Indian Free Speech Has US Text and UK Habits

Article 19(1)(a) of the Indian Constitution looks, on the page, like an American free-speech provision. It says all citizens have the right to freedom of speech and expression, and it imposes no internal qualifications on that right at all. The qualifications come in Article 19(2), as a separate clause, and they read like a checklist: the security of the State, public order, decency or morality, friendly relations with foreign States, contempt of court, defamation, incitement to an offence. Read together, the text invites a familiar American reading. State the right broadly. List the restrictions narrowly. Force the State to justify each restriction against the broad right. That is, roughly, how the First Amendment is read in the United States. It is not how Article 19(1)(a) is read in India. The Supreme Court has been candid about the gap. In Life Insurance Corporation of India v Manubhai D Shah (1992), the Court declared that Article 19(1)(a), unlike the First Amendment, is not an absolute right and must be exercised in a way that does not jeopardise the rights of another or clash with 'the paramount interest of the State or the community at large'.¹ What that formulation describes is not American free-speech law. It is a balancing power. Indian courts cite Holmes when they want to sound brave, and Strasbourg when they want to convict.

The text is American. The habit is European.

The doctrinal mismatch is older than most of the case law it now distorts. It begins with the First Amendment to the Constitution itself, passed in 1951 in direct response to Romesh Thappar v State of Madras (1950) and Brij Bhushan v State of Delhi (1950).² Both decisions had read Article 19(2) the way an American court would have read a similarly worded provision: narrowly, against the State, with the burden of justification on the government. Both struck down public-order laws because public order was not, at that point, on the list of permissible restrictions. Within a year, Parliament had added 'public order', 'incitement to an offence', and the qualifier 'reasonable' before 'restrictions'.³

The First Amendment is usually described as a widening of Article 19(2). It was. But it was also something more specific. It was the moment when the constitutional text moved from a structure that resembled the American absolutist model — broad right, listed exceptions, narrow construction — to a structure that resembled the European proportionality model — broad right, broadly listed exceptions, restrictions to be judged 'reasonable'. The word 'reasonable' is what does the work. It is the same word that does the work in Article 10(2) of the European Convention on Human Rights, where Strasbourg has built an entire jurisprudence of necessity and proportionality on it.⁹ It is also the word that gives Indian judges, in practice, a balancing power that American judges would refuse to exercise.

The point about 'reasonable' is worth pressing. An American court reading the First Amendment does not ask whether a restriction is reasonable. It asks whether the restriction is content-based or content-neutral, whether strict or intermediate scrutiny applies, whether the restriction is narrowly tailored to a compelling interest. None of these tests deploys 'reasonableness' as a primary criterion. By contrast, a Strasbourg court reading Article 10(2) does ask whether the restriction is necessary in a democratic society — which is, in operation, a reasonableness inquiry with proportionality built in. The Indian Constitution after 1951 has the European structure on the page. That structure invites the European method. The method has arrived. The rhetoric, however, has stayed American, and that is where the trouble starts.

What was inherited from the United States was the text and the rhetoric. What was inherited from the United Kingdom — and, after 1950, from the wider Council of Europe — was the habit. Indian free-speech jurisprudence has spent the seventy-five years since the First Amendment trying to reconcile the two, and has not succeeded. I think it is time to stop pretending that the reconciliation is conceptually coherent.

Sedition: a colonial instrument with a constitutional disclaimer

The clearest place to see the mismatch is in the law of sedition. The constituent-assembly record on sedition is unambiguous. Seth Govind Das, Somnath Lahiri, Damodar Swarup, and others objected to the inclusion of 'sedition' as a ground of restriction in Article 19(2) precisely because sedition was a colonial instrument that had been used to imprison Gandhi and Tilak.¹⁴ The word was deliberately struck from the draft. The First Amendment did not put it back. Section 124A of the Indian Penal Code — the sedition provision — survived because the Court read it down in Kedar Nath Singh v State of Bihar (1962), holding that it covered only those words which have the 'pernicious tendency or intention of creating public disorder or disturbance of law and order'.⁴

