Words and worlds

This work is licensed under the Creative Commons | © John Leavitt. ISSN 2049-1115 (Online). DOI: http://dx.doi.org/10.14318/hau4.2.009


Words and worlds

Ethnography and theories of translation

John LEAVITT, Université de Montréal

If different languages orient the speaker toward different aspects of experience, then translation can be seen as a passage between lived worlds. This paper traces out some key moments in the history of translation theory in the modern West and argues that translation and ethnography require each other. Free of the constraint that professional translators produce easily digestible texts for the target audience, anthropologists are particularly well placed to carry out translations that take context seriously into account, as well as ethnographies centered on texts. Such “ugly” translations (Ortega y Gasset) can force the reader to work to reorient him- or herself, to cross a boundary into what is potentially another world, initially another language-world. Through the history, we seek to distinguish the translation of referential content, something that is always possible, and translating stylistic and indexical (contextual) elements, something that has often been declared impossible. The paper draws some of the implications of these arguments on the basis of text-artifacts constructed from Central Himalayan oral traditions.

Keywords: translation, ethnography, history, poetics, Central Himalayas


Of the human sciences, only anthropology has consistently taken seriously the idea that beside or beyond the evident fact that all human beings live in the same world, there is a sense in which they live or have lived in different worlds. Since Boas, these have been called “cultures,”1 and have recently returned as “ontologies.” As it happens, the cultures that seem to be most different from the ones that have produced most anthropologists are those that have been colonized and transformed. Are we to think of an Australian or African tribal group as living in a different world, or rather as peripheral lumpen proletarians? How important is what is distinctive? As Bridget O’Laughlin wrote (1978: 103), what is distinctive about the zebra is its stripes, but stripes are hardly the most important aspect of a zebra’s life.

Yet what is most important to people themselves depends in part on their historical situation and their expectations; these need not always be the same everywhere, at every time. Existing in a situation of starvation and disease is not the way any human being, or any organism, wishes to live. But once you can eat and feed and protect your children, then other factors come into play. These factors may be conscious or close to consciousness; or, as in the case of the language one speaks, they may be largely unconscious. And the record shows that these world-making factors can be extraordinarily diverse.

This means that human values, goals, and conceptualizations are themselves diverse, and maybe even incommensurable (Povinelli 2001): insofar as conceptions differ, lived worlds may be said to differ. Probably the most blatant diversity is that of human languages, this very diversity being one of the important language universals. This is a simple fact: the species speaks or has spoken thousands of different mutually unintelligible languages. How seriously should this diversity be taken? Some say not at all: according to Noam Chomsky, if a Martian arrived on Earth and observed how children learn language, he (in these parables it is always a male Martian) would say that there is only one human language with, in Steven Pinker’s oxymoron, mutually unintelligible dialects (Pinker 1994: 240). On the other side, both George Steiner (1975) and Claude Hagege (1985) use the same parable to opposite effect: both say that if a Martian arrived on earth and observed human physical variation, he would assume that there were maybe a dozen human languages—instead of the five to ten thousand we actually find (for discussion of the Martian parable paradox, see Leavitt 2011: 5–6). Diversity goes all the way up and down the levels of language structure, from sound to the dividing of experience in lexicon, to the directing of attention to one or another aspect of experience by pervasive grammatical categories, to genres of discourse, the kinds of things one is expected to say and hear in a given society. Both the music and the meanings of languages are organized differently for each one—and, again, there are thousands. This gives the continuing impression, through the generations, that different languages convey different worlds (Sapir 1921: 120–21; Friedrich 1986: 16).

The idea that ethnography is a kind of translation (Asad 1986; Rubel and Rosman 2003; and see the essays in this issue) has largely been treated metaphorically, as an implication of the broader movement to see cultures as texts to be interpreted (Geertz 1973; Ricoeur 1973). Now (Hanks and Severi, this issue), we are seeing an attempt to use the idea of translation as an alternative both to fixed cognitive universals and to an ontological pluralism of sealed worlds, implying instead a focus on the processes of exchange and transfer not only between societies (“cultures”), but also between different social classes, genders, caste groupings, within societies, which in fact define all social interaction.

There is, however, a risk here, as there is in any analogy. By extending the idea of translation out of language into culture more broadly, we weaken its force of surprise. Languages that are mutually unintelligible are absolutely so. To recap Roman Jakobson’s functions of language (1960): when someone speaks, say, Turkish—one of the approximately 9,990. languages that I absolutely do not know,2 I do not understand a single word. The referential function is missing, or rather was never there. A unilingual Turkish friend and I may be very phatic together over coffee, I may feel strong aspects of his expressivity, he can be effectively conative in getting me to jump out of the way of an oncoming bus, I may even sense some of the rhythm and alliteration if he recites a poem, but as long as he keeps speaking Turkish, there is no referential content, unless the two of us start working hard and for a long time on the metalinguistic function—which is to say that we start learning each other’s languages.

Note how different this is from what is called translating culture. Since what we call culture includes virtually all aspects of life, as soon as I arrive in, again, say, Turkey, I think I recognize all kinds of things: houses, restaurants, markets, places of worship; men and women, children and adults, dressing somewhat differently from each other and occupying somewhat different roles. I can immediately begin to form impressions on what might someday become ethnographies of the economy, politics, social organization, religion. We can call these translations, and they share characteristics with interlingual translations. But there is a precious opacity to languages that other human practices do not share, or do not share so unavoidably. Staying close to actual linguistic translation practice offers ethnographers a salutary reminder of how much, in all aspects of a culture, they do not understand.

As against the model or metaphor of the text, an alternative tradition, of philology and ethnopoetics, has opted, rather, for what we might call the metonym of the text: texts exist in cultures and in particular linguistic media. In this view, texts are seen not as “fragments of culture” (Silverstein and Urban 1996: 1), but as cultural foci, incomprehensible outside of their cultural and social situations, and, above all, made of language, which is to say, always a particular distinct language.

The fact of linguistic difference involves every level of language, from the untranslatability of phonetics and phonology (Webster 2014) to that of the universe of presuppositions out of which the source text arises, with levels of lexicon and grammatical architecture in between. Moving from one language to another and trying to have coherent “normal” texts in both must necessarily mean suppressions and additions—what A. L. Becker (1995), following Ortega y Gasset, calls deficiencies and exuberances; what John McDowell (2000) calls muting and adding. Yet it seems to go without saying that modern Western translation, whether literary translation or the mass of “useful,” that is, business, government, and other organizational translation, has as its central, often its only, goal to convey what are considered the necessary or useful aspects of the referential meaning of a text into another language.

But producing such a usable text, that is, a normal-sounding or normal-reading text in the target language (to use the rather frightening terminology of translation theory), requires alterations of what was there in the source text, even in its referential meanings, at every level of language. The original is thinned out (deficiency), or else one feels it has been thickened up with material that wasn’t there in the original (exuberance). This situation is all the more acute if we hope to convey not only the referential sense, but also something of the style and poetics of the original, or its role in its own society (Silverstein 2003).

Sensitive or suspicious readers have long felt that this transmogrification involves loss or betrayal, as is conveyed in the many derogatory aphorisms about translation: traduttore traditore, “translator [=] traitor” (a classic bit of metapragmatics that illustrates itself in its patterned untranslatability); a translation is a “phantom limb” (Robinson 1996 [1997]); “a translation is like a stewed strawberry” (attributed to Harry de Forest Smith in Brower 1959: 173). This last simile conveys both the maintenance of something through translation and at the same time the loss of freshness, that is, of complexity, and also, evidently, of structure. But the converse, conveying all of the meanings carried in the source statement, and only those, requires creating a “monstrosity” (McDowell 2000) in the target language. In this theoretical sense, of full transference from one to another normal linguistic frame, translation is impossible. And yet in practice it is both possible and necessary. In Franz Rosenzweig’s words, “Translating means serving two masters. It follows that no one can do it. But it follows also that it is, like everything that no one can do in theory, everyone’s task in practice” (1926 [1994]: 47).3

Given this reality, “normal,” ordinary translation, what Jakobson (1959a) calls interlingual translation or translation in the strict sense, already carries a huge conceptual and cultural load: any translation is a translation of cultures or worlds. If we take language diversity seriously at all, then translators are on the front line; they are pilots traversing a relativistic linguistic universe.

Here I will offer a brief run through some modern Western ideas about translation, highlighting what seem to me to be particularly acute formalizations, and propose a radical view of translation that seems particularly suited to anthropologists. Going over this material seems important for an anthropological audience: the history of translation theory in the West parallels that of anthropology as a history of attitudes toward other ways of saying and thinking.4

The following part of the paper will explore a single example drawn from a ritual performance carried out in the Central Himalayas of northern India.

Renaissance diversity

From the sixteenth century, over a relatively short time, the Western European world, based on an agricultural economy, a fixed hierarchical social organization and concomitant ideology, and the cultural dominance of Latin, was transformed.

