Writing with Fragments and Silences: An Ethnographer’s Anxiety and Responsibility

Download PDF: Flueckiger, Writing with Fragments and Silences


* Editors’ Note: This essay is part of the Roundtable on Ethnography and Religion.

Robert Orsi articulates two anxieties that I have often experienced in the writing of ethnography: the pressure to articulate a “certainty of truth,” to neatly tie up frayed ends, in order to formulate theory (see the epigraph of his essay); and the question of what to “do” with unsettling experiences that we don’t fully understand—our own and/or those of our interlocutors on the field—that may undercut that theory or our interpretive frameworks. Most ethnographers of religion on the ground would admit that they often begin to write their ethnographies with lacunae in their “data,” with fragments. These may be created by the serendipity of fieldwork encounters, created consciously by those with whom we work when they leave out what they don’t want us to know or write, by assumptions made by our interlocutors of what they think we do know/understand, the passage of time and vagaries of memory, the context-specificity of conversations or performances, or even illness/madness. How do we write—or should we write at all—with fragments in what we know or what we can ethically publish?


I noticed MR’s frozen smile and manic-like speech (often directed to no one in particular) across a small, congested courtyard where a crowd was gathered to see the south Indian goddess Gangamma in her festival human-guise when she returned from her hours-long perambulations through the heated streets of Tirupati.1 MR caught my eye and walked over to me, identifying herself as someone who was “crazy [picci; lit., mad] for the goddess.” She invited my fieldwork associate and me to come home with her, where, she said, she would show us her story as written down by her daughter.

MR’s manic speech shifted quickly from subject to subject and back again, often with no obvious connections, making it difficult to follow. My fieldwork associate was both greatly annoyed with and a little afraid of MR, not trusting the stated religious nature of her madness, and she insisted that we decline her invitation that day. I found my way to MR’s one-room flat the next day with an older female friend who had more patience and life experience within which to contextualize MR’s picci character. When we got to her flat, MR was conflicted: she had a story, she said, but didn’t have permission from the goddess to tell it. She reiterated:

People call me picci. I’m picci for temples. We all have our picci, and this [my own research on Gangamma] is yours. People around me berate me [saying], “You go around like a mad woman.” There isn’t a single person who hasn’t scolded me…. People don’t pay attention to me because they think I’m picci.

Gradually, over several years, I heard fragments of MR’s story, a story caught between silence and articulation. It’s a heart-breaking story of both empowerment by the goddess, who has seemingly given space for MR’s picci, and the personal toll this relationship with the goddess, often characterized as “too much to bear,” has taken on MR.

I returned to Tirupati several years after my initial meeting with MR to conduct year-long fieldwork on Gangamma narrative and ritual traditions, during which time I met her in her home or at the temples she frequents. Not only did our conversations continue to be fragmented—picci—but I also learned that MR’s living arrangements were also fragmented, as she moved from rental room to rental room, a different one each time I met her. So, when I went to say goodbye to MR before returning to the U.S., I wasn’t surprised when her neighbors told me she had moved once again. They directed us to a room just down the road, but forewarned us that her daughter, herself a Gangamma devotee, had died only days earlier.

When we entered her small room, MR was visibly distraught, her sari and hair in disarray; but she seemed eager to tell us what had happened. Her daughter, she explained—in fits and starts—had moved back to her maternal home after having been thrown out of her in-laws’ home by her father-in-law; he had accused her devotion to Gangamma of interfering with his own ritual magical powers (jadu), powers that MR suggested he used against his enemies. The daughter’s husband had visited her periodically at MR’s home, but he then died under mysterious circumstances. Her daughter had started eating less and less; and then she had stopped eating altogether—literally starving herself to death. MR saw this as an act of courage and devotion, telling us proudly, “Even while we were taking her [to the cremation ground] everyone fell at her feet, declaring her wisdom even at such a young age.”

