Persian verse in Istanbul and Agra: a borrowed high culture (c. 1500)

What the receiving courts had before the canon arrived

Konya, 1277: an order against a language

On 13 May 1277, having taken Konya from the Seljuk administration, the Karamanid ruler Mehmed Bey issued a decree remembered in a single sentence: from that day forward, in the council, in the dervish lodge, in the court, in the assembly and in the public square, no tongue but Turkish was to be spoken ¹. In the Republic of Turkey the firman is taught as the founding document of Turkish as a language of state. It is more useful read the other way round — as evidence of what it was fighting. The apparatus it was trying to reform conducted itself in Persian. Four years earlier, in the same city, Jalāl al-Dīn Rūmī had died having composed some twenty-five thousand couplets of the Masnavī in that language, and his household would go on writing in it for another century ². Anatolian Turkish in 1277 was what people said. Persian was what was written down when something needed to last.

Fifty-three years later a dervish of Kırşehir named Âşık Paşa finished a long didactic poem called the Garib-nâme and explained in it why he had written the thing in Turkish at all: Türk diline kimsene bakmaz idi / Türklere hergiz gönül akmaz idi — nobody looked at the Turkish tongue, no heart ever inclined toward Turks ³. The couplet is quoted in Turkish schoolbooks as patriotism. Read as evidence it is something sharper: a writer in 1330 defending, to an audience he expects to object, the decision to use the language his audience spoke.

What Anatolian Turkish had, and what it did not

Anatolian Turkish in 1400 was not a mute language. It had a poetry with its own metre, its own occasions and its own saints.

• Syllabic metre (hece), counting syllables rather than weighing them — the natural prosody of a language with no phonemic vowel length.
• Yunus Emre (d. c. 1320), whose hymns in plain Turkish were sung in dervish lodges for six centuries and still are.
• A performance circuit outside the palace: the lodge, the village, and later the itinerant âşık singer with a long-necked lute.
• Narrative verse in Turkish for audiences that needed it, including Ahmedî's İskender-nâme, finished about 1390.

What it did not have was an apparatus. There was no canon whose members every educated person could name; no body of criticism that could say why one couplet was better than another; no biographical anthology certifying who counted as a poet; no fixed stock of images whose meanings a listener could decode instantly; and no metre whose mastery was itself the proof of education. All of those existed, fully built, three weeks' ride to the east.

Hindustan in 1400: a chancery, a cosmopolis, and no vernacular lyric

North India's situation was different in every particular and identical in the one that mattered. Persian had been the chancery language of Indian Muslim polities since the Ghaznavids took Lahore in the eleventh century, and the Delhi Sultanate had produced in Amīr Khusraw (1253–1325) a Persian poet good enough to be read in Iran. Alongside it stood Sanskrit, which Sheldon Pollock calls "the dominant culture language of precolonial southern Asia outside the Persianate order" — still producing commentary, still the medium of learning, still the language a king's inscription used when it wanted to sound like a king ⁴. Beneath both ran the vernaculars: Apabhramsha verse, the emerging regional literatures, and the devotional song of the bhakti poets, which was oral, popular and religiously charged.

What India in 1400 did not have was a vernacular lyric on the Persian model — a poetry in a language people actually spoke that used the quantitative metres, the fixed forms, and the closed image-stock of the Persian canon. It would take another three hundred years, and when it arrived it would be called Urdu.

The thing that was about to arrive

By 1400 the Persian literary tradition had been assembling itself for four and a half centuries and was very nearly finished. Ẕabīḥ-Allāh Ṣafā's history of Persian letters needs four volumes to reach the point at which this record begins, and what it describes is not a library but a canon: a short list of books that every educated person in the Persographic world was expected to have read, to be able to quote, and to be examined on ²⁷.

• Ferdowsī, Shāhnāma (c. 1010) — some fifty thousand couplets, and with them the entire repertoire of kingship, legitimacy and Iranian antiquity that any later ruler could borrow.
• Niẓāmī, Khamsa (d. 1209) — five narrative poems, including Laylā va Majnūn and the Iskandarnāma, which became the standard set for illustration in every workshop from Tabriz to Agra.
• Rūmī, Masnavī (d. 1273) — the mystical epic, and with it the vocabulary in which erotic love could be read as theology and theology as erotic love, in either direction, at will.
• Saʿdī, Būstān (1257) and Golestān (1258) — the school texts. Wherever Persian was taught, from Bosnia to Bengal, these were what beginners read ³⁰.
• Ḥāfeẓ, Dīvān (d. 1390) — the ghazal at its limit, and the book people opened at random when they wanted an answer.
• Jāmī (d. 1492) — the last, and the one whose death allowed the anthologists to close the list.

