Affective temporalities of presence and absence: musical haunting and embodied political histories in an Algerian religious community

ABSTRACT What does it mean to be musically haunted? In the Algerian popular Islamic ritual called dīwān, one can be haunted by the deep, bass-register melodies of spirits, saints and historical figures of the trans-Saharan slave trade. Musical haunting is affective haunting. Melodies are not only felt emotionally as recurrent fear, dread and ambiguous loss (Boss 1999. Ambiguous Loss: Learning to Live with Unresolved Grief. Cambridge and London: Harvard University Press.) but they are also simultaneously physically arresting for the body and senses, erupting into uncomfortable sensations like prickling skin and knots in the stomach, eventually precipitating into registers of trance. Here, musical affects manifest spectrally – both directly as non-human entities or spirits and indirectly through strong emotions that tend to ‘take over’. The haunted are never eventually ‘healed’ in ritual, in the sense of completion; suffering always comes back in some form. Rather, dīwān is a modality of continually inhabiting and embodying various tumultuous, political histories perpetually resounding through the daily lives and physical bodies of the dīwān community. By way of non-Western understandings of affect, music, and ritualised temporality, this essay illustrates an intertwined, spectral interdependency of music, affect and politics.


Introduction
Sergu comes in black. He is most commonly understood as a sub-Saharan entity associated with wildness, danger and the unknown. He appears draped in a black satin cloak, carrying a long, pointed spear and his head is wrapped in a turban so that only his eyes peer out from bundles of cloth. Many understand him to embody the Tuareg, a desert ethnolinguistic group who trafficked slaves across the Sahara Desert. It is probably for this reason that, in ritual, he is depicted as a hunter: his adepts take up spears and move them in stabbing motions. He seems to be hunting humans.
What does it mean to be musically haunted?
In the Algerian popular Islamic ritual called dīwān, one can be haunted by the deep, bass-register melodies of historical figures, saints, prophets, spirits and spectres of the trans-Saharan slave trade. 1 One can also be haunted by nondescript, atemporal and painful feelings that may or may not have a connection to personal histories or identifiable events. Nevertheless, like spectres, these feelings circulate in public, can invade a person and revisit over and over again. All of these visitations are realised by music through the recurrent, ritual performance of specific, categorised melodies that manifest the spectral. 2 All musical hauntings are affective phenomena. My use of 'affect' represents a broad notion of feeling (ih  sās) in this Arabophone context, including the senses, felt energies, atmospheres, and emotions. Affect as an umbrella category is useful here because affective dynamics subsume all other modes of feeling, including emotion. While emotions invariably implicate affect (e.g., fear with bodily clenching), not all affects are labelled 'emotional'. Moreover, 'emotions' in dīwān are not necessarily divergent from affect in terms of involving 'interior', biographical and personal dimensions, nor are they always 'recognisable as one's own' (e.g., Massumi, see Reisnour and Desai-Stephens 2020). Rather, in dīwān, emotions are consistently described as intersubjective, felt as circulating energies in the air: qualities often considered particular to affect. Indeed, discrete affects (e.g., chills or paralysis) in dīwān, contrary to some Deleuzian theorisations, are not autonomous (Massumi 1995), nor do they precede signification or representation (Thrift 2008) or 'follow a different logic' than emotion (Massumi 2002: 27). Rather, dīwān's affective dynamics, in their entirety as felt wholes, are predominantly mediated (see Mazzarella 2009), purposefully cultivated and biographically meaningful.
While it can be useful to specify phenomenological differences between emotion and affect, such as what we index as 'anger' versus 'bodily heat', felt modes in dīwān are so deeply entangled and co-terminous that parsing out boundaries can be more confusing than clarifying. As pointed out in the introduction to this special issue, affect scholars have progressively critiqued sharp distinctions between affect and emotion (Desai-Stephens and Reisnour 2020). Likewise, Cassaniti points out that overlapping feeling categories are not new to either scholars of affect or anthropologists of emotion. Rather than debate semantics, the important issue is rather 'about what kinds of qualities are attended to in seeking to understand experience ' (2015: 139). In musical haunting, the deep, bass tones of melodies act as sonic forces, physically impacting bodies and senses, often erupting into uncomfortable sensations like prickling skin, knots in the stomach and temporary paralysis. Haunting melodies are also felt as fear, dread, grief, or a kind of subjectless suffering. Thus, a spectre is not just marked by his or her melody and relationship to the past (e.g., historical figure or ancestor) but by the patterned feelings that a spectre generates in the haunted.
