فارسی

Persian has 32 letters: the 28 of Arabic, plus four invented for Persian sounds with no Arabic equivalent.

The pattern is worth pausing on, because it is the clearest example anywhere of how an alphabet gets extended: take the nearest existing letter, add a mark. Three of the four use the same trick, tripling the dot. It is also why the Persian keyboard puts these four in awkward places. They were latecomers to the alphabet and they are latecomers to the board.

The more interesting half of the story is the letters Persian did not discard. Arabic distinguishes sounds Persian does not have, and rather than drop the letters, Persian kept them for Arabic loanwords and collapsed the pronunciations.

The consequence is a spelling system where a Persian schoolchild has to memorise which /z/ a word takes, the answer being whichever one Arabic used. It is roughly the problem an English speaker has with there, their and they're, running through the whole learned vocabulary.

It also makes the script a permanent record of which words came from Arabic. If a word contains ث ح ذ ص ض ط ظ ع, it is almost certainly an Arabic loan. That is close to a complete test, and it means a page of Persian can be read as a map of its own borrowing.

Persian, like Arabic, does not normally write short vowels. The three diacritics exist, zabar, zir and pesh, but they appear in the Qur'an, in dictionaries, in children's books and almost nowhere else. A reader supplies them from knowing the word. This is why Persian is unusually hard to read aloud at a beginner's level and unusually fast to read once you are fluent.

Two Persian mechanics have no Arabic equivalent, and both are invisible on the page. The first is the ezafe. Persian links a noun to what follows it with an unwritten -e, so ketāb-e man, my book, is written کتاب من, which on the page is just "book" then "me". The grammar is carried by a vowel that is not there, which is the hardest thing about reading Persian and the one thing you cannot point at.

The second is the zero-width non-joiner. Persian breaks words internally more than Arabic does, using an invisible character, U+200C, to stop two letters joining while keeping them one word. The verb می‌روم, mi-ravam, "I go", is one word with a deliberate seam in it. Get it wrong and the word is still readable but looks illiterate, roughly like writing "does'nt". It also breaks naive text processing, since the character is invisible and still counts.

Arabic's default hand is Naskh, which sits on a level baseline. Persian's is Nastaliq, which hangs each word from upper right down to lower left, so a line of text descends in a series of sloping cascades. It developed in fourteenth-century Persia and became the standard hand for Persian, and later for Urdu, while Arabic stayed with Naskh.

Nastaliq is very hard to render digitally, because the shape and vertical position of every letter depend on the whole word. Most Persian on a screen is therefore set in Naskh-based fonts, including this page, which is why Persian on a screen looks more Arabic than Persian on paper does. Urdu, which resisted the compromise, is the reason Noto Nastaliq exists at all.

Pashto is not a Persian dialect. It is a separate Iranian language, further from Persian than Persian is from Kurdish, and it is written in the same Perso-Arabic script extended a second time. Pashto uses 45 to 46 letters: the Persian 32 plus retroflex and other consonants Persian lacks, ټ ډ ړ ږ ښ ڼ ځ څ, several of them introduced by Bayazid Pir Roshan in the sixteenth century.

So the script layer of the Afghan language question is not a division at all. Persian and Pashto share an alphabet. The fight is over which language gets written in it in public, and that is the subject of the second module.

The differences between Iranian Persian and Dari are smaller than the political noise implies, and they run in an unexpected direction. Dari and Tajik are the conservative varieties; Iranian Persian is the innovator.

Classical Persian distinguished long ē from ī, and long ō from ū. Iranian Persian merged both pairs; Dari and Tajik keep them, so shēr, lion, and shīr, milk, are two words in Kabul and one in Tehran. Dari keeps ق and غ apart where most Iranian speech merges them, and Dari and Tajik keep the classical /w/ where Iranian Persian has moved to /v/.

Where the three genuinely part is the borrowed layer. Iranian Persian took its modern technical and administrative vocabulary largely from French and later English, Tajik took its from Russian comprehensively, and Dari takes some from English and increasingly from Pashto and Urdu. A word like "aeroplane" is a different word in each.

