Balinese

Vowel signs

ᬓᬶ ki U+1B13 BALINESE LETTER KA + U+1B36 BALINESE VOWEL SIGN ULU

A vowel sign is attached to a consonant base to express a following vowel sound. Sometimes vowel signs have multiple parts, which are displayed on different sides of the base consonant or cluster. They are known as 'matras' in Sanskrit.

Balinese uses the following dedicated combining marks for vowels. They are all vowel signs.

11

ᬶiii1B36

ᬷiːiī1B37

ᬸuuu1B38

ᬹuːuū1B39

ᬾe ɛé ee1B3E

ᭀo ɔoo1B40

ᭂəě eə1B42

ᭃəː ə̄1B43

ᬵɑːaɑ̄1B35

   

ᬿaːiaiaʲ1B3F

ᭁaːu aʷ1B41

To represent the sounds rə or lə, Balinese uses vocalic letters. A sequence such as *ᬭᭂ U+1B2D LETTER RA + U+1B42 VOWEL SIGN PEPET is not used. See Vocalics.

Six of the vowel signs are spacing marks, meaning that they consume horizontal space when added to a base consonant.

All vowel signs are typed and stored after the base consonant, and the glyph rendering system takes care of the positioning at display time. The glyphs used to represent vowels, whether alone or in multipart vowels, are arranged around an orthographic syllable, which may be 2 consonants, rather than just around the immediately preceding consonant. See Pre-base vowel signs and Circumgraphs.

Composite vowel signs

Composite vowel signs are only produced when text is decomposed; 5 of the circumgraphs split off the ᬵU+1B35 VOWEL SIGN TEDUNG glyph, to create the following pairs:

5

ᭀavoido ɔ o1B3E
1B35

ᭃavoidəː ə̄1B42
1B35

ᭁavoidaːu aʷ1B3F
1B35

ᬻ rəː r̥̄1B3A
1B35

ᬽ ləː l̥̄1B3C
1B35

Pre-base vowel signs

ᬓᬾ ke U+1B13 BALINESE LETTER KA + U+1B3E BALINESE VOWEL SIGN TALING

Two vowel signs appear to the left of the base consonant letter or cluster.

both

ᬾe ɛé ee1B3E

ᬿaːiaiaʲ1B3F

These are combining marks that are always typed and stored after the syllable-initial consonant. The rendering process places the glyph before the consonant for display or printing.
Click on the following word to see the sequence of characters in storage.

ᬘᬾᬮᬾᬂ ˈcɛlɛŋ pig

These vowel characters are actually placed before the start of the orthographic syllable. This means that a word with a consonant cluster at the start separates the pre-base vowel from any post-base vowels by more than one consonant character (see Figure 1).

Circumgraphs

When a single vowel sign code point produces glyphs on more than one side of the consonant base, it is referred to here as a circumgraph.

ᬓᭀ ko U+1B13 BALINESE LETTER KA + U+1B40 BALINESE VOWEL SIGN TALING TEDUNG

Five vowel or vocalic sounds are represented by a vowel sign that is a single code point in memory, but when displayed it has visually separate parts that appear on different sides of the preceding consonant or cluster.

6

ᭀo ɔoo1B40

ᭃəː ə̄1B43

ᭁaːu aʷ1B41

ᬻrəː r̥̄1B3B

ᬼlə l̥1B3C

ᬽləː l̥̄1B3D

This section includes some vowel signs described in the section Vocalics.

Like pre-base glyphs, these are combining marks that are always stored after the base consonant. The rendering process places the glyphs around the base consonant, as needed.
Click on 'Show composition' to see the sequence of characters in storage for the following word.

Glyphs can appear on up to 3 sides of the base. Some of the glyphs merge with the base character's glyph (see Context-based shaping & positioning).

These circumgraphs have canonically equivalent decomposed forms (see Encoding vowel signs).

Vowel sign placement

Vowel length

The sources are not very clear about whether Balinese vowels vary in length during pronunciation (see Vowel sounds). The Balinese vowel sign repertoire does, however, contain glyphs that distinguish between short and long vowels (see Vowel summary table).

Nasalisation

If Balinese nasalises any vowel sounds, it is not explicitly marked in the orthography.

Independent vowels

How does the orthography handle vowels that are not preceded by a consonant?