The diverse pantheons of spectres who appear in dīwān and haunt the living were generated from and are still precipitated by multiple historical periods of violence, 1 For Arabic transliteration, I use the system of the International Journal of Middle Eastern Studies with some adaptions for Algerian dialect. 2 Fieldwork was based in Oran, Algeria and the western corridor cities south to Béchar, supported with regular travel across Algeria to study other dīwān communities, including those in the Mzab valley, Algiers, and the east. Fieldwork spanned 18 months total between 2013, [2014][2015]2016. Data gathered included participant-observation and attendance of dīwān rituals or other dīwān spectacles (stages, weddings, festivals), as well as ongoing conversations, interviews with musicians, ritual experts, ritual attendees, and friends. All interviews were conducted by myself primarily in Algerian Arabic and were transcribed and translated primarily by myself, sometimes with the assistance of Algerian friends.
upheaval and suffering. Thus, coinciding with the existential reach of hauntology, dīwān rituals attend to the numerous psychic and affective ways that the impacts of history and geopolitics are never entirely 'in the past'. Concerning this issue's thematic nexus of music, affect and politics, this essay demonstrates how intersubjective musical phenomena grip bodies with political phantoms, thus setting 'the political at the heart of the psychological' (Good et al. 2008). Indeed, hauntology has always paired the psychological and political: it was conceptualised as political from the beginning. The very notion of it came out of Derrida's efforts in Specters of Marx (1994) by first examining ways that a particular political world viewthat of communismhad or had not survived. From this starting point, Derrida engaged with the temporal instability of memory, politics and transmission across generations. Similarly, while dīwān has largely been absent from Algerian studies or trans-Saharan history (except Dermenghem 1954;Lapassade 1982;Pâques 1964;Turner 2017Turner , 2020aTurner , 2020b, the history of dīwān itself is one of rupture, displacement and loss: dīwān coalesced from centuries of the trans-Saharan slave trade and this history has always enfolded complications of memory and forgetting, presence and absence. Alongside trans-Saharan trafficking, multiple political histories haunt dīwān. Given that ritual insiders, ūlād dīwān (literally 'children of dīwān') are primarily black Algerians, dīwān stands out as a racially marked practice in a region where racism is alive and well, bringing various levels of stigma and prejudice from outsiders. Moreover, while dīwān's iteration of Sufism belongs to a wider, political context of Sufi family groups (locally considered 'popular Islam' or 'folk Islam'), Sufism in the region as a whole has always been wrapped up in both state politics and internal politics (Werenfels 2014;Willis 2012). 3 Many active Sufi orders and their practices in Algeria occupy a politically unstable position; Islam itself is understood and experienced as a political field as much as a religious one.
Finally, colonial and post-colonial histories still haunt Algerians (cf. Lazali 2018), including ūlād dīwān. From the revolutionary war against the French and upheaval after independence to the Civil War in the 1990s, disruption and violence of Algeria's shifting politics directly impacted dīwān communities. Many of my interlocutors spoke of the displacement of their communities after Independence, upsetting their neighbourhoods, rights and roles in society. Others recounted the terror of the 1990s when dīwān musicians were sometimes targeted and killed by extremists for their 'heterodox' practices. With these multiple histories in mind, echoes of loss and rupture reverberate through the political backdrop in which marginalised dīwān communities laboured to live. Affective accumulations press into the present through spectres, both as supernatural entities and as hauntings of painful feelings. They are made physical and rendered audible by music, by musical hauntings that never go away.

Why hauntology?
Haunted places are the only ones people can live in. (Michel de Certeau, The Practice of Everyday Life, 1984: 108) In French, 'ontology' and 'hauntology' are antithetical homonyms: antithetical because the empirical certainty implied by uttering 'ontology' is instantly undone by sounding like 'hauntology' which acutely deconstructs certainty. Derrida's jeu de mots expressly enacts the overall deconstructive project of hauntology. As well ontological certainty, haunting deconstructs the seemingly oppositional existential binaries of time/space, material/immaterial and past/present. It brings to the fore the inexplicable and bewildering ways that the past can impose on the 'present' and, in the world of politics, that events like historical ruptures, traumas, wars and other tragedies reverberate and resonate in ways that might escape language, writing, or linear time: this is precisely the case in dīwān.
Indeed, hauntology has accelerated into an especially productive framework in postcolonial or psychological anthropology (Csordas 2019;Good et al. 2008Good et al. , 2012bHollan 2014Hollan , 2019Pinto 2018;Rahimi 2019 and others), 'as a language of human subjectivity in time' (Rahimi & Good 2019: 410). Relevant to this case study, haunting is so appealing and productive for the ways that it 'seems to subvert rationality … and demands from us an admission of our inability to know and understand everything'; ultimately, it 'goes beyond the scope of verifiable reality, scientific certainty, definite and definable experience' (Lorek-Jezińska and Więckowska 2017: 9-11).
The last point is critical because dīwān involves a wide range of nuanced, ephemeral and supernatural phenomena, such as spirits interacting with humans, that are difficult if not impossible to describe, narrate, or explain. Historically, such phenomenaparticularly supernatural oneshave been especially perplexing to academic inquiry that typically subscribes to secular humanist and positivist paradigms. Unwilling or unable to entertain occult or supernatural ontologies, Western analyses have often rendered them as semioticplaceholders for 'something else really going on'through demystified explanatory models such as non-discursive resistance or, at worst, pathological delusion. Such a crisis of conceptual translation and ethical representation is precisely where hauntology can help: its own cognitive crisis already questioned 'reason, systems, structures, and meanings' and, thus, 'the very ability to fully understand reality' (Lorek-Jezińska and Więckowska 2017: 9-10). 4 Likewise, this essay presupposes the suspension of judgement toward the supposed 'real/unreal', particularly when describing the experience of others. Attuning to the postcolonial affordances of hauntology, subverting the still-haunting-us colonial dichotomy of 'their beliefs' and 'our knowledge' (Good 2012a: 516), I engage hauntological possibilities with dīwān dynamics in order bypass 'the twin pitfalls of subjectivism and positivism' (Gordon 1997: 18). That is, to open up an epistemological space in order to engage more generously with certain inscrutable phenomena that are critical to human experience.
With these concerns in mind, this essay explicitly intertwines non-Western and culturally specific notions of affect with its respective hauntological modes. Regarding the latter, Hollan (2019) has pointed out that much hauntological scholarship has neglected or confounded divergences between literal and metaphorical hauntings, what I delineate below as 'direct' and 'indirect' hauntings. Dīwān involves both of these forms, sometimes simultaneously, through the categorical ambiguity of the ritual pantheon. Indeed, by mapping hauntological theory via affective dynamics, hauntology's spatio-temporal possibilities remarkably align with the affordances and priorities of ritual. Open temporalities of ancestors and descendants conjured through song parallel hauntology's affectively impactful challenge to closed, linear time.

Political affects and the background of dīwān
At least five songs in the dīwān ritual conjure Sergu. From the first few notes of his melody, trancers hit themselves, run out of the ritual, fall to the ground, begin moaning, writhing, or become paralyzed. Hasna, one of his victims in her mid-forties, tells me that his songs are some of the 'heavier' ones and she describes him as 'cruel'. He has tormented her as long as she can remember. Sergu, a hunter of humans, struck me as the quintessential haunting of the trans-Saharan slave trade and, thus, one of the most prominent spectres haunting dīwān.
Dīwān (lit. 'assembly') is a nocturnal music and trance ritual belonging to one of the many popular Sufi orders in North Africa, the Bilāliyya. 5 Over centuries and through various periods in history, numerous sub-Saharan ethnolinguistic groups were forcibly taken across the Sahara through complex slave trading networks. Over the centuries of the trade, sub-Saharan Africans brought their spirit pantheons and practices with them which gradually coalesced into local esoteric practices, whether Sufi or otherwise. While some sub-Saharan populations were Muslim before being traded as slaves, others converted en route or after their arrival in North Africa. 6 In brief, we can say that the historical emergence of dīwān, along with its better known sister traditions, the Moroccan gnāwa and Tunisian st ambēlī, is part of the gradual instantiation of 'black popular Islam' in North Africa; that is, a racially marked, stratified assemblage of esoteric and religious practices. Because dīwān evolved via ancestral lineages, rituals are still mostly attended by born-in-the-tradition insiders, joined by extended kin, close friends and fans or connoisseurs (muh  ebbīn). 7 Dīwān rituals manifest a rich and varied pantheon of spirits, saints and other supernatural entities from sub-Saharan pantheons (e.g., Hausa, Songhay) that were long ago incorporated into the Algerian context but are no longer understood or remembered, even within dīwān communities. Accordingly, ūlād dīwān disagree about what kind of supernatural entities populate rituals and which ones exist or not, if any exist at all. Social attitudes toward The Unseen (al-ghayb) are complex and rife with tension. Nevertheless, we can speak of two categories of spectres in dīwān: spectres who were once human (a saint or historical figure) and spectres who were never human (sub-Saharan spirits or jnūn). 8 'Once human' spectres manifest indirect hauntings; they haunt via the strong and sometimes overwhelming emotions associated with them. That is, they are 'emotional' hauntings. For example, empathy as felt, emotional resonance with the suffering and martyrdom of ʿAlī, the son-in-law of the Prophet, can expand into haunting. But in cases like this, it is the emotional connection that produces haunting, not ʿAlī himself. Similarly, a reputable ritual musician in Oran, Houari, spoke to me of seeing visions during rituals of his loved ones and teachers who have passed on. Others reported similar visual and auditory phenomena, saying that these instances bring up profound emotions that precipitate trance.
