Towards the end of Paul Scott's A Division of the Spoils, the final novel of The Raj Quartet, and in the television series as well, Indian people make an appearance and commit acts of unmotivated and horrible violence. The British heroine comments, “Such a damn, bloody, senseless mess … the mess the raj had never been able to sort out.” Making sense, sorting out, was supposed to be the special vocation of British rule, yet here were all the seething, primordial conflicts rising to the surface again in the Hindu versus Muslim partition of India in 1947.

Since the publication of Benedict Anderson's Imagined Communities almost a decade ago, historians and anthropologists have become accustomed to think of the nation—that bedrock of our existence as moderns—as an imagined entity.1 Yet it is problematic, at least to this author, that while the spotlight is (once again) on the nation, the other key actor in Anderson's formulation, the language through which a specific nation is subjectively conceived and imagined, appears to have been relegated to the sidelines. By proposing that languages, too, be considered cultural and symbolic entities that are themselves enrolled into projects of imagination, this essay assumes its beginnings in questions provoked by Anderson's analysis. Because languages themselves are subjected to multiple imaginations, how indeed can they enable the harmonious and contention-free imagination of communities? How indeed do they acquire the capacity to generate feelings of “contemporaneity,” “solidarity,” and “belonging,” as Anderson proposes? Do they not just as well have the power to scatter and disaggregate such sentiments? These are questions occasioned by more than just theoretical curiosity. In many parts of the world, the diverse imaginations that cling to languages have frequently sparked off contestation, divisiveness, and even violence, especially in multi-lingual pluralities like India, the focus of this article.

Culture, one of the keywords of our time, became common, as Raymond Williams has suggested, in Western discourse in the early nineteenth century. Subsequently, pushed by both anthropological and literary-aesthetic studies and extended to global dimensions, the concept of culture, which supposedly expresses primordial naturalness and the irrational, is often played off against its counterpart from the beginning, the calculated mechanicalness of civilization, or the rational. More recently, in the burgeoning field of cultural studies, the boundaries between the two terms have become increasingly blurred. Now civilization, too, is seen as the domain of the irrational, masking itself in socalled rational representation.

When I was a girl, I was taught not to gossip by a school game. We would sit in a circle, and someone would whisper a phrase into the ear of the person sitting next to her. By the time the phrase was returned to the first speaker, it was totally deformed—hilarious proof that hearsay distorted facts. I had already published a book based extensively on oral interviews when I realized how insidious this game was, that it rested on two extremely authoritarian principles: Information should be transmitted passively, and no one has the right to alter or amend received statements.

Although the concept of social banditry has received much critical attention since E. Hobsbawm published his book in 1969, the debate so far has perhaps been narrowly focused (Blok 1972; Vanderwood 1981; Dreissen 1983; Hart 1987; Koliopoulos 1987). Too often that debate has been framed in terms of whether bandits either express pre-political sentiments or prevent the emergence of peasant solidarity, rather than in terms of which types of rural structures encourage its emergence and reproduction, the psychology and sociology of terror, and how bandit myths may be created and used within nation-states or societies aspiring to nationhood.

Revolutionary ideologies of both the left and the right have led to genocide and mass murders of unprecedented scope in this century. The survey by Harff and Gurr shows that genocides and politicides—killings of people in groups targeted because of organized political opposition—between 1945 and 1980 may have caused in this same period over twice as many deaths than wars. Such state killings claimed up to 2.6 times the number of lives lost in the aftermath of natural disasters between 1967 and 1986 (1987). Similarly, Rummel argues that four times as many people have been killed in democides, or state mass murders, of citizens in the twentieth century than have been killed in wars (1992).

Emmanuel LeRoy Ladurie has observed that in medieval Montaillou ”time was always vague,” so that “history was absent or almost absent from Montaillou culture.” According to the inquisition record on which LeRoy Ladurie depended, Montaillou villagers often found it difficult to determine exactly how long ago a given event had occurred. They described past events as having happened “three or four years ago,” “seventeen or eighteen years ago,” “twenty or twenty-four years since.” Others quantified the past by referring to “the time when the heretics predominated in Montaillou, before the round-up by the Inquisition in Carcassonne.” Unsurprisingly, the same imprecision governed their sense of age. In rural Montaillou, the rhythm of transhumance more readily marked time than did any abstract calendar (LeRoy Ladurie 1979:280–1).