10

ᬇiiị1B07

ᬈiː ị̄1B08

ᬉuuụ1B09

ᬊuː ụ̄1B0A

ᬏe ɛéẹ1B0F

ᬑo ɔoọ1B11

ᬅaaạ1B05

ᬆɑːaɑ̣̄1B06

   

ᬐaːiaiạʲ1B10

ᬒaːuoạʷ1B12

At the beginning of a word, most standalone vowels are represented using one of the 10 independent vowel characters. The set includes a character to represent the inherent vowel sound.

ᬉᬱᬥ usadə traditional medecine

ᬆᬤᬶ adi first

The vowel signs for ə (ᭂU+1B42 VOWEL SIGN PEPET) and əː (ᭃU+1B43 VOWEL SIGN PEPET TEDUNG) don't have an independent form, and have to be used after ᬳU+1B33 LETTER HA at the beginning of a word, ie. ᬳᭂU+1B33 LETTER HA + U+1B42 VOWEL SIGN PEPET and ᬳᭃU+1B33 LETTER HA + U+1B43 VOWEL SIGN PEPET TEDUNG, respectively, eg.

ᬳᭂᬫ᭄ᬧᬢ᭄ əm.pat four

In Sasak, independent vowel ᬅU+1B05 LETTER AKARA can be followed by an explicit ᭄U+1B44 ADEG ADEG in word- or syllable-final position, where it indicates the glottal stop. Other consonants can also be subjoined to it. eg. ᬳᬫᬅ᭄ hmạ͓ amaʔ

Vowel signs

11

ᬳᬶiihi1B33
1B36

ᬳᬷiː hī1B33
1B37

ᬳᬸuuhu1B33
1B38

ᬳᬹuː hū1B33
1B39

ᬳᬾe ɛehe1B33
1B3E

ᬳᭀo ɔ ho1B33
1B40

ᬳᭂə hə1B33
1B42

ᬳᭃəː hə̄1B33
1B43

ᬳᬵɑː hɑ̄1B33
1B35

   

ᬳᬿaːi haʲ1B33
1B3F

ᬳᭁaːu haʷ1B33
1B41

Typically, a standalone vowel is represented by a vowel sign attached to ᬳU+1B33 LETTER HA, which acts as a carrier, eg. ᬤᬳᬾᬭᬄ daerah development

Without a vowel sign the letter ᬳU+1B33 LETTER HA may represent a, eg. ᬳᬮᬲ᭄ alas forest

However, it may be unclear from the written text whether ᬳU+1B33 LETTER HA represents the sound h or is used as a carrier for a vowel, eg. compare ᬳᬶᬕ higa rib ᬳᬶᬕᭂᬮ᭄ igel dance

Plain vowels

Diphthongs and other combinations

Combining marks

However, there is also a set of combining marks for syllable-final consonants that don't need to be followed by the adeg adeg.

3

ᬄ-hhh̽1B04

ᬂ-ŋngŋ̽1B02

ᬃ-rrr̽1B03

ᬂU+1B02 SIGN CECEK and ᬄU+1B04 SIGN BISAH only appear at the end of a word, eg.

ᬓᭂᬋᬂ kěrěng eat a lot

ᬫᬗᬄ mangah logic

unless the word involves repetition, eg.

ᬘᬾᬂᬘᬾᬂ cengceng musical instrument

ᬃU+1B03 SIGN SURANG can appear at the end of any syllable.

ᬓᬃᬡ karna ear

A syllable-final diacritic may appear above a stack. It is typed and stored after the other components in the stack, eg. ᬩᬗ᭄ᬓᬸᬂ bangkung pig

When the syllable has a spacing vowel sign, any above-base final-consonant mark appears over the base character, rather than over the vowel sign. This is positioned by the font; the final consonant mark is still typed and stored after the other syllable components, eg. ᬕᭂᬤᭀᬂ ɡədoŋ building

See also Modre symbols.

Conjunct formation

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

Stacked and conjoined consonant clusters are referred to as conjuncts.

᭄1B44

In Unicode, the stacking and conjoining behaviour is achieved by adding ᭄U+1B44 ADEG ADEG between the consonants. The font hides the glyph automatically when a conjunct is formed.

In some cases, however, the adeg adeg remains visible (see Visible adeg adeg).

Stacking

To represent consonant sounds without intervening vowels, the non-initial consonant letter is typically drawn below the initial consonant letter, and with a slightly different shape. These subjoined forms are called gantungan (ᬕᬦ᭄ᬢᬸᬗᬦ᭄).

