Democracy as Abstraction: The Three-Judge Dissent in Abhiram Singh on Section 123(3) and Political Discourse

On 2 January 2017 a seven-judge Constitution Bench held by 4:3 that a candidate who appeals to voters on the basis of the voter's religion or caste commits a corrupt practice. The three dissenters — D.Y. Chandrachud, Adarsh Kumar Goel, and Uday Umesh Lalit JJ. — read the section narrowly. A reading of the dissent that nearly was the law.

On 2 January 2017, a seven-judge Constitution Bench of the Supreme Court decided that a candidate who appeals to voters on the basis of the voter’s religion, race, caste, community, or language commits a corrupt practice under section 123(3) of the Representation of the People Act, 1951. The vote was four to three. The three dissenters — D.Y. Chandrachud J., joined by Adarsh Kumar Goel J. and Uday Umesh Lalit J. — read the provision differently. On their reading, the word “his” in section 123(3) refers only to the religion of the candidate himself, or of a rival candidate. An appeal to a voter about the voter’s religion was not, in their view, the corrupt practice Parliament had described.

The line on which the disagreement turns is a single pronoun. The line on which the dissent rests is wider. To hold that a person who seeks to contest an election is prohibited from speaking of the legitimate concerns of citizens that the injustices faced by them on the basis of traits having an origin in religion, race, caste, community or language would be remedied is to reduce democracy to an abstraction.¹

The sentence carries the weight of the dissent. It frames the textual disagreement — what “his” refers to — as a question about what kind of political speech the Constitution permits. The majority’s reading proscribes appeals to voters’ religion or caste even when the appeal speaks to historical discrimination that voters share. The dissent’s reading preserves space for that speech. The disagreement is not really about grammar. It is about whether democracy in India can engage with religion and caste as facts of social life, or whether the Constitution requires that elections be sanitised of them.

The stakes that put weight on every word

To begin where the dissent does is to begin with the consequence. A finding under section 123(3) is not a finding of mere statutory contravention. It is a finding that the High Court hearing an election petition must record under section 99 of the Act, naming the candidate (and any person who acted with the candidate’s consent) as having committed a corrupt practice. The consequence under section 8A of the Act is a disqualification of up to six years, declared by the President on a reference from the Election Commission. The disqualified person stands debarred from voting at any election for the same period. The name of such a person is removed from the electoral roll under section 16. The person ceases to be an elector and ceases to be qualified to fill a seat in Parliament or the state legislatures for the period of disqualification.

The architecture is, in the dissent’s reading, quasi-criminal. “Election petitions alleging corrupt practices have a quasi-criminal character,” the dissent observes at paragraph 11, “where a statutory provision implicates penal consequences or consequences of a quasi-criminal character, a strict construction of the words used by the legislature must be adopted.” The rule reaches back to Tolaram Relumal v. State of Bombay, a 1954 Constitution Bench decision, and is reinforced by later cases on the standard of proof in election trials, which “veers close to that which guides a criminal trial.”

What this does to the interpretive question is to raise the threshold. A provision read broadly enough to bring within its sweep speech that the Constitution otherwise protects — speech about caste discrimination, about religious minorities, about language rights — cannot, in the dissent’s view, be allowed to expand by purposive construction. The legislature drafted what it drafted. The word it used is “his.” The interpretive question is what “his” means, read strictly, in a quasi-criminal statute with disenfranchising consequences.

The textual argument

Section 123(3) divides, the dissent suggests, into three parts. The first identifies who may commit the corrupt practice: a candidate, the candidate’s agent, or another person acting with the candidate’s consent. The second identifies what the act consists in: an appeal to vote, or to refrain from voting, for any person. The third identifies the basis on which the appeal is made: “on the ground of his religion, race, caste, community or language.”

The pronoun “his” is the hinge. Grammatically, the dissent argues at paragraph 15, it refers back to its nearest antecedent: “any person” in the second part of the section — meaning the candidate in whose favour the appeal is made, or the rival candidate against whom the appeal is made. It is impossible to construe sub-section (3) as referring to the religion, race, caste, community or language of the voter. The structure of the sentence does not support that reading. The voter is not the person identified anywhere in the section. The voter is the recipient of the appeal, not its subject.

