Joining Type
Unicode property controlling how Arabic and Syriac characters connect to adjacent characters. Types include Right_Joining, Left_Joining, Dual_Joining, and Non_Joining.
What is Joining Type?
Joining Type is a Unicode character property that describes how a character participates in the cursive joining behavior of Arabic-script (and other scripts that use connected cursive forms, such as Syriac, N'Ko, and Mongolian). It is fundamental to correct text shaping for Arabic and related scripts, because Arabic letters change their visual form depending on their position in a word — specifically, whether they connect to the letters before and after them.
The Unicode standard defines the Joining_Type property with six values, stored in ArabicShaping.txt in the Unicode Character Database.
The Six Joining Types
| Value | Abbreviation | Meaning | Example |
|---|---|---|---|
| Right_Joining | R | Connects to the preceding character only (right side) | ا (Alef, U+0627) |
| Left_Joining | L | Connects to the following character only (left side) | Rare; some extensions |
| Dual_Joining | D | Connects to both preceding and following characters | ب (Ba, U+0628) |
| Non_Joining | U | Does not connect to either neighbor | Space, Latin letters |
| Join_Causing | C | Causes joining but may not be visible | Zero Width Joiner (U+200D) |
| Transparent | T | Ignored for joining purposes; acts as if absent | Combining marks, some diacritics |
How Arabic Letters Change Shape
An Arabic letter's joining type, combined with its position in a joining sequence, determines which of up to four contextual glyph forms (allographs) the font renders:
- Isolated form: The letter appears alone or between non-joiners
- Initial form: The letter begins a joining sequence (connects to the right, toward the next letter)
- Medial form: The letter is in the middle of a sequence (connects on both sides)
- Final form: The letter ends a joining sequence (connects to the left, toward the previous letter)
For example, the Arabic letter Ba (ب, U+0628, Dual_Joining) appears as:
| Context | Form | Appearance |
|---|---|---|
| Standalone | Isolated | ب |
| Word-initial | Initial | بـ |
| Word-medial | Medial | ـبـ |
| Word-final | Final | ـب |
This shaping logic is implemented by text shaping engines like HarfBuzz, which is used by virtually every modern application that renders Arabic text (browsers, LibreOffice, Android, iOS).
Transparent Characters
Characters with Joining_Type=Transparent are critical for correct shaping. A Transparent character between two Dual_Joining letters does not break their connection — the joining algorithm "looks through" it. Arabic diacritical marks (tashkeel, like fatha ◌َ U+064E and shadda ◌ّ U+0651) are Transparent, allowing vocalized text to join exactly as the unvocalized base text would.
Zero Width Joiner and Non-Joiner
Two special-purpose characters control joining explicitly:
- ZWJ (U+200D, Zero Width Joiner) — Join_Causing: forces a joining connection where it would not otherwise occur
- ZWNJ (U+200C, Zero Width Non-Joiner) — causes a break in an otherwise-connecting sequence
Quick Facts
| Property | Value |
|---|---|
| Unicode property name | Joining_Type (jt) |
| Data file | ArabicShaping.txt in UCD |
| Primary affected scripts | Arabic, Syriac, N'Ko, Mongolian, Hanifi Rohingya |
| Dual_Joining letters | Most common Arabic consonants (ب، ت، ث، ج…) |
| Right_Joining letters | Alef (ا), Waw (و), Reh (ر), Zain (ز), Dal (د) |
| Shaping engine | HarfBuzz (universal), Uniscribe (Windows) |
| Force join | U+200D Zero Width Joiner |
| Break join | U+200C Zero Width Non-Joiner |
関連用語
プロパティ のその他の用語
文字が最初に割り当てられたUnicodeバージョン。システムやソフトウェアバージョン間での文字サポートを判断するのに役立ちます。
Unicode property (UAX#11) classifying characters as Narrow, Wide, Fullwidth, Halfwidth, Ambiguous, or …
Unicode property listing all scripts that use a character, broader than the …
文字を大文字・小文字・タイトルケースに変換するルール。ロケール依存の場合があり（トルコ語のI問題）、1対多のマッピングもあります（ß → SS）。
文字が属する文字体系（例：ラテン、キリル、漢字）。Unicode 16.0は168個のスクリプトを定義し、Scriptプロパティはセキュリティと混在スクリプト検出に重要です。
サポートしていないプロセスで目に見える効果なく無視できる文字で、異体字セレクター・ゼロ幅文字・言語タグなどが含まれます。
名前付きの連続したコードポイント範囲（例：基本ラテン = U+0000〜U+007F）。Unicode 16.0は336個のブロックを定義し、すべてのコードポイントはちょうど1つのブロックに属します。
RTLコンテキストでグリフを水平に反転すべき文字。例：( → )、[ → ]、{ → }、« → »。
すべてのコードポイントを30個のカテゴリ（Lu・Ll・Nd・Soなど）の1つに分類する体系で、7つの主要クラス（文字・記号・数字・句読点・記号・区切り・その他）にグループ化されています。
同じ抽象的内容を持つが外観が異なる場合がある2つの文字シーケンス。正規等価より広い概念。例：fi ≈ fi、² ≈ 2。