"Dari" is an old word. It appears in Arabic and Persian sources from the tenth century, generally taken to derive from dar or darbār, court, and to have meant the formal court language of the Sasanians. What is recent is its use as the official name of the language of Afghanistan, which dates from the 1964 constitution.

Whether that renaming was a linguistic tidy-up or a political act is contested. Nile Green argues the impulses behind renaming Afghan Persian as Dari were more nationalistic than linguistic. Zalmai Nishat frames it as one move in a longer project of anchoring Afghan sovereignty in Pashtun identity, running back through Mahmud Tarzi's Ottoman-influenced nationalism to Amir Abdul Rahman; on that reading, a separate name detached Afghanistan's Persian speakers from the wider Persian world and made Persian look like a local Afghan language rather than the regional one. The case on the other side is that pluricentric languages routinely carry different names in different states, that dari is not an invention, and that a state may reasonably name its own official standard.

What is not in doubt is that many Afghan Persian speakers themselves prefer "Farsi" and treat "Dari" as imposed. When the BBC rebranded its Afghan service BBC Dari rather than BBC Persian in 2017, more than 180 cultural figures from Afghanistan, Iran and Tajikistan signed a protest calling the change misleading and divisive.

Afghanistan has had no reliable census since 1979, and the estimates that exist are contested along the exact ethnic lines the argument is about. The range is the finding.

The first two lines disagree by a factor of two on the same question. Two things are nonetheless widely agreed: Persian is Afghanistan's lingua franca, the language strangers from different groups use with each other, and its second-language reach is much wider than its native base; and Pashto is the language of the group that has held the state for most of its modern existence.

There is a cartographic wrinkle worth noticing. Persian speakers in Afghanistan are routinely mapped as three separate ethnic blocks, Tajik, Hazara and Aimaq, while Pashto speakers are mapped as one. The effect, as Martin Lewis has pointed out, is that Persian looks smaller on a map of Afghanistan than it is. The Hazara case deserves its own line: Hazaras speak Hazaragi, a Persian variety with retroflex consonants and a distinct vocabulary layer, and are Shia in a Sunni-majority state. They are Persian speakers, but they are not Tajiks, and treating "Persian speakers" as a bloc hides that their marginalisation is religious as much as linguistic.

The Taliban did not invent this. Promoting Pashto as the language of the Afghan state is a project of the 1930s, run by a monarchy, and it long predates any Islamist politics. Pashto was given official status by royal decree under Zahir Shah in 1936, and by 1937 Prime Minister Muhammad Hashim was declaring it the only language of state, mandating it in government and education.

Through the 1930s to the 1960s, state-funded bodies, notably the Anjuman-i Adabi-yi Kabul, built Pashto technical and administrative terminology specifically to displace Persian. This is where pohantun, university, comes from, coined against Persian dāneshgāh. That word is now the centre of the fight, which is a neat historical irony: it was manufactured for this purpose in the first place.

The 1964 constitution named Dari and Pashto both official, and renamed Farsi to Dari. Persian regained ground between 1979 and 1992 under communist rule and the Tajik-led mujahideen government, and the national anthem shifted from Pashto to Persian. Regional languages gained official status where they held a local majority, first in 1980 and then in the 2004 constitution, which named Uzbek, Turkmen, Balochi, Pashayi, Nuristani and Pamiri as third official languages in their areas. The first Taliban government, 1996 to 2001, resumed the 1930s policies.

So the pattern is a state repeatedly trying to make its administrative language match its ruling group's language, against a population that mostly does not speak it. The unusual feature of Afghanistan is not that the state promotes a language; every state does. It is that the state promotes a minority language against the country's own lingua franca.

The Taliban have not formally removed Dari's official status. On paper, Dari and Pashto remain co-official and the third-language provision stands. The pressure is administrative and symbolic rather than legal, which makes it harder to point at and harder to reverse.