The small windowless room became oppressive as the story unfolded. My fieldwork associate, whose own sister had died under mysterious circumstances (reported as suicide, but the family thought murder) only a year before, became visibly distraught and kept questioning MR about why she hadn’t force-fed her daughter. (My associate later told me she felt like she was being smothered in that small room, “like something was holding my hand tightly; I couldn’t breathe; I couldn’t think.”) MR countered, “Of course, I tried to feed her. Am I not her mother? But she was very stubborn. She didn’t eat for more than forty days.” My fieldwork associate and I returned to my flat feeling helpless, vulnerable, and agitated—having experienced, to use Ruth Behar’s phrase, “anthropology that breaks your heart.”2

MR’s picci character was identified by others, in part, through her incessant speech; and yet she was unable to speak the full truth of her experience. She both felt compelled to tell the story of her relationship to the goddess and was prohibited by the goddess from fully doing so; she was caught between speech and silence. I became caught in a dilemma similar to that of MR: to tell the story or not, and if so, how to tell it. I knew only fragments of MR’s life, primarily through her own representation and narratives, with some observation of her ritual performances. I wondered about the ethics of publishing fragmented narratives of a vulnerable,picci woman who certainly could not have been aware of its implications on the printed page.

I have decided to publish MR’s heart-breaking, fragmented story, making clear that it is a story whose meaning neither MR herself, according to her own narration, nor I fully understand. (When once I had tried to clarify a particular story, she told me that I hadn’t understood because she herself didn’t understand it.) Without including MR’s narrative, the Gangamma imaginative and social worlds I was writing about would be incomplete—perhaps too optimistic in my focus on the empowering identity between women and the goddess. This conclusion was supported by statements such as that of a female sweeper, who had confidently asserted when I asked whether women were afraid of this goddess on the final day of her festival, when she was at the height of her power,

No, because she [the goddess] is shakti [female power], and we [women] have shakti. You have shakti; I have shakti. But men are different. They don’t have it, so they’re afraid. It’s like we’re talking now; just like two women can talk together. So women aren’t afraid. Men are different, so they’re afraid.

But MR’s story gives us a different perspective—yes, she could bear the goddess, but this relationship had taken a toll. Perhaps because of its picci form and content, MR’s story helps us to begin to understand on an experiential level how or why the goddess Gangamma can be, finally, “too much to bear.”

My unsettling experiences with MR leave us with one kind of fragment—an incomplete and picci narrative. But there is another kind of ethnographic fragment: when what we know, if we publish it, may indirectly negatively impact those with whom we work. This knowledge may be based on observation, on direct conversations with those involved, on conversations about them (hearsay), and/or on our own intuitions based on weaving different fragments of knowledge together. These are often circumstances that an Institutional Review Board could not imagine.

The Gangamma traditions that I have observed and worked with for twenty years are undergoing rapid change under the impact of a rapidly growing Indian middle class. Traditions such as those of the village goddess Gangamma, often associated with lower castes (and in this case, artisan rather than land-owning castes), are being re-formed and over-written by more dominant Sanskritic, upper- and land-owning caste, and middle-class aesthetics, morality, and sensibilities. I experienced a gnawing anxiety when I thought about my ethnography being read in this Indian context, fearing the judgment that may be cast on the actors of the ethnography, and was faced with several decisions about how to write with fragments in order to protect those with whom I worked.

I had attended three annual Gangamma festivals and had already lived in Tirupati for several months at the beginning of my year-long fieldwork before I heard of the tradition of matammas: women who have been offered to the goddess as babies or little girls when they or the village are experiencing goddess-related poxes, rashes, and fevers. When these girls reach puberty, their families celebrate a wedding-like ritual in which the girls and goddess exchange wedding pendants. Thereafter, these women are traditionally free to enter relationships with men with or without marriage. Today, this tradition of offering girls to the goddess has been outlawed, and these women and the children born of their alliances outside of marriage are scorned, socially marginalized, and otherwise at risk. The living tradition has gone underground or has been corralled into the “safe” confines of NGOs working to reform matammas and rehabilitate/educate their children; and thus I did not learn about the tradition until well into my fieldwork.

I sensed a possible connection to Gangamma traditions, but did not know how to go about meeting matammas, thinking it would not be appropriate to attempt to meet them through the NGOs working to eliminate this tradition. But once I learned about the matamma tradition, I began to see and hear multiple indirect traces of it, began to ask Tirupati residents more direct questions about it, and ultimately met two women who had exchanged wedding pendants with the goddess, whose personal narratives became the basis of a chapter of my ethnography. However, they were unusual in that they had married human males, and perhaps this was a reason they were willing to tell me their stories.

