Perso-Arabic Alphabet: Which Language Uses Which Letter

Arabic Language > Perso-Arabic Alphabet

The Perso-Arabic alphabet is the Arabic alphabet as Persian extended it: the 28 Arabic letters plus پ چ ژ گ for p, ch, zh and g, 32 letters in all. Urdu, Pashto, Sorani Kurdish, Sindhi and Uyghur built on it and added letters of their own. Paste a text below to see which language it is.

The identifier reads only the letters, so it works on a sentence you cannot read yet. No Arabic keyboard? Type on our online Arabic keyboard and paste the result in. The eight samples, and how often the identifier gets a sentence right, are listed further down.

The Perso-Arabic family at a glance

Seven languages in this table write with the Arabic script and add letters for sounds Arabic does not have. Persian came first, and the next five built their alphabets on it. Jawi shares only چ with Persian and has its own letters for p and g. The last column matters most for telling them apart: it shows which Unicode ye and kaf each language uses.

Sources: Wikipedia's articles on the Persian alphabet, Urdu alphabet, Pashto alphabet, Kurdish alphabets (the Kurdo-Arabic section), Sindhi language (writing systems), Uyghur Arabic alphabet and Jawi script, read on 14 October 2026; code points from the Unicode Arabic chart (U+0600 to U+06FF) and, for Jawi's ݢ, the Arabic Supplement chart. The Urdu count of 39 or 40 is the one on our Arabic vs Urdu comparison; Wikipedia calls the number debated, because ة is sometimes counted as the 40th letter. The Sorani count includes وو, which the Kurdish Academy's order gives its own place. Sindhi and Uyghur write in other scripts too, and Jawi is co-official beside the Latin alphabet; this page covers only their Arabic-script alphabets.

Arabic is in the table as the base the others build on. Its written form is shared across the Arab world, while its spoken dialects differ from country to country, as our map of the Arabic dialects shows.

The four letters Persian added

Persian added four letters for sounds Arabic lacks: p, ch, zh and g. Each is an Arabic letter with extra marks, so a reader who knows the Arabic letters can guess the family resemblance at a glance.

Names and sounds follow Wikipedia's Persian alphabet article and our comparison of Arabic and Persian, which gives an example word for each. Persian letters join the way Arabic letters do. Seven of them never connect to the letter after them, ا د ذ ر ز ژ و, the six Arabic non-connectors plus ژ. Our guide to how Arabic letters join covers the rules the whole family shares.

Letter-by-language matrix

Twenty letters are shared by two or more of the eight alphabets, or come in look-alike versions with different Unicode code points. They are the letters to check first when you meet an unfamiliar text. A tick means the language uses the letter; a tick in brackets means it does so only in some words, places or spellings, explained under the table.

Sources for the bracketed ticks. Pashto: Wikipedia's Pashto alphabet article says ک is also commonly written ك, g is written ګ or گ, Peshawar spelling uses ے where Afghan spelling has ی, and ں appears in some dialects. Sorani: the Kurdistan Region standard uses ک, but Wikipedia notes ك is still in use; Sorani's h is ھ, yet we counted ه 692 times in the Kurdish Wikipedia's own article on the Kurdish language. Sindhi uses ژ in loanwords only. Urdu writes ة in a few Arabic loanwords. Jawi uses ة mainly in Arabic loanwords and ى mainly in European ones; Wikipedia's Jawi table writes kaf as ک, while the Jawi UN text we used types ك. Persian ۀ varies by region, and many writers type ه plus a small hamza instead.

Two pairs are easy to mix up. ي and ی look the same at the start or in the middle of a word, and so do ك and ک, but they are different characters. A search for a word typed with one will not find it typed with the other, and a tattoo or a sign can end up with a Persian letter in Arabic text. Our checker for Arabic tattoo text flags exactly that.

Letters only one language uses

These letters each belong to one of the seven languages and to none of the others in this table. One of them in a text names the language almost on its own, which is why the identifier looks for them first. Sounds are given in plain words and in the IPA, from each language's Wikipedia article.

Urdu also writes ۂ ۓ ۃ, forms of ہ ے and ة used in a few spellings. Sindhi has two signs that stand for whole words, ۽ (and) and ۾ (in). Jawi's ga is ݢ in Wikipedia's table, but the Jawi UN text we used types it ڬ (U+06AC), so the identifier accepts both. Uyghur and Sorani also write almost every vowel as a letter, where Arabic, Persian and Urdu usually leave short vowels out, so their words look longer: Uyghur ئادەم (adem, human being), the second word of Article 1 in the Uyghur UDHR, spells out both of its vowels.

