Unicode® Standard Annex #53
Unicode Arabic Mark Rendering
This document specifies an algorithm that can be utilized during rendering
for determining correct display of Arabic combining mark sequences.
This UAX makes no change to Unicode normalization forms, and does not propose a new normalization form. Instead, this is
similar to the processing used in [
MicrosoftUSE
]:
a transient process which is used to reorder text for display in an internal rendering pipeline.
This reordering is not intended for modifying original text, nor for open interchange.
Status
This document has been reviewed by Unicode members
and other interested
parties, and has been approved for publication by the Unicode Consortium.
This is a stable document and may be used as reference material or cited as
a normative reference by other specifications.
A Unicode Standard Annex (UAX)
forms an integral part of the
Unicode Standard, but is published online as a separate document. The
Unicode Standard may require conformance to normative content in a Unicode
Standard Annex, if so specified in the Conformance chapter of that version
of the Unicode Standard. The version number of a UAX document corresponds to
the version of the Unicode Standard of which it forms a part.
Please submit corrigenda and other comments with the online reporting
form [
Feedback
]. Related information that is useful in
understanding this document is found in Unicode Standard Annex #41,
Common References for Unicode Standard Annexes
.”
For the latest version of the Unicode Standard see [
Unicode
].
For a list of current Unicode Technical Reports see [
Reports
].
For more information about versions of the Unicode Standard, see [
Versions
]. For any errata which may apply to this annex, see [
Errata
].
Contents
The assignment of Canonical_Combining_Class values for Arabic combining characters in Unicode is different than for most other scripts. It is a mixture of special classes for specific marks plus two more
generalized classes for all the other marks. This has resulted in inconsistent and/or
incorrect rendering for sequences with multiple combining marks since Unicode 2.0.
The Arabic Mark Transient Reordering
Algorithm (AMTRA) described herein is the recommended solution to achieving correct
and consistent rendering of combining character sequences containing Arabic marks. This algorithm provides
results that match user expectations and assures that canonically equivalent sequences are
rendered identically, independent of the order of the combining marks.
Rules and recommendations for the correct display of combining marks are discussed in a number of places in the Unicode Standard,
including Section 5.13,
Unknown and Missing Characters, Section 7.9
Combining Marks
, and
Section 9.2
Arabic
in [
Unicode
]. Some general principles include:
Canonically equivalent sequences should display the same.
Combining marks from the same combining class are normally displayed using the
inside-out
rule, that is, from the base outward.
Combining marks from different combining classes (other than ccc=0) may be re-ordered with respect to each other if that helps to achieve the desired display.
In the Unicode Standard, the Arabic script combining marks include eleven different non-zero Canonical_Combining_Class values, as shown in
Table 1
. When a combining character sequence includes marks from more than one of these classes, the rendering system has to determine a display order in which to position these marks on the base character.
While it might be tempting to just use NFC or NFD, neither of these normalization forms will yield what Arabic readers expect. For one example that will be easily understood by all readers of Arabic script, given a combining character sequence including a
shadda
(ccc=33) and
damma
(ccc=31), NFC and NFD will move the
damma
before the
shadda
—at which point the default inside-out rendering rule would place the
shadda
above the
damma
, which is incorrect.
Some cases are obvious to readers of languages written with Arabic script, and thus will likely get the same display from various rendering implementations. However, many of the combining marks, especially those with ccc=220 and ccc=230, are not commonly understood. Different rendering implementations have made different decisions regarding display order, resulting in inconsistent behavior between one system and another.
AMTRA defines a method to reorder Arabic combining marks in order to accomplish the following goals:
The inside-out rendering rule will display combining marks in the expected visual order.
Ensure identical display of canonically equivalent sequences.
Provide a mechanism for overriding the display order in exceptional cases.
Table 1: Canonical_Combining_Class Values for Marks Used in Arabic Script
Canonical_Combining_Class (ccc) Value
Combining Marks in this Class
Combining grapheme joiner, combining alef overlay
27
fathatan, open fathatan
28
dammatan, open dammatan
29
kasratan, open kasratan
30
fatha, small fatha
31
damma, small damma
32
kasra, small kasra
33
shadda
34
sukun
35
Superscript alef
220
All other below combining marks except
small low noon with kasra
230
All other above combining marks, small low noon with kasra
The algorithm starts by reordering combining marks according to one of the Unicode Normalization forms, and then makes adjustments by moving certain marks closer to the base.
