Communal and Caste-Based Conflict and Violence

Most criminology treats the individual as the unit of offending. Communal and caste violence forces a different starting point: the offender acts as a member of a group, against a victim chosen because of their membership in another group, and the violence is legitimated within the offender's own group as defence, retaliation or restoration of honour rather than as ordinary crime.

This is the in-group out-group structure that social psychologists trace back to Henri Tajfel and John Turner's social identity theory of the 1970s and 1980s: once a boundary between us and them is activated, in-group loyalty and out-group hostility rise together, and ordinary moral restraints on harming a stranger weaken when the stranger is coded as belonging to the rival group.

Religion and caste are two of the sharpest identity boundaries available in South Asia because both are ascribed at birth, visible through name, dress, locality or occupation, and backed by dense networks of kin and community that can be mobilised quickly.

A land dispute between two individuals of different castes, or a personal quarrel between neighbours of different faiths, can therefore escalate far beyond the original parties because each side experiences an attack on one of its own as an attack on the whole community, inviting a collective response rather than an individual one.

This group logic changes what a criminological account has to explain. It is not enough to ask why a particular person picked up a weapon; the more useful questions are who benefits from converting a local dispute into a group conflict, which actors have the capacity to mobilise a crowd on short notice, and why the state's ordinary deterrent and protective functions so often fail at exactly the moment they are most needed.

Those questions carry through the rest of this topic, from the riot as an event, to caste atrocity as a distinct legal category, to the state's record of intervention.

A communal riot is rarely a spontaneous explosion of pent up hostility. Empirical work on riots in Indian towns, most influentially the political scientist Paul Brass's study of Uttar Pradesh published in his 2003 book The Production of Hindu-Muslim Violence in Contemporary India, argues that many riot-prone towns sustain what he called an institutionalised riot system.

In such a town a standing network of local actors, small-time politicians, criminal operators, and sometimes sympathetic police or administrative officials, has both the motive and the capacity to convert a minor, often trivial, incident into a full riot whenever it becomes politically useful to do so.

Brass describes the sequence as three recognisable phases rather than a single trigger. A preparation phase is continuous in riot-prone towns: local actors keep communal tension alive through minor inflammatory acts long before any single incident sets off violence, so a precipitating incident, a scuffle, an accident, an argument over a loudspeaker or a procession route, lands on ground already prepared for escalation rather than causing the riot by itself.

An activation phase follows, often timed to political mobilisation or an election, in which organised actors, sometimes linked to political formations, bring prepared crowds onto the street, and it is at this stage that the character of the violence shifts from chaotic to targeted, with attackers able to identify which shops, homes or persons belong to the rival community.

An explanation phase follows the violence itself, in which politicians, media and sometimes commentators contest how its causes are described, often presenting it as spontaneous and unpredictable in a way that diffuses responsibility away from those who organised it.

The institutionalised riot system model matters because it shifts the explanatory weight away from ancient hatred, a framing that treats communal violence as an inevitable eruption of deep-seated animosity, and towards production, the idea that riots require organisation, resources and often political incentive to occur, and can therefore in principle be prevented by removing that capacity rather than by managing sentiment alone.

Rumour control and early, visible administrative presence at the precipitating incident stage are the two interventions most consistently associated with a riot not occurring, which is a testable, falsifiable claim rather than a general statement about communal harmony.

Caste based violence is treated in Indian criminal law as a distinct category rather than as an ordinary assault, theft or homicide charged under the general penal code. Other jurisdictions reach the same problem through hate crime law, which treats an offender's selection of a victim for their group membership as an aggravating fact and which criminology analyses through the Levin and McDevitt typology of bias offenders, sorting offenders into thrill, defensive, retaliatory and mission types.

The Scheduled Castes and the Scheduled Tribes (Prevention of Atrocities) Act, 1989 was enacted after Parliament concluded that the existing law was not deterring violence against Dalits and Adivasis, and that ordinary criminal procedure, run through the same local police and courts that were often themselves complicit in caste hierarchy, was failing these communities specifically.

