Beyond Tālaprastāra in Indian Music: Prosody as a Generating Function of Rhythmic Complexity in Aruṇakirinātar's Tiruppukaḻ

"A generating function is a clothesline on which we hang up a sequence of numbers for display."

-Herbert Wilf²

The use of permutation as a tool to generate melodic patterns is discussed in early Indian texts on musicology such as the Dattilam and Bṛhaddeśī (Widdess 1995). But the earliest reference to an analogous scheme for rhythmic cycles (tāla)-called Tālaprastāra-is found only in later texts, the Samgīta Samayasāra of Pārśvadeva and the Samgīta Ratnākara of Śarṅgadeva (c. 13th century CE). The chapter on tāla in the Samgīta Ratnākara contains more than 70 verses on this topic, and subsequent treatments of the subject follow Śarṅgadeva. A discussion of Tālaprastāra is found in many Sanskrit and Tamil texts on music written between the 15th and 17th centuries and a few later texts as well. The theme figures prominently in the Telugu musicological tradition, where one finds Tālaprastāra discussed in texts written as late as the 20th century. A table showing a particular scheme of Tālaprastāra is found in a manuscript attributed to Śyāma Śāśtri, one of the composers who constitute the so-called musical trinity of Carnatic (South Indian) music.³ The perceived importance of Tālaprastāra stems from the notion that it provides the musician a scheme to generate different tālas.

Despite the extensive treatment of Tālaprastāra in medieval literature, there are no examples of musical forms (compositions) that allow us to study either the necessity or the practical utility of this scheme. Nor is its use to be found in contemporary music. Indeed, it is fair to say that Tālaprastāra as described in these texts has remained largely, if not entirely, within the theorist's realm. Are we then to conclude that the generation of rhythmic variety and complexity, and their practical realization in the sense of inventing new tālas, lie solely within the domain of percussion music? The purpose of this paper is to argue that the scheme was not restricted to the domain of percussion as there is at least one example in Indian music, viz., the works of the 15th century Tamil poet Aruṇakirinātar,⁴ where rhythmic variety is realized in an entire corpus of compositions, albeit without seeking recourse to the cumbersome and often clumsy machinery of Tālaprastāra described in the texts. Given the elegance of Aruṇakirinātar's methods and his innovations in the Tiruppukaḻ, some of which will be discussed later in this paper, it is not surprising that this particular corpus of compositions has survived in the performance traditions of Indian music and dance. A study of Aruṇakirinātar's compositions provides us with valuable insights into some of the methods used by poets who excelled in composing rhythmic verse, and the connection between these methods and the grammar of rhythm. Aruṇakirinātar's works thus present an opportunity to study the relationship between prosody and tāla, a subject that has not been discussed in any detail within the context of Indian music.

The paper is organized as follows. The remainder of this section is devoted to a brief summary of the manifestation of variety in tāla, the origins of the method called Tālaprastāra, and the historical and musical context in which I study Aruṇakirinātar's contributions. In the following section, I discuss the rules of prosody used by Aruṇakirinātar. The link between prosody and rhythm is explored next through a series of examples that illustrate how Aruṇakirinātar generates rhythmic complexity through prosody, and the final section is devoted to a discussion of the interplay between melody and rhythm in Tiruppukaḻ.

Two defining attributes of a tāla are the number of mātrās, which characterizes the duration of a tāla cycle, and the manner in which these mātrās are partitioned into substructures of a tāla that may be termed as building blocks. The absolute unit of the mātrā varies between texts and traditions,⁵ and this need not concern us. For the purposes of this discussion, it suffices (1) to specify the ratio between the mātrā counts of these building blocks and (2) to note that this ratio is fixed. Hence, in the remainder of this paper I choose to work with "dimensionless units," in terms of which the three building blocks-laghu, guru, and pluta-correspond, respectively, to 1, 2, and 3 units.⁶ A tāla is then immediately specified by a particular arrangement of these building blocks. For example, a tāla whose arrangement is guru, guru, laghu, pluta corresponds to 8 units. I emphasize, however, that the total number of units is subservient to the particular arrangement of the building blocks in the characterization of the tāla. This is only to be expected since the practical realization of keeping time to a tāla either through the action of the hand(s) or through a percussive instrument usually reflects not merely the total number of units in a tāla, but also the arrangement in which they are partitioned therein.

Keeping the total number of units in a tāla fixed, variety and complexity may be envisaged in the following ways: (1) partitioning the total number differently; for example, keeping the total number of units fixed at 8, a new tāla is obtained by partitioning these 8 units into a guru, laghu, guru, laghu, guru arrangement. This arrangement is different and distinct from the previous example, viz., guru, guru, laghu, pluta. (2) Permuting a given arrangement, thus yielding a new tāla with the same units as the original; for example, permuting the arrangement guru, guru, laghu, pluta to guru, laghu, guru, pluta yields a different arrangement of 8 units and hence considered a distinct tāla. Thus, any rearrangement of building blocks automatically constitutes a new tāla. (3) Further variety ensues upon expanding the set of building blocks, guru, laghu, pluta, to include new elements with different units.⁷ Note that I omit the possibility of dilating the internal units (building blocks) either equally or unequally. It will be seen later that the effects of such operations are accommodated easily within Aruṇakirinātar's scheme.

The method of Tālaprastāra provides a systematic way to enumerate all possible operations (partitions and permutations) that leave the total number of mātrās invariant.⁸ While this clever combinatorial scheme may be useful to understand tāla variety a posteriori, its appearance in the Samgīta Ratnākara likely led to the unfortunate (and perhaps, predictable) effect of classification gaining precedence over content. In this context, it is interesting to note Śarṅgadeva's definition of a tāla.⁹ He says, " . . . tāla is measured time which determines the duration of gīta etc. by means of the individual measurement of laghu etc. . . ." (5.3).

Clearly, Śārṅgadeva understands tāla as a durational measure of song that did not have an independent existence. Despite this definition of the tāla, the method of prastāra, subsequent to its appearance in the Samgīta Ratnākara, seems to have degenerated into an exercise in enumerative combinatorics. Since the number of tālas that could be generated through permutation increases as a function of the total number of mātrās constituting the tāla, there are no obvious reasons either for generating specific tālas without compositional forms set to these tālas, or to terminate the process after the generation of a certain number of them. Typically, tāla compilations choose an arbitrary number (e.g., 108), that have little or nothing to do with the grammar of tāla, and we have no examples of musical forms that employed these; viz., these tālas have no "meaningful existence" (Ramanathan 1997) without a compositional form to measure.¹⁰ The situation is summarized aptly by Sathyanarayana, " . . . the evolution of deśī tāla lost its equilibrium, reached a point of extreme proliferation in which system, science, convenience and practicability were chased away by fancy, caprice and invention for invention's sake. This led to anarchy, license and exuberant redundance in the fourteenth-fifteenth centuries AD" (Sathyanarayana 2004).