To be clear, haunting by once-human spectres does not mean that the deceased person revisits as a ghost. In fact, while local beliefs vary, official discourse in Islam asserts that after death, a human soul cannot be seen or heard by living humans. Thus, in cases of 'once-human' spectres, it is feeling that does the work of haunting. In contrast, a 'nonhuman' spectre such as a jinn or spirit is understood to actually be present as energy and haunt humans. They can be thought of as 'direct' hauntings in comparison to 'indirect' emotional haunting. Direct hauntings by non-human spectres generally involve more intense, painful affective states (paralysis, seizures) and sometimes relentless physical and emotional suffering.
All of these affects roughly map onto varying modalities of consciousness, including what is locally considered 'emotional trance' ( jedba) and 'inhabitation' (bori) or 'spirit possession' trance by non-human entities. While it may seem that the latter are more directly tied to a typical idea of haunting, inhabitation is only a single node within a dynamic spectrum of trance modalities. 9 In terms of symptoms and treatment of various 'causes' of trance (a spirit or overwhelming emotions), little difference is made between categories of saints, spirits and jnūn. In addition, modes of consciousness in dīwān ebb and flow into one another and can be difficult to parse out even amongst ritual experts. In kind, I am not restricting my exploration of musical haunting to the 'direct' or literal haunting of 'inhabitation trance'. Both human and non-human specters, at the level of felt experience, haunt in imbricated ways.
The haunting presence (whether emotionally-charged memory or a jinn manifesting in ritual) always brings with it unpleasant affects (chills, paralysis, physical pain, clenching), memories and corporal patterns of movement, putting the haunted person in a vulnerable position and demanding bodily engagement with the feelings (see Turner 2020a), whatever their 'cause'. Secondly, all of these spectre typeswhether human or non-humaninvariably 'haunt' the living as musical modalities that resist closure. Whether a certain spirit is present or absent, such ritual pantheons 'exist only through their musical performance' (Jankowsky 2010: 69, emphasis mine). Quite importantly, it is not only the literal or direct hauntings of spirits that involve a 'musical motto', as it has classically been theorised in music and trance studies (e.g., Rouget 1985); emotional, indirect hauntings (e.g., jedba) are also dependent on melodies, on the performance, hearing of, and bodily engagement with the song.
Just like hauntological structures, dīwān rituals are conducted over and over, repeated throughout a person's lifetime. The repetition of pain is the rehearsal of possibilities, of possible ways one can relate to pain and suffering of the past and presentagain, whatever its 'cause' may be. The repetition, the chronic nature of ritualised trance dancing as a response to haunting, is itself the therapy: the never-ending process of relating to haunting and engaging with it as an essential part of what it means to be human in a musical world densely populated with ancestors, prophets, saints and spirits.
Haunting in dīwān is a valued experience. It does things: helpful, hurtful, therapeutic and painful things. These are all critical ways of being in and negotiating the world. The music that scaffolds dīwān ritual is not a means of 'resounding transcendence', a crucial phenomenon undergirding other traditions (Engelhardt & Bohlman 2016). Rather, as Derrida noted, 'a spectre is always a revenant' (Derrida 1994: 11) coming again and again. It cannot be transcended. This unrelenting nature is also the nature of dīwān; it, too, has no cure. It is that which will never go away. It is not meant to.

Ritual makeup
Mūsa comes in green or deep blue. His songs constitute the suite Mūsawiyyn that is arguably the most important in a larger grouping of several suites known as 'Bah  āra' (ocean, sea) referencing the water element. Munir, haunted by this series, makes fluid swimming motions during his trance. For some adepts, Mūsa is the Jewish prophet Moses whose parting of the Red Sea links him to water. But for others, Mūsa is fundamentally 'of the water' as some order of water spirit.
To be haunted by a spectre is to be haunted by a song. Ontologically speakingand hauntologically as wellthe song is not a representation of a spectre (such as Mūsa) nor is it symbolic of its namesake but, rather, the sonic materiality of the songthe vibration of sound wavesis experienced as the very vibration of the spectre. Songs are made up of the same 'material' as spirits, spectres and feelings: immersive, vibrating energy.
