The Musical Poetry of Endangered Languages

Kota and Wakhi Poem-Songs in South and Central Asia

Oral Tradition, 35 (2021):103-66

Many peoples of the world conceptualize what English speakers call poetry and music or song as a single integrated unit, what I will call the “poem-song.” Poem-songs may function as models or molds, opening up possibilities for singers, poets, and composers to structure and remember texts, and to convey their ideas through familiar melodies and other channels. Singers may “compose and recompose the phrases for the idea of the moment on the pattern established by the basic formulas” (Lord 1960:5). They may also innovatively break, counter, or question the molds by changing typical metric structures, line lengths, context-appropriate tunes, and so forth.1 I will be considering the poem-song in two endangered, minority languages of South and Central Asia, Kota and Wakhi, and showing how, in the absence of strong revitalization movements, small-scale acts of individual creativity contribute to the vitality of these languages.2 As Gregory Nagy states in another context, “an oral tradition stays alive though its variations and

1 See also Margaret Mills, who analyzes the transmission of Afghan tales and finds that variation does not merely consist of substituting “structurally equivalent objects, personages, and events into a single narrative ‘framework,’” but that variation is structural, leading to varied outcomes in the tales themselves (1990:233 and passim).

2 The two regions under discussion are both long-term field sites of mine. I have been conducting field research in Tamil Nadu, India, since 1982, and specifically in the Nilgiri Hills, where the Kota people live, since 1990. My two years of doctoral research (1990-92) were conducted in all of the (at that time) seven Kota villages, and I lived continuously in one of them, Kolmel, for more than a year. From 1992 until the time of this writing, I have been conducting shorter field excursions to the Nilgiris, each lasting from several days to several months. In 2012 I began a new project in Tajikistan, initially focused on musicians who sing and accompany themselves on plucked lutes, with attention to how aspects of text are brought out on these instruments. Part of my research was conducted in the Wakhan, a border region of southeastern Tajikistan. My interest was in how speakers of the minority Wakhi language compose new poetry (which is always sung), drawing from models of classical Persian. I spent an academic year in 2012 working in Tajikistan as well as in Pakistan, where some Wakhis had migrated. Since 2014 I’ve been returning to Tajikistan several times a year to continue this research. Beginning in 2015 I extended this research to the Wakhan region of northern Afghanistan, across the river from the Wakhi-speaking part of Tajikistan.


reworkings” (1996:27).3 The Kota and Wakhi cases are not unique, but rather illustrate the possibilities of the poem-song in any society in which “poets” are always creating texts meant to be sung. They also serve as a reminder that “starting with the text” need not necessarily blind one to the richness of a text’s musical realization.4 I was motivated to write the present article both after observing striking parallels between the song traditions of these otherwise vastly different cultures, and upon reflecting on my own process of coming to learn the languages and sing songs in these languages.

In one sense, the poem-song is unremarkable. Songs are totalities that include melody, rhythm, and maybe instrumentation—they are generally not just song texts, although the musicality of a written poem may challenge even English-language distinctions between poem and song.5 The melody of a conventional song melody will probably remind us, however imperfectly, of its lyrics, just as in ancient Greece a melody could stimulate a performer to remember its associated dance poses and words (Nagy 2010:382). We don’t usually hear song lyrics performed as speech, and when we read familiar ones, we might “hear” them sung in our heads.6 Singer-songwriters are typically the wordsmiths of these kinds of poem-songs. In the languages many of us speak, poetry and verse also call for their own formal contexts—the poetry-reading, the play, the speech—in which the poet, the actor, the schoolchild, the politician, recite or quote the poem, usually in a manner distinct from ordinary speaking. This poetry can be set to music, too, but the poem and its setting come about through distinct phases of creation, and usually through the agency of different people. The poem-song becomes worthy of special attention when an autonomous tradition of spoken recitation does not exist. This is the case in the genres I explore in this article.

“Speech” often differs from “singing” in degree rather than kind. George List, in “The Boundaries of Speech and Song” (1963), was not the first to notice that words and concepts corresponding to speech, song, chant, recitation, oration, and so forth vary considerably from

3 Nagy paraphrases the conclusions of Ramón Menéndez Pidal, who writes in his study of La Chanson de Roland (1960:67-68),

En conclusion, toute oeuvre qui demeure, des siècles durant, dans la tradition anonyme, tantôt interprétée sourtout par les chanteurs professionels, tantôt surtout par les simples amateurs (geste, ballade, conte . . . etc. . . . ), toute oeuvre de cette catégorie ne revête pas une forme fixe, sculpturale, mais une forme vivant, continûment renouvelée dans ses éléments constitutifs; c’est une oeuvre qui vit de variantes et de remaniements.

The key difference here is that the traditions under discussion are not entirely “anonymous,” and not merely “interpreted” by performers; in some cases, performers are in a more substantial sense composers and authors.

4 See the debates on this matter laid out in Rodgers 2017:316-21.

5 Elissa Guralnick calls the poem “Now Sleeps the Crimson Petal,” by Alfred Lord Tennyson, an example of a song written without music (Guralnick 2009). I have no quarrel with acknowledging the deeply musical qualities of some texts, whether explicitly poetic or not. But Guralnick seems to imply a poem can be a song even before it has been performed aloud, and that is to imply that even a spoken recitation of the poem is “singing.” The metaphor of music extends here beyond the ordinary use of “song” in English—and it locates the musicality entirely within the text. For a different view, see Bickford, who regards singing and verse as “organiz[ing] utterances in forms not native to language” and singing as “a layer of discourse built on top of language” (2007:439, 466).

6 See Aaron Fox (2004) for a detailed ethnography of a Texan community whose members integrate the messages and styles of country music into their everyday conversations, as well as an analysis of everyday “country” tropes that are taken up in country music.


society to society, and many scholars have since reiterated this point (for example, Sherzer and Urban 1986:6; Seeger 1986).7 Nevertheless, List (1963:1) could consider speech and song under a broad encompassing rubric because both are vocally produced, linguistically meaningful, and melodic. List’s framework was designed to take account of an entity only in its moment of performance and not in its state of potential; it did not track the possibilities and limitations of a performance style. Does textual content dictate melodic possibilities? Does a tune, rhythm, or matter of enunciation suggest particular texts appropriate to it? In the cases considered here, the answers are often, “yes.” This mutual implication is part of what gives the poem-song its coherence, recognizability, and force.

In surveying a range of scholarship on music-and-language relations published between 1994 and 2012, Paja Faudree argues for “viewing music and language as variably constructed distinctions in a total semiotic field” in order to promote holism in anthropological practice without sacrificing the analysis of text. She uses the term “language-music” for this “unified expressive field” (2012:519-20).8 I find this formulation problematic: language is an entire sign system, present and recognizable in every society, that differs from other sign systems in particular ways. For Roman Jakobson, language is distinguished by the important role of the phoneme (1978:66-67) and by six functions, including the “poetic” (1960). For Steven Pinker (and Darwin), language is an “instinct”: knowing it means “knowing how to translate mentalese into strings of words and vice versa” (Pinker 1995:82). Music, by contrast, cannot be defined across societies (except by the observer), and even if we provisionally agree upon sounds that are musical across societies, those sounds do not necessarily serve a common set of functions. Studies exploring language-music relations inevitably explore some aspect of language in relation to a particular kind of music.9 I am not convinced that “language” and “music” constitute parts of a unified expressive field. I emphasize this, lest in the term “poem-song,” “poem” be taken as synecdochical for “language” and “song” be taken as synecdochical for “music.” Rather, I offer “poem-song” as shorthand for the following:

A poem-song combines verbal expressive forms, such as poetry and verse, with special uses of pitch, timbre, melody, rhythm, and, possibly, sound-producing instruments (sometimes called

7 List created a “chart for classifying forms intermediate to speech and song,” taking into account relative degrees of intonation and scalar structure (1963:9).

8 While sympathetic to a holistic analysis of verbal expression, I take issue with the way Faudree and some others invoke “language.” For example, some authors focus on langue (e. g., syntax or grammar) and others on parole (e. g., everyday speech rhythms) without acknowledging that these are but some of the attributes contained in “language,” while they are not language itself. We can’t substitute “language” for “speech,” or “music” for “song,” and speak of a language-to-music continuum. Although the term “language” can be used very loosely, it refers to a system and not to an act. Definitions of music today are often fraught with ideological concerns, and in some languages, need not include vocal production at all (and indeed, it is common to hear the English word music used in reference only to instrumental performance).

9 There are too many examples to mention, but to give a small idea of the range, these include studies of the “semantics” of one musical system (Powers 1976); general arguments about the constraints imposed by tonal languages on musical settings—based on insights from one language (for example, Agawu 1988); analyses that stem from speakers’ claims about the “musicality” of their language (Faudree 2013); and rhythmic analysis of purely instrumental repertoires that, according to nationalist sentiments, resemble particular European languages (Patel 2008:159-68).


music). While recognizing the poem-song as a unity, performers and listeners are usually capable of singling out elements or attributes for discussion, demonstration, or critique.

This article is divided into three main parts. Part I briefly discusses the ways in which music and poetry have been conceptualized as parts of larger encompassing wholes, especially in South, Central, and West Asia, and, because of its historical connections with these regions, ancient Greece. Part II is a general introduction to the Kota and Wakhi people, their languages, and their musical traditions. Part III, which forms the central body of the article, is itself divided into two sections. The first considers Kota speech, the life and songs of the modern composer A. K. Rangan, and the āṭḷ or song of grief in historical perspective. The second section considers the Wakhi bыlbыlik and its transformations, and the life and songs of a modern bayd composer, Qurbonsho. The South-Indian-Kota and Central-Asian-Wakhi case studies shine a light on one another, bringing into focus processes of composition and performance that transcend the particularities of one time or place. I conclude by considering the poem-song in light of Ibn Khaldun’s image of the “mold” or “loom” in reference to Arabic poetry.

I. Encompassing Wholes: Music and Poetry

Theorists of music in the Arabic- and Persian-speaking world drew upon and reinterpreted ancient Greek writings on music. Aspects of what they wrote about continue to be important to discourses on music in Central and South Asia today. Especially relevant to the poem-song is the idea of suitability between textual subject and musical setting.10 The root of the word “music” in English and mūsīqī in Arabic is the ancient Greek term mousikē. But in ancient Greece mousikē meant much more than just music. It was the art or craft of the Muses, “goddesses who inspire the special state of mind required to create the special language that they control.” Mousikē, the “special language” over which the Muses presided, embraced words, movement, and melody. Even performing on musical instruments was “in fact an aspect of verbal art” in ancient Greece (Nagy 2010:370). One of the reasons for this integration was the existence of both metric and melodic accent in ancient Greek (Nagy 2010:384; Allen 1973:3-5, 86)—so that a text was something akin to a musical score.11 Mousikē, then, was grounded in a notion of divine inspiration with an emphasis on the verbal.

10 This is not to imply that there is a single intellectual history of such ideas—composers in many parts of the world think about how music suits a text, and may even use disjunctions between music and text as an expressive tool. Stephen Rodgers writes, “The greatest conflicts between text and music have to do with subverting the natural declamation of the text” (2017:336). The Nāṭyaśāstra is a touchstone for ideas regarding verbal and musical correspondence in Sanskrit poetics: “Consonants, vowels, euphonic combinations (sandhi), case-endings (vibhakti), nouns (nāma), verbs (ākhyāta), prefixes (upasarga), particles (nipāta), secondary suffixes (taddhita), and syllabic and moric metres always relate to the verbal themes of music (pada)” (Nāṭyaśāstra 28:16-17, trans. by Ghosh 1950:II, 4). More specific indications of the kinds of dhruvā songs that are appropriate for particular dramatic themes are given in Chapter 32 of the Nāṭyaśāstra, especially lines 422-70 (Ghosh 1950:II, 149-54). One such statement is, “In the case of crows, monkeys, swans, and peacocks, the Dhruvās should be made up of light syllables with swift movement, and of heavy syllables with their slowness” (32:430-31, trans. by Ghosh 1950:II, 150).

11 In the classical period accents were not marked in the text, as they are today (Probert 2006:15).


The Greek term melos approximated what we now call music or song. In the writings of Aristotle’s student Aristoxenus and others, melos had the broad meaning of “melody, rhythm and words” (Barker 1989:126) but could also refer more narrowly to melody or scale.12 Music theorists writing in Arabic and Persian such as Al Farabi (872-950) and Safi ul din Urmawi (1216-1294), who drew on their knowledge of Greek writings, used the term alḥān (plural of laḥn) to refer to a range of concepts corresponding to the Greek melos. Mūsīqī, in the Arab Middle Ages, was neither the Greek mousikē nor the art of sound, but was rather a branch of mathematics, the science of composing alḥān (Farmer 1913-36; Urmawi 1960-61). Al Farabi called melodies (aḷhān) “perfect” (kamilah) if they were performed by the human voice and in some cases on instruments. Through their association with poetic discourses, perfect melodies could inspire listeners to seek wisdom and act ethically (Madian 1992:245, 351-52). Mūsīqī, then, was the science of making melodies etc., the most valued of which had a special relationship with poetry.13

In ancient India, prior to the fourth century BCE, Sanskrit writers used separate terms to designate song, instrumental music, drama, and dance. Owing to the close association of these arts, according to Manmohan Ghosh, the term sangīta emerged in about the third century CE, “signifying . . . all the different phases of music including dance” (1950:5). Just as the word “music” today has a more narrow meaning than mousikē, the term sangīta today refers more narrowly to instrumental music and song.

In contrast to the Sanskrit tradition, the South Indian concept encompassing music was—as in ancient Greece—language. In South India, that language was Tamil, and since at least the seventh century CE, Tamil grammarians have been representing their language as threefold, comprising poetry, music, and drama.14 In the view of many writers, classical Tamil poetic works

12 Barker (1989:126) translates the beginning of Aristoxenus’s Elementa Harmonica, Book 1, as follows: “The science concerned with melody has many parts and is divided into several species, of which the study called Harmonics must be considered one: in order it is first, and its character is like that of an element.” Barker comments (126 n. 1):

Melos, here translated “melody,” can mean: (i) song, broadly conceived to include melody, rhythm and words (a common usage, sometimes expanded by technical writers into the phrase teleion melos, “complete melody,” e.g., 12 Arist. Quint. De Mus. 28.8-10); (ii) melody conceived apart from the other elements (e. g., 35.24, 38.21 below), or (iii) the melodic series or scale on which a melody is based (e. g., 5.11-12, 39.20-3).

13 For a deeper and more detailed discussion of these and other matters pertinent to musical knowledge in the Muslim world, see Blum 2013.

14 Kamil Zvelebil notes that evidence for the term muttamiḻ (“three-Tamil”) dates back only to the late-sixth or early-seventh century CE, although the texts in which the term appears claim a more ancient origin for it. The three Tamils are iyal (speech and writing), icai (song or music), and nāṭakam (drama). In Zvelebil’s view (1992:141-42),

it is a conception of and approach to language according to which language use is manifested in its totality when expressed as speech (or spoken → written word), as song (the word sung, speech joined with musical sound), and as enacted; it is a conception of and approach to language in its dynamism and functional entirety; that is to say, language (in this case, the Tamil language) is not “just” speech, not only spoken/written word (as in iyal, poetry) but also, simultaneously, song, music, word combined with musical sound, sung word, and again simultaneously, word enacted in performance.


were not just musical, but themselves constituted forms of music.15 Grammars, such as Kuṇacākarar’s commentary on Amitacākarar’s tenth-century work, Verses on the Precious Jewel Prosody, employed specific terms for rhythmic feel and intonation in their analysis of prosody. These terms were often compounds that included the Tamil term used elsewhere for music, icai (Niklas 1993:141 and passim); in Tamil prosody, icai referred to a subdivision of rhythm (Niklas 1988:180).16 The recourse of grammarians to icai and related terms speaks to the lack of a clear boundary between pure text and musical sound in Tamil poetry of this period.17

In many performance traditions of the world, words and music bind, limit, and provide space for one another. Musical considerations at times clarify the metric organization of poetry that has no regular pattern in terms of syllable count or quantity. What Mary Boyce called the “cultivated . . . . imaginative and evocative” sung poetry of pre-Islamic Persia (1957:35-36 and passim) was, in her view, based on the musical placement of stresses. Ehsan Yarshater made a similar point, extending it to much modern Iranian folk poetry as well (1974:62 and passim). John Marr (1985:409) and Lynn Ate (1984) have analyzed Tamil poetry of the early and the early-medieval periods along the same lines, writing that syllable groupings do not always conform to the poetic feet of Tamil and Sanskrit classical meters, but are rather arranged according to the way they must have been sung. In both the Iranian-speaking and Tamil lands, music and poetry have been linked deeply both in concept and performance. The idea of poetry as an object separated from music is, in some cases, a modern one. But the idea of the poem-song, poetry and music as an integrated unit, has by no means disappeared.18

II. Regions and Languages: The Nilgiri Hills and the Wakhan

The Nilgiri Hills are located at the juncture of three South Indian states, Tamil Nadu, Kerala, and Karnataka, and were one of the many hill stations in colonial India developed as respites from the heat of the plains.

15 Or as David Shulman put it on the first page of his “biography” of the Tamil language, “Indeed, ‘music,’ or ‘the Tamil that is music,’ icaittamiḻ, is one of the meanings of the name [Tamil] itself in the ancient grammatical and poetic sources” (2016:1).