The Act lists a defined set of atrocities, including forced consumption of noxious substances, sexual exploitation, dispossession of land, forced labour, and social boycott, that attract enhanced punishment when committed against a member of a Scheduled Caste or Scheduled Tribe because of that status, and it creates special courts intended to try these cases more quickly than the general system allows.

The Act has been amended twice in response to gaps exposed by litigation and enforcement experience. The Scheduled Castes and Scheduled Tribes (Prevention of Atrocities) Amendment Act, 2015, notified in force from January 2016, expanded the list of offences and clarified procedural duties on police and public servants.

A further amendment in 2018 responded to a Supreme Court ruling that had introduced a preliminary inquiry requirement and an approval step before arrest, provisions the amendment restored to their pre-ruling position after Parliament judged that the added procedural steps were being used to delay action rather than to prevent misuse; the Supreme Court upheld the 2018 amendment's validity in 2020.

The rationale for a separate statute, rather than relying on the general law of assault or murder, is that caste atrocity is understood as an offence with an identity motive built into its definition, similar in structure to hate crime legislation in other jurisdictions, and that ordinary criminal law does not register or punish that motive as an aggravating element on its own.

Whether the Act has closed the enforcement gap it was designed for is contested, and conviction rates under it remain a subject of continuing debate among legal scholars and human rights researchers, a point taken up further when this topic turns to state response.

Nepal took a comparable legislative route for the same underlying problem. Its Caste-Based Discrimination and Untouchability (Offence and Punishment) Act, 2011 prohibits caste-based discrimination and untouchability in both public and private life, sets enhanced punishment for a public official found responsible, and allows a victim to claim compensation from the perpetrator, the same basic choice India made in 1989 to treat caste harm as a distinct statutory offence rather than leave it to the ordinary law of assault or defamation.

Caste atrocity cases cluster around a recognisable set of triggers rather than occurring at random across social life, and identifying the pattern is useful both for prevention and for understanding why courts and commentators keep returning to the same categories of dispute.

Land is the most consistently documented trigger: disputes over agricultural land allotted to Dalit families under land reform or welfare schemes, over grazing rights, or over encroachment by dominant caste landholders, recur across decades of reported cases because land ownership in much of rural India remains closely tied to caste status, and a Dalit family asserting a legal land right is frequently read by dominant caste neighbours as an assertion against the existing social order itself, not merely a property claim.

Honour is a second recurring trigger, most visibly in cases of inter-caste marriage or relationships that cross caste lines without family or community sanction. Violence in these cases, sometimes organised through informal caste or community bodies acting outside the formal legal system, is directed at punishing the couple or their families for breaching an endogamy norm, and the harshest cases have involved killings that legal and media commentary in India refers to as honour killings, a term that describes the stated justification offered by perpetrators rather than a legal defence recognised by any Indian court.

A third recurring trigger is access to public space and public resources on equal terms: entry to temples, access to a shared water source, the right to draw water from a common well, use of a common cremation ground, or simply walking through a locality on equal footing, particularly at the time of a local festival or procession.

Because these are precisely the forms of social equality that the constitutional abolition of untouchability under Article 17 was intended to guarantee, violence triggered by an assertion of this kind is frequently read by both victims and courts as a direct challenge to caste hierarchy rather than an incidental dispute, which is part of why the Prevention of Atrocities Act frames these acts as a distinct wrong.

State response to both communal riots and caste atrocities has a documented pattern of failure that recurs across decades of official inquiry. Police complicity or inaction is the most frequently cited failure: officers drawn from the locally dominant community have, in numerous documented riots and atrocity cases, delayed responding to distress calls, refused to register a first information report promptly, or in the worst documented instances participated in or facilitated violence rather than preventing it.

Delayed intervention compounds this problem even where outright complicity is absent, because the institutionalised riot system model shows that a riot's window for containment is narrow, concentrated in the rumour and early mobilisation stages, and a police force that arrives only after violence has peaked has missed the point at which intervention was effective.

India's standard institutional response after a major riot has been to appoint a commission of inquiry under the Commissions of Inquiry Act, 1952, to establish facts and assign administrative or political responsibility.

