The space of translation

This work is licensed under the Creative Commons | © William F. Hanks. ISSN 2049-1115 (Online). DOI: http://dx.doi.org/10.14318/hau4.2.002

The space of translation

William F. HANKS, University of California, Berkeley

This paper explores the space of translation spanning cross-cultural description and the verbal act of rendering in one language what is expressed in another. We make a three-way distinction between translation as a method of revealing difference and similarity, cultural interpretation, which is related but distinct, and endogenous translation that takes place within a single language or culture. Intracultural translation plays a constitutive role in the social life of any human group, and not only in mediating between different groups and languages. This is evident in all varieties of reported speech, paraphrase, commentary, and exegesis. These share with translation two features that distