3. Trajectories in the continuing study of parallelism

Introduction

When my paper ‘Roman Jakobson and the comparative study of parallelism’ was published in 1977, I imagined—or perhaps, it would be better to say, I hoped—that Jakobson’s influence and his persuasive argument for the critical importance of parallelism in understanding poetic language would prompt an explosion in the study of parallelism at a wider comparative level. Yet no such explosion occurred.

Instead, since 1977, there has been a steady development in the study of parallelism occasionally marked by new materials and flashes of analytic insight. This development has occurred, for the most part, in a number of separate intellectual silos defined in terms of particular language groups with relatively little reference to other comparative linguistic traditions of parallelism. A program for the study of parallelism as a ‘near-universal’ (as Jakobson phrased it) has barely begun. In my own work over the past decade, I have to recognise that I, too, have concentrated my attention primarily on the comparative occurrence of canonical parallelism within the Austronesian language family.

For this volume, I feel that it is appropriate to update my 1977 paper: 1) to highlight important developments in the study of parallelism, particularly canonical parallelism; 2) to document the continuing recognition of new traditions of canonical parallelism; and, in the process, perhaps 3) to venture comparative comment on parallelism as a general but situated linguistic phenomenon. It is the contention of this chapter that there is a host of similar semiotic processes evident in the use of parallelism in distinct traditions throughout the world but it is only when one examines the range of these traditions that it becomes possible to glimpse these similar semiotic processes. I offer this summary excursion as an addendum to my original paper and as a prelude to my own research included in this volume. Given my concern in attempting to recognise similarities across different traditions, I will be strategically selective in my comparative examination and commentary but will try to be more inclusive in citing references to recent work within the field and to issues that relate to my own research on parallelism.

Biblical scholarship: The Hebrew and Ugaritic traditions

No area of study has produced a greater volume of research touching on parallelism than that of Biblical studies. Yet despite the fact that the comparative study of parallelism was initially given impetus by Robert Lowth’s researches, the extensive and erudite study of Biblical parallelism remains largely a self-referencing field and only rarely draws upon, or contributes to, a wider global discussion. Nevertheless, the study of Biblical parallelism offers considerable value for comparative consideration.

In this regard, there has been a succession of publications since 1977 that require consideration. The first of these is James J. Kugel’s The Idea of Biblical Poetry: Parallelism and its history (1981); the second is Wilfred G. E. Watson’s Classical Hebrew Poetry: A guide to its techniques (1984), which includes a particularly useful chapter on parallelism; and the third is Adele Berlin’s The Dynamics of Biblical Parallelism (1985), which is the most significant and relevant of recent studies. Two other volumes that take a specific, detailed look at parallelism are also worth including in this list: Robert Alter’s The Art of Biblical Poetry (1985:1-28) and David L. Petersen and Kent Richards’ Interpreting Hebrew Poetry (1992: 21-35).

Kugel’s volume is a work of considered, conventional Biblical scholarship. His focus is on what he calls the ‘parallelistic line’, which he defines as the ‘basic feature’ of the Hebrew poetic tradition for the expression of songs, sayings, proverbs, laws, laments, blessings and prayers. The parallelistic line consists of two brief clauses, separated by a pause with the second clause associated with, linked to, or corresponding with the first clause. The connection between these two clauses can correspond closely in syntactic structure and semantics but can also be loosely associated with the first clause. In Kugel’s assessment, the ‘majority of parallelisms in the Bible fall between such extremes…complete correspondence is relatively rare’ (Kugel 1981:2–3). Kugel cites Jakobson’s 1966 paper on parallelism, but is concerned mainly but only briefly with comparative considerations of Semitic parallelism.

Drawing upon a considerable range of scholarly research, Watson’s volume provides a more incisive discussion of parallelism and its occurrence in the Hebrew, Ugaritic and Akkadian traditions of poetic composition. His discussion covers a range of general issues in parallelism as well as others that are specific to Semitic parallelism. Thus, for example, he calls attention to the common occurrence of ‘gender-matched parallelism’ in both Hebrew and Ugaritic. This form of parallelism was first identified by Umberto Cassuto in his study The Goddess Anath: Canaanite epics of the patriarchal age (1971), and was subsequently shown to be a recurrent feature in Hebrew compositions as well. A number of studies have been done on this form of parallelism (Berlin 1979; Watson 1980).

Watson gives various examples of this ‘gender-matched parallelism’.

From Isaiah (49, 22):

They shall bring your sons in their embrace (m.),

and your daughters shall be carried on their shoulders (f.).

From Jeremiah (48, 37):

On every pair of hands (f.), a slash (f.),

On every pair of hips (m.), sackcloth (m.).

From Ugaritic poetry:

Scarcely had his word (m.) issued from his mouth (m.),

From his lips (f.), his word (f.).

And from Akkadian:

In the city the young girl’s song (m.) is altered

In the city the young man’s tune (f.) is altered.

Watson also compares ‘number parallelism’ in Hebrew and Ugaritic—a subject first broached by Ginsberg and discussed extensively by Gevirtz (1963:15–24, 29–30). In Hebrew, only some of the numbers from 1 to 7 can pair with their successor digit, thus: (x//x+1); 7//8 is of frequent occurrence but 8//9 is not used.

From the Psalms (62, 12):

Once hath God spoken,

Twice have I heard this.

From Proverbs (30, 15):

Three things are never satisfied,

Four never say: ‘Enough’.

From Proverbs (6, 16):

Six things doth YHWH hate,

And seven are his disgust.

From Ecclesiastes (11, 2):

Give a portion to seven,

And also to eight.

The pair 7//8 can be multiplied by 10 or more often by 11 to indicate larger quantities: 70//80 or 77//88. A variation on this combination is ‘sevenfold//seventy-seven’.

From Genesis (4, 24):

For sevenfold is Cain avenged

but Lamech, seventy and seven!

The Hebrew tradition also uses the pair: thousand//ten thousand.

From I Samuel (18, 8):

They have given to David ten-thousands

And to me have they given the thousands.

The Ugaritic tradition uses virtually the same number pairs, similar multiplication by 11 and combines one thousand with ten thousand.

Let bread be baked in the fifth

Food for a sixth month.

Seven years may Baal fail

Eight, the Rider of the Clouds.

Sixty-six cities did he seize

Seventy-seven towns.

He took a thousand pitchers of wine

Ten thousand he mixed in his mixture.

Akkadian, by contrast, makes less use of number parallelism and most occurrences of this parallelism are in incantations. Watson notes the existence of number parallelism in Phoenician and Aramaic, but makes no reference to similar forms of parallelism in other non-Semitic traditions. Such number parallelism is common in other traditions of parallelism and thus offers a specific point of comparison.

Much of Watson’s discussion of parallelism is taken up with a consideration of what he calls ‘parallel word pairs’. This is the term he prefers to a plethora of other terms in the Biblical literature—‘standing pairs’, ‘fixed pairs’, ‘A-B pairs’ or ‘parallel pairs’—used to identify the canonical semantic pairs.

The critical focus on these pairs began with the discovery in 1928–29 of the repository of Ugaritic texts at Ras Shamra in northern Syria and the decipherment of these texts, written, as they were, in a distinctive alphabetic script using cuneiform. Ginsberg (1935) was the first to note these pairs, which he called ‘standing pairs’. He described them as ‘certain fixed pairs of synonyms that recur repeatedly, as a rule in the same order’ (1936). Others followed Ginsberg’s lead, most notably Gevirtz in his seminal study Patterns in the Early Poetry of Israel (1963). Recognition of these pairs and their compilation has itself become a substantial undertaking. Michel Dahood has made the most prominent contribution to this task. Initially in ‘The grammar of the Psalter’ (1970), written with Tadeusz Penar, Dahood compiled an extensive list of Hebrew/Ugaritic parallel word pairs that occur specifically in the Psalms; he then went on to publish three further compilations (Dahood and Penar 1972; Dahood 1975, 1981), bringing to 449 his count of parallel pairs.

In this connection, Watson notes the use of rare terms that combine with more commonly recognised terms. Thus, in an A-B pair, one term, the A word, whose occurrence is relatively common, can combine with other ‘rare and esoteric words’. An understanding of these esoteric words derives from their relation to their more common partnered terms. Also implied in this is the possibility that any particular term can form a pair with a number of other terms. This opens up the possibility of analysis that goes beyond the listing of fixed word pairs. The combination of two terms—one common and the other, in some way, unusual—is of frequent occurrence in other traditions of semantic parallelism; in some cases, the ‘rare’ term may originate from another dialect or even from another language.

Watson also discusses the use of chiasmus—the reversal or inversion of word order in consecutive verses—as a common device for poetic composition in Hebrew (1984:201–13). It was Dahood, however, who had earlier linked this compositional technique to the use of synonymous pairs. Basing his argument on a host of examples taken from Job, Dahood argued that ‘when the poet uses the chiastic word order, the synonymy of the parallel members tends to be stricter than when the order is not chiastic’ (1974:120). A couple of examples may serve to illustrate this feature in Hebrew poetry.

From Job (28:2):

Iron is taken from ore,

And from smelt rock, bronze.

From Job (32:14):

I shall not marshal against him your arguments,

And with your words I shall not rebut him.

In his discussion, Watson also considers what might be called the ‘oral hypothesis’ adopted by Biblical scholars to explain the variability in the use of fixed word pairs in different parts of the Bible. He considers in particular a paper by Perry B. Yoder, ‘A-B pairs and oral composition in Hebrew poetry’ (1971), which articulates this position—a position that can in fact be traced back to Ginsberg and Gevirtz.

For Yoder, the ‘stock of word pairs was not the work of any individual poet…but it was the poetic inheritance of each generation of poets’ (1971:472). The use of this inheritance varies considerably in different passages of the Bible. Yoder cites Psalm 54 as an example of the ‘high-density’ use of such pairs. Watson, in his translation, italicises those word pairs on which the psalm is constructed: name//might, save//defend, hear//give ear, prayer//words of my mouth, foreigners//vicious men, risen up against me//seek my life, helper//supporter, evil recoil//destroy, sacrifice//praise your name, enemies//foes.

Psalm 54:

O God, by your name save me

and by your might defend me

O God, hear my prayer

give ear to the words of my mouth

For

foreigners have risen up against me

vicious men seek my life;

See!

God is my helper

the Lord really is the supporter of my life

Making evil recoil on my slanderers

In truth, he really destroyed them

For generosity I will sacrifice to you

I will praise your name, Yahweh, for it is good.

For

from all my enemies he rescued me

and my eyes gloated over my foes.

Yoder regards ‘the technique of parallelistic composition by the use of traditional word pairs as a technique developed by oral poetic traditions to meet the needs of oral poets’ (1971:483). Invoking the work of Milman Parry on oral formulaic expressions in Homer, Yoder then goes on to argue that for poets in the Ugaritic–Hebrew tradition the use of fixed pairs was equivalent to the use of formulaic expressions in the Homeric tradition. Since there is a great deal of variation in the ‘density’ of the use of fixed pairs, particularly in the Bible, it follows, as Yoder indicates, that wherever this density is high, the text reflects a strong oral poetic influence.

Yoder’s analysis is essentially an extended supportive restatement of Gevirtz’s early argument that ‘the poets of Syria and Palestine had at their command a body of conventionally fixed pairs of words upon which they might freely draw in the construction of their literary compositions’ (1963:38).1

Watson also enumerates and briefly summarises the varieties of forms of parallelism that occur in Hebrew and Ugaritic compositions: staircase parallelism, synonymous-sequential parallelism, noun–verb parallelism, and vertical parallelism. One of the useful aspects of Watson’s summary discussion of research on parallelism within the Semitic field is his inclusion of short, relevant bibliographies for each of the topics he discusses.