Whether or not a song vibrates a once-human or never-human spectre, all dīwān songs feature a recognisable melodic theme that is understood to be the primary carrier of haunting. Like the most insidious, oppressive earworm, melodies haunt not only when they are played in ritual but outside of ritual, when they are heard in one's 'head' or felt wriggling somewhere in the body. They ebb and flow at their own will. They become physically entangled and weigh heavily as a perpetually lingering sonic shadow. Adepts spoke of intense urgency to hear a song and dance to it. Haunting melodies would not abate otherwise and adepts might grow increasingly sick (mrīd  ). Ritual musicians described energies or spirits coming into their hands, so that the notes 'come on their own' (yjīū wah  edhūm) or that the spirits themselves 'took over' the songs.
Such ontological collapsehuman and non-human spirit, song and spectreis common in dīwān. Central to the notion of haunting as well, there is no need for something to be this or that, just as there is no need for something to be here or there, then or now. Mūsa is particularly interesting here: no clear difference is made between the personage of Mūsa as the prophet Moses and Mūsa the water spirit. He is experienced as both in ritual, sometimes simultaneously by the same people. Indeed, within many of the once-human and never-human pantheon categories, ontological categories overlap within songs. Another suite of songs, Brahīm (Abraham), invokes the father of Abrahamic religions in one moment and, in another moment, a song for a jinn of the same name. Thus, both direct and indirect, metaphorical and literal hauntings intertwine. Sergu also slips between the categories of the Tuareg ethnolinguistic group, and something spirit or jinn-like; in Algiers, he is categorised within a 'pagan' group of sub-Saharan spirits associated with Hausa pantheons (Khiat 2014;Turner 2020b) although he also turns up as the Tuareg 'Mahama Surgu' in Songhay pantheons (Rouch 1989). Today, his dīwān personage enfolds a nebulous assemblage of myth, fractured memory and energy. 10 Like with Hasna, he haunts with his melody and it is understood that he is his melody; he emerges and is brought forth, revenant through and with his melody.
These melodies vibrate and spiral outwards from the main ritual musical instrument, the ginbrī, a three-string, spiked lute. The instrument is said to be alive, made of living materialswood, camel skin, sheep intestine stringscharged with life energy. In dīwān, a single, amplified ginbrī is played by the mʿallem (master), the ritual expert of the musical dimensions of ritual while double-headed metal clappers help support the driving metric cycles of the ginbrī. Sitting next to the mʿallem, the main singer leads the call section of the call-and-response singing and directs the members of the chorus who flank the side of the mʿallem, play the clappers and belt out response phrases.
A robust and consistent musical repertoire structures the ritual, lasting anywhere from six to nine hours. The repertoire's eighty or more songs constitute at least ten song suites, divided by the hagiographic ranking of the song's namesake (such as prophets and saints), by historical figures, by ethnolinguistic origin of the song ('Hausa' songs from sub-Saharan Africa) and by various spirit pantheons inherited from sub-Saharan African ancestors. Each song suite uniquely engages the 'tactile-kinesthetic-affective body' (Sheets-Johnstone 2018: 3), including an associated colour, appearing mostly in the clothing used during that song, one or more ritual objects that are held in the hands of dancers, a food that must be ingested during the songs and always specific bodily gestures that indicate the type of personage of the song(s). Sergu, like other spectres, has aesthetic needs, quintessential gestures and specific haunting demands such as coming in black, holding long spears and trance-dancing with the spears. The laboured, slow movements and gestures of the saint ʿAbd el-Qādr Jīlānīintended to depict and an old man with a staffare distinct from those of other religious figures such as ʿAlī (cousin and son-in-law of the Prophet) who bears a sword and rides his horse into battle.
Whether human or nonhuman spectre, this 'favourite song' generates an affective response intense enough to spill over into trance. 11 Throughout the night, as the musical repertoire unfolds, the mʿallem (lead musician) watches certain people whom he knows to be haunted by particular spectres. He prioritises experienced trancers, those who have been haunted for decades. His musical working of the affects follows the needs of those trancers: more of a certain phrase, more qrāqeb (metal clappers) or less, or a quicker tempo.
Most dīwān trance varieties are understood as precipitating from the musical ignition of pain and suffering ('ghabina'), the way song vibrates and brings into the moment memories and affects, sensations and emotions associated with painful feelings. In other words, this is not ecstasy or rapture but the musical ignition of burdensome affects. When Yusef was haunted by ʿAlī, he described feeling the suffering that ʿAlī endured during his lifetime at the same time that Yusef would also feel his own suffering in any way that resonated with ʿAlī's. This means that musical hauntingthe collapsing of spectre and song and the collapsing of 'then' and 'now'is fundamentally a collapse of 'that pain' and 'this pain'. This musically generated interaffectivity (cf. Fuchs 2017), is a notion of subjectivity released from a single, physical body; an atemporal, transpersonal subjectivity swelling with affective resonance. The term resonance is key here to pinpoint the types of vibrating, immaterial energymusic, spectres and circulating affectsthat, intertwined together, resonate dīwān bodies and spaces. Melodies are conductors in the transmission of affect (see Brennan 2004) or, as Kapchan puts it, 'the sound body … resonates (with) its environment, creating and conducting affect ' (2015: 41).