However, a survey of the literature shows that many tālas enumerated in the Samgīta Ratnākara did have a "meaningful existence," at least prior to their being classified; viz., these tālas were specific to certain performance traditions. An example is from the Tamil tradition, which is of immediate relevance to us. In his commentary on the Tamil text Cilappatikāram (Iyer [1892] 2001), Atiyārkunallār (c. 13th century CE) writes,

. . . those forty one tālas, ranging from the eka tālam comprising four mātrās to the pārvatilōcaṉam with sixteen mātrās, belong to the puṟakkūttu. Āṟaṉmaṭṭam, eṭṭaṉanmaṭṭam, ṭālavōriyal, taṉṉilaiyōriyal, as well as the eleven varieties of pāṇi seen in oṉṟaṉpāṇi, eṇkūttuppāṇi, etc., mutalaṭivāram, etc., all belong to the akakkūttu . . . ¹¹

Note the unambiguous connection made by the commentator between specific dance forms (kūttu) and certain tālas (of which both eka and pārvatilōcaṉam appear in Śarṅgadeva's list of deśī tālas). A similar connection between tālas and specific compositional forms can also be seen in the works prior to the Samgīta Ratnākara (such as the Mānasollāsa) that list a variety of tālas in vogue. In some cases, the names of the tālas are used interchangeably with compositional forms (prabandha),¹² which suggests yet again that at least some of these tālas were tied to specific compositional or dance forms. It is then reasonable to surmise that Śarṅgadeva's efforts toward systematization of the so-called deśī tāla were motivated by its popularity. It is also likely that such systematization ultimately led to the deśī tāla system inflating itself to death, all examples of rhythmic variety in composition lost irrevocably. More importantly, we do not know anything regarding the devices used by composers to impart rhythmic variety in song and dance. What remains in the literature is the prastāra scheme, which I argued above, is devoid of content in the absence of specific examples in composition.

It is in this context that I seek to study Aruṇakirinātar's Tiruppukaḻ and describe how he used prosody to generate meaningful variety in the expression of tāla. I shall show that the methods used by Aruṇakirinātar realize the goal of Tālaprastāra in a very natural way, and that the rhythmic structure in his songs exhibits some important characteristics of the deśī tāla described by Šārṅgadeva and others.

The Tālaprastāra scheme is presented in the Samgīta Ratnākara, within the text's discussion of the deśī tāla. The phrase "deśī tāla" occurs for the first time in this text (though some of the tālas listed in the Samgīta Ratnākara as "deśī" are mentioned in earlier texts) and is contrasted with the "mārga tāla" mentioned in Bharata's Nāṭya Śāśtra. These terms have been the subject of investigation for a long while now.¹³ Rowell's comments (Rowell 1992) on the deśī tāla are particularly relevant in this context. He sees the emergence of deśī tāla as part of a broader "phenomenon of deśī-the dynamic expansion of musical resources during the second half of the first millennium" during which time "a large number of popular regional traditions were collected, codified, partly homogenized, and set alongside the venerable mārga tradition (which in the end they supplanted). . . ." Rowell speculates further that the deśī tālas were closely associated with songs and poetic traditions and that the new tālas were characterized by short, repeatable sequences of fixed durations. Chaudhary (1997) notes that the mārga and deśī tālas differ in the basic units they use and in the value of the laghu (fixed in mārga, and variable in deśī). In emphasizing the structural differences between these two kinds of tālas, neither author seems to have paid much attention to another aspect of the deśī tāla-its manifestation as variety in sound. This feature of the deśī tāla is evident in Śarṅgadeva's characterization and Kallinatha's commentary on this section of the Samgīta Ratnākara (5.237-238). I discuss these below and argue that the features identified by Śarṅgadeva and Kallinatha, as unique to deśī tāla, are realized in the Tiruppukaḻ.

Śarṅgadeva identifies two important characteristics of the deśī tāla: (1) the flexibility of its structure, which arises from the variable value of the laghu and (2) the importance given to its expression. According to him, " . . . the deśī tāla comprises the action of laghu and other durations and finds aesthetic expression in the sound of cymbals . . . " (5.237). Śarṅgadeva's phrase " . . . yathāśobhaṃ kāmsyatāladhvananādikaya . . . " in the above verse has been elaborated by Kallinatha in his commentary. Kallinatha chooses to interpret yathāśobhaṃ-as required by (considerations of) beauty-as the fundamental requirement behind both the rhythmic organization of the deśī tāla and its expression. He writes, for instance:

. . . bound by that beauty which arises from the sound of cymbals, sometimes restrained, and at other times not, some times in the strokes of specific hand gestures and some times in the pattern of strokes, the variety and complexity of sound engenders the wondrous deśī tāla . . .

(5.237)

Kallinatha also states clearly that the variety in deśī tāla is a direct consequence of variety in arrangement and the number of units, " . . . it should be understood that the variety in tāla arises both from the variety in the arrangement (of laghu, etc.) and in its measure . . . " (5.238).

Kallinatha's commentary reveals two features of the deśī tāla that are particularly significant for a discussion of Aruṇakirinātar's works. The first is that the structure of the deśī tāla was flexible enough to sustain and illustrate rhythmic variety. However, Śarṅgadeva's choice of the phrase "yathāśobhaṃ" underscores the fact that the structure (of the deśī tāla) was less important than considerations of rhythmic aesthetics. Next, his allusion to the cymbals (kamsyatāla) and Kallinatha's exposition thereof lead us to conclude that in the deśī tālas, rhythmic variety was more apparent in its manifestation as sound than as complexity in the underlying structure of the tāla.¹⁴ Indeed, the latter condition is neither necessary nor sufficient to satisfy the former,¹⁵ and Śarṅgadeva's criterion for the beauty of deśī tāla may be considered as "rhythmic color," in complete analogy with Mātaṇga's description of a rāga as possessing "tonal color."¹⁶

In the Tiruppukaḻ, one finds the manifestation of rhythmic variety as sound at two levels. The first is found in the connection between textual flow (meter) and the rhythmic arrangement. This aspect will be discussed in detail in the following section. The second is seen in the profusion of onomatopoeic terms in the Tiruppukaḻ verses, and I will discuss this point briefly. The specific theme of expressing tāla variety through sound (especially sound from cymbals) recurs throughout the works of Aruṇakirinātar. As a devotee of the Tamil god Murukaṉ (Zvelebil 1991), Aruṇakirinātar hears tāla in the sound produced by Murukaṉ's anklets. To the poet, these anklets express the "infinite variety of sound" (7.1219).¹⁷ Similarly Murukaṉ's feet are described as being adorned by "gold anklets that produce music of infinite variety" (1.15, 4.204, 7.703). The analogy with Śarṅgadeva's "kāmsyatāladhvananam" is obvious. The importance Aruṇakirinātar attaches to the expression of tāla as sound is further reflected in his use of onomatopoeic terms such as "kiṇkiṇI," (1.16, describing the sound of the cymbals) or phrases that serve the dual purpose of expressing sound (usually of some musical instrument) and specifying the rhythmic structure of the composition. Some examples are "ṭuṭuṭuṇṭuṭu ṭuṇṭuṭu ṭuṇṭeṉa" (1.2, describing the sound of drums) and "tokukukuku tokukukuku toṅkat toṅkat tokutīto" (2.17, describing the sound of the drum accompanying Siva's dance). Aside from these explicit allusions to tāla as infinite variety in sound, the perception of tāla variety in the Tiruppukaḻ is seen most clearly in its prosody, which I discuss in the following section.

