Bangla (draft)

Bengali

Letters

Combining marks

Numbers

Punctuation

Symbols

Other

Consonant-based orthographic syllables

[C[n]h] [C[n]h] C[n] [h | v (n)] [f]

The core of a consonant-based syllable is a base consonant character, which may or may not additionally represent an inherent vowel if it stands alone.

There is no inherent vowel if it is followed by a vowel sign, eg. কী kī কি ki কো ko or hasant, eg. ক্ k͓ At the end of a word, there may or may not be an inherent vowel, even if there is no hasant.

Any base consonant may be a combination of consonant code point plus nukta.

The base consonant can be preceded by up to two consonant+hasant pairs (where the consonant may also be a combination of consonant+nukta), but only if those consonants form conjuncts (ie. the hasant is invisible), eg. ক্ক k͓k ম্প m͓p ক্ষ k͓ʃ̇ ন্ত্র n͓t͓r If the preceding consonants carry visible hasant symbols, those are treated as separate orthographic syllables.

Likewise, the variable use of the hasant in Bengali means that a phonetic cluster of consonants can constitute a larger series of orthographic syllables. For example, this word for cymbal has two phonetic syllables, but 3 orthographic since the rt combination is not combined: করতাল

A vowel sign may optionally be followed by a nasalisation diacritic.

Unless the base consonant is followed by a hasant, the syllable may be terminated by a final consonant repesented by khanda ta, anusvara, or visarga.

Vowel-based orthographic syllables

Vowel-based syllables begin with a standalone vowel, which is represented by a single independent vowel or vocalic.

An independent vowel may be followed by an anusvara, visarga or candrabindu (nasalisation), eg. উঃ, আঁ ụh̽, ɑm̽

Vowel sounds

Consonant sounds

	labial	dental	alveolar	post- alveolar	retroflex	palatal	velar	glottal
stops	p b	t d			ʈ ɖ		k ɡ
aspirated	pʰ pf bʰ	tʰ dʰ			ʈʰ ɖʰ		kʰ ɡʰ
affricates				t͡ʃ d͡ʒ
aspirated				t͡ʃʰ d͡ʒʰ
fricatives	f		s z	ʃ				ɦ h
nasals	m		n				ŋ
approximants	w		l			j
trills/flaps			r ɾ		ɽ
aspirated					ɽʰ

pf, pʰ and f are alternative pronunciations for the same phoneme, depending on where the speaker is from, and all are written using ফ [U+09AB BENGALI LETTER PHA],

True retroflex (murdhonno) consonants are not found in Bengali. They are apical postalveolar in Western Dialects. In other dialects, they are fronted to apical alveolar.wp

r occurs word-initially, whereas ɾ occurs medially and finally. Both sounds are written using র [U+09B0 BENGALI LETTER RA].wp

s and ʃ are often merged. z is found mainly in foreign words.wp

In the Bangla spoken in Dhaka, ɾ and ɽ are often indistinct phonemically,wp eg. the following two words can be homophonous: করা কড়া

j and w are pronunciations of য় [U+09AF BENGALI LETTER YA + U+09BC BENGALI SIGN NUKTA] when it appears between certain vowels.

Inherent vowel

The inherent vowel is typically transcribed as a, and pronounced as ɔ or o. (And sometimes halfway between these two, when influenced by surrounding sounds.) Bengalis are not always aware of these sound differences – thinking of this as one sound. So kɔ or ko are written by simply using the consonant letter.

ক kɔ or ko [U+0995 BENGALI LETTER KA]

There is also a vowel sign pronounced o. This can lead to inconsistent spellings, eg. bhalo, good, well, can be spelled either ভালো or ভাল. Verb forms tend to be particularly inconsistent, sometimes basing the rationale on what looks good in a particular context.

The rules for determining the sound of the inherent vowel are not simple. Partly it is a question of vowel harmony. The following two tendencies can help:

• In words with inherent vowels in two consecutive syllables, the sound will usually be ɔ..o, not o..ɔ, eg. গরম However, exceptions occur for prefixes, such as prɔ-, ɔ-, and sɔ-.r,8

• When pronounced at the end of a word after a conjunct consonant, the inherent vowel is always o,r,8 eg. যুদ্ধ

• The pronunciation tends to be o when followed by a one of i, j, u, w either immediately or in the next syllable, but ɔ otherwise.d,400

Vowel signs

Non-inherent vowel sounds that follow a consonant are represented using vowel signs, eg.