In August 2021, immediately after the takeover, Persian was removed from public billboards and the word dāneshgāh was taken off university signage nationally; Balkh University's trilingual signboard in Pashto, Persian and English was replaced with Pashto and English only. That October a lawyer in Kabul reported that ministries were conducting official communications in Pashto alone, that ministry logos were being redrawn with the Dari text removed, and that the central bank was writing to commercial banks in Pashto only. The same dispatch records a Taliban official correcting a journalist's dāneshgāh and demanding pohantun.

By September 2022, Herat University's new signboard read Herat Pohantoon, Pashto only, and the Sanaie weekly in Ghazni had switched from mixed Persian and Pashto to Pashto alone, in a majority-Persian-speaking province. In August 2023 the Taliban deputy minister of information and culture told graduates in Bamiyan not to waste their time learning Persian literature. Bamiyan is a Hazara, Persian-speaking province, which is presumably the point.

June 2024 brought a broad renaming campaign. Hamid Karzai International Airport became Kabul International Airport; Great Masud Road, Martyr Mazari Square and Burhanuddin Rabbani University lost their names; Charikar was renamed Imam Azam, and Education Square in Herat became Iqra Square. By January 2026 the new National Cancer Diagnostic and Therapeutic Hospital in Kabul was carrying Pashto and English signage only.

The Taliban's Arabic layer is real and it sits at the top of the register. The state calls itself an Imārat, an emirate, an Arabic word for an Arabic institutional form, in preference to jomhūrī, republic. Its leader's title is Amīr al-Muʾminīn, Commander of the Faithful, Arabic and deliberately caliphal. The flag carries the shahada in Arabic. The movement's official media has long run an Arabic edition alongside Pashto, Dari, Urdu and English.

The renaming campaign drew its replacements from Arabic and Islamic vocabulary rather than from either Pashto or Dari: Iqra, "recite", the first word of the Qur'anic revelation, replacing "Education"; Imam Azam, the honorific of Abu Hanifa, replacing Charikar. And the April 2026 decree pairs "Islamic" with "Afghan" as the two acceptable sources of terminology, which puts Arabic-derived religious vocabulary on the approved side of the line alongside Pashto.

The mechanism is educational. Taliban leaders were trained in Pakistani Deobandi madrasas where instruction ran in Pashto or Urdu, and where medieval Arabic texts and their translations replaced the Persian Islamic literature that had been the vehicle of Islamic learning in Afghanistan for centuries. Afghan Islam was historically Persianate; its theology, poetry and jurisprudence were transmitted in Persian. Deobandi Islam is not, and the cohort now running the state was educated in the second tradition. Arabic is in the titles, the naming and the seminary; it is not in the paperwork, and it has no official or working status in the administration.

Which is why a "de-Persianising, pro-Arabic" reading of this is too simple. Persian is already saturated with Arabic. Something in the order of 40% of the Persian lexicon is Arabic in origin, and the entire religious and legal register is Arabic; every Persian speaker in Afghanistan already says ketāb, hokūmat, qānūn, madrasa. Nobody is introducing Arabic into a language that lacked it.

What is happening is a register shift inside a shared Arabic-Islamic vocabulary: away from the Persianate tradition, in which Arabic religious terms sit inside Persian sentences and Persian carries the commentary, and towards a Deobandi one, in which the Arabic source text and its Urdu and Pashto gloss are the transmission line and Persian is surplus. The target is not the Arabic words. It is Persian's standing as a language of Islamic learning.

There is a practical asymmetry underneath the identity argument, and it is the objection Afghan academics actually make. Pashto is a poor substitute in that role. It has a much thinner classical literature than Persian, and its modern technical vocabulary was largely built to order in the twentieth century. Replacing Persian in higher education means replacing it with a language that, for now, has fewer books in it.

Put as its proponents would put it: Pashto is the language of a plurality or a majority of Afghans depending on whose figures you take, and it has been the state's official language since 1936, so using it in government is not novel. After ninety years of policy its standing in administration remains weaker than Persian's; from a Pashtun-nationalist view that is the anomaly being corrected rather than an injustice being done. The renaming of places named for mujahideen commanders is, on the Taliban's account, a state removing factional names after defeating those factions, which is ordinary state behaviour however unwelcome to the renamed. And no formal legal change to Dari's status has been made.

Against which: removing Persian from signage, from a hospital, from a university and from a provincial weekly, in Persian-speaking provinces, is hard to read as anything but a policy of public erasure, and the people affected are not a faction but roughly half the country. The Nowruz bans of both Taliban periods point at a cultural target broader than administrative language.

Tajik has changed alphabet three times in a hundred years, each time by state decision, each time for political reasons. Until the 1920s it was written in the Perso-Arabic script, and was not thought of as a separate language at all, simply Persian. Soviet reform simplified the Persian script from 1923 and then abandoned it; the Decree on Romanisation made a Latin alphabet law in April 1928, based on the Turkic Yañalif. The aim of that one is worth naming: cut Central Asia off from the Persian and Arabic textual world, and from the clergy who controlled literacy in it.

Cyrillic came in during the late 1930s, and after 1939 Perso-Arabic materials were banned. Near-universal literacy was achieved by the 1950s, in Cyrillic, which is the fact that makes reversal so hard; the script change worked. A reform in 1998 dropped ц, щ, ы and ь, leaving 35 letters.

Those six letters exist precisely because Persian has sounds Russian does not, in the same way پ چ ژ گ exist because Persian has sounds Arabic does not. The same language has now been patched into two different foreign alphabets, and you can read the patches to see what the language needed both times.

At the height of perestroika nationalism in 1989, a language law declared Tajik the state language, formally equated Tajik with Persian by placing the word Farsi alongside it in the law, and called for the gradual reintroduction of the Perso-Arabic alphabet. The banning of the Islamic Renaissance Party in 1993, and the civil war, stalled it; the script had become associated with the Islamist side. In 1999 the word Farsi was removed from the state-language law, which is the moment Tajikistan chose a separate national language over a shared Persian one. A new law in 2009 made Tajik the sole state language and removed Russian's standing as the language of interethnic communication.

In 2025, and again in May 2026, President Rahmon ordered expanded teaching of the Perso-Arabic script in schools and universities. It is taught as "Ancestral Alphabet and Texts", twice weekly in years 7 and 8, framed as access to the manuscript heritage rather than as a change of script. Read carefully, this is heritage policy and not script reform. Cyrillic has never been abandoned and there is no live proposal to abandon it. The Perso-Arabic script is being taught as something ancestral, which is to say as something past.

Two populations matter, and neither is in Tajikistan. Tajiks are Afghanistan's largest Persian-speaking group, estimated at a quarter to a third of the population, concentrated in Kabul, the northeast, Herat and Mazar-i-Sharif, and they write in Perso-Arabic rather than Cyrillic. So "Tajik" names an ethnic group across a border and a Cyrillic written standard on one side of it.

Samarkand and Bukhara, meanwhile, are historically Persian-speaking cities that ended up inside Soviet Uzbekistan when the borders were drawn in the 1920s. Official statistics put Tajiks at about 5% of Uzbekistan's population; scholarly work suggests substantial undercounting, because registering as Uzbek carries social and economic advantages. Tajik-language schooling has contracted, from 318 schools nationally in 2001 to 282 by 2004, and in Samarkand city down to a single mixed primary school.

Bukhara adds a third thread. Bukharian, the variety spoken by Bukharan Jews, is a Tajik-Persian dialect with a Hebrew-script tradition, now almost entirely diasporic in Israel and New York. Which makes three scripts for one language.

The two ends of the Persian world are moving in opposite directions at the same moment, and the symmetry is exact enough to state plainly. In April 2026 the government of Afghanistan ordered foreign terms purged from a Persian-speaking administration. In May 2026 the government of Tajikistan ordered wider teaching of the Perso-Arabic script as ancestral heritage.

One state is removing Persian from the script it is written in. The other is teaching a script it does not write in, to reach the Persian it can no longer read. Neither is really about language. Both are about which past a state wants to claim.