Modern Standard Arabic

Letters

Combining marks

Numbers

Punctuation

Symbols

Other

Vowel sounds

The diphthong aj is colloquially pronounced more like ej, however in this page we will continue to transcribe it phonetically per the official pronunciation.

Consonant sounds

	labial	dental	alveolar	post- alveolar	palatal	velar	uvular	pharyngeal	glottal
stops	p b	t d				k ɡ	q		ʔ
ejective		tˤ dˤ
affricates				d͡ʒ
fricatives	f v	θ ð	s z	ʃ ʒ		x ɣ	χ ʁ	ħ ʕ	h
ejective		ðˤ	sˤ zˤ
nasals	m		n
approximants	w		l ɫ		j
trills/flaps			r

Modern Standard Arabic covers many territories, most of which have their own dialects or languages, and these tend to influence the local pronunciation of Standard Arabic. In the chart above, we remain conservative, only mentioning variants that tend to apply to the standard pronunciation. For a slightly more detailed set of notes, see Wikipedia.

p and v are sometimes pronounced by some speakers for foreign words, such as باكستان فيروس

Sometimes alternative letters are used for such words (see foreign).wp

Although most dialects include it as a phoneme, ɡ is only used in Modern Standard Arabic as a marginal phoneme to pronounce some dialectal and loan words.wp

The sound ɫ occurs as a phoneme in a handful of loanwords, though not in all pronunciations. It also occurs in the name اللّٰه

The sound, d͡ʒ is used in Algerian, Hejazi, Najdi, Iraqi, and Gulf regions, whereas ʒ is used in Moroccan, Tunisian, Egyptian, Levantine, and Israeli regions. In both cases, the sound is written using ج [U+062C ARABIC LETTER JEEM].wp,#Local_variations_of_Modern_Standard_Arabic

ظ [U+0638 ARABIC LETTER ZAH] is pronounced zˤ in some regions, rather than ðˤ. 

Tone

Arabic is not a tonal language.

Structure

The following notes on structure are taken from Wikipedia.wp,#Phonotactics

[C1] [S1] V [S2] [C2 [C3]]

Arabic syllable structure consists of an optional syllable onset, consisting of one or two consonants; an obligatory syllable nucleus, consisting of a vowel optionally preceded by and/or followed by a semivowel; and an optional syllable coda, consisting of one or two consonants.

The following restrictions apply:

Onset

C1 can be any consonant, including a liquid (l r). The onset is composed only of one consonant; consonant clusters are only found in loanwords. Sometimes an epenthetic a is inserted between consonants.

Nucleus

Includes S1 V S2.

Coda

C2 and C3 can be any consonant.

More vowel details

The following table summarises the main vowel to character assignments, for vocalised text. Vowel representation in non-vocalised text can be derived by ignoring the combining marks.

In normal (non-vocalised) text a vowel location is simply not indicated where only a diacritic is shown in the table. Because the sukun is also dropped in non-vocalised text, where a mater lectionis remains it only implies a vowel location, since it may alternatively represent a consonant or a glide.

In word-initial position vowels are attached to ا. However, more often than not, a hamza is also attached to the alef to indicate the glottal stop. Although this actually constitutes a consonant plus vowel, in unvocalised text the alef (plus any hamza) signals the location of a vowel. The table therefore shows these maximal combinations.

More information can be found by clicking on the characters in the list box below, or in the section vowel_mappings.

The sounds eː, o, and oː are only used for transliterations of foreign words, and are spelled identically to iː u, and uː, respectively.

The letter ى, used as an alternative for a final aː, is a dedicated vowel character (see vletter).

This is the full set of characters needed to represent the Arabic language vowels, including the alef+hamza combinations. The last 3 combining marks are only found in decomposed text.

Vowel letter, alef maksura

0649 represents aː at the end of many words when it is written with YEH instead of an ALEF. In this case, YEH has no dots below, and this code point produces the requisite shape.

حتى

It also produces aː in informal speech for some words that end with nunation (though formally the ending is pronounced -an).

معنى

If any suffix is added, the spelling reverts to the normal alef, eg.

معناهم mæʕnaː-hum

Combining marks used for vowels

In situations where it is necessary to unambiguously indicate the underlying vowel sounds, short vowels can be expressed using diacritics called harakat, eg. العَرَبِيَّة

However for languages such as Arabic, Persian and Urdu they are typically not used unless there is a particular need to help the reader understand the pronunciation. The previous example would therefore usually be written العربية

On the other hand, when the script is used for some other languages (such as Uighur, Kashmiri, or Hausa), all vowels are shown, as a matter of course. These diacritics are also used in the Qur'an (though not originally), to reduce ambiguity.