One of the most valuable studies of Biblical parallelism to be published to date is Adele Berlin’s The Dynamics of Biblical Parallelism (1985). Berlin takes her cue from Lowth—not by arguing over his specific claims but rather in recognising, as she writes, that Lowth ‘was right about the essence of parallelism: it is a correspondence of one thing with another’ (1985:2). More significantly, Berlin explicitly adopts Jakobson’s perspective on poetic language as her framework of analysis—quoting Jakobson’s famous dictum that ‘the poetic function projects the principle of equivalence from the axis of selection into the axis of combination’. She describes this pronouncement as ‘piercingly insightful and maddeningly general’ (1985:7), but she carefully elaborates and develops on Jakobson’s ideas, drawing on the exegeses on these ideas by Linda Waugh (1980). Her stated goal ‘is to present an overarching, integrated and linguistically based description of biblical parallelism’ (1985:29).

Berlin refers to my 1977 paper on the comparative study of parallelism but her primary interest is in a thoroughgoing analysis of the Hebrew tradition of poetry, and much of her book is taken up in a dialogue with a variety of other Biblical scholars—in particular, Kugel—who have adopted differing conceptions of the parallelism and its significance. She divides her study into sections on the examination of: 1) grammatical, 2) lexical-semantic, and 3) phonological parallelism, and each of these sections is further subdivided into varieties of parallelism. Thus, for example, gender parallelism, noun/pronoun parallelism, or contrasts in person, number or case are separately discussed and illustrated under the rubric of grammatical parallelism. Berlin also examines what she calls positive–negative parallelism: pairings in which a corresponding line or phrase is just the ‘negative transformation’ of the previous line or phrase:

From Proverbs (6, 20):

Guard, my son, the commandment of your father

And do not forsake the teaching of your mother.

From Deuteronomy (9, 7):

Remember, do not forget.

Berlin’s discussion of lexical-semantic parallelism leads to a partial but not particularly successful attempt to explain the specific pairing of words by different linguistic rules such as marking or minimal contrast. She concludes her study with a valuable consideration of the ‘expectation’ and ‘effect’ of parallelism. The book, as a whole, is a remarkably rich work of detailed Biblical scholarship written in a new key within the discipline. As such, its conclusion is worth quoting:

Parallelism, then, consists of a network of equivalences and/or contrasts involving many aspects and levels of language. Moreover by means of these linguistic equivalences and contrasts, parallelism calls attention to itself and to the message which it bears. Parallelism embodies the poetic function, and the poetic function heightens the focus on the message. (Berlin 1985:141)

The earliest evidence of parallelism: The Sumerian tradition

Since 1875, Sumerian studies have been engaged in piecing together the scattered fragments of cuneiform tablets stored in museums and other repositories around the world in an effort to restore the diverse literary creations of Sumer. No-one has contributed more to this immense effort and to the translation of works of great importance than Samuel Noah Kramer. His book From the Poetry of Sumer (1979) is a landmark study that provides a glimpse into this poetic tradition, much of it characterised by pervasive parallelism.

Despite evident parallelism, the fragmentary nature of these cuneiform texts often leaves the interpretation of their mythic meaning of many passages uncertain. One of these fragmentary poems that describes a fecund world impregnated after a great flood—a fragment that Kramer specifically labelled as ‘perplexing, problematic and enigmatic’ (1979: 30)—nonetheless provides a good illustration of this complex Sumerian parallelism:

The old man instructed, the old man exhorted

After the rain had poured down, after it had demolished walls;

After hailstones and firebrands had poured down,

After man had confronted man defiantly,

After there had been copulation—he had also copulated.

After there had been kissing—he had also kissed,

After the rain had said: ‘I will pour down,’

After it had said: ‘I will demolish walls,’

After the Flood had said: ‘I will sweep everything away,’

Heaven impregnated, Earth gave birth,

Gave birth to the numun-plant, also,

Earth gave birth, Heaven impregnated,

Gave birth to the numun-plant, also,

Its luxuriant reeds kindled fires. (Kramer 1979:32)

Some years after the publication of From the Poetry of Sumer, Kramer began a productive collaboration with the poet and folklorist Diane Wolkstein. Together they produced a volume, Inanna: Queen of Heaven and Earth (1983), which draws together an extensive assemblage of ritual compositions regarding the goddess Inanna and puts these retranslated pieces together in a coherent fashion. While Kramer and Wolkstein each contributed separate background essays, the translation is a joint effort. The result is a work of exceptional scholarship lifted to a literary plane. Virtually every passage in the volume is expressed in parallelism.

Some of the most remarkable poetic passages in Inanna: Queen of Heaven and Earth can be found in the dialogue between between Inanna and her lover and consort, Dumuzi, the shepherd. The parallelism in these passages is as explicit as it is erotic and multiple metaphors it produces reflect Sumerian ritual concerns with fertility and fecundity. The following are excerpts of dialogue from the Courtship of Inanna and Dumuzi:

Inanna spoke:

“What I tell you

Let the singer weave into song.

What I tell you,

Let it flow from ear to mouth,

Let it pass from old to young:

My vulva, the horn

The Boat of Heaven,

Is full of eagerness for the young moon

My untilled land lies fallow.

As for me, Inanna

Who will plow my vulva?

Who will plow my high field?

Who will plow my wet ground?

Dumuzi spoke:

“Great Lady, the king will plow your vulva.

I, Dumuzi the King, will plow your vulva.”

Inanna sang:

“Make your milk sweet and thick, my bridegroom.

My shepherd, I will drink your fresh milk.

Wild bull, Dumuzi, make your milk sweet and thick.

I will drink your fresh milk.

Let the milk of the goat flow in my sheepfold.

Fill my holy churn with honey cheese.

Lord Dumuzi, I will drink your fresh milk.”

Dumuzi spoke:

“My sister, I would go with you to my garden.

Inanna, I would go with you to my garden.

I would go with you to my orchard.

I would go with you to my apple tree.

There I would plant the sweet, honey-covered seed.”

Inanna called for the bed:

“Let the bed that rejoices the heart be prepared!

Let the bed that sweetens the loins be prepared!

Let the bed of kingship be prepared!

Let the bed of queenship be prepared!

Let the royal bed be prepared!”

(Wolkstein and Kramer 1983: 36-42)

Parallelism in Sumerian poetry relies as often on triplet sets as on dyadic sets and often these triplet sets consist of a succession of an initial general term followed by more specific terms: garden, orchard, apple tree.

As a student of Stanley Kramer, Adele Berlin’s first major publication was a monograph on a critical Sumerian text: Enmerkar and Ensuhkešdanna: A Sumerian narrative poem. This narrative poem, made up of 282 lines, features a contest between the ruler of Uruk and the ruler of the Aratta over the right to marry the goddess Inanna. For this study, Berlin identified the cuneiform tablets from which she constructs her text and then provided a transliteration and translation of her constructed text with an extensive critical commentary. She devoted considerable attention to the use of parallelism throughout the text. She refers to the succession of parallel terms, as in the triplet sets noted by Wolkstein, a ‘progression’ of parallel terms in successive linesand provides examples of this kind of parallel progression:

Like a wild donkey of Sakan, he runs over the mountains

Like a large powerful donkey, he races

A slender donkey, eager to run, he rushes forth.

(Berlin 1979:lines 45–7)

Such parallel enumeration can continue for many lines.

Berlin’s interest is in the varied uses of parallelism in composition, rather than identifying recurrent formulae or particular fixed pairs. In the current state of textual reconstruction, emphasis continues to focus on the scholarly construction and expansion of the corpus of Sumerian poetry. This corpus may not yet be at a stage at which it is feasible to identify the full range of recurrent semantic pairs.

Thorkild Jacobsen is another distinguished scholar of Sumerian culture. His book, The Harps That Once…: Sumerian poetry in translation (1987) is a rich compendium of Sumerian ritual and poetic texts with valuable annotations and commentary. In his collection of texts, many in parallelism, is a ritual text that is believed to date from the time of the restoration of the Sumerian capital of Ur after the fall of the Third Dynasty and the sacking of Ur around 2004 BC. This cuneiform text is a harp-lament for the destruction of Ur but is thought to have been used in a ritual for the restoration of Ur around 1940 BC. The entire lament consists of 413 lines, almost all in strict parallelism. It is a formulaic composition that progresses by the ordered recitation of the names of particular gods or goddess and of the specific sites of the temples dedicated to them. As such, it is almost certainly the earliest example of what I describe in Chapter 11 of this volume as a ‘topogeny’—an ordered recitation of successive place names, the spatial equivalent to the ordered recitation of personal names in a genealogy. Although not confined to traditions of parallelism, the recitation of topogenies is a common feature of many traditions of parallelism throughout the world.

This Sumerian topogeny is a recitation of abandonment. The invasion by the Elamites is likened to a storm that forces a shepherd to abandon his byre and sheepfold. The storm strikes one Sumerian city after another and destroys its temple, forcing the god or goddess of that city to abandon it. The topogeny is thus an ordered recitation of the cities of Sumer, the main temple in each city and the gods who held sway in those temples. The ‘Lament for Ur’ begins with the following stanza and continues in steady succession with only slight alteration to the same formula:

His byre he was abandoning,

and his sheepfold, to the winds

the herder was abandoning his byre

and his sheepfold, to the winds,

the lord of all lands was abandoning it

and his sheepfold to the wind,

at the temple close, Enlil was abandoning Nippur

and his sheepfold to the wind,

his consort Ninlil was abandoning it

and her sheepfold, to the winds

at their dwelling house, Ninlil was abandoning Kiur

and its sheepfold, to the winds. (Jakobsen 1987:447–74)

This formula is repeated to produce an ordered progress—a procession—of gods and goddess, each of whom abandons their city and temple. At its conclusion, the text launches into yet another long topogeny based on another formula that proceeds systematically through the cities and temples of Sumer, again linking a city, its temple and its god or goddess. A brief excerpt from this second topogeny gives a sense of the beauty of its reiterative parallelism:

Brickwork of Ur, bitter is the wail

the wail is set up for you!

Ekishnugal, bitter is the wail,

the wail is set up for you!

Temple close Agrunkug, bitter is the wail

the wail is set up for you!

The evidence of parallelism in early Egyptian

Parallelism is also evident in early ritual compositions in Egypt. Writing of Egyptian literature, Miriam Lichtheim identifies three literary styles, one of which she designates as an ‘intermediate style’ between prose and poetry:

The intermediate style…is characterized by symmetrically structured sentences. It was employed exclusively in direct speech. Hence I call it ‘symmetrically structured speech,’ or, the ‘orational style’…In Egyptian as in biblical literature, the principle [sic] device that activates the orational style is the parallelism of members. (Lichtheim 1973:11)

One of the oldest examples of this orational style comes from hieroglyphics on the walls of the sarcophagus chambers adjoining rooms and corridors of Unas, the last king of the Fifth Dynasty of the Old Kingdom, dating from approximately 2300 BC. Known as the Pyramid texts, this corpus of incantations was intended to direct the passage of the deceased king to join the sun-god, Re, and to promote his transformation.