কী ki [U+0995 BENGALI LETTER KA + U+09C0 BENGALI VOWEL SIGN II]

Bengali vowel signs are nearly all combining characters. One consonant is also used in a special configuration described below. In principle a single character is used per base consonant, but 2 vowel signs decompose to more than one character (see circumgraphs). All vowel signs are typed and stored after the base consonant, whether or not they precede it when displayed. The glyph rendering system takes care of the positioning at display time.

An orthography that uses vowel signs is different from one that uses simple diacritics or letters for vowels, in that the vowel signs are generally rendered relative to an orthographic syllable, rather than just applied to the letter of the immediately preceding consonant (see prebase_vowels for an example).

Almost all of the vowel signs are spacing combining characters, meaning that they consume horizontal space when added to a base consonant.

See also vowelligatures and vocalics.

Combining marks used for vowels

Bangla uses the following dedicated combining marks for vowels.

Bengali has lost the distinction between short and long vowels in pronunciation, but retains the difference in spelling.

The variation in pronunciation for the vowel signs can often be explained by vowel harmony.

  ো   [U+09CB BENGALI VOWEL SIGN O] was originally pronounced ʊ, and that pronunciation sometimes persists alongside the o that came from Sanskrit, eg. নোংরা

Consonants used for vowels

When it occurs as the last member of a consonant cluster য [U+09AF BENGALI LETTER YA] has a special shape seen in orange in fig_yophala, and is called ʤɔ-pfɔlɑ (য-ফলা). One of its functions is to create the sound æ.

details

হ্যাঁ

There are exceptions to the previous rule, when the a-kar produces its normal value, eg. ব্যাখ্যা

সন্ধ্যা

The sound æ can also be produced when there is no vowel sign, eg. ব্যথা bætʰɑ pain

According to Radice, vowel harmony may phonetically produce e when the following syllable is i.

Unusually for Indian scripts, y̌ɔ-phɔla can also be used after independent vowels to create the standalone sound æ. The sequences অ্যা [U+0985 BENGALI LETTER A + U+09CD SIGN VIRAMA + U+09AF LETTER YA + U+09BE VOWEL SIGN AA] and এ্যা [U+098F BENGALI LETTER E + U+09CD SIGN VIRAMA + U+09AF LETTER YA + U+09BE VOWEL SIGN AA] are used mostly in transliterations of borrowed words, eg. অ্যাটর্নি এ্যাডভোকেট

See also composite_vowels, which explains how independent vowels are used for the off-glide of diphthongs.

Pre-base vowel signs

Three vowel signs appear to the left of the base consonant letter or cluster, eg. বাকি

These combining marks are always stored after the base consonant, ie. the codepoints follow the order in which the items are pronounced. The rendering process places the glyph before the base consonant without changing the code points.

The vowel sign is actually placed before the start of an orthographic syllable. In fig_prebase the sequence of glyphs for the orthographic syllable is rendered VCC, whereas the pronunciation is CCV. In conjuncts with 3 consonants, it will still be rendered before all the consonants.

details

ব্যক্তি

প্লিজ

However, if the cluster doesn't form a conjunct, or is split by a visible virama, the cluster becomes two orthographic syllables and the pre-base vowel sign appears after the last consonant with the virama. The sequence of displayed glyphs is now CVC. If the conjunct contains 3 consonants, the displayed order will be CCVC.

details

বালতি

প্‌লিজ

Circumgraphs

Two vowels are represented by circumgraphs, producing glyphs on opposite sides of the consonant onset.

details

অক্টোবর

These circumgraph vowel signs are normally a single code point, but in decomposed text the base consonant can be followed by two code points , eg. compare the following equivalent ways of writing ko:

Composite vowels & diphthongs

Composite vowels in Bengali may occur when ো [U+09CB BENGALI VOWEL SIGN O] and ৌ [U+09CC BENGALI VOWEL SIGN AU] are decomposed, however this is not common. See encoding for more details.