What does that test do? On its surface, it does American work. It distinguishes between mere criticism of the government and speech which is causally tied to disorder. In its operation, it does European work. The test is not Brandenburg. Brandenburg v Ohio requires that advocacy of force be 'directed to inciting or producing imminent lawless action' and be 'likely to incite or produce such action'.⁸ Imminence and likelihood are doing the work, and they are doing it with a strict-scrutiny lens that puts the burden on the State. Kedar Nath's 'pernicious tendency or intention' has neither imminence nor likelihood. It has a generalised tendency-and-intent gloss that can absorb almost any politically inconvenient speech if the prosecutor is willing to argue that disorder was foreseeable. The Court keeps citing 'clear and present danger' in the same breath as 'pernicious tendency'. The two are not the same thing. The first is a Holmesian standard with bite. The second is the colonial 'bad tendency' rule with a constitutional disclaimer attached.

Lawrence Liang, in his treatment of free speech in the Oxford Handbook, traces this inconsistency carefully. He observes that Indian standards run between 'bad feelings', 'bad tendency', and 'clear and present danger', with the Court choosing one or another according to the case it has in front of it.¹⁵ The choice is not articulated. The choice is rarely defended. The result is that two cases on similar facts can produce opposite outcomes, and the jurisprudence as a whole offers no predictable rule to citizens or prosecutors. That is the European habit. Strasbourg has been candid about it: Handyside v United Kingdom (1976) explicitly endorses a 'margin of appreciation' that hands national authorities significant discretion, with the Court reviewing only proportionality.⁹ What Strasbourg owns up to, Indian courts perform. The difference is that Strasbourg has structured the discretion. India has not.

Public order and proximity: a doctrine that points two ways at once

The proximity-and-proportionality test that has become, in name, the central doctrine of Article 19(2) review tells the same story. The test came together across three cases. Romesh Thappar required that restrictions on speech be directed solely at undermining the security of the State.² Ram Manohar Lohia demanded that the restriction have 'a reasonable relation to the object that the legislation seeks to achieve' — proximity, with the federal-court precedent of Rex v Basudeva behind it.⁵ Rangarajan then formulated the 'spark in a powder keg' standard, requiring that speech be 'inseparably locked up with the action contemplated', a perlocutionary test in which there should be no temporal gap between utterance and effect.⁶

Read together, those three formulations look like an Indian Brandenburg. They almost are. They are not, because the same Court that has articulated them has also continued to apply the older 'in the interests of' standard from Ramji Lal Modi v State of Uttar Pradesh (1957), under which restrictions are permissible if the speech has a 'calculated tendency' to disrupt public order — a tendency test, in other words, that pulls the doctrine back towards the colonial register. Lawrence Liang notes that Kedar Nath cited Ramji Lal Modi while ignoring Ram Manohar Lohia. The proximity strand and the tendency strand have continued to coexist for sixty years, with the Court picking the strand that fits the result it wants.

This is not balancing. Balancing has rules. The Strasbourg version of Article 10 has at least an articulated structure: necessary in a democratic society, prescribed by law, in pursuit of a legitimate aim, proportionate to that aim. Handyside makes that structure explicit. Indian Article 19(2) review has the appearance of structure — proximity, proportionality, reasonableness — but each of these is loose enough that the Court can apply it to support either result. The 'spark in a powder keg' language reads as a free-speech protection in Rangarajan. It reads as a censorship justification in cases the Court has chosen to forget.

The Court's own treatment of the 'clear and present danger' standard tells against the supposed alignment with American doctrine. In Babulal Parate v State of Maharashtra (1961), the Court explicitly rejected the American clear-and-present-danger test, on the ground that fundamental rights under Article 19(1) are not absolute and are subject to the restrictions in the subsequent clauses. That is, the Court has rejected the strict imminence-and-likelihood test that would actually discipline 'pernicious tendency', while continuing to use 'clear and present danger' as decorative language when it wants to sound liberal. The vocabulary is American. The operating principle is European, except that the operating principle has not been articulated as a principle. It has been articulated only as a list of factors. A list of factors is not a principle.