With the books came the machinery, and the machinery is the part that travelled. The quantitative ʿarūż metres supplied a measure of technical competence that could be assessed by anyone trained in them. The takhalluṣ, the poet's pen name signed into the closing couplet, supplied attribution. The radīf, the word or phrase repeated after the rhyme, supplied the constraint against which skill was visible. The naẓīre, the answering poem, supplied both apprenticeship and competition. The tazkira, the biographical anthology, supplied certification: a list of who counted, with samples, in an order that was itself a judgement.

What mattered about this system was not that it was beautiful. It was that it was complete. A court adopting it acquired, in a single motion, a syllabus, a set of examinations, a critical vocabulary, a supply of certified experts trained elsewhere at somebody else's expense, and a way of telling good work from bad that did not depend on the taste of whoever happened to be paying. Nothing the Ottoman or Mughal courts could have built themselves in a century would have done that. This is what portable means, and it is why the borrowing looks, from a distance, like surrender and was, up close, a purchase.

The transmission: Herat, and the two roads out of it

The court that became the model, 1469–1506

Sulṭān Ḥusayn Bāyqarā ruled Herat from 1469 to his death in May 1506, and the establishment he maintained there is the single most-copied court in Islamic history ⁵. Its personnel can be listed:

The court's self-understanding is worth stating plainly, because it explains why it exported so well. Herat did not think of itself as Persian in any ethnic sense. Its ruler was a Turk descended from Timur; its greatest patron wrote poetry in Chaghatay Turkic; its language of literature was Persian; its language of law was Arabic. Maria Subtelny's account of the Timurid settlement describes exactly this arrangement — Turko-Mongol military power resting on Persian administrative and literary culture, each party clear about what the other supplied ⁵. The model was therefore portable to any court with the same shape, which by 1550 meant most of the Islamic world between the Danube and the Bay of Bengal. Herat was not a Persian court. It was a Turkic court that had decided what Persian was for — and that decision, rather than the Persian language itself, is the thing the other courts copied.

The protest from inside, December 1499

The best evidence that this arrangement was felt as an arrangement — rather than as the natural order — is that Herat produced its most articulate critic. In December 1499 Nawāʾī completed the Muḥākamat al-lughatayn, "the judgment between the two languages," written in Chaghatay Turkic and arguing that Turkic was the superior literary instrument ⁶. His central demonstration is philological rather than rhetorical: he assembles a list of roughly a hundred Turkic verbs (ninety-nine in most manuscripts, a hundred in the Fatih and Budapest copies) for which, he says, Persian has no single equivalent and must resort to a phrase ⁶.

Nawāʾī's argument did not win. It could not: he was making it in a court whose prestige economy ran on Persian, and he made it at the end of a life spent financing Persian poets. But the Muḥākamat matters for the atlas because it is a contemporary, named, dated statement that speakers of the receiving language knew their language was being treated as the lesser instrument, and minded. Every subsequent claim in this record about vernacular devaluation rests on it rather than on hindsight.

The road west: Istanbul and the answering poem

No conquest carried the canon to the Ottoman court, and no single carrier can be named, because the traffic was continuous. Anatolian courts had been Persographic since the Seljuks. What changed across the fifteenth and sixteenth centuries was scale, deliberateness, and the emergence of an Ottoman Turkish lyric that could stand comparison with its models.

The mechanism was the naẓīre — the answering poem, written to the metre, rhyme and refrain of an existing poem, which is simultaneously homage, apprenticeship and competition. An Ottoman poet learned by answering Persian poems; he demonstrated mastery by answering them well enough to be answered in turn. The practice explains the shape of the whole tradition: the Ottoman gazel is not an imitation in the weak sense but an entry in a running contest whose earlier rounds had been held in Shiraz and Herat ⁷.