Consonants used for vowels

Diphthongs & glides

The 2 diphthongs aj and aw are written using a combination of short a with the semivowels ي [U+064A ARABIC LETTER YEH] and و [U+0648 ARABIC LETTER WAW],wp,#Vowels. In vocalised text this usage can be detected by the presence of sukun, but in non-vocalised text it is not so obvious.

عين

عود

Composite vowels

The 5 composite vowels listed here only appear in vocalised text. Three represent long vowels, where the vowel diacritic is followed by a letter. Two more represent standalone vowels, with alef used as a carrier for the vowel diacritics. Diphthongs and glides are not included here.

Click on the letters for examples.

Standalone vowels

In principle, Arabic has very few true standalone vowels, since vowels are nearly always preceded by a glottal stop or other consonant. However, when ا appears without any hamza diacritic at the beginning of a sentence, it may be pronounced a, for the definite article, or i in various other circumstances.

When preceded by another word these sounds are elided, although the spelling remains unchanged.

See initialletters for more details.

Vowel length

Long vowels are generally distinguished from short vowels by the use of matres lectionis (see matres).

Nasalisation

Nasalisation is not a phonemic feature of Arabic.

Tones

Arabic is not a tonal language.

Vowel sounds to characters

This section maps Modern Standard Arabic vowel sounds to common graphemes in the Arabic orthography. 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.

Sounds listed as 'infrequent' are allophones, or sounds used for foreign words, etc.

Graphemes in normal spelling are shown to the left, and vocalised spelling is shown on the right.

'Initial' vowels also show what the vowel looks like when preceded by a glottal stop, which is common at the beginning of a word.

Basic consonant letters

The main Unicode Arabic block contains 153 letters, with 77 more in the extended blocks. As shown in the previous section, only a small subset of those are used to write a given language. The others represent special characters added to the repertoire for one or other of the many languages for which the Arabic script is used.

The vast majority of letters represent consonants. A few represent long vowels.

The following letters are those generally recognised as constituting the alphabet for the Standard Arabic language.

Of those, as mentioned earlier, some letters represent long vowel locations or combinations of consonant plus vowel.

Supplementary letters

Other Unicode letters regularly used in Arabic include:

Most of the above letters with diacritics decompose in Unicode Normalization Form D (NFD), however ة [U+0629 ARABIC LETTER TEH MARBUTA] does not.

Alef

Formally speaking, ا [U+0627 ARABIC LETTER ALEF] has no sound of its own. It is really a vowel lengthener and carrier. Its main uses in arabic orthography are:

• as a carrier letter for a word-initial vowel, eg. الآن انتباه

• as a lengthening sign for the a-vowel, eg. إنتاج

• as a carrier letter for the hamza (see أ [U+0623 ARABIC LETTER ALEF WITH HAMZA ABOVE], إ [U+0625 ARABIC LETTER ALEF WITH HAMZA BELOW], آ [U+0622 ARABIC LETTER ALEF WITH MADDA ABOVE], and ٱ [U+0671 ARABIC LETTER ALEF WASLA]).

That said, its presence usually indicates the location of a vowel.

It also has one or two minor functions such as in conjunction with tawiin (nunation) (see U+064B ARABIC FATHATAN ً ).

Certain parts of the arabic verb end in a long u-vowel that is conventionally written with a following alef that has no effect on pronunciation, eg. كتبوا ktbwɑ kætæbuːThe alef is omitted if a suffix is added, eg. كتبوها ktbwhɑ kætæbuː-haa

Hamza

ء [U+0621 ARABIC LETTER HAMZA] represents the glottal stop sound. For historical reasons, it is treated as an orthographic sign rather than as a letter of the alphabet. It sometimes stands alone, but usually appears with a 'carrier' letter - ALEF, WAW, or YEH for which separate precomposed characters are available in Unicode ( أ إ ؤ ئ ). Examples of use include أنكر نائم بناء

In modern printed arabic, the hamza is rarely shown when it occurs at the beginning of a word, but may appear in conjunction with another character. When the hamza is above another character you should typically use ٔ [U+0654 ARABIC HAMZA ABOVE] with the appropriate base character, although there are a number of exceptions, and for the Arabic language all the needed combinations are available as precomposed characters. For more details, see the character description.