Going into trance as part of haunting is understood by trancers as part of the therapeutic process of dealing with pain, hardship and sufferingthe ruptures, loss and suffering of ancestors, of communities and of one's own personal journey in the present. Describing what trance feels like, most individuals describe 'something else' taking over and losing control of their bodies: 'I just don't feel well' (ma h  asītsh mlīh  ) or 'I cannot control myself' (ma h  akemsh fī ruh  ī). As trancers lose more and more agency to the music, ritual helpers utilise important aesthetics (smells, tastes) to deepen the experience. Trancers must take on all of these aesthetics; that is, songs have to be trance-danced according to the respective qualities and affects associated with the spectres. Saints and prophets, for example, involve 'lighter' varieties of trance (milder bodily motion) than spirits (more vigorous).
From her early teenage years, Hasna began reacting to Sergu's songs in ritual. Her reaction ranged from bursting into tears to suddenly becoming paralyzed or feeling dread. Like with other adepts, such affective ripples occur at the beginning of a song, at the first iteration of the main theme. Meriem, another adept, spoke of feeling sad 'for no reason' and Mehdi, a young man said that he had 'picked up' grief from someone else. When Hasna falls over, begins to moan, or even reaches up to cover her face or bury her head, she is losing agency to some other presence, some kind of spectre, whether it be a jinn or a painful memory-feeling. This means that, first, she is haunted by the spectre through affective-bodily reactions in response to the song.
Hasna then must approach the ritual space in front of the musicians and begin moving her body in semi-codified trance-dance movements in order to be musically worked into an associated trance state. Haunting, thus, involves several stages of feelings and responses for the haunted. As Hasna indicated, this can be exhausting work. Sergu's song suite alone can last an hour or more.
Being worked by and responding carefully to the musicians, trance-dancers may finish either by collapsing of exhaustion, by losing consciousness and falling over, or by feeling that the trance-dancing has relieved the build-up of pain enough that he or she can retreat from the ritual space. How long this relief lasts depends on the spectres. Painful feelings may come up again or supernatural agents may assert their own will over the haunted. Like many modes of suffering, relief is ultimately temporary. When one is musically haunted by a song/spectre, s/he will likely be haunted for life. As a person ages, hauntings return, over and over again, as an ongoing relationship. Haunting presses into the future, too.
This incessant recurrence is built into the structure of the music. Albeit a temporal art seeming to move us 'forward' in time, music in dīwān, however, is cyclical. As a song progresses, the musical textures deepen, repeating themes over and over, weaving layer upon layer of singers, qrāqeb (metal clappers) and ginbrī, and gradually compressing and accelerating. Every song and every suite intensifies in this way. With a cyclical musical structure that thickens and thins with variable sonic textures (textures of clappers, singers, and overall speed), ritual time also circles in on itself, moving in spirals of sensory intensity. Because these songs are energetically conjoined with spectres, music makes sonically palpable the ritual lifeworld's recurring and 'hauntological structure' (Csordas 2019: 521). In other words, by way of musical affect, particularly the rumbling bass tones of the ginbrī, ritual time is felt as an intensive magnitude (Schmitz 2020), wholly immersive rather than as successive moments between then and now. Growing ever intense, these affects modulate into trance which also intensifies and eventually peaks. This public, witnessed nature of trance engages a social field as much as a personal one. It is precisely in the intercorporality between personal and social registers of feeling and being that haunting does so much of its work (Figure 1).

What haunting does personally and socially
Jangari Mama comes in red. Meriem, one of his adepts, is ritually covered in a red cloak and wears a red sash-like belta hizamathat must always be tied around the abdomen. This is because Jangari Mama requires his adepts to take up large butcher knives and, at particular moments designated by musical cues, the adepts turn the knives on themselves, slicing side to side on the abdomen. Even when she is barely conscious and paralyzed, others move Meriem's arms for her, enacting these gestures until she regains enough presence to do them herself.
Dīwān (also used as plural) are semi-public occasions for the community to assemble and to enjoy the music and social atmosphere. 12 In this social nexus, who is haunted by whom is public knowledge. Being haunted by a particular song becomes wrapped up in one's social identity and self-identity. Attendees and fellow trancers sometimes discuss reasons for being haunted by a particular song/spectre but conclusions are typically tentative. Like with Meriem's case, speculation usually ends with, 'only God knows why'. There are not always clear reasons why someone might be haunted by a song as the connection does not necessarily derive from one's ancestral origins or other explicit ties. For example, Hausa songs can haunt non-Hausa descendants. For one adept, Munir, who is not part of a family lineage of dīwān adepts, haunting began when he was a child, after exposure to his first dīwān. He recounted how often others watched him, noting how he was reacting in each ritual. Even when he was 'not in the mood', he was nevertheless encouraged to get up anyway; sometimes ritual participants physically made him get up and dance. 13 Indeed, ritual leaders, helpers, musicians, connoisseurs and the public explicitly contribute to a successful dīwān. Experts and helpers keep an eye on trancers, who are carefully monitored throughout their trance process. Cases of paralysis, for example, require special attention from ritual leaders. Friends and family coach their loved ones who suffer through difficult states, patting them on the back, tying back restrictive clothing, propping up their limp bodies and carrying them to the side of the ritual space when they lose consciousness. At other times, if a trancer is frozen, unresponsive, or unable to move, a ritual expert may physically move the stuck body, as in Meriem's case mentioned above. The key is to get moving whatever feelings are vibrating or ceasing to vibrate.
The local approach to affect here is that difficult and painful feelings are viewed as purposeful and an important aspect of being human, particularly as a Muslim subject humble before God. 14 As Asad has argued about agency in Islam (2003Islam ( , 2009, pain can be a 'kind of action', and, thus, not necessarily a position of disempowerment or victimhood. That is, pain can be purposeful; it does things for individuals and communities. Here, pain is not so much a thing but a process spanning time. In kind, the embodiments of musical hauntings, the affects leading to trance, are intended to be worked (ykhedem) through the body in trance-dancing until the affects have run their course for the time being (see Turner 2020a).
Musical hauntings are viewed as productive for other reasons, too. Because hauntings are considered part of one's identity, the ritual realisation of musical haunting is a mode of self-actualisation for the haunted and is, therefore, important to sustain. The public display of pain and suffering through laborious trance-dancing, regardless of its 'cause', warrants respect and empathy from others and, depending on the spectre in question, it can then implicitly classify people via their suffering. For example, because a jinn can be particularly unruly and merciless, a direct musical haunting of a jinn requires more care than indirect, emotional haunting.
The public performance of all registers of musical haunting, then, also means the public performance of pain and suffering. But what kinds of painful feelings or 'suffering' (ghabina) are we talking about? In the first instance, many adepts spoke of their daily sufferings and challenges as wrapped up in their hauntings, from a death in the family to relationship or financial struggles. Hammitu, a musician and ritual expert, insisted that one can be afflicted by the present. He cited pain and suffering that emerges from within the current intersubjective matrix of the haunted person, even joking that one could be afflicted by 'too many taxes'. Firstly, socio-cultural norms both within and surrounding the dīwān community discourage the discussion of painful things, lest they encourage gossip and meddling in the world of 'only-Godknows'. Secondly, the majority of histories precipitating haunting are typically those that are unspeakable or unsayable. Here, suffering is often beyond the linguistic reach of narrative (cf. Desjarlais 1997) but the predictable, temporal structure of the ritual's robust musical architecture offers an expressive, sensory-rich medium in which pain can emerge without the need to 'talk about it' or for it to be 'coherent' or linear. 15 These unspoken sufferings are often considered as transgenerational affective debts, emerging from unresolved, collective suffering. Sometimes precise ancestors were mentioneda parent or grandparent and in a handful of cases, a distant relative who was enslavedand at other times, these debts were more general. Musical hauntings, indeed, give the haunted special, enigmatic and especially affective access to other time-spaces although, to be clear, ritual actors would never overtly claim this. Ritual knowledge and understandings are implicit and embodied, passed on in non-verbal ways. The body is key, I was consistently told. It is not simply an expression of these ambiguities, but a guide. In this regard, I got the overwhelming sense that, in dīwān rituals, the body could 'remember things' that the rest of the self could not. Letting the body unfold into trance was a prime way of accessing this history.