Although 2 of the vowel signs (and independent vowels) represent diphthongs (oi̯ and ou̯) with a single code point, most of the many diphthongs are represented by a sequence of vowels,wa,#Vowels eg. কেউ

Diphthongs typically represent the off-glide using one of the following:

• ই [U+0987 BENGALI LETTER I]
• উ [U+0989 BENGALI LETTER U]
• ও [U+0993 BENGALI LETTER O]
• য় [U+09AF BENGALI LETTER YA + U+09BC BENGALI SIGN NUKTA] (for e̯).

The following are examples of diphthongs. Hyphens indicate a consonant with inherent vowel. The detailed character notes contain examples of Bangla words containing the diphthongs shown.

Standalone vowels

Bengali represents syllable-initial vowels using a set of independent vowel letters, eg. ওস্তাদ উট ঔষুধ

Standalone æ appears mostly in loan words. It involves the use of য [U+09AF BENGALI LETTER YA]. See consonant_vowels for more details.

Vowel elongation

ঽ [U+09BD BENGALI SIGN AVAGRAHA] is a Sanskrit-derived symbol that is used in modern Bengali to lengthen vowel soundsws, eg. কিঽঽঽ? kiiii Whaaatt? শুনঽঽঽ ʃunooo Listennn

Nasalisation

ঁ [U+0981 BENGALI SIGN CANDRABINDU] nasalises the vowel in a syllable, eg. হ্যাঁ হাঁপান Nasalised vowels include ĩ ũ ẽ õ ɛ̃ ɔ̃ ã.

The candrabindu should be placed over the top of an independent vowel, but over the base consonant when a vowel sign is attached – not over the vowel sign. In the code point sequence, however, this should occur after any combining vowel sign associated with the same syllable. Note how the base consonant is identified correctly in the second word of fig_candrabindu, even though the candrabindu is 4 code points away. Some fonts do not position the candrabindu correctly.

Consonants with no following vowel

Bengali uses ্ [U+09CD BENGALI SIGN VIRAMA] (called হসন্ত hʃ̈n͓t hasant hɔsonto in Bengali) to indicate that the inherent vowel is not pronounced after a consonant, eg. the following explicitly represents just the sound k.ক্ k͓

The hasant is rarely seen. It is not used at the end of a word even though the inherent vowel is pronounced at the end of some words and not others, eg. গরম gɔrôm, hot vs. গড়ান gɔɽɑnô, to roll. There is no real way to tell when the inherent vowel is pronounced and when not in this position, except that it is usually pronounced following a word-final consonant cluster.

Within a word also, some clusters don't use the hasant in Bengali, and the reader simply has to know that the inherent vowel is not pronounced, eg. করতাল

This is particularly common at morpheme boundaries, for example in verb forms.

Consonant clusters that are represented by conjunct forms use the hasant between consonants to invoke the shape changes. If the font has the glyphs needed to produce the conjunct the hasant is hidden (see clusters).

Refs: Radice 3, 7-8, 21, 148; Daniels 400

Vowel sounds to characters

This section maps Bengali vowel sounds to common graphemes in the Bengali orthography, grouped by whether they are vowel signs ( vs ), or standalone ( s ). Click on a grapheme to find other mentions on this page (links appear at the bottom of the page). Click on the character name to see examples and for detailed descriptions of the character(s) shown.

Hyphens indicate a consonant with inherent vowel.

Basic consonants

Repertoire extension

় [U+09BC BENGALI SIGN NUKTA] is used to create 3 additional letters, eg. the dot changes ড ɖ to ড় ɽ. Here is a list of graphemes that combine nukta with an existing consonant.

To reveal detailed notes about usage see the list of precomposed characters a little lower.

য় [U+09AF BENGALI LETTER YA + U+09BC SIGN NUKTA] represents j, w or e̯, depending on what vowels occur alongside it. See the character notes for details.

The nukta should always be typed and stored immediately after the consonant it modifies, and before any combining vowels or diacritics.

The Unicode Standard recommends that content authors use decomposed sequences for these letters. However, the Unicode block also contains the precomposed code points shown below.

Decomposed sequences are not recomposed by Unicode Normalisation Form C (NFC).

Final consonants

One letter and 2 diacritics represent syllable-final consonant sounds.

In a sequence of characters, these should all occur after any combining vowel sign associated with the same syllable. None carry vowel signs.