The certifying apparatus came too, and it arrived late enough to be dated. In 1538 Sehī Bey completed the Heşt Bihişt, the "Eight Paradises," the first Ottoman tezkire: a biographical anthology of 241 poets, arranged in eight sections by rank, with samples of each man's work ⁷. It is modelled directly on the Persian tazkira tradition, it was received with enthusiasm by the Ottoman elite, and it founded a genre that Ottoman critics would keep writing until the nineteenth century. That is the moment at which the transmission is complete in the technical sense: the receiving culture no longer merely produces poems in the borrowed system, it now judges them, ranks them and certifies them by the borrowed system's own rules, in its own language, without reference to Iran.

Sultans entered it personally. Meḥmed II (r. 1451–81) kept Persian poets on salary. Selīm I (r. 1512–20) — the conqueror of Egypt, the man who fought the Safavids at Chaldiran — left a dīvān of some 599 poems composed almost entirely in Persian, under the pen name Selīmī, and a large part of it consists of answering poems to the Persian canon, with Ḥāfeẓ the most frequent target ⁸. It was first printed at Istanbul in 1889 ⁸. There is no clearer single fact about the Ottoman relationship to Persian than this: the sultan who broke Safavid Iran at Chaldiran wrote his private verse in the language of the people he had broken, and signed it with a pen name.

The road east: exile, recruitment, and an administrative order

The Indian road runs through two men's misfortunes. Bābur, a Timurid prince squeezed out of Central Asia, took Kabul in 1504 and Delhi in 1526; he wrote his memoirs in Chaghatay Turkic, and the standard by which he judges every garden, poem and person in them is Herat, which he visited in 1506 and never stopped describing ⁹. His son Humāyūn lost India in 1540, spent years as a guest of the Safavid Shāh Ṭahmāsp, and when he returned in 1555 brought two painters out of the Tabriz workshop — Mīr Sayyid ʿAlī and ʿAbd al-Ṣamad — who founded the Mughal atelier ¹⁰.

Under Akbar the arrangement became administrative. From 1582 Persian was made the language of the revenue accounts throughout the empire, which meant that any Indian who wanted a place in government — Hindu or Muslim, Brahman or Khatri — had to learn it. Muzaffar Alam's account of the decision treats it as a deliberate act of political construction rather than a cultural preference: Persian was chosen precisely because it belonged to no Indian community and could therefore belong to the state ¹¹. The consequence was a Persian-literate Indian service class that outlived the dynasty by a century and a half.

Patronage did the rest. Iranian poets came to India in numbers across the sixteenth and seventeenth centuries — ʿUrfī of Shiraz (d. 1591), Naẓīrī of Nishapur (d. 1614), Ṭālib Āmulī, Abū Ṭālib Kalīm, Ṣāʾib of Tabriz (d. 1676) — because Mughal money was available and Safavid money had been redirected toward Shiʿi religious scholarship ¹². They did not come as missionaries of a culture. They came because India paid.

The dated skeleton

What changed, and what it pushed aside

A Turkish rebuilt to scan

The Persian metrical system, ʿarūż, is quantitative: it measures syllables as long or short, and a long syllable is one with a long vowel or a closed syllable. Persian has phonemic vowel length and supplies long syllables in abundance. Turkish does not have phonemic vowel length at all. To write Ottoman poetry in the inherited metres, poets therefore had to do two things: stretch native Turkish vowels artificially where the metre demanded length (imāle), and — far more consequentially — import Persian and Arabic vocabulary wholesale, because borrowed words came with the long vowels the metre required ⁷.

The cumulative result was a court register, Ottoman Turkish, whose grammar was Turkish and whose lexicon was so heavily Persian and Arabic that an Anatolian farmer could not follow a state document or a court poem. This is not a retrospective judgement; it is the structural fact that made the twentieth-century language reform both possible and, to the reformers, necessary. Walter Andrews's study of the gazel makes the corresponding literary point: the Ottoman lyric was not a private outpouring but a public, codified performance in which a shared vocabulary of images did the work, and competence in that vocabulary marked membership of a governing class ¹³.