The feudal economy and social order were being replaced by an expanding capitalism; nation-states were on the rise with the valorization of state vernacular languages against Latin; Western societies were conquering and colonizing whole sections of the globe, setting up a world political-economic system. New philosophical systems and experimental sciences were coming into being, and the great political- ideological structure of the church was being challenged from within. For many scholars of the century, new worlds, in many senses of the word, were opening up. One focus of debate was the relative value of vernacular languages versus Latin and Greek; another was the accessibility of both scientific and biblical truth to the masses of people; and much of the discussion turned on questions of translation.

The dominant medieval view had been that the diversity of languages was a curse, God’s punishment for the human pride manifested in the construction of the Tower of Babel. Both Protestant reformers in the center and north of Europe and Catholic language activists in France and Italy challenged this view and saw language diversity either as a positive good or as a problem that could be handled through translation. I will begin with the better-known Protestant theories, but devote more time to Italian views of diversity, since I presume that these are less familiar to anthropologists.

Translation, for some Reformation thinkers, such as John Calvin (1509–64), at least allowed some mitigation of the curse of Babel. What one could translate was the meaning, the sense of the text, and this could be conveyed in vernacular tongues as well as in classical ones. The question of whether and how to translate Holy Writ into vernacular languages became a central religious and political debate, one of the central issues of the Reformation. With the Bible translations of the sixteenth century, translation itself came to be an essential part of a double goal: to reveal the referential meaning of Holy Writ to anyone who could read the vernacular, thus dispossessing the clergy of their monopoly on sacred texts; and at the same time to valorize the vernacular itself as a national tongue, since a part of the Protestant argument was for national churches headed in each case by the secular head of the country, not by the pope.

In Protestant translation practice, what is important is not the particular form of the language, but that the referential meaning be conveyed in a colloquial vernacular that Everyman can understand. This view was expressed clearly by Luther: the goal is “to produce clear language, comprehensible to everyone, with an undistorted sense and meaning” (“Preface to the Book of Job,” cited in Rosenzweig [1926] 1994: 48); “in speech the meaning and subject matter must be considered, not the grammar, for the grammar shall not rule over the meaning” (“Translator’s letter,” 1530, cited in Bassnett 2014: 59). This view has continued to inspire Protestant translation theory up to this day, notably that of the Summer Institute of Linguistics: true conversion requires that God’s Word be accessible in the speaker’s own language (see, e.g., Handman 2007).

In Italy, parallel issues arose around the Questione della lingua: what language should serve as the cultural cement for Italy, now thought of as at least a potential national entity in spite of its political fragmentation. Here the basic argument was that the multiplicity of vernaculars was an aspect of the diversity of the world, in itself a good thing, an expression of God’s infinite creative power. As Claude-Gilbert Dubois puts it about this period in Italy and in a France heavily influenced by Italian thinking, we “witness a kind of ennobling of the multiple. . . . Multiplication is not corruption, but fecundity” (1970: 119).

This is an enormous shift from the medieval view of an ordered hierarchy: it is no longer easy to tell what is superior and what is inferior. In the words of the Florentine humanist Benedetto Varchi (1503–65), “In the universe there must exist . . . all the things that can exist; and nothing is so small or so ugly (tanto picciola nè così laida) that it does not contribute and add to the perfection of the universe” (cited in Stankiewicz 1981: 181). If even the smallest and ugliest thing adds to the perfection of the universe, this implies a fundamental incommensurability: instead of a clear hierarchy, perfection is an infinite scatter. And this refusal of hierarchy applies to languages as well: one cannot assert the superiority of some languages over others. In his Dialogo delle lingue ([1542] 2009: 35), the Paduan humanist Sperone Speroni (1500–1588) has his mouthpiece assert: “I hold it certain that the languages of every country, such as the Arabic and the Indian, like the Roman and the Athenian, are of equal value.”

The implication of this argument is, then, to challenge the Babel story and say that the multiplicity and diversity of languages is neither a curse nor a mere felix culpa, but a positive good, a gift from God. This was said explicitly by the Sienese humanist Claudio Tolomei (1492–1556): “Fecund nature (la feconda natura) did not accept that there should be a single form of speech in this great and varied edifice of the world” (1531, cited in Stankiewicz 1981: 180). Varchi, for his part, concludes: “I say that nature could not, and indeed perhaps should not, have made a single language for the whole world” (cited in Demonet 1992: 513).

Edward Stankiewicz sums up the views of this period: “One of the central notions in this program is the recognition that each of the individual, living languages has a place in the overall scheme of things, and that each of them is endowed with linguistic properties which make up its distinctive character, or ‘genius’” (1981: 180).

The second part of this statement raises the notion of the distinctiveness of each language. Languages are not only valuable, but their value lies in their differing specifics, the distinctive character or genius of each one. Dubois (1970: 119) makes an explicit link between the diversity of words and the new idea of an infinite plurality of distinct worlds proposed by Giordano Bruno later in the century: “The plurality of languages, like the plurality of worlds invented by Giordano Bruno during the same period, demonstrates the immense richness of the spirit.”

But the evocation of Bruno (1548–1600) raises a different point. We have seen that for Luther, and for Reformation Bible translators generally, what mattered was a clear conveyance of referential meaning—precisely not the exploitation of the specific riches of each language. Bruno’s own defense of translation was based on a similar view of philosophical and what we would call scientific discourse. Bruno said that it was perfectly all right to read Aristotle in an Italian translation rather than the original Greek, in that either of these permitted the reader to test Aristotle’s theories empirically and so recognize Aristotle’s errors. Scientific knowledge was based not on the “pedantry” of knowing Greek and Latin, not on the word, but on knowledge of the world.

Unlike the Bible translators, the Italian theorists went on to justify a different side of language use from that of either Scripture or science: in literary and poetic discourse, the unique genius of each language comes into fruition as a form of social play. The Florentine shoemaker-sage Giovan Battista Gelli (1498–1563) writes that “every language has its particularities and its caprices (capresterie)” ([1551] 1976: 201–2); Varchi writes that Florentine has a “certain peculiar or special or particular property, as do all other languages” and boasts of “certain qualities and capricious turns (certe proprietà e capestrerie) in which the Florentine language abounds.” And these turns are particularly apparent in everyday popular discourse (cited in Stankiewicz 1981: 182–83). Language cannot be encompassed in a definite set of rules, according to Varchi, because the rational faculty is incapable of capturing the expressive power of language in everyday speech. “Reason should prevail and win in all matters, except in language, where if reason is contrary to usage, or usage to reason, it is usage, not reason, which must always take precedence” (cited in ibid.: 183).

The most famous phrasing of this distinctive quality was in Joachim du Bellay’s 1549 Defense et illustration de la langue frangoyse. Most of this text is a translation from Speroni’s Dialogo, but it is du Bellay who specifies that each language possesses a certain je ne scais quoy propre—and he makes this point as part of a discussion of the difficulties of translation, as indeed do a number of Italian authors. Each language has its own force and beauty (forze e bellezza) which cannot be carried over into another tongue. Gelli ([1551] 1976: 201) writes: “They say that discourses (cose) that are translated from one language into another never have the same force or beauty that they had in their own. . . . To say things in one language in the style of another has no grace at all.”

Repeatedly, the discussion of the distinctiveness of each language is part of the discussion of the difficulties of translation. The examples used are usually literary ones: Homer, Virgil, and Dante or Petrarch—and notably not Aristotle—none of whose work can be translated without losing its most important quality: the distinctive beauty of each. We find the example in the De subtilitate (1550) of the peripatetic mathematician-physician Girolamo Cardano (1501–76):

The utility of the diversity of languages is that they can thereby express all the affects of the soul. The proof is that one cannot express Homer’s sentences in Latin or in our mother tongue, nor Virgil in Greek or in our mother tongue, and even less the poems of Francesco Petrarch in Latin or Greek. (Cited in Dubois 1970: 118)

Implicit in the discussion is a distinction between two goals of discourse, that of conveying information and that of projecting the force and beauty of a language—to use anachronistic terms, between the referential and poetic functions of language. Translations from any language into any language can convey what we could call referential meaning but fail to convey play, capresterie, the force and beauty deriving from specific circumstances of social life. The distinction is central for Bruno, and Gelli writes, “I know that translation is done for the sake of sciences, and not to display the power and beauty of languages” ([1551] 1976: 202, cited in Stankiewicz 1981: 182). The formal distinction of the two modes was made by Sir Francis Bacon (1561–1626) between literary grammar, which deals with languages and which cannot really render the je ne scai quoy of each language in translation; and philosophical grammar, which deals with “the analogy between words and things, or reason” (Hüllen 2001: 213) and is in fact adequate for the advancement of learning.

Already, then, different goals of translation were being distinguished in the seventeenth century. For Protestants, the Sacred Scripture plays the role of science for Gelli or Bruno: in both cases, the goal is to convey clear meanings that are of immediate spiritual or secular use. But both presume the possibility of another attitude, one that fully recognizes the genius of each language. This possibility is developed above all by the Italians, with their recognition that linguistic worlds are constructed by bringing out the possibilities of each language in social interaction and in play.