Consonant clusters

Consonant clusters are written using:

1. No special rendering. This is a common occurrence in Bengali.
2. Conjuncts. There are a number of possibilities here.
   1. Fused vertically : Reduce the component shapes and combine them vertically, usually approximately within the normal character height.
   2. Conjoined : The two consonants sit side by side, but the first consonant has an altered shape.
   3. Ligated : A fusion of the component letter shapes where it may be difficult to tell them apart.
   4. Special forms : These apply to cluster-final YA (y̌ɔ-phɔla), the letter RA, and cluster-final MA.
3. A visible virama below the non-final consonants in the cluster. In Bengali this may be a conscious decision, and not just a gap in font support.
4. Final consonant letters or marks followed by another consonant. There is no interaction between the finals and the following character (see finals).

See also consonant_length.

Conjunct formation

See a table of 2-consonant clusters.
The table allows you to test results for various fonts.

্

To produce a conjunct, ্ [U+09CD BENGALI SIGN VIRAMA] is added between the consonants in the cluster. There are exceptions, but this type of virama is usually not displayed, eg. the sequence ক + ্ + ষ [U+0995 BENGALI LETTER KA + U+09CD SIGN VIRAMA + U+09B7 LETTER SSA] produces ক্ষ k͓ʃ̇

The font usually determines how a cluster is rendered, although it is possible to influence this (see formatting).

Different fonts may combine the same letters in different ways. The following figures shows characters that are combined in different ways by different fonts.

Quite often, clustered consonants are pronounced differently than you would expect. In particular, conjuncts ending with ব [U+09AC BENGALI LETTER BA] or ম [U+09AE BENGALI LETTER MA] tend to not pronounce the latter, but double the length of the consonant before it (see consonant_length).

Nasals in conjuncts tend to conform to phonological patterns. Velar consonants (k, kh, g, etc) combine with ঙ ŋɔ, palatal consonants (c, ch, ..) combine with ঞ ñɔ, retroflex ণ ɳɔ, dental ন nɔ, and labial ম mɔ.

The sections below show examples of the various types of conjunct forms. The lists are not exhaustive. The shapes shown are by default those contained in the Noto Sans Bengali webfont. Other fonts may combine components in different ways. Click on the characters if you want to see the components.

Vertical conjuncts

Conjunct shapes are most commonly formed by arranging the components vertically, reducing and combining the shapes of the individual components as needed.

sthô

llô

The conjuncts in fig_conjunct_vertical used in words: আস্থা ঝিল্লি

Examples of components arranged vertically.

ব্ল␣ক্ল␣ক্ব␣ঘ্ন␣ফ্ল␣ক্ক␣ক্ব␣গ্গ␣গ্ধ␣গ্ন␣গ্ল␣গ্ব␣হ্ল␣হ্ণ␣থ্ব␣ঞ্ছ␣ঞ্জ␣ঞ্ঝ␣ল্ক␣ল্গ␣ল্প␣ল্ব␣ল্ল␣ণ্ণ␣ণ্ম␣ণ্ব␣ণ্ড␣ন্ত␣ন্থ␣ন্ন␣ন্ব␣ন্ত্ব␣ম্ন␣দ্ভ␣দ্ব␣ত্ত␣ত্ত্ব␣ণ্ড␣ণ্ড্র␣ত্ন␣ত্ব␣ট্ট␣ক্ট␣প্প␣প্ন␣প্ত␣প্ল␣শ্ন␣শ্ল␣শ্ব␣ষ্ক␣স্ক␣স্খ␣স্ত␣স্থ␣স্ন␣স্ল␣স্ব␣স্ত্র␣ম্ন␣ম্ল␣ম্ভ␣ম্ব␣ঙ্ক

Conjoined conjuncts

Many conjuncts are formed by combining components horizontally.

mpô

ccô

The conjuncts in fig_conjoined used in words: ক্যম্পাস উচ্চারণ

Other examples of components arranged side-by-side, frequently with simplification of the initial consonant.

ব্ব␣ব্দ␣ধ্ব␣ল্ট␣ল্ফ␣ণ্ট␣ণ্ঢ␣ণ্ঠ␣ন্দ␣ন্স␣ল্ড␣দ্দ␣ড্ড␣ঙ্খ␣ঙ্ম␣জ্জ␣জ্জ্ব␣জ্ব␣জ্ঝ␣চ্চ␣চ্ছ␣চ্ঞ␣ণ্ঠ␣প্ট␣প্স␣শ্চ␣শ্ছ␣ষ্প␣ষ্ফ␣স্ট␣স্প␣স্ফ␣ম্প␣ম্ফ␣হ্ব

Ligating conjuncts

A small set of conjuncts combine the consonants into a ligated shape, where individual components can't always be easily discerned.

mpô

ccô

The conjuncts in fig_conjunct_ligated used in words: খ্রিষ্টান ক্ষণ

Examples of conjuncts arranged in a way that involves ligation, significantly altering one or more of the components.