Compare what an actual structured proportionality test would have required in Kedar Nath. The Court would have had to ask whether Section 124A pursues a legitimate aim — yes, public order. Whether it is rationally connected to that aim — yes, on the face of it. Whether less restrictive alternatives are available — and here the Court would have had to confront the fact that ordinary public-order laws, including provisions of the CrPC and the offence of incitement, already cover the field. Whether the restriction is proportionate stricto sensu — and here the colonial provenance of the section, its uneven prosecutorial record, and the chilling effect on political speech would have entered the analysis. None of this happened. The Court read down the provision to save it. Reading down is a fudge that proportionality cannot tolerate. Indian doctrine relies on the fudge.

Obscenity and the import of dead English law

If sedition shows the doctrinal mismatch in action, obscenity shows it in time-lapse. Ranjit D Udeshi v State of Maharashtra (1965) considered whether D H Lawrence's Lady Chatterley's Lover was obscene under Section 292 IPC, and applied the Hicklin test from R v Hicklin (1868) — a Victorian standard that English courts had themselves abandoned by the time the Indian Court applied it.¹⁰ Hidayatullah J added a layer of the American Roth test on top, and an 'artistic redemption' gloss of his own, producing the sort of compound that no jurisdiction had ever endorsed as a single rule.

The strangeness of this is worth dwelling on. Hicklin asked whether the matter charged tended to deprave and corrupt those whose minds are open to such influences — a vulnerable-populations rule. It is the kind of rule that a colonial authority might apply to a subject population. The American Roth v United States (1957) replaced this with a community-standards test that took the average person, not the most susceptible person, as the reference point. Miller v California (1973) refined this further, dropping the 'utterly without redeeming social value' element that Roth had introduced. The Indian Court in Udeshi picked Hicklin and bolted Roth onto it. The result was a test that policed both the work and its likely effect on hypothetical vulnerable readers, with the Court reserving to itself the role of literary critic.

The Court only formally moved off Hicklin in Aveek Sarkar v State of West Bengal (2014).¹⁰ That is fifty years after Udeshi. The English original had been doctrinally dead since the Obscene Publications Act 1959. Why did Indian courts continue to apply a Victorian rule for half a century? The answer cannot be that the courts genuinely believed that the obscenity question turned on whether material would deprave the most susceptible reader; if they had, they would have said so consistently. The answer is closer to this: the Court wanted the result Hicklin allowed (censorship of works the bench found offensive), the rhetorical respectability the American tests offered (the language of community standards and artistic merit), and the discretion that an inconsistent doctrine permitted. Hicklin gave the result. Roth supplied the cover.

Hate speech and Section 295A: the executive as literary critic

The pattern is even sharper in hate-speech jurisprudence. Ramji Lal Modi read Section 295A's 'deliberate and malicious intention of outraging the religious feelings' as importing both an 'aggravated form' requirement and a 'calculated tendency' requirement. This is a structured-balancing test on paper. In application, it has not been one. Baragur Ramchandrappa v State of Karnataka (2007) is the case I would point at for anyone who wants to see the doctrinal mismatch operating on a literary work in real time.¹¹

The novel Dharmakaarana speculated, in a section, on the unclear parentage of Basavanna — a historical figure on whom scholars are not agreed. The Karnataka government issued an order under Section 95 CrPC forfeiting the book. The Supreme Court upheld the order. It did so by holding that the word 'appears' in Section 95(1) does not require the government to establish that all ingredients of the underlying offence, including mens rea, are made out. The government's view that the ingredients 'appear to be present' is enough, with the burden shifting to the petitioner to disprove it. The Court also cited Rangarajan's own test — that the relevant audience is 'an ordinary man of common sense and prudence and not that of an out of the ordinary or hypersensitive man' — and then proceeded to ignore that test, privileging the claims of a small subset of an offended community over the work read as a whole.