Prosody of Tiruppukaḻ

The prosody of Tiruppukaḻ is best understood by breaking down its structure into three components: (1) pā (meter), (2) cantam (tāla), and (3) vaṇṇam (sound class). The poem is vested with these qualities upon the imposition of appropriate constraints. The primary constraint is on the lyrical structure (the meter of a poem) to which cantam and vaṇṇam are imposed as secondary and tertiary constraints, specifying tāla and class of sound (hard, soft, medium). Next, I argue that the imposition of these constraints along with the introduction of cyclicality turns the poem into a composition set to a specific tāla.

Meter

The Tiruppukaḻ songs are composed in a meter called āciriya viruttam. It has four lines and each line is made up of six or more feet. The foot can be di-, tri-or polysyllabic.¹⁸ The basic constraints of the viruttam meter are the following: (1) all lines should have the same number of feet; (2) the arrangement of the syllables inside a foot in the first line should be identical to the syllabic arrangement of the corresponding foot in the second line (e.g., the second foot of the first line should have the same syllabic arrangement as the second foot in the second line); and (3) similarly corresponding feet in the third and fourth lines must have the same syllabic structure. The poem is considered superior if all four lines have the same syllabic arrangement, and the Tiruppukaḻ songs fall under this latter category. The syllabic arrangement inside a foot is most conveniently characterized by the unit of metrical duration. The second column of Table 1 shows the metrical duration of sounds in Tamil grammar.

Table 1.

The second and third columns show speech and rhythmic durations of sound in Tamil prosody that determine the characteristics of meter and tāla, respectively. The final column shows the classes of sounds: hard (H), soft(S), and medium (M) used in determining the vaṇṇam. See text for a discussion.

The constraint on the arrangement of syllables in a line is the first step toward what I call rhythmization of a poem.¹⁹ However, this step does not suffice to transform the poem into a song set to a specific tāla. This is because syllabic arrangement in a poem is determined by the metrical duration (of a speech sound) that, in turn, determines the meter. The key point here is that the metrical duration of a sound is not necessarily equal to the duration of the same sound in music (say, as measured by the beats of a tāla), which fact is noted in the earliest extant Tamil grammar, the Tolkāppiyam (verse 33, Eḻuttatikāram; Pavanar 1966, 36). Thus, for example, the two syllables "tat" and "ta" have the same metrical duration (of speech sound-second column in Table 1) and will be considered equivalent insofar as the determination of meter is concerned. But these two syllables do not have the same rhythmic duration; the addition of the consonant "-t" to "ta" acts as a stop in the rhythm. It is important to understand this difference when analyzing rhythmic verse.

Throughout this paper, I will use the following notation to indicate rhythmic duration:

ta ta ta ta      tatta    tat tat      tā na

In this method of indicating durations, the first line represents 4 "ta" syllables, and we will take its rhythmic count to be 4 (in units of "ta," which will be the fundamental rhythmic unit 1). The second line, "tatta," can be thought of as "ta+t+ta." Here, the consonant "-t" has zero metrical value; that is, its occurrence is irrelevant in considerations of meter. However, the same consonant "-t" has a rhythmic duration of 1 unit and, consequently, the rhythmic duration of "tatta" will be taken to be 3. Similarly, "tattat" in line 3 represents a rhythmic count of 4 and "tāna," 3. These assignments represent the fundamental counting rule for all rhythmic verse in Tamil. The failure to recognize the distinction between metric and rhythmic counts has led to some confusion in the literature (Selvamony 1991).

The distinction between the metric and rhythmic durations of a sound does not violate the rules of basic prosody that determine the meter. In composing rhythmic verse, the poet imposes an extra set of constraints based on the rhythmic duration of the same syllables that determine its metrical structure. Hence, the poem is rhythmized without losing its metrical structure; metrical structure or meter is governed by the metrical duration of the sound, while rhythmic structure or tāla depends on the rhythmic duration of the same sounds. To state this more concisely, all rhythmized verses belong to some metrical class, but a verse composed in a specific meter is not necessarily rhythmic.

Tāla

The second step in rhythmization is to invoke a set of independent constraints on the rhythmic duration of the syllables in the poem. Examples of this practice are seen in older literature, for example, Cilappatikāram (e.g., verse29.22, Iyer [1892] 2001) as well as in the devotional literature of other Tamil poets who preceded Aruṇakirinātar. However, to my knowledge, a formal distinction between rhythmic and metrical durations first appears in Tamil grammar only between the 10th and 11th centuries. Both the Yāpparuṅkala Virutti (c. 10th century) and the Vīracōḻiyam (c. 11th century) make this distinction clear. For instance, the Vīracōḻiyam defines the rhythmic duration of sound using the terminology of Sanskrit grammar thus: "The long vowel, the long vowel followed by a consonant, the short vowel followed by a consonant are called guru; a solitary short vowel is called the laghu, but at the end of a line, it may assume the function of the guru" (verse 130; Gopalaiyar 2005, 476).

The same opinion is voiced by the author of the Yāpparuṅkala Virutti (see Ilankumaran 1973, 477).²⁰ By stating the rhythmic duration of speech syllables in units of the laghu and guru, these texts provide an unambiguous prescription to relate the duration of speech and rhythmic durations. Thus, for example, we see that "ta" has metrical and rhythmic duration of 1. The metrical duration of "tat" is again 1 (the consonant has no metrical value) whereas its rhythmic duration is 2. Consequently, the two syllables are only metrically equivalent.²¹ Further note that though the short vowel has a rhythmic duration of 1 (laghu) it "may assume the function of a guru," that is, assume a duration of 2, at the end of a line. A similar rule applies at the end of a half-line if the two halves have identical rhythmic structures (Mutaliar 1984). The origins of this rule lie in the flexibility required in the melodic expression and/or recitation of a song, a point I will discuss later in the context of the Tiruppukaḻ.

Having fixed the rhythmic duration of sounds in terms of the laghu and guru, the constraint leading to rhythmization is obvious. We demand that every line of the verse has the same rhythmic arrangement in addition to having the same syllabic arrangement; that is, any foot in a given line must have the same arrangement of guru and laghu as the corresponding foot in the other lines. Consequently, all lines have the same rhythmic duration and more importantly, the mātrās are partitioned identically in all the lines and hence the rhythmic structure can be denoted by a tāla.