16 Since poetry and music are mutually constitutive in this grammar, it is problematic to define icai simply as “music.”

17 Examples of these terms for rhythmic variation are ēnticai (“rising”), nkicai (“swinging”), and ōḻukicai (“flowing”). (Ulrike Niklas translates ēnticai with the word “eminent,” which doesn’t lend itself well to describing rhythm.) The same term icai, along with the terms ōcai and oli, are elsewhere implemented to indicate phonological uses, leading Niklas to translate them as “tone” (1988:193).

18 See also Crosson, who discusses the value Gaelic poets place on traditional music—and how they have consciously turned to music to inform their contemporary writings, recognizing the “issues of a broken tradition and community” (2008:15 and passim).


            Map 1. The Nilgiri Hills of South India (from Wolf 2009:238).

The British particularly enjoyed the Nilgiris for the similarity of their often rainy climate to that of England. The larger Nilgiri environment is populated by immigrants from the plains and other parts of India who own businesses and tea plantations, or work in them; a robust Indian and international tourism business operates there as well.


Fig. 1. Blacksmithing in Kolmēl village, 2001 (photo by Richard K. Wolf).

The special inhabitants of the region are “tribal” populations known as ādivāsis (“original inhabitants”), who once shared a caste-like system of economic and ritual interdependence.19 Among those tribes, the Kotas were known for their skills as craftsmen, potters, jewelers, blacksmiths, and musicians, which they practiced alongside farming, hunting, and raising cattle (Wolf 2005).

The Kota language belongs to the South Dravidian group and emerged as an independent language no later than about 2,000 years ago, when Tamil and Malayalam became separate languages.20 It is currently listed as “Critically Endangered” in the UNESCO Atlas of the World’s Languages in Danger, meaning “the youngest speakers are grandparents and older, and they speak the language partially and infrequently” (Moseley 2010). This is a mischaracterization, however. Despite the small number of speakers (roughly 2,000), most Kotas are fluent in their

19 The governmental designation for ādivāsi is Scheduled Tribe, a term which contrasts with Scheduled Caste (dalit or so called untouchable), Other Backward Caste, and other labels, many of which remain in flux.

20 Historical linguists attest to this using the so-called palatalization rule in which Proto-Dravidian /k-/ becomes /c-/ before front vowels, with some exceptions. Kota preserves the velar /k-/ from Proto-Dravidian (Emeneau 1995).


native language and use it for day-to-day communication with one another—particularly in Kota villages.

Unlike for Wakhi, as I shall discuss, no dramatic geographical factors account for the development of Kota in isolation from the plains languages of Tamil and Malayalam. Colonial writers on the region would commonly romanticize the isolation of the region, exaggerating the role of the mountains in separating the tribes from the plains population. Kota could well have developed in the context of relative social isolation, brought on by the caste-like intertribal system of economic and ritual interaction and the strict rules of endogamy.

Fig. 2. The Kota village of Kolmēl in the late 1990s (photo by Richard K. Wolf).

The Nilgiri mountain range is not vast, covering only about 1,000 square miles. In contrast, the home of the Wakhis and other Ismaili Pamir groups is much larger: the Gorno-Badakshan autonomous region of Tajikistan covers about 25,000 square miles, and Badakhshan, Afghanistan, is about 17,000 square miles in size. The mountain peaks of the Nilgiris are in the 7,000-8,000 foot range. These are lower than even the valleys in Wakhan, some of which are as


high as 11,000 feet.21 Some degree of isolation—social if not also geographic—allowed the Nilgiri languages to develop, but this was happening in an environment in which at least some members of the communities were conversant in several Dravidian languages of the region.

The population of roughly 2,000 Kotas is now spread out over six villages (a seventh village is now defunct), and some Kotas live and work in other parts of India, returning to their villages mainly for festivals and funerals. The community is strictly endogamous in the sense that those who marry outside the community are ostracized: there is no way (now) of “becoming” a Kota through marriage or residence.22 Most Kotas live as peasant farmers in their villages or work in a variety of private and government jobs throughout India. Some own tracts of land with tea and other cash crops that yield significant income.

A small number of women maintain the craft of making special pottery for everyday and ritual use, and men still play and make musical instruments; both men and women dance, and women’s singing is an important concluding act at various ritual junctures.

21 One of the possible isolating factors for Kota and its neighboring tribal languages was the presence of malaria in the moist, thickly forested foothills. Although malaria transmission can technically occur at all altitudes in the Nilgiris (Bishop and Litch 2000:157), it is commonly believed not to. In any case, those who regularly walked through the dense malarial zones would have developed a degree of immunity, as would those who lived in plains areas where malaria was prevalent.

22 There are legendary exceptions to this in the distant past. For example, some of the exogamous clans of Kolmel village are said to have originated when a Kota and a non-Kota were married and had children.


Fig. 3. Kota woman making a clay vessel for use at the varldāv, a secondary mortuary ceremony. Mēnāṛ village, the Nilgiris, 1991 (photo by Richard K. Wolf). 

The principal musical repertoires are named for and tightly tied to ritual contexts. The ensemble of double-reed koḷ and drums plays one repertoire for funerals, one for god-related rituals, and a dance-music repertoire that to an extent cuts across these and other contexts. Unlike in most South Asian societies, no repertoire exists for weddings. Songs fall under the local categories of “god songs” (devr pāṭ,mainly sung by women as they dance in a circle), lullabies (jo jo pāṭ), and āṭḷ, songs of grief and loss that are personal and do not form part of any ritual repertoire.


Fig. 4. Kotas playing music during a funeral in Kolmel village, January, 2015 (photo by Richard K. Wolf).

Men and women compose many songs that do not fall into generic categories, both in Tamil and in Kota, often adapting melodies they have heard from films, the radio, and local devotional singing. Kotas are occasionally called upon to perform their music at municipal and state functions, both in the local capital and in Delhi. However the kinds of songs to be discussed here are domestic and personal, performed alone or among a small group of neighbors and relatives out of grief, remembrance, in the course of storytelling, or for the sheer joy of singing. Although a few women composers have also sung their songs on a local radio station, this is not the norm.


The population and geographic spread of Wakhi speakers is at another order of magnitude. While official accounts vary, some Wakhis estimate their numbers to be as high as 80,000. The UNESCO Atlas of the World’s Languages in Danger classifies Wakhi as “definitely endangered,” meaning, “children no longer learn the language as mother tongue in the home” (Moseley 2010). This too is a mischaracterization. Some children living outside of Wakhi-populated areas may not learn Wakhi as a mother tongue in the home. In my experience, Wakhis speak to one another (including children) in Wakhi.


Fig. 5. Wakhan River Valley near Yur village, Upper Wakhan, Afghanistan (photo by Richard K. Wolf, January, 2020).

The Wakhi homeland is the Wakhan River Valley, beginning high in the Pamir Mountains along the border of Afghanistan and Pakistan. Wakhi settlements dot both sides of the Wakhan River and continue west until the Wakhan River joins the Panj River, which borders Tajikistan. The Panj turns southwest, and Wakhi settlements continue on both sides of that river, ending just before the administrative centers of Sultan Ishkashim in Afghanistan and Ishkashim in Tajikistan.23 The Wakhan and Panj rivers are headwaters of the Amu Darya or Oxus River. Significant Wakhi diasporas also exist in Pakistan, China, and Russia.

23 The rivers are not shown explicitly in Map 2, but the Wakhan River joins the Panj River from the east in Qala-e Panja, and from there onward (southwest, then north) the dotted line indicating the national boundary between Tajikistan and Afghanistan corresponds to the Panj River.


Map 2. Wakhi settlements and migration (from Kreutzmann 2017:197; reproduced with permission of the author).

Traditionally, Wakhis have subsisted on farming in their home villages near the river, but seasonally much of the village has regularly migrated to one or more high mountain settlements for the summer to graze its cows, sheep, and goats. Owing to heavy taxation within Afghan Badakhshan, and religious discrimination against this Ismaili Muslim community by various Sunni rulers in Afghanistan, the Wakhis were forced to move into some of the harshest and least productive lands in the region and at times, to sell their own people into slavery (Kreutzmann 2015:205-08, 210-12, 220, 248-58, 273, 295-98). During the eighteenth and nineteenth centuries, populations of Wakhis migrated in waves to southwest Xinjiang, China, and to northern Pakistan (Kreutzmann 2015:208-09).

Unlike Kotas, Wakhis are free to marry non-Wakhis—the more important consideration being whether or not the prospective spouse is Ismaili. However, Ismaili-Sunni alliances are not unknown, and over generations non-Wakhis, even Sunnis, have become de facto Wakhis by living in the area, practicing Ismaili religion, and speaking the Wakhi language. Similarly, those Wakhis who were sold into slavery a century ago have apparently merged into the surrounding population of Badakhshan in Afghanistan, and their descendants may not even know that their ancestors were Wakhi.

The Wakhi language, called x̌ik zik in Wakhi, belongs to the Pamir group within the Eastern Iranian language family. It developed independently from the language that, farther to the west, became Modern Persian. Some two to three thousand years passed before Modern Persian spread eastward, reaching the region of today’s Tajikistan in the eighth century CE (Perry 2006). It probably took more time for the form of Persian that is now called Tajik in Tajikistan and Dari in Afghanistan to make inroads into the mountains where Wakhis live. Over the centuries, Persian speakers from farther west—that is, from modern-day Iran—settled in relatively small numbers in the Wakhan. Their descendants continue to live in Tajik-speaking


villages of Wakhan in Tajikistan today. Later, national projects led to the teaching of Tajik and Dari in Tajik and Afghan schools. Despite the relatively modern advent of spoken Persian in the region, Ismaili Muslims of the Pamirs and Badakhshan have been using Persian-language poetry as part of their religious practice for at least a millennium.

Fig. 6. Summer settlement in the upper pasture above the Wakhi village of Zumudg, Tajikistan. The river Panj is visible just beyond the descending mountains, and the Pamirs of Afghan Wakhan are on the other side of the river (photo by Richard K. Wolf, August, 2019).

 In the Wakhi homeland, some of the prototypical song forms include the bыlbыlik, a three-line song, usually expressing longing for a distant or dead loved one; wedding songs (tuyāna or bayd), often accompanied by frame drum (dā’ira, Wakhi dorya) and sung in Persian or Wakhi; and funeral laments, sung either as quatrains in Persian or alternating between the Persian and Wakhi languages.


Fig. 7. View of Zumudg village and upper pasture from the Afghanistan side of the Panj River in winter (photo by Richard K. Wolf, January, 2020).

Modern Wakhi songs, called bayd, are often set in rhyming forms that correspond to the ghazal (rhyming couplets AA BA CA DA and so forth) and rubā‘ī (quatrains in AABA, AAAB, and other combinations). As differences in vowel length aren’t phonemically significant in Wakhi, Wakhi poetry does not draw upon the ‘aruz metrical system. New Wakhi song tunes and performance styles are drawn from the musical environment in which the performer lives. Tajik Wakhis sing songs embracing some of the diversity of the cosmopolitan Soviet era, and Afghan Wakhis use tunes and styles they’ve heard in Afghanistan. Radio and television have also been conduits for transmitting the tunes of popular singers such as Ahmad Zahir (1946-1979) in Afghanistan and Dushanbe Pallaev (1950-2017) in Tajikistan. Wakhis of Gilgit-Baltistan in northern Pakistan sometimes describe the melodies of their newly composed sung-poetry in terms of the language in which they think the song was originally sung—as if the language sticks to the music. The musical styles in this part of Pakistan are distinctively different from those in Tajikistan and Afghanistan.24

24 Georg Buddruss and Sigrun Wiehler-Schneider (1978) found regularities in the singing of Wakhis in Hunza that led them to speculate on the existence of an indigenous music theory.


Fig. 8. Left to right: Poet-singer Daulatsho playing ghijak and singing bayds,accompanied by Zarkub (rəbob) and Mamad Ziya (dorya). Meadow called Sir on route to the upper pasture in Yur, Afghanistan (photo by Richard K. Wolf, July, 2016).

While poetry and music are deeply integrated in this region, melodies can nevertheless be detached from the particulars of a song and played alone or used for other poems. So, the integration of poetry with singing is not necessarily dependent upon a particular melody—poet-singers (and singers in general) may express themselves according to their choices of musical settings, based on their personal moods and the musical and social contexts in which a particular text comes to mind. This doesn’t seem to be the case for the early-medieval Tamil texts discussed in the Verses on the Precious Jewel Prosody, where the musical characteristics are intrinsic to the text—that is, not separable from it at all. In the repertoires with which I work, particular texts and melodies together form well-defined genres or distinct pieces within those genres. This seems to be particularly true for songs of loss or sorrow, such as the āṭḷ in Kota and the bыlbыlik in Wakhi.


Fig. 9. Karakoram mountains in Gilgit-Baltistan, Pakistan (photo by Richard K. Wolf, July, 1997).

III. Kota and Wakhi Poem-Songs

Kota and Wakhi, like all languages, include many borrowed words and constructions, the foreignness of which not all speakers recognize. The thorough interpenetration of Tamil and Persian into each language, respectively, and the further embedding of the words those languages had already incorporated from Arabic, Sanskrit, and other languages, make for a complicated linguistic picture with many openings for creative etymology. The Arabic word ma‘mul (“customary”) has made its way into both Kota and Wakhi. While Wakhi speakers would probably recognize the term as Persian or Arabic, Kotas generally see it as indigenous. Ironically so, for māmūl in Kota is the word for “traditional” or “the old way.” (The contrasting term, meaning “modern,” is ocmūl).25 The mixed vocabularies of each language, along with exposure to popular and literary traditions of the dominant language in each case, have opened up avenues for new forms of verbal composition that transcend the boundaries of any one language. Supporting the vitality of a language, endangered or not, need not mean purging it of foreign

25 The folk etymology of māmūl is —from a Tamilization of the Sanskrit mahā, meaning “great”—plus the Kota word mūl, meaning “direction.” This yields “old way” or “great tradition,” as it were. To distinguish modern innovation from the ways of the ancestors, Kotas substituted with the term oc, meaning “new.” This yields ocmūl, “new way”—a modern innovation.


influences—despite the various attempts of ideologically driven political movements to do just that. Some of the most influential composers in Kota and Wakhi over the course of the twentieth century have experienced some form of separation from their homelands. This has, I think, given them perspectives and data from outside their village frames of reference to expand the linguistic and musical possibilities of their own languages.

Kota Songs

In this section we will explore the life and music of an influential Kota song composer as well as analyze perhaps the most well-known Kota āṭḷ, using recordings from 1922 and 2019. I will point out the ways in which Kota singing differs from speech, how Kotas create textual and musical parallelism, and ways in which they modify tunes, rhythmic patterns, and metric tendencies to work with a given text, as well as create texts in the first place.

A. K. Rangan

An introverted, imaginative Kota man named A. K. Rangan created a new style of Kota composition using popular tunes from Tamil films and the like, which were melodically more complicated and wider in tessitura or range than those of Kota māmūl genres. His compositions, of which about five are still sung, harked back to film songs from the 1940s onward, but by the time I met Rangan in 1990, he no longer sang or remembered any of his songs. From a young age, having fallen victim to the quarrel between the māmūl faction and the ocmūl faction of the village, Rangan was outcasted. After spending years wandering penniless through the cities and towns of the Indian plains, he eventually served in the army in far northeast India. The accompanying video excerpt of Rangan recounting his life story (Example 1 in the eCompanion) and the transcription below give an idea of the sound of Kota speech, shot through with many consonant clusters.

Rangan insisted on preparing his life story in advance rather than responding to an interview. My research assistant R. Kamatn sits to his right, responding attentively at the appropriate moments (the responder performs an important role in storytelling among the Kotas as in many cultures). All the untranslated responses mean roughly “yes” or “then what” and more generally “I’m listening.” The text begins just after Rangan describes being born, and the recording starts from the bolded words and ends at the underlined word.

Excerpt from A. K. Rangan’s life story:

AKR: alk enne mog vecko. mog vecvīḷḷe, ayṇ engicko? enne mupād mog āypīḷḷ kārmōṛk parykm calckēmīrr ayk tūykir ōcīkmūḏ,

Then [she] gave birth to me. When [she] was giving birth to me, what did my father do? He said that when I became thirty [days old], they’d take me to Karamadai [temple on the plains] and fulfill a vow. So, it is said, they carried me there.


RK: cer

AKR: ayk oyṯke pūjm gījm elm koṭṯk kev kutkōṛo iḏṯe, kev kuytīḷḷe kutilk enn viṛkīnāykmūḏ

They said “we have to go there and perform a puja and everything and pierce his ears.” While piercing my ear, according to what they said, it wouldn’t pierce.

RK: e

AKR: ayk koṛikn kevatk ītr oygbaṭk, maykkir oygmūḏ. anmāyṭk añjīṭ, kutlk payḷad iṯ ītr vadko,

Then when they were removing the piercing instrument, I fell unconscious, they say. Then they became fearful because the piercing failed, and then they brought me home.

RK: cere

AKR: alk ītr vadṯke mm . . . anme itko. itrkveḏmēle, alk ān doḍa doḍa añj-ār varcm āyko, alk ayṇ inanm maṇḍn kirāp vecko. alle koṭ veykvōṛo, koṭ iḏmēle, naṛmaṇḍlike uṉck mīrn viṛkōṛo

So they came back and so that’s how it was. In the meantime I got bigger and bigger, reaching the age of five or six, then they cropped my hair. They were supposed to leave one tuft in the center with a little bit of hair [according to a custom related to receiving divine advice through an oracle],

RK: m micm pūrām mayṇḍr ōṛo

The rest should be shaved . . .