This specification defines a group of combining marks called “Modifier Combining Marks” (MCM) for use by this algorithm. MCM are combining characters that are normally used to modify the base character before them, and should normally be rendered closer to the base character than
tashkil
(supplementary diacritics, including vowels). The MCM characters are not formally classified as
ijam
(consonant pointing/nukta, and so on) in the Unicode Standard, but they are usually perceived by users as
ijam
The complete list of MCM characters is defined in the Unicode Character Database (see [
UCD
]) file PropList.txt.
The set of MCM characters is intended to be stable. Adding an existing Unicode character to the list of MCM could change the rendering of data that assumes the implementation of AMTRA. Additional characters may be added to the MCM at the time they are encoded (see Section 5.4
Rationale for Exclusion of Some Marks
).
In the following specification, parenthetical definitions, for example (D56), refer to definitions in the Unicode core specification.
Input:
A Combining Character Sequence (D56) containing one or more Arabic combining marks.
Output:
A canonically equivalent Combining Character Sequence reordered for rendering using inside-out stacking.
Steps:
Normalize the input to NFD
Within the result, for each maximal-length substring, S, of non-starter (D107) characters, re-order as follows:
Move any shadda characters (ccc=33) to the beginning of
If a sequence of ccc=230 characters begins with any MCM characters, move the sequence of such MCM characters to the beginning of
(before any characters with ccc=33).
If a sequence of ccc=220 characters begins with any MCM characters, move the sequence of such MCM characters to the beginning of
(before any MCM with ccc=230 or ccc=33).
Implementation note:
Considering that most Arabic fonts have higher quality glyphs for precomposed characters, implementations may try to recombine base characters with a combining character immediately following them if that would result in a precomposed Unicode character. For example, if after running AMTRA the first two characters of the output are U+064A ARABIC LETTER YEH and U+0654 ARABIC HAMZA ABOVE, an implementation may want to replace them with U+0626 ARABIC LETTER YEH WITH HAMZA ABOVE. (This also helps make sure that the dots of U+064A are not displayed, even if the font is not aware of the Unicode requirement for U+064A losing its dots when combined with U+0654.)
When this step is done, implementations should not skip combining marks. For example, if the output of AMTRA is the sequence
The following figure demonstrates the algorithm using an artificial sequence of characters:
The default display order implemented by the AMTRA will be correct for most uses. However in
situations where a different mark order is desired, U+034F COMBINING GRAPHEME JOINER (CGJ)
can be used to achieve the desired display order. The following sections give examples of the use of CGJ.
The following examples demonstrate why each of the respective characters is included in the MCM.
U+0654 ARABIC HAMZA ABOVE and U+0655 ARABIC HAMZA BELOW
The use of combining hamza above and below is discussed in
Section 9.2
Arabic
in [
Unicode
].
Example 1
Quran1
] (page 9, end of line 5)
In Example 1, AMTRA puts a
damma
over a
hamza above
If an orthography needs to place the
hamza above
over the
damma
, the text should be encoded as
AMTRA place the
kasra
below a
hamza below
If an orthography needs to place the
hamza below
under the
kasra
, the text should be encoded as
U+0658 ARABIC MARK NOON GHUNNA
Regarding inclusion of this mark in the MCM, Kew says “The ARABIC NASALIZATION MARK is considered equivalent to a ‘nukta’,
as it is a modifier
that binds tightly to the underlying letter.
” (italics added for emphasis) [
Kew
]. The character is the character encoded as U+0658 ARABIC MARK NOON GHUNNA.
U+06DC ARABIC SMALL HIGH SEEN and U+06E3 ARABIC SMALL LOW SEEN
ARABIC SMALL HIGH SEEN is included in MCM because most Quranic orthographies use the character as an MCM only. Orthographies that place the
small seen
differently will need to use a CGJ.
Example 2a
Al-Hilâlî
Example 2b
Al-Hilâlî
In Example 2a, the
small high seen
is rendered below the
sukun
, while in Example 2b, it is rendered over it.