Sound Class

Imposing a constraint on the rhythmic duration of sound completes the process of rhythmization. However, another independent constraint when imposed on sound class leads to further variety in the vocal expression of sound. This constraint can be understood in the following way. As discussed earlier, the constraint on rhythmic duration only demands that corresponding feet (across different lines) must have the same arrangement of guru and laghu. Now note that a given arrangement (of guru and laghu) can correspond to any class of sounds-hard, soft, or medium. The constraint on sound class demands that the arrangement of syllabic sounds in terms of their sound class must also be identical in every line. Consequently, an entire verse is constructed with a specific rhythmic arrangement and may have a preponderance of one sound class. This feature has been called vaṇṇam by modern grammarians in the context of rhythmic verse (Tantapani 1987) and is related to a more general feature discussed in the Tolkāppiyam. The word vaṇṇam means color and here it may be understood to be the "color of sound."²² Recall here, my earlier remarks on variety in tāla and rhythmic color in the deśī tāla. The Tamil grammarians invoke this connection by bridging meter, tāla, and sound class.

A verse that satisfies all three (primary, secondary and tertiary) constraints above is called a vaṇṇac canta viruttam. The Tiruppukaḻ corpus arguably represents the best vaṇṇac canta viruttam found in Tamil literature.

Rhythmic Variety, Sound Classes, and Cyclicality in Tiruppukaḻ

From our previous discussion, it follows that variety in the composition of verse (subject to the constraints discussed above) automatically satisfies the goal of Tālaprastāra-namely, creation of variety and complexity in tāla. Indeed, the prastāra of creating verse (rather than tāla) is discussed in the Yāpparuṅkala Virutti, and the discussion there is very similar to that of Tālaprastāra in the Samgīta Ratnākara. In the following, I discuss how the Tiruppukaḻ exhibits rhythmic variety in its construction.

Most of the Tiruppukaḻ songs have the following structure.

• • A verse has four lines that are broken up into eight semi-lines. These semi-lines have identical speech sound arrangement, rhythmic arrangement, and sound class.²³

• • Each semi-line has several feet (di-, tri-, or polysyllabic) and the final foot of every semi-line acquires a special status. Grammarians call the final foot toṇkal (lit. hangs) and I will refer to it as the "final foot" with the understanding that it refers to the final foot of every semi-line.

Variety in Rhythm and Sound Class: The Role of the Foot

The variety in rhythm and sound class of the Tiruppukaḻ is conveniently defi ned and described by the building blocks of a rhythmic verse as shown in Table 2 (Tantapani 1987). The respective mātrā counts (the rhythmic duration) are also specified in the table. Note that the half consonants acquire the status of a laghu-their rhythmic duration is 1 mātrā, though they do not matter in determining the metrical structure.

Table 2.

Duration, sound class, and the notation of the eight building blocks of rhythmic verse.

The following points may be noted:

• • To the set of eight building blocks above, one can add another set of eight by elongating the final vowel (e.g., taṉā [3 mātrās], tāṉā [4 mātrās], etc.).

• • The rhythmic structure and duration of each foot (and consequently, of the entire verse) will be represented by a combination of the above eight. Examples are taṉātaṉā (a foot with 5 syllables and 5 mātrās), tantat (2 syllables, 4 mātrās), tayyatanta (4 syllables, 6 mātrās), and so on. The reduction of the basic rhythmic syllables to building blocks of 2 or 3 in Aruṇakirinātar's works is important in view of what I described earlier. As the examples below will show, the rhythmic structure underlying the tālas generated by Aruṇakirinātar is simple, and the richness of variety is exhibited truly as variety in sound patterns.

• • Since the rhythmic structure of the entire verse is represented by that of a semi-line, it suffices to represent the semi-line with the rhythmic syllables taṉā tāṉa, and so on, shown above. The grammarians call this notation cantak kuḻippu. It has been used consistently by various authors and scholars who compiled Aruṇakirinātar's work and also found in manuscript versions of some songs.²⁴

• • Thus, for example, the semi-line (1.1),

  muttait tiru pattit tirunakaiattikiṟai satticcaravaamuttikkoru vittuk kurupara-eṉavōtum

  will be represented in cantam notation as

  tattat tana tattat tanatanatattat tana tattat tanatanatattat tana tattat tanatana-tanatānā

• • The threefold occurrence of a rhythmic structure before the final foot, as in the above example, is most common in the Tiruppukaḻ. In such cases, it is enough to notate the fundamental nonrepetitive phrase. For the above example, cantam notation would then read,

  tattat tana tattat tanatana

  The final foot (see below) will always be shown separately (following the "-"). Since it has a fixed rhythmic duration independent of the rhythmic phrase(s) that precede it, I will ignore it in some examples below.

• • The cantam notation is admittedly cumbersome in Roman script, but has the advantage that it reflects the variety in sound class. Thus, for example, while the syllables tatta, tanta, and tayya are rhythmically and metrically equivalent, they lead to three different representations of 3 mātrās as sound. The choice of tayya (as opposed to the other two) in the rhythmic representation of a given verse automatically means that the foot (or part of it) occurring at that portion of the semi-line (and all subsequent semi-lines) is made up of two consonants or short vowels separated by a consonant of medial sound class (the set ya ra la va ḻa ḷa); for example, only words such as alli, payya, and so on, will figure in this position of the song. Analogous rules hold for the hard and soft equivalents.

I now turn to a few examples that demonstrate the generation of rhythmic variety in Tiruppukaḻ and demonstrate how the goals of Tālaprastāra are realized in these songs. I am not aware of a complete enumeration of all the distinct rhythmic patterns found in the thousand songs available; nor do I attempt this task.²⁵ Instead, I describe some examples that illustrate the central themes of this paper.

Tālaprastāra in Practice

I begin with some examples where the rhythmic units are multiples of four or eight. The first example is verse 3.118, which exhibits a simple pattern of fours. Below, I show an entire semi-line (with the final foot). Note the threefold repetition of a fundamental rhythmic phrase and the presence of the two long vowels at the end of the final foot.

tānā tana tana tānā tana tana tānā tana tana-tanatānāātā ļikaḷ puri kōlā kala viḻi yālē yamu teṉu-moḻiyālē

In this example, an 8 mātrā pattern of (two) fours occurs as two long rhythmic syllables followed by four short ones. A slightly modified pattern is seen in 3.121, where the 8 mātrās are partitioned into (35). I show this semi-line below (without its final foot, which has the same rhythmic structure above).

tāna tanatanana tāna tanatanana tāna tanatananacīṟa lacaţaṉviṉai kāraṉ muṟaimaiyili tīmai purikapaṭi

A different partition of 16 is to be found in 7.753, where feet of 3, 4, 5, and 4 mātrās are stacked together in the fundamental rhythmic phrase shown below. This phrase occurs thrice in a semi-line before the final foot.

tanana tānana tānanāt tanantaaṟivi lātava riṉarpēc ciraṇṭu

Note that the first and the fourth feet are metrically equivalent in the above example. The "ṇ" in the fourth foot has no metrical value. However, its presence leads to a rhythmic duration of 4 for the fourth foot (as compared to 3 for the first foot). In terms of sound class, the presence of "ṇ" (rather than, say, "t") means only a consonant of soft sound class shall occur at this position of the verse (i.e., the middle consonant in the fourth, eighth, and twelfth feet of every semi-line); for example, kuṇaṇka, kalankaṇ, and so on.