AKR: mayṇḍrkvōṛo. alk inm kirāp vecrkēḏbaṭk, am kōkāl janmele, aṛi, piṛi, “koṭ veyād, kōnāṭōn dākl kristvan dākl kirāp vecko” irr, ceriyāna pōrātṃ pōracṯ ūyrk,

Should be shaved . . . then when I was wearing the crop like this, the people of our village were punching me and grabbing me and saying “without keeping a tuft you’re wearing a crop like a Tamil plainsperson, like a Christian,” they fought fiercely and threw me out.

RK: cere

AKR: ad ayṇ anme ēmāytr ad nāḷke naynke irre, ēmaytre vadk

Then my father deceived them, saying tomorrow or the day after he would [cut his own hair, leaving a little bit in the center, to make up for how he had his son’s hair cut].

RK: ceriya


AKR: alk eṉk pat varcm āyk alkī, tūj vāydyarayṇ tamīṇkdēnm kirāp veck, alk id entl galāṭāykūrdartm, kōkāll galāṭāykōre jāyt viṭiṯ tayḷṛkōṛo iḏṯke …

And then I reached ten years of age, and the teacher Tuj also cropped his son’s hair, which meant real trouble! In the village they said it would be necessary to outcaste [the ones who had had their hair cropped] …

Rangan is relatively quiet and subdued in the recording in comparison with most Kotas, who tend to speak in a more animated verbal style. However a close examination of the video reveals expressive hand gestures, referring to concrete objects (his ear, cropped head), the ends of sentences, thoughts, or moments in time (movements with distinct stops for closure), and more abstractly, the failure of the ear-piercing to work (a gesture outward from the chest). Rangan’s manner of raising and lowering his fist is a common South Indian gesture used to ask questions—here, “why did you shave your head like a Tamil?” (it did not refer to the fighting).26 Then at the end of the recording Rangan waves his hand in rhythmic coordination with the contour of his words, “ad ayṇ anme ēmāytr ad . . . .” All of these kinds of gestures are common to speaking, narrating, and singing the song genre āṭḷ.

Rangan describes a major event in the history of Kolmel village that had ramifications for the gods worshipped (modern versus ancient), the manner of conducting mortuary ceremonies (two or one, with or without bovine sacrifice), and what came to be perceived as a long-term rift in the village between its māmūl and ocmūl factions (Mandelbaum 1954:86). A. K. Rangan puts himself in the center of this story. By contrast, anthropologist David G. Mandelbaum narrated this moment in Kota history from the perspective of his main informant, Sulli (referred to above by his Kota name, Tuj), who situated himself as the major actor (Mandelbaum 1960:276).

The song sung by Rangan that we examine differs from most traditional (that is, māmūl) Kota songs in terms of its wide melodic range and its seemingly constant shifts in meter and rhythmic pattern. Based on my impression of the melody, and having identified film-song models for some of his other songs, I suspect this too was modeled on a film song. Kotas find some of Rangan’s songs difficult to sing, producing results that come off as rhythmically and melodically meandering. The singer P. Kamatn, who died in the late 1990s, was considered a fine instrumentalist and singer. He and his sister Mādi (mother of R. Kamatn) both used to sing this song, but I haven’t come across it otherwise. In general, Rangan would complain that people sang his songs incorrectly, but he could not remember how he used to sing them, nor had he committed the texts to writing. This leaves us in doubt about the orderliness of the “original” textual and musical parts of his influential compositions—if a stable original ever did exist (see Lord 1960:100 for a discussion of “originals” in oral performance). The first version of the song transcription shows the text as it would be spoken. Doubled consonants are pronounced individually, unlike in English. The text as sung is provided separately in a musical transcription.

26 In the Tamil interrogative gesture, the thumb sticks straight up from the fist.


Erdd tigḷe,” composed by A. K. Rangan and sung by P. Kamatn (May, 1992; consult the eCompanion, Example 2):


erdd tigḷe nīymn aṛdg tigḷē
enāḷṇ oypd āḷār ēdā?
aṛdād oypd adecym ednā?
erdd tigḷe nīymn aṛdg tigḷē

O you shining moon, tell me
Where is my husband headed?
Why did he go without telling anyone?
O you shining moon, tell me

Verse 1:

āne anḏirlil nayḷilk oygē 
tigḷe veḷpil āne ayāre vadmēl āgādām

I go to play in the evening,
in the moonlight, at a time
you said it’s not good for me to go there

Verse 2:

alle nīye mañjuḷk uykṯe
avn entk cēytṯ taḏīyē

Then you hid in the mist
and let him draw closer

Verse 3:

enayṇk anjṯ nī alāda cīmk oypiyo
paṇm īlādn unaciṯ nī paridēcāyṛ oypiyo

Fearing my father, you’ve gone away somewhere
Thinking you have no money, you’ve gone along like a beggar

Verse 4:

gav īlāde gaṇmog perpē
aṛdād oypd adecym ednā

With the appearance of an unloved boy
Why did he go without telling anyone?

A wife asks the moon to identify the route on which her husband has embarked. She alludes to a tryst, when the moon hid in the mist so she could meet her lover under the cloud of darkness. Āḷṇ means not only husband but any classificatory affine in the same generation as a Kota woman—that is, a Kota man with whom romantic relations would not be incestuous. So the lover is the āḷṇ, but perhaps they haven’t been married. The boy-lover ran off in fear of the girl’s father, perhaps because, as a husband, he wouldn’t have had the money to support her. The feeling of alienation probably resonated with Rangan’s own experiences as an outcaste.

Certain prosodic features found here are common to all Kota songs. For instance, the phonemic distinction between long and short syllables is maintained rigorously, and consonant clusters are broken up a little bit by extending the release of each: for example, erududu and aṛudugu create space between the consonants in the clusters d-d and d-g in erdd and aṛdg, respectively. More distinctive to modern Kota poetry, and likely borrowed from Tamil, is the use of rhythmically striking alliteration, consonance, and assonance to form internal rhymes and other parallelisms—and these parallelisms serve poetic functions (Jakobson 1960). In the refrain, for example, the shining of the moon is linked to the telling of the moon by the parallel sequence of vowel—liquid/flap-voiced consonant—voiced consonant: erdd and aṛdg. The husband, his


route, and his “not telling” are all brought into poetic equation through the use of long ā vowels: enāḷṇ, āḷār, and aṛdād. Notably missing is any kind of poetic meter; rather, the tune holds the text together and creates line breaks.

The melodic-rhythmic patterns alternate between duple and triple meters. The first line establishes a straightforward pattern of four units of six pulses based on the speed with which the syllables are uttered and the way they are accented. In the following notation, metric units are demarcated by slashes, and each pulse receives a syllable or a dot. The first line in each set indicates the scale degree in relation to the tonal center (taken as C). The second line indicates the text syllables, with vowels added to show how each one is articulated. Bold indicates dynamic accent, underline indicates word-beginnings, which receive stress in some positions more than others. Italics on the pitch positions indicate obvious points of metrical stress—such indications become more important in lines 2 and 3, where accents change. An asterisk indicates a position of metrical stress that does not correspond to a syllable articulation. As the first line is metrically ambiguous, only the first syllable of each six-pulse unit is italicized.

Each unit of six pulses can be heard as three sets of two. In this hearing, two units would be equivalent to a single 6/8 measure in Western staff notation. However, one can also hear each six-beat unit as 3 + 3 (over the 2 + 2 + 2 scheme). Many folksong types in South India, including some Kota women’s dance songs, maintain a steady two-against-three feel. In the text itself, both interpretations are supported: the position of ti, the first syllable of tigl (“moon”) in pulse position 5, reinforces an interpretation of each group of six pulses as 2 + 2 + 2. The word ni (“you”) in pulse 4 of the second set of six can support a 3 + 3 hearing, although ni is not strongly stressed.

As for the “color” of the musical articulations, the two halves of the first line (units 1 and 3) begin with the vowels [e] and [a]. In Tamil, Kota, and other Dravidian languages, the onset of [e] is a glide, indicated in the notation as “ye.” In the case of [a], the nasal from the previous word nīymn joins with it to yield “n‿a.” Both [e] and [a] return in transformed fashion in lines 2 and 3. As soft onsets, these vowels do not strongly emphasize metric units. By contrast, the [t] in tigl, as an unvoiced stop, is naturally harder and lends accentual force. The melodic line also reinforces the 2 + 2 + 2 grouping in the third unit, where the scale degree 3♭ is emphasized through upper and lower neighbors. The small melodic peak on ti in unit 3, pulse position 5, serves to reinforce its metric accent, and gives ti more rhythmic drive than in unit 1.

Key to notation:

Unit X
Pulse positions (not labeled):1 2 3 4 5 6
Scale degrees (in numbers): /x x x x x x /
Syllables:   x x x x x x


First line of “erdd tigḷe,” as sung:

          Unit 1                                Unit 2                  Unit 3                                   Unit 4
 /7♭  7♭   7♭   7♭|   7♭   7♭     / 1   .   .  1   1    2      / 3       4   3♭  2    3♭ 1    /1 . . . . .
 yer-də-  də-  də    ti   -gə   – le   .   .  ni-yu-mə-     n‿a – ṛə-də-gə   ti-gu – ḷe . . . . .

The next line of the refrain reinforces the reading of each six-beat unit as two groups of three. The central units of lines 2 and 3 follow the 6 + 6 pulse duration set up in line 1, but only if we understand the extra syllables ye and aṛə as “pickup” notes, and and as downbeats of final units that are not fully realized. That is to say, the metric character of the singing is strongly active within the phrase but does not continue from phrase to phrase (cf. Foley 2002:33). Buddruss and Wiehler-Schneider note a similar phenomenon in their analysis of Wakhi songs from Hunza, Pakistan, where extra syllables are accommodated not by singing them more quickly, but by expanding the period, adding extra beats if necessary (1978:108).27 I’ve found this to be more characteristic of Kota songs than Wakhi ones, especially if they are not accompanied by a time-keeping instrument.

Lines 2 and 3 of the refrain with extra syllables framing the two-unit core:

                     Unit 1                         Unit 2
1      /   1  .   5      5   .    5   / 5       6♭   5   *     3     2   /  3
ye –  nā  –  ḷəṇ   woy – pə  / dā    –    ḷā   –     r‿ē   –  /  . . . . .
1  2  / 3♭   4 * . 3♭    2  / 1      2   3   2       2      1  /  1 . . . . .
a-ṛə-      – – d woy-pə- da-de-cy-mə –  ne -də – ?

The syllable yen, meaning “my,” that precedes the first syllable of unit 1 is, sensibly, an anacrusis to the strong beat on āḷṇ “husband.” Most of the words in lines 2 and 3 beginning with vowels, including āḷṇ, are sung from the final consonant of the previous word (indicated as above with an undertie ‿). The word-initial [o] in oypd has a bilabial glide, making the sound wo. The complex duple metric feel in line 2 is supported by word-initial consonants or vowels on three of the four strong beats, positions 1 and 4. The syllable r‿e (unit 2, position 5), which is offset by a pulse, allows , the end of the question word, to fall on a strong beat. Unit 2 can also

27 “Der Text einer Strophe bewirkt dann eine deutliche Variantenbildung, wenn seine Silbenzahl nicht mit der für das Versschema ungefähren Norm übereinstimmt. In der 2. und 4. Strophe des 1. Liedes hat Phrase A 1 eine Silbe mehr als in den anderen Strophen. Um diese überzählige Silbe unterzubringen, wird der sonst etwa drei Achtel lange Abschlußton unterteilt, in der 2. Strophe als Viertel + Achtel, in der 4. Strophe als Achtel + Achtel. Die gesamte Zeitdauer kann sich bei einer solchen Unterteilung durchaus verändern. Es scheint, als wird nicht so sehr nach dem divisiven, sondern mehr nach dem additiven Prinzip verfahren, indem der Grundschlag als metrisches Element beibehalten und die auf Grund der Silbenzahl erforderliche Anzahl an Tönen aneinandergereiht wird. Reicht die in den anderen Strophen aufgestellte normative Zeitdauer nicht aus, wird sie so weit verlängert, bis alle Silben untergebracht sind” (Buddruss and Wiehler-Schneider 1978:108).


be heard as 2 + 2 + 2 with each syllable lasting two pulses, driving forward to the strong beat on . The metric shift corresponds to the singer’s wondering about the route (āḷār) her husband (āḷṇ) is traveling (oypd).

Line 3 is a melodic answer to the question posed in line 2, resolving the third scale degree to the first. The pickup syllables are the stem of the verb “to say” (aṛ-). The strong beat at the beginning of unit 1 is a verbal ending expressing negation, meaning, here, “without informing anyone.” The extended syllable [ā] and the word following it, oypd, give the impression that the rhythmic pattern of line 2 will be repeated. But this turns out to be an illusion, as woy falls on pulse position 5 (not 4, as in line 2), and initiates a very clear pattern of alternating stresses. This reinforces the metric feel of both units as 2 + 2 + 2. In line 2 the woman is wondering where her husband is, whereas in line 3 she is asking a rhetorical question—the move from tension to resolution in the music is justified textually because of the enunciative force of the rhetorical question.

Listening with the principles laid out above in mind, the song is musically a set of melodic questions and answers, at different pitch registers, with one line of the pair favoring a two-against-three feel and the other emphasizing regular alternation of strong and weak beats    (2 + 2 + 2 for each unit). The structure of the verses is as follows:

Verse 1: Line 1, Tune A, question, 2 + 2 + 2 feel
               Line 2, Tune B, answer, two-against-three feel

Verse 2: Line 1, Tune C (lower in tessitura), question, two-against-three feel
              Line 2, Tune D, answer, 2 + 2 + 2 feel

Verse 3: Same structure as Verse 1, using tunes A and B

Verse 4: Same structure as Verse 2, using tunes C and D

To summarize, the melody in this song holds together and reinforces poetic aspects of the text. The relationship between the text per se and the song as sung is not one of perfect conformity. The singer uses a couple of melodic ideas and structuring principles to create coherency for lines that otherwise vary in syllable length. Each line is separated by a pause that is not accounted for in terms of an overall musical meter, but on its own, each line has a metric feel. Among the reasons for this are the pickup notes, which reinforce the sense of meter within the lines but not between them. That focal meter consists of two sets of six pulses, divided into pulse groupings of two or three. The flow of the song is animated by these two different ways of dividing units of six pulses, and is held together by the repetition of verse melodies and the use of a refrain.


In the āṭḷ, or Kota song of grief, a flexible tune creates continuity from line to line, even when those lines vary considerably in length. The metric variation in the āṭḷ and the degree to


which the singer may use her own words is greater than in the more strictly (and complexly) composed modern songs of A. K. Rangan and others. Generally, to refer to more regular metric songs (pāṭ), Kotas use the related verbal form for singing, pāṛ-/pāc-.28 In contrast, Kotas tend to use different verbs for the āṭḷ. One verb, which means “to render a tune vocally or instrumentally,” et-/eyt-, has non-musical meanings that include “picking up and carrying,” “putting hands together in salutation,” and “building a house.” According to my observation, “rendering” involves the performer’s on-the-spot negotiation between several components of a performance—which leads to considerable variation. In the āṭḷ, the singer negotiates the song by fitting her mental image of a story into the textual and melodic mold of an āṭḷ. In the case of instrumental music, the shawm player makes the melody fit the drum pattern and vice-versa (the process is not always straightforward). Āṭḷs also have a related verb, āṛ-/āc-,meaning to speak or move (and, in other contexts, dance)—and this makes some sense, as the singer tells a story, sometimes with gestures.

The āṭḷ employs stereotypical forms of address and often refers to the deceased as one who didn’t heed sensible advice. The causes of the death are generally inferred, not told—the sorcery of a neighboring tribe being the proximate cause, perhaps brought into actuality by a wild animal or an illness. The song may be as short as about a minute or as long as about ten. Longer songs will likely narrate a series of place names and events that track the path of the subjects as they meet their fate. Shorter songs do not tell stories so much as allude to them; but even the longer ones seldom provide all the information needed to follow the underlying story, and thus rely on a paramusical tradition of storytelling to be fully understood.

Āṭḷs are not technically what many scholars call laments, in the sense of spontaneous, tuneful expressions of grief at the side of a corpse. Kotas call the latter “crying” (agl). Rather, āṭḷs are songs composed by particular people in response to particular events, and their styles are sometimes inspired by funeral lamentation. However, just because āṭḷs are particular, composed songs does not mean that they are performed the same way each time, nor does it mean that the singer is responding to the same event originally alluded to in the song. Rather, unless the singer is the composer and is thinking about her own life tragedies, she sings for aesthetic enjoyment and to pass the time, alone or with others. The singer may, in singing, both evoke and pacify her own pain and fear. Unlike most of the named Kota musical genres, āṭḷs have no ritual context.

In December of 2019 I traveled to the Nilgiris and initiated a survey of current Kota singing practices with an eye toward changes over the last thirty years. Kotas are apparently singing many of the same songs as they were in the early 1990s, as well as composing new songs; the new wave of composition brought on by A. K. Rangan more than forty years ago is still strong. Āṭḷs are still being sung even though some of these point to situations that are no longer so relevant to young people (such as the imminent threat of wild-animal attack). Āṭḷ singing styles are out of keeping with the song styles currently popular in South India. Even in 1991 I was struck by the perspectives of a young male musician, who criticized the āṭḷ because

28 The Dravidian Etymological Dictionary (Burrow and Emeneau 1984) does not include this version of the verb in entry 4065, which suggests to me that the verb might have been created on the analogy of the Tamil headword, pāṭu, pāṭi, subsequent to Emeneau’s fieldwork in the late 1930s.


of its cut-up articulation and offered the smooth continuity of film songs as a favorable alternative.29

Fig. 10. Kotas were valued guides for British colonial hunting expeditions. This photo was displayed prominently in the home of a religious officiant in Kurgoj village (reproduced by the author, December, 2019).