The examples are indeed from the same document (Al-Hilâlî and Khân 1996), just two pages away. The
small high seen
has
different roles: in Example 2a it is a hint that the base letter,
sad
, should be pronounced as if it was a
seen
; in Example 2b, it is a pause-related hint.
Example 2a (characters and properties):
Running AMTRA on this string does not result in any changes.
Example 2b (characters and properties):
Running AMTRA on the string in Example 2b resulted in an undesired change. It puts a
sukun
over a
seen above
If an orthography needs to place the
seen above
over the
sukun
, the text should be encoded as
U+06E7 ARABIC SMALL HIGH YEH
Example 3
Milo
] (page 9, line 11)
In Example 3, AMTRA puts a
shadda
over a
small high yeh
If an orthography needs to place the
small high yeh
over the
shadda
, the text should be encoded as
Running AMTRA on this string does not result in any changes.
U+08F3 ARABIC SMALL HIGH WAW and U+08D3 ARABIC SMALL LOW WAW
U+08F3 ARABIC SMALL HIGH WAW “is functionally similar to the already-encoded
U+06E7 ARABIC SMALL HIGH YEH” and therefore
small high waw
is included in MCM
Pournader
]. In available examples,
small high waw
and
small low waw
are functionally equivalent and, because they emphasize the vowel, are strongly bound to the body of the word. For these reasons they are both included in MCM.
U+06E8 ARABIC SMALL HIGH NOON
Example 4a
Quran2
Example 4a has a
sukun
over a
small high noon
. AMTRA puts a
sukun
over a
small high noon
If an orthography needs to place
small high noon
over
sukun
, the text should be encoded as
Example 4b
Example 4b shows a practical orthography that uses
small high noon
for nasalization. It is theoretically possible for a vowel to appear
above the
small high noon
in this practical orthography. In such a case, AMTRA puts the vowel (in this case
damma
) above
small high noon
In order to force the
small high noon
above the vowel, use the CGJ (
Example 5
Quran3
Example 5 has a
fatha
over a
large round dot above
. AMTRA puts a
fatha
over a
large round dot above
If an orthography needs to place
large round dot above
over
fatha
, the text should be encoded as
NFD assures that sequences such as
composes
The Canonical_Combining_Class for
shadda
(ccc=33) is higher than most vowels; however, it should be displayed closer to the base than the vowels.
AMTRA is able to handle the special ligation of
kasra
and
kasra-like
characters which are ligated with a
shadda
or
hamza
in some styles and appear just below them instead of below the base letter; they still logically follow the
shadda
or
hamza
Meem above
(ccc=230),
meem below
(ccc=220) and other similar characters are not
included in the MCM because their behavior already meets normal expectations. Examples 6a-6c show that the
combining meem
is
normally stored after
fatha
kasra
or
damma
, whereas including
meem above
and
meem below
in MCM would have the undesirable effect of moving them in front of
fatha
kasra
or
damma
Example 6a
Quran1
] (page 11)
Example 6b
Quran1
] (page 21)
Example 6c
Quran1
] (page 19)
Sukun Alternate Forms
There are three
sukun-like
marks encoded at U+06DF..U+06E1 that are used in some Quranic orthographies to
denote different entities–they may not always represent a
sukun
. The Canonical_Combining_Class of these marks is 230,
so their ordering in the presence of other combining marks is not affected by AMTRA. However, because the combining class for the
sukun
is 34, these
sukun-like
marks will
not
be treated like a normal
sukun
in all cases. Users
who create data using these alternate
sukun
characters will have more flexibility than when using the normal
sukun
AMTRA does not make them equivalent to U+0652 ARABIC SUKUN, as that would make the algorithm unnecessarily complex and
make the usage of CGJ more frequent.
Maddah
Neither U+0653 ARABIC MADDAH ABOVE (ccc=230) nor U+06E4 ARABIC SMALL HIGH MADDA (ccc=230)
are MCM because they are normally displayed above vowel marks.
Combining Alef Overlay
U+10EFC ARABIC COMBINING ALEF OVERLAY (ccc=0) cannot be MCM because any character with a ccc=0 is not moved by AMTRA.
The input for the example below must be
lam
fatha
alef overlay
madda
, and it will not be reordered by the algorithm.