An example where the basic rhythmic pattern is simple yet exhibiting variety in sound expression is seen in 3.115. Here, the semi-line is broken into two smaller pieces, of which the first is expressed by

tantam tantam tantam tantamkuṉṟuṅ kuṉṟuṅ ceṇṭuṅ kaṉŗum

Note the alternating soft consonants, yielding variety in sound class. This construction implies that the first half of every semi-line in the verse should have a similar arrangement of alternating soft consonants. The second piece of the semi-line is also made up of feet with 4 mātrās, but the construction of these 4 mātrās is slightly different:

tanatana tanatana tanatana tanatanapaṭivaṇar mulaiyiṇil mrukamata meḻukiyar

Thus, we find rhythmic variety even while the partition of units is homogeneous in the tāla cycle.

The relationship to Tālaprastāra can now be seen clearly. Consider, for instance, verse 5.394. The fundamental rhythmic phrase that occurs thrice in a semi-line is 16 mātrās and partitioned into (3+5+4+4). Its structure is given by

tāna tanattat tanatta tantanakōti muṇittuk kaṇatta koṇṭaiyar

As discussed earlier, the goal of Tālaprastāra is to generate new structures by permuting the partition of units. In the Tiruppukaḻ, this is realized by the permutation of syllabic arrangement. So, in addition to the example above, we also find songs where the basic pattern is (5+3+4+4; verse 5.231), (5+4+3+4; 7.991), (4+3+5+4; 2.53), (3+5+4+4; 5.394), and so on. Further, Aruṇakirinātar goes beyond such permutation by creating variety within each of these patterns. For instance, consider the following three examples (verses 7.1197, 5.414, and 5.394) where the same pattern (3+5+4+4) occurs.

tatta tanattat tanatta tattanavaṭṭa mulaikkac caviṇttu vaittuḷa

tanana tanatanā tanatta tānanakutalai moḻiyiṉār nitikkoḷ varaṇi

tāna tanattat tanatta tantanakōti muṭittuk kaṉatta koṇṭaiyar

In each of the phrases above, the blocks themselves (of 3, 4, and 5 mātrās) are constructed differently, thus yielding variety within the same partition scheme.

As is evident from these examples, the possibilities arising from the twin devices of permutation and repartitioning are infinite and bound only by rules of prosody and the composer's poetic skills. Verse 6.441, for example, is constructed out of rhythmic phrases of 17 mātrās partitioned as (6+3+3+3+2) and shown below.

tānatatta tāna tanā tanā tanacīrcirakku mēṉi pacēl pacē leṉa

Different realizations of 17 mātrās (different partitions) are found in other verses-for example, verse 7.1154 where the 17 mātrā phrase is partitioned into (4+4+5+4), (4+5+4+4; 7.1140), (5+4+4+4; 7.631), and so on.

In all the examples given above, the fundamental rhythmic phrase occurs thrice in a semi-line. In some verses though, the entire semi-line is captured within a single rhythmic phrase. An example is verse 5.336, whose structure is shown below (along with the final foot). In this verse, the semi-line is divided into two components, (3+4+3+4) and (4+4+4+4). Note that in the final foot, the syllable "a" is given a rhythmic duration of 2, in accordance with the rule we discussed earlier concerning the laghu in the final position.

tanana tāttana tanana tāttanakumuta vāykkaṉi yamuta vākkiṉar

tānā tānā tānā tānā-tanatānākolē vēlē cēlē polē-aḻakāṉa

Some verses from the Tiruppukaḷ have the same rhythmic structure as tālas currently in vogue. An example is the Tiruppukaḷ (4.225), "pāti matinati . . . " This verse alternates between feet with 3 and 4 mātrās and is structurally isomorphic to the popular tāla, Miśra Cāpu. The entire semi-line (without the final foot) is given below.

tāna tanatana tāna tanatana tāna tanatanapāti matinati pōtu maṇicaṭai nāta raruļiya

Stresses

An example that constitutes a variant of this class (of 7 mātrās) is verse 7.1177. Here, the fundamental rhythmic phrase is of 14 units and partitioned into (5+4+5). The rhythmic structure of the verse is given by

tanan'atta tan'atta tan'attanapuruva'ttai neṟittu viḻikkayal

In terms of the total number of mātrās, the above phrase can, in modern parlance, be thought of as a mixture of 5 (called khaṇḍa) and 9 (caṇkirṇa) and sung accordingly. However, this is not a faithful representation of the verse because typical rhythmic patterns used to represent 5 and 9 mātrās do not reflect the way these units are partitioned. Further, a natural place for stress (marked by "^/") in this verse is the third beat as shown above, where the vowels a, i, and i marked in the words puruva^/ttai, neṟi^/ttu, and viḻi^/kkayal are stressed while singing. This feature is lost entirely when the verse is shoehorned into a 5+9 structure.

Another example where the verse offers positions for rhythmic stresses is found in verse 5.263. Here again, the rhythmic structure of the semi-line is simple-all feet are of 5 mātrās-however, the stresses make it sound very different from the typical patterns of 5. The structure of the semi-line (with the final foot) is shown below.

tana'ttana tanattamtana'ttana tanattamtana'ttana tanattam-tanatānācina'ttavar muṭikkumpakai'ttavar kuṭikkuṅceku'ttava ruyirkkuṅ-cinamākac

It is evident from our discussion that the Tiruppukaḻ verse and its fundamental rhythmic structure are strongly intertwined. Perhaps to underscore this feature, Aruṇakirinātar incorporates syllables mimicking the rhythmic notation explicitly in some songs. An example is verse 2.69, which describes the rhythm to which Murukaṉ's peacock dances. One of the semi-lines reads as follows (I include the final foot for completeness):

taṉaṉataṉa tānta ṉantatteṉanaṭaṉa mārnta tuṅkattaṉimayilai yūrnta cantat-tirumārpā

The first part of the semi-line above describes the dance of Murukaṉ's peacock using the syllables taṉā, and so on. But a comparison with the other two parts shows clearly that this phrase is also the underlying rhythmic structure of the semi-line. Similar examples are found in verses 5.354, 5.404, and so on. Indeed, the connection between the metric and rhythmic structures is so strong and explicit in the Tiruppukaḻ that attempts to bring this corpus of songs into "standardized" tāla frameworks (such as the 108 tāla or the 35 tāla schemes) are neither fructuous nor appropriate. The rhythmic structures of the Tiruppukaḻ and these tāla schemes are "additive," a term employed by Powers to underscore the fact that the tāla cycle is made up of components of unequal length²⁶ (as opposed to "divisive" rhythms of melody wherein the rhythmic cycle is usually subdivided into subsections of equal length²⁷). However, a Tiruppukaḻ verse presents both the total rhythmic count and the manner in which it is partitioned in a tāla cycle, and this may not necessarily coincide with the partitioning of the components in any of the tāla schemes.