The āṭḷ discussed below, arguably the most famous, is associated with several distinct stories and has been adopted into the instrumental repertoire for funerals.30 The version on the accompanying video (Example 4 in the eCompanion) shares both the story and the melody with one recorded by the Gramophone Company of Calcutta under the auspices of the Linguistic Survey of India in 1922 (Gramophone Company of Calcutta n.d.; Gravely 1927:30-31; consult the eCompanion, Example 3).

A corrected transcription of the 1922 recording appears below, followed by the printed version from 1927. The singer probably dictated the words of the song to the fieldworker and likely a Tamil-speaking assistant. This dictation seems to be the basis of the printed version, for it omits many sung details and adds lines not present in the recording. I have transcribed the sung version in tristichs corresponding to the melodic strophes.

29 See Wolf 2000/2001:161 for a description of the kind of disarticulation common in āṭḷs.

30 For a case study using this āṭḷ, see Wolf 1997:360-84 and 2000/2001:160-64. See also my more general treatment of āṭḷs in Wolf 2000/2001 and 2005:59-63 and passim.


“Kōṭa. Song:—Story of Mathi.” 1922 version, as sung (consult the eCompanion, Example 3):


la la la la la la . .
la la la . . la la . . a
la la la a la la


ēṭk ōkōme Māydē
iḏre Māydē
ōkēḏmēle Māydē

“Let’s go (collect) reeds, Mādi!”
having said (that), Mādi
while going, Mādi


pīc aḍal āyko Māydē
vīdīke Māydē
ōkēḏmēle Māydē

a cat crossed, Mādi
the path from the house, Mādi
while going, Mādi


nāy aḍal āyko Māydē
ayo yanga Māydē
enekēno Māydē?

a dog crossed (the path), Mādi
Woe, younger sister Mādi!
What to say, Mādi?


Kārgālke Māydē
ōkēḏmēle Māydē
kāk aḍal āyko Māydē

To Kārgāl (a place in Ticgāṛ village), Mādi
while going, Mādi
a crow crossed (the path)


ayo yanga Māydē
inekēnego Māydē?
enangne Māydē?

Woe, younger sister Mādi!
What to say, girl?31 Mādi
My younger sister Mādi


oḏ alāde aṇe
mūnḏe aṇe
cagnme āyko aṇe

“Not one, elder brother
three, elder brother
there were three omens, brother


tirgirkōme aṇe
iḏmēleke Māyde
nīyāne Māyde

Let’s turn around and go back, elder brother”
When you said that, Mādi
I said, “you Mādi”


tiriguḷigo Māyde
ānāne Māyde
ojāḷe Māyde

“You turn around, Mādi”
“I, myself,” Mādi
“alone,” Mādi


oyṯe Māyde
vakēnego Māyde
iḏīḷḷe Māyde

“Will go,” Mādi
“And come [back to the village],” Mādi
At the time of saying this, Mādi

31go is a suffix used for addressing a female. It’s not a meaningful variation here.



paṛdgēne Māyde
iḏmēleke Māyde
aṇōr īṛāḷe Māyde

I said [this], Mādi
And while I was saying it, Mādi [said:]
“The two of us, brother and sister [will go together],” Mādi


oyṯe Māyde
itlārēne Māyde
eridire Māyde

I will not go [alone], [said] Mādi

“I’m afraid,” [said Mādi]


vakēḏmēlē Māyde
kōkālekē Māyde
vakēḏmēlē Māyde

While coming back, Mādi
To the village, Mādi
While coming back


ananm ala aṇē
porañjeke aṇē
āype aṇē

“It’s not like that, elder brother”
“I have to go ‘outside,’ brother”


iḏre Māyde
paṛdgēn iḏmēl Māyde
ayo yanga Māyde

so said Mādi
while I was telling her, Mādi
Woe, younger sister Mādi


Fig. 11. “Kōṭa. Song:—Story of Mathi.” 1922 version as printed (Gravely 1927:30-31).

The text printed with the 1922 recording (Gravely 1927:30-31) describes the journey of a girl and her brother, who meet three animals identified as bad omens. When they return home, the girl dies. The song as recorded on gramophone record (Example 3 in the eCompanion) is clearly rehearsed and more tidy than one tends to find now, and the printed text exaggerates this orderliness. Several key features of the āṭḷ are left out of the 1927 published transcription even though they are present in the recording: address of the deceased with terms of endearment (here, “my younger sister”), and rhetorical expressions of futility—here, “what to say?” This version, even as sung, is unusual in explicitly identifying the animals crossing the path as bad omens. Singers will sometimes insert the names of other animals—which suggests that, at least for some, the animals are not really serving as omens at all. The animals may serve as variables that can be augmented or substituted according to the imagination of the singer and provide content to propel the melody onward. The theme of the girl telling the boy that this trip is not a good idea fits with the theme of blame common to many āṭḷs. In this case, Mādi warns her brother of what turns out to be grave danger.

The opening events in my 2019 recording conform to those of the song a century before, with a cat and a porcupine crossing the path. An element of the story normally not sung but told


or explained separately, involves Mādi going off into the reed forest to defecate. Normally the euphemism “go outside” is used in Kota, although direct reference to defecation is perfectly acceptable in everyday speech.32 Maṇimāla made the act of defecation explicit in her 2019 rendering with a laugh, perhaps to clarify the meaning for the small children present. In the 1922 recording, Mādi refuses to go back to the village all by herself. She tells her brother on the way back, “I have to go outside,” and that’s all. No explicit reference to her death. Listeners are supposed to fill in the rest from their knowledge of the story. The 2019 recording provides quite a bit more detail than the 1922 recording, as actually sung.

Mādi āṭḷ, as sung by Maṇimāla, age 55, Kurgoj village (December 22, 2019; consult the eCompanion, Example 4):


enavā māyde,
enkēna māyde,
la la la, la la,
li li lo, li lo

O mother Mādi,
what to say?


āne vārādidēn vadī,
vēḏārāme enāde,
ammā nī māyde

I told you not to come out but you did

Mother, Mādi


paylene vārādi,
pice kurke iṯo,
vārādī idēna,
vadiyo, māyde

“Don’t come out of the house”
a cat crossed the path
I told you not to come out
You came anyway, Mādi


la la le la le . . . .
li li lo li li lo . . .
kalavāleke vadī
ammā vārā iḏe
muḷgōyte kuruke,
iḍtirā ammā

You came into the yard,
I said not to
A porcupine crossed the path


vēḏa dēnā vadi,
la la la . .  la le . .
li li lo . .  li lo . .

I said don’t, but you came

32 The everyday word for “shit” used here, kipās, is also one of the many humorous nicknames Kotas use. The “Parable of the Prodigal Son,” printed immediately following “Kōṭa. Song:—Story of Mathi” in the Gravely anthology (1927:32-35), is attributed to “Kippas.”

33 The singer melds the end of the previous strophe with the peak pitch of the next one—hence the lone word in the first line of this stanza.



arikōṭeke onāmā,
ay enene kurk iṯida
vēṭa dēnene,
ammā māyde

When you went to Arikot
I crossed your path there
and said, don’t!
mother Mādi


enavā māyde
enkēnā māyde34

My mother Mādi,
what to say?


ayk mallāṛe oygo,
ayke oyṯe ammā,
porañjāre kipāc okvem

You went into the bamboo forest
having gone there, mother,
[you said] “I’m going to take a poop”

(At this point, the singer laughs, and a female listener makes a sound of affirmation and encouragement.)


kipāck okvē
paṛdiya amma
dūrtk vēṭa
otikene ūrvere

“I’m going to take a poop”
you said, mother.
“Don’t go far”


paṛidēne ammā
mantekēḷāde, oygiyā
ennavā māyde

I told you, mother
You went without listening
my mother, Mādi


eṛēne aṇḍire
vadoma māyde
paykene vadomā

After picking reeds
we came, Mādi
we came home


porañjekepere oygiya,
enavvēne māyde
enekēnā māyde

You “went outside”
My mother Mādi
what to say, Mādi?


andelāyṛēne vadiṯe
porañjike aype aṇē dī,
enavva māyde
enekēna māyde

Having come back in the evening
You said, “elder brother, my menses have started” 
My mother Mādi
what to say, Mādi?


nāḷ uṇāyṛe vadēne,
ammā enave māyde,
vaḏāme tēlvalke

I came the next morning
my mother Mādi
when I came to the menstrual hut

34 Only the first half of the melody is sung here. The melody starts again in the next stanza.



kākvālēne iḏ jāma
enave māyde
enīne vitiṯe

At cockcrow
my mother Mādi
you left me


kaṛdiviyā māyde
enavvanī māyde

You departed
my mother Mādi


The 2019 rendition of the āṭḷ presents the brother telling Mādi not to go far off but Mādi going anyway, “without heeding advice.” “One who doesn’t heed advice” is a formulaic form of address in the āṭḷ that presages events to follow. In this case, Mādi is violated by a member of another tribe, one universally feared in the Nilgiris as sorcerers. In other stories, the victim is impaled on the tusks of a wild board, trampled by a bison, or torn apart by a panther or a tiger. The listener knows that when the girl is alone, something bad will happen.

Mādi then returns home and says she has started to menstruate (expressed as “become outside”) and goes to the place in the village reserved for menstruation and childbirth. At dawn next morning, Mādi is found there, dead. Instead of saying that Mādi is dead, the song says that she “left” and “departed” (lit. “crossed over”)—to the world of the dead.

Thanks to the recording sponsored by the Linguistic Survey of India, we know that this song’s basic melody has been in circulation since at least 1922.35 The arrangement of text into units of the melody, however, varies considerably. On the 1922 recording, a complete iteration of the tune consisted of three phrases, with up to six syllables corresponding to the six main pulses of the tune (in a few places an extra syllable is inserted). Each line ends with the addressee, either Mādi or the elder brother, and this textual repetition combined with the placement of the addressee within the structure of the tune creates and fulfills the listener’s expectations. These expectations are set up in the very structure of vocables at the start of the 1922 recording, with five or six la syllables distributed over eight beats (rests and sustained notes indicated with dots, underlines indicating double speed):

Vocable structure and text structure of Example 3, compared:

la la la la la la . . (six syllables)

la la la .  la la . . a (3 + 2 syllables, with a melisma on the third and an articulatory [a])

la la la a-a la la . .  (same as above, but two double-time articulatory [a] vowels after the third la)

35 Usually the text to this song (meaning the text and story) remains attached to this very melody; however, in the fieldnotes of David Mandelbaum, I found a description of this story linking it to a different melody (Wolf 1997:362; Mandelbaum fieldnotes (n.d.), song 27 and cylinder 22; this recording is archived as part of Mandelbaum 1938). As of 1992 I had collected four versions of the present song melody attached to this story, and sixteen more examples with different texts/stories, adding up to a total of six stories (Wolf 1997:363-64). Since then I have heard more versions of the song than I can count. Instrumental versions of this tune exist as well. This is significant, because instrumental versions of songs are rare among the Kotas. The tonal system of the double-reed doesn’t conform well to the Kota system of singing, which is generally diatonic.


ēṭk ō-kō-mə Māy-dē . . (six syllables)

i-ḏi-re . Māy-dē . . (3 + 2 syllables, with a melisma on the third)

ō-kēḏe-mē-le Māy-dē . . (same as above, but two double-time syllables in position 2 (in italics))

Unlike our previous example, with its many instances of assonance and alliteration, the parallelisms in this example tend to be exact repetitions of words or phrases. A rare exception is in line 5, where the word Kār-ə-gā-lə-ke (“to Kārgāl”) appears in the expected position of the words aḍal ayko (“crossing happened”), deferring the latter phraseto the third line of the tristich.

Although the 1922 rendition sounds highly polished and regular, the logical boundaries for segments of content do not always conform to the boundaries of the tristich. For example, in stanzas 7-10, the content would suggest sets of four and three lines rather than sets of three:

Mādi addresses her brother, “bad omens, let’s go home”:

oḏ alāde aṇe
mūnḏe aṇe
cagnme āyko aṇe
tirgirkōme aṇe

“Not one, elder brother
three, elder brother
there were three omens, brother
Let’s turn around and go back, elder brother”

Brother responds, “you go home”:

iḏmēleke Māyde
nīyāne Māyde
tiriguḷigo Māyde

When you said that, Mādi
I said, “you Mādi
You turn around, Mādi”

Brother tells Mādi what he proposes to do, “I’ll collect the reeds and come home”:

ānāne Māyde
ojāḷe Māyde
oyṯe Māyde
vakēnego Māyde

“I, myself,” Mādi
“alone,” Mādi
“Will go,” Mādi
“And come [back to the village],” Mādi

The fact that micro-level beginnings and endings provided by the three-line musical unit do not match up with the beginnings and endings of narrative units does not seem to be an issue for Kotas.

 In the 2019 recording, the syllable counts in each line vary considerably more than in the 1922 recording. There are two main variations of the tune, one with three phrases and one with four phrases; occasionally Maṇimāla will sing just the two first phrases that both versions hold in common. The third phrase of the 1922 recording corresponds to the fourth phrase of the 2019 recording.

In the following outline of rhythm and melody in the 2019 recording, the note durations and/or rests are indicated with dots (as above), but a comma is added to indicate breaths and pauses between lines. (These are more substantial than the pauses we heard between the lines of


A. K. Rangan’s song and those in the 1922 recording of this song.) Underscoring is used to indicate a doubling of speed.

First stanza of “Māyde” āṭḷ, transcribed (consult the eCompanion, Example 4):

/ 1    4   4    3 2   5  .  .  /
 yen av-vā  māy-de,               

/ 4    4    5   4 3   2        1  .  2 /
  ye-ne- kē- na    māy -de .  . 

/ 1  3  2 .  2  1  . 2  , /
  la la la . la  lo . .  , 

/  1 3  2  . 2 1  . .  /
   li li lo . li lo . .

Despite a rough start, Maṇimala presents the four-phrase musical structure of her version of the āṭḷ right at the outset—two lines with words, two lines with vocables.36 The lines present metric variations on the basic tune, one tending toward a seven-beat phrase and one tending toward an eight-beat one. In the first stanza, Maṇimāla introduces the text using seven-beat phrases but completes the strophe with eight-beat phrases on vocables. The seven-beat line is characteristic of the version of this āṭḷ in Kurgōj village (where the song was recorded and Maṇimāla lives), and I will call it the “model” version. Versions in other villages tend to employ straightforward eight-beat lines. Compare the rhythm of Maṇimāla’s “model” version with that of the 1922 version (beats are numbered above the syllables):

Comparison of seven- and eight-beat first lines in Examples 3 and 4:

     1   2    3    4      5       6 7  (8)
   yen av-vā  māy-de       .  .          2019 recording
   ēṭk  ō-  kō- mə  Māy- dē .    .     1922 recording

     1      2   3    4       5      6    7   (8)
    yene kē na  māy- de     .     .        2019 recording
     i-     ḏi-re    .     Māy- dē    .     .   1922 recording

36 The beginning of the recording was interrupted because the singer was uncomfortable with the sound of her voice, causing her to pause a few times. You can hear another singer join in after the first line, and then Maṇimāla starts again, completing the tune on the vocables la, li, and lo.


The word Māyde, along with its melodic contour, appears in beats 4-5 in the model version (2019) and in beats 5-6 in the 1922 version. Because this difference leads to a seven- versus eight-beat line, the song feels very different. This baseline difference between the two versions has relevance for this āṭḷ in general, but there are many other possibilities for stretching and compressing the melody.

Stanza 2 of Example 4, transcribed:

/  1  4   4   4   4   4    2   5  . . 4 , /  eleven beats
    ā-ne vā-rā-di-dēn va-dī . . .  , 

4   / 4   5   43 21  21     1   1  .  2  /  nine beats (preceded by a vocalization on [a])
a     vē-ḏā-rā-me  ye-   nā-de  . . 

/ 1     3    2  . ,  1  2      1  .  2  , /  eight beats (with an internal pause and vocalization on [a])
   am-mā nī . ,  a māy-de .  .  ,

In the first line of this tristich, the formerly seven-beat melody extends over eleven beats to accommodate the extra syllables. I suspect that Maṇimāla prefixed the word “ān,” meaning “I,” to the line—this was unnecessary because the verb was conjugated for the first person.37 Had she not, the melody would have conformed melodically and rhythmically to the model (that is, vā-rā-di-dēn va dī . = seven beats). These kinds of prolongation don’t seem to bother anyone, so my “correction” is merely theoretical.38

In the second line of the tristich, Maṇimāla begins by briefly intoning [a]. Such onsets are stylistic markers peculiar to the āṭḷ, and they create the discontinuity alluded to in the musician’s comment above. Once the words start, the sung unit is still a little bit longer than the model, owing to a prolongation of the tonic on the word yenāde. This three-syllable word could have been replaced with the name of the addressee, Māyde, the word that typically appears in the line-final position in this song. As a two-syllable word, it would have kept the metric model intact (see, for example, Wolf 2000/2001:163-64). The third line falls within the model and ends with a rise to the second degree—which usually cues a fourth line that returns to the tonic. Here the third line is left (musically) hanging, and the next stanza begins.

As Maṇimāla warms up, she starts stanza 3, a quatrain, in close accordance with the model. The first line occupies seven beats and the second, eight beats:

37 The e added to ān is a form of emphasis, and in general, vowels are added in singing.

38 As Albert Lord cautioned, “we must be content with the texts that we have and not endeavor to ‘correct’ or ‘perfect’ them in accordance with a purely arbitrary guess at what the original might have been” (Lord 1960:100). In this case, I have heard enough versions of the song to consider plausible alternatives that more closely approximate the normative metrical structures. It is not a guess about an “original.”