What the Ottoman canon produced

The tradition that resulted is not derivative in any sense a reader of the poems would accept. Between roughly 1450 and 1600 it produced work that stands with its models:

• Aḥmed Paşa (d. 1497) and Necātī (d. 1509), who naturalised the Persian forms in Turkish and made the answering poem a Turkish genre.
• Fużūlī (d. 1556), who wrote three complete dīvāns — in Turkic, Persian and Arabic — and whose Turkish Leylā vü Mecnūn is the version of the story that Turkish speakers still know.
• Bâkî (1526–1600), whose elegy for Süleyman is the canonical Ottoman poem of grief for a ruler.
• Nefʿî (d. 1635), the satirist, whose invective got him executed at the palace.

Fużūlī's career also supplies the cleanest single instance of the whole transmission. His prose martyrology Ḥadīqat al-suʿadāʾ — the Garden of the Blessed, on the killing of Ḥusayn at Karbala — is a Turkish remaking of Kāshifī's Rawżat al-shuhadāʾ, written at Herat around 1502 ¹⁴. A Persian book made at the Timurid court within a decade of Jāmī's death becomes, within two generations, the Turkish text through which Ottoman and Azerbaijani Muslims mourn Karbala.

Husayn at the Bedside of the Dying Hasan, folio from a Ḥadīqat al-suʿadāʾ of Fużūlī, late 16th century. The Metropolitan Museum of Art, New York, sf1979.211. CC0 via Wikimedia Commons. · CC0

One convention of the inherited system should be stated plainly rather than stepped around, because the whole image-stock rests on it. The beloved of the Persian ghazal, and therefore of the Ottoman gazel and later of the Urdu ghazal, is grammatically ungendered and conventionally male — a beautiful adolescent, cruel by convention, unattainable by definition, and readable at the poet's discretion as a boy, a woman, a patron, a ruler or God. Walter Andrews and Mehmet Kalpaklı's study of the sixteenth-century material argues that this convention sat inside a real Ottoman culture of same-sex desire that had its own vocabulary, its own venues and its own scandals, and that reading the poems as pure allegory is as much a distortion as reading them as pure autobiography ²⁹. The convention is not a curiosity at the edge of the tradition. It is the hinge on which the tradition's central ambiguity — is this love poem about God? — actually turns.

The Indian style, and the reverse traffic

In India the Persian the immigrants wrote changed under them. What emerged in the seventeenth century — later labelled sabk-i hindī, the Indian style — is dense, metaphor-compounding and intellectually strenuous, built on the far-fetched conceit, the compressed abstraction and the line that has to be unpacked rather than heard. Shamsur Rahman Faruqi's defence of the style argues that the Iranian critical tradition's dismissal of it is a judgement about place rather than quality, and that its greatest practitioner, Mīrzā ʿAbd al-Qādir Bīdil (d. 1720), is simply absent from the Iranian canon ¹⁵. Ehsan Yarshater framed the same question as a choice between "progress or decline" and declined to settle it ¹⁶.

Meanwhile traffic ran the other way, and a record that ignored it would be dishonest. Akbar had the Mahābhārata translated into Persian as the Razmnāma between 1584 and 1586, by a bureau in which Sanskrit pandits and Persian secretaries worked side by side; Dārā Shukōh had fifty Upanishads rendered into Persian in 1657 as the Sirr-i Akbar, and it was through that Persian that Europe first read them. Audrey Truschke's study of Sanskrit at the Mughal court documents Sanskrit intellectuals as participants at Akbar's and Jahāngīr's courts rather than as subjects of a translation project done to them ¹⁷. Pollock records the corresponding fact on the Sanskrit side: Akbar's highest official, ʿAbd al-Raḥīm Khān-i Khānān (1557–1630), wrote verse not only in the vernacular but, modestly, in Sanskrit ⁴.

Nanhā (painter), ʿAbd al-Raḥīm ʿAnbarīn-Qalam (calligrapher). Invention of the Mirror in the Presence of Alexander the Great, from a Khamsa of Niẓāmī, Mughal India, 1595. Walters Art Museum, Baltimore, W.613 fol. 316b. Public domain via Wikimedia Commons. · Public domain

Rekhta becomes Urdu

The third movement is the one that outlived the other two. Across the sixteenth and seventeenth centuries, poets in the Deccan sultanates began writing the Persian forms in the local Indic speech — a language called by turns Hindi, Hindvi, Dakani, Gujri and rekhta, "poured together," which was also the name of a macaronic genre mixing Persian and Indic in varying proportions ¹⁸. The decisive moment came around 1720–21, when the dīvān of Valī "Dakhanī" reached Delhi and the Persian-writing poets of the capital discovered that the thing could be done in a language people spoke ¹⁸.