Many of the subjoined forms are just slightly smaller versions of the original, but several have very different shapes altogether, most of which ligate with the cluster initial consonant by joining strokes.

There can be up to 3 consonants combined in this way, but the third consonant must be one of ya, ra, la or wa.

The lists below show consonants in their normal and subjoined forms

16

ᬩ᭄ᬩb1B29
1B44
1B29

ᬢ᭄ᬢt1B22
1B44
1B22

ᬤ᭄ᬤd1B24
1B44
1B24

ᬘ᭄ᬘt͡ʃ1B18
1B44
1B18

ᬚ᭄ᬚd͡ʒ1B1A
1B44
1B1A

ᬓ᭄ᬓk1B13
1B44
1B13

ᬕ᭄ᬕg1B15
1B44
1B15

ᬳ᭄ᬳh1B33
1B44
1B33

ᬫ᭄ᬫm1B2B
1B44
1B2B

ᬦ᭄ᬦn1B26
1B44
1B26

ᬜ᭄ᬜɲ1B1C
1B44
1B1C

ᬗ᭄ᬗŋ1B17
1B44
1B17

ᬯ᭄ᬯw1B2F
1B44
1B2F

ᬭ᭄ᬭr1B2D
1B44
1B2D

ᬮ᭄ᬮl1B2E
1B44
1B2E

ᬬ᭄ᬬy1B2C
1B44
1B2C

6

ᬞ᭄ᬞt1B1E
1B44
1B1E

ᬟ᭄ᬟd1B1F
1B44
1B1F

ᬠ᭄ᬠd1B20
1B44
1B20

ᬔ᭄ᬔk1B14
1B44
1B14

ᬙ᭄ᬙ t͡ʃ1B19
1B44
1B19

ᬛ᭄ᬛd͡ʒ1B1B
1B44
1B1B

7

ᬝ᭄ᬝt1B1D
1B44
1B1D

ᬣ᭄ᬣt1B23
1B44
1B23

ᬥ᭄ᬥd1B25
1B44
1B25

ᬖ᭄ᬖg1B16
1B44
1B16

ᬰ᭄ᬰs1B30
1B44
1B30

ᬡ᭄ᬡn1B21
1B44
1B21

ᬪ᭄ᬪn1B2A
1B44
1B2A

Conjoined consonants

In conjoined clusters, the consonant glyphs remain side by side, but the non-initial consonant is reduced on the left side. These conjoined forms are called gempelan (ᬕᬾᬫ᭄ᬧᬾᬮᬦ᭄).

This list shows consonants in their normal and conjoined forms

3

ᬧ᭄ᬧp1B27
1B44
1B27

ᬲ᭄ᬲs1B32
1B44
1B32

ᬋ᭄ᬋrə1B0B
1B44
1B0B

both

ᬨ᭄ᬨp1B28
1B44
1B28

ᬱ᭄ᬱs1B31
1B44
1B31

The conjoined ᬲU+1B32 LETTER SA is unusual in that it also adds a stroke below the initial consonant (see Figure 5). This helps distinguish it from the conjoined p.

Visible adeg adeg

Because there is no word separator, consonants at the end of one word and beginning of the following word are normally stacked, too.

In some cases this leads to ambiguity about whether this is one or two words. If you really want to make clear which is which, you can use an explicit adeg-adeg, eg. compare ᬧᬓ᭄ᬭᬫᬦ᭄ pakraman membership ᬧᬓ᭄‌ᬭᬫᬦ᭄ Pak Raman Mr Raman

The Unicode Standard recommends the use of U+200C ZERO WIDTH NON-JOINER (ZWNJ) after the adeg-adeg in order to prevent conjunct formation. However, not many people understand the function of ZWNJ or can access it easily from the keypad. It also doesn't introduce line-break opportunities. A better solution may be to use U+200B ZERO WIDTH SPACE (ZWSP). This character is needed anyway on most systems in order to allow line-breaking, and it appears to work equally well for this.

A somewhat ambiguous situation arises where conventions prevent certain combinations stacking. For example, the name of the village tamblung should not stack the mbl, but should look as follows.

ᬢᬫ᭄‌ᬩ᭄ᬮᬂ

The Unicode Standard advises to use a zero-width non-joiner after ma, to achieve this.