Whence Variety?

Thus far, I discussed a few examples from the Tiruppukaḻ to illustrate how the construction of rhythmic verse realizes the goals of Tālaprastāra in practice. While it is clear that infinitely many distinct rhythmic patterns could possibly be generated along these lines, it is instructive to ask the question from the point of view of a composer; viz., among the infinitely many patterns available, how is a particular rhythmic pattern chosen? Interestingly enough, the answer is also related to the theme we discussed so far, the strong connection between verse and tāla in the Tiruppukaḻ. In most of the songs, Aruṇakirinātar indicates the name of the place where the song was composed. This name usually occurs as one word in the verse (or at times, split across two words), and places a constraint on the appearance of at least one rhythmic unit which corresponds to the name;²⁸ for example, requiring the word Āviṉaṉkuṭi (the older name for Paḻani, a town situated 70 miles from the city of Maturai in Tamil Nadu) to appear in the verse forces one foot of the semi-line to have the corresponding rhythmic structure, tānatantana. Seen this way, it is not surprising that the verse 3.103 has the following rhythmic structure (below, I show the last semi-line of the verse to illustrate the point made here).

tāna tantana tāna tantana -tanatānāāvi ṉaṉkuṭi vāḻvu koṇtaruṇ-perumālē

At first sight, imposing this constraint might seem severely restrictive; but Aruṇakirinātar uses devices such as adding an honorific to the name, declining the noun (the name of the temple/town), and so on. This yields a distinct rhythmic unit that is then woven into a different rhythmic phrase. Consequently, there is still a wide variety of rhythmic structures among the songs sung at the same place. As an example, consider verse 3.109, which also addresses Murukaṉ in Āviṉaṉkuṭi. In this verse, Aruṇakirinātar uses words to mean "(Murukaṉ) in Āviṉaṉkuṭi." As seen below, this constraint determines the rhythmic structure of the entire verse! I reproduce the last semi-line to show the difference between this structure and the previous example.

tanatāna tantananatiruvāvi ṉaṉkuṭiyiltanatāna tantananavaruvēḷca vuntarikatanatāna tantanana-tanatānācekamēlmey kaṇṭaviral-perumālē

Sometimes this additional constraint is also reflected in the sound class of the verse along with the rhythmic units. For example, consider verse 5.315, which addresses Murukaṉ in Vaḷḷimalai (located near the town of Vellore in Tamil Nadu). The rhythmic structure (5+3+3) is given below, where again I show the last semi-line of the verse (note the presence of the medium sound class). Here, Aruṇakirinātar chooses to describe a "Murukaṉ who climbed Vaḷḷimalai," which construction determines the rhythmic structure of the verse and the variety in sound class.

tayyatta tāna tantavaḷḷikku lāma ṭarntatayyatta tāna tantavaḷḷikkal mītu ceṇṟutayyatta tāna tanta-tanatānāvaḷḷikku vēṇai koṅṭa-perumālē

A fascinating set of examples is found in the songs composed by Aruṇakirinātar in Citamparam. Pillai (1965) lists 64 songs and notes that these verses correspond to 60 distinct rhythmic structures. It is likely that Aruṇakirinātar was motivated by the connection between rhythm, dance, and Citamparam, the abode of the dancing god Naṭarāja.²⁹ In generating this variety, Aruṇakirinātar employs several different words to name (or signify) Citamparam such as Puliyūr, Kaṉakacapai, Kaṉakampalam, Cempoṉṉampalam, Ciṟṟampalam, Tiruvampalam, and so on. The entire exercise is conceived and executed brilliantly.

So far, I discussed the rhythmization of verse as a general principle and the generation of rhythmic variety in the Tiruppukaḻ. I now turn my attention to Aruṇakirinātar's innovation in the Tiruppukaḻ-the final foot.

Cyclicality and the Final Foot

The uniqueness of the Tiruppukaḻ as envisaged by Aruṇakirinātar is apparent when the structures of his other works (also rhythmic verses) are analyzed. Consider, for example, the Pūta Vētāḷā Vakuppu (9.6) consists of 10 distinct rhythmic lines that are repeated in the same order throughout the verse (which has 80 lines). However, there is an important difference between these verses and the Tiruppukaḻ in that the former do not have the final foot. In introducing the final foot in the Tiruppukaḻ, I propose Aruṇakirinātar was motivated by another ingredient necessary to transform verse to song-cyclicality.

Cyclicality in Tamil verse is much older than the Tiruppukaḻ. For instance, one finds verses in the Cilappatikāram, called paṭṭu (lit. song), where a specific line is used at the end of every verse. Based on this, it is reasonable to speculate that the last line served the purpose of a refrain. Similar examples are also found in Tamil bhakti literature prior to the Tiruppukaḻ. The utility of cyclicality lies in its ability to direct the expression of song toward a motif-rhythmic or melodic. I will focus my attention on the latter, since rhythmic cyclicality is evident in the construction of the Tiruppukaḻ, and also because melodic cyclicality leads me to the final topic of discussion in this paper, the interplay between rhythmic structure and melodic expression.

Today, the performance of Tiruppukaḻ is mostly heard in temples where groups of singers gather to sing the poems.³⁰ Usually, the group features a lead singer who sings the semi-line (or the line). The same semi-line is then repeated by the group. In this performance, the final foot, which acts effectively as the termination of the semi-line, is used as a transition between the lead singer and the group (and vice versa). Beyond this obvious function of connecting the (text of the) semi-lines in a verse, and its performance, the final foot directs the singer back to the beginning of the (same) semi-line. Thus, it also provides a natural way to introduce melodic variations. This feature is used mostly in concerts (where a single vocalist performs). The last syllable of the final foot is often a long vowel (even if it is not, the rule for the canta viruttam allows for an extension in its duration and the short vowel "may assume the function of the guru"). In practice, this means the singer gets more room for melodic embellishment and allows for a dilation of the rhythmic cycle. Similarly, when a Tiruppukaḻ is set to music and choreographed, the semi-line with the final foot affords a convenient unit for elaboration.

When a Tiruppukaḻ verse is sung as part of a concert repertory, the musician often uses the flexibility offered by the final foot either to make the rhythmic structure of the entire semi-line (and thus, the song) homogeneous or to fit the rhythmic structure into one of the tālas belonging to the concert repertory. Some examples of this will be found in the following section.

Interplay between Melody and Rhythm

In this final section, I discuss the interplay between rhythm and melody, a connection that can be restated, in the light of my discussion thus far, as one between prosody and melody. Though not the central theme of this paper, I mention a few salient points on this interesting subject, with the hope that they lead to further research. With the exception of a few general remarks, I shall restrict my discussion in this section to Aruṇakirinātar's Tiruppukaḻ. I emphasize that unlike the rhythmic expression of Tiruppukaḻ (which I argued can be deduced from the metrical structure), there are no references in the literature as to melodic expression (i.e., we do not know the rāgas in which specific verses were sung).³¹ Consequently, I will focus my attention only on general themes that concern the interplay between rhythm and melody.