Classical arabic distinguishes between 'cutting' and 'joining' hamza. 'Cutting' means always pronounced, 'joining' means frequently elided. The joining hamza is of little practical importance in modern arabic pronounced without the old case endings. When it does appear in modern Arabic, ٱ [U+0671 ARABIC LETTER ALEF WASLA] is used to indicate a joining hamza.

Teh marbuta

ة [U+0629 ARABIC LETTER TEH MARBUTA] usually has no sound, eg. مَدْرَسَة

However, it is sometimes pronounced t in specific grammatical contexts.

It is used for historical reasons to indicate the feminine ending, a, and is only used in final position. The dots are borrowed from TEH (ت). If any suffix is added, the ending is spelled with ت [U+062A ARABIC LETTER TEH], eg. مَدْرَسَتْنَا

In modern arabic it is not uncommon to find the two dots omitted, particularly on masculine proper names that have the feminine ending, eg. طلبه

Vowelled text may omit the short a diacritic before the TEH MARBUTA, because the sound is always the same.

Letters for foreign sounds

The following characters are not part of the standard Arabic language set but are occasionally used to represent foreign sounds.

Two of the above are borrowed from Persian/Urdu.

Other letters

The following characters also have the general property of Letter, but are less commonly used for modern Arabic language text.

ڢ [U+06A2 ARABIC LETTER FEH WITH DOT MOVED BELOW] and ڧ [U+06A7 ARABIC LETTER QAF WITH DOT ABOVE], are alternative forms that are used in Northwest Africa. ࢲ [U+08B2 ARABIC LETTER ZAIN WITH INVERTED V ABOVE is used for Berber.

ٱ [U+0671 ARABIC LETTER ALEF WASLA] is described in the section hamza. Whereas many of the above letters with diacritics decompose in Unicode Normalization Form D (NFD), this letter does not.

ﷲ [U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM] is a letter from the Arabic precomposed block used to write the name of Allah. The composition of this character differs from font to font in terms of glyph forms. With some fonts it is necessary to add diacritics, whereas with others it is not. 

ـ [U+0640 ARABIC TATWEEL] is used to stretch words for simple justification, or to make a word or phrase a particular width, or as a form of emphasis. For more information see justify.

ڢ [U+06A2 ARABIC LETTER FEH WITH DOT MOVED BELOW] and ڧ [U+06A7 ARABIC LETTER QAF WITH DOT ABOVE], are alternative forms that are used in Northwest Africa. ࢲ [U+08B2 ARABIC LETTER ZAIN WITH INVERTED V ABOVE is used for Berber.

ٱ [U+0671 ARABIC LETTER ALEF WASLA] is described in the section hamza. Whereas many of the above letters with diacritics decompose in Unicode Normalization Form D (NFD), this letter does not.

ـ [U+0640 ARABIC TATWEEL] is used to stretch words for simple justification, or to make a word or phrase a particular width, or as a form of emphasis. For more information see justify.

Characters in the Arabic Presentation Forms blocks should not normally be used, but they contain just a few characters that are not just for compability use, including the following. (Click on the characters for more details.)

ﷲ [U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM] is a letter from the Arabic precomposed block used to write the name of Allah. The composition of this character differs from font to font in terms of glyph forms. With some fonts it is necessary to add diacritics, whereas with others it is not. 

Arabic definite article

The pronunciation of ال (alif followed by lām) varies when it represents the Arabic definite article.

The lām is not pronounced if it precedes one of the following characters, but instead the following sound is doubled, eg. السلام علیکم

These are called 'sun letters' in Arabic. The other letters are 'moon letters'.j,32

The alif is also not pronounced if the preceding word ends with a vowel or h. It is, however, written.j,32

Vowel absence

When text is vowelled, ْ   [U+0652 ARABIC SUKUN] can be used over a consonant to indicate that it is not followed by a vowel sound. Like other vowel diacritics, this is typically not used in modern text, unless it is necessary to clarify pronunciation.

مَكْتَب

Onset clusters

TBD

Final consonants

Final consonants in Arabic are simply written using ordinary consonant letters.

Consonant clusters

Consonant clusters in Arabic are simply written using a sequence of consonant letters. In vowelled text the consonant(s) without a following vowel may carry a sukun (see novowel).

Consonant length

The diacritic ّ  [U+0651 ARABIC SHADDA] doubles the value of the consonant it is attached to, which is phonemically significant in Arabic, eg. تجّار

Like the short vowels, it, too, is not often used, although sometimes it appears when vowel signs don't.