Observation: Note that this may also be achieved by intelligence in the font, as was actually the case when I generated this example (click on it to see). It's not clear to me what is the preferred approach: put ZWNJ in only when the font doesn't do what you want, or use it always. The latter may lead to more consistent content where different fonts are applied to the text (eg. after cut and paste). In theory, this shouldn't affect searching and sorting, although some applications may not ignore the ZWNJ as they should.

Dedicated final marks

Balinese represents some final consonants using dedicated marks (see Finals). Such final marks are followed by ordinary consonant shapes in consonant clusters. There is no visual indication of missing vowel sounds other than the use of the mark itself.

Grapheme clusters

Base Combining_mark* Joiner?

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

1. Nukta (see Rerekan)
2. Dependent vowels (see Vowel signs and Vocalics)
3. Final consonants (see Finals)
4. Virama (adeg adeg) (see Consonant clusters and novowel)

Any of the above may occur after a consonant base. Independent vowel bases usually only have final consonant marks.

The following examples show a variety of grapheme clusters:

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

	ᬢᬷᬃᬢ holy water
	ᬅᬃᬣ wealth
	ᬓᬺᬰ᭄ᬡ Krĕsna Krishna
	ᬤᬍᬫ᭄ daləm deep
	ᬤᬦ᭄ᬢ dantə tooth

Note how grapheme clusters break up the conjuncts. This is not usually desirable (see Larger typographic units just below).

Larger typographic units

(Consonant Rerekan? Adeg_adeg)* Grapheme_cluster

Balinese commonly stacks or conjoins glyphs, to form conjuncts. The conjuncts represent consonant clusters, which can arise (a) where one phonetic syllable ends in a consonant letter and the following syllable begins with a consonant, or (b) when most medial consonants are written, since Balinese uses conjunct forms for sequences such as Cr-, Cy-, Cw-, Cry-, etc. The cluster of consonants that make up the conjunct are all encoded with adeg adeg between them (see Consonant clusters).

Balinese is unusual in that these conjuncts occur across word boundaries, so the word-final consonant of the first word may be stacked above the word-initial consonant of the second. See Figure 9 for an example.

Grapheme clusters terminate after a sequence of marks containing an adeg adeg, 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+1B44 ADEG ADEG, and the final grapheme cluster begins with a consonant.

The following are examples.

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

	ᬤᬦ᭄ᬢ dantə tooth
	ᬢᬶᬫ᭄ᬧᬮ᭄ timpal friend
	ᬩ᭄ᬭ᭄ᬬᬕ᭄ bryag laughter
	ᬰᬵᬲ᭄ᬢ᭄ᬭ sastrə writing

Note that one of the characteristic features of the Indic category of Virama is that the adeg adeg is visible when not followed by a consonant, but invisible when a consonant does follow (creating a stack). This means that the adeg adeg sometimes participates in a simple grapheme cluster, but when followed by a consonant it becomes the 'glue' that creates an orthographic syllable.

On the infrequent occasions when an adeg adeg 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. A zero-width space can achieve that.

ᬧᬓ᭄​ᬭᬫᬦ᭄ pak.ra.man Mr Raman

Browser behaviour

Cursor movement. Move the cursor through the text.
Gecko steps through the whole text using grapheme clusters. It takes 2 or more steps (depending on the number of GCs) to get through the stacks, one grapheme cluster at a time. Blink and WebKit step through all words using the orthographic syllables described here (ie. they step over a stack and all associated combining characters in one jump).

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, except for WebKit, which deletes one grapheme cluster at a time.

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 appears to segment on orthographic syllable, per the description here, except for one case where the complex stack is split. WebKit and Blink appear to sometimes wrap inside stacks and other times not. It's not obvious why, but both segment in the same way.

Phrase & section boundaries

What characters are used to indicate the boundaries of phrases, sentences, and sections?

See type samples.

6

᭞1B5E

᭝1B5D

᭟1B5F

᭚1B5A

᭛1B5B

᭜1B5C

Balinese has its own punctuation marks.

᭝U+1B5D CARIK PAMUNGKAH is used as a colon, and ᭞U+1B5E CARIK SIKI and ᭟U+1B5F CARIK PAREREN are used as comma and full stop respectively. U+1B4E INVERTED CARIK SIKI and U+1B4F INVERTED CARIK PAREREN were introduced in Unicode v16 to express finer distinctions than the former, used in some manuscripts. 9

Both ᭚U+1B5A PANTI and ᭛U+1B5B PAMADA are used to begin a section in text. U+1B7F PANTI BAWAK was introduced in Unicode v16 to represent finer subdivisions in some manuscripts.