The parallelism in utterances 273–4 of these Pyramid texts, which give expression to Unas’ transformation, provides a striking example of the early orational style in Egyptian:

Sky rains, stars darken

The vaults quiver, earth’s bones tremble,

The planets stand still

At seeing Unas rise as power,

A god who lives on his fathers

Who feeds on his mothers…

He has encompassed the two skies,

He has circled the two shores;

Unas is the great power who overpowers the powers,

Unas is the divine hawk, the greatest hawk of hawks,

Whom he finds on his way he devours whole.

Unas’s place is before all the nobles in lightland,

Unas is god, oldest of the old,

Thousands serve him, hundreds offer to him,

Great-Power rank was given him by Orion, father of gods.

Unas has risen again in heaven,

He is crowned as lord of lightland. (Lichtheim 1973:36–8)

The wider Chinese and South-East Asian traditions of parallelism

The role of parallelism and the specific extent of semantic parallelism in early Chinese writing remain sources of speculation. Following in the footsteps of Marcel Granet, who emphasised the insistent binary nature of Chinese traditions, Léon Vandermeersch has argued that the earliest forms of Chinese divination that derived its answers through an interrogation of a symbolic delineation of the turtle’s carapace provided the prototype for literary parallelism (1989). More substantially relevant is C. K. Wang’s examination in The Bell and the Drum (1974) of the ‘stock phrases’ utilised as ‘conventions of composition’ in the creation of the Shih Ching, The Book of Odes. In his paper on parallel structures in the canon of Chinese poetry, David J. Liu has argued that ‘for the ancient Chinese, the concept of parallelism was implicit in the very structure of the universe, manifesting itself in an all pervasive complementary bi-polarity’. He goes on to note that ‘one of the earliest bodies of writing, the I Ching (“Book of Changes”), compiled circa 1100 BC, contains oracular statements in parallel couplets’ (1983:642).

A similar pervasive parallelism marks the verses of the Yuandao, the Chinese treatise on the Dao that forms a crucial component of the Huainanzi, an early Han dynasty compendium of knowledge written in the second century BC around the year 139 BC. The treatise presents the Dao as the ‘oneness of all things’, and the complementary dualism that endows the Dao pervades the parallel language in which it is described.

The opening lines of the Yuandao exhibit parallelism both between lines and within lines:

It [the Dao] shelters the heavens and supports the earth

Extends beyond the four points of the compass

And opens up the eight points of the compass.

It is high beyond reach

And deep beyond reckoning

It envelops the cosmos

And gives to the yet formless.

Flowing from its source it becomes a gushing spring

What was empty slowly becomes full;

First turbid and then surging forward,

What was murky slowly becomes clear. (Lau and Ames 1998:1)

The sixth-century literary commentator Liu Hsieh (456–522), in his grand survey of Chinese literature, now translated as The Literary Mind and the Carving of Dragons (Shih1959), concludes his discussion of the importance of linguistic parallelism in Chinese poetry with an instructive tsan in parallel form from the Li-Chi, The Book of Rites:

A body requires its limbs to be in pairs

A phrase, once forged, must have its counterpart.

With the left hand one lifts; and one holds with the right,

To attain both the essence and the flavor,

Parallelism gleams and dazzles like flowers which are entwined,

Reflecting without distortion like a calm mirror,

It flows in two streams, smooth as jade

Giving rhythm as do the pendant jewels. (Shih 1959:194)

Also intriguingly pertinent to a discussion of parallelism is Hua Wu’s assertion that the seventeenth-century literary critic Jin Shengtan, in his critical discourse on The Water Margins, recognised parallelism as fundamental to all composition. As Hua Wua claims: ‘for Jin Shengtan, parallelism is the basic principle underlying the role of all writing, including that of The Water Margin’ (1988:172). When examined more specifically, this supposed parallelism is itself, in Jin Shengtan’s analysis, based on a set of abstract principles: 1) repetition (chongfu), 2) synonymous substitution (jiaohu ercheng), 3) contrast (jinshu duisheng), 4) gradation (chuanje erchu), and 5) contiguity (chengu pinfa)—all of which are prototypically exemplified in The Analytics of Confucius (1988:171).

R. A. Stern, in his discussion of ancient Tibetan poetry (1972:252-259), compares modes of expression in this poetry to that of the alternating songs of ancient China found in the Shih cheng. He asserts that the “principle of parallelism is so strong that same sentence is often uttered, once with a descriptive expression, once with a corresponding name”. He gives as an example of this in the following lines:

Of the enemy, he cut out the heart;

of the wild yak Karwa, he cut out the heart;

For the kinsman, he achieved vengeance;

for the brother Yikyi Dangcham,

he achieved vengeance.

Stern (1972:253)

Parallelism occurs in a great variety of forms, often with each limb separated by a caesura ni, which Stern translates as ‘yes’:

With a wide mouth, yes, grass he eats,

with a wide neck, yes, water drinks.

Stern (1972:254)

Many of these forms of parallelism have carried on into contemporary Tibetan ritual traditions.

Of notable interest in this regard is the occurrence of ‘binary ritual language expressions’ among various Tibeto-Burman tribal groups in the central and eastern Himalayas. Nicholas Allen was the first to call attention to this form of parallelism (1978). Since then, Martin Gaenszle and his research colleagues have documented the occurrence of ‘binominals’—the term they give to ‘expressions of paired words’—in the ritual speech of other related groups now living in Nepal (Gaenszle et al 2005; 2011). There is evidence of traditions of parallelism as well among the Tibeto-Burman speakers of Nocte, Tangsa and Singpho and among a variety of Tai speakers in Upper Assam and in Arunachal Pradesh in India (see Morey nd). Similarily Anthony Walker has documented the use of parallelism among the Lahu Nyi or Red Lahu, a Tibeto-Burman-speaking hill tribe living in southwestern Yunnan, western Laos and northern Thailand (1972).

Of considerable relevance also to the study of canonical parallelism is the work of David Holm. In his research among the Donglan Bouyei Zhuang, a Tai-speaking population in the north-western highlands of Guangxi, he came upon manuscripts of indeterminate age written in Chinese script for the performance of key rituals: texts for buffalo sacrifice, for funerals and for summoning the souls of the dead (Holm 2001, 2003, 2004). The salient feature of all these texts is a pervasive parallelism. Holm has translated and published these texts with valuable notes and commentary on their background and context. A short excerpt recounting the beginnings of creation taken from an early segment of the text (lines 46–63) for the buffalo sacrifice gives an indication of this parallelism:

In the beginning the world was suddenly dark and suddenly light

All at once it was suddenly heaven and suddenly earth.

No one yet knew of night nor evening

They knew not of short nor long,

They knew neither crosswise nor straight ahead…

The Emperor had not yet been created

Family Property had not yet been established

Not yet had everything under the sky been created

Not yet had everything on earth been created

Not yet had the moon and the sun been established

The great spirit in the sky looked down from above,

The Sky gods made a decision on high.

They made a Seal and issued it,

Then they sent down a certain King Pangu

The sky then opened into two halves,

The sky then transformed itself into two paths.

It formed a road for him to descend,

It formed a path for him to come on. (Holm 2003:108)

Holm notes that

while the texts are written artefacts, transmitted by copying, they show signs of oral composition. On the other hand, the fact that they were written down helped make them ritually powerful, given the pervasiveness of Buddhist and Taoist notions of the efficacy of texts and the enormous prestige of the Chinese script. (2003:36)

It is certainly evident from the texts themselves that they were meant to be recited. Holm describes the ritual masters (boumo) who take it in turn to recite these scriptural texts and includes a CV with recitations in his 2003 and 2004 publications. An illustrative example of the ritual speaking voice occurs in many places in the texts, as, for example, in the buffalo sacrifice:

‘Whittle a pair of chopsticks from nanmu wood

While I speak of the times before

Whittle an arrow of Tree of Heaven wood,

While I talk about times before,

While I speak of those fabled times,

While I tell you the story of the Emperor’s Inheritance.’ (Holm 2003:108)

Holm calls particular attention to linguistic borrowings that occur in the creation of synonymous pairings: ‘A particularly interesting feature of the parallelism is the way in which indigenous Zhuang concepts are frequently brought into parallel relation with terms borrowed from the Han’ (2003:36). This is the same phenomenon that Watson and other Biblical scholars discuss under the rubric of ‘rare and esoteric words’, especially when they cannot identify the origin of an esoteric word. The composition of pairs based on strategic borrowing across distinct dialects and different languages can be considered a fundamental feature of many—perhaps most—complex systems of semantic parallelism (see Fox 1974:80–1; Chapter 4, in this volume). It extends comprehension and communication beyond single speech communities especially in areas of linguistic diversity.

Since 1977, a variety of studies of the oral literatures of the highland populations of mainland South-East Asia has shed new light on less well-known traditions of parallelism. Frank Proschan (1984, 1989, 1992 has written on parallelism in the verbal arts of the Khmhu as has Håkan Lundström (see Lundström and Tayanin 2006; Lundström 2010), while Jacques Lemoine (1972) and Jean Mottin (1980) have noted the importance of parallelism in various poetic genre among the Hmong. (Kenneth White (1983) has provided an English translation of a Hmong Kr’ua Ke mortuary chant that Lemoine recorded.) Both Proschan and Mottin point to the use of parallelism in ‘love songs’—a speech genre identified by Granet in his Festivals and Songs of Ancient China (1919) as prone to the use of alternating parallel expressions. However, the use of parallelism in mortuary and ancestral propitiation is also wide spread.

The traditions of parallelism among the languages of Middle America

One of the most intellectually exuberant areas for the study of parallelism in the past several decades has been that of Middle America. The developments have coincided with the decipherment of the Mayan glyphs (see Coe 1992), and the addition of a new historical dimension has given impetus to the study of the diverse oral traditions of parallelism in the region. Reflecting on the decipherment of the glyphs, Floyd Lounsbury has signalled the importance of parallelism in the process:

And a native literary convention, the ‘poetic couplet,’ grounded in an oral tradition that is still very much alive today and that in ancient times was carried over even into the unwieldy medium of hieroglyphic inscriptions, also serves to bring out equivalences. Passages which are in this form present parallel strophes, ‘rhyming’ (so to speak) in meaning rather than sound, and conveying their message twice but in different phrasings. It was the recognition of this device that led to the discovery of the equivalence between some of the diverse forms of ‘birth glyphs’. (Lounsbury 1989:233)

Adding to this historical dimension for the study of parallelism has been a renewed attention to colonial manuscripts, many of which were written in parallelism, as well as considerable oral repertoire reflecting the Mayan past. In addition to his invaluable translation of the Popul Vuh (1971) based on a manuscript written between 1550 and 1555 by a Mayan noble of Quiche, Munro Edmonson has gone on to translate the Book of Chilam Balam (2008). Denis Tedlock has done his own translation of the Popul Vuh (1985;Revised: 1996) as well as a translation of the Rabinal Achi (2003), the remarkable Mayan dialogue and dance drama first recorded in 1856, but performed locally from the sixteenth century to the present. Similarly, Gary Gossen, whose work has focused on parallelism among the Tzotzil Maya, has edited and translated in parallel verse form a collection of some 74 separate Tzotzil tales, spanning four cycles of creation, destruction and restoration, to create a single vast ‘epic’ (2002). Gossen’s work is reminiscent of Elias Lönnrot’s nineteenth-century efforts in the creation of the Kalevala, compiled from various sources to become one of the longest continuous compositions in parallelism in world literature.

The opening verses of Edmonson’s translation of the Popul Vuh provide an illustration of Mayan parallel composition that often interpolates triplet semantic sets in the midst of a string of dyadic semantic sets:

This is the root of the former word

Here is Quiche by name

Here we shall write then,

We shall start out then,

The former wordsThe beginnings

And the taproots

Of everything done in Quiche town,

The tribe of the Quiche people.

So this is what we shall collect then,

The decipherment,

The clarification,

And the explanation

Of the mysteries

And the illumination

By the Former

And Shaper;

Bearer

And Engenderer are their names,

Hunter Possum

And Hunter Coyote

Great White Pig

And Coati,

Majesty

And Quetzal Serpent,

The Heart of the Lake

And Heart of the Sea,

Green Plate Spirit

And Blue Bowl Spirit.

(Edmonson 1971: 3)

One can compare this semantic parallelism with the parallelism in the final soliloquy by the warrior Cawek in Tedlock’s translation of the Rabinal Achi:

‘Alas, then, Sky!

Alas, then, Earth!

If I am truly dead

if I am lost

at the navel of the sky

navel of the earth

then I shall resemble

that squirrel

that bird

that died on the branch of a tree

in the flower of a tree

while searching

for his meals

his morsels

here at the navel of the sky

here at the navel of the earth.

You then, Eagle

you then, Jaguar

Come now!

Do your duty

Do your work.

Do it now with your teeth

your claws.

But you certainly won’t stand my hair on end

In the blink of an eye

because

I am truly brave

coming as I do

from my mountain

my valley.

May Sky and Earth be with you too

Little Eagle, little Jaguar.’ (Tedlock 2003:121–2)

In a paper on the semantics of Mayan parallelism, Edmonson examines the ‘associational chains’ among 42 pairs that occur in the first 94 lines of the Popul Vuh, arranging these sets on a continuum between the universal and the particular. Some, such as, for example, word//name, root//tree, heaven//earth, mother//father, he classifies as ‘universal’; others, such as plant//root, tribe//town, lake//sea, bowl//plate, he classifies as ‘widespread’; while others, such as white//book, white//tribe, tree//know, heart//breath, possum//coyote and pig//coati, he describes as ‘particular’. The universal categories, he views, as grounded in ‘general human experience’; the widespread categories are ‘contingent upon experiences common to many, but not all cultures’, while the particular categories are ‘found only in Middle America, some of them perhaps only among the Quiche Maya of the sixteenth century’ (1973:239).

For Edmonson, the ‘exotic associations’ equating ‘white’ with ‘book’ and ‘tribe’ are of special interest, revealing metaphors central to Quiche religion. ‘The particularistic semantics of the Popul Vuh suggests that details of expression in that work are significantly conditioned by ideas peculiar to the Quiche Maya and organized around their special view of cornfarming, hunting, priesthood and parenthood’ (1973:242).

Tedlock has taken a different approach to the Popul Vuh. Working with a native speaker of Quiche, Andrés Xiloj, who was also a ritual specialist, Tedlock offers a reading—or rather a ‘hearing’—of the Popul Vuh as if it were recited in a contemporary ceremonial setting. He is interested, importantly, in the ‘movement’ of the text/recitation: ‘the vertical movement of verse and the horizontal movement of prose’ (1987:151). Here he disagrees, as in his translation (1985), with Edmonson’s attempt to render the whole of the work in couplet verse.

Although the pervasive parallel semantics of the text is retained, its flow, which involves a semantic inventory of occupations, takes on a different form:

FulfilI your names—

Hunahpu Possum

Hunahpu Coyote

bearer twice over, begetter twice over,

great peccary, great tapir,

lapidary, jeweller, sawyer, carpenter,

maker of green plates,

maker of green bowls,

incense maker, master craftsman,

grandmother of day, grandmother of light.

(Tedlock 1987:153–4)

Tedlock also points to the importance (noted already by Hymes 1980) of the role of ‘framesetting’ terms that indicate the beginnings of a poetic phrase. He emphasises the importance not just of parallel couplets but also of parallel triplets, arguing that these triplets, which are also common in contemporary ritual discourse, ‘often occur at or near the beginning or end of a series of couplets’ (Tedlock 1987:159). He gives this example from the Popul Vuh:

Now it still ripples,

now it still murmurs

ripples,

it still sighs,

it still hums,

and it is empty

under the sky. (Tedlock 1987:158)

Whether these are all genuine semantic triplets and thus composed of three separate elements, or simply ‘performative’ triplets and thus single expressions of a recognised doublet, is an empirical question that probably requires more systemic analysis. Nevertheless, as in his other work, Tedlock’s performance-oriented perspective provides useful insights into the composition and use of parallelism.

One of the most important recent studies on Maya parallelism is the PhD thesis by Kerry Hull, ‘Verbal art and performance in Ch’orti’ and Maya hieroglyphic writing’ (2003). In this thesis, Hull provides an extensive documentation of Ch’orti’ Mayan ritual language, locating this speech among the oral narrative genres in Ch’orti’ and linking its poetic expressions and rhetorical devices to similar features in Mayan hieroglyphic writing.

More recently still, together with Michael Carrasco, Hull has put together and edited a remarkable collection of 17 papers on Mayan ethno-poetics entitled Parallel Worlds: Genre, discourse, and poetics in contemporary, colonial and classical Maya literature (2012). This is a volume that includes virtually all of the leading researchers in the field with an exceptional range of expertise on poetics and parallelism in historical sources and contemporary usage. As such, it deserves particular consideration.

In the introduction to his paper in Parallel Worlds, Hull expounds on the importance of parallelism and its pervasiveness in formal speaking among the Maya:

The preferred vehicle for literary expression among all modern Mayan languages is paralleled discourse. Quite simply, parallelism defines poetic or ornate discourse in the minds of the Maya themselves. Within its seemingly strict confines, the Maya are able to elaborate profoundly complex cultural knowledge by means of associative connections. Exploiting these relationships, be they complementary, contrastive or otherwise, Maya poetics operate at times outside the boundaries of the line, usually only showing their true literary qualities when in juxtaposition…in modern Maya languages, a general adage applies: the more form the discourse, the more parallel structures appear…It is undoubtedly on ritual and formal speech occasions where the full flowering of parallel structures can be found among the Maya…Both knowing which word combinations are acceptable and understanding the metaphorical extensions that often accompany such groupings are crucial elements in one’s communicative competence in many Maya societies. (2012:74–5)

In his paper, Hull examines a dozen of what he calls ‘diphrastic kennings’: the pairing of two distinct elements to produce another, more abstract concept. The term ‘diphrastic’ is borrowed from Ángel Maria Garibay, who was the first to use the term ‘difrassimos’, to describe such parallel terms in Nahuatl literature (1953:112). Among the list of these diphrastic kennings are pairs or dyadic sets that Edmonson might have classified as universal—sky//earth, male//female, stone//wood, green//yellow—and others that he would have regarded as more particular: bread//water, flint//shield and throne//mat. Hull traces each of these pairs from their use in Mayan hieroglyphs through their appearance in colonial Mayan writings to their continuing use among different present-day Mayan groups.

Hull’s intention is to demonstrate the ‘poetic tenacity’ of these pairings but his analysis shows much more. Thus, for the pair sky//earth, the linkage of these elements is retained in whatever language is used to expressed the concept: cham//kab in the inscriptions in the Palenque Palace, kaj//ulew in the Rabinal Achi, can//cabal among the Yuketekan Maya, or even borrowed Spanish terms, syelo//mundo, as used as among Ch’orti’ healers. There is persistence of a semantic category that utilises different lexical elements for its expression.

Parallel Worlds is a significant publication and indispensable for the study of Mayan parallelism. While a number of papers offer a dense analysis of the historical underpinnings and complex mythology expressed in the inscriptions of the classic period and of post-classic codices, other papers focus on the poetics of Mayan ritual language.

Michael Carrasco considers the ‘high poetic value’ of the ‘rhetoric of political propaganda’ incised on the Temple of Inscriptions at Palaque, whose construction dates from between 678 and 690. The parallelism of the inscriptions on behalf of K’inich Janaab Pakal, the Divine Palenque Lord, is stunning, even if the full significance of these assertions may be obscure:

His second stone-seating

[was] 12 Ajaw 8 Keh

the Eleventh K’atun

The face-of-the-sky Maize God had become lord

the face of the jewel tree had sprouted

the face of the ‘Five origins’ tree had sprouted.

The heavenly bundle of jade

the earthly bundle of jade;

the necklace,

the earspool...

(Carrasco: 2012:136)

Lloyd Anderson notes the importance of ‘coupleting’ to the decipherment and translation of Mayan inscriptions: ‘After coupleting was recognized as common in Mayan discourse, both spoken and written, this feature claimed enormous attention’ (2012:175). He also argues, however, for the importance of the syntactic placement of temporal markers and the arrangement of text as critical for a full understanding. Similarly, in his contribution to the volume, Tedlock looks at Mayan glyphs in inscriptions and vase painting as ‘graphic poetry’ whose arrangement is critical to the message conveyed.

Examining a Ch’orti’ manuscript of the colonial period, Danny Law looks at the various borrowings from Spanish that combine with an appropriation of native terms to incorporate new Christian concepts in mixed-language couplets. Thus, for example, in addressing the Christian God, the couplet ‘mother-father’ (na’//mi) is combined with the Spanish Dios and the Ch’orti’ term for ‘lord’ (ahaw):

natz’et ka-na’ ka-mi	You (sing) are our Mother, our Father
natz’et hun-te’tak Dios, noh-noh ahaw	You are the one God, big, big Lord. (Law 2012:280)

The creation of new concepts through the recombination of existing dyads is a hallmark of traditions of parallelism and is commonly noted in the comparative literature on parallelism. The ritual engagement with Christianity and assimilation of Christian ideas in parallel compositions among the Maya provide a particularly rich field for comparison with other traditions of parallelism that have had a similar religious engagement, as, for example, among the Rotenese (see Chapters 15, 16 and 17, in this volume).

Luis Enrique Sam Colop presents a superb and subtle reconsideration of the poetics of the Popul Vuh, giving his own interpretation of the compositional ordering of its lines with accompanying illustrative translation of its opening lines. The translation is arranged to highlight parallels in couplets, triplets and quatrains:

This is the root of the Ancient Word of this place called K’iche

Here we will write

we will implant the Ancient Word,

the beginnings

and the source of everything done in the town of K’iche

the nation of the K’iche people

Thus, we will take up the teaching

the clarification and

the account of what has been hidden

of what has been revealed

by the Maker

Modeler

Bearer

Begetter…

(Colop 2012:302)

Threading his way between Edmondson and Tedlock, Colop concludes with highly specific observations on the compositional structure of the Popul Vuh:

[V]erbal art in K’iche Mayan is organized in parallel lines. Parallelism, however, does not mean contiguity of two lines only…In couplets, the modifier is generally dropped in the second line, and in triplets it is dropped more frequently in the second of the three lines. In quatrains it is the last verse that generally breaks the paradigm to move the discourse into prose. Thus Maya texts integrate both verse and prose without one excluding the other. (2012:307)

Following Colop, Allen Christenson examines chiasmus as a poetic device used by the K’iche in the colonial period. Charles Hofling also looks at chiasmus in connection with the use of parallelism repetition in a comparison of the narrative styles of two different Mayan groups: the Mopan and the Itzaj. In a region where the possibilities of comparisons in the use of parallelism among different Mayan groups is considerable, Hofling’s paper is the only one to consider, if only briefly, similarities and differences in performative styles. In fact, with the exception of several references in the paper by Colop and in another paper by Auore Bequenlin and Alain Breton, there is little comparative focus beyond the region and little recognition or acknowledgment of similar linguistic processes, such as chiasmus, in other traditions of parallelism. The quality and depth of research on Maya parallelism are, however, encouraging. This is particularly the case where native Mayan-language speakers take up this research. Hilaria Cruz de Abeles’ linguistic examination of Chatino oratory (2009) is a well-nuanced example of this kind of research.