What followed was fast. Mīr Taqī Mīr (1723–1810), Sawdā, and then Ghālib (1797–1869) built in two generations a lyric tradition in a vernacular that used Persian metre, Persian forms, the Persian image-stock and the Perso-Arabic script — carried on an Indic grammatical skeleton. Ghālib himself, who is now read overwhelmingly for his Urdu, spent his life insisting that his Persian was the better work.

The structure of the result is worth setting out, because the whole later quarrel turns on it:

• Grammar and core vocabulary: Indic (Khaṛī Bolī, the speech of the Delhi region).
• Prosody: Persian ʿarūż, quantitative, with the same adaptations Turkish had needed.
• Forms: ghazal, qaṣīda, masnavī, marsiya — Persian throughout.
• Image-stock: rose and nightingale, moth and candle, wine and tavern, the cruel beloved — Khurasan and Fars, unchanged.
• Script: Perso-Arabic, in the nastaʿlīq hand developed for Persian.
• Critical apparatus: the tazkira, the ustād–pupil relation, the mushāʿira contest.

The institutions came across as completely as the forms. A poet apprenticed himself to an ustād, submitted drafts for correction, and was known thereafter by his master's line as much as by his own. The mushāʿira — an assembly at which poets recite in turn, each on a metre and rhyme announced in advance, the candle passing down the row from speaker to speaker — is the majlis of Herat and Shiraz, moved to Delhi and Lucknow and conducted in a different language. It is still held. There is no comparable case in world literature of an entire critical and pedagogical institution being transplanted intact into a language unrelated to the one it was built for, and surviving the transplant by three hundred years.

What was pushed aside

Three displacements can be documented, and they are of very different weights.

The first is the Turkish vernacular tradition, which was not destroyed but was placed permanently beneath notice. Court criticism had no category for syllabic verse; the tezkires that certified poets did not list âşık singers. A living tradition continued for six hundred years in the position of folklore.

The second is Sanskrit, and it must be stated precisely, because the tempting version is wrong. Pollock does not argue that Persian killed Sanskrit literature. His argument is that Sanskrit literary culture chose its own irrelevance: "In terms of both the subjects considered acceptable and the audience it was prepared to address, Sanskrit had chosen to make itself irrelevant to the new world," and, of the poets of the seventeenth and eighteenth centuries, that "Sanskrit writers had ceased to make literature that made history" ⁴. Persianate competition is one condition among several in his account, alongside the rise of the vernaculars and the narrowing of Sanskrit's own patronage base. What the record can state without controversy is the balance of demand. In William Adam's survey of Bengal schooling in the 1830s, five districts held 353 Sanskrit schools with 2,555 students, almost all Brahmans; in Burdwan district alone, 899 students were studying Persian at Muslim schools, half of them non-Muslim and about a third of those Brahmans ⁴. Persian was what you learned to get a job. Sanskrit was what you learned to be a Brahman.

The third is the Indian vernaculars in their own courts, which lost the revenue and legal domains to Persian after 1582 and did not recover them — until, in the nineteenth century, they got them back from a colonial administration that had taken them from Persian first.

The bill, and when it fell due

The transmission itself cost almost nothing. There is no conquest in it, no deportation, no forced conversion of a literature, and no documented victim. It ran on patronage, ambition and admiration, in both directions, for four hundred years. Nile Green's formulation is the one to hold onto: "by the fifteenth century Persian had become a language of governance or learning in a region that stretched from Burma to the Balkans, from Siberia to southern India," and "between the sixteenth and eighteenth century Persian reached the zenith of its geographical and social reach" ¹⁹. He also supplies the qualification that turns out to be the whole cost story: "It is a rule of thumb for learned lingua francas that their reach is geographically broad but socially shallow" ¹⁹.

A socially shallow prestige language is a debt instrument. For as long as the courts that maintained it existed, nothing was owed. When they stopped existing — in India from 1837, in Turkey from 1922 — the bill arrived in full, and it was presented to people who had borrowed nothing.