When both shadda and kasra are attached to the same base consonant, a common, though not universal, practice is to display the kasra below the shadda, rather than below the base consonant, eg. مُمَثِّلْ Some fonts, such as Amiri, don't do this. (See also gpos.)

Consonant sounds to characters

This section maps Modern Standard Arabic consonant sounds to common graphemes in the Arabic orthography. 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.

Sounds listed as 'infrequent' are allophones, or sounds used for foreign words, etc.

Symbols

The following infrequently used character with a general category of symbol is the phrase recited before each sura (chapter) of the Qur'an (except for the ninth), and means "In the name of God, the Most Gracious, the Most Merciful".

Ligatures

The combination ل + ا [U+0644 ARABIC LETTER LAM + U+0627 ARABIC LETTER ALEF] is always written as a ligature. The underlying code points are, however, preserved. The form of this ligature that joins to the right is ‍لاand unjoined it is لا

Observation: When diacritics are used with this ligature, they sometimes appear to be over the ALEF, rather than over the LAM, eg. قليلاً This would require a typing order that is different from the spoken sequence.

Other combinations of characters are likely to also ligate (see gsub). The number of ligatures in text typically depends on the font used, but ligation can also be used as a device to manage justification, in which case it needs some degree of manual control

Formatting characters

Modern Arabic text makes use of a relatively large set of invisible formatting characters, especially in plain text, many of which are used to manage text direction. Descriptions of these characters can be found in the following sections:

• shapingcontrols
• directioncontrols

Digits

See type samples.

A set of arabic-indic digits are typically used in Middle Eastern and Gulf countries, whereas North African countries tend to use European digits. In neither area, however, is one digit style used exclusively.

The Unicode bidi_class property for these native digits is Arabic_Number, which makes them behave differently from ASCII digits, and differently from the set of extended digits used for Persian, Urdu, etc. For more information, see expressions.

Arabic script has its own number separators, which are used in Arabic language text when the non-European digits are used. They are ٫ [U+066B ARABIC DECIMAL SEPARATOR] and ٬ [U+066C ARABIC THOUSANDS SEPARATOR].

Arabic also has its own characters for ٪ [U+066A ARABIC PERCENT SIGN] and ؉ [U+0609 ARABIC-INDIC PER MILLE SIGN]. The ASCII % [U+0025 PERCENT SIGN] and ‰ [U+2030 PER MILLE SIGN] are also used.

The CLDR standard-decimal pattern is #,##0.###. The standard-percent pattern is #,##0% or #,##0٪.c

See also expressions about directional implications for handling expressions or sequences of numbers.

Extended-Arabic digits. Still in the basic Unicode Arabic block, as mentioned, there is a second set of digits in Unicode for use in languages such as Persian and Urdu.

۰␣۱␣۲␣۳␣۴␣۵␣۶␣۷␣۸␣۹

The glyph shapes are typically different for 3 of the digits (although not always the same 3 digits) in Persian, Urdu and Sindhi.

Arabic
Persian
Urdu
Sindi

Currency

Unicode has a character for the rial: ﷼.

Managing text direction

Unicode provides a set of 10 formatting characters that can be used to control the direction of text when displayed. These characters have no visual form in the rendered text, however text editing applications may have a way to show their location.

‫ [U+202B RIGHT-TO-LEFT EMBEDDING] (RLE), ‪ [U+202A LEFT-TO-RIGHT EMBEDDING] (LRE), and ‬ [U+202C POP DIRECTIONAL FORMATTING] (PDF) are in widespread use to set the base direction of a range of characters. RLE/LRE comes at the start, and PDF at the end of a range of characters for which the base direction is to be set.

In Unicode 6.1, the Unicode Standard added a set of characters which do the same thing but also isolate the content from surrounding characters, in order to avoid spillover effects. They are ⁧ [U+2067 RIGHT-TO-LEFT ISOLATE] (RLI), ⁦ [U+2066 LEFT-TO-RIGHT ISOLATE] (LRI), and ⁩ [U+2069 POP DIRECTIONAL ISOLATE] (PDI). The Unicode Standard recommends that these be used instead.

There is also ⁨ [U+2068 FIRST STRONG ISOLATE] (FSI), used initially to set the base direction according to the first recognised strongly-directional character.

؜ [U+061C ARABIC LETTER MARK] (ALM) is used to produce correct sequencing of numeric data. Follow the link and see expressions for details. 