Example 7
Quran4
] (page 502)
Some rendering engines will insert a dotted circle for what they understand to be an invalid sequence. This is a problem in Arabic script
because something that appears invalid may actually be valid text in some lesser-known orthography of a minority language or in the Quran.
For example, the Microsoft Windows text rendering engine, described in [
Microsoft
], inserts a dotted circle in
combinations of certain Quranic marks that are known to appear with each other in the Quran.
Such spell-checking processes are best implemented at a higher level than a rendering engine. Also, a dotted circle insertion algorithm that
displays all canonically equivalent sequences identically is hard to design and the result may be counter-intuitive for its users.
Implementations of the algorithm may be adapted to insert dotted circles by applying the algorithm first and then inserting the dotted circles.
AMTRA is not intended or expected to be applied to stored text. However, there may
be situations unrelated to rendering where AMTRA may be useful, and this UAX does not prohibit such use.
For example, when a text editor is processing a backspace key, a decision
has to be made about what character(s) to remove from the text. For sequences involving combining marks, if the desire is to remove one mark
at a time, users may expect that the
outermost
marks should be removed first. For Arabic script the AMTRA could be used to
identify outermost marks.
When new combining marks are encoded, 220 should be used for below marks and 230 for above marks. In the special cases where an
alternative version of the basic
tashkil
is encoded, the same Canonical_Combining_Class as the
tashkil
could be used, but extreme care should be taken.
U+08D9 ARABIC SMALL LOW NOON WITH KASRA
When it was added to Unicode 9.0, the
small low noon with kasra
(which appears below the text) was mistakenly given a ccc=230 (mark above).
It should have been 220 (mark below), but that cannot now be changed. When used with other combining marks, there are a number of issues:
When used with any ccc=220 (marks below), in the absence of any
combining grapheme joiner
characters, the reordering by AMTRA will always place the
ccc=220 marks between the base character and the
small low noon with kasra
. If this is not desired then a
combining grapheme joiner
can be used
to prevent the reordering.
Combining class sequences containing both U+08D9 and another character of ccc=230 might have the same display but not be canonically equivalent.
For this reason it is recommended that U+08D9 be encoded at the very end of the combining mark sequence.
In Quranic orthographies where U+08D9 appears between a base letter and a
kasra
, such as Example 8 below, a
combining grapheme joiner
must be used in order to control the display order so that U+08D9 is not reordered by AMTRA to move after the
kasra
. The input for Example 8 must be
Font developers should make sure that
small low noon with kasra
is treated as if it was a mark below and therefore has no impact on rendering of any
marks above.
Example 8
Quran5
] (page 635)
Roozbeh Pournader authored the initial concept. Bob Hallissy and Lorna Evans assisted Roozbeh Pournader in turning that concept into this technical report. The three co-authors and co-editors continue to contribute to this document, making various technical and editorial changes, including the classification of newly encoded Arabic combining marks.
Thanks to David Corbett, Behnam Esfahbod, Asmus Freytag, Ned Holbrook, Richard Ishida, Thomas Milo, and Ken Whistler for feedback on and contributions to this document, including earlier versions.
The following summarizes modifications from the previous revisions of this
document.
Revision 11:
Reissued
for Unicode 17.0.0
Modifications for previous versions are listed in those respective versions.
© 2018–2025 Unicode, Inc. This publication is protected by copyright, and permission must be obtained from Unicode, Inc. prior to any reproduction, modification, or other use not permitted by the
. Specifically, you may make copies of this publication and may annotate and translate it solely for personal or internal business purposes and not for public distribution, provided that any such permitted copies and modifications fully reproduce all copyright and other legal notices contained in the original. You may not make copies of or modifications to this publication for public distribution, or incorporate it in whole or in part into any product or publication without the express written permission of Unicode.
Use of all Unicode Products, including this publication, is governed by the Unicode
. The authors, contributors, and publishers have taken care in the preparation of this publication, but make no express or implied representation or warranty of any kind and assume no responsibility or liability for errors or omissions or for consequential or incidental damages that may arise therefrom. This publication is provided “AS-IS” without charge as a convenience to users.
Unicode and the Unicode Logo are registered trademarks of Unicode, Inc., in the United States and other countries.
US