“Les Belles Infidèles”

By the early 1600s, the French language and French style were dominating Europe. Already in the previous century, French discussion of translation had tended to be nationalistic and, notably, contrasted the “direct order” of the French sentence (Subject-Verb-Object, with adjectives coming for the most part after the noun) with the horrid mixup of Latin word order. With the rise of Cartesian rationalism, the assumption came to be strongly reinforced that all human beings think in the same way, languages being a mere means of external formulation and communication. When Descartes himself was challenged (by Hobbes) as to whether what he called reason could really be merely a local way of linking words, he replied: “Who doubts that a Frenchman and a German can have the same thoughts or reasonings about the same things, despite the fact that the words they think of are completely different?” (Descartes 1984: II, 125–26). In 1669, Descartes’ follower Louis Le Laboureur claimed that since everyone in the world thought in the direct order of French, the Romans too must have conceived their sentences in French, then added a step to scramble the syntax before speaking or writing.

This basic position, as well as a strong sense of cultural self-sufficiency, motivated French translation theory. The argument was that the translator should reproduce not just the words of the text, but also the effect it had on its original readers. Since the text was not exotic or strange-sounding for its intended public, we should translate it in a way that is not disturbing for our public. If this requires changing meanings, so be it: such changes will in fact often be an improvement on what are sometimes needlessly repetitive, longwinded, incoherent, or indelicate originals. And this applies equally to scientific, religious, and literary works. In the words of the great translator of the century, Nicolas Perrot d’Ablancourt (1606–64),

It would be a Judaic superstition to attach oneself to the words and abandon the design for which they were used.... It is only rendering half of an Author if one cuts off his eloquence. As he was agreeable in his own language, he should be so in ours as well, and insofar as the beauties and graces of the two are different, we should not hesitate to give him those of our own country, since we are removing his. Otherwise we would be making an ugly (meschante) copy of an admirable original. (1638, cited in Horguelin 1996: 76)

When Perrot’s translation of Tacitus—in which he says he sought “to look not so much to what [the author] said, as to what he should say “ (cited in ibid.: 78)—came out during the 1640s, the general opinion was that it was a great improvement on the original. But a more extreme statement accompanies his 1654 translation of the quite scatological Greek satirist Lucian of Samosata. Plainly there is plenty in Lucian that would shock seventeenth-century French sensibilities.

It was . . . necessary to change all this, to make something agreeable; otherwise, it would not have been Lucian. . . . For this reason I do not always attach myself to the words or the ideas of this Author; and, maintaining his goal, I arrange things to our air and our style. Different times require not only different words, but different thoughts; and it is the custom of Ambassadors to dress in the fashion of the land to which they have been sent, for fear of appearing ridiculous to those they aim to please. It is true that this is not, strictly speaking, Translation; but it is better than Translation. (Cited in Horguelin 1996: 80)

Indeed, Perrot would write that his translation of Thucydides had brought his author back from the dead in the form of a modern Frenchman; this is something for which Thucydides might well have been grateful. On the contrary, failing to change the original sufficiently, writes Perrot, produces a “monstrous body” (corps monstreux; cited in Venuti 1995: 47).

Not everyone’s arguments were as intelligently put as Perrot’s. Pierre Perrin (c. 1620–75), the so-called Abbé Perrin, claimed that his 1648 translation of Virgil brought Aeneas back as a French cavalier, “dressed up with pomp and plume and bangles” (cited in Ladborough 1938: 86).

There were a few detractors of this method, mainly among philologists and what were called the traducteurs scrupuleux (as opposed to Perrot and the traducteurs libres). Among these, Gilles Ménage (1613–92) is reported to have said that Perrot’s translations were like a lady he had once known in Tours: she had been belle, but infidèle. Since then the whole movement of elegant domesticating translation has been known as “Les Belles Infidèles.”

Many worlds

Unlike the French, who saw French as the closest linguistic approximation to universal values of reason and art, seventeenth-century German scholars valorized German for its very particularity and distinctiveness, its German-ness. Justus Georg Schottel (1612–76) devotes a section of his Extensive work on the German language (1663) to “How one should en-German (verteutschen)”—this is, characteristically, an alternative German word for übersetzen, “translate”:

To achieve this [the improvement of German] the arts, the sciences, history, and other realia should be dressed and become known in truly natural German, so that our language . . . will not smell of un-German Spanish pride, or Italian splendour gone stale, or French pronunciation and circumlocutions that are neither here nor there, or of other things, but it will have its own short and euphonious German nature, rich in meaning and sense. (Cited in Lefevere 1977: 11)

Leibniz’s writings on German go in the same direction, arguing, for instance, that it would be a pity if German ended up denatured like English (Leibniz [1697] 1916: 30).

The reaction against French models, and against all universal abstract standards, was part of a wider movement to create a cultural Germany even in the absence of a political one. These tendencies reached their fulfillment in the work of Johann Gottfried Herder (1744–1803) and the following generation of Romantics (see Berman [1984] 1992). Herder’s view of human history was one of clear celebration of diversity. In explicit rejection of French translation models, he insists on fidelity (Treue ) to the distinctiveness of the source language and the source text, and, in an echo of Renaissance theorists, writes:

I certainly do not believe that there exists any language in the world that can convey the words of another language with the same force and equivalent words. . . . The richest and the most useful language is one that best lends itself to word-for-word translations, translations that follow the original step by step. (Cited in Sdun 1967: 21)

Herder goes so far as to propose that the best thing for a language’s unique development might be to keep it pure of all translations, all influence from other languages, “like a young virgin” (cited in ibid.: 26). These apparent contradictions grow out of a single problematic. To argue for foreignizing translations that make other languages and cultures accessible to us, each in its unique character, or to argue that we should stay clear of other languages and cultures so as to preserve our own unique character, are conclusions from a single set of premises: that there are a multitude of distinct and unique language-cultures, each one valuable. This is very different from the assumptions behind “Les Belles Infidèles.”

In the early nineteenth century, the Romantics mounted a full-scale defense of “faithful” translation. A. W. von Schlegel criticizes what he calls French translation for making the foreign visitor “dress and behave according to [French] customs” (cited in Lefevere 1992: 79), and, in the period’s translation manifesto (1813), Friedrich Schleiermacher (1768–1834) says that to translate a foreigner as he would have written had he not been foreign is like painting a portrait of a man not as he actually looks, but as he would have looked if he had had a different father. And it was Schleiermacher who posed the great dichotomy in a formulation that has been cited in most of the major statements on translation since then:

The true translator, who really wants to bring together these two entirely separate (ganz getrennten) persons, his author and his reader, and to assist the latter in obtaining the most correct and complete understanding and enjoyment possible of the former without, however, forcing him out of the circle of his mother tongue—what paths are open to the translator for this purpose? In my opinion, there are only two. Either the translator leaves the author in peace as much as possible and moves the reader toward him, or he leaves the reader in peace as much as possible and moves the author toward him. ([1813] 1992: 41–42, translation modified)

Typically Romantic is the insistence here on the separateness and uniqueness of persons, and by implication of their worlds.

Schleiermacher presents his division as balanced, but his sympathies are with translation that moves the reader toward the author. Where the reverse type offers the reader no surprises, this has the potential of opening up his or her sensibilities, and with it that of his or her culture and language. After all, Schleiermacher was also the founder of hermeneutics as a discipline. German Romantic foreignizing challenged French hegemony and called for making Germany a center of intercultural ferment at the heart of Europe:

Our nation may be destined, because of its respect for what is foreign and its mediating nature, to carry all the treasures of foreign arts and scholarship, together with its own, in its language, to unite them into a great historical whole, so to speak, which would be preserved in the centre and heart of Europe, so that with the help of our language, whatever beauty the most different times have brought forth can be enjoyed by all people, as purely and perfectly as possible for a foreigner. (Schleiermacher cited in Venuti 1995: 110)

So Germany’s openness to others sets it up to be the center and arbiter of world culture.

Since the positing of clear positions in the seventeenth through the early nineteenth centuries, translation practice in the West has wavered between that of the treacherous but lovely “Belles Infidèles” and that of the faithful Romantics; these tendencies have been called domesticating and foreignizing (e.g., Bassnett 2014), remarkably untranslatable terms that correspond, for instance, to the equally ugly French cibliste and sourciste.

Boas and translation

When he came to America from Germany, Franz Boas (1858–1942), the founder of modern North American anthropology and linguistics, brought the German assumption of diversity with him, but in a more radical form. Given his practice of doing research by actually living with “primitive” people, and his goals of attending to the history of all peoples (Bunzl 2004) and treating all verbal traditions with the philological respect usually given to Greek and Latin (Boas 1906), Boas shared neither the nationalist agenda nor the hierarchization of peoples and languages that marked most Euro-American thinking about language and culture. Instead, his own political agenda, like that of his students, was largely contestatory, seeing exposure to alternative ways of living and constructing the world through language as inherently challenging the prevailing assumptions of the Euro-American societies of his time.