Turkey, 1928–1932: the severance

The Turkish Republic inherited a state language it could not use for a nation-state's purposes. Its response was two reforms in four years. On 1 November 1928 the Grand National Assembly adopted a Latin alphabet for Turkish, ending the Perso-Arabic script overnight for all official and eventually all public purposes. On 12 July 1932 the Türk Dili Tetkik Cemiyeti — later the Türk Dil Kurumu — was founded, and at the First Turkish Language Congress at Dolmabahçe that September it set about replacing Persian and Arabic vocabulary with Turkish words, revived, dialectal or newly coined ²⁰.

Geoffrey Lewis's history of the reform is titled The Turkish Language Reform: A Catastrophic Success, and the title is the argument ²⁰. It worked. Modern Turkish is a functioning, flexible, widely literate national language, and the literacy gains after 1928 were real. The cost is that the corpus on the other side of the change became unreadable — not difficult, not archaic, but unreadable — inside two generations.

The proof does not require an estimate or a contested percentage. It is a fact about a single book. Mustafa Kemal's Nutuk, the thirty-six-hour speech of 1927 that is the founding narrative of the Republic and was written by the man who ordered the reform, had to be re-rendered into contemporary Turkish in 1963, and again in 1986, because Turkish readers could no longer follow the original ²⁰. If the founder's own account of the founding needs translating twice within sixty years, six centuries of Ottoman dīvān poetry need no argument at all: they are simply gone from general readership, accessible in modern Turkish paraphrase or to specialists, and to nobody else.

That is what this transmission cost Turkey. The Persianate lexical layer was not a decoration that could be removed; it was load-bearing, because the metre had required it. Cutting it out was the only way to make the language national, and cutting it out severed the readership from its own literature.

The severance also cut both ways, and the second cut is less often noticed. A Turkish reader who cannot read Bâkî cannot read the Ottoman archive either — the court registers, the endowment deeds, the petitions, the provincial correspondence, six centuries of the ordinary paperwork of a state that governed twenty-odd modern countries. Ottoman palaeography became a specialist's skill in a generation, which is why Ottoman history is now read in Turkey largely through the mediation of people trained to read it, rather than directly. And the reform did not stop at what the metre had required. The Sun Language Theory, which dominated the Third Turkish Language Congress in 1936 and held that Turkish was the ancestor of the world's languages, was the point at which the project stopped being linguistic housekeeping and became something else; Lewis's account of that congress is the least affectionate passage in his book ²⁰. The theory was quietly abandoned. The vocabulary it had authorised was not.

India: when script became sect

The South Asian bill was presented differently and is harder to bound, so it should be stated in stages.

Stage one, the colonial substitution. Act No. XXIX of 1837 empowered the Governor-General in Council to dispense with the regulations requiring Persian in judicial and revenue proceedings, with effect from 1 December 1837 ²¹. Persian was replaced, in the North-Western Provinces, by Hindustani written in the Perso-Arabic script — which is to say by the language the Persianate order had produced. The Persianate order's own successor inherited its offices.

Stage two, the quarrel. In 1867 petitioners at Banaras asked the government to replace the Perso-Arabic script with Devanagari in the courts of the North-Western Provinces. Christopher King's study of the Hindi movement traces how, over the following three decades, a difference of register and script between two forms of one spoken language was built into a difference of communities, with organisations, journals, census campaigns and school curricula on each side ²². In April 1900 Lieutenant-Governor Antony MacDonnell granted Devanagari equal official status with the Perso-Arabic script in the province ²².

Stage three, the hardening. Tariq Rahman's history of the Hindi–Urdu split argues that the two languages were separated by social and political work rather than by linguistic distance, and that the separation was largely accomplished between 1867 and 1947 ²³. By the end of it, Urdu — a literary inheritance that in 1750 had been the shared property of Delhi's poets regardless of religion — was a marker of Muslim identity, and the script it was written in was the marker's visible form.