Stanza 3, line 1, from Example 4, transcribed:

/ 1     4    4    32     5  .  4  , /
   pay-le-ne   vā –   rā . dī ,    

/ 4   5   4    3 2-1 2 1  . 2  , /
  pi-ce  ku-ru-ke  i  ṯo . .  ,

This time she completes both parts of the second half of the melody as well, but in both cases prolongs the melody by several pulses.

Throughout the song, the singer draws on one melodic idea with two basic templates consisting of seven-beat or eight-beat lines. But she is not bound by these metric frames. She simply tells the story, inserting here and there formulae and motives that are specific to the āṭḷ genre. The result is a song that unfolds flexibly in time, as if reproducing the process of remembering—a process of recounting rather than reciting. I can’t be sure if this halting cadence of performance was one of the elements, along with Maṇimala’s appreciated vocal tone, that led all those present to praise her. Indeed outside the frame of the camera, several women were weeping as Maṇimāla sang. They could not have been crying over the anonymous Mādi depicted in this old song (Mādi is the generic name of all Kota females), but might have been weeping for something in themselves, perhaps touched by the sensitivity and beauty with which Maṇimāla sang.


What can we say about this poem-song and its role vis-à-vis Kota as an endangered language? Sir George Grierson, the British officer in charge of the Linguistic Survey of India, recommended the documentation project in 1918 in part “to form valuable records of languages which are liable to change, and which in future years may be extinct” (Gravely 1927:3).39 In fact, these recordings and their transcriptions have not been readily available to Kotas, except when I played for them the cassette copies I’d made from an antique gramophone in the Madras Museum in 1991-92. While the printed text could potentially have been helpful (and it is now available digitally), it was completely inadequate as a documentation of the poem-song; rather, like the English “translation” at the end, it was a summary of an ideal-type of the song. And it left out almost every single marker of the performance: it lacked indication of where grammatical or melodic lines began and ended, and the characteristic features of the genre, as sung, were not represented. The recording is a valuable documentation of one performance in the past, but the project has not directly aided the Kotas in preserving their language. Rather the persistence of the tune and the formulaic nature of the āṭḷ have allowed and encouraged a form of transmitting stories—and in more generalized terms, a way to give voice to emotion—that is unique to the Kota language.

39 For a detailed analysis of the Linguistic Survey of India as a colonial project, see Majeed 2019a and 2019b.


Wakhi Songs

This section considers what melodic and poetic features consistently make up Wakhi poem-songs, how individuals can be creative with these forms, and what kinds of variation typically occur. Wakhis consider bыlbыliks quintessential examples of their verbal-musical art. Beginning with bыlbыliks, in light of our discussion of āṭḷs, we proceed to the more general category of Wakhi poem-songs called bayds, examining how individual instances transcend genre boundaries and/or challenge conventions.


In their classic study of the Wakhi language published in 1976, Aleksandr Leonovich Griunberg and I. M. Steblin-Kamenskiĭ preface their collection of 56 bыlbыliks with these observations (1976a:23):

Only one genre of actual Wakhi poetic folklore remains—that of the small-song bul’bulik, consisting of three lines that rhyme according to the scheme a-b-a. The refrain bəlbəlik tar nolam or bəlbəl tar nolam, “I sing the bul’bulik to you,” is common to all bul’buliks. Bul’bulik is a specifically female type of folklore. Women sing bul’bulik in the summers, standing on the edge of a hill, facing the village and plugging their ears with their fingers.40

Steblin-Kamenskiĭ revises this interpretation of the refrain in his Etymological Dictionary of the Wakhi Language, writing that it can also mean “I sing to you as a nightingale” (Steblin-Kamenskiĭ 1999:107; emphasis mine). The source of this oft repeated earlier interpretation of the refrain may have been S. I. Klimchitskiĭ’s 1936 publication, “Wakhi Texts,” in which twelve songs are identified as bыlbыliks.41 In 1965, Tatiana Nikolaevna Pakhalina (1928-1995) collected examples of bыlbыlik sung by women and girls aged seventy-six, eighteen, and fifteen. This was in Vrang, now the administrative center of Wakhan in Tajikistan. Pakhalina mentions nothing of a refrain and calls the poems “tristich song texts” (Pakhalina 1973 and 1975).42 The

40 “По сути дела сохранился лишь один жанр собственно ваханского поэтического фольклора. Это жанр короткой песенки бульбулик, состоящей из трёхстишия, рифмуемого по схеме aba и припева: Bəlbəlík ta-r nólam или bəlbə́l ta-r nólam ‘Я пропою тебе бульбулик,’ общего для всех бульбуликов.

Бульбулик является специфически женским видом фольклора. Женщины поют бульбулики на летовках, встав на краю склона, обратившись лицом к селению и заткнув пальцами уши” (Griunberg and Steblin-Kamenskiĭ 1976a:23).

41 He wrote that each three-line strophe is followed by the refrain, “bülbülik tār nōləm ē,” which he translates as “I sing the nightingale song to you” (Klimchitskiĭ 1936:89). The most prominent scholar and native speaker of Wakhi, Bogsho Lashkarbekov, also translates the refrain as “I sing the nightingale song to you” (1972:144), although he writes it “wuz bыlbыl tar noləm e,” beginning with the word “I,” which I have never heard sung.

42 Interestingly, in her 1973 samples, which are more extensive than those in her 1975 book, Pakhalina (1973:176) omits an outlier for rhyme in Wakhi—a case where the first line ended with and the third with əṣ̌k. In an edited volume entitled, Problems in Oriental Versification, one might expect this to have been included as a “problem.” No other example in Pakhlina’s later, 1975, publication was left out of the earlier one.


only version of a refrain containing the word “nightingale” (bыlbыl) I have encountered during fieldwork has been “bыlbыl tar noləm,”“[I] the nightingale, sing to you,” or more accurately, “lament to you,” which is what nightingales are said to do. Sometimes other refrains are used, and in some places no refrain is used at all (particularly in Afghanistan).43

Fig. 12. Bыlbыlik singers near upper pasture above Yur village, Afghanistan. They are facing the author, but would normally be facing the other direction (photo by Katherine Freeze, July, 2016).

I find it peculiar, in the first interpretation of the refrain, that Wakhis would think it necessary to state the name of the genre they are performing. In the case of an extreme variant performance—we shall examine one—the common refrain could serve as an argument that, “Yes, believe it or not, this is a bыlbыlik.” But in the vast majority of cases, Wakhis know that the singer is singing a bыlbыlik, and those who do not know the Wakhi language would not understand the refrain anyway. One might say the refrain performs the function of “finalization”: marking the end of one woman’s utterance and creating a space for taking turns (Bakhtin 1986:76-81). The second interpretation of the refrain frames the performance as a mimesis of the prototypical nightingale, familiar not only in the wider Persian literary world, but also, since

43 Hiromi Lorraine Sakata made recordings in Afghan Wakhan in 1972, including example 31 in her book Music in the Mind (2002:164). She was kind enough to share all her recordings from that field visit, and none of the bыlbыliks on the recording had refrains.


ancient Greece, in the West and beyond.44 The nightingale continuously laments, each song different from the last.

As several variant refrains exist in the performance of bыlbыlik today, variations may have been even more common in the past, crystalizing into a favored one over time. This would have happened by the early 1930s, or at least by the time Klimchitskiĭ conducted research for his 1936 publication—the first collection of bыlbыlik and the first mention of the refrain of which I’m aware (Klimchitskiĭ 1936). I can’t help but wonder whether the self-referentiality of this refrain was fostered by korenizatsiya, the Soviet nationalities policy instituted in 1923, which encouraged individual ethnic groups (“nationalities”) to emphasize if not invent outward markers of their cultural specialness (Martin 2001:12-13 and passim). The apparent absence of the refrain in Afghanistan suggests that it developed only in modern Tajikistan. Whether or not this reflexivity was a response to Soviet ideology, the refrain is an example of what Greg Urban calls “metaculture” (2001:3 and passim). The refrain is not only “about” the bыlbыlik, it is also a part of a process whereby Wakhis in Tajikistan have come to define bыlbыlik over time—to the point that the editors of a recent collection felt the need to print the refrain after each of the 257 bыlbыliks it contains (Matrobov and Mirboboev 2015:58-94).45

The verb used in the common refrain, nola car-, means “lament, groan, complain,” as well as “sing, embellish.”46 Although the nightingale may “lament” the bыlbыlik, singers aren’t normally said to “lament” it. The verb for most kinds of singing in Wakhi is x̌an-/x̌at-, meaning “to say, sing, or tell” (Buddruss and Wiehler-Schneider 1978:90; Griunberg and Steblin-Kamenskiĭ 1976a:505).47 X̌an- is “telling” in the sense of expressing, and that expression can be mental. One couplet by the Wakhi poet-singer Qurbonsho, whose poetry will be examined in detail below, uses the term in this more abstract sense:

sad nolai yəm ẓ̌ə rəbobi tu x̌an
yəm ẓ̌ə jonət ẓ̌ə jigari kəbobi x̌an

You imagine the 100 laments of my rubob,
and my roasted heart and soul.48

44 As Gregory Nagy notes, mimesis of the nightingale’s lament forms part of Homeric epic (1996:86): “So also with laments that are quoted, as it were, in Homeric performance. When the rhapsode performs Andromache’s lament, he is Andromache, singing her lament, just as he is Homer when we hear in the Homeric poems: ‘tell me, Muse.’ So also, finally, with the lament of the nightingale in Odyssey 19.518-523: the songbird’s beautiful sad song is being chosen by an epic character as a model for her own epic self-expression. Moreover, in narrating the lyric lament of the nightingale, epic imitates it as a model. This way, epic is not only imitating but actually re-enacting lyric, drawing on its own resources of mimesis.”

45 The discussion of the refrain in this work is plagiarized from Lashkarbekov 1972.

46 I am providing a wide definition here to take in possible usages of nola brought into Tajik Wakhi from Tajik. The word nola is used for a musical ornament in Tajik.

47 One contrastive verb is joy-/joyd-, which would be used for reading something aloud in a spoken form, or to say a prayer (o joy-). Some verbal forms can also use the verb “to do,” car-/kərt-, including the former example of a prayer, as well as “explain” (mano car-), “recount” (naql car-), “tell a story” (qisa car-), or perform one of the genres of funeral lament (rəboi, naw’agari, sifat).

48 Literally, the “kabob” [made of] my soul/heart and liver.


If performance involves some kind of projection, whether in vocally present form or as an abstraction of the mind, creating a song seems to be the process in reverse. The common verb meaning “to compose” any verbal art form is nix̌v-/nix̌ovd-, which means also to “pull out,” “extract,” or “kick out (someone you’re mad at).”49

The Wakhi bыlbыlik is analogous to the āṭḷ in some ways. However, unlike the āṭḷ, which is only sung with an āṭḷ tune, the three lines of the bыlbыlik can be imported into other song formats. Like the āṭḷ, the bыlbыlik is a song of loss, not a lament, and is cryptic as regards any specific turn of events. Its generic conventions, like those of the āṭḷ, make it readily identifiable, and prosodically speaking, the bыlbыlik is more “poetic” than the āṭḷ in its form and use of metaphor. Unlike āṭḷs, which have variable stanza lengths, bыlbыliks are all fundamentally tristichs; the singer usually repeats the third line to fill out a four-phrase melody.50 In general, bыlbыliks are more fixed than āṭḷs. Āṭḷs have irregular parallelisms; generic āṭḷ features can appear in more than one place in a line or stanza; and new āṭḷ tunes and texts continue to be composed. Bыlbыliks by contrast follow an A-B-A rhyme scheme and employ end rhymes. Generally speaking, the “rhyme” in a bыlbыlik consists of either a final vowel or a final vowel followed by a consonant, but near rhymes are also found occasionally—for example, sarγuč and niyəṣ̌k (Pakhalina 1975:176). (Kota songs use a variety of internal rhymes common to both the Tamil and Kota languages.) Each bыlbыlik line tends toward seven syllables—though differences are easy to accommodate—and these syllables correspond to the seven “beats” of the single bыlbыlik tune (not counting pauses between lines).51 As John Miles Foley put it in the context of a different form of oral syllabic poetry, “Music and silence [aren’t] adjunct phenomena; they [are] part of the line” (2002:33). The first line of a bыlbыlik is an image with wide scope for interpretation; the next two lines allude to a person and describe some kind of action.

Both the bыlbыlik and the āṭḷ are gendered as female in their respective cultures, as is funerary crying. Kota women cry copiously and conspicuously at a funeral, men less so. At Wakhi funerals, women lament together using a tune very close to the bыlbыlik, while men gather for prayer led by the Khalifa, and sing or listen to maddo.52 However, there is no restriction on men singing or composing the “women’s” genres of āṭḷ and bыlbыlik, so the

49 The range of the Persian verb barāvardan captures both the “pull out” and the “compose” meanings of nix̌v-/nix̌ovd- (Ḥayyīm 1934-36 s.v. āvardan, bar āvardan). Klimchitskiĭ indeed defines the related verb nix̌n-/nix̌t- as barovardan in Tajik (1936:109). Nowadays some people write their compositions down and will use the verb “write” (nəwiš-, nəwišt-) for composing, just as in English. And, drawing from Tajik, if someone wants to say in a more elevated sense, “create a work (of art),” they could use the verb ejod car-/kərt-. I thank Zohir Piltaboev for this clarification.

50 The form of repetition, where line 3 is repeated with a small variation to complete a four-part melody, is only one of several possibilities explored in Tajikistan, Afghanistan, and Pakistan (see Mock 1998:114, 126).

51 Griunberg and Steblin-Kamenskiĭ (1976b:9-10) write that there is no consistent syllable structure, as the lines vary from four to eight syllables in length; this is technically correct, but the melody is articulated in such a way that seven-syllable lines fit best, and as Karomatov et al. put it, is “structured according to the rules of a musical work of art” (2010:III, 12).

52 Drawn from the Arabic word for praise, maddo is a genre of Ismaili devotional music. It consists of sung Classical Persian verse accompanied by two types of rubāb.


gendering is more ideational than practical. The matter of how and when men take up these genres is beyond the scope of the present article.

Singers vary in the degree to which they compose their own bыlbыlik texts. Often the frame of a longstanding bыlbыlik is made personal with the name of a loved one and becomes for all intents and purposes the composition of the singer. In Example 5, which I recorded in Vrang, Tajikistan, the singer sitting in the middle, Kobulmo Rajabova, leads what she identifies as her own bыlbыlik. Only the insertion of her brother Xushjon’s name makes it her own; the text is otherwise one of the most common.53

Text of “sətkək bə iwət buy.” Sung by Ra’no Rojibekova, Kobulmo Rajabova, and Bibiniso Shogunbekova in Vrang, Tajikistan (January 14, 2014; consult the eCompanion, Example 5):

e sətkək bə iwət buy
ar bor ki Xыšjon x̌anəm
          e lol
e woy aftboron arəm ẓ̌ə ruy
          aftboron arəm ẓ̌ə ruy
          e lol

e bыlbыl tar noləme

Little beads, one after the other
Whenever I say “Xushjon”
          O, little brother!
Woe! torrents of rain stream down my face
torrents of rain stream down my face
          O, little brother!

I, the nightingale, lament to you

The opening line refers to beads, that is, rosary beads, with the diminutive –ək—meaning something like, “dear little beads”—and alludes to counting, literally, “one and two.” Hearing this terse phrase, the listener imagines the experiencer counting beads as if reciting prayers, passing the moments contemplatively while remembering somebody who is far off or dead. I say the “experiencer,” because the singer may be voicing her emotion through images common to bыlbыliks, not necessarily images that pertain literally to herself—the persona created as the experiencer in the bыlbыlik is not usually named. In the second line Kobulmo uses the Wakhi-ized Tajik word for “each time,” ar bor. This provides alliterative and internal rhyme with aft boron in line 3. Aft boron is also Wakhi-ized Tajik, meaning literally “seven rains”; in the bыlbыlik it is a metaphor for heavy crying.

The story behind Kobulmo’s song concerns the tragic death of her brother, Xushwaxtsho. He’d lent money to a village-mate who after many excuses finally agreed to repay him. The debtor invited Xushwaxtsho to wait in a small room of his house. He got a gun, a plastic bag, and an axe, distributed fireworks to the neighborhood children, and connected a live electrical wire to the door latch. Once inside, masked by the sounds of the fireworks, he shot Xushwaxtsho. The injured man staggered toward him, grabbed the live wire and died. The murderer chopped him into pieces and dropped the bagged pieces of his dismembered body into a hole he had dug ostensibly for an apricot tree. Then he planted the tree. For some time the killer claimed that Xushwaxtsho had gone to Afghanistan, but eventually he was forced to confess to the murder.

53 A version of this is the first bыlbыlik appearing in Pakhalina (1975:173; Russian translation on page 174).


This sent a wave of shock through the Wakhi community, where such violence is almost unheard of. In the context of this song, though the text of the song is hardly distinctive, the mention of Xushjon, “dear Xush,” is enough for a Wakhi familiar with the incident to recall the story and the sister’s grief.

A bыlbыlik tune operates within the interval of a minor second and is rhythmically flexible.54 The singer generally gives each syllable (if there are seven) the same length, draws out the last syllable on a single pitch, and employs her own style of delivery and embellishment.