‏ [U+200F RIGHT-TO-LEFT MARK] (RLM) and ‎ [U+200E LEFT-TO-RIGHT MARK] (LRM) are invisible characters with strong directional properties that are also sometimes used to produce the correct ordering of text.

For more information about how to use these formatting characters see How to use Unicode controls for bidi text. Note, however, that when writing HTML you should generally use markup rather than these control codes. For information about that, see Creating HTML Pages in Arabic, Hebrew and Other Right-to-left Scripts.

Expressions & sequences

This section is about sequences of numbers, rather than a sequence of digits. Sequences of numbers are sets of numbers separated by punctuation or spaces, such as 10–12–2022. Sequences of digits, such as 123, in Arabic text run LTR automatically.

A sequence of numbers used to express a range of values generally runs right to left in the Arabic language (and languages using the Thaana or Syriac scripts), whereas for Persian language text (and in Hebrew, N’Ko or Adlam scripts) it runs left to right.

This also tends to apply to expressions such as 1 + 2 = 3.

fig_range_ar shows Arabic text which is right-to-left overall, containing an ASCII-digit numeric range that is also ordered RTL, ie. it starts with 10 on the right and ends with 12 on the left.

In Persian, however, the sequence would generally run LTR, so 10 would be on the left, and 12 on the right. The underlying order of the characters that make up the expression, and the order in which they are typed, remain the same. (Click on each figure to see the underlying character sequences.)

However, the preferred order for a sequence of numbers may also depend on the context. For ISBN numbers, telephone numbers, and so forth, a left-to-right sequencing is likely to be preferred.

The default direction for a sequence in an application that implements Unicode fully will depend on:

1. the digits used (ASCII, Arabic or Extended Arabic),
2. whether or not the sequence is preceded by Arabic script text, and
3. the separators used.

Writing styles

Arabic orthographies can be grouped into a number of writing styles, some of which are more common for particular languages, while others can be used interchangeably for the same language. Sometimes the variations are adapted to usage, for example book text vs. inscriptions; sometimes the variants reflect regional, cultural or stylistic calligraphic preferences.

The different styles include Naskh, Nasta'liq, Ruq'a, Thuluth, Taʻlīq, Kufi, Diwani, Maghribi, Kano. The examples in this page use a naskh writing style. For a brief introduction to font styles, with examples, see the following pages: Text layout requirements for the Arabic script, the home page of arabicfont.org, and

The naskh writing style is the most prominent style for the Arabic language, and has become the default form of Arabic language content in most contexts. It has clearly distinguished letters, which make it easy to read, and can be written in small sizes.

The ruq’ah writing style was designed for use in education, in official documents, and for every-day writing. It is known for its clipped letters composed of short, straight lines and simple curves, as well as its straight and even lines of text. It is a functional style of writing that is quick to write and easy to read. It also doesn’t extend baselines, like a naskh font does. In 2010's rebranding of Amman a ruq'ah font family was created to serve as an italic face. Monotype has an interesting article on the development of ruq'ah.

The nasta’liq writing style is the standard way of writing Urdu and Kashmiri, and is also often a preferred style for Persian text. Key features include a sloping baseline for joined letters, and overall complex shaping and positioning for base letters and diacritics alike. There are also distinctive shapes for many glyphs and ligatures.

The kano writing style is a common way of writing Hausa in Nigeria in the ajami script, and like other East African writing it is based on Warsh (Warš) forms, which incorporate Maghribi characteristics. Some sources describe an alternative Hafs (Ḥafṣ) orthography, used with hand-written adaptations for the newspaper Al-Fijir.

The kufi writing style is the original style used for the Koran, but is not used for newspapers or official content today. However, it is used in modern content for logos and other stylised applications.

Cursive script

See type samples.

Arabic script is always cursive, ie. letters in a word are joined up. Fonts need to produce the appropriate joining form for a letter, according to its visual context, but the code point used doesn't change. This results in four different shapes for most letters (including an isolated shape). Ligated forms also join with characters alongside them.

The highlights in the example below show the same letter, ع [U+0639 ARABIC LETTER AIN], with three different joining forms.

Most Arabic script letters join on both sides. A few only join on the right-hand side: this involves 4 basic shapes for Modern Standard Arabic.

ء [U+0621 ARABIC LETTER HAMZA] doesn't join on either side.

Context-based shaping & positioning

Context-based shaping

See just above for shaping related to cursive joining.

In all but the most basic fonts, glyph shapes are highly variable for Arabic letters. For example, fig_mishafi_ka shows a wide variety of shapes produced by default in the Mishafi font for ك [U+0643 ARABIC LETTER KAF] when followed by various letters.