From his paper “On alternating sounds” (1889) through the Introduction to the Handbook of American Indian languages (1911) and into his later writings, Boas focused on examples of misunderstandings between languages and cultures (discussion in Leavitt 2011: 130–32). His is an anatomy of misreadings—of sounds, of the meanings of and relationships among words, of grammatical patterns. In every case, the speaker of a given language expects to find familiar patterns when confronting a new one, and the speaker’s errors can be shown to result not from the inferiority of the other language, its lack of structure or vagueness or excessive fussiness, as several generations of evolutionists had claimed marked primitive languages, but from the mutual interference of different coherent systems. We—that is, all speakers—mishear sounds, misattribute meanings, and misconstrue grammatical forms in a different language because of our own legitimately acquired prejudices as speakers of our own language; and speakers of the other language will make exactly the same kinds of mistakes—not the same mistakes, but the same kinds of mistakes—when they try to handle ours.

This means that the site of illumination is on the edge, on the frontiers between systems: foreseeable misunderstandings are evidence for the existence of distinct systems. There is a theory of translation inherent in this approach, although Boas did not formulate it as such. Once differences have been established, the researcher should seek to grasp the forms of the new language as they are actually used, not replace them with familiar forms. This is the method propounded by Wilhelm von Humboldt (1767–1835), as, for instance, when he attacked Jesuit manuals of Japanese for using largely inapplicable Latin parts of speech (Humboldt [1825] 1906). Boas (1900) similarly criticizes a missionary grammar of Kwak’wala for drawing parts of speech from English and forcing pieces of Kwak’wala into this structure.

Such a view of languages as coherent and different systems means that the analyst must not be satisfied with translation equivalents, nor take the possibility of glossing as proving that languages are “really” the same. The fact that an English sentence involving tense might be the best, in the sense of the most normal-English, gloss of a Hopi sentence without tense does not prove that Hopi has “tacit tense.” A. L. Becker has put this point elegantly in a book with the significant title Beyond translation:

Our general tendency is to “read into” our experience of a distant language the familiar things that are missing, all the silences, and then we claim that these things are “understood,” “implied,” or “part of the underlying logical structure” of these languages. . . . It takes a while to learn that things like tenses, and articles, and the copula, are not “understood” in Burmese, Javanese, or Malay. . . . In Burmese these things aren’t implied; they just aren’t there. (1995: 7–8)

If required, Burmese is perfectly capable of conveying the information carried in tenses, articles, and the copula by using its own methods. As Boas insisted, one can speak of any subject in any language—at the limit, one might have to add some vocabulary. This means that referential content can be translated, if only sometimes via elaborate paraphrases: the fact of translatability of referential content is indeed universal.

Boas further elaborated Humboldt’s recognition of the importance of obligatory grammatical categories, which oblige the speaker to choose alternatives within a given realm of experience. In English, for instance, most nouns require a choice of singular or plural forms; finite verb forms require a specification of person and tense, that is, relationship between the time of speaking and the time of the event spoken about. While it is possible to specify this information when speaking any language, plenty of languages do not require their constant specification. And other obligatory categories may exist: you may be required to specify how you know what you are talking about (evidentiality); or every noun may be classified by its shape (Jakobson 1959a, 1959b; Friedrich 1969). This immediately raises a question of translation: How much does one translate obligatory grammatical categories? Should they be suppressed because they are obligatory and therefore “invisible”? How does one judge that they are coming into “visibility”?

This is the central question of Jakobson’s often-quoted paper on translation (1959a). He gives the example of French gender. We are told in our high-school French classes that gender is a purely formal part of French grammar: you just have to memorize that it’s la table and le mur, but there is no sense in which a table is really thought of as feminine and a wall as masculine (this position is stated in Brown and Lenneberg 1954). Against this, Jakobson shows that in some circumstances gender does play a role in ideation and certainly in connotation.5

If we accept this, then gender-meaning is there, hovering, potentially marking every noun. Should we be translating la table as “the table (it’s a girl)”? Paul Friedrich tried something like this in his treatment of the tale of Uncle Peanut and Aunt Onion (1969: 30; 1979: 485) from Purépacha (earlier called Tarascan by outsiders), a Mexican language that has fourteen obligatory markers of shape. Friedrich’s English translation adds little tags indicating the shape of activities. The English of this text is odd indeed, and in fact requires a paragraph of grammatical annotation for each paragraph of translation. It would be hard to tell this story as a story in Friedrich’s rendering—but, on the other hand, his treatment does tell you an enormous amount about the Purépacha language and, I dare say, the Purépacha world.

Many of the central arguments in the writings of Boas’ students Edward Sapir (1884–1939) and Benjamin Lee Whorf (1897–1941) deal with obligatory grammatical categories, using the same kind of treatment of examples that Boas does. In fact, this whole Boasian literature on language differences, given that it is taking place in English, can be read a series of reflections on translation.

The “Sapir-Whorf hypothesis,” first proposed in the 1950s well after the deaths of both Sapir and Whorf, is often taken to claim that each language is a closed and sealed system unto itself. This is utterly different from the Boas-Sapir-Whorf practice of operating on the borders of systems with the understanding that passage between worlds is always possible but requires work. The reading of Sapir and Whorf as linguistic determinists led to easy refutations: first, that Whorf contradicted himself by seeking to describe Hopi concepts in English; second, that the very fact that translation is possible proves that all languages are carrying the same content (Black [1959] 1962; Davidson 1974). This last view should be contrasted with the reverse one, maintained by German linguists who did believe that one could never really escape the horizon of one’s native language: for them, the fact that there are any difficulties in translation proves that languages are profoundly different (Bynon 1966: 472).

Misery and splendor

During the same decades that Boas and his students were active, Saussure-inspired linguists and philosophers on the Continent were developing their own theories of the systematicity of languages, whether through Prague School linguistics or a developing French structural linguistics represented in the work of Émile Benveniste (1902–76). Here, as in other fields, the recognition of multiple systems typical of structuralism parallels the views of the Boasians (see, e.g., Jakobson 1959b).

Within translation, one statement of the radical differences among languages is found in the dialogue “The misery and splendor of translation” ([1937] 2000) by the Spanish philosopher José Ortega y Gasset (1883–1955), another of the most anthologized pieces in the translation canon. Here is how the case is put in the last part of the dialogue, “El esplendor”:

“The simple fact [is] that the translation is not the work, but a path (un camino) toward the work. If it is a poetic work, the translation is not one, but rather an apparatus, a technical artifice that brings us closer to the work without ever trying to repeat or replace it. . . . I imagine, then, a form of translation that is ugly (que sea fea), as science has always been, that does not claim to wear literary garb; that is not easy to read but is very clear indeed, although this clarity may demand copious footnotes. The reader must know beforehand that when he reads a translation he will not be reading a literarily beautiful book, but rather will be using a fairly boring apparatus (un aparato bastante enojoso); one, however, that will truly help him transmigrate within the poor man Plato (transmigrar dentre del pobre hombre Platón)” (Ortega y Gasset [1937] 2000: 61, 62, translation modified)

I put this quote within quotes because in his dialogue Ortega is not, in fact, putting these words into his own mouth—which he could well have done, since he presents himself as one of the speakers, and he has himself talk quite a lot. The scene takes place in Paris, apparently in 1937, during Ortega’s exile from the Spanish Civil War. It happens at the Collège de France and is said to involve professors and students of the Collège. This section of the text is provoked by Ortega’s asking one of his interlocutors, described as “a brilliant scholar of linguistics of the new generation,” to give his own views on translation. If there is any historical reality to this setting, the only person this could possibly have been was Benveniste, who was elected to the Collège that very year at the age of thirty-five. If we take this attribution seriously—as far as I know, Benveniste never laid claim to it—then the real source of these ideas is the other great structural linguist of Jakobson’s generation: this ringing repudiation of “Les Belles Infidèles” in favor of ugly translation and respect for the otherness of the source text—and the source world—comes not out of Spanish existentialism, but French structuralism. The Spanish philosopher is the channel and, of all things, the translator of Benveniste’s views on translation.

Reference and style

We have seen that Reformation and Renaissance traditions distinguished between translating for what we would call referential meaning, which is held to be appropriate for sacred and philosophical texts, and the much less evident, perhaps impossible, process of seeking to capture the full exploitation of the resources of a language as exemplified in literary texts; and we have seen Ortega/Benveniste propose a radically scientific approach to the otherness of poetic language. Scholars of the Boasian tradition, for their part, focused almost exclusively on referential meaning, which, as they maintained, can always be translated, if you are willing to use enough circumlocutions.