The matamma tradition suggested to me an explanation for the matriarchal focus and lack of visibility of a husband in one of the ritual families with whom I have worked over many years. When we first met this family and asked about the husband of the matriarch, a family member told us he was “in the village.” But he never returned for the festivals in which I participated, nor did I hear any further reference to him. A young male cousin at the time was more straightforward about this man: “He left the house and went off. They don’t know if he’s alive or dead.” Because the family itself did not talk openly about this male figure, I assumed they did not want me to know about him and I did not feel free to press them for further details. Only a few days before I left Tirupati at the end of that year, however, while waiting in a temple courtyard for a particular festival ritual to begin, a female cousin of the family told me that there was, in fact, no husband and that the matriarch was a matamma. What was I to do with this hearsay information that suddenly clarified the matriarch’s ritual authority and independence and the silences of the family? The uneasy choice I made was to write in fragments myself and to leave it to observant readers to put the pieces together, providing enough of the cultural and contextual background to enable them to do so.

The caste identity of this particular family, too, was unclear. We were first told by a local anthropologist, who has written a book in Telugu about Gangamma’s festival, that this family belongs to a weavers’ caste. Many years later, an in-marrying male relative told us that what we had been told was a caste name was actually derived from the family’s ritual responsibilities at a large Vishnu temple in the town, and that the family belonged to a dominant local caste much higher on the caste hierarchy than the weavers’ caste is. Just two years ago, in response to my continuing confusion about their caste, the primary male ritual specialist of the family affirmed that they were not weavers but rather belonged to this dominant caste. The different caste identities articulated in different contexts and over a period of twenty years suggest a fluidity of caste identities, particularly under the growing influence of middle-class ideologies in relationship to ritual roles, status, and hierarchies. I tried to write about this fluidity descriptively and indirectly, fearing the possible consequences of fixing a caste identity in time on the marriage possibilities of the younger, English-educated generation of this family, but also desiring to “perform” the fluid dynamics of caste.


Subaltern historians Ranajit Guha and Gyanendra Pandey have cautioned against the construction of a “grand narrative” of history that would ignore fragments that disrupt or make rough an imagined coherency of the Indian nation state or national culture. Pandey distinguishes two uses of the term “fragment”: fragments of society—i.e., the subaltern groups who live on the margins of a “mainstream” (Hindu, Brahminic) Indian culture—and fragmentary expressions that lay outside of “a ritualized account, a collective memory or record that has been generated on behalf of the entire community.”3 Pandey writes:

Part of the importance of the “fragmentary” point of view lays in this, that it resists the drive for a shallow homogenization and struggles for other, potentially richer definitions of the “nation” and the future political community.4

The dominant nationalist historiography … needs to be challenged also because of its privileging of the so-called “general” over the particular, the larger over the smaller, the “mainstream” over the “marginal” …5

Acknowledging the ethnographer’s impulse for narrative coherency—even when focusing on (sometimes “marginal”) individuals and communities—I return to the questions: how can we, or should we, write with ethnographic fragments? What are the consequences of leaving fragments out of our writing, and who and what do we leave out by doing so?

One principle that I have used to make decisions about whether or not to include fragments that I do not totally understand or that may be misunderstood by readers is to consider their importance to the argument of the ethnography. There have been many conversations, interactions, even descriptions of physical space, that I have left out of each of the ethnographies I have written when I felt they may be misunderstood by non-specialist readers and are not central to the issues I am addressing—although they may be for another ethnography. For the fragments I do include that shape, reshape, and disrupt my arguments, I have tried to contextualize them in wider repertoires of rituals, narratives, and conversations that are themselves indigenous commentary on the fragment.

Long-term ethnographic research is another way to address particular fragments the ethnographer encounters, since over time, a fragment may be filled out or woven together with other fragments into a more “coherent” story. However, given the academic pressure to publish, single-project research and writing over a period of many years is not always realistic. So how do we write to account for the fact that what we are representing/describing is itself simply a fragment in time, which may shift over years? One way is to look for and accentuate in our writing those places of flexibility within the traditions we are engaged with, to listen for narratives of change, to notice the margins, and to make clear the fragmental, time/context-specific nature of our data.