Stage four, the verdict from inside. The Persianate inheritance was not only attacked from outside it. In the decades after 1857, Urdu's own critics turned on the tradition they had been raised in. Muḥammad Ḥusain Āzād's Āb-e ḥayāt (1880) — the most reprinted and most widely read Urdu book of its century, and the work that fixed the shape of the Urdu canon — and Alṭāf Ḥusain Ḥālī's Muqaddama-e sheʿr-o-shāʿirī (1893) together argued for a "natural poetry": plain, sincere, referential, addressed to real conditions, and free of what they took to be the artificiality of the inherited Persian conventions. Frances Pritchett's study of the episode shows what that programme required its authors to do to their own tradition — to describe the ghazal's conventional beloved, its formal virtuosity and its whole apparatus of indirection as decadence, in terms borrowed substantially from the English criticism their British colleagues had taught them ²⁸. The Persianate ghazal survived the verdict and is still the most popular poetic form in South Asia. But its own critical establishment had convicted it, and the conviction is what every Urdu student since has had to read first.

The cost billed to this transmission is precisely this and no more: that the Persianate apparatus of Urdu — its script, its lexicon, its whole inherited surface — was available to be read as a religious marker when somebody wanted one, and was so read. The record does not bill Partition to a fifteenth-century Herat court, and it does not bill the students shot in Dhaka on 21 February 1952 during the protests against Urdu as the sole state language of Pakistan. Those deaths belong to the politics of 1952. They are named here because they are the visible edge of what language politics in South Asia became, and because an atlas that stopped at 1900 would leave the reader thinking the quarrel had been about spelling.

The cost the source culture imposed and did not pay

One item belongs on the bill in the other direction. The poets who made sabk-i hindī wrote in Persian, at the highest level, for Indian patrons — and the Iranian canon did not keep them. Ṣāʾib, ʿUrfī, Naẓīrī, Kalīm and above all Bīdil are marginal or absent in Iranian literary histories, while Bīdil is a national poet in Afghanistan and Tajikistan and a staple of Persian-language education there ¹⁵. Annemarie Schimmel's survey of Indian Islamic literatures registers the same asymmetry from the Indian side ²⁴. Mana Kia's account of Persianate belonging before nationalism shows how far this is a modern effect: in the eighteenth century, origin and place did not work as the exclusive categories that later national canons made them ²⁵.

This is the characteristic cost of a lingua franca with a metropolitan centre. The centre gets to decide, retrospectively, who was writing the language and who was merely using it — and it makes that decision after the periphery has done the work.

What is not on this bill

For completeness, four things this record declines to charge to the transmission:

1. The death of Sanskrit literary culture. Contested, multi-causal, and argued by its principal scholar to be substantially internal ⁴.
2. Partition, 1947. Language politics is one thread in it; a fifteenth-century poetic canon is not a cause of a twentieth-century constitutional catastrophe.
3. The Bengali Language Movement deaths of 21 February 1952. Named above, not billed.
4. The Ottoman Empire's own political failures. Divan poetry did not lose the First World War.

What remains, and what the cost rating rests on, is a real but bounded charge: two vernacular literary traditions held for centuries in a position of inferiority in their own courts; one national readership severed from six hundred years of its own literature within two generations; one shared literary inheritance converted into a communal marker; and one body of first-rate poetry orphaned by the canon of the language it was written in.

Coda: what the borrowing bought

It should be said last, because the cost framing can swallow it. What the Ottoman and Mughal courts got for this was not decoration. They got a fully specified system for saying difficult things — an inherited vocabulary of images precise enough that a poet could compress an argument about God, desire, power and death into fourteen syllables and be understood exactly, by a listener in Sarajevo or Golconda, without explanation. Fużūlī used it to write the grief of Karbala; Bâkî used it to bury a sultan; Mīr used it to write the most unguarded love poetry in South Asian literature; Ghālib used it to doubt in public. Bert Fragner's term for what made this possible is Persophonie — a written Persian domain held together by shared practice rather than by any state ²⁶. Nothing else in world literary history has covered that much ground, for that long, in a language that so few of its writers learned at home.

And it did not leave. The metres went into Urdu and stayed; the Urdu ghazal went into the Indian film song, which is the most widely heard poetry on earth. The nastaʿlīq hand developed for Persian is still the script of Urdu, and is the reason a Pakistani newspaper looks unlike an Arabic one. Persian vocabulary is load-bearing in Turkish, Urdu, Bengali, Punjabi, Sindhi, Pashto and Bosnian, in every case for words about paper, government, love and God. Two of the four things a lingua franca can leave behind — its words and its forms — are still in daily use by something over a billion people. The other two, its readership and its prestige, are what the twentieth century took away, and this record is mostly an account of who was holding them when it did.