When a trancer is working through their haunting in trance-dancing, they occupy an overlap 'between presence and absence, self and other' (Jankowsky 2016: 81). It is a psychical territory of innumerable horizons, between things lived and remembered, things not lived but remembered and things 'remembered' or experienced but that never happened. What is 'real' is determined by what is felt as meaningful, however it may come. Such phenomena are not unique to dīwān. Indeed, 'the notion of the between has deep roots in Sufi thought' (Stoller 2009: 5), in the Arabic concept of barzakh, an isthmus between two places. It is understood in many other psycho-therapeutic settings (Crapanzano 2004;Jankowsky 2016;Kapchan 2007Kapchan , 2015Pandolfo 1997;Stoller 2009). The bothand mechanics of musical haunting are analagous to a barzakh: the 'simultaneity' or energetic overlap of 'then' and 'now' mirrors the way song and spectre energetically coalesce to vibrate the body.
Sonically gripped in the temporal envelope of cyclical song, the therapeutic dimension of musical haunting precisely emerges from these simultaneous felt modes of presence and absence, here and now. That is, when musical hauntings grip bodies with tranceitself a barzakh between cognitive presence and absence that obscures the thresholds of memory and amnesiafeelings are meaningful in and of themselves, whether or not they follow temporal 'logic' or can be 'cognitively' understood.

Concluding thoughts
Rima comes in black. He is usually presumed to be male and is a Hausa, sub-Saharan 'spirit', but he can also be found in Songhay spirit pantheons, indicating that he has a long history across and within sub-Saharan Africa. Aziz trances to all of Rima's three consecutive songs that are musically slower and simpler than the other dīwān songs. Like me, Aziz, finds this simplicity especially poignant. I feel a strong sense of longing and sadness when I hear Rima's third song. But these are not my ancestors. This is not my history. Nevertheless, one night I found myself moved to tears by his song, not understanding why. Aziz warned me not to listen to Rima's songs too often: if I started to feel the urge to dance, I, too, could be haunted by things I had never experienced myself. The spirits might take my suffering as an invitation.
Today in Algeria, the recognition of the trans-Saharan slave trade and the trans-generational suffering it created is muted. The violent origins are understated in dīwān as well. Unlike other ethnographic explorations dealing with such monstrous spectres (Kapchan 2007;Shaw 2002), few ūlād dīwān speak openly about the trade and many vehemently deny its existence. Despite admirable literature on slavery within Africa and across the Sahara (notably Fisher 2001; Lovejoy 2000Lovejoy , 2004Wright 2007), the details of this particular part of Algerian history have not been closely examined. There are rough estimates as to numbers and general ideas about what sorts of lives slaves and their descendants had over centuries of forced migration but little else. Furthermore, while dīwān are still widely and regularly practiced in Algeria and attract steady general interest from the state as music festivals and 'folklore', dīwān is largely invisible to the average Algerian. Elders worry that it is gradually disappearing. In 2013, a well-known moqedm (ritual elder and leader), Jalūl Mot am, told me, 'With the independence of Algeria, dīwān disappeared'. Many other ritual experts echo this sentiment, citing the post-Independence displacement and destruction of the grābaracially segregated communities where dīwān developedwhere post-Independence changes are said to have eroded the cohesion of communities that were also musical and ritual communities. But again, dīwān has always involved the push and pull between memory and forgetting.
In light of all of this, the phenomena of haunting are especially helpful in understanding how unwritten, unspeakable, forgotten, revised and repressed histories still deeply matterin flesh, in bones, in musical and ritual practice. As Kleinberg notes: How can we account for the missing portions of the past without simply assuming them to be the missing part of a larger whole whose properties and scope we have already determined? On my account, the haunting … allows for the presence of this absence without predetermining the "what" of what the ghost or haunting is, without supposing it to be something known and determinable but simply absent. (Kleinberg 2017: 10;my emphasis) And while slavery has ended, 'something of it continues to live on, in the social geography of where people reside, in the authority of collective wisdom and shared benightedness … Such endings that are not over is what haunting is about' (Gordon 1997: 139).
Likewise, dīwān are not ways of settling 'unfinished business' nor is the 'finishing' the goal. 16 After all, dīwān is not a cure. The haunted are never eventually 'healed' in the sense of completion; suffering always comes back in some form. Rather, dīwān is a modality of continually inhabiting and embodying various tumultuous, political histories perpetually resounding through the daily lives and physical bodies of ūlād dīwān. These historieswhether personal or collective documented histories, memories, or visionsreturn again and again through the affective mechanics of music.

Notes on contributor
Dr. Tamara Turner is a music anthropologist specialising in North African popular Islam, trance rituals, and affect studies with supporting areas in psychological anthropology and the anthropology of consciousness. Her award-winning doctoral thesis was the first research to document the musical repertoire, practice, and history of Algerian dīwān, a nocturnal trance ritual of the Bilaliyya Sufi Order. She is now a research fellow at the Max Planck Institute, Center for the History of Emotions in Berlin.