The exception within this tradition was in the work of Edward Sapir. Sapir, unlike Boas and Whorf a student of literature and a poet himself, felt he recognized an implicit poetic character in the very structure of a given language: a music in phonetics, resonances of suggestion in the choice of words and how words are put together, and figures implicit in grammatical categories. For Sapir, once again, the ideal position is on the edge, on the front line between linguistic worlds:

Since every language has its distinctive peculiarities, the innate formal limitations—and possibilities—of one literature are never quite the same as those of another. The literature fashioned out of the form and substance of a language has the color and the texture of its matrix. The literary artist may never be conscious of how he is hindered or helped or otherwise guided by the matrix, but when it is a question of translating his work into another language, the nature of the original matrix manifests itself at once. All his effects have been calculated, or intuitively felt, with reference to the formal “genius” of his own language; they cannot be carried over without loss or modification. (Sapir 1921: 227)

Sapir distinguishes different literary styles, more or less dependent on the specifics of their linguistic material versus their referential content and so more or less satisfactorily transferrable between languages. This of course recalls the distinction made in the Renaissance and Reformation, except that these earlier authors reified the problem as one of different kinds of texts, rather than different functions of language, as suggested by Sapir and made explicit by Jakobson (1960).

The development of a North American school of ethnopoetics in the 1960s and 1970s, a cooperative enterprise among anthropologists, linguists, and poets, was predicated on recognizing the Boasian failure to grasp the poetic dimensions of the texts being recorded (e.g., Hymes 1999 on Boas). The ethnopoetic work of Dell Hymes, Dennis Tedlock, and their colleagues, the effort to rediscover or re-present poetic form from non-Western texts, can be seen as an attempt to add new dimensions to translation, bringing the reader closer to the singer or teller of tales. The great arguments in ethnopoetics (e.g., Hymes 1981; Tedlock 1983; cf. Mason 2008) have been about how to translate and how to present a translated text. From the side of the poets, the inspiration for ethnopoetics was exemplified in Jerome Rothenberg’s experiments in what he called “total translation,” seeking, again, to add contextual and interpretive dimensions to his English renderings of Seneca ritual texts (see Rothenberg 1968).

Sapir’s idea of a poetry implicit in the grammar of a language was developed by Paul Friedrich (e.g., 1986) in ways that had direct implications for translation, and particularly for the dichotomy between the poetic and the supposedly purely referential. If a given grammar lends itself to certain connections, metaphors, metonyms, rather than to others, then a translator is dealing with an underlying poetics that is there in the very weave out of which any text is made, however referential it may look.

In his work on the necessarily situated nature of any enunciation, Michael Silverstein has shown how reference is only one dimension of what are necessarily far more complex communicative acts. In particular, he has drawn attention to aspects of language that are cross-culturally more or less available to the metalinguistic awareness of speakers themselves (Silverstein [1981] 2001), thus granting a greater role to the effect of language ideologies on languages themselves than had been allowed in the pure Boasian model. Silverstein’s work on translation (2003) distinguishes among translation proper, that is, transfer of referential meaning including grammatical meaning; what he calls transduction, the attempt to transfer stylistic and indexical qualities of usage; and transformation, a shift in semiotic modality to try to render some aspect of the situational meaning of the source event.6 Silverstein is led to call for a necessary role for ethnography in going as far as transduction:

How does one capture the “tone,” i.e., indexical penumbra, of a word or expression in a source text by one in a target language used in a highly distinct culture? Clearly, something on the order of a cultural analysis of both systems of usage is a prerequisite to finding a route of transduction, in analytic terms that reveal both the similarities and the differences, so as to be able to navigate a proper transduction from the source to the target. (2003: 89)

Translation studies and reality

What is usually referred to as translation theory or translation studies has come into being since the 1970s as an interdisciplinary field. Two developments mark it off from earlier debates, which, as we have seen, focus primarily on how one should translate. One is the recognition that translation is always carried out within what has been called the target culture, for reasons peculiar to the target society. Since this is the case whatever strategy of translation one adopts, foreignizing translation serves purposes that are defined by the target society as much as does domesticating translation, or else it would not be taken on.

Another of the developments of recent translation studies has been to recognize the hard reality of translation practice. Instead of simply asking how one should translate, these theorists are also asking what a society chooses to translate and why. While contemporary translation theory tends to come down in favor of foreignizing translations, which respect the original and seek to transform the reader (Venuti 1995; Bassnett 2014), normal translation practice continues to be very much like “Les Belles Infidèles”—after all, translation is a business, and most translations are made to sell. It is clear (Venuti 1995; Bellos 2011) that the mass of translation going on in the world is from English to other languages, with a much smaller percentage going in the other direction, and this primarily from a very small number of tongues. In translations into English, domestication dominates. In other words, what actually gets translated, in substance and style, tends to reinforce the hegemony of Anglo-American and, to a lesser degree, Western European culture.

The question, then, is what we want to translate and why. Anthropologists are in a particularly strong position to bring little-considered texts into the current world tradition, and to do so in a way that allows serious consideration both of the implications of language structure and of the anchoring of the text in particular values and particular interpersonal situations.

Twilight swinging

Traditionally, anthropologists have worked in societies that fall outside the main concerns of Western academic history. Many of these societies have only recently adopted systems of writing and still function to a large degree in oral media. For linguistic anthropologists working with oral texts, the question of translation involves yet another layer of complexity. It requires an imagined parcours from performance and reception to recording of one kind or another to written fixation in one or more text-artifacts to interpretation and translation (cf. Silverstein and Urban 1996). These problems have emerged for me in the most telling way in working on texts emerging out of oral performance traditions of the Central Himalayan region of Kumaon in what is now the state of Uttarakhand in northern India. These are largely performed by professional—that is, remunerated—drummer-singer-poets who are masters of a large number of named genres, some narrative, some ludic, many highly ritualized. Some of the ritual genres are performatively effective in bringing gods to be present and to manifest themselves in human bodies in what the Western tradition calls possession.

In Kumaon, the language spoken by the bulk of the population in their homes is one or another dialect of Kumaoni, a member of the Pahari branch of the Indo-Aryan language family, itself a member of the Indo-European family. Government, education, and commerce are carried out primarily in Hindi, the official language of the state. Most Kumaonis are Hindus who interact with a host of regional divinities as well as with the great Hindu gods. The regional divinities manifest themselves “in person” in nocturnal possession rituals called jāgar or “waking”; their appearance is provoked and controlled by the professional drummer-poet (Gaborieau 1975; Leavitt 1997). All ritual devoted to these regional gods is carried out in the Kumaoni language.

In what follows, I raise some of the questions that have come up in considering a small part of one of the early sections of a jāgar performance I recorded in 1982 in the northern part of District Nainital. Here the drummer-poet was Sri Kamal Rām Arya, one of the area’s better-known performers in a number of oral-poetic genres. The utterances come from the poet’s opening invocation to the gods and to the coming night, called sandhyā, “twilight” or “evening.” As is usually the case in this tradition, this is an oral performance; until recently, most of the poets were illiterate, and while Kamal Rām knows how to read and write, as far as I can see writing plays no part in his art. For the would-be translator this means that there is a series of steps, each one riddled with potential pitfalls, which must be crossed over before one has a “source” text(-artifact) with which to work—and such an artifact is never the only possible one. In the case of the utterances presented here, there was a recording in situ; there was an initial transcription of the recording by Sri Indar Singh Negi, an inhabitant of a neighboring village and native speaker of the language, who had gone through some months working with me and so had an idea of what transcription involved. Then transcriber and translator—and this, of course, meant that the translator already had done the basic work of phonological and grammatical analysis of the language—read over the transcription together while listening to the recording, making corrections and asking for explanations. In some cases, we returned to the poet for explanations, but these were rarely forthcoming: for the poet himself, his performative skill was never considered as separate from its exegesis (as is the case, for instance, of Yugoslav bards: Lord 1960; cf. Tedlock 1983: 3). After various iterations and revisions, we had a text-artifact in “the” Kumaoni language: “the” is in quotes because our transcription represented only the dialect and distinctive diction of this singer, and to some degree only on this occasion.

Note one of the crucial differences between this kind of ethnographic pretranslation, involving the creation of a text-artifact, and standard text translation: the former involves living performers and living interpreters, and so must necessarily be collaborative in nature (cf. McDowell 2000). How collaborative it is, in what direction the questions and answers go, depends on the situation; but it has the potential to shake up the usual flows of information.

The initial questions are thus those which center not on how to translate from code to code (i.e., from Kumaoni to English), but on how, in Jakobson’s terms, to operate an intersemiotic transfer (i.e., from speech to writing). Can we present the transcribed text in a way that preserves some of the qualities of a performance that has rhythmic, dynamic, and melodic dimensions? Even if a “total” transcription is impossible, Dennis Tedlock (1983) has shown that experimenting with transcription can allow the indication of elements of a performance which are lost in a standard prose or verse transcription.

Here is an initial transcription of some verses, with interlinear treatment of the first, including indications of grammatical marking.7 Kumaoni nouns have three cases, direct, oblique, and vocative—which are only overtly distinguished in some nouns—two genders, and two numbers. Like other modern Indo-Aryan languages, Kumaoni makes constant use of compound verbs, in which the first element carries the referential meaning while the second modifies or orients the way of being or doing and carries the grammatical information.

Figure 01
Figure 02

ghōl ki caṛi jo cha ghōl mẽ lhai gai cha, isvar mero bābā, sandhyā kā bakhat mẽbāti batauv ko dyār jo cha, isvar mero bābā, bātai bādi go, isvar mero bābā, sandhyā kā bakhat mẽ godi ko bālak jo cha god mẽ sukālo hai go, isvar mero bābā, sandhyā kā bakhat mẽ tumāro nām lhinũ, isvar mero bābā.