This is not American balancing, and it is not European proportionality. It is administrative censorship with a doctrinal veneer. The American First Amendment would not have permitted this outcome: a content-based prior restraint on a published novel, on the basis that the executive thinks an offence might appear to be present, is not the kind of restriction that strict scrutiny would tolerate. The European Article 10 test would not have permitted it either: Strasbourg's necessity requirement demands a pressing social need, and a speculative aside in a historical novel is not a pressing social need by anyone's measure. India, holding the doctrinal vocabulary of both systems, produced an outcome that neither would have produced. That is the cost of the mismatch.

The hate-speech jurisprudence makes one further pattern visible. Indian courts have, in Ramji Lal Modi and in cases that follow, drawn a distinction between speech that has a calculated tendency to disrupt public order and speech that merely offends. The distinction sounds proportionality-friendly. In practice, the burden of separating the two has been put on the offended group rather than on the prosecutor. Lawrence Liang reads this as a structural feature of the post-colonial public sphere: a deliberative-democratic ideal in Article 19(1)(a), and a nervous, hypersensitive public-order imagination in Article 19(2). The Court could choose to police the imagination, requiring that hurt-sentiment claims survive a serious necessity test before they justify forfeiture or prosecution. It has not chosen this. It has, instead, ratified the hurt-sentiment claim and converted it into a constitutional ground of restriction.

The counter-argument: doctrinal pluralism as a constitutional virtue

The strongest defence of the Indian approach does not deny the mismatch. It defends the mismatch as a virtue. The argument goes like this. India is not the United States. It is not the United Kingdom. It is not Council-of-Europe Europe. It is a country with a recent history of colonial speech-suppression, a fragmented public sphere, and a thicket of communal sensitivities that the founders themselves recognised when they wrote Article 19(2) more broadly than the First Amendment. A doctrine that imports rigid American absolutism would be a doctrine wrong for the country it has to govern. A doctrine that imports purely European balancing would surrender the textual ambition of Article 19(1)(a). The Indian Court's hybrid is a working accommodation. It uses American language where the speech is clearly political and the State is clearly the threat — Shreya Singhal is the example.⁷ It uses European language where the speech is religious or communal and the threat is more dispersed. The vocabulary is borrowed; the judgment is local.

I want to take this argument seriously, because it is the version of the defence that survives in legal-academic literature and that judges of the present Court would themselves probably articulate. There is force in it. The First Amendment, as it is practised in the United States, treats speech that an Indian observer might think dangerous as constitutionally protected — neo-Nazi marches, cross-burning short of true threat, calls for political violence that fall short of imminence. The European model is more comfortable with limits on hate speech and Holocaust denial. India, with deeper communal pressures than either, may need a third path. Sahara v SEBI (2012), where the Court read Chintaman Rao 'along with developments in European jurisprudence' to demand proportionality and necessity, is honest about the borrowing.¹³ Mark Tushnet, looking at India from outside, suggests that India's borrowed doctrines often acquire a domestic logic that is more important than the original source.¹²

But the defence proves less than its defenders think it does. If the Indian doctrine were a structured hybrid — if the Court had said, for instance, that political speech against the State gets American-style strict scrutiny, while communal speech among private parties gets European-style proportionality — then we would have a defensible third way. We do not. The Court has instead pretended that the same doctrine governs both cases, and has used the language of strict scrutiny in the rhetoric and the language of balancing in the analysis, with the choice between the two unarticulated. The litigant cannot predict which standard the case will be decided under. The prosecutor does not know what to charge. The publisher does not know what to publish. Doctrinal pluralism is a virtue when it is owned. The Indian variant is doctrinal incoherence dressed as pluralism.

What an honest doctrine would look like

An honest doctrine would say what it does. There are three changes I would push for, and I think each is constitutionally available without an amendment. Before I get to those changes, two preliminary observations are worth making, because both of them push back against the easy assumption that the present jurisprudence is settled.