ষ্ণ␣ব্ধ␣হ্ন␣ঞ্চ␣ত্থ␣ন্ধ␣ন্ধ্র␣ক্ষ্ম␣ক্স␣হ্ম␣ঙ্গ␣দ্ধ␣দ্ধ্ব␣ক্ত␣ঞ্ঝ␣ব্জ␣জ্ঞ␣ষ্ঠ␣ষ্ট␣ট্ট

Consonant lengthening

There are a number of ways of producing a lengthened consonant sound in Bangla.

A straightforward approach is to duplicate the consonant sound in conjunct form. For example, a long l can be written ল্ল [U+09B2 BENGALI LETTER LA + U+09CD SIGN VIRAMA + U+09B2 LETTER LA], eg. ঝিল্লি

Another common way of doubling the length of a consonant is to use a conjunct ending with ব [U+09AC BENGALI LETTER BA] or ম [U+09AE BENGALI LETTER MA], eg. ভস্ম বিশ্ব

The y̌ɔ-phɔla (্য [U+09CD BENGALI SIGN VIRAMA + U+09AF BENGALI LETTER YA]) can also lengthen the consonant it follows, eg. জন্য

ঃ [U+0983 BENGALI SIGN VISARGA] can also lengthen the following consonant, with no aspiration, eg. নিঃশব্দ

Consonant sounds to characters

This section maps Bengali consonant sounds to common graphemes in the Bengali orthography, grouped according to whether they are regular ( r ), conjuncts ( c ), or final ( f ). Click on a grapheme to find other mentions on this page (links appear at the bottom of the page). Click on the character name to see examples and for detailed descriptions of the character(s) shown.

Vowel signs

The 2 circumgraphs can be written as a single character, or as two characters (in decomposed text).

The single code point per vowel sign is the form preferred by the Unicode Standard and the form in common use for Bengali. The parts are separated, however, in Unicode Normalisation Form D (NFD), and recomposed in Unicode Normalisation Form C (NFC), so both approaches are canonically equivalent.

Whichever approach is used, the vowel signs must be typed and stored after the consonant characters they surround. In the case of decomposed vowel signs, the order is also important and must be as shown above.

Independent vowels

The single code point on the left should be used, and not the sequence on the right, because they are not made the same by normalisation, and they are not semantically equivalent. Using the right-hand sequence will cause searches and machine understanding of the data to fail.

Nuktas

The way the Unicode Standard recommends to type and store graphemes with nuktas is slightly inconsistent for Bengali. Here we look at alternative strategies for all uses of the nukta in the Bengali block (usage recommendations for Bangla are given in the section extendedC), and consider the effects of normalising the text using Unicode Normalisation Form D (NFD), and Normalisation Form C (NFC).

For the following alternatives the decomposed form is recommended by the Unicode Standard. NFC does not recombine the parts into precomposed characters. Instead, both NFC and NFD normalisation produces decomposed forms.

In the next case, the Unicode Standard recommends using the precomposed form. Neither form is converted into the other by normalisation, so they are not equivalent. It is therefore best to use the precomposed form only, so that text matches other text as expected.

In practise, it's hard to envisage content authors being aware of, let alone respecting, rules about whether they should use precomposed or decomposed forms. Keyboards or other input mechanisms, or perhaps sometimes applications that automatically normalise can perhaps guide users to the recommended practise, but it's likely that Bengali text will always contain a mixture of forms for these graphemes, and matching algorithms will need to be prepared to equate them all.

Codepoint sequences

Nuktas must immediately follow the base consonant they modify.

When 2 vowel signs are used for a circumgraph, the encoded order of the combining marks should match the displayed order, left to right.

Currency

৳ [U+09F3 BENGALI RUPEE SIGN] is the Bengali rupee sign.