As mentioned earlier, the metrical structure of the Tiruppukaḻ can be represented by the sequence, A₁ B₁ A₂ B₂ A₃ B₃ A₄ B₄. Here, A (B) represents the first (second) half of a semi-line. Recall that all semi-lines have the same rhythmic count and share the same structure. Further, alliteration in the first and second syllables is found between A₁ . . . A₄. For these reasons, the obvious and the most typical melodic structure of the Tiruppukaḻ is also the simplest bipartite structure, M₁ M₂ M₁ M₂ M₁ M₂ M₁ M₂, where an entire verse comprises two melodic units that are repeated. This melodic structure may be contrasted with that of the contemporary kṇti in Carnatic music, represented by M₁ M₂ (M₁) M₃ (M₁).

It may be asked why the Tiruppukaḻ does not allow for a structure with, say, 3 or more melodic units as in the kṟti format. The answer lies in the factor that we have not considered so far in this paper-the meaning of the text. Typically, the final foot of a semi-line leads to the beginning of the following semi-line. As I remarked in the previous section, Tiruppukaḻ performance today is mostly in temples where groups of singers/devotees gather to sing these poems. In these renderings, the meaning of the text gains precedence over melodic expression. Often, the lead singer construes parts of a semi-line with parts from a different semi-line to build a semantically complete unit.³² Such exercises place a severe restriction on the melodic structure. Furthermore, there are many examples in the Tiruppukaḻ where one semi-line with the final foot can be repeated without compromising the meaning of the text. However, there is no regularity in the distribution of such semi-lines in a verse. For these reasons, a simpler melodic structure with 2 units is the most natural choice and widely seen in practice. I mentioned earlier that the final foot offers the performer some scope for melodic improvisation, but this is only in terms of returning to the same semi-line. Sometimes this compromises the meaning of the text by impeding the textual flow (which would require the singer to proceed to the second half of the semi-line). Again, this is in contrast with the format of the kṇti, where the text usually allows the performer to go back to M₁ after singing M₂.

These points are well illustrated in the melodic expression of the popular verse, "kaittāla niṇaikaṇi . . . " (1.1). My analysis below³³ and as notated in Figure 1 is based on M. S. Subbulakshmi's vocal rendering of this song, which can be heard online.³⁴ Subbulakshmi sings this verse in rāga Nāṭṭai. The rhythmic structure of the semi-line (with the final foot) is tattana tanatana tattana tanatana tattana tana-tana-tanatānā. Following the rules outlined earlier, this structure corresponds to 30 mātrās partitioned into (4+4+4+4+4+4+6). Now, the modern unit of measure, the akṣara is 1/4 of the units I use in this paper, and hence this song should (by this count) correspond to a tāla of 7 1/2 akṣaras. However, this song is almost always sung (as in Subbulakshmi's version) in the popular āti tāla, which has 8 akṣaras. From the point of view of a performing musician, this transformation is almost asking to be done. The āti tāla comprises 32 mātrās (using the system of units followed in this paper) partitioned into (4+4+4+4+4+4+4+4); so if the final foot of the verse can be extended by 2 mātrās, the verse would fit into a tāla which is part of standard concert repertory. Here, the long vowel(s) in the final foot play an important role, allowing the musician to dilate it, consequently transforming the rhythmic structure of the verse. Further note the melodic variations in the final foot and also how the final foot is used to return to the beginning of the semi-line for melodic variations of the semi-line (cyclicality). The bipartite melodic structure of Tiruppukaḻ is also evident in this example. Subbulakshmi employs a device used often in such cases to relieve monotony-the semi-lines 3 and 4 are sung at twice the tempo of the preceding semi-lines.

Figure 1.

M. S. Subbulakshmi's vocal rendering of "kaittala niṟaikaṉi . . . " in rāga Nāṭṭai (1.1).

The example depicted in Figure 1 typifies the changes made to the rhythmic structure by performers. These changes are motivated by the deśīre to homogenize the partitioning of mātrās in nearly homogeneous structures, and (or) to facilitate melodic variation within a bipartite melodic structure. This set of changes effected by musicians (in the performance of Tiruppukaḻ and more generally, any rhythmic verse) follows the dictum that rhythmic structure begins to play a less important role when melodic expression gains precedence. Consequently, the "additive" structure of the Tiruppukaḻ is turned into a "divisive" one with the components of the tāla having equal duration. A corollary is that the meaning of the text also becomes less important under these conditions. For instance, in the above example, textual flow would require the singing of the second half of the semi-line immediately following the final foot (of the first half). But this requirement is abandoned for the sake of introducing melodic variety. Indeed, while making such changes, the musician follows the same set of steps that led to the evolution of genres such as the kṛti where melodic expression often gains precedence over rhythmic structure and textual flow.

The putative connection between expressions of tāla and rāga has been debated extensively in the context of the Tēvāram (a corpus of bhakti literature in Tamil) without any consensus. Surprisingly, the question has not been raised with reference to Tiruppukaḻ, which given its rhythmic variety, would seem the obvious place for investigation. Although we do not know if Tiruppukaḻ verses were sung to specific ragas in the times of Aruṇakirinātar, this lack of information need not deter me. For the question at hand does not concern authenticity of melodic expression; rather I seek the relationship between rhythmic and melodic structures.

During the last 50 years, collections of Tiruppukaḻ have been set to music (and notated) by a few musicians. Consequently, there is enough material available to make such a study feasible. It would be interesting to study the influence (if any) of the rhythmic structure on the melodic contours fashioned by the musician, say by comparing verses set in the same rāga, but sung in different tālas, or by comparing the melodic structures given by different musicians to the same verse (and by definition, the same tāla). To my knowledge, this study has not been undertaken yet. Below, I present two examples that illustrate the questions that can be raised by such a study.

The first example is a verse set to music by A. S. Raghavan (see Figure 2 for notation). The rhythmic structure of this verse is tanatana tanatana tanatana tanatana tanatana tanatana-tanatānā. The sixfold repetition of taṉātaṉa (4 units) is to be noted. Raghavan makes the same minor transformation to the rhythmic structure of the verse as in the previous example; viz., two mātrās are added to the final foot, making the total rhythmic count 32, and the verse is sung in ati tāla. In what follows, it is only the structure of the semi-line that matters and so, this modification of the rhythmic structure is irrelevant. Raghavan chooses to sing this verse in the rāga Harikāmpōti. Unlike the previous example, the melodic structure is of the form M₁ M₂ M₃ M₄ M₁ M₂ M₃ M₄ (i.e., the first four semi-lines have distinct melodic lines). However, the melody is strongly tied to the rhythmic structure as seen in the notated sample (see Figure 2). Raghavan's treatment is quite interesting in that he chooses to emphasize the partitioning of the verse (and equivalently, the tāla) in units of 4. This partition is forced on the melodic line. As a result, the distribution of svaras follows the same distribution of 4 mātrās in the rhythmic structure, leading to phrases such as SS nD nn DP DD PM PP MG (scale degrees 88 b76 b7b7 65 66 54 55 43) in measure 1 (see Figure 2), leading to an aural perception of an atypical Harikāmpōti. Indeed, were it not for the underlying rhythmic structure, such phrases would almost invariably sound incongruous in the presentation of this rāga. Note the requirement of melodic structure mimicking the rhythmic structure (svara following the tāla) leads to an awkward RGMGS (23431), a phrase that is avoided in typical renderings of Harikampōti in favor of GMGRS (34321).