The little bird of the nest has gone into the nest, Lord my father, at twilight time, the traveler on the road, Lord my father, has tied his camp on the road, Lord my father, at twilight time the child of the lap has become happy in the lap, Lord my father, at twilight time we take your name, Lord my father.

What we have offered is a fairly straightforward transfer of the referential meaning of this fragment of a text-artifact. But even here a great deal has been added and a great deal supressed. Kumaoni, like other North Indian languages, has masculine and feminine gender, and feminizing a word is often used as a diminutive: here the “little bird” is my pragmatic reading of what in fact is “female bird.” I have suppressed the repeated use of jo cha, “who is,” which seems to me to be a nonsemantic filler in Kumaoni bardic performance. How to handle the semantic penumbras of isvar, borrowed from Sanskrit Īśvara, a title of God associated with certain schools of Hinduism but not others and which I have simply given as “Lord,” or ābā, both the vocative of bāp, “father” and the word for a guru or spiritual “father”? Kumaoni having no determinative article, I have put in “the” where it seemed appropriate, although the translation could as well read “a little bird has gone into a nest.”

This is a series of utterances that have only been heard, never written down except by Sri Indra Singh Negi by hand in prose presentation on sheets of lined paper. Basing myself on the poet’s apparent phrasing rather than his pauses, I could as well transcribe it:

ghōl ki cari jo cha
ghōl mẽ lhai gai cha,
isvar mero bābā,
sandhyā kā bakhat mẽ

and so forth, rendering it something like:

as twilight swings
what works are worked
Lord my father
at twilight time
Lord my father
cows, calves have been tied up
nest’s little bird
in the nest has gone
Lord my father
at twilight time

The basic syntax of the Kumaoni sentence, as we have seen, is Subject-Object-Verb. This means that in unmarked discourse, the verb acts as a sealant of sentences or clauses, which appear as packages with the complements tucked in between the subject and the verb. In Kumaoni narrative poetry, this structure is nice and tight: each line is a clause, usually a sentence, ending unambiguously with the verb. In the sandhyā, too, which is emphatically nonnarrative, many clauses do end with verbs, again helping to mark each one off as clearly as the stress on the first syllable marks off half-clauses; but syntactic ambiguity comes in when we try to find a place to put nonsentence clauses like “Lord my father” which must be connected to what precedes or what follows but give us no hint as to which option is to be preferred.

Thus the first transcription and tentative translation, in paragraph-like blocks, does convey something of the text’s weird syntagmatic structure, a string of clauses that can be connected up in several ways, making it impossible to tell where one sentence ends and the next begins. Like the relentless repetitions, the ambiguous clausal pattern of this opening recitative—a pattern that is completely different from that of narrative parts of the ritual—serves to hook in and focus the attention of the listeners, whether divine or human. What in more standard discourse would be clearly distinguished sentences here overlap to form much longer units, along which the listener is pulled. This quality is conveyed better, I think, in blocks rather than, say, in breath-lines (Olson [1950] 1997), or in a transcription strictly based on timing. The poet seems to be at pains to push the line of language through his breaths to create somewhat longer units, themselves divided by periods of his voice’s silence and the continuing beating of the drum.

But already this choice of transcription, using clauses separated by commas rather than, say, imposing a standard sentence-structure, involves a preliminary analysis of the relations among the participants in the ritual: that the singer is singing to gods and ancient gurus, that the audience is there as privileged over-hearers. In the ritual world, the point of this section is to invite and pull in the gods, into what we may call the empirical world, that available to immediate observation: the most evident thing about the performance of the sandhyā is the way it seems to trap the attention of the assembled people, who as it begins are chatting away, but are silent and attentive five minutes later, when the chant has taken over. The point of this part of the ritual is to capture and hold the attention of all these interlocutors, divine and human. The choice of how to present a transcription and a translation, in other words, already presupposes what Silverstein calls a transductional analysis.

Looking at other levels of language adds to the complexity of the task. Phonetics and phonology represent the most evidently untranslatable level. The Kumaoni text is full of alliteration: this short passage is extremely heavy with initial bs and gs, both unaspirated and aspirated. Other parts of this section rely heavily on rhyming, which will either be lost in translation or, to be maintained, will require suppressions of and additions to elements of the referential sense.

Rhythm and stress are also lost in translation. Kumaoni has a stress accent on the first syllable of most words. In this text there is a steady beat of hyperstressed syllables at the beginning of each half-clause (gāi bachan kā/bādan lāgĩ), maintaining a pattern that allows varying numbers of relatively unstressed syllables to be inserted between the stresses. A translation, even into another word-stress language such as English, loses the accent that here metricalizes half-clauses; this might be indicated by adding boldface to the translation, as I have added it to the transcription above, or, if in an oral rendition, by reading aloud with the right stresses. But here other problems, this time from the “target” language, impose themselves: to present something like normal English, determiners are unavoidable, adjectives precede nouns, and word order is Subject-Verb-Object rather than the Subject-Object-Verb order of Kumaoni and almost all other South Asian languages. If we stress the first syllable of each half-clause in English, different words are accentuated than in the Kumaoni, and these turn out to be words that don’t “deserve” the stress: “The cow and the calf/have been tied up, // the little bird of the nest/ has gone into the nest.” If we stress the words corresponding semantically to those stressed in the Kumaoni, we lose the distinction among half-clauses: “The cow and the calf/have been tied up, // the little bird of the nest/has gone into the nest.” And if we force the English words into a Kumaoni order, we have a monstrosity, but one that maintains something like the rhythm: “Cow calf’s/tying have-done, // nest’s birdie/nest-in has gone.” This is the only way to give the reader an idea of what is going on. It hardly can stand by itself, but requires explanation with reference to the transcription-artifact, to which the reader must have access. It is a case of going “beyond translation.”

Word choice, too, is always problematic: we can find English glosses for the important words in the Kumaoni text, but to understand what they are doing there, they must be explained or annotated. An exegetical paratext must surround the text. The oft-repeated word sandhyā, which is also the name given to this section of the ritual, means far more than simply twilight or evening: in orthoprax Hinduism, it is the word used for both of the junctures between day and night, which are holiest moments of the daily cycle, and for the prayer said at dawn and dusk. Its phonology also marks it as a borrowing from Sanskrit, the sacred language, rather than an inherited Kumaoni word: the Kumaoni word for evening, which has undergone the expected transformations from Sanskrit, is sã̀s. Thus the word sandhyā has religious connotations that the English word twilight (or alternatives: dusk, evening) does not. This choice of the word ties the immediate perceptible scene into a cosmic scene, which is exactly what this section of the ritual is seeking to do.

Kumaoni oral poetry is not memorized, but composed in performance by combining fixed formulas (see the ideal model in Lord 1960).8 Giving that this is a living tradition, other poets from the same tradition are available to be listened to, and are being listened to, by thousands of people today. The performance of any given poet can change from twilight to twilight: while Kamal Rām’s sandhyās show great stability in style and wording, they too are longer or shorter depending on the occasion. And each poet has his own style or styles (they are mainly men), each drawing on a treasury of formulas that are largely shared across the region. The intertextuality is extraordinarily dense: this rendition of the sandhyā by this poet on this night echoes others at many degrees of distance. The reality of this variation could certainly be called to the reader’s attention, which means that the translated text is no longer the sole text, but becomes one element in a potential multitext of variations (as now exemplified for Homer: see the Homer Multitext website). In the single recording we possess of one of his jāgars, the most famous Kumaoni oral poet of recent generations, Gopi Das of Kausani (c. 1900–75),9 begins his sandhyā:

dhārā duba dinā, devo,/gārõ pari chāyā
hari jagadīsā, o devo,/gārõ pari chāyā

The day has sunk down on the mountain crests, gods,/in the valleys shadow has fallen
Oh Lord, oh gods,/in the valleys shadow has fallen

And he too uses the trope of birds returning to the nest at twilight time:

cārā oro kā panchī lai/cārā oro bāso lhai cha
ghōlā ko panchī lai/ghōlõ mē bāso lhai cha

Birds of the four directions/have settled in the four directions.
Birds of the nest/have settled in the nest.

The poet Jay Rām, whom I was able to record in a village close to that of Kamal Rām the previous year, chanted:

ghōl ki cari ghōl mẽ baith rai cha, gāy bāchõ kā badan bāndhi jānĩ, nau lākh
tārā khul jānĩ, jangal kā ghasyāri ai jānĩ, pāni kā panār ai jānĩ.

The little bird of the nest is sitting in the nest, the cow and the calves have
been tied up, the nine hundred thousand stars have opened up, the grass-cutting
girl has come from the forest, the drawers of water have come from the water.

And each of these singers had his own rhythm, his own melody for the sandhyā quite different from those of the others.