Teeth letters

A good font will constantly change the shape of glyphs slightly so as to create a more aesthetically pleasing, and in some cases a more easily readable, flow.

teeth_letters shows an example where the same word is displayed using different fonts.a,#h_teeth_letters The font on the left applies rules to distinguish the letter bases clearly. Note, in particular, that although there are 3 letters which are repeated, none of those letters uses the same shape twice.

Special joining forms

In more traditional fonts, you will also often see the join between certain characters above the baseline. Compare the highlighted character joins in fig_raised_joining, showing the same sequence of letters but with joins above vs. along the baseline. (The first font is Mishafi, and the second Scheherazade New.)

But actually a good font will typically have a range of shapes and placements for a given letter, depending on the adjoining letter. This is illustrated in noon_joining_forms.a,#multi-context-joining

Characters within a word may also combine vertically in certain groupings. See the example in vertical_joining.

Context-based positioning

When vowel or shadda diacritics are used they can be placed in different positions, according to the context.

When both shadda and vowel signs are combined with a base character, a more complicated set of rules may be applied. Depending on the font used, some vowel diacritics may be placed relative the shadda diacritic, rather than relative to the base character.

Font styling & weight

See type samples.

Case & other character transforms

Arabic has no case distinction.

However, as mentioned in numbers, Arabic sometimes uses ASCII digit glyphs and other times uses local digit glyphs. Some fonts and authoring applications allow you to choose which glyph shapes to use for the same underlying characters.

Arabic fonts may also have alternative shapes for glyphs, which can be turned on in certain circumstances. For example, some fonts have a set of swash forms for certain characters, which can be used for justification, or just for effect.

Grapheme clusters

In most cases Arabic text uses precomposed characters and omits vowels. Therefore grapheme boundaries are consistent with individual letters. Where this is not the case, the additions are combining marks, and the Unicode grapheme cluster is designed to span combinations of base character plus any number of following combining marks.

Larger typographic units

One potential complication is that fonts often render sequences of characters as ligated forms. The ligated forms are a font-specific feature, whereas grapheme clusters are based on code point sequences: some fonts may display the same sequence of characters without a ligated form. Most applications tend to move character by character through the text, producing situations like the cursor position in fig_gc_ligature_cursor.

This approach allows for easy deletion or insertion of any component of a ligated form.

Many Brahmi-derived scripts are segmented by units that incorporate more than one grapheme cluster, for operations such as in-word line-breaking, justification, letter-spacing, and initial letter highlighting. It is not clear whether such typographic units are needed for Arabic language text, since there is usually no hyphenation, and no initial letter highlighting, and letter-spacing and justification follow quite different rules, extending the baseline. It may be worth looking at the very rare examples of vertically-set lines with upright Arabic letters to check whether ligatures like lam-alif or others are kept together. (My expectation is that lam-alif is not split, but the others may be.)

Apart from lam-alif, where one can expect a rule to apply consistently, another issue is that an application that wants to keep ligatures together as a single unit would have to be aware of the rendering behaviour of the particular font in use, since some fonts have ligatures for a given code point sequence and others don't. There is no way of deriving this information from the code point sequence itself, since that is always exactly the same.

Browser behaviour

The last word on each line (only) has a decomposed sequence for the lam+hamza.

Cursor movement. Move the cursor through the text.
Gecko , Blink, and WebKit browsers steps through the text using grapheme clusters. This means that it takes 2 steps to get past the lam-alif ligature. The decomposed sequence in the last word is treated like any other grapheme cluster.

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.

Word boundaries

Words are separated by spaces.

In Arabic, small words like 'and' (و) are written alongside the following word with no intervening space (eg. الجامعات والكليات means 'universities and colleges', but there is only one space). Such small words are handled typographically as part of the word they are attached to.

Phrase & section boundaries

Arabic language uses a mixture of ASCII and Arabic punctuation. Other languages using the Arabic script may use different punctuation, such as the full stop in Urdu.

Arabic language text also uses ‐ [U+2010 HYPHEN], – [U+2013 EN DASH], and — [U+2014 EM DASH].

Bracketed text

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

Mirrored characters

The words 'left' and 'right' in the Unicode names for parentheses, brackets, and other paired characters should be ignored. LEFT should be read as if it said START, and RIGHT as END. The direction in which the glyphs point will be automatically determined according to the base direction of the text.

