Rejoinder: The retroflex, like the Marathi ‘na’, has little do with caste

Shailaja Paik’s recent piece, “How the hard ‘na’ insists on Marathi’s caste hierarchy”, makes an argument worth engaging with seriously.

She observes that the Pune Brahmin register of Marathi has functioned as a social credential, leaving speakers from other parts of Maharashtra feeling as though their Marathi is lesser.

The phenomenon she describes is real: pronunciation can, in many contexts, function as a social marker. Yet she ends by writing, “The hard ‘na’ is a small sound with a long history and deep politics of caste creating segregated sociability.” In doing so, she conflates social hierarchy with linguistic structure, attributing caste politics to the phoneme itself. That is a grave mistake.

What is the retroflex

Retroflex consonants are produced when the tongue curls back toward the hard palate. They correspond, in Devanagari, to ट, ठ, ड, ढ, ण, ळ, and their equivalents across dozens of Indian scripts.

If retroflexion were truly a Brahmin marker, we would expect it to diminish sharply in the speech of non-elite communities across South Asia. The linguistic record shows the opposite: retroflexion is pan-Indian, structurally embedded, and socially non-exclusive.

A wide range of Indian languages employ the retroflex ṇa. In my mother tongue, Gujarati, the consonant is shared across caste names, ranging from Chāraṇ and Vaṇajārā, to Vāṇiyā, Luvāṇā, and Brāhmaṇ.

In Tamil and Malayalam, retroflexes are central to both identity and sound. The letter ழ (zh, /ɻ/), as in Tamiḻ (தமிழ்), and ള (ḷ), as in Malayāḷam (മലയാളം), are emblematic of Dravidian phonological depth and are often mobilised as markers of linguistic pride. This retroflexion is tied to Dravidian cultural assertion and anti-Brahmin ideological formation, the precise opposite of upper-caste prestige.

Gondi, spoken by roughly three million people, preserves a fully functional retroflex system traceable to Proto-Dravidian. The name Gōṇḍ itself contains the retroflex cluster ṇḍ, illustrating how deeply this feature is rooted in the language. The same is the case with Bhili, an Indo-Aryan language, and Santali, an Austroasiatic Munda language, both of which maintain retroflex consonants as core phonological features among Adivasi communities. In all three, their corresponding writing systems and scripts explicitly encode these distinctions, reinforcing their salience in both speech and representation.

Retroflexion is, in fact, one of the defining features of the South Asian Sprachbund, a geographic zone of linguistic convergence cutting across genetic language families. As Murray Emeneau argued, and as scholars like George Cardona, Colin Masica, and DNS Bhat have documented, these sounds are not inherited from Proto-Indo-European.

They were acquired through millennia of contact with Dravidian, Munda, and other substrate languages of the subcontinent. The hard ṇa entered Indo-Aryan speech through contact with the very communities that Brahmin prestige culture has historically marginalised.

To call it a Brahmin sound, then, is historically inaccurate.

‘Correct Marathi’

Paik’s pluralist conclusion compounds the problem. She ends her piece by declaring: “There is no single, correct Marathi, only Marathis.” This is meant to be liberatory, but it overlooks a basic constraint.

Phonemic distinctions such as the difference between na (न) and ṇa (ण), between la (ल) and ḷa (ळ) exist because they carry meaning. Mana (the mind) and maṇa (a unit of weight) are different words. A framework that treats all variation as equally valid, with no basis for recognising that distinctions exist and matter risks eroding language itself.

Phonetic prescriptivism has real costs when wielded as a social weapon, but the answer can be better access: the goal should be that every Marathi speaker, from Vidarbha, from Konkan, from Marathwada, has full command of the language’s phonological richness – not that that richness is declared politically suspect, informally abandoned, or allowed to erode through neglect. Inclusion is the anti-caste position.

What is actually being lost?

This shared pan-subcontinental phonology encodes the names of rivers, villages, and persons in a way that survives translation into no other sound system. The retroflex ṇa is in the name of Haryana and Konkan and Chennai. It is embedded in the names of people ranging from Jignesh Mewani to Zohran Mamdani. It resonates in the word Mahāparinirvāṇa, in Gurabāṇī, and in Gaṇatantra itself.

To let these distinctions erode is a loss which compounds: first from the classroom, then from the home, then from the poem, then from the prayer. The retroflex ṇa has already begun disappearing from spoken Hindi, flattened into the dental na.

The distinction between śa (श, palatal sibilant) and ṣa (ष, retroflex sibilant) has largely collapsed in most Indian languages. Pha (फ, aspirated bilabial) is merging into fa (फ़, labiodental fricative); ga (ग) is absorbing ġa (ग़, voiced uvular fricative). Our ghazals are becoming gajals, our phūl are becoming “fool”, and most Hindi speakers now get two out of three consonants in Gaṇeśa wrong.

The real problem

The anti-caste movement has produced rigorous thinkers, and it is worth recalling the standard they set. BR Ambedkar concluded that access, education, and inclusion needed to be democratised, and that Dalits should have full command of their languages rather than diminished versions of them.

Abandoning a phoneme is how a people lose the sounds of their own speech. Neither is mispronunciation resistance nor is language loss justice. The anti-caste movement deserves a more honest target than a hard “na”.

Aalok Thakkar is Assistant Professor, Department of Computer Science, Ashoka University.