Commissions following major riots, including the inquiry into the 1984 anti-Sikh violence in Delhi and the inquiry into the 2002 Gujarat riots, produced detailed factual records, but critics, including legal scholars and human rights organisations, have long pointed out that a commission's findings are recommendatory rather than binding, that criminal prosecutions arising from these events have frequently taken decades to conclude, and that conviction rates in riot related prosecutions remain markedly lower than for comparable violent crime, a gap frequently attributed to witness intimidation, delayed or compromised investigation, and the practical difficulty of proving individual culpability within a crowd.

The same pattern of delayed and partial accountability appears in caste atrocity enforcement. Government crime data compiled annually by India's National Crime Records Bureau has repeatedly shown conviction rates under the Prevention of Atrocities Act well below the conviction rate for the penal code generally, and researchers attribute the gap to some of the same causes documented in riot cases, under-registration by police reluctant to record a case under the Act, weak investigation, and pressure on witnesses and victims to withdraw complaints, particularly where the victim and accused continue to live in the same village after the case is filed.

Kenya's response to the violence that followed its disputed December 2007 election shows the same commission of inquiry pattern outside South Asia. The Commission of Inquiry into Post-Election Violence, chaired by Justice Philip Waki and known as the Waki Commission, documented roughly 1,500 deaths in its October 2008 report and named suspects for prosecution, but its recommendation for a local tribunal was never implemented and the matter went to the International Criminal Court instead.

The resulting cases against Deputy President William Ruto and President Uhuru Kenyatta were vacated in 2015 and 2016 after numerous witnesses withdrew or recanted, citing threats and intimidation, the same failure mode documented in Indian riot and atrocity prosecutions.

Placing India's experience alongside sectarian and ethnic conflict elsewhere clarifies what is general to group conflict and what is specific to the Indian case. Northern Ireland's Troubles, a sectarian conflict between the mostly Protestant unionist community and the mostly Catholic nationalist community, ran from roughly 1968 to 1998 and killed approximately 3,600 people.

The conflict shared the institutionalised structure Brass documents for India: paramilitary organisations on both sides, the Provisional Irish Republican Army and loyalist paramilitary groups, sustained recruitment, funding and targeting infrastructure over decades, converting what began as a civil rights dispute into a durable armed conflict.

It ended not through a single security victory but through a negotiated political settlement, the Good Friday Agreement of 1998, which created power sharing institutions and cross border cooperation mechanisms, an outcome that underlines how political accommodation, not policing alone, resolved a conflict that policing alone had failed to end for thirty years.

The wars that followed the breakup of Yugoslavia in the 1990s show a more extreme version of the same group logic operating along ethnic and religious lines.

In Bosnia and Herzegovina, the July 1995 massacre of more than eight thousand Bosniak men and boys at Srebrenica by Bosnian Serb forces, later recognised by international courts as genocide, illustrates how an institutionalised capacity for organised violence, backed by a state or paramilitary structure, can escalate group conflict to a scale far beyond anything documented in Indian riots.

The war ended with the Dayton Agreement of December 1995, which divided the country into two largely autonomous entities along the ethnic lines the conflict had hardened, a settlement that stopped the killing but is widely assessed by researchers as having frozen ethnic division into the country's constitutional structure rather than resolving it.

Read together, the comparative cases point to three prevention approaches that recur across contexts regardless of which identity marker is at stake. Early-warning systems that track rumour, incendiary local speeches, and unusual crowd gathering before mobilisation are strongest where local police maintain routine contact with community leaders rather than intervening only after an incident occurs.

Peace committees composed of respected leaders from both sides of a local divide, used in several Indian cities since the 1980s and structurally similar to community relations bodies used in Northern Ireland after 1998, work best when they have a standing relationship with the local administration rather than being convened only in a crisis.

Confidence-building measures, joint festivals, shared civic infrastructure, transparent policing, address the underlying in-group out-group dynamic directly rather than only managing its symptoms, and researchers studying both the Indian and Northern Irish cases agree that sustained contact between groups under conditions of equal status is the single most consistently supported intervention across the literature on reducing intergroup hostility.