At the same time, an oral text like this is only realized in a unique situation: ethnographic information about it and around it are a necessary part of the philology of the text. This is a song sung at (and about) nightfall, a sacred time when rural families do in fact come back together from the day in the fields, in the forests, or at a job, just as birds go back to their nests, just as the grazing cows come back to be tied up in the cow-basement. In this song of crepuscular invitation, the gods are being asked to enter this human house at this moment of daily reuniting. The text is, in fact, the part of the ritual that indexically centers this spot, where the ritual is taking place, in a transformed landscape, now to be identified with the gods of regional mythology. The different realizations of this theme are capitalizing on this lived fact of Kumaoni rural life, and perhaps on a fantasy of idealized hill life; they are conveying it in beautiful rhythmic language; and they are doing so to specific transformative effect.

In a case like this, a verbal object, a text, illuminates aspects of wider life; at the same time, there is no way to understand what the text itself is doing in its world without an ethnographic exposition of broader aspects of that world. To what extent is this true of any translation?


Unlike translators and publishers, who have to make a living proposing palatable equivalences, anthropologists have their own agenda. In the North American tradition, at least, this has involved the attempt to understand nonmodern, non-Western ways of living and speaking in a way that usually constitutes at least an implicit critique of the idea of the universality, naturalness, and correctness of our own. In terms of translation, anthropologists—those who are prepared to treat texts not as a model for, but as anchoring points in the movement of life we label society and culture—could offer a radically sourcist, radically contextualizing, and collaborative translation practice which seeks to open up a text and its world to the reader, rather than replacing it with an easy-to-assimilate rendering of what some intermediary decides is its relevant referential content (cf. Leavitt 2005).

Such revelation recalls the methods of the old philology: it requires going into the text in its specificity, but also into the specificity of its language and its generic tradition; it demands a contextualization of the text in its world—a world the existence which you become aware of because you sense its boundary effects on your own. Anthropologists are uniquely suited for working on the edges of languages and worlds, making the differences manifest, keeping the paradoxes sharp and alive. As Dell Hymes (1981) called for an anthropological philology, perhaps this would be a philological anthropology. For such a text-focused practice, translation would also be exegesis; exegesis requires ethnography, and ethnography requires collaboration.

Professional translators, it is clear, have to avoid monstrosities and “unreadable” renditions. But anthropologists need not fear monsters; they are uniquely positioned to respond to Ortega/Benveniste’s call for ugly translations. They are free to experiment with forms of “experience-close” transcription, so as to include aspects of orality and the particular circumstances of a performance, and translation, to try to bring worlds together. If this means having to deal with “English monstrosities” (McDowell 2000: 225), then so be it: we should seek not only “to acquire a sense of how natives hear their poetry” (ibid.: 229), but also to change ourselves and our readers so that what at first sight is a linguistic monstrosity can come to sound like poetry. This means a violation of the literary, and to a degree of the social, canons we have all internalized.

And this would not take place in a social void: there is a small but real public in today’s societies, if made up only of anthropologists’ students and colleagues—but it certainly goes beyond them—for trying to understand other worlds.


I am grateful to the organizers of and other participants in the “Translating Worlds” colloquium for their invitation and comments, and to the anonymous readers for HAU for their very helpful suggestions. Special thanks to Luke Fleming for his close reading and perspicacious advice. This paper is dedicated to the memory of Alton L. Becker (1932–2011). When not marked or cited from published translated works, translations are my own.


Asad, Talal. 1986. “The concept of cultural translation in British social anthropology.” In Writing culture: The politics and poetics of ethnography, edited by James Clifford and George E. Marcus, 141–64. Berkeley: University of California Press.

Bassnett, Susan. 2014. Translation studies. Fourth edition. London: Routledge.

Becker, A. L. 1995. Beyond translation: Essays toward a modern philology. Ann Arbor: University of Michigan Press.

Bellos, David. 2011. Is that a fish in your ear? London: Allen Lane.

Berman, Antoine. (1984) 1992. The experience of the foreign: Culture and translation in Romantic Germany. Translated by S. Heyvaert. Albany, NY: SUNY Press.

Black, Max. (1959) 1962. “Linguistic relativity: The views of Benjamin Lee Whorf.” In Models and metaphors, 244–57. Ithaca, NY: Cornell University Press.

Boas, Franz. 1889. “On alternating sounds.” American Anthropologist 2: 47–53.

———. 1900. “Sketch of the Kwakiutl language.” American Anthropologist (N.S.) 2: 708–21.

———. 1906. “Some philological aspects of anthropological research.” Science 23: 641–45.

———. 1911. “Introduction.” In Handbook of American Indian languages, edited by Franz Boas, 1–83. Washington, DC: Government Printing Office.

Boroditsky, Lera, Lauren A. Schmidt, and Webb Phillips. 2003. “Sex, syntax, and semantics.” In Language in mind, edited by Dedre Gentner and Susan Goldin-Meadow, 61–79. Cambridge, MA: MIT Press.

Brower, Reuben A. 1959. “Seven Agamemnons.” In On translation, edited by Reuben A. Brower, 173–95. Cambridge, MA: Harvard University Press.

Brown, Roger, and Eric H. Lenneberg. 1954. “A study in language and cognition.” Journal of Abnormal and Social Psychology 49: 454–62.

Bunzl, Matti. 2004. “Boas, Foucault, and the ‘native anthropologist’: Notes toward a Neo- Boasian anthropology.” American Anthropologist 106: 435–42.

Bynon, Theodora. 1966. “Leo Weisgerber’s four stages in linguistic analysis.” Man (N.S.) 1: 468–83.

Davidson, Donald. 1974. “On the very idea of a conceptual scheme.” Proceedings and Addresses of the American Philosophical Association 47: 5–20.

Demonet, Marie-Luce. 1992. Les voix du signe: Nature et origine du langage à la Renaissance (1480–1580). Paris: Honoré Champion.

Descartes, René. 1984. The philosophical writings. Translated by John Cottingham, Robert Stoothoff, and Dugald Murdoch. Cambridge: Cambridge University Press.

Dubois, Claude-Gilbert. 1970. Mythe et langage au seizième siècle. Bordeaux: Ducros.

Finnegan, Ruth. 1977. Oral poetry. Cambridge: Cambridge University Press.

Friedrich, Paul. 1969. “On the meaning of the Tarascan suffixes of space.” International Journal of American Linguistics, Memoir 23.

———. 1979. “Poetic language and the imagination.” In Language, context, and the imagination, 441–517. Stanford: Stanford University Press

———. 1986. The language parallax. Austin: University of Texas Press.

Gaborieau, Marc. 1975. “La transe rituelle dans l’Himalaya central: folie, avatar, méditation.” Purushartha 2: 147–72.

Geertz, Clifford. 1973. The interpretation of cultures. New York: Basic Books.

Gelli, Giovan Battista. (1551) 1976. “I caprici del Bottaio.” In Opere. Edited by Delmo Maestri. Third edition, 125–288. Turin: Union Tipografico-Editrice Torinese.

Hagège, Claude. 1985. L’homme de paroles. Paris: Fayard.

Handman, Courtney. 2007. “Access to the soul: Native language and authenticity in Papua New Guinea Bible translation.” In Consequences of contact: Language ideologies and sociocultural transformations in Pacific societies, edited by Miki Makihara and Bambi B. Schieffelin, 166–88. Oxford: Oxford University Press.

Horguelin, Paul A. 1996. Traducteurs français des XVIe et XVIIe siècles. Montreal: Linguatech.

Hüllen, Werner. 2001. “Reflections on language in the Renaissance.” In Language typology and language universals, Vol. I, edited by Martin Haspelmath, Ekkehard König, Wulf Oesterreicher, and Wolfgang Raible, 210–21. Berlin: Walter de Gruyter.

Humboldt, Wilhelm von. (1825) 1906. “Notice sur une grammaire japonaise imprimée à Mexico.” In Gesammelte Schriften, V. Edited by Albert Leitzmann, 237–47. Berlin: B. Behr.

Hymes, Dell. 1981. “In vain I tried to tell you”: Essays in Native American ethnopoetics. Philadelphia: University of Pennsylvania Press.

———. 1999. “Boas on the threshold of ethnopoetics.” In Theorizing the Americanist tradition, edited by Lisa Philips Valentine and Regna Darnell, 84–107. Toronto: University of Toronto Press.

Jakobson, Roman. 1959a. “Linguistic aspects of translation.” In On translation, edited by Reuben A. Brower, 232–38. Cambridge, MA: Harvard University Press.

———. 1959b. “Boas’ view of grammatical meaning.” In The anthropology of Franz Boas, edited by Walter Goldschmidt, 139–45. Washington, DC: American Anthropological Association.

———. 1960. “Linguistics and poetics.” In Style in language, edited by Thomas E. Sebeok, 350–77. Cambridge, MA: MIT Press.

Kalmar, Ivan. 1987. “The Völkerpsychologie of Lazarus and Steinthal and the modern concept of culture.” Journal of the History of Ideas 48: 671–90.

Konishi, Toshi. 1993. “The semantics of grammatical gender: A cross-cultural study.” Journal of Psycholinguistic Research 22: 519–34.