Melody and rhythm of “sətkək bə iwət buy”:

/ 2♭  2♭    2♭   2♭ 2♭ 2♭   2♭ ——– 1 /
  e    sət – kək bə  i – wət bu ——– y

/ 2♭ 2♭  2♭ 2♭     2♭  2♭    1   /
  ar  bor ki Xыš- jon x̌a – nəm

          /2♭ ——- 1  1  /
           e  ——– e lol

/2♭  2♭     2♭ 2♭   1    2♭  1     1   2♭             /
  e    woy aft bo-ron  a – rəm ẓ̌ə ru———y

/2♭   2♭   2♭   2♭   1    1   1           /
 aft   bo- ron  a –  rəm ẓ̌ə ru—–y

          /2♭——- 1  1  /
           e ——– e lol

/2♭  2♭   2♭ 1   2♭ 1    1   1     /
  e   bыl – bыl  tar no- lə-me         

          /2♭ ——- 1  1  /
            e  ——– e lol

Kobulmo delivers the initial syllables of each line rapidly, in the manner of intoned speech. E lol is one of several generic insertions between lines and doesn’t always correspond, as in this case, with the subject of the song (the singer’s younger brother). It is rather an expression of grief and love. Each three-line unit is a complete bыlbыlik, and when women sing them

54 Mark Slobin noted the melodic focus of some falaks in Afghan Badakhshan around a minor second (1976:208). This is indeed a stressed interval, both on the Tajik and Afghan sides of the border.


together, they generally take turns leading. The other women provide owoγ̌ or vocal support. The art in bыlbыlik singing is to maintain perfect unison, on the one hand, and create delicate filigree patterns in appropriate places, on the other.

A second example of bыlbыlik (Example 6) illustrates a more measured approach to the initial syllables, allotting almost equal weight to each and intoning them without any hint of speech. Both women in this video, taken in Shitkharv village in 2012, sing in unison—varying only in ornamentation. Izatmo Salimova (on the left) creates the filigree pattern against the steady intonation of Maliknoz Ghulomiddinova (on the right).55 The image in the first line also appears in bыlbыliks other women have created.

Text and translation of “šəmolək bə kuboda” (consult the eCompanion, Example 6):

šəmolək bə kuboda
o-yi sar bə məxaṣ̌ e lol
o-yi sar-i γamboda
e o-yi sar-i γamboda e56

e bыlbыl tar nolame e lol

Dear wind, beating against the mountain
Don’t draw out a heavy sigh
A heavy sigh is grief

I, the nightingale, lament to you

In Wakhi, even the wind can be addressed in the familiar, diminutive, using the –ək suffix. Listening to the wind hitting the mountain—remember, the singers are usually in the high mountain pastures during the summer when they sing these songs—the experiencer perhaps does not want to be reminded of sadness. Is she addressing the mountain, herself, or someone else?


In Tajikistan and Afghanistan, Wakhi songs sung to the accompaniment of a frame drum, rəbob, or other instrument are called bayd (from Arabic bayt,“house,” “family,” or “verse”). In Wakhi, there is no separate word for poem. Bayd means both “poem” and “song,” and like the bыlbыlik the bayd is “told” or “sung.” Texts in praise of religious figures or on explicitly religious themes are usually performed in a different style, and in Persian rather than in Wakhi. Bыlbыliks are not considered kinds of bayd, although bыlbыlik texts can be incorporated into bayds.57

Many bayds in Tajikistan and Afghanistan are in the poetic form of the ghazal, but they can also take the form of quatrains. An excerpt of a performance I documented on my first visit

55 This very common text is found in collections of Griunberg and Steblin-Kamenskiĭ (1976a:25), Pakhalina (1975:13), and Matrobov and Mirboboev (2015:77).

56 Although “o-yi sar-i γamboda” was clearly sung, and also appears that way in several collections, I am told that the line should be “o-yi sar bə γamboda” in order to make sense. Of the collections I’ve searched, only that of Klimchitskiĭ (1936) gives the variant.

57 According to John Mock (1998:105), however, bыlbыliks are included in the category of bayd among Wakhis in Pakistan.


to Yur village, Afghanistan, in 2015 illustrates one method for incorporating a bыlbыlik text (or any three-line Wakhi poem) into the framework of a bayd. The first stanza in the excerpt is the refrain; the second is a verse consisting of the bыlbыlik discussed above, with an additional line that ties the object of the poem to the beloved represented in the refrain. One meaning of the Wakhi verb γ̌ir-/γ̌ird- is “go around a place” or “turn around,” but in this context it also means “devote oneself” or “sacrifice oneself.” I have translated the repeating Wakhi term with differing English ones to capture some of the nuances in each context. The style of rendition, the addition of a line, and the “finalization” with the refrain “I am obsessed with my ‘Leaf Flower’” serve to mark this as a bayd and not a bыlbыlik, despite their shared verse content. Note that a one-line refrain completes the quatrain and that a separate refrain quatrain is inserted between each verse. The recording begins with the refrain quatrain. Bayd singers incorporate bыlbыlik texts into their songs in several ways, often spontaneously.58

Excerpt from Wakhi bayd sung by Baxmal, Mamadsharf, Shirxon, and Yusuf Ali in Yur village, Upper Wakhan, Afghanistan (August 6, 2015; consult the eCompanion, Example 7):

wuz x̌ə dələr γ̌irəm
x̌ə bыlbыlər γ̌irəm
wuz yaw junər γ̌irəm
yaw tatхunər γ̌irəm

ti satki iwət buy
ar bor taw yod carəm
aftboron rəm ẓ̌e ruy
bargigыlər γ̌irəm

I sacrifice myself for my beloved
I circle around my nightingale
I wander about for the sake of her soul
I devote myself to her father’s house

Your beads, one after the other
Whenever I think of you
Streams of tears run down my face
I am obsessed with my “Leaf Flower”

In the context of the bayd, this bыlbыlik is no longer the lament of, say, a mother or a sister for a close family member who has died or traveled far away, but rather that of a love-crazed young man—yes, the beloved could be far away in fact, but it is just as likely that the beloved is far away in a metaphorical sense. And here the lover identifies the beads as those of the beloved—so rather than contemplating his far-away loved ones while counting rosaries, he may be looking at her actual necklace sitting out on a surface.

58 John Mock illustrates this same process in a bыlbыlik text adopted by the singer Fazal Rahman of Chipursan valley in Hunza (1998:130). The refrain is essentially the same as that in the present example, however the tune Fazal Rahman uses (according to my own recordings of him) is of a type used in Northern Pakistan and not normally in Afghanistan (although Afghan Wakhis do sing tunes they identify as originating in Pakistan). Mock was convinced that Fazal Rahman obtained the texts via photocopies of Leonovich Griunberg and I. M. Steblin-Kamenskiĭ’s published collection of bыlbыliks in their monumental work on the Wakhi language from 1976 (Griunberg and Steblin-Kamenskiĭ 1976a). This may have been the case, but Fazal Rahman told me that he obtained the text from a handwritten notebook in the Persian script with the heading, Bыlbыlik. It is possible that the texts were passed down in writing via the Persian script, but it is also possible that someone a generation ago transliterated from the romanized Soviet text into Persian. Mock is not correct in saying that Fazal Rahman “changed . . . the order in which the stanzas are sung” (1998:128), because bыlbыliks are self-contained units; they are not stanzas as such, and in any case can be sung in any order.


On this 2015 recording, the singers are all disciples of the most respected Wakhi poet-singer in Afghanistan at the moment, Daulatsho. They sing to a stock tune and accompany themselves on dorya, rəbob, and ghijak. The bыlbыlik text begins at 0:14 in Example 7 (consult the eCompanion).

Notation of verse excerpt, Example 7, derived from “sətkək bə iwət buy”:

/3♭ 2♭ 1  2♭ .   7     1 .  /
 ti  sat-ki i –     wət buy

/3♭ 2♭  1    2♭   .  7   1 .  /
 ar bor taw yo-d carəm

/3♭ 2♭   1    2♭.   7  1 .  /
 aft-bo-ron rəm  ẓ̌e ruy

/3♭  2♭  1    2♭.  7   1 .  /  
 bar-gi  gы-lər  γ̌i-rəm

The refrain is a common poetic mold found in Tajikistan, Afghanistan, and Pakistan that can accommodate different melodies. It need not take verses that originate from a bыlbыlik but can use any tristich of a similar length. For instance, in Pətr, the Wakhi village closest to Sultan Ishkoshim (Afghanistan), there are no Wakhi composers, and for this reason, according to Ghulam Nasir, singers such as himself know only a few Wakhi songs (they sing primarily in Dari). One of those songs was a version of the bayd above, in this case accompanied by harmonium and dorya. I recorded this in the Afghan bazaar, a no-man’s land between Tajikistan and Afghanistan near Ishkoshim next to the bridge-border, in 2013. Being at a national border I could only video record inside an empty godown, not in the open where the market and the border security were visible.

Excerpt from “wuz x̌ə dələr γ̌irəm,” performed by Ghulam Nasir (vocal and harmonium) and Amonbek (dorya and vocal), transcribed (consult the eCompanion, Example 8):

wuz x̌ə dələr γ̌irəm
x̌ə bыlbыlar γ̌irəm
wuz yaw junər γ̌irəm
yaw tatxunər γ̌irəm

ẓ̌ə gimjoni da sarsin
dam kaf tiloyi dərbin

I sacrifice myself for my beloved
I circle around my nightingale
I wander about for the sake of her soul
I devote myself to her father’s house

My lover in the mountain green59
with golden binoculars in hand

59 In this case, Sarsin is the name of a particular mountain green in Afghanistan, but the referent of this word varies by location.


iloy kartət maẓ̌i win60
wuz yaw junər γ̌irəm

Look at me one time!
I wander about for the sake of her soul


/ 3♭   2♭  3♭ 2♭ 1   7    1  /
 wuz  x̌ə  də-lər γ̌i-rəm-e

(The other three lines of the refrain use variations on this melody.)


/ 2♭  3♭  3♭ 3♭  3♭  3♭  3♭  . /
  ẓ̌ə  gim jo-ni  da sar-sin 

/ 2♭    3♭   2♭ 2♭ 3♭  2♭   2♭  . /
  dam kaf  ti- lo  yi dər  bin

/ 3♭    3♭  3♭   3♭   3♭   2♭   1  . /
   i   – loy kar-tət  ma-ẓ̌i  win

/ 3♭     2♭   1   7    1  1   7 /
 wuz  yaw  ju-nər γ̌i-rə-me

Clearly this version draws on the same melodic resources as the version from Yur; however, here the verse is differentiated from the refrain through the use of 3♭ as a reciting tone on the first line. In this performance, none of the verses were based on bыlbыliks, but each of them was formally similar to a bыlbыlik: a tristich with each line containing seven syllables. However the tristich rhyme scheme in this example is A-A-A and not A-B-A. The fourth line returns to the refrain—this time using the third line of the refrain quatrain (and not a different line, as in our previous example)—and ending on the seventh below the tonic as a lead-in to the refrain quatrain. Another version of this setting is provided from Vrang, where a group performs this song in national costume as an accompaniment for a girl’s dance (consult the eCompanion, Example 9).61 This orchestral version is the type one would encounter when Wakhis are featured in variety shows or at Ismaili cultural events. Whatever the version, the four-beat frame for each

60 There is a problem with this line. My native-speaking Wakhi collaborators indicated the word is wind here, which would put it all in the past tense and make the tense agree with the first part of the line “did one time.” However, the singer didn’t pronounce the “d,” and if he did, he would have broken the rhyme.

61 Recording details: Oshurmamad Sabzaliev (balandzikom), Shohqirghiz Mulloev (rubobi badaxshi), Azizxon Gulmirzoev (accordion), Imronsho Mirzoev (rubobi badaxshi), Bodurxon Rahmatshoev (chang), Mirboz Rahmatshoev Mirboz (balandzikom), Tohir Odinabekov (dorya), Nilufari Shoqirghiz (dancer). Performed for the author in Vrang, October 26, 2012.


line lends itself to the syllable structure of bыlbыlik and other traditional three-line Wakhi poetic texts.

As a general matter, Wakhi musicians see the role of musical instruments as adding interest or excitement to the rendition, but it doesn’t, for them, fundamentally alter a song to use one or another accompanying instrument. Weddings call for louder accompaniment and ideally for amplification. Electric instruments are universally seen as modern, but individuals vary in their aesthetic preferences for electric versus acoustic instruments. The balandzikom is used almost exclusively for maddo singing. The drumming patterns depend to an extent on the melody, and choices of melody can vary according to the mood of the performer and the context. However, Central Asian music in particular is known for taking a single melodic idea and creating contrast in a suite of successive items by altering the rhythmic cycle or zarb. The appropriateness of one zarb or another may depend on such features as line length or poetic meter (in Persian), as well as its perceived suitability to the meaning of the text (a lively dance groove would not be suitable for a song contemplating the transience of human life, for instance).

New Wakhi Songs

In 1992, the Vrang native Qurbonsho (born c. 1964) introduced a new style of Wakhi song composition that drew on contemporary musical styles of Tajikistan and Afghanistan, often incorporating Pamiri rhythmic grooves. Many of today’s Wakhi poets and poet-singers in Tajikistan, Afghanistan, and Russia view him as tantamount to an ustod (master/teacher) because he opened up a world of Wakhi creativity unknown to living generations of Wakhis. Over the course of his career, his “muses” have been two (imaginary) women named Sitora and Sharora, at times considered as one single beloved (“Richard, for one heart there is one beloved, right?” Qurbonsho would ask me); at other times, two (“One wears traditional dress, and the other wears modern dress,” he would say). Cassettes and MP3s of his recordings circulate widely in the Wakhi world, and his tunes and lines of poetry wind up in the songs of others. In 2016, Qurbonsho was, he thought, bringing his life as a composer to a close. Sitora and Sharora had, in the end, failed him, and he was going to stop composing. Instead, by inhabiting his second persona by the name of Arghovon, he would give up his hermit-like existence and take up the project of traveling the world, singing his songs.

Now several Wakhi pop bands and many modern Wakhi poets perform using a variety of styles in Wakhi—that is, there is no single “Wakhi music.” Given Qurbonsho’s importance in the efflorescence of music in the Wakhi language, it is worth examining one bayd of his own. His compositions are products of his lifetime goal to raise the level of Wakhi artistry. Qurbonsho does not want to break from the traditional songs of the past, but wants rather to pull from the future and from the past to make Wakhi songs worthy of today. The following bayd (Example 10), which he composed in 2016, expressed a sense that Qurbonsho’s poetic love affair was coming to an end: he was ready to shed his identity as Qurbonsho and take on the persona of Arghovon. Qurbonsho employs the Persian poetic form of the rubā‘ī (here with the rhyme scheme A-A-B-A) and a symbolic vocabulary common to the Persianate world: the moth and the flame as representative of the lover and the beloved (and in mystical poetry, the individual in


relation to God); the idea of love as burning, scorching, and destroying; and the special role of songbirds (not here the nightingale).

Text and translation of Qurbonsho’s bayd, “nəxod ki tar γam nast” (July 31, 2016; consult the eCompanion, Example 10):


nəxod ki tar γam nast
də ti dida γal nam nast
nəxod ki wuzəm parwonaət
marət ẓ̌ə šam nast

Don’t you really care?
There are still no tears in your eyes
Is it possible that I am your moth,
but that you are not my candle?


tawiṣ̌ə wuz qыwəm
bəroi tawṣ̌ə nыwəm
nəxod ki bifoida ruyi dənyoṣ̌
wuz x̌ati ϑыwəm

I’m calling out for you!
I’m crying for you!
Could it be that, roaming the world uselessly,
I’m burning myself?


čizər yəm nowaxtí
kərẓ̌əpčiš ẓ̌axti
nəxod ki ẓ̌ə gыli arγəwon
wudg naγ̌di ṣ̌paxti

Why, unexpectedly,
do the magpies called out “ẓ̌ax ẓ̌ax”?
Is it really possible, that my Arghovon flower
has blossomed tonight?

In stanza 1, Qurbonsho addresses Sitora and Sharora, suggesting that they are not the ones he is meant to be pursuing, even though, from their point of view, he is acting like their “moth.” In stanza 2, he seems to suggest that it is not the beloved that is causing him to be scorched, but his very act of calling out and crying for his beloved—that is, his own acts of singing. Stanza 3 alludes to the Wakhi belief that when magpies make the sound “ẓ̌ax ẓ̌ax,” it’s a sign of good luck. Asking why the magpies are making these sounds sets up the ambiguity in lines 3 and 4. “Arghovon’s flower” could, of course, be the beloved, and so the magpies are crowing to celebrate the blossoming of his love. But, as Qurbonsho refers to himself as Arghovon here, and given the context of this poem in Qurbonsho’s poetic life history, these lines suggest that he is leaving behind Qurbonsho and the troubles of Sitora and Sharora, his “useless” activities. Now he is blossoming as Arghovon (which is also the name of a flower)—in the new phase of his life as a performer.

Virtually all of the songs Qurbonsho sings in Wakhi can be sung to more than one tune and with the accompaniment of more than one instrument. So in analyzing this particular example, we are addressing only one possible manifestation of the song. At the main kind of venue for his public performances, tuys (celebrations associated with weddings and circumcisions), Qurbonsho will usually accompany himself on synthesizer with drum machine, or on accordion (in the latter case, he’ll have backup singers and a frame-drum player).


Fig. 13. Qurbonsho performing at a wedding (tuy) in Vrang, August, 2019 (photo by Richard K. Wolf).