Two further textual markers reinforce the point. The provision uses “the ground”, not “a ground.” The definite article suggests exclusivity — the appeal must be made on the ground of religion, meaning solely or substantially on that basis. Pre-1961, the section read “grounds” in the plural; in 1961, Parliament substituted the singular. And the structure of the surrounding text refers throughout to the candidate, to the candidate’s agent, and to the person rendered electorally accountable by the section. The voter, as an entity whose characteristics might bring section 123(3) into play, simply does not appear.²

The textualist reading does not rest on grammar alone. The dissent develops a constitutional rationale for why Parliament drafted the section the way it did. A representative of the people, the dissent observes at paragraph 17, represents the entire constituency. The Constitution rejected separate electorates. A candidate who solicits votes on the basis of his own religion compromises that representative function. A candidate who attacks a rival on the basis of the rival’s religion does the same. But a candidate who speaks to a constituency about the social discrimination its members have suffered is doing something the Constitution does not merely tolerate but, in numerous provisions, contemplates.

The constitutional architecture of identity

The dissent’s argument widens at paragraph 18. The Constitution is not, the judges observe, indifferent to religion, caste, or language. “There is no wall of separation between the state” and these facts of social life, the dissent observes, putting the proposition against the U.S. constitutional metaphor it invokes by contrast. The text places these facts of social life within the structure of constitutional protection itself.

The catalogue is dense. Article 15(1) prohibits discrimination against any citizen only on grounds of religion, race, caste, sex, or place of birth, but clause (4) permits special provisions for socially or educationally backward classes and for Scheduled Castes and Scheduled Tribes. Article 16(4) permits reservation in public employment. Article 17 abolishes untouchability. Article 25 protects religious freedom, but clause (2)(b) authorises social welfare and reform laws including the throwing open of Hindu religious institutions. Articles 29 and 30 protect linguistic and religious minorities’ rights to conserve culture and to administer educational institutions. Articles 330 and 332 reserve seats for Scheduled Castes and Scheduled Tribes in Parliament and state legislatures. Article 341 empowers the President to specify the Scheduled Castes. Articles 343 and the Eighth Schedule address language.

What this catalogue establishes, the dissent argues, is that the Constitution treats religion, caste, and language not as private matters to be excluded from political life but as facts of social life with which the constitutional order itself engages. “Religion, caste and language,” the dissent observes at paragraph 20, “are as much a symbol of social discrimination imposed on large segments of our society on the basis of immutable characteristics as they are of a social mobilisation to answer centuries of injustice.” Electoral politics, on this reading, is one of the legitimate domains in which that mobilisation occurs. “Social mobilisation is a powerful instrument of bringing marginalised groups into the mainstream. Access to governance is a means of addressing social disparities.”³

From here the dissent moves to its strongest formulation. To prohibit a candidate from speaking to voters about historical discrimination on the basis of religion or caste is to read the Constitution as requiring electoral discourse to be evacuated of the very categories the Constitution itself works with. The result, in the dissent’s rendering, is a democracy reduced to abstraction — a politics that cannot speak of the realities it exists to address.

Speech, discourse, and what the section actually proscribes

The distinction the dissent draws at paragraph 21 is between an appeal and a discussion. An appeal under section 123(3), on the narrow reading, is a solicitation of votes founded on the candidate’s religion or caste. “If a candidate solicits votes on the ground that he is a Buddhist that would constitute an appeal on the ground of his religion. Similarly, if a candidate calls upon the voters not to vote for a rival candidate because he is a Christian, that constitutes an appeal on the ground of religion.” That conduct, the dissent agrees, is what section 123(3) was drafted to prohibit.

The Constitution does not deny religion, caste, race, community or language a position in the public space. Discussion about these matters — within and outside the electoral context — is a constitutionally protected value and is an intrinsic part of the freedom of speech and expression. The spirit of discussion, debate and dialogue sustains constitutional democracy.D.Y. Chandrachud J., para 21, joined by Goel and Lalit JJ.

The line maps a doctrinal distinction onto a constitutional principle. Discussion of religion, caste, or language during an election is protected speech under Article 19(1)(a). The corrupt practice provision in section 123(3) operates at the boundary of that protection — it proscribes a specific kind of speech (the appeal to vote founded on the candidate’s own identity) without extending its reach to political discourse generally. Reading “his” to cover the voter’s religion, on the dissent’s analysis, collapses the distinction. It absorbs into the corrupt practice category speech that the Constitution otherwise protects.