The study of parallelism in Nahuatl, Kuna, Quechua and Navajo

The study of parallelism in Middle America is by no means confined to Mayan research. Some of the earliest research on parallelism began with Garibay (1953), and this research continues to the present. In his presidential address to the annual meeting of the Linguistic Society of America, William Bright took the opportunity to examine the parallelism of a classical Nahuatl text dating from 1524. This text, Coloquios y doctrina Cristiana, edited with a translation, commentary and facsimile of the original manuscript by Miquel Leon-Portilla (1986), is a record preserved by Father Bernardino de Sahugun of a theological disputation between a group of Aztec priests and their spiritual counterparts, a dozen Franciscan friars. From the language used in this disquisition, this encounter was conducted in high formality.

Although the Aztecs acknowledged the death of their gods and the rightness of Christianity, they pleaded for the preservation of their ritual way of life. The text, which Bright quotes at length, reveals some remarkable aspects of semantic parallelism of this period. Within less than a decade of the overturn of their world in 1519, the Aztecs had already begun to fashion new forms of semantic parallelism to communicate their situation: either by using traditional pairs to designate new referents or by creating new dyads with constituent elements of previous dyadss. Thus, for example, Christ is referred to as ‘He who is Night//who is Wind’ (in youalli//in ehecatl); the Bible is ‘His book//His writing, the word of heaven (lit. ‘the heaven-word)//the divine word (the god-word)’. The Aztecs thus were able to converse in new theological terms using a formal language that had previously been reserved for their own sacred mysteries. This process of refashioning ritual languages to incorporate Christian concepts occurred at all levels of society throughout Middle America and was probably one of the reasons such parallelism has remained as a spoken heritage to the present.

In the Americas, the study of parallelism is not confined to areas of Nahuatl or Mayan languages. Joel Sherzer, whose earlier research highlighted the importance of parallelism among the Kuna (1972, 1974), has gone on to situate these forms of speaking in two notable studies: Kuna Ways of Speaking: An ethnographic perspective (1983) and Verbal Art in San Blas (1990). Bruce Mannheim, who has written on parallelism in various contexts (1986, 1987), has intriguingly examined the use of parallelism in a document, El primer nueva corónica i buen gobierno, written by Felipe Guaman Poma de Ayala in 1615 (1986). This document, in Quechua and Spanish, purports to describe the festival of the Inka in Cuzco at which the Inka sang to a tethered llama draped in a red mantle in thanksgiving for a successful harvest. As Mannheim explains, ‘“[s]emantic couplets” are a peculiarly Quechua poetic device in which two otherwise morphologically and syntactically identical lines are bound together by the alternation of two semantically related word stems’ (1986:52).

Of particular interest is Mannheim’s examination of various of these semantic couplets: ‘tears’ (weqe) pair with ‘raindrops’ (para), ‘woven shawl’ (lliklla) pairs with ‘skirt’ (aqsu), ‘to carry’ (apay) pairs with ‘to guide’ (pusay), ‘to look at’ (rikuy) pairs with ‘to watch’. Most interesting, however, is his discussion of the synonymous pairing of two terms for ‘water’, unu and yaku: unu, the word for ‘water’ in the Cuzco area, and yaku, the word for ‘water’ from the Ayachuco dialect of southern Peruvian Quechua. Mannheim’s discussion thus documents the use of dialect terms in the creation of a particular pairing—a phenomenon often alluded to in other traditions of parallelism but rarely so clearly examined.

Another area with a significant focus on parallelism is the study of Navajo poetry and ritual.

Building on the work of Gladys Reichard, particularly her monograph Prayer: The compulsive word (1944), various researchers (Witherspoon 1977; Gill 1980; Webster 2008) have examined the poetics of Navajo curing rituals. In this regard, Margaret Field and Taft Blackhorse have called attention to the role of ‘metonymy’ in Navajo prayer (2002). The most common feature of Navajo curing is the combination of repetition and parallelism: the reiterative use of a single compositional frame in which a succession of ordered semantic elements is inserted. Field and Blackhorse provide a simple example of this compositional sequence:

I will be healthy,

Wind will be beneath my feet,

Wind will be beneath my legs,

Wind will be beneath my body,

Wind will be beneath my mind,

Wind will be beneath my voice.

(Field and Blackhorse 2002:124)

This is a compositional feature that Navajo shares with other traditions of parallelism, as, for example, the Kuna. (An example of this sequential mode of composition among the Kuna was discussed in my 1977 paper; see Chapter 2, in this volume.) The important semantic aspect of this Navajo example of parallelism is the ordered relationship among enumerated elements of the person: feet, legs, body, mind, voice. In Field and Blackhorse’s terminology, this sequential relationship is designated as ‘metonymy’; for Gill (1980:41), these terms are part of a ‘catalogue’ that depends on the particular curing ceremony that is conducted; for Paul Friedrich, these terms form an inventory of contiguity-based tropes (1991). For the Navajo, there is a directionality to this ordering of terms that is conceptualised as ‘upward-moving’. Ritual requires a specifically ordered sequence as in the naming of the four directions—east, south, west and north—in a clockwise order.

The semantic relations among these ‘equivalent’ units are part of a yet wider semiotic process whereby a sequence of terms and their succession can follow a more extended path. This process involves a combination of parallelism and repetition. Sherzer, for example, writes of the ‘Kuna passion for listing people, within frames provided by literary formulas and parallel lines and verses, mapped on to a narrative’. He goes on to argue that the ‘use of lists in order to generate forms of discourse is so widespread in the world that it clearly constitutes a universal principle of oral discourse, of course as an instance of the most general universal principle, repetition’ (1990:48–9).

This mode of discourse can result in extended genealogical lists, but it can also frequently follow an ordered sequence of places—what I have described as as a ‘topogeny’ (see Fox 1997:8–17; and in this volume, Chapter 11). There are common features to all such parallel recitations: the use of a set frame—often requiring one substitution, or possibly two, per line—combined with an ordered secession of terms: personal names, placenames, special categories or colours, sequential actions, and a variety of fixed inventories. In all such sequences, attention is focused on the set relationship among the successive elements of the sequence—the significance of a particular cultural metonymy.

Finnish and Ural-Altaic studies in parallelism

In recognising similarities between Finnish and Hebrew oral composition, early Finnish scholarship can be credited with initiating the comparative study of parallelism. E. Cajanus’s Åbo dissertation, ‘Linguarum ebraeae et finnicae convenientia’, published in Lund in 1697, and Daniel Juslenius’s Oratorio de convenientia lingua Fennicae cum Hebraea et Graeca, published in Stockholm in 1728, both predate the work of Bishop Lowth, while Henricus Gabriel Porthan’s Dissertationis de Poesi Fennica, published in Åbo in 1766—the most explicit of these studies in its examination of Finnish parallelism—was carried out independently of Lowth.

Elias Lönnrot’s compilation of oral recitations—possibly from as many as a dozen different singers—took shape over several decades in the nineteenth century to become the present Kalevala. An initial version of these recitations, known as the Old Kalevala, was published in two volumes in 1835–36; the present expanded version, the standard version, appeared in 1849. This compilation became Finland’s national epic and perhaps the best-known narrative composition in parallelism. Although there have been many translations of the Kalevala into English, Francis Peabody Magoun’s translation (Lönnrot 1963), though in prose, provides an exceptional glimpse into one of the world’s literary classics. The singer’s voice in the opening lines of the Kalevala is reminiscent of the composer of the Popul Vuh, whose intention was also to reveal: It is my desire, it is my wish

to set out to sing, to begin to recite,

to let a song of our clan glide on, to sing a family lay.

The words are melting in my mouth, utterances dropping out,

coming to my tongue, being scattered about on my teeth.

(Magoun translation of Lönnrot 1963:3)

Wolfgang Steinitz’s Der Parallelismus in der finnisch-karelischen Volksdichtung (1934) set the basis for future research on Finnish-Karelian parallelism that has continued to the present. Based on research with one particular Karelian singer, Steinitz’s monograph is a model of linguistic analysis that gives particular attention to the semantics of canonical composition. Much of the current research on parallelism is published in Finnish and is therefore less accessible to a wider readership. A key survey document in English on recent research is the volume Song Beyond the Kalevala: Transformations of oral poetry, edited by Anna-Leena Siikala and Sinikka Vakimo (1994). This is a useful compilation of papers on a variety of issues in Kalevala scholarship. With the exception of Pertti Anttonen’s paper, ‘Ethnopoetic analysis and Finnish oral verse’, the scholarship in this volume, however, is less concerned with semantic parallelism and more with other distinctive features of the Finnic oral tradition, in particular, its metrical structure and its consistent use of alliteration. Frog and Eila Stepanova identify this combination of poetic features: ‘The strichic trochaic tetrameter characterized by alliteration and mutual equivalence in parallel lines, called “Kalevala metre” in Finnish and regivärrs in Estonian, is found among Finns, Karelians, Ingrians, Votes and Estonians’ (2011:204). Pentti Leino (1986) provides an excellent introduction to this metrical system; Janika Oras (2012) has added a further dimension to these studies by examining how musical patterning is interwoven with verse parallelism in the Estonian runo songs known as regilaul. Among various publications, given Berlin’s examination of negative parallelism in Hebrew, it is also pertinent to cite the work of Felix Oinas, ‘Negative parallelism in Karelian-Finnish folklore (1976). Oinas’ Studies in Finnish Folklore: Homage to the Kalevala (1985) is also concerned with Kalevala parallelism.

Steinitz extended his study of Finno-Ugric parallelism by further research in Russia on Khanty (or Ostyak) and published his results in a two-volume work, Ostjakische Volksdichtung und Erzählungen aus zwei Dialekten, which first appeared in 1939–41but has since been reprinted with a foreword by Roman Jakobson in 1975–76 and expanded to four volumes. Continuing in this line of research, Brigitte Schulze has written a dissertation on semantic parallelism in Khanty, ‘Der Wordparallelismus als Stilmittel der ostjakischen Volksdichtung’ (1982). To honour Steinitz on his eightieth birthday, Ewald Lang and Gert Sauer published a volume, Parallelsimus und Etymologie (1987). In this volume, Lang puts forward his argument for parallelism as a ‘universal principle of overall structural composition’ (1987).

Parallelism in Australian and New Guinea languages

There are various indications of the use of parallelism in Australian and New Guinea languages. The most important contribution to the study of parallelism in Australia is T. G. H. Strehlow’s masterly examination of Aranda songs, Songs of Central Australia (1971). As the son of a missionary, Strehlow was raised among the Aranda and had a native speaker’s knowledge of the language. He was also acquainted with the literature on Hebrew parallelism. His work among the Aranda was the first study to examine the varied use of parallelism in ritual compositions and it remains the most extensive analysis of parallelism in any Aboriginal language. Strehlow’s summary description makes clear the different levels at which this parallelism operates:

[I]n a native song words and word-weaving receive as much attention as the rhythms and tonal patterns which accompany them…the Aranda couplets (or quatrains) tend to consist of two individual lines which, musically and rhythmically, stand in a complementary relation to each other: the second line of a couplet is either identical in rhythm and construction with the first, or it balances the first line antithetically and rounds off the couplet by a contrasting rhythm of its own. The relation of parallelism and antithesis also characterizes the language of the songs. As a general rule each couplet, like a Hebrew psalm verse, falls into two halves: the second half either reiterates or restates, in slightly different words, a subject already expressed by the first half, or it introduces a new thought or statement, thereby advancing or completing the subject that has been expressed in the first half. (1971:109–10)

As an example of this ‘word-weaving’ parallelism, Strehlow cites the five-couplet lines of a song in which the ancestor Ankōţa gives voice to his performance in a sacred ceremony:

I am red like a burning fire;

I am covered with glowing red down.