Context-based shaping & positioning

Vowel ligatures

Sometimes vowel signs (particularly U) form ligatures with a preceding base consonant. fig_lig shows ligated (top) and non-ligated (bottom) forms for several combinations. In certain contexts it may be less appropriate to ligate (eg. newspapers and modern typefaces). Both forms are equivalent in every way but visually.

The default behaviour of a given font can be modified using the zero-width non-joiner character before the vowel to produce the simple form. See fig_ligated_simplified.

See a matrix of consonants followed by vowel signs for Bengali. A few clusters that are often ligated are pre-highlighted.

Font styling & weights

tbd

Grapheme clusters

Base ZW(N)J? Combining_mark* ZW(N)J?

Combining marks may include zero or more of the following types of character.

1. Nukta [1] (see extendedC) Only one per grapheme cluster, typed and stored immediately after the base consonant.
2. Dependent vowels [10] (see combiningvowels and vocalics) Usually a single code point, but in decomposed text can be two (see circumgraphs).
3. Nasalisation mark [1] (see nasalisation) Occurs over an independent vowel, or over a consonant when it is followed by a vowel sign. It is nevertheless typed and stored after any vowel signs.
4. Final consonant marks [2] (see finals) One of 2 possible combining marks, at the end of a grapheme cluster sequence. May also occur after independent vowels.
5. Virama (hasant) (see clusters and novowel) Normally occurs immediately after a consonant (and optional nukta) at the beginning of a cluster, but also occurs after independent vowels, particularly when writing the sound æ. It may also occur after RA+ZWJ to force a particular rendered shape for RA (see below).

In some cases, a ZWJ is inserted between RA + hasant followed by YA in order to specify special shaping rules (see special_forms).

A base consonant may be followed by ZWNJ before vowel sign code points where the author wants to prevent ligation of the following vowel sign (see vowelligatures). A ZWNJ may also be used after a virama to prevent conjunct formation and force the virama to be rendered visibly to the reader (see visiblevirama).

The following examples show a variety of grapheme clusters:

Click on the text version of these words to see more detail about the composition.

	বালতি
	ইংরেজ
	নিঃশব্দ
	ঝিল্লি
	ক্যম্পাস

Note how grapheme clusters segment the parts of a conjunct after the virama. This is not always desirable (see orthographicS just below).

Larger typographic units

(Consonant Nukta? Virama)* Grapheme_cluster

Bengali commonly stacks or conjoins glyphs, to form conjuncts (see clusters). The conjuncts represent consonant clusters or gemination.

Grapheme clusters terminate after a sequence of marks that includes a hasant, but editorial operations that change the visual appearance of the text, such as letter-spacing, first-letter highlighting, line-breaking, and justification, should never split conjunct forms apart. For this reason, an alternative way of segmenting graphemes is needed. This may not apply, however, for some other operations such as cursor movement or backwards delete.

Where conjuncts appear, a typographic unit contains multiple grapheme clusters. The non-final grapheme clusters all end with ্ [U+09CD BENGALI SIGN VIRAMA​], and the final grapheme cluster begins with a consonant.

The following are examples. Some examples were shown in the previous section: here the conjunct is treated as a single typographic unit.

Click on the text version of these words to see more detail about the composition.

	ঝিল্লি
	ক্যম্পাস
	ক্ষণ
	কুর্তা

Note that consonant clusters aren't always rendered as conjuncts, and in those cases the individual letters remain as separate graphemes. For example, consonants without an inherent vowel may be written using a consonant letter on its own at the end of a word or in other locations (see novowel).

Click on the text version of these words to see more detail about the composition.

	রিকশা
	খাননা

On the occasions when a virama needs to be visible even though it is followed by another base, an invisible character must be added to prevent it joining with the following base. ‌ U+200C ZERO WIDTH NON-JOINER can achieve that.

	ফ্‌ল্যাট
	লান্‌চ

Complicating factors

Behaviour is font-dependent. If a font doesn't have a conjunct form for a particular combination of characters it will make the virama visible.

What's important to note here is that it is normally possible to break a line after the grapheme cluster containing the virama when the virama is visible. This is currently difficult to manage because the decision as to whether the text is segmented into 2 graphemes or one depends only on the capabilities of the font used (ie. the rendered result); the code point sequence is identical for both cases, and gives no clues to which approach to segmentation is applicable.

Visible viramas can also affect vowel sign positioning. For the purposes of illustration, see fig_prebase_position, where the placement of the pre-base vowel varies. In the conjunct form on the left, the vowel sign is rendered to the left of the whole conjunct. If the sequence is not rendered as a conjunct, as in the second example, the pre-base glyph precedes the TA, not the SA.

Browser behaviour

Cursor movement. Move the cursor through the text.
Gecko steps through the text using grapheme clusters with one exception. It takes 2 or more steps (depending on the number of GCs) to get through the conjuncts, one grapheme cluster at a time. In some cases, especially with RA, the cursor may look as if it is in the wrong position. The exception is the sequences with yo-phala (eg. ল্যা), which are skipped over as if they were a single grapheme cluster. Blink and WebKit step through all words using the orthographic syllables as described here (ie. they step over a stack and all associated combining characters in one jump), except that WebKit doesn't include the ZWNJ in the grapheme cluster it ends, requiring an extra jump to get past it (although the cursor doesn't move).

Selection. Place the cursor next to a character and hold down shift while pressing an arrow key.
The behaviour is the same as for cursor movement.

Deletion. Forward deletion works in the same way as cursor movement. The backspace key deletes code point by code point, for all browsers.

Line-break. See this test. The CSS sets the value of the line-break property to anywhere. Change the size of the box to slowly move the line break point.
Gecko sets break opportunities in the same places as cursor stops. WebKit and Blink both wrap on orthographic syllable boundaries.

Word boundaries

Words are separated by spaces.

Phrase & section boundaries

Bangla uses a mixture of ASCII and Bengali punctuation.

The danda, । [U+0964 DEVANAGARI DANDA], is used for sentence final punctuation.

Observation: I haven't seen much evidence for the use of the double danda, ॥ [U+0965 DEVANAGARI DOUBLE DANDA].

Western punctuation, such as commas, semicolons, colons, quotation marks and hyphens are also used quite commonly.

Bracketed text

Bengali commonly uses ASCII parentheses to insert parenthetical information into text.

Quotations & citations

Bengali texts typically use quotation marks. Of course, due to keyboard design, quotations may also be surrounded by ASCII double and single quote marks.

Emphasis

Italicisation, bolding, and underlining are not traditionally features of Bengali text.

Abbreviation, ellipsis & repetition

The bisɔrgô ঃ [U+0983 BENGALI SIGN VISARGA​] is sometimes used to mark initial abbreviations.

A sign called urdha-comma can be used to indicate truncation of words, eg. কʼরেʼপরে The Unicode Standard recommends use of ʼ [U+02BC MODIFIER LETTER APOSTROPHE].u,460

Observation: Wikipedia seems to use a normal apostrophe.

The Unicode Bengali block also has the punctuation ৽ [U+09FD BENGALI ABBREVIATION SIGN]. It is possible that … [U+2026 HORIZONTAL ELLIPSIS] is used.

Observation: Information is needed about how it is used. It doesn't appear to be in common use.

Inline notes & annotations

tbd

Other punctuation

৺ [U+09FA BENGALI ISSHAR] is used alongside the names of deceased persons.

Other inline text decoration

tbd

Line breaking

Bengali is preferably wrapped at word boundaries.

Text alignment & justification

tbd

Text spacing

tbd

Baselines, line height, etc.

tbd

Bangla uses the so-called 'alphabetic' baseline, which is the same as for Latin and many other scripts.

It also has a 'hanging baseline', which may be used for text alignment in things such as initial letter highlighting. The hanging baseline is based on the top bar that joins the letters.

Bangla requires slightly more vertical space than Latin text. To give an approximate idea, fig_baselines compares Latin and Bangla glyphs from Noto fonts. The basic height of Bangla letters is typically slightly higher than the Latin x-height, including conjunct stacks, however certain letters and combining marks extend beyond the Latin ascenders, creating a need for larger line spacing.

fig_baselines_other shows similar comparisons for the Bangla MN and Vrinda fonts.

Counters, lists, etc.

You can experiment with counter styles using the Counter styles converter. Patterns for using these styles in CSS can be found in Ready-made Counter Styles, and we use the names of those patterns here to refer to the various styles.

The modern Bangla orthography uses ASCII digit numbering, but also has a native numeric style.

Page & book layout

This section is for any features that are specific to Bengali and that relate to the following topics: general page layout & progression; grids & tables; notes, footnotes, etc; forms & user interaction; page numbering, running headers, etc.

References