Which ye, kaf and h each text really uses

The letter tables say what each alphabet allows. The counts below say what real texts type. We counted every look-alike ye, kaf and h in the UN's Universal Declaration of Human Rights in each language, and in the Central Kurdish Wikipedia article on the Kurdish language for Sorani, which has no Arabic-script UDHR in the collection we used.

In these texts, Persian and Arabic never share a ye or a kaf: one text uses only ی ک, the other only ي ك. Urdu does not use the Arabic ه at all; its h is ہ or ھ. And the Pashto text types the Arabic ي throughout, although Wikipedia gives ی for Afghan spelling, so one language can appear with either ye. The letter count is every Arabic-script letter in the text. The Sindhi figures come from the text layer of the OHCHR PDF.

How the identifier decides, and how often it is right

The identifier names a language outright only when the text holds letters that prove it, and it marks a result based on missing letters as "most likely". It works in three steps:

1. It collects the letters that differ between the eight alphabets (everything in the two letter tables above) and keeps only the languages whose alphabet has all of them.
2. If any of those languages has its own letters in the text, a letter neither Arabic nor Persian has, those languages win. One left: "This looks like" that language. Two or more: it says it cannot tell and lists them.
3. Arabic and Persian have no letters of their own, so they are named only from absence, in a text of 25 letters or more where no other language's own letter appears, and the result says "most likely". A shorter text with no telling letter gets "cannot tell".

Before step 1, presentation forms (pre-shaped letters from PDFs and old software) are turned back into plain letters. If a long text fits no alphabet until ي ك and ی ک are swapped, it says so: that is Persian typed on an Arabic keyboard, or the reverse.

We learned Arabic, not the seven languages on this page, so every letter and sound here comes from the linked sources, and a script checked every letter against its stated code point and every matrix cell against the identifier's own letter lists. To test it, we split each source text into sentences and ran every sentence of 25 letters or more through the identifier. The letter lists come from Wikipedia and were checked against these same texts, so this measures consistency rather than performance on text it has never seen.

All nine Sindhi sentences it could not name contain the Urdu letter ٹ, which is not in the Sindhi alphabet but appears in seven different words of the PDF's text layer, so the identifier saw letters from two alphabets and said so. The three Sorani misses contain Arabic words or an Arabic quotation, with ة or ي. The one wrong answer was a list of words in another Kurdish dialect, written with no Sorani letter, which it called most likely Persian.

Other languages written with these letters come out as the alphabet they share. The Dari UDHR is named most likely Persian, which is right, since Dari is a form of Persian. Punjabi in the Shahmukhi script is named Urdu, because it uses Urdu's letters. Saraiki gets "cannot name one language", because it mixes Urdu letters with Sindhi ones. A text that pastes two languages together is named after the one with its own letters, and a short phrase rarely carries enough letters to decide. Names change spelling across the family too, as our page on how to write Sarah in Arabic shows for one name.

The eight samples are Article 1 of the Universal Declaration of Human Rights from the UDHR in Unicode collection (Arabic, Persian, Urdu, Pashto, Uyghur and Jawi), Article 1 of the Sindhi translation on the OHCHR website, and for Sorani the first two sentences of the Central Kurdish Wikipedia article on the Kurdish language. For the 28 shared letters in all four of their positions, see our chart of the 28 Arabic letters in every position.

Frequently asked questions

More free tools

All free tools →

• ض ص ث قArabic Keyboard: Type Arabic Online
• مرحباArabic Transliteration: Convert Arabic, Arabizi and English Letters
• وشمArabic Tattoo Checker: Catch Stencil Errors Before the Ink
• اسمYour Name in Arabic Script, With Checked Spellings

Keep learning Arabic

• Up one levelArabic LanguageArabic is the fifth most popular language in the world. Learn more about the Arabic language and how you can get started learning.
• In this topicArabic vs Hebrew: Same Family, Very Different WorkloadsArabic vs Hebrew for English speakers: which one is faster to learn, the 22 letters they share, and 30 shared words, each checked against Wiktionary.
• In this topicArabic vs Persian: One Script, Two Unrelated LanguagesPersian (Farsi) is not Arabic: different language families, one shared script, 8,000 loanwords. See the letters, the grammar gap, and which is faster to learn.
• Also on OpenArabicIslamic Calendar 2027: Ramadan, Eid and Every Hijri DateIslamic calendar 2027 (1448-1449 AH) by Umm al-Qura: Ramadan from 8 February, Eid al-Fitr 9 March, Eid al-Adha 16 May; sighting can shift a day. Free PDFs.