Ladborough, R. W. 1938. “Translations from the Ancients in seventeenth-century France.” Journal of the Warburg Institute 2 (2): 85–104.

Leavitt, John. 1997. “The language of the gods: Craft and inspiration in Central Himalayan ritual discourse.” In Poetry and prophecy, edited by John Leavitt, 129–68. Ann Arbor: University of Michigan Press.

———. 2005. “Thick translation: Three soundings.” In Language, culture, and the individual: A tribute to Paul Friedrich, edited by Catherine O’Neil, Mary Scoggin, and Paul Tuite, 79–108. Munich: LINCOM.

———. 2011. Linguistic relativities: Language diversity and modern thought. Cambridge: Cambridge University Press.

Lefevere, André. 1977. Translating literature: The German tradition from Luther to Rosenzweig. Assen: Van Gorcum.

Leibniz, Gottfried Wilhelm. (1697) 1916. “Unvorgreifliche Gedanken betreffend die Ausübung und Verbesserung der deutschen Sprache.” In Deutsche Schriften, I, edited by Walther Schmied-Kowarzik, 25–54. Leipzig: Felix Meiner.

Lord, Albert Bates. 1960. The singer of tales. Cambridge, MA: Harvard University Press.

Mason, Catharine. 2008. “Ethnographie de la poétique de la performance.” Cahiers de Littérature Orale 63–64: 261–94.

McDowell, John H. 2000. “Collaborative ethnopoetics: A view from the Sibundoy Valley.” In Translating Native Latin American verbal art, edited by Kay Sammons and Joel Sherzer, 211–32. Washington, DC: Smithsonian Institution Press.

Meissner, Konrad. 1985. Mälushähi and Räjulä: A ballad from Kumäün (India) as sung by Gopi Däs. Wiesbaden: Otto Harrassowitz.

O’Laughlin, Bridget. 1978. Review of Marshall Sahlins, Culture and practical reason. Dialectical Anthropology 3: 97–104.

Olson, Charles. (1950) 1997. “Projective verse.” In Collected prose, 239–49. Berkeley: University of California Press.

Ortega y Gasset, José. (1937) 2000. “The misery and splendor of translation.” Translated by Elizabeth Gamble Miller. In The translation studies reader, edited by Lawrence Venuti, 49–63. London: Routledge.

Pinker, Steven. 1994. The language instinct. New York: William Morrow.

Povinelli, Elizabeth. 2001. “Radical worlds: The anthropology of incommensurability and inconceivability.” Annual Review of Anthropology 30: 319–34.

Ricoeur, Paul. 1973. “The model of the text: Meaningful action considered as a text.” New Literary History 5: 91–117.

Robinson, Doug. (1996) 1997. “Translation as phantom limb.” In What is translation?, 11331. Kent, OH: Kent State University Press.

Rosenzweig, Franz. (1926) 1994. “Scripture and Luther.” In Scripture and translation, Martin Buber and Franz Rosenzweig. Translated by Lawrence Rosenwald and Everett Fox, 47–69. Bloomington: Indiana University Press.

Rothenberg, Jerome, ed. 1968. Technicians of the sacred. New York: Doubleday.

Rubel, Paula G., and Abraham Rosman, eds. 2003. Translating cultures. Oxford: Berg.

Sapir, Edward. 1921. Language. Boston: Houghton Mifflin.

Schleiermacher, Friedrich. (1813) 1992. “On the different methods of translating.” Translated by Waltraud Bartscht. In Theories of translation, edited by Rainer Schulte and John Biguenet, 36–54. Chicago: University of Chicago Press.

Sdun, Winfried. 1967. Probleme und Theorien des Übersetzens in Deutschland. Munich: Hueber.

Silverstein, Michael. (1981) 2001. “The limits of awareness.” In Linguistic anthropology: A reader, edited by Alessandro Duranti, 382–402. Malden, MA: Blackwell.

———. 2003. “Translation, transduction, transformation: Skating ‘glossando’ on thin semiotic ice.” In Translating cultures, edited by Paula G. Rubel and Abraham Rosman, 75–108. Oxford: Berg.

Silverstein, Michael, and Greg Urban. 1996. “The natural history of discourse.” In The natural history of discourse, edited by Michael Silverstein and Greg Urban, 1–17. Chicago: University of Chicago Press.

Speroni, Sperone. (1542) 2009. Dialogo delle lingue/Dialogue des langues. Translated by Gérard Genot and Paul Larivaille. Paris: Les Belles Lettres.

Stankiewicz, Edward. 1981. “The ‘genius’ of language in sixteenth-century linguistics.” In Logos semantikos, Vol. I, edited by Jürgen Trabant, 177–89. Madrid: Gredos.

Steiner, George. 1975. After Babel. London: Oxford University Press.

Tedlock, Dennis. 1983. The spoken word and the work of interpretation. Philadelphia: University of Pennsylvania Press.

Stocking, George W., Jr. 1968. Race, culture, and evolution: Essays in the history of anthropology. New York: Free Press.

Venuti, Lawrence. 1995. The translator’s invisibility. London: Routledge.

Webster, Anthony K. 2014. “In favor of sound: Linguistic relativity and Navajo poetry.” Paper presented to the Symposium about Language and Society—Austin (SALSA) XXII, Austin, Texas.

Mots et mondes: Ethnographie et théorie de la traduction

Résumé : Si les différences de langues orientent les locuteurs vers différents aspects de l’expérience, alors la traduction peut être conçue comme un passage entre différents mondes vécus. Cet article retrace les moments clés d’une histoire récente de la théorie de la traduction en Occident et défend l’idée que la traduction et l’ethnographie ont besoin l’une de l’autre. N’étant pas sujets à l’injonction qui requiert des traducteurs professionnels des textes facilement assimilables pour un public ciblé, les anthropologues sont particulièrement bien placés pour mener à bien des traductions qui rendent compte du contexte avec attention, et des ethnographies centrées sur des textes. De telles traductions « laides » (Ortega y Gasset) peuvent contraindre le lecteur à se réorienter lui-même, à franchir la frontière avec ce qui constitue potentiellement un autre monde, c’est à dire, initialement, un autre langage-monde. En considérant l’histoire, nous cherchons à distinguer la traduction de son contenu référentiel - ce qui constitue une ambition réalisable - et à traduire des éléments stylistiques et indexicaux (contextuels) - ce qui a souvent été déclaré impossible. Cet article explore les implications de ces arguments en s’appuyant sur des artéfacts textuels rédigés à partir de traditions orales de l’Himalaya central.

John LEAVITT teaches in the anthropology department of the Université de Montréal, specializing in linguistic anthropology. He has conducted field research in the Central Himalayas of northern India and in Ireland and has published on comparative mythology, oral poetry, and the history of linguistic relativity.

John Leavitt
Département d’anthropologie
Université de Montréal
C.P. 6128, Succursale Centre-Ville
Montréal, Québec
Canada H3C 3J7


1. Boas is the first to have used the word “culture” in the plural in English (Stocking 1968). The plural use in German (Kulturen) had been established in the late nineteenth century by Lazarus, Steinthal, and the school of Völkerpsychologie (Kalmar 1987).

2. I use the figure of ten thousand attested human languages. The figure ranges from five to ten thousand, depending on what is considered a language and what a dialect.

3. “Übersetzen heißt zwei Herren dienen. Also kann es Niemand. Also ist es, wie alles, was theoretisch besehen niemand kann, praktisch jedermanns Aufgabe.” Cf. Sapir’s statements that two languages could be incommensurable, while it was clear that he never questioned the possibility of translation. To be incommensurable means to be operating with different coordinate systems. While ultimate and exact congruence cannot be achieved in this situation, one can find equivalences that will serve all practical purposes (Leavitt 2011: 136–38).

4. For histories from within translation studies, see Berman ([1984] 1992); Venuti (1995); Bassnett (2014).

5. Going farther in this direction, some psychologists (e.g., Konishi 1993) have started to find clear sex-based connotations for differing grammatical genders. Most recently, Lera Boroditsky has run experiments on native speakers of German and Spanish that show that gender connotation remains even when they are speaking in English (Boroditsky, Schmidt, and Phillips 2003).

6. While it is justified with an analogy to energy transducers, the choice of the word “transduction” is translationally problematic, since some version of it is already the standard word for translation in Romance languages (e.g., tradução, traducción, traduction, traduzione).

7. The following abbreviations are used here for grammatical terms: N = noun, A = adjective, P = pronoun, V = verb, pp = postposition, ppl = present participle, m = masculine, f = feminine, s = singular, p = plural, dir. = direct, obl. = oblique, voc. = vocative, conj. = conjunctive verb form, pres. = present, past = past; numbers indicate first, second, or third person.

8. This is not always the case with oral poetry. The Veda, Inuit poetry, and Somali poetry (Finnegan 1977) were or are orally composed, then memorized word for word.

9. Gopi Das was recorded in 1970 by Marc Gaborieau, to whom I am grateful for granting me access to the audiotape. An epic sung by Gopi Das, and a long interview with the singer, can be found in Meissner (1985).