Figure 2.

A. S. Raghavan's vocal rendering of "tirumoli yuraipera . . . " in rāga Harikāmpōti.

Both examples above indicate the melodic limitations posed by a strict adherence to rhythmic structure. In some instances, however, the rhythmic structure can be used to enhance specific melodic features. This is illustrated in the following verse (7.1297), paravaikkettaṉi, sung commonly in the rāga ūrṇacantirika.³⁵ The rhythmic structure of the verse (including the final foot) is shown below (see Figure 3 for melodic notation).

tanatat tattana-tanatānāparavaik kettaṉāi-vicaitūtu

Figure 3.

Metric expansion of the verse, "paravaikkettaṉāi," to accommodate melodic features of Tyagaraja's rāga Purṇacantirikā.

There are 14 mātrās in this verse, according to the rules described earlier. However, in performance this duration is expanded to 24 mātrās. The expansion (for reasons that will be discussed shortly) is uneven and the rhythmic structure changed; the tāla is expanded at specific points. Unlike the previous examples, these changes are made for considerations of melody. The verse is sung in the rāga Pūrṇacantirikā. The characteristic phrases of this rāga are RG M- R- (234-2-) with stops M-(4-) and R- (2-) following RG (23), which is sung at a faster tempo, PDPS' (5658). In general, the rāga follows the following contours: SRGMPDPS' (12345658); S'NPMGMR-S (87543421). The phrase PDPS' (5658) in the ascent and PMGMR- (54342-) with a rest at R in the descent are crucial in establishing the identity of the rāga. This particular form of Purṇacantirikā was first introduced by the composer, Tyagaraja, who also composed in an allied rāga, Janarañcaṇi.³⁶Figure 3 shows how this form is realized in the verse. The expansion at the end of the first semi-line is to emphasize the position of R (2) as a note of rest. Similarly, note the expansion on Ŕ (9) at the end of the third semi-line, following the triplet ŔŔḾ (91011). Note the important role played by the rhythmic duration of the consonant syllables in this verse; for example, the rhythmic duration of the syllables vaik and ket (in the first semi-line paravaikkettaṉāi) is 2, which provides the necessary stops for the svaras M and R in the key phrase, RGM-R-. A detailed study of the interplay between melody and rhythm in the Tiruppukaḻ verses along these lines is yet to be done.

Summary

In this paper, I discussed the construction of rhythmic verse in Tamil, choosing the works of the Tamil poet Aruṇakirinātar as examples. I argued that (1) the imposition of an additional set of constraints to the fundamental rules of prosody and (2) the introduction of cyclicality transform a verse into song, a process I called rhythmization. The set of rules used in the construction of such rhythmic verses also allows the composer to introduce rhythmic variety. This is achieved by a one-to-one mapping between syllabic and rhythmic structures made precise by the rules of Tamil prosody. Thus, variety in verse is also rhythmic variety. However, rhythmic variety in the Tiruppukaḻ goes beyond structural complexity and is more apparent in its manifestation as variety in sound; this latter property is identified as the defining characteristic of the deśī tālas by the 13th century musicologist Śarṅgadeva. Thus, the Tiruppukaḻ verses represent some of the best examples of variety in rhythmic and aural color in Indian music. Elsewhere, I have argued that the construction of the Tiruppukaḻ may be considered as instantiating the notion of "three-fold Tamil" or muttamiṇ, the functionality of language which allows for multiple modes of expression as in speech, song, and mimetic rendering (Muthukumar 2008).

In the case of rhythmic verses that are performed or sung such as the Tiruppukaḻ, the basic connection between prosody and prastāra (the exercise of introducing tāla variety) discussed in this paper exemplifies the method of Tālaprastāra described in the 13th century musicological texts such as Parsvadeva's Samgīta Samayasāra and Śarṅgadeva's Samgīta Ratnākara. Two points merit my attention here. (1) What are the origins of Tālaprastāra? There are no references to this technique in the literature on musicology prior to the two texts mentioned above. Some scholars (Sathyanarayana 2001) have speculated that Śarṅgadeva follows the Jain tradition (of Pārśvadeva) in his discussion of Tālaprastāra. One may then ask what Pārśvadeva's sources were. I have attempted to show in this paper that the method of introducing tāla variety can be identified with the method of introducing variety in rhythmic verse.³⁷ Thus, the roots of prastāra may well lie in the grammar of poetry. Indeed, detailed accounts on the prastāra of verse are found in the Yāpparuṅkala Virutti and the Vīracōḻiyam, both being works on Tamil grammar. The former work and its primary text (Yāpparuṅkalam) are attributed to grammarians in the Jain tradition. I am thus tempted to speculate that the discussion of prastāra in Tamil grammar preceded the discussion of Tālaprastāra in musicology. (2) Śarṅgadeva introduces Tālaprastāra as a scheme to classify the so-called deśī tāla, and as I argued in this paper, the rhythmic structures seen in Aruṇakirinātar's Tiruppukaḻ and the manner in which their varieties abound are in complete agreement with Śarṅgadeva's description of the deśī tāla. It may then be asked if the tālas found in the Tiruppukaḻ are indeed the deśī tālas of Śarṅgadeva. This question is rendered moot by the fact that we do not have examples of compositions (in vogue during Śarṅgadeva's times) in deśī tālas. Furthermore, as the examples discussed in this paper show, the rhythmic structure of the Tiruppukaḻ verse can be identified readily using the rules of grammar. Hence, embarking on an exercise of fitting each Tiruppukaḻ verse to any of the deśī tālas available in tāla compilations seems entirely superfluous.

I conclude with a remark concerning the influence of Tiruppukaḻ on subsequent poets and composers. Aruṇakirinātar's influence is evident in some later works modeled on the same lines as the Tiruppukaḻ. Examples include poetry composed by Kācim Pulavar (c. 18th century) and the Cantat Tiruppukaḻ of Aptul Kātir.³⁸ However, a more subtle influence of Aruṇakirinātar's work may well lie in the emergence of the so-called coṟkaṭṭu in (South Indian) compositional forms prevalent in the 17th and 18th centuries. These compositions feature lines of rhythmic syllables inlaid within a song. These rhythmic patterns are either identical to the underlying rhythmic structure of the song or complement it. The possible connection between these compositions and the Tiruppukaḻ is an interesting topic that needs further research, and will be discussed elsewhere.