In Example 10 he is playing on a fancifully styled rəbob, gifted to him by me and made by a nearby instrument maker, Imronsho Mirzoev. He uses the rhythmic pattern (zarb) called rapo, which consists of four beats, each beat distinguished by a particular form of attack or sequence of attacks:62


In the following notation, beat 1 conforms to the first beat of the pattern as Qurbonsho teaches it. The beginning of the strumming pattern does not need to correspond to the beginning of a line of text—indeed, in my experience, it usually doesn’t. The rhythmic placement of some of the syllables is approximate. For example, in the first line, the second word in both nəxod ki and tar γam comes earlier—giving the impression of a multisyllable word (an amphibrach in the first case and a trochee in the second) rather than separate words. Lining up the melody in this way shows how line 3—the non-rhyming line—diverges from the others. (The use of rhyming lines to maintain a basic melody or range and non-rhyming lines to diverge is common in Central and South Asia.) The tonic, which does not appear in the vocal line until the end of the whole song, is iterated constantly only by the rəbob (its three courses are tuned: 5ve 2 – 5 5 – 8 8).

62 Qurbonsho gives more-or-less equal duration to each beat, but in the wider Pamirs, performers sometimes swing the rhythmic pattern considerably. The strumming pattern is an essential (though also variable) part of the definition of the zarb, wherever it appears.


The vowel [e] introducing lines 1 and 3 comes in earlier and on the beat for line 3, initiating a deliberate and stretched-out delivery of syllables in that line. This leads to a melodic climax on the words “Is it really possible?” (5 and 6♭), followed by an undulating pattern downward on the words “that I am your moth.” The melody and text in this line stretches beyond the length of lines 1 and 2 by two full beats. The fourth and final line of this stanza is shorter, and, as a rhyming line, returns to the initial range. While I would not wish to claim that this fluttering descent is really meant to imitate a moth, we can observe that in each of the three stanzas there is a key phrase highlighted with the melody at that point, following the question “Is it really possible?” The musical drama of line 3 is also set up by the repetition of the first pair of lines in each stanza.

The syllabic structure in lines 1, 2, and 4 is closely in keeping with much traditional Wakhi singing—the first is six syllables, rounded out to seven with the vocalic [e] at the beginning. Line 2 is seven syllables. And line 4 starts late, as if making room after the long line 3, with five syllables. Counter to what Buddruss and Wiehler-Schneider (1978:108) observed in the Wakhi songs they examined in Pakistan (but similar to what we discovered in some of the Kota examples), the “period” does not get extended simply because more syllables need to be accommodated. Qurbonsho’s steady maintenance of the zarb allows him to play with line length and melodic placement without losing the periodicity of the cycle.

Transcription of the first stanza of the baydnəxod ki tar γam nast,” performed by Qurbonsho (July 31, 2016; consult the eCompanion, Example 10):

          4              1       2              3          4          1            2            3

/         3♭    3♭     3♭    3♭            2♭        3♭        3♭           .             .        /
/         e      nə-    xod  ki             tar       γam      nast        .             .        /

/         3♭     3♭     3♭    3♭             2♭       3♭        3♭           .             5       /
/         də    ti       di-   da             γal      nam      nast        .             e       /

(The couplet above is repeated.)

/         5               6♭       .             4  4      3♭        3♭ 3♭    4  3♭      4  2♭   /
/         nə             xod                  ki-wu   zəm     par-wo- na —— ət       /

/           .              .        3♭ 3♭      2♭        3♭        .            2♭           .         /
/           .              .        ma-rət    ẓ̌ə         šam     .            nast         .         /

The same melody and line repetition structure is used for each of the three stanzas, and in the final stanza Qurbonsho repeats the last line and resolves the lowered second degree to the tonic on the vowel [e]. This form of cadence derives from the singing of Ismaili religious song,


called qasoid or maddo, and has been widely adopted in the music of Ismaili musicians in the Pamirs.

Deformation in a Traditional Genre

Returning to the bыlbыlik, one might ask whether traditional genres leave room for innovation. In other words, must all creativity be channeled within the existing generic frameworks as commonly understood? To address this, we look at a second composition of Kobulmo’s, again regarding the tragic death of her brother. In the previous bыlbыlik, she had personalized it by inserting her brother’s name. This bыlbыlik goes a step further.

Text and translation of Kobulmo’s “bыlbыlik” (consult the eCompanion, Example 11):

ey Xəšǰoni noziyon e: lol
ey woy dəṣ̌man tawi nosti pərarmon e:

ey Nišastaǰon vit ayron e:
e kumər ẓ̌ə širin tatjon e:
e woy cart maẓ̌i məktab rəwon e:
e woy cart maẓ̌i məktab rəwon e: lol
e bыlbыl tar noləme:: lol

O my indulged Xushjon!
Woe, the enemy destroyed you, with all your
          hopes and dreams
Oh, Nishastajon became shocked,
“Where is my sweet dear father,”
Woe! “to send me off to school?”
Woe! “to send me off to school?”

I, the nightingale, lament to you

Kobulmo addresses this poem-song to Xushjon, regarding the reaction of his daughter, Nishastajon. As she recounts it,

In this bыlbыlik I said that my brother used to take his daughter to school because his children were the light of his life. People would say, “But you have no son, only three daughters!” To which he’d reply, “to me my daughters are my sons,” adding jokingly, “I will make my ‘sons’ march and send them to the army. When I take my daughter to school, I’ll invite all my relatives and make a parade.” The enemy killed him and he didn’t realize his wish, but my brothers and I carried out his wishes: we brought her to school in a parade. But no matter how much we danced and had fun the girl didn’t smile even for a moment. Her father’s “taste” was not there. No matter how much we cajoled her, she didn’t laugh even once. She was looking around asking, “where is my father?”

The second line of this bыlbыlik refers to the killer, who dashed the hopes of Xushjon. The child is shocked at the events and looks for her dad, wondering who will take her to school.

The style and structure of this song are not those of a typical bыlbыlik. Rather than being open-ended, this text alludes to a particular event and names individuals. It is five rather than three lines long, and the syllables come out in even more of a burst than in her performance of a straightforward bыlbыlik. Line 2 in particular has to accommodate nine syllables (if we leave out the vocative e woy or woe at the beginning). The overall tune is very close to that of a bыlbыlik:


it focuses on the very low flat second degree, touching the first degree once in the middle of each line, and resolving on it. But there are subtle differences regarding the antecedent-consequent structure of the melody and the use of repetition and insertions to articulate the usual parts of the stanza.

The first line approximates the grouping of typical bыlbыlik lines into 3 + 4 syllables, here “ey Xəš-jo-” plus “-ni no-zi-yon.” The line nevertheless comes off as subtly different owing to the addition of e lol to the first line (the downward arrow indicates a spoken drop in pitch). The second line of a bыlbыlik will usually have e lol or some similar extension, and the line will not rhyme. Instead we have a long rhyming line, and just a hint of a drop to the sixth below at the end with no e lol. Lines 3-5, which also rhyme, have seven central syllables (I am ignoring the introductory [e] and concluding [e] or e lol), but without the repeating third line, which forms a melodic conclusion in the bыlbыlik, this does not satisfy the expectations for a bыlbыlik. Overall, the melody and its articulation are similar to those of the laments Wakhi women perform at funerals—genres called rəboi and nawhagari. But those laments are either in Tajik (in the case of the rəboi) or a mixture of Tajik and Wakhi (in the case of nawhagari), and their poetic forms are different. In the repetition of the fifth line, Kobulmo lands confidently on the lowered sixth degree below the tonic. This is perhaps the strongest marker of the difference between a bыlbыlik and a rəboi. And yet, the three women conclude the stanza with the bыlbыlik refrain, leaving no doubt that they do, in fact, wish to represent this as a bыlbыlik.

Transcription of Kobulmo’s “bыlbыlik” (consult the eCompanion, Example 11):

1   2♭  2♭     1  1  2♭2♭          1    1 ↘ 
ey Xəšjo . . ni noziyo———–n e- lol

2♭   2♭    2♭   2♭    2♭   2♭   1     2♭   2♭   2♭ 2♭             1   1       (6)
ey   woy dəṣ̌-man ta-  wi   nos-ti    pər-armo————n e——

2♭   2♭   2♭   2♭   2♭  1     2♭ 2♭                     1    1
ey   Ni- šas- ta-  jon  vi-t ay-ro——————n e——-

2♭   2♭   2♭    2♭ 2♭   2♭   1    2♭                 1     1   
e     ku-  mər ẓ̌ə  ši-  rin  tat-jo—————–n e——–                         

2♭   2♭    2♭    2♭  2♭   1   1      2♭   2♭ 2♭                 1        1        (6♭)
e     woy cart ma- ẓ̌i    .   mək-tab  rəwo——————-n e—–

2♭   2♭     2♭    2♭    2♭ 1   1       2♭   2♭2♭                   1      1          6♭
e     woy  cart  ma-  ẓ̌i  .   mək- tab  rəwo——————-n e—— lol

2♭    2♭    2♭    1    2♭ 1     2♭  1                 2♭  1 1
e      bыl  bыl   .    tar  no  lə    m e————- e lol



As we have seen, a vigorous tradition of contemporary bayd composition, spurred on in part by Qurbonsho, does not rely on the bыlbыlik for its poetic basis—drawing instead from the poetic forms that circulate widely in Persian, Urdu, and related languages of Central and South Asia. But this does not diminish the significance of the bыlbыlik as a site in which Wakhi language and emotion crystalize into a single form. Qurbonsho and others, however, question whether the bыlbыliks handed down from generations past have heartfelt meaning for those who sing them today. In Qurbonsho’s view, life conditions have improved to such an extent that singers cannot draw from hardship in their lives to generate true performances. Bыlbыliks become simply songs. Qurbonsho claims never to have heard a bыlbыlik as it’s meant to be sung—from the bottom of one’s heart. We should be skeptical of such a claim, for how could Qurbonsho know the real bыlbыlik if he has never heard it? Still, his claim rings true as we see young girls don their folkloric white satiny costumes and perform bыlbыlik on stage, drawing out the self-conscious refrain “I, the bыlbыlik, lament to you.” But then there are singers like Kobulmo, who will never be able to rid herself of grief, no matter how much she sings; the mere insertion of her brother’s name into the most common of bыlbыliks renders it intensely personal.


What light do these examples shed on the poem-song? Tajik and Afghan musicians commonly use the Persian (ultimately Arabic-derived) term qālab, “model” or “mold,” in reference to the form of a melody, piece, poem, and so forth. The qālab could be the basic note-structure of a melody, including the repeating strumming pattern if it is played on a musical instrument, and it can mean the prosodic rules of a poem. As musicians sometimes put it, after mastering the qālab, it is easy to pour in personalized content—whether that means ornamentation, the correct placement of text within the rhythmic structure, or the invention of new text. The use of qālab thus has an implication for expression. Mastery of form allows one to focus on the message itself as well as on the delivery of the message—even if the message is one of melody without words. The qālab itself determines, or is at least supposed to determine, the emotional-moral framework for the content—thus musicians will say such and such text is compatible with such and such a musical setting and not compatible with another. But in other cases, the qālab can shift our understanding of the message—as it did in the case of the bыlbыlik text framed within a bayd focusing on a worldly beloved.

The poem-song is one in which the qālab and the textual message it contains are in harmony. An effective poem-song in performance requires something extra—this may involve overt facial gestures or sounds of emotion, imaginative treatment of the melody or rhythm, but it also may involve something external to the performance itself—the knowledge of context and narrative that extends beyond the words of a song.

I’d like to close by considering the role of form—call it qālab or by any other name—in relation to the question of endangered languages. The fourteenth-century Tunisian philosopher-historian Ibn Khaldun, often credited as the first sociologist and anthropologist, frequently invokes the image of a loom and a mold in his famous work, the Muqaddimah. Uslūb, or


“method,” Khaldun writes (1970:373-74; translation from Rosenthal 1967:376; see also Blum 1975:223),

is used to refer to a mental form for metrical word combinations which is universal in the sense of conforming with any particular word combination. This form is abstracted by the mind from the most prominent individual word combinations and given a place in the imagination comparable to a mold or loom. Word combinations that Arabs consider [proper] are then selected and packed . . . into (that form), just as the builder does with the mold or the weaver with the loom. Eventually, the mold is sufficiently widened to admit the word combinations that fully express what one wants to express.

While it would be a mistake to say that people who speak endangered languages lack theories of prosody and music, it is likely that such theories are not articulated in the explicit forms we find in the codified traditions of Tamil, Sanskrit, Persian, and Arabic, or the methods of singing the classical music of South India, Iran, and Iraq. Nevertheless, highly complex musico-poetic traditions exist even in languages with a small number of speakers.63

The methods for composing such poetry are not given in grammars or dictionaries but are rather imbibed through existing models, which become looms or molds in Khaldun’s terms. The key here, I think, is the idea that the molds can be “sufficiently widened” to accommodate what the poet wishes to express. The cases we’ve seen are meant to be examples of possibility rather than the basis for generalization. I see the poem-song as a kind of mutually implicating mold: the “musical” side shapes, without fully determining, the kinds of verbal sounds the poem-song contains—and also helps the imagination fill in what might not be said explicitly. The textual side, through parallelisms, juxtaposition, terseness, or prolixity, can work within, create, or, as in the last case, break the usual molds. The fact that in many traditions, originary versions of any particular text or melody cannot be recovered or never existed in the first place in no way weakens the illusive mold—rather it leaves a great responsibility with the builder every time she builds, the weaver every time he weaves—rather than with the creator of the loom, the mold, or any product from them.

Changing life experiences of singers in the modern world can, perhaps, turn the woven products from the looms into lifeless objects, but we can’t be sure of it. While new experiences continue to stimulate poet-singers to press their speech into molds they have abstracted from the modern media, from musico-poetic exemplars in other languages, and from as-yet unrealized possibilities in their own languages, they enrich their languages and inspire others to do so as well. Ibn Khaldun wrote, “The desire to press speech into the molds of poetry sharpens the mind” (Rosenthal 1967:375). Perhaps we can say, “the desire to press poem-songs into action sharpens the language.”

Harvard University

63 Take for instance the Todas of the Nilgiris, a tribal group of about 1,200 who live in small hamlets alongside their Kota neighbors. Murray B. Emeneau wrote a thousand-page book on their verbal art (1971), which continues to occupy the minds of contemporary linguists such as Peri Bhaskararao (Nara and Bhaskararao 2003).


Guide to Transliteration and Musical Notation:

Transliterations follow the ALA-LC Romanization Schemes for non-Roman scripts or draw on the same principles (sometimes simplified: ligatures are not used for Russian here). Retroflexes are indicated by underdots (ṭ, ḍ, ṛ, ḷ, ṣ̌, ẓ̌, ṇ), alveolars are indicated by underscores (ṟ, ḏ, ṯ, ṉ) (except the [ḻ] in Tamil), long vowels are indicated by macrons. Tajik, Dari, and Farsi are all versions of Persian, however Tajik is currently written in Cyrillic and is transliterated here from the Cyrillic. The Tajik [o] is equivalent to the Persian [ā], and [e] is equivalent to the Persian [i], in the ALA-LC system. In Kota, the difference between [ṛ] and [ḍ] is length—the latter is held longer than the former. In Kota, the phoneme represented by [c] varies: English [s], [ch], and [ts] are all possible depending on speaker. Wakhi is transliterated according to the system established by Griunberg and Steblin-Kamenskiĭ 1976a, which uses a few Greek and Cyrillic characters as well as special signs. [γ] is a voiced velar fricative ([gh] in the ALA-LC tables for Persian), and [x] is an unvoiced velar fricative ([kh] in Persian). The same characters with a hachek, [γ̌] and [x̌], are pronounced with less vocal friction than their counterparts without the hachek. Wakhi [š] is like English [sh], but pronounced toward the front of the mouth (somewhat like the Russian [щ]), and [ṣ̌] is a retroflex [sh] (similar to [ш] in Russian). [ϑ] (theta) is pronounced like [th] in “three.” [δ] (delta) is pronounced like [th] in “the.” [ž] is pronounced like [j] in French “je.” [ẓ̌] is a retroflex version of the same sound. [č] is English [ch] pronounced in the front of the mouth, while [č̣] is a retroflex version of the same sound. [ʒ] is approximately [dz], as in “two heads are better than one.” [c] alone in Wakhi is [ts]. [ə] and [ы] are very similar sounding vowels, and even though in some words they are seemingly interchangeable (depending on speaker and dialect), they are nevertheless separate phonemes. [ə] is close to the [u] in “but.” [ы] is pronounced like the letter represented by that character in Russian: position your lips to say “oo,” stiffen them, and say “ee.”

The musical notation indicates scale degrees taking the tonic as C. So, 1 is C, 2 is D, 3 is E, and so forth. Flats are applied accordingly to indicate lowered scale degrees, so 3♭ is E♭. Rests and prolongations are indicated with dots, and double speed is indicated with an underline. Not all the music falls into “measures” in a Western technical sense. I have indicated relevant metrical units using slashes. Further details regarding notation appear alongside the descriptions of particular musical examples. In all cases, I recommend listening to the recordings provided with this article.


I delivered a version of the present article as a keynote address at a symposium on the music and poetry of Kashmir, March 20, 2020, University of North Texas; in the wake of COVID-19, a scaled-down version of the symposium was held on Zoom. I thank Sadaf Munshi and Vivek Virani for inviting me to speak at the symposium and inspiring me to take my work in a comparative direction. I express my thanks to the performers whose singing and compositions I cite in this article. I am especially grateful to Qurbonsho for the many hours he spent discussing his Wakhi poetry with me and teaching me to sing it. Zohir Piltaboev and Chorshanbe


Goibnazarov have provided me invaluable help in my efforts to learn Wakhi, hosted me in their family homes in Wakhan, and facilitated my encounters with many singers. R. Kamatn (also known as Duryodana), who became my Kota research assistant in 1991, has long been my friend and consultant. Most of the transcriptions of the Kota songs in this article were the result of our careful listening together and his acts of follow-up with others in the village when words or meanings were not clear. I also wish to thank the anonymous peer reviewer for this journal, as well as Stephen Blum and Gregory Nagy, for their helpful comments. David Shulman inspired me to look further regarding the musicality of Tamil and directed me to useful sources in this regard. Martha Selby read a late draft of the article with a critical eye on my discussion of Tamil poetics. To all these friends and colleagues, I owe a debt of gratitude.