I am red like a burning fire;

I am gleaming red, glistening with ochre.

I am red like a burning fire;

Red is the hollow in which I am lying.

I am red like the heart of a fire

Red is the hollow in which I am lying.

A tjurunga is standing upon my head

Red is the hollow in which I am lying. (1971:110)

In Strehlow’s view, this poetry is simple but effective.

Myfany Turpin, who has done more recent research on Aranda song, confirms Strehlow’s analysis of the structure of these songs. Following Linda Barwick (1989:18), she designates this basic dual prosodic structure of Aranda songs as consisting of ‘text-line pairs’ (TLP) composed of two nearly identical ‘text-lines’ for the insertion of lexical elements (2007:103). While concentrating on the melodic and rhythmic structure of these songs, Turpin defers to Strehlow’s work for analysis of their semantic complexity.

A number of researchers have taken up Paul Friedrich’s idea of ‘polytropy’ in their analysis of parallelism in Aboriginal song. In his analysis of Murriny Patha songs, Michael Walsh cites the use of parallelism (and repetition) as an instance of Friedrich’s ‘formal macrotrope’ (2010:126–7). Similarly, Peter Toner enumerates a range of compositional features in Yolngu songs—repetition, synonymic repetition, synonymic parallelism and formulaic parallelism—as different kinds of ‘formal tropes’ (2001:145–6). As an example of synonymic and formulaic parallelism, he provides the text of a song known as ‘jewfish’ (makani) that would appear to be a kind of topogeny—an ordered recitation of placenames. In this case, the places named are the beaches that were created by an ancestral jewfish as it swam past them. A few lines of this song provide an idea of its structure:

[J]ewfish/jewfish/jewfish/jewfish

gone/my/[beach] of Buburru

gone/my/[beach] of Gumbula

gone/my/[beach] of Nawurapu

getting closer/over there/[beach] of Aluwarra

getting closer/over there/[beach] of Amadhadhiltj

getting closer/over there/[beach] of Nyinybini

gone/my/[beach] of Aluwarra

gone/my/[beach] of Marrurru. (Toner 2001:145–7)

Elsewhere Toner discusses ‘the prolonged inventory of places named’ in other songs as an instance of Friedrich’s ‘contiguity tropes’ (2001:152).

Overall it would appear that though parallelism does occur in Australian Aboriginal compositions, particularly songs, such parallelism is not a pervasive or canonical feature of such compositions as is the case in other traditions of parallelism.

The use of parallelism is also reported in various New Guinea languages. Documented examples of parallelism are to be found in spells and in some song poems among the Foi (Weiner 1991), in the narrative genre known as Tom Yaya Kange among the Ku Wara in Highland New Guinea (Rumsey 2001, 2002, 2010) and in the poetic song traditions of the Ambonwari in East Sepik (Telban 2008).

In one of his papers on Ku Wara parallelism, Alan Rumsey gives examples of what he describes as ‘pairing compounds’: two nouns combined in sequence followed by a suffix that marks them as a pair. These pairs, as Rumsey notes, carry a meaning that encompasses more than their parts. Thus the pair ‘marsupial-bird’ connotes ‘hunted animals of the forest’; the pair ‘pig-dog’ can be used to refer to all ‘domestic animals’ but can also connote ‘people with uncontrolled appetites’; while the pair ‘sugarcane-banana’ carries a connotation of ‘snack food’ (2002:278–80). A particularly interesting pair is the combination ‘(Catholic) mass-(Lutheran)prayer/invocation’, which together is used to refer to Christianity. The semantics that Rumsey describes for such compounds is in fact a critical feature of innumerable dyadic sets in different traditions of canonical parallelism. Of further interest from a regional perspective is the occurrence of specific pairing compounds such as ‘yam-taro’ and ‘sugarcane-banana’, which are among the most widespread canonical pairs that occur in traditions of parallelism throughout much of eastern Indonesia. These pairs are generally used as botanic icons for persons (see Fox 1971:242–4; 1988:23; 1992; see also Nakagawa 1988:233).

Similarly, in his study of the Melpa of Mount Hagen, the German missionary ethnographer Hermann Strauss has emphasised the fundamental importance of pairing compounds—what he refers to as ‘a concept of complementation’: ‘Complementation to form a pair-unit is a very central feature of the religion and culture of the Mbowamb [Melpa]’ (1990:12). Certainly the texts that he has included in his ethnography give evidence of parallelism but not of an extensive or elaborated tradition as such.

In New Guinea, as in Australia, it would appear that the use of parallelism is definitely a feature of oral composition but this feature does not appear to have been developed to the same level of elaboration, as occurs, for example, throughout much of the Austronesian-speaking world.

The study of parallelism in Austronesian languages

The use of parallelism occurs across the whole of the Austronesian-language family. As a speech form, parallelism is, as elsewhere, a feature of elevated discourse and is invariably used as a vehicle for ritual communication and as a means of preserving sacred knowledge. The use of parallelism was a living tradition in many areas of the Austronesian-speaking world until recently and this traditional use continues particularly in the islands of eastern Indonesia.

Nowhere, however, does there exist the historical depth for the study of parallelism that exists in Middle America. The oldest existing texts in parallelism date from the middle of the nineteenth century. The Bible translator A. Hardeland, in his Versuch einer Grammatik der Dajackschen Sprache (1858), was the first to note the Hebraic parallels in Dayak ‘spirit language’ (Basa Sangiang), but he provided only a single illustration of this ‘spirit language’ with translation and commentary; however, the posthumous publication of the Swiss missionary H. Schärer’s two-volume compilation of texts in Basa Sangiang, Der Totenkult der Ngadju Dajak in Süd-Borneo (1966), advances Hardeland’s initial work by providing a substantial corpus for study. This corpus on the Ngaju has been further extended by the substantial work of Sri Kuhnt-Saptodewo (1993, 1999).

Other large compilations of ritual-language texts are those assembled by D. Dunselman (1949, 1950, 1954, 1955, 1959, 1961) from the Kendayan and Mualang Dayak of West Borneo; by H. Lagemann (1893, 1906) and W. L. Steinhart (1934, 1937, 1938, 1950, 1954) from Nias; by W. Dunnebier (1938, 1953) from Bolaang Mongondow; by H. van der Veen (1929, 1950, 1965, 1966) and by Dana Rappoport (2013) from the Sa’dan Toraja; and by P. Middelkoop (1949) from the Atoni Pah Meto of West Timor and by A. Quack (1981, 1985), D. Schröder and A. Quack (1979), with a more recent collection of ritual texts by Josiane Cauquelin (2008), for the Puyuma of Taiwan. Another recent study of a priestly language—that of the transvestite bissu priests of the Bugis of South Sulawesi—is Gilbert Hamonic’s Le Langage des Dieux (1987), which includes a considerable number of prayers in parallelism. Most of these collections of texts, with accompanying translations and commentary, were produced by their various authors primarily for the purposes of cultural exegesis.

H. van der Veen’s monograph The Merok Feast of the Sa’dan-Toradja (1965) provides a good illustration. The main Merok feast chant, which takes the form of a journey, consists of more than 790 double-lines filled with paired metaphoric images in an array of successive formulaic frames. It has many of the features that one encounters in other traditions of parallelism: admonitions, ritual directives, inventories and topogenies. An example of this is the inventory for the spiritual journey that it sets forth. This semantic inventory is repeatedly framed in the double-lines: ‘As sustenance for the journey//as provisions on the way’ (ammi pokinallo ilalan//ammi pokokong dilambanan):

As sustenance for the journey, take the three-eared rice

As provisions on the way, take the cut one, branched in three…

As sustenance for the journey, take the gold kris of great size

As provisions on the way, take the piece of beadwork, with cords hanging low…

As sustenance for the journey, take an auspicious dream

As provisions on the way, take a pregnant nocturnal vision. (van der Veen 1965:37–9)

After the long recitation of these provisions, the chant sets forth the path that should be followed, using another simple dyadic frame that combines ‘path’ and ‘way’ (lalan//lambanan):

Then shalt thou take the rainbow as thy path

Thou shalt make thy way along the arch of the sky. (van der Veen 1965:39–40)

The longest section of this chant, which runs for more than 140 double-lines, is a topogeny that invokes and then invites the gods from all directions and from all of the Toraja territory and beyond, named and ordered in dyadic form:

The God of the Leatung region,

the Lord of the district of Mangkaranga

The God of the Patua’ region

the Lord of the district of Mila’.

(van der Veen 1965:42–61)

The recitation of this invitation to the gods among the Toraja—a requirement in many ritual performances—is similar in function and form to the extended Zhuang invitation to their gods and spirits recorded by Holm (2003:60–9).

Lexicons of ritual language

Hardeland’s concern with parallelism was primarily philological. He thus attempted to distinguish different classes of vocabulary (1858:4–5): 1) slightly altered Dayak words; 2) Malay words, also slightly altered; and 3) special words, whose meaning and form were confined exclusively to use in the ‘spirit language’. His dictionary (1859) includes some 900 of these Basa Sangiang words. M. Baier ( see Baier et al 1987) continued this work of more than a century by publishing a dictionary of Basa Sangiang based on materials from both Hardeland and Schärer.

Following Hardeland, B. F. Matthes (1872) noted the use of similar sorts of lexical items in the parallel language of the transvestite priests of the Bugis, as did the missionary linguist N. Adriani, who wrote a brief comparative paper in Dutch on ‘Indonesian priest languages’ (1920, reprinted 1932:III, pp. 1–21) and another similar paper in German on ‘magic speech’ (1926, reprinted 1932:III, pp. 167–75).

The concern with the constituent origin of the lexicon of these ritual languages has continued to interest researchers to the present. In his study of Dusun ‘sacred language’, Evans (1953:495–6) distinguished five classes of words that made up this register: 1) ‘ordinary Dusun words’; 2) ‘special but easily recognizable forms of ordinary words—poetic forms—derived from ordinary Dusun’; 3) ‘words not usually current in the village…but found in other villages, near or far away’; 4) ‘loan words from Malay’; and 5) ‘words used, as far as known, only in the “sacred language”, for which derivations are not obtainable’.

Writing of the Timugon Murut priestesses (babalian) whose ritual curing performances require a strict pairing of all ritual utterances, D. J. Prentice gives particular attention to the origin of the lexicon of these utterances. In such performances, which can last for days, the principal priestess utters the first line to which her acolyte, who accompanies her, responds with the second. Prentice describes the first line of each couplet as ‘couched in a style of language which has much in common with the poetic style normally used in non-ritual singing’, whereas the ‘response’ (taam) is ‘couched in a special ritual language’ (1981:130–1). Prentice identifies various sources for such response words: 1) ‘a small number of cases where the ritual substitute is a synonym that already exists in the normal language…numerals may also be reckoned as belonging to this category since a numeral “X” is replaced in the taam by the numeral “X + 1”’; 2) ‘more frequently, the taam-word exists in normal Timugon, but with a different (sometimes even contradictory) meaning’; 3) ‘taam-words which exist in normal language, but in a different form (i.e. a derivative rather than as a base, or vice versa)’; 4) ‘a very large number of ritual terms…known only from cognates in some other Murut dialect or in another language’; and, finally, as in the case with Evans’ analysis of Dusun, 5) ‘a number of substitute words for which no connections are known, either in Timugon or in any other dialect or language’ (1981:133–5).