At the end of a section, ᭜U+1B5C WINDU is usually used between two other punctuation marks that vary according to the section opener. Typical sequences include carik siki ᭞᭜᭞, carik pareren ᭟᭜᭟ (sometimes called pasalinan), panti ᭚᭜᭚, and carik agung ᭛᭜᭛.9

End of text markers include ᭽U+1B7D PANTI LANTANG and ᭾U+1B7E PAMADA LANTANG, or a combination of those or their shorter counterparts with ᭜U+1B5C WINDU, such as ᭽᭜᭽U+1B7D PANTI LANTANG + U+1B5C WINDU + U+1B7D PANTI LANTANG or ᭚᭜᭽U+1B5A PANTI + U+1B5C WINDU + U+1B7D PANTI LANTANG.9

Line breaking & hyphenation

Are there special rules about the way text wraps when it hits the end of a line? Does line-breaking wrap whole 'words' at a time, or characters, or something else (such as syllables in Tibetan and Javanese)? What characters should not appear at the end or start of a line, and what should be done to prevent that? Is hyphenation used, or something else? What rules are used? What difficulties exist?

Because there are no spaces between words, and because the end of one word and the beginning of another often form conjuncts (see Figure 9), Balinese doesn't wrap at word boundaries. See Graphemes for a description of the typographic units that are used for line break opportunities.

Unfortunately, modern browsers are often unable to detect appropriate break points for Balinese, so in the sample text at the beginning of this page U+200B ZERO WIDTH SPACE is used at places where the line could be broken. Otherwise, the line would continue, unbroken off the right side of the page.

Pameneng

᭠1B60

In lontar texts where a word must be broken at the end of a line (always after a full syllable), the sign ᭠U+1B60 PAMENENG is inserted. This sign is not used as a word-joining hyphen; it is used only in linebreaking.

Observation: The images appear to show a gap before the pameneng.

In online use, an application would need to insert the pameneng, rather than the content author. As line-length is changed by stretching a window, or as content is added earlier in the same paragraph, the location of the word relative to the line edge will change. The insertion of pameneng is only appropriate at those instants when the appropriate sequence of characters appears at the line end.

For an application to use this correctly, it would need to know where the word boundaries are in the text, and then put this character at the end of the line only when a multisyllabic word is broken. This would require a dictionary to be applied to the text, since it would not be appropriate to insert the pameneng at the boundary of 2 words.

Observation: Aditya Bayu Perdana has found instances in lontar where ᬄU+1B04 SIGN BISAH is moved to the beginning of a line, alone, while a pameneng appears at the end of the previous line. If this is not just a scribal inconsistency (eg. it's not clear why you wouldn't put the bisah at the end of the line if there's space for a pameneng), it may indicate that this letter should not be a combining mark in Unicode; however, the usage needs to be verified first. See pictures.

Text alignment & justification

Does text in a paragraph needs to have flush lines down both sides? Does the script allow punctuation to hang outside the text box at the start or end of a line? Where adjustments are need to make a line flush, how is that done? Does the script shrink/stretch space between words and/or letters? Are word baselines stretched, as in Arabic? What about paragraph indents?

According to Sudewa, full justification is not a feature of Balinese text in traditional palm-leaf manuscripts, and only left, or occasionally centred or right alignment is relevant.

Baselines, line height, etc.

Does the script have special requirements for baseline alignment between mixed scripts and in general? Is line height special for this script? Are there other aspects that affect line spacing, or positioning of items vertically within a line?

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

Figure 11 shows glyphs from the Noto Serif fonts. The basic height of Balinese letters is the same as the Latin x-height, however extenders and combining marks, extend well beyond the Latin ascenders and descenders, creating a need for larger line heights.

General page layout & progression

How are the main text area and ancilliary areas positioned and defined? Are there any special requirements here, such as dimensions in characters for the Japanese kihon hanmen? The book cover for scripts that are read right-to-left scripts is on the right of the spine, rather than the left. When content can flow vertically and to the left or right, how to specify the location of objects, text, etc. relative to the flow? Do tables and grid layouts work as expected? How do columns work in vertical text? Can you mix block of vertical and horizontal text? Does text scroll in a different direction?

Traditionally, Balinese was written on thin, landscape palm-leaf manuscripts, called lontar.

The text was packed in without paragraph breaks.