Example 1

A. K. Rangan (right) recounts his life story (see pp. 129-31)

Example 2

Erdd tigḷe,” a song composed by A. K. Rangan, is sung by P. Kamatn (see pp. 131-35)

Example 3

Mādi āṭḷ, as recorded by the Gramophone Company of Calcutta under the auspices of the Linguistic Survey of India in 1922 (see pp. 137-47)

Example 4

Mādi āṭḷ, as sung by Maṇimāla in Kurgoj village, India, in 2019 (see pp. 137-47)

Example 5

Ra’no Rojibekova, Kobulmo Rajabova, and Bibiniso Shogunbekova sing “sətkək bə iwət buy” in Vrang, Tajikistan, in 2014 (see pp. 152-53)

Example 6

Izatmo Salimova (left) and Maliknoz Ghulomiddinova (right) sing “šəmolək bə kuboda” in Shitkharv village, Tajikistan, in 2012 (see p. 154):

Example 7

Baxmal, Mamadsharf, Shirxon, and Yusuf Ali sing a Wakhi bayd in Yur village, Upper Wakhan, Afghanistan, in 2015 (see pp. 155-56)

Example 8

Ghulam Nasir (vocal and harmonium) and Amonbek (dorya and vocal) perform “wuz x̌ə dələr γ̌irəm” near Ishkoshim in 2013 (see pp. 156-57)

Example 9

Oshurmamad Sabzaliev (balandzikom), Shohqirghiz Mulloev (rubobi badaxshi), Azizxon Gulmirzoev (accordion), Imronsho Mirzoev (rubobi badaxshi), Bodurxon Rahmatshoev (chang), Mirboz Rahmatshoev Mirboz (balandzikom), Tohir Odinabekov (dorya), and Nilufari Shoqirghiz (dancer) perform a Wakhi bayd in Vrang, Tajikistan, in 2012 (see p. 157)

Example 10

Qurbonsho performs“nəxod ki tar γam nast” in 2016 (see pp. 158-61)

Example 11

Kobulmo (center) sings a bыlbыlik in Vrang, Tajikistan (see pp. 162-63):


Agawu 1988
V. Kofi Agawu. “Tone and Tune: The Evidence for Northern Ewe Music.” Africa, 58.2:127-46.

Allen 1973
W. Sidney Allen. Accent and Rhythm: Prosodic Features of Latin and Greek: A Study in Theory and Reconstruction. Cambridge: Cambridge University Press.

Ate 1984
Lynn Ate. “Literary Meter vs. Rhythmic Beat in Tamil Alvar Literature.” Journal of South Asian Literature, 19.2:1-7.

Bakhtin 1986
M. M. Bakhtin. “The Problem of Speech Genres.” In Speech Genres and Other Late Essays. Ed. by Caryl Emerson and Michael Holquist. Trans. by Vern W. McGee. Austin: University of Texas Press. pp. 60-102.

Barker 1989
Andrew Barker, ed. Greek Musical Writings. Vol. 2. Cambridge: Cambridge University Press.

Bickford 2007
Tyler Bickford. “Music of Poetry and Poetry of Song: Expressivity and Grammar in Vocal Performance.” Ethnomusicology, 51.3:439-76.

Bishop and Litch 2000
R. A. Bishop and J. A. Litch. “Malaria at High Altitude.” Journal of Travel Medicine, 7.3:157-58.

Blum 1975
Stephen Blum. “Towards a Social History of Musicological Technique.” Ethnomusicology, 19.2:207-31.

Blum 2013
______. “Foundations of Musical Knowledge in the Muslim World.” In The Cambridge History of World Music. Ed. by Philip V. Bohlman. Cambridge: Cambridge University Press. pp. 103-24.


Boyce 1957
Mary Boyce. “The Parthian Gōsān and Iranian Minstrel Tradition.” Journal of the Royal Asiatic Society, 89.1-2:10-45.

Buddruss and Wiehler-Schneider 1978
Georg Buddruss and Sigrun Wiehler-Schneider. “Wakhi-Lieder aus Hunza.” Jahrbuch für Musikalische Volks- und Völkerkunde, 9:89-110.

Burrow and Emeneau 1984
T. Burrow and M. B. Emeneau. A Dravidian Etymological Dictionary. 2nd ed.Oxford: Clarendon Press.

Crosson 2008
Seán Crosson. “The Given Note”: Traditional Music and Modern Irish Poetry. Newcastle, UK: Cambridge Scholars.

Emeneau 1971
M. B. Emeneau. Toda Songs. Oxford: Clarendon Press.

Emeneau 1995
______. “The Palatalizing Rule in Tamil-Malayalam and Telugu.” Journal of the American Oriental Society, 115.3:401-09.

Farmer 1913-36
George Henry Farmer. “Mūsīḳī.” In Encyclopaedia of Islam, First Edition (1913-1936). Ed. by M. Th. Houtsma, T. W. Arnold, R. Basset, and R. Hartmann. Leiden: Brill.

Faudree 2012
Paja Faudree. “Music, Language, and Texts: Sound and Semiotic Ethnography.” Annual Review of Anthropology,41.1: 519-36.

Faudree 2013
______. Singing for the Dead: The Politics of Indigenous Revival in Mexico. Durham, NC: Duke University Press.

Foley 2002
John Miles Foley. How to Read an Oral Poem. Urbana: University of Illinois Press.

Fox 2004
Aaron A. Fox. Real Country: Music and Language in Working-Class Culture. Durham, NC: Duke University Press.

Ghosh 1950
Manomohan Ghosh, trans. The Nāṭyaśāstra: A Treatise on Ancient Indian Dramaturgy and Histrionics, Ascribed to Bharata-Muni. 2 vols. Bibliotheca Indica, 272. Calcutta: Asiatic Society.

Gramophone Company of Calcutta n.d.
Gramophone Company of Calcutta. “Kōṭa. Song:—Story of Mathi.” Records of the Languages and Dialects of the Madras Presidency. Recorded c. 1922 and documented in Gravely 1927.

Gravely 1927
F. H. Gravely. Gramophone Records of the Languages and Dialects of the Madras Presidency: Text of Passages. Madras: Government Press.


Griunberg and Steblin-Kamenskiĭ  1976a
Aleksandr Leonovich Griunberg and I. M. Steblin-Kamenskiĭ. Языки Восточного Гиндукуша. Ваханский Язык: Тексты, Словарь, грамматический Очерк. (Languages of the Eastern Hindu Kush. The Wakhi Language: Texts, Dictionary, Grammar.) Moscow: Nauka.

Griunberg and Steblin-Kamenskiĭ 1976b
______. Сказки Народов Памира. (Tales of the Pamir People.) Moscow: Nauka.

Guralnick 2009
Elissa S. Guralnick. “Sound and Song in Poetry: Music as a Form of Explication.” English Language Notes, 47.1:173-80.

Ḥayyīm 1934-36
Sulaymān Ḥayyīm. New Persian-English Dictionary: Complete and Modern, Designed to Give the English Meanings of over 50,000 Words, Terms, Idioms, and Proverbs in the Persian Language, as well as the Transliteration of the Words in English Characters. Together with a Sufficient Treatment of all the Grammatical Features of the Persian Language. Tehran: Librarie-imprimerie Béroukhim.

Jakobson 1960
Roman Jakobson. “Closing Statement: Linguistics and Poetics.” In Style in Language. Ed. by Thomas A. Sebeok. Cambridge, MA: Technology Press of Massachusetts Institute of Technology. pp. 350-77.

Jakobson 1978
______. Six Lectures on Sound and Meaning. Cambridge, MA: MIT Press.

Karomatov et al. 2010
F. M. Karomatov, Nizom Nurjonov, and B. Kabilova. Музыкальное искусство Памира. (Musical Art of the Pamir.) 5 vols. Bishkek, Kyrgyzstan: University of Central Asia.

Khaldūn 1970
Ibn Khaldūn. Al-Muqaddimah li-Ibn Khaldūn: Al-juzʾ al-awwal min Kitāb al-ʻibar wa-dīwan al-mubtadāʾ fī ayyām al-ʻArab wa-al-ʻAjam wa-al-Barbar wa-man ʻāṣarahum min dhawī al-sulṭān al-akbar: Wa-huwa taʾrīkh. Beirut: Dār al-Qalam.

Klimchitskiĭ 1936
S. I. Klimchitskiĭ. “Vakhanskie Teksty.” (“Wakhi Texts.”) Trudy Tadzhikistanskoĭ Bazy, 3:75-124.

Kreutzmann 2015
Hermann Kreutzmann. Pamirian Crossroads: Kirghiz and Wakhi of High Asia. Wiesbaden: Harrassowitz.

Kreutzmann 2017
______. Wakhan Quadrangle: Exploration and Espionage during and after the Great Game. Wiesbaden: Harrassowitz.

Lashkarbekov 1972
B. Lashkarbekov. “Ba’ze Namunahoi Nazmi Vakhonī.” (“Some Examples of


Wakhi Verse.”) In Zabonhoi Pomirī va Fol’klor I. Ed. by Dodkhudo Karamshoev and Svetlana Vasilevna Khushevnova. Dushanbe: Akademiai Fanhoi RSS Tojikiston, Instituti Zabon va Adabiyoti ba nomi A. Rudaki. pp 143-47.

List 1963
George List. “The Boundaries of Speech and Song.” Ethnomusicology, 7.1:1-16.

Lord 1960
Albert Bates Lord. The Singer of Tales. Harvard Studies in Comparative Literature, 24. Cambridge, MA: Harvard University Press.

Madian 1992
Azza Abd al-Hamid Madian. “Language-Music Relationships in al-Farabi’s Grand Book of Music.” Unpubl. Ph.D. dissertation, Cornell University.

Majeed 2019a
Javed Majeed. Colonialism and Knowledge in Grierson’s Linguistic Survey of India. New York: Routledge.

Majeed 2019b
______. Nation and Region in Grierson’s Linguistic Survey of India. New York: Routledge.

Mandelbaum 1938
David G. Mandelbaum. “India, Kota and Toda, 1938.” Archives of Traditional Music, Indiana University, Bloomington.

Mandelbaum 1954
______. “Form, Variation, and Meaning of a Ceremony.” In Method and Perspective in Anthropology: Papers in Honor of Wilson D. Wallis. Ed. by Robert F. Spencer. Minneapolis: University of Minnesota Press. pp. 60-102.

Mandelbaum 1960
______. “A Reformer of His People.” In In the Company of Man: Twenty Portraits by Anthropologists. Ed. by Joseph B. Casagrande. New York: Harper. pp. 274-308.

Mandelbaum n.d.
______. “David Goodman Mandelbaum Papers, 1899-1991, 1933-1986.” BANC MSS 89/129 cz. Bancroft Library, University of California, Berkeley.

Marr 1985
John Ralston Marr. The Eight Anthologies: A Study in Early Tamil Literature. Madras: Institute of Asian Studies.

Martin 2001
Terry Martin. The Affirmative Action Empire: Nations and Nationalism in the Soviet Union, 1923-1939. Ithaca, NY: Cornell University Press.

Matrobov and Mirboboev 2015
S. Matrobov and A. Mirboboev, eds. Wux̌ diyor ganj: Ganji Vaxonzamin. Majmu‘ai osori mardumī. (Treasure of the Wakhan Territory. Collected Words of the People.) Dushanbe, Tajikistan: Gulu Giyoh.


Menéndez Pidal 1960
Ramón Menéndez Pidal. La Chanson de Roland et la tradition épique des Francs. 2nd ed. Trans. by Irénée-Marcel Cluzel. Paris: Picard.

Mills 1990
Margaret Ann Mills. Oral Narrative in Afghanistan: The Individual in Tradition. New York: Garland.

Mock 1998
John Howard Mock. “The Discursive Construction of Reality in the Wakhi Community of Northern Pakistan.” Unpubl. Ph.D. dissertation, University of California, Berkeley.

Moseley 2010
Christopher Moseley, ed. Atlas of the World’s Languages in Danger. 3rd ed. Paris: UNESCO.

Nagy 1996
Gregory Nagy. Poetry as Performance: Homer and Beyond. Cambridge: Cambridge University Press.

Nagy 2010
______. “Language and Meter.” In A Companion to the Ancient Greek Language. Ed. by Egbert J. Bakker. Malden, MA: Wiley-Blackwell. pp. 370-87.

Nara and Bhaskararao 2003
Tsuyoshi Nara and Peri Bhaskararao. Songs of the Toda: Text, Translation & Sound Files. Osaka: Endangered Languages of the Pacific Rim.

Niklas 1988
Ulrike Niklas. “Introduction to Tamil Prosody.” Bulletin de l’École française d’Extrême-Orient, 77.1:165-227.

Niklas 1993
Ulrike Niklas, ed. and trans. The Verses on the Precious Jewel Prosody Composed by Amitacākarar, with the Commentary by Kuṇacākarar (Amitacākarar iyaṟṟiya yāpparuṅkalakkārikai, Kuṇacākarar iyaṟṟiya uraiyuṭaṉ). Pondichéry, India: Institut français d’Indologie.

Pakhalina 1973
T. N. Pakhalina. “Образцы ваханской народной поэзии.” (“Samples of Wakhi Folk Poetry.”) In Проблемы восточного стихосложения. (Problems of Oriental Versification.) Ed by I. S. Braginskiĭ. Moscow: Nauka. pp 68-76.

Pakhalina 1975
______. Vakhanskiĭ Iazyk. (Wakhi Language.) Moscow: Nauka.

Patel 2008
Aniruddh D. Patel. Music, Language, and the Brain. Oxford: Oxford University Press.

Perry 2006
J. R. Perry. “Tajik Persian.” In Encyclopedia of Language & Linguistics. Vol. 12.2nd ed. Ed. by Keith Brown. London: Elsevier. pp. 484-87.


Pinker 1995
Steven Pinker. The Language Instinct: The New Science of Language and Mind. London: Penguin Books.

Powers 1976
Harold S. Powers. “The Structure of Musical Meaning: A View from Banaras (A Metamodal Model for Milton).” Perspectives of New Music, 14.2:308-34.

Probert 2006
Philomen Probert. Ancient Greek Accentuation: Synchronic Patterns, Frequency Effects, and Prehistory. Oxford: Oxford University Press.

Rodgers 2017
Stephen Rodgers. “Song and the Music of Poetry.” Music Analysis, 36.3:315-49.

Rosenthal 1967
Franz Rosenthal, trans. The Muqaddimah: An Introduction to History. 2nd ed. Princeton, NJ: Princeton University Press.

Sakata 2002
Hiromi Lorraine Sakata. Music in the Mind: The Concepts of Music and Musician in Afghanistan. Washington: Smithsonian Institution Press.

Seeger 1986
Anthony Seeger. “Oratory is Spoken, Myth is Told, and Song is Sung, but They Are All Music to My Ears.” In Sherzer and Urban 1986:59-82.

Sherzer and Urban 1986
Joel Sherzer and Greg Urban, eds. Native South American Discourse. Berlin: Mouton de Gruyter.

Shulman 2016
David Shulman. Tamil: A Biography. Cambridge, MA: The Belknap Press of Harvard University Press.

Slobin 1976
Mark Slobin. Music in the Culture of Northern Afghanistan. Viking Fund Publications in Anthropology, 54. Tucson: University of Arizona Press.

Steblin-Kamenskiĭ 1999
I. M. Steblin-Kamenskiĭ. Этимологический словарь ваханского языка. (Etymological Dictionary of the Wakhi Language.) Saint Petersburg: Saint Petersburg State University.

Urban 2001
Greg Urban. Metaculture: How Culture Moves Through the World. Public Worlds, 8. Minneapolis: University of Minnesota Press.

Urmawi 1960-61
Ṣafī al-Dīn al-Urmawī. “Kitāb al adwār fi al mūsīqī.” Trans. (from Arabic into Persian) by Mīrzā Muḥammad Ismāʿīl b. Muḥammad D̲j̲aʿfar Iṣfahānī. Majallah-i mūsīqī, 3:42-49 (1339 AH) and 3:50-58 (1339-40 AH).

Wolf 1997
Richard K. Wolf. “Of God and Death: Music in Ritual and Everyday Life. A Musical Ethnography of the Kotas of South India.” Unpubl. Ph.D. dissertation, University of Illinois at Urbana-Champaign.


Wolf 2000/2001
______. “Mourning Songs and Human Pasts among the Kotas of South India.” Asian Music, 32.1:141-83.

Wolf 2005
______. The Black Cow’s Footprint: Time, Space, and Music in the Lives of the Kotas of South India. Delhi: Permanent Black.

Wolf 2009
______, ed. Theorizing the Local: Music, Practice, and Experience in South Asia and Beyond. Oxford: Oxford University Press.

Yarshater 1974
Ehsan Yarshater. “Affinities between Persian Poetry and Music.” In Studies in Art and Literature of the Near East: In Honor of Richard Ettinghausen. Ed. by Peter J. Chelkowski. Salt Lake City: Middle East Center, University of Utah. pp. 59-78.

Zvelebil 1992
Kamil Zvelebil. Companion Studies to the History of Tamil Literature. Leiden: Brill.