The Article 19(1)(a) anchor is doing real work in this argument. Section 123(3) is, at its widest interpretation, a restriction on political speech imposed by a statute carrying quasi-criminal consequences. Restrictions on Article 19(1)(a) must fit within Article 19(2). They must be reasonable. And under the doctrine the Court itself has developed, restrictions on political speech operate at the heart of what 19(2) is most reluctant to permit. A reading of section 123(3) that proscribes ordinary political engagement with caste-based discrimination or religious minorities’ concerns is, in the dissent’s view, in tension with what Article 19(1)(a) is for.

Legislative history, and what the 1961 amendment did

Section 123(3) was substantively amended in 1961. Before the amendment, it covered “any systematic appeal” to vote or refrain from voting on grounds of caste, race, community, or religion. The 1961 amendment removed the requirement of a “systematic” appeal, added language as a ground, and introduced the pronoun “his.” Sub-section 3A, prohibiting promotion of feelings of enmity or hatred on those grounds, was inserted at the same time.

The dissent treats the legislative history as confirming the narrow reading. The deletion of “systematic” widened the section: a single appeal could now suffice, where before a pattern of conduct had been required. The addition of “his” narrowed the section in a different direction: the appeal had to be on the ground of the candidate’s religion or that of a rival candidate, not on the ground of anyone’s religion. Both moves can be read together, the dissent argues, as Parliament’s considered judgment about how far it wished to extend the corrupt-practice category. The 1961 amendment expanded ambit in one dimension and constrained it in another.

The Law Minister at the time, A.K. Sen, addressed the introduction of “his” in a speech in the Lok Sabha that the dissent reproduces. The dissent treats Sen’s remarks as direct evidence of the parliamentary intention behind the pronoun — evidence not always available in Indian statutory construction, and treated as cautiously admissible under the modern doctrine reflected in Pepper v. Hart and adopted in Indian practice by judgments including Justice J.S. Verma’s opinion in Dr Ramesh Yeshwant Prabhoo v. Prabhakar Kashinath Kunte.⁴

The Select Committee that preceded the 1961 amendment had received two notes of dissent — one from Smt. Renu Chakravartty, one from Shri Balraj Madhok — each disputing aspects of the redrafting. The dissent treats this material as further evidence that Parliament had carefully considered the scope of section 123(3) and had drafted what it drafted. Paragraph 36 reads the legislative history as showing that Parliament, in omitting the requirement of a systematic appeal, deliberately widened the section’s ambit — and was entitled, having widened it, to define its boundaries by introducing the pronoun “his.” The 1961 amendment, on this reading, was a single coherent legislative move that expanded reach in one dimension and constrained it in another.

The precedent chain

The dissent’s fourth movement reviews the Supreme Court’s decisions on section 123(3) over the five decades since the 1961 amendment. The cases include Jagdev Singh Sidhanti (a Constitution Bench reading section 123(3) against Article 29(1)’s linguistic protection), Kultar Singh v. Mukhtiar Singh (a Constitution Bench on the section’s purpose), Ambika Sharan Singh (an appeal on the ground of the candidate’s Rajput identity), Ziyauddin Bukhari v. Brijmohan Ramdas (a three-judge decision on a Muslim candidate attacking a rival Muslim candidate’s religious bona fides), and Dr Ramesh Yeshwant Prabhoo (a 1996 three-judge decision on an appeal made for the candidate’s religion). Across these decisions, the dissent argues, the Court has consistently read “his” as referring to the candidate, not the voter.

The nine-judge decision in S.R. Bommai v. Union of India, which the 2014 reference order identified as containing observations possibly inconsistent with the narrow reading, is addressed at paragraph 43. Bommai was a case on Article 356 and the dismissal of state governments; the meaning of section 123(3) was not directly in issue. Justice P.B. Sawant’s observations on the section, the dissent argues, are obiter and do not have the binding force of a holding. Other judges in Bommai did not address the section. The reference order’s premise — that Bommai had unsettled the prior reading — is therefore, on the dissent’s view, mistaken.