The fragments I’ve written about above are faced by all ethnographers. The ethnographer in religious studies, however, may face unique issues when writing about one particular fragment: the deity. The deity is, almost by definition, never totally knowable and, thus, from a human perspective, a fragment—or, in some cases, may be imagined as the whole that encompasses all fragments. If we assume that one of the goals of ethnography is to draw readers into a particular world—in this case a world of religious rituals, narratives, and actors—then, for those traditions that have deities, how do religious studies ethnographers write the deity as an actor/agent in that world? How do we represent a world in which a hotel clerk, knowing I had gone uphill to visit the temple of the god Venkatesvara, asked, “How is god today?”  and in which the deity was imagined as having shifting moods that a worshipper would be able to discern? This has been a matter of debate in religious studies; for two opposing views, see Russell McCutcheon’s argument for religion as social formation and Robert Orsi’s definition of religion as “a network of relationships between heaven and earth involving humans of all ages and many different sacred figures together.”6

One example of the differences between the ways many anthropologists and scholars of religious studies have approached this issue of agency can be found in studies of possession traditions. Anthropologists have tended to focus on the human needs (particularly of disempowered humans) that may be addressed through possession, but they rarely address the agency of the deity him/herself in this process. The ethnographer in religious studies who takes seriously the agency of the deity that is assumed by the communities she studies may include a different set of questions: why does the deity choose to possess certain persons (even if it is the same set of persons anthropologists have identified as often becoming possessed), and what benefits does the deity accrue when s/he enters a human body?

Both approaches to human/divine agency have their place, but in my own work, I have chosen to try to understand and imagine the worlds of my interlocutors in which the deity—known by both his/her worshippers and the ethnographer herself only through fragments—is active. This approach does not preclude analysis and theory, but may be subtly performed through ethnographic writing itself: in part, by giving authority to the words of those with whom we work (including in the ways in which their words are laid out on the printed page in relationship to our own) and through narrative ethnography that leaves space for fragments and ambiguity.

Many of my interlocutors have understood the creative, fluid, and even fragmental space of narrative when they have responded to my rather direct questions with a story that hasn’t always answered the question I meant to ask. When I asked one female singer/storyteller what the connection was between the two-and-a-half hour long song she had just sung and the festival whose story I had asked for, she looked at me quizzically and said, “Do you want me to sing it again?” Her answer to my question had been the performance itself and gave me an important cue in my own analysis of the narrative; a narrative response left room for interpretation and ambiguity that a straightforward discursive answer would not have.

In another context, a Muslim pir answered my question of how he “understood” the Hindu deity Ganesh with the well-known story of the god’s creation. I thought I was asking him whether or not he “believed” in the god whose festival was being celebrated all around him; but he answered my blunt question much more subtly, rather than with an assertion of theology. His narrative response suggested a shared narrative repertoire and worldview across Hindu-Muslim identities that he may have been unable or unwilling to speak directly. These instances suggest the usefulness of an ethnographic narrative style that leaves the reader with imaginative possibilities, a genre of writing that leaves room for multiple interpretations (those of our interlocutors and our own analytic ones) and encourages active reader engagement. Finally, while an ethnographer needs to draw theoretical conclusions that may be relevant beyond the immediate contexts in which she works, the fragment suggests that we write with humility and awareness of the subjunctive nature of our conclusions.

Feature photo by Joyce Flueckiger



  1. This ethnographic description is drawn from a chapter in my forthcoming bookWhen the World Becomes Female: Guises of a South Indian Goddess(Bloomington: Indiana University Press, 2013).
  2. Behar, Ruth,The Vulnerable Observer: Anthropology That Breaks Your Heart (Boston: Beacon Press, 1996).
  3. Pandey, Gyanendra. “In Defense of the Fragment: Writing about Hindu-Muslim Riots in India Today.”Representations 37, (1992). 12.
  4. Ibid., 3.
  5. Ibid., 29.
  6. Russell McCutcheon,Critics Not Caretakers: Redescribing the Public Study of Religion (Albany: SUNY Press, 2001); Robert Orsi, Between Heaven and Earth: The Religious Worlds People Make and the Scholars Who Study Them (Princeton University Press, 2005), 2.


By Joyce Burkhalter Flueckiger
Joyce Burkhalter Flueckiger is Professor of Religion at Emory University.