Arabizi and Franco-Arabic to Arabic script: what 3, 7 and 5 mean

What Arabizi is, and why nobody writes it the same way

Arabizi, also called Franco-Arabic or simply Franco in Egypt, is Arabic written in Latin letters. Some Arabic sounds have no Latin equivalent, so writers borrowed digits that look like the Arabic letters: 3 resembles «ع», 7 resembles «ح», and 2 resembles the hamza «ء».

It spread through text messages and chat at a time when Arabic keyboards were not available on every phone and computer, and plenty of people still use it for quick messages, comments and usernames.

There is no single standard, though. One person writes 7abibi, another habibi. The letter qaf may be written q or 8; Egyptians often write it as 2 because it is pronounced as a glottal stop in Cairo, and Gulf writers often use g because that is how it sounds there. So no converter can understand every writer. The Transliterate tool uses a published table you can inspect before you trust the output, and leaves the rest to you.

The digit and letter table the tool uses

This is the table exactly as the tool applies it. You can also open it under Limits & mapping below the tool:

An apostrophe after a digit adds a dot: 3' is ghain, 6' is zah and 9' is dad. 7', which some people use for kha, is not in the table: it comes out as ha followed by the apostrophe. Write 5 or kh instead.

The remaining single Latin letters map to the nearest Arabic letter: b ب, t ت, d د, r ر, z ز, s س, f ف, k ك, l ل, m م, n ن, h ه, w و, y ي. c and x are not in the table and stay as they are.

The tool reads two-letter pairs before single letters, so sh becomes shin rather than seen plus ha. That occasionally misfires: ashal (“easier”) becomes «َشَل». Watch g as well: it always becomes jeem, so if the writer meant qaf, as in Gulf galbi, fix it.

Convert Arabizi to Arabic, step by step

1. Open Transliterate.
2. Set Scheme to Arabizi · explicit phonetic mapping and Direction to To Arabic.
3. Paste your text into Original text.
4. Under Single a e i o u in Arabizi, choose how vowels are handled. The next section explains the options.
5. Read the Transliteration preview, correct the words, then copy the result. For long texts, Export lines CSV gives you each source line next to its result.

salaam 3alaykum

سَلام عَلَيكُم

Notice the difference: aa in salaam became an alef, while each single a became a fatha.

Vowels: marks, long letters or nothing?

Arabizi usually writes short and long vowels with the same letter. In mar7aba the first a is a short fatha and the last one is a long alef. The tool does not know the word, so it applies one rule to every single vowel letter:

• Short vowel marks: a becomes fatha, e and i kasra, o and u damma.
• Approximate long-vowel letters: a becomes alef, e and i ya, o and u waw.
• Omit: single vowels are dropped; doubled long vowels such as aa, ee and oo are kept.

mar7aba

مَرحَبَ

mar7aba

مارحابا

mar7aba

مرحب

None of them gives «مرحبا» directly. Omit comes closest: add one alef at the end and you are done. That makes it the best default for chat messages, since written Arabic normally leaves short vowels out. Approximate long-vowel letters suits words full of long vowels:

3'ali

غالي

Choose Short vowel marks when you want to see a rough pronunciation before writing the word yourself.

Why the result always needs editing

The tool converts symbols one by one. It knows no words and no dialects, and it never guesses missing vowels. These are the fixes you will make most often:

• The article: el or al comes out as «ِل», «َل» or «ل». Write «ال».
• Words that start with a vowel: with short vowel marks or Omit, ana and ahmad get no initial alef, because the first a became a mark or was dropped; with long-vowel letters they start with a bare alef and no hamza. Add the hamza yourself: «أنا», «أحمد».
• Word endings: Arabizi does not distinguish «ـة» from «ـا», so salama ends in a fatha, an alef or nothing, depending on the option. Restore the ta marbuta yourself: «سلامة».
• Punctuation: Latin commas and question marks are kept; replace them with «،» and «؟».
• Dialect: el5eer and el5air are the same word, and an Egyptian writer may type 2albi for «قلبي», which comes out as «ءَلبِ». The tool writes what it sees.
• English words: every Latin letter is converted, so meeting becomes «ميتِنج». Remove English words before converting or put them back afterwards.

mar7aba, kifak? ana b5air

مرحب, كفك? ن بخر

After editing: «مرحبا، كيفك؟ أنا بخير». Small fixes, but none of them can be automated without understanding the sentence.

9aba7 el5eer

صبح لخير

meeting el sa3a 3

ميتِنج ِل سَعَ 3

In the last example the 3 stayed a digit because it stands alone, which is what the next section covers.

Digits: when they stay numbers

Messages contain times, prices, room and phone numbers next to digits that stand for letters. So by default the tool leaves a standalone digit alone, meaning one with no letters attached. A digit inside a word, like the 7 in mar7aba, is always converted.

Turn on Also map standalone Arabizi digits only if the writer uses a lone digit as a letter. Every real number in the text will be converted too:

room 302

روم ع0ء

0, 1 and 4 are not in the table and never change. Links and email addresses stay intact while Protect detected links & email is on, even if they contain a 3 or a 7.

The other direction: Arabic to Arabizi

Choose From Arabic to write Arabic text in Latin letters and digits, for example for a friend who understands spoken Arabic but cannot read the script. The tool runs the same table backwards: ha is 7, ain 3, kha 5, qaf 8, hamza in all its seats (ء أ إ ؤ ئ) 2, alef aa and ta marbuta h. Vowel marks carry over when present, and a shadda doubles the letter.

مُحَمَّد

mu7ammad

صباح الخير

9baa7 aal5yr

Unvowelled text comes out without short vowels, and the article becomes aal. It is a letter-by-letter conversion, not a pronunciation guide. Do not use it for a name on a passport or an official document: those follow romanization systems the tool does not implement, such as ALA-LC or ISO 233, or the spelling the person already uses.

Buckwalter: a transliteration that loses nothing

Arabizi is easy to read but throws information away: every hamza becomes 2, and both ta marbuta and ha become h. Send «مدرسة» (school) there and back:

مدرسة

mdrsh

mdrsh

مدرش

Buckwalter transliteration solves that. It gives every Arabic letter and every diacritic its own single ASCII character. The Quranic Arabic Corpus describes it as a one-to-one mapping with Unicode, which is why text it covers can go back and forth without loss:

مدرسة

mdrsp

Some of its symbols: $ for shin, * for dhal, H for ha, E for ain, x for kha, p for ta marbuta, Y for alef maqsura, > for hamza above alef, < for hamza below, ~ for shadda, o for sukun, and F, N, K for tanween. Case matters: s is seen and S is sad; z is zay and Z is zah.

شُكْرًا

$ukorFA

kitAbN

كِتابٌ

The tool implements the core symbols plus the letters پ چ ڤ گ, not the extra Quranic annotation symbols. Use Buckwalter when Arabic has to pass through a system that accepts only ASCII, or for corpus linguistics and text-processing work. Do not hand it to general readers; without the table it is close to unreadable.

Limits of the tool

• Up to 20,000 text units at a time.
• Symbol conversion only: no translation, no spell-checking, no guessing of vowels or grammar.
• No academic romanization standards such as ALA-LC, DIN 31635 or ISO 233.
• Ordinary Latin text is converted too, so review mixed-language text.
• Unknown characters are left unchanged.

If the Latin text is pure gibberish, like hgsghl ugd;l, it is not Arabizi at all but Arabic typed with the English layout active. See the wrong keyboard layout guide.