First, on the freedom of assembly and association, Menaka Guruswamy has shown that the Court has been doctrinally vague in exactly the way it has been vague on Article 19(1)(a) — upholding Section 144 CrPC without principled limits, denying any right to strike to whole classes of citizens, treating freedom-of-association cases as factual disputes rather than rights questions. The pattern is the same. American-sounding rhetoric — the freedoms are constitutional cornerstones, the State must justify restrictions — is paired with case-by-case balancing that looks like the European margin of appreciation. Even the structure of Article 19(3) and 19(4), with their open-textured 'in the interests of sovereignty and integrity of India or public order or morality' formulations, encourages a balancing register. That is what the Court has performed. It has just not admitted the performance.

Second, the comparative literature on Indian constitutional law has noticed the borrowing without always noticing the disjunction. Mark Tushnet's chapter on the Indian Constitution seen from outside discusses how doctrines like basic-structure review have travelled abroad with their abstraction intact and their rigour stripped.¹² Free-speech doctrine is not in his list of major exports, and there is a reason for that. The Indian free-speech jurisprudence does not present, to the outside scholar, as a coherent body of comparative material. It presents as a set of negotiated outcomes whose underlying principle is hard to extract. A doctrine that is hard to extract is a doctrine that has not been articulated. That is the problem I am trying to name.

With those observations in place, the three changes:

The first is to acknowledge that Article 19(2) post-1951 is closer to ECHR Article 10(2) than it is to the First Amendment, and to apply a structured proportionality test consistent with that text. Sahara already begins this work; Shreya Singhal consolidates it for the vagueness limb. The Court should commit to a four-step test — legitimate aim, prescribed by law, rationally connected, proportionate — and apply it consistently to every Article 19(2) case. That includes sedition, obscenity, and Section 295A. The result might be more restrictive of speech than the American baseline, but it would be coherent.

The second is to abandon Kedar Nath's 'pernicious tendency' formulation in sedition cases and replace it with a properly Brandenburg-style imminence-and-likelihood test. The First Amendment makes a particular kind of bet about the relationship between speech and order, and that bet has been examined and defended over a long period. The Indian sedition jurisprudence has the bet without the examination. If the Court is willing to use the proportionality test for the rest of Article 19(2), it should use the imminence test for the sub-class of speech that is most clearly political — speech against the government as government. Shreya Singhal's 'discussion, advocacy, incitement' triad already gestures at this distinction. Make it doctrinal.

The third is to stop importing Hicklin, and to acknowledge that Aveek Sarkar's partial move to Roth needs to be carried through to Miller. Obscenity doctrine should track the works rather than the readers, and should require that material be patently offensive by community standards and lack serious literary, artistic, political, or scientific value. The half-import of Roth without its bracketing tests is exactly the kind of half-borrowing that has produced the present mess. A similar discipline is needed on hate-speech and 295A doctrine: a community-standards test that takes the offended audience to be the ordinary reasonable reader and not the most easily offended sub-group, and that requires the prosecutor to show the calculated tendency, rather than presuming it from the executive's say-so under Section 95 CrPC.

None of this is radical. None of it requires the Court to repudiate its prior cases. Shreya Singhal, Rangarajan, and Ram Manohar Lohia have already done most of the conceptual work. What is missing is the discipline to commit to one register at a time and to say so.

I would put the choice this way. Either the Court is reading a First-Amendment-shaped Article 19(1)(a), in which case it must accept that more speech the State wants to prosecute is constitutionally protected, including sedition and a wider range of communal speech than the present doctrine tolerates. Or the Court is reading an Article-10-shaped Article 19(2), in which case it must accept the Strasbourg method in full — necessity, proportionality, margin of appreciation, and the open trade-offs that come with them. The hybrid that pretends to be both is the doctrine we have. It is the doctrine that produced Baragur Ramchandrappa, the long survival of Hicklin, and the use of sedition as a routine instrument of prosecutorial harassment well into the twenty-first century, even as the Court was striking down Section 66A for vagueness in the same period. Indian free-speech jurisprudence cites Holmes for its bravery and Strasbourg for its results. The mismatch is the source of the dysfunction. It is also a fixable thing, if the Court is willing to say which constitution it is reading.

Get the week's biggest decisions, distilled.

A weekly digest for lawyers, students and the legally curious. Every Friday morning. Free.