatment of the issues for alignment of such scripts with other fonts see Steve Zilles' talk, 'Internationalized Text Formatting in CSS and XML' in the proceedings of IUC22.

There are other practical considerations related to enabling Indic script input and display. Keyboards must, of course, provide access to all needed characters, but consider standardisation of layout. On-screen display must support adequate resolution and line height, as well as proportional spacing.

For information about collation of Indic scripts, see Unicode Technical Note #1.

Glossary

abugida
A word used to describe scripts where consonant letters represent syllables with an inherent vowel. See Consonants and inherent vowels.
addak
A diacritic used in Gurmukhi to lengthen the following consonant sound. See Consonant clusters.
anusvara
A diacritic used to represent nasalisation of vowels and/or to represent nasalised consonants. See Nasalisation and alternative nasal letter representations.
articulatory
Related to the production of speech sounds.
aspirated, aspiration
Aspirated consonants are those produced with an audible expulsion of breath. Note that a non-aspirated consonant, such as a b, is produced with much less aspiration than a similar sound in English.
bindu
A diacritic used to represent nasalisation of vowels and/or to represent nasalised consonants. See Nasalisation and alternative nasal letter representations.
combining character
A character that graphically combines with a preceding base character. Combining characters are not usually used on their own, They include combining accents, diacritics, vowel signs, etc.
conjunct forms
A special graphical representation used to display a combination of consonants without intervening vowels. See Consonant clusters.
diphthong
A pair of vowels considered to be a single phoneme where the tongue moves from one to the other in such as way as to cause continual change in vowel quality.
glyph
The visual representation of one or more underlying characters. A font is made up of a set of glyph images.
half-form
A reduced version of a consonant glyph (typically missing the vertical stem) used to represent a consonant without a following vowel. See Consonant clusters.
homorganic
A consonant articulated at the same point in the vocal tract as a consonant in another class. For example, ŋ is the homorganic nasal of k.
independent vowel
A vowel used at the beginning of a word or within a word immediately after another vowel sound. See Independent vowels.
inherent vowel
In Indic scripts a consonant character represents a syllable that includes the consonant followed by a default (inherent) vowel sound. This vowel sound varies by language and script. See Consonants and inherent vowels.
ligature
A glyph representing a combination of two or more characters.
logical order
The order in which text is usually typed on a keyboard. For the most part, logical order corresponds to phonetic order.
nukta
A diacritic used in several indic scripts to extend the range of sounds covered by the alphabet. See Consonants and inherent vowels.
plosive
A sound produced by the mouth in such as way as to temporarily block the passage of the air, eg. a p.
phoneme
A minimally distinct sound in the context of a particular spoken language. For example, in UK English p and b are distinct phonemes because 'pat' and 'bat' are distinct.
repha
A glyph representing the character RA as the initial consonant in a cluster. The repha appears to the right of the consonant cluster. See Consonant clusters.
retroflex
Retroflex sounds are those made with the tongue being curled upwards.
tippi
A diacritic used in Gurmukhi to represent nasalisation of vowels and/or to represent nasalised consonants in the following syllable. See Nasalisation and alternative nasal letter representations.
virama
A combining mark used to indicate a consonant without a following vowel. See Consonants and inherent vowels.
visarga
A character used most commonly to transcribe Sanskrit, but also sometimes having additional uses such as in Tamil where it is used in conjunction with other characters to create sounds not in the basic repertoire. See Visarga.
vowel sign
A combining character used to indicate the replacement of the inherent vowel associated with a consonant with another vowel sound. See Vowel signs.

References

Sources

  1. The Unicode Consortium, The Unicode Standard -- Version 3.0, ISBN 0-201-61633-5. (See http://www.unicode.org/unicode/standard/versions/Unicode3.0.html.)
  2. P Daniels, W Bright, The World's Writing Systems, ISBN 0-19-507993-0
  3. R Gillam, Unicode Demystified, ISBN 0-201-70052-2
  4. R Snell, S Weightman, Teach Yourself Hindi, ISBN 0-340-42464-8
  5. Kalra, Purewall, Teach Yourself Panjabi, ISBN 0-340-70129-3

Other references

  1. C Wissink, Unicode Technical Note #1, http://www.unicode.org/notes/tn1/
  2. K Zia, Mapping of National Urdu Standard to Unicode, Proceedings of 22nd International Unicode Conference, 2002
  3. M Dürst, F Yergeau, M Wolf, R Ishida, T Texin, Character Model for the World Wide Web 1.0, (See http://www.w3.org/TR/charmod/.)

Acknowledgements

Many thanks to Cathy Wissink, Mark Davis and Joe Becker for reviewing the initial version of this paper at very short notice and still making numerous useful comments and suggestions.

First published 15 Aug 2003. This version 2022-10-20 19:31 GMT.  •  Copyright r12a@w3.org. Licence CC-By.