The number of characters that are mirrored in this way is around 550, most of which are mathematical symbols. Some are single characters, rather than pairs. The following are some of the more common ones for Arabic.

(␣)␣<␣>␣[␣]␣{␣}␣«␣»␣‹␣›

Quotations & citations

See type samples.

Arabic texts may use guillemets or quotation marks around quotations. Of course, due to keyboard design, quotations may also be surrounded by ASCII double and single quote marks. Note, however, that the order of use is different from that in LTR text, because they are not automatically mirrored.

Arabic language text uses 2 different sets of quotation marks. Sometimes they are mixed in the same text. The example in fig_quotation_marks uses both in a single sentence.

When using bracketing quotation marks, « [U+00AB LEFT-POINTING DOUBLE ANGLE QUOTATION MARK] is used at the start, and » [U+00BB RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK] is used at the end. The shapes are typically rounded, a shown in fig_quotation_marks.

When an additional quote is embedded within the first, the quote marks are ‹ [U+2039 SINGLE LEFT-POINTING ANGLE QUOTATION MARK] and › [U+203A SINGLE RIGHT-POINTING ANGLE QUOTATION MARK].

The other quote marks are ” [U+201D RIGHT DOUBLE QUOTATION MARK] at the start, and “ [U+201C LEFT DOUBLE QUOTATION MARK] at the end.

When an additional quote is embedded within the first, the quote marks are ’ [U+2019 RIGHT SINGLE QUOTATION MARK] and ‘ [U+2018 LEFT SINGLE QUOTATION MARK].

Unlike the bracketing quotation marks, these characters are not mirrored during display. This means that LEFT means use on the left, and RIGHT means use on the right.

Emphasis

Emphasis can sometimes be expressed by stretching the baseline of one or more words. See the section on justification below for more information about baseline stretching.

Abbreviation, ellipsis & repetition

Arabic uses … [U+2026 HORIZONTAL ELLIPSIS].

Inline notes & annotations

tbd

Other punctuation

Other punctuation marks used in Arabic include the following.

Other inline text decoration

Underlines & overlines

Underlines and overlines in Arabic text are usually further from the baseline than they are in Latin text. This is because the Arabic letters extend further from the baseline, and because there are also sometimes diacritics beyond those long extensions. Typically, the line will be drawn so that it is further from the baseline than any other glyphs reach.^§

In some cases, however, while still keeping the line further from the baseline than in Latin text, typographers don't clear the glyphs. In this case, the lines usually skip the ink of the other glyphs.

When skipping ink it is important to avoid leaving very short remnants of the line between glyphs, since these may look like dots or diacritics.

The Qur'an tends to use overlines, rather than underlines.

Line breaking & hyphenation

Lines are normally broken at word boundaries.

They are not broken at the small gaps that appear where a character doesn't join on the left.

Like most writing systems, certain characters are expected not to start or end a line. For example, periods and commas shouldn't start a line, and opening parentheses shouldn't end a line.

Text alignment & justification

See type samples.

Arabic script justification can be implemented using a number of different techniques, which ideally are applied in combination. These include:

• Adjustment of inter-word spaces.
• Micro-adjustments to the intra-word spaces that occur where characters don't join with the following letter.
• Use of wider glyph forms or swash forms for certain characters.
• Stretching the baseline connectors between letters, which we'll refer to here as kashida.
• Application of ligated forms to reduce the width of certain words.

(In hand-written manuscripts it is also possible to find instances where the letters the would appear at the end of the line are squeezed above the last word in the line, or hang into the margin.)

The application of the various techniques is generally subject to rules governing the frequency and location of use of particular methods. Rules can differ by writing style – for example, elongation is not normally used at all for ruq'a fonts. Where baseline stretching is applied, the rules for what can be stretched, and how much, are complicated. Unlike space-based width adjustments, baseline extension is not a question of simply adding equal-length extensions across the line. The rules tend to differ across orthographies, and eminent typographers of the past also had their own preferred or idiosyncratic rules.

The baseline extension character ـ [U+0640 ARABIC TATWEEL] is sometimes suggested as a way of producing justification by extending the baseline, however when a browser window is resized, or when new text is added near the start of a paragraph, lines wrap differently and all the places where tatweel would be needed have to be recalibrated. Thus tatweels only work for static text with fixed dimensions.

Better quality justification systems stretch glyphs, rather than adding baseline extensions. This dynamic stretching of glyphs is often called 'kashida'. In some typesetting systems, such as InDesign, the stretching can be produced automatically without the need for tatweel characters. InDesign has controls to vary the preferred length of the extensions.