The dissent closes the precedent section with stare decisis. The narrow reading has held the field since the 1961 amendment. Elections have been held; governments have changed; majorities in Parliament have shifted. The provision has remained untouched while others around it have been amended. There is merit in ensuring a continuity of judicial precedent. The interpretation that the Court has earlier placed on section 123(3), the dissent concludes at paragraph 49, “is correct and certainly does not suffer from manifest error. Nor has it been productive of public mischief.” The threshold the Constitution Benches in Keshav Mills and Supreme Court Advocates on Record Association had set for revisiting settled interpretations — manifest error, public mischief, or material change in legal context — was not met.⁵

What the majority and the dissent disagree about, and what they don’t

It is worth marking what the dissent does not say. It does not say that appeals to voters on the basis of their religion are constitutionally protected. It does not say that the legislature could not, by clear language, prohibit such appeals. The position is narrower: that section 123(3) as drafted does not prohibit them, and that the consequences of a finding under the section — disqualification, disenfranchisement — are too serious to permit an expansive construction not supported by the text.

The majority’s position is that section 123(3) is to be read purposively, against the background of secularism as a basic feature of the Constitution recognised in S.R. Bommai. The provision’s object, on the majority’s reading, was to cleanse electoral discourse of appeals founded on identity. Restricting that object to appeals about the candidate’s own identity, while leaving appeals about the voter’s identity unregulated, would be artificial — it would leave the larger evil (the deployment of religious or caste mobilisation in elections) untouched. The majority therefore reads “his” to include the religion of any of the relevant persons: candidate, agent, or voter.

The dissent’s response to the purposive reading is twofold. First, that the purposive interpretation does not necessarily produce the result the majority claims it does — a court can read “his” purposively as carrying the boundary Parliament drew between widening the scope of corrupt practice (the deletion of “systematic”) and constraining it (the addition of “his”). Second, that even if the majority’s purposive reading were one available interpretation, the strict construction of a quasi-criminal statute requires the court to prefer the narrower of two available readings. “It must be left to the legislature to amend or re-draft the legislative provision, if it considers it necessary to do so.”

What the disagreement comes down to, in this respect, is a question about the institutional location of expansion. The majority is willing to expand the reach of section 123(3) by construction, in the service of a purpose the Constitution endorses. The dissent treats that expansion as legislative work, properly Parliament’s to do. The argument is, in part, the recurring argument about the boundary between judicial reading and judicial drafting.

What the dissent preserves

The majority’s judgment is the law. As of January 2017, an appeal to a voter on the basis of the voter’s religion, race, caste, community, or language is a corrupt practice under section 123(3), with the same consequences — six-year disqualification, disenfranchisement, removal from the electoral roll — that have always attached to findings under the section.

What the dissent preserves is twofold. The first is an articulated free-speech reading of the section’s operation. The dissent records, for future benches that may examine the boundary between political discourse and electoral corrupt practice, a textualist position grounded in Article 19(1)(a) and in the constitutional architecture of identity. The position holds that political speech about religion, caste, and language is not merely permitted but constitutionally protected, and that section 123(3) operates at the boundary of that protection rather than as a general sanitisation of identity from electoral life.

The second is an articulated stance on the role of strict construction in election law. The chain of authorities the dissent assembles — Tolaram Relumal, Baldev Singh Mann, and the line of decisions on the quasi-criminal character of election trials — reaches back to the early years of the Constitution. The dissent treats this body of doctrine as continuing to bear on how section 123 should be read. Whether future benches treat it as so binding will depend, in part, on how the boundaries of election law are reshaped by amendment or by further constitutional interpretation. The provision itself remains as the 1961 amendment left it.

One observation is worth recording. Three of the seven judges on the bench in Abhiram Singh read the section narrowly. The majority’s reading prevailed because four judges read it broadly. A different composition of the bench, on the same record, could have produced the opposite result. What the dissent’s position requires future interpreters to engage with is not a marginal view but a textualist reading that very nearly was the law. The line between the two readings runs through a single pronoun. The interpretive choice it presents — between an electoral law that sanitises identity from political discourse and an electoral law that prohibits only the candidate’s deployment of his own identity to solicit votes — is, on the dissent’s framing, a choice about what kind of democracy India’s Constitution contemplates.