The rules for the pairing of terms combined with the required use of special terms in the make-up of many pairs are evidence that these ritual languages are not simply the elevated pairing of ordinary speech but rather are special registers that invariably require a considerable learning process.

Parallelism in Austronesian speech genres

At the same time that in the nineteenth century some missionary linguists were focusing attention on priestly registers among the Austronesians, other missionaries had begun to recognise the more general nature and widespread use of parallel compositions. Thus, for example, in 1880, writing of Malagasy, the British missionary James Sibree noted: ‘It is most pleasant therefore to listen to a native orator, especially as in the more formal Malagasy speeches the parts of every sentence are regularly balanced in construction, form a kind of rhythm very closely resembling the parallelism of Hebrew poetry’ (1880:148).

As a special register for oratory, prayer, poetry or song, such formal socially elevated composition is not necessarily confined to a specific class of practitioners and is therefore more likely to be used in a variety of settings. Whereas in the case of ‘priestly registers’ identification is generally couched in relation to particular performers or religious performances, oratorical registers (based on similar modes of composition but often with a less obscure lexicon) are generally discussed in terms of different speech genres or literary productions. Often, however, oral poets in Austronesian societies are regarded as ‘spirit-enlightened’ figures whose roles are similar to, and often overlap with, that of priests, priestesses or shamans. Based on his research among the Berawan of Sarawak, Peter Metcalf has published a collection of prayers in parallelism by which, for example, individuals instruct sacrificial animals to convey their words to specific deities (1989, 1994). In a different context, Margaret Florey has studied the preservation of a local genre of ‘incantations’ among the Alune of Seram in Central Maluku, who are undergoing rapid social and linguistic changes (Florey 1998; Florey and Wolff 1998).

The literature on these various poetic registers is diverse. Thus, for example, L. Sabatier has published the text of a major ‘epic’ in pervasive parallelism for the Rhade (1933); in his commentary on the Salasilah of Koetai (1956), W. Kern has commented on the comparative significance of parallelism in Malay penglipur-lara tales, Middle Malay anday-anday and Minangkabau kaba (1956); and Nigel Phillips has produced an exemplary study of Sijobang, the sung narrative poetry of West Sumatra (1981). For the Philippines, H. J. Wigglesworth has emphasised the role of parallelism as a rhetorical device in Manobo narrative discourse (1980); as has Nicole Revel among the Palawan (2013). J. P. B. de Josselin de Jong has written a critical study of eastern Indonesian poetry (1941). Timo Kaartinen has examined the parallelism of a song tradition among the displaced ‘Bandanese’ in the Kei Islands (1998) and Aone van Engelenhoven has provided a preliminary study of the parallelism of ‘royal speech’ and ‘song speech’ among the people of Leti in the southern Moluccas (1997). His article includes the translation of a parallel text first recorded in 1846. Similarly, Gillian Sankoff has analysed parallelism in poetry of the Buang, an Austronesian-speaking population of Papua New Guinea (1977). The Kumulipo, the long Hawaiian narrative chant translated by Martha Beckwith (1951), provides an excellent example of similar forms of extended literary parallelism. While most of these compositions are in strict canonical parallelism, others show parallelism in varying degrees. The Iban, for example, whose ritual traditions have been particularly well documented by Benedict Sandin (1977), James Masing (1997) and Clifford Sather (2001), use parallelism to a lesser extent than other Borneo populations such as the Kendayan or Berawan.

The prominence of parallelism in societies on the islands of Sumba, Flores and Timor has meant that most ethnographic accounts of these populations discuss the use of parallelism in social and ritual contexts. This is particularly true of the island of Sumba with its closely related yet distinct groups: Danielle Geirnaert-Martin (1992), for example, examines the use of parallelism in the Paddu rituals of Laboya; Janet Hoskins (1993) the parallel language of the calendric rituals of Kodi; Webb Keane (1997) considers the whole rhetoric of performance and representation among the Anakalang; while Joel Kuipers has written two important studies (1990, 1998) on Weyewa ritual speech. Although the Weyewa tradition appears to lack an extended narrative genre like some found on neighbouring islands, the use of ritual speech extends to a variety of what Kuipers refers to as political, religious and personal genres of speech. Kuipers (1998:37) identifies no less than 19 such genres. He also addresses the question of ongoing changes in the use of ritual speech styles, noting the marginalisation of an aggressive performance genre expressing the anger of politically powerful figures in favour of nostalgic laments and school-sanctioned children’s performances (1998). Adding to this array of studies on parallelism in Sumba, Elvira Rothe, in an unpublished dissertation for Ludwig Maximilian University in Munich (2004), has produced a remarkably detailed linguistic examination of the Poddu rituals of Loli.

The same is true for Flores and Timor. The ethnographies by Philipus Tule for the Keo of central Flores (2004) and by Karl-Heinz Kohl for the Belogili of eastern Flores rely on compositions in parallel language in their expositions. Tule’s ethnography, which focuses on a mixed Catholic–Muslim population, includes prayers in parallelism both for the ordination of a priest and for the safe journey of someone going on the Haj. Kohl’s ethnography records the rituals of the rice cycle. Three as yet unpublished PhD theses—the first by Michael Vischer (1992) focusing on the island of Palué, the second by Penelope Graham (1991) on the community of Lewotala in eastern Flores, near the Belogili, and the third by David Butterworth (2008) on the Krowe community in central eastern Flores—provide further detailed ethnographic evidence of the use of parallelism in Flores.For Timor, the use of parallelism with critical attention to the cultural significance of key semantic pairs forms the basis of the ethnographies by Andrew McWilliam on the Atoni Pah Meto (2002), by G. Tom Therik on the southern Tetun at their ritual centre in Wehali (2004), and by Elizabeth Traube on the Mambai (1986), while Gregor Neonbasu’s ethnography (2011) is organised specifically around the oral traditions of the Atoni of Biboki. Herbert Jardner’s monograph Die Kuan-Fatu-Chronik (1999) records and examines parallelism in the narratives of the Atoni of Amanuban. The unpublished PhD thesis by Benjamin de Araujo Corte-Real (1999) on the verbal arts of the Mambai offers a further important contribution to the study of parallelism in Timor.

Of particular comparative significance is the publication of a substantial collection of Bunaq texts collected by Louis Berthe (1972). The Bunaq are a non–Austronesian-language group situated in—indeed surrounded by—Austronesian-speaking populations. Historically, the Bunaq have been drawn into the ritual spheres of neighbouring Kemak and Tetun and their parallelism reflects this historical involvement. In their sacred texts, numerous Bunaq canonical pairs combine Bunaq and Tetun/Kemak terms, thus bridging language families by coupling lexical elements of Austronesian and non-Austronesian. Engelenhoven (2010) has documented the use of parallel discourse among the Fatuluku, another non-Austronesian language group at the eastern end of Timor-Leste, comparing its specific pairs with those of other neighbouring Austronesian-languages groups within the region.

An ethnographically focused comparison of the ritual languages of eastern Indonesia can be found in the volume edited by Fox, To Speak in Pairs: Essays on the ritual languages of eastern Indonesia (1988). This volume consists of 10 studies of different ritual languages on Sumba, Flores, Rote and Sulawesi. The collection as a whole provides an indication of the diversity of social contexts in which these registers continue to be used in societies of eastern Indonesia: for divination and spirit communication, prayers and sacrificial invocations, bride-wealth negotiations, ordeals, mortuary ceremonies, complex origin-cycle rituals, and a variety of other critical oratorical contexts. Several authors focus on the ritual-language traditions of different groups in west Sumba: Janet Hoskins examines the formal etiquette of communication with spirits in Kodi; David Mitchell explicates the metaphoric couplets in oratory among Wanukaka speakers; Brigitte Renard-Clamagirand focuses on the speech used in a ceremony for ‘banishing transgression’; and Joel Kuipers examines the pattern of prayer among the Weyewa. Gregory Forth contrasts the invocatory and oratorical speeches in Rindi in eastern Sumba. For Flores, Eriko Aoki provides a detailed examination of a sequence of divination and ordeal to discover stolen objects among Lio speakers; Satoshi Nakagawa considers the formal idioms of marriage among Endeh speakers; while Douglas Lewis examines the performance of a narrative history in Tana Ai. Fox examines the formulaic structure of a long mortuary chant on Rote and Charles Zerner and Toby Volkman consider the language of one ritual in a complex cycle of Toraja ceremonies.

The prominence of the ritual voice

While each of the papers in To Speak in Pairs examines a distinct tradition of parallelism in a different speech community, there is an underlying commonality to all of these traditions in eastern Indonesia. In addition to the broad similarity of many shared semantic pairs, there occurs in ritual speech throughout the region what might be called a distinct ‘ritual voice’. The chanter, poet or speaker takes on and merges with the persona or personae of the performance and personally directs the ritual message. Seve ral examples may illustrate the variety of modes of this ritual voice.

The words of Mo’an Robertus Rapa of clan Ipir in Wai Brama, as quoted by Douglas Lewis (1988:271), give an indication of one form of ritual voice—the proud ritual voice:

I am clan Ipir, the great ebony tree

I am exalted like the large birds of the mountains

The domain of Wai Brama and Wolobola

Raja as far as Balénatar

Speaking to the sky and earth

Addressing the sun and moon.

In contrast with this ritual voice, there is the ritual voice—the humble ritual voice—in the ‘deference’ mode of speaking described by Janet Hoskins (1988:54–5) among the Kodi:

Only we are left, spiders with no livers

Only we are left, snails with no thoughts…

There is no one here to tell us the many long myths

There is no one here to teach us more of our customs

I am alone like a child holding the net for a discus toy

I am alone like a child grasping the rope on a spinning top.

This is the voice that even the most respected Kodi chanters adopt in pleading with the ancestral ‘spirits of the inside’ to whom they are indeed children.

Among the Atoni of Timor, knowledgeable ritual speakers take on the persona of their clan embodied in the clan name: the kanaf. The history of the clan can thus be recited as a journey of the name—a dense personal topogeny that moves through a specific landscape. The narrative of the difficult ‘path’ of the Nabuasa clan, as recounted by Andrew McWilliam (2007:107; see 2002:80), gives an idea of this ritual voice:

As a great man,

one who kicks the land and

one who kicks the water

travelling past

the slippery bamboo and

the swollen stream

Amsam Noetasi

Monam Saibet.

Not staying long

not remaining [but]

turning back

returning from

The rising moon,

The rising day.