Note that the result of the automatic justification in fig_kashida_justification is different from that in the newspaper clipping. For one thing, the kashida effect is only applied once per word (but is applied to most words). The rules determining which combinations of characters receive baseline stretching, and the extent of that stretching also differ.

InDesign also allows fonts to substitute long swash variants for certain characters, which soak more some of the horizontal space.

Well justified text would apply a mixture of swash characters, space stretching, kashidas and ligatures to achieve a visually appealing and effective justification. Also, the baseline stretching in fig_kashida_justification is flat. A more advanced system would instead produce elegantly curved kashidas more like handwritten text.

It is very common to see baseline stretching in modern Arabic text where a word or phrase is stretched to fill a particular space, eg. the Arabic tag line (الابداع المتجدد Creativity renewed) below the word Lexus in fig_lexus is stretched to be the same width.

Observation: Text that is stretched in this manner very often has multiple kashidas per word. This is perhaps understandable, given that usually only a small number of words are involved.

Text spacing

See type samples.

Spaces are not added between characters, with the exception of micro-spacing during justification, which is applied to word-medial letters that don't join to the left. On the other hand, the baseline within words is often stretched.

It is quite common to see Arabic text stretched to fit a given width, as shown in fig_lexus, but that type of stretching is more akin to justification than the typical letter-spacing that is applied to other scripts. The amount of stretch is determined by the area that needs to be filled.

In some cases, it may be that elongation of words is driven by stretching the distance between letters rather than matching an external template, for example to express emphasis or prolonged sound. However, as for justification, this is not normally based on an even amount of stretch between all letters, as letter-spacing tends to be in other scripts.

Baselines, line height, etc.

The alphabetic baseline is a strong feature of Arabic script on the whole, since characters tend to join there. This is not always the case: for example, some adjacent pairs or ligatures have joins above the baseline, and initial letters in some fonts may start slightly above the baseline, but for most cases it remains a strong feature.

The nastaliq writing style, on the other hand, uses arrangements of joined glyphs that cascade downwards from right to left, and ressemble a strongly sloping baseline.

However, even writing styles with an ostensibly flat baseline may, in good quality fonts, draw words on a slightly slanted baseline, or multiple baselines, as shown in fig_word_baselines.a,#multi-level-baselines

Characters within a word may also combine vertically in certain groupings, as mentioned in the previous section.

Line height and multi-script positioning. Even without the deviations from the baseline described above, the ascenders and descenders of Arabic letters tend to travel further from the baseline that is usual in Latin script text. Allowances for this need to be made for line height settings on a page, but also it can be problematic when combining Latin and Arabic text on the same line using different fonts for each.

If the Arabic font supports the needed Latin letters, the font design will already take into account the relative sizes of the letters, and their placement relative to the baselines of each script. If different fonts are used, though, it's important to match the baselines and harmonise the font sizes used.

Arabic letters have ascenders and descenders that tend to be longer than the Latin ones. fig_baselines shows ascenders and descenders for Arabic letters in the Scheherazade New font. With the addition of diacritics above and below the letters, the line height needs to be significantly higher than for Latin script text.

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 Arabic language uses 1 numeric and 2 fixed styles. Wikipedia lists 2 more styles: an old maghrebi sequence and the hijai sequence.

Styling initials

It is possible to find cases where Arabic enlarges and styles the first character at the beginning of a paragraph, but it is quite rare.

Observation: It is not clear whether it is appropriate to maintain the joining forms of the initial letter and the following letter. A good proportion of the examples seen have the initial letter in a box, in which case it appears to be in isolated form. For further discussion see this thread and this one.

General page layout & progression

Arabic books, magazines, etc., are bound on the right-hand side, and pages progress from right to left.

Columns are vertical but run right-to-left across the page.

Grids & tables

Tables, grids, and other 2-dimensional arrangements progress from right to left across a page.

Forms & user interaction

Form controls should display Arabic text from right to left, starting at the right side of the input field. Form controls should also usually be arranged from right to left.

fig_form shows some form fields from an Arabic language web page. Note the position of the labels relative to the input fields and the checkbox, mirror-imaging a similar page in English. Note also that the input text in the first field appears to the right of the box.

The position of a scrollbar should depend on the user's environment, not on the content of a page. A non-Arab user viewing a web page in Arabic shouldn't have to look for the scroll bar on the left side of the window. In a system that is set up for an Arab user, however, the scrollbar can appear on the left.