A ritual voice can also be so specifically engaged in the details of a ceremony that it is almost incomprehensible without careful exegesis of the speaker’s intention and the context of his speaking. Joel Kuipers has examined a sequence of ceremonies among the Weyewa (1990:115–16)—a placation ritual that followed an earlier divination ritual by the speaker and specialist Mbani Mata. He was acting on behalf of someone, Malo Dunga, who, it was determined, had failed to carry out the promises of his ancestors, Mbili and Koni. Excerpts from this ritual that make reference to Malo Dunga as ‘cucumber flower’//‘gourd fruit’ and the problem that Mbani Mata has uncovered by ‘cutting the tree to its base’//‘searching for water to its source’ provide a glimpse of the complex involvement of a ritual voice:

You cucumber flower offspring there

you gourd fruit descendant there

he arrived at shadow of the house eaves

he arrived at the porch

and if I cut the tree

to its base

and if I search the water

to its source

don’t hide the source of red ants

don’t hide the fish in a bamboo jar…

I say to you right now…

to help us listen to Mother’s voice

to help us listen to Father’s voice…

do not deviate from the path

do not raise your sword at me.

As Kuipers explains, a speaker may change names—and thus personae—many times in the course of a single ritual.

The ritual voice may be revelatory but it may also be intentionally obscure—a voice that struggles to express ancient mysteries that can not be fully comprehended. Thus, among the Mambai of Timor-Leste, the Ritual Lord who combines the conjoined attributes of male and female is conceived of as a new born infant. As Elizabeth Traube explains, this representation figures prominently in ritual oratory: ‘Drawing on images of birth, performers liken themselves to babies who have just emerged from womb:

My brow is soft

My navel is bloody

There is oil over my brow

There is blood over my eyes.

My ears are still full

My head is still whole,

I show with the mouth

I reveal with the eyes.

(Traube 1986: 105)

In these and other contexts, the ritual voice is that of the speaker transformed, no longer a mere individual, but rather a collective instrument of formal annunciation and communication whose formal language is culturally established, semantically ordered and recognisably meaningful.

Comments on the comparative use of parallelism

This survey of directions and diversions in the comparative study of parallelism serves several purposes. It demonstrates the prevalence and pertinence of parallelism as a linguistic phenomenon. It also indicates its considerable persistence as an oral tradition as well as its continuance as a literary tradition. It is also notable that parallelism emerged in ritual compositions at the beginnings of so many traditions of early writing—in Sumer, in Egypt, in China and also among the Maya. The library unearthed at Ras Shamra contained cuneiform texts that linked the Ugaritic tradition of parallelism to the Akkadian and ultimately Sumerian traditions. Research has similarly shown the persistence of Mayan traditions of parallelism—even the persistence of fixed pairs—from the Palenque inscriptions to present-day Mayan ritual performances. Chinese, it could be argued, has also maintained a continuous tradition of parallel composition.

Although most of the research on parallelism has been carried out largely independently on different language traditions, it is evident that there are similarities in many of the features that define parallelism as a linguistic phenomenon. There are also striking gradations in the use of parallelism in composition. These gradations are evident between different traditions as well as within particular traditions. In their simplest forms, parallel compositions may consist of a repeated formulaic frame with a succession of single semantic substitutions. In other traditions, linked parallel lines may consist of three, or possibly even four, semantic pairs. The most elaborate of these traditions are generally based on culturally specific requirements for (or at least the expectation of) the canonical pairing of semantic elements.

Parallelism is invariably an elevated speech register; in many traditions, the greater the emphasis on parallelism and the set pairing of words and phrases, the more elevated is the level of speaking and the greater is the significance of the message this speaking is intended to convey. Parallelism is thus a device of ritual and high oratory but also a device for the preservation of sacred knowledge.

Parallelism can occur at many levels and there is a great variety of modes of composition both oral and literary. The semantics of parallelism is, however, paramount and offers the widest scope for research. There is nothing either simple or mechanistic about this semantics. The pairing of terms is more than the sum of its individual elements. In Jakobson’s framework, the metaphoric axis of similarity is projected onto the metonymic axis of continuity. Although this describes the formal process of pairing, the process of pairing creates new meanings.

The link between two words produces a metaphoric connection that may have multiple meanings and interpretations. In this semantic relationship, the complementarity of elements at the literal level enunciates a further interpretative level. As Hull notes, ‘older brother’//‘younger brother’ can refer more widely to ‘all men’; ‘mother’//‘father’ to a complex set of relations of nurture or superiority. Among the Rotenese, when ‘elder brother’//‘younger brother’ are paired, their referent is specifically to ‘(male) members of the same clan’; when ‘mother’//‘father’ are paired, the sense, as among the Maya, is that of nurture and superiority, but the referent is often specifically to a ruler or government official.

Or, for example, where in this translation of the Popul Vuh, Edmonson gives a literal rendering to the couplet ‘green plate’//‘blue bowl’, Sam Colop in his translation interprets this pair in its fuller metaphoric sense as ‘green earth’//‘blue sky’. Among the Weyewa, ‘cucumber flower’//‘gourd fruit’ is a botanic metaphor for a ‘descendant’ or ‘offspring’, and as Joel Kuipers explains, the recurrent pair ‘horse’//‘dog’ or the even more expressive parallel phrase ‘horse with a standing tail’//‘dog with a black tongue’, while striking in itself, is in fact intended as a reference to the orator or narrator of a composition, allowing him to insert his presence within his composition.

Underlying all such pairings is the notion of the ‘canonical’: ‘fixed word pairs’, ‘conventional couplets’, ‘diaphrastic kenning’, ‘dyadic sets’—however they are identified, the interpretable semantic stock of pairs recognised by a particular speech community. The range of these pairs and their relationship to one another are fundamental to the understanding of any tradition of parallelism. To date, Dahood’s ‘Ras Shamra Parallels’—a listing with commentary of shared Ugaritic and Hebrew pairs—is the most extensive compilation of such canonical elements. My own, as yet unpublished, dictionary of Rotenese dyadic language has at present more than 1750 such pairs and, as it continues to expand, may well run to more than 2500 pairs.

Another recurrent observation made of traditions of complex canonical parallelism is their utilisation of what in the literature on Hebrew parallelism are referred to as ‘rare and esoteric words’. The sharing of pairs between Ugaritic and Hebrew is an aspect of this linguistic phenomenon. It indicates the historical inheritance and preservation of a common semantic tradition but it is also an indication of a readiness to borrow from other linguistic sources. In the Middle East, the shared inheritance of semantic sets evident in Hebrew may well extend to the earlier Akkadian/Sumerian traditions of parallels whereas a similar sharing of semantic couplets in the Mayan region extends from the earliest inscriptions to the present. The borrowing of new terms from other languages and the recombination of existing terms to create new semantic concepts are equally fundamental to the continuing development of traditions of parallelism. Together, all of this is evidence of the development of traditions of elaborate semantic parallelism in areas of linguistic diversity and indeed the exploitation of linguistic diversity to create an increasingly rich repertoire of pairings.

Seen from the perspective of a single speech community, the tendency has been to consider this use of ‘external’ linguistic resources as an impediment to normal intelligibility—the creation of specialised, exotic, priestly registers. In situations of multiple language use, however, and of dialect diversity, parallel languages can actually promote communication—especially ritual communication—across linguistic boundaries. My own research on the island of Rote with its chain of dialects gives evidence of this ‘inter-dialect’ aspect of its ritual language. To appreciate this ‘dialect concatenation’, it is essential to examine the function of a tradition of parallelism from beyond the confines of a single speech community and instead to consider it from the perspective of the multiple speech communities that have a share in it.

In the literature on parallelism, considerable attention is given to varied forms of composition: the arrangement of lines, couplet, triplet and quatrain-structures, chiasmus, negative pairs, internal pairing or the ordering of inventories. This important focus on the varieties of dual structures that occur at the surface level of composition does not, however, give attention to the possible semantic networking of the terms that compose these dyadic structures. In traditions of canonical parallelism, many of the lexical terms that form particular pairs also occur in other pairings. There is thus more to the study of parallelism than the identification and listing of specific pairs or couplets.

Key terms may link with other terms and can, in this way, be embedded within complex semantic networks. These networks are themselves of interest in understanding the ideas of a particular culture. Thus, for example, in Chapter 5 in this volume, I examine the semantic network formed by verbs for ‘speaking’ in Rotenese, and in Chapter 6, I consider the network of the key terms—or primary symbols—in Rotenese ritual language defined as those terms that have the greatest linkages among themselves and with other terms in ritual language.

In my ‘Introduction’ to the volume To Speak in Pairs, I also compared the semantic ‘embeddedness’ of terms for ‘earth, land, ground, country’ in Rindi, a dialect of east Sumbanese, with those for the same term in Rotenese (1988:25–6). In Rindi, this term is tana and in Rotenese, it is dae. In his paper for the volume (1988:129–60), Gregory Forth listed the various terms that paired with tana, and I did the same for the term for dae in Rotenese.

Rindi	Gloss	Rotenese	Gloss
tana//ai	tree, wood	dae//ai	tree, wood
//wai	water	//oe	water
//watu	stone	//batu	stone
//awangu	sky	//lai(n)	sky
//rumba	grass	//tua	lontar palm
//luku	river	//dale	inside
//pindu	gate	//de’a	outside
//paraingu	domain	//dulu	east

This simple illustration indicates that Rindi and Rote share a common set of ritual associations that link ‘earth’, ‘water’, ‘stone’ and ‘tree’. For both there also exists a contrastive association between ‘earth’ and ‘sky’. For the east Sumbanese, other associations of ‘earth’ or ‘land’ are with ‘grass’ and ‘river’ whereas among the Rotenese there is an association of ‘earth’ with the ‘lontar palm’, a primary source of livelihood. In Rotenese, the term dae can function as a directional indicator meaning ‘below’, and through this sense, it links to other directional terms such as ‘inside’, ‘outside’ and ‘east’. In contrast, in Rindi, the further association of tana is with ‘gate’ and ‘domain’. Although this illustration provides only an initial set of linkages for these terms, it shows how formal comparison might be made between different systems of meaning. Both terms form part of yet larger semantic networks and these networks can be examined either in terms of the ‘embedding’ of particular elements or in terms of their larger overall structures. Chapter 6 in this volume provides another perspective on the embedding of dae as a primary ritual symbol. Using the evidence of the linkages among parallel terms, it is certainly possible to imagine comparisons among the different cultural traditions of eastern Indonesia but it is also possible to conceive of comparisons, for example, between eastern Indonesia and the Mayan region—both areas of still vital traditions of parallelism.

This survey makes clear that parallelism is no simple phenomenon. In its many manifestations, it varies considerably in its semantic complexity and in the varieties of its performative expression. What is most evident from this survey is that parallelism is not always immediately evident. What constitute culturally defined notions of equivalence, opposition or complementary—in Robert Lowth’s phrase, the ‘correspondences…answering to one another’—can only be interpreted from within particular traditions. Criteria for categories of gendered opposition, positive-negative pairings, rules of tonal complementarity and of alliteration, metrical congruence and the reversals defined by chiasmus define the nature of particular traditions of parallelism.

The semantic complexity of such traditions of parallelism range from a successive iteration of particular terms, often comprising a set repertoire of related items, through a plethora of culturally specific registers that use a variety of semantic pairings to yet more elaborate systems of canonical parallelism hedged with ritual injunctions that demand adherence.

A great deal of the literature on parallelism documents traditions that have ceased to exist as oral traditions—that are now literary objects of analysis that may, at best, hint at earlier stages of their composition. However, in parts of the world, particularly in Middle America and Southeast Asia, there still exist living oral traditions of parallel composition—some of considerable complexity and vitality. Although these traditions are under threat, they remain a special area for continuing research.

This survey of recent research on parallelism has the purpose of providing a prelude to the study on a single, still active performative tradition of canonical parallelism—that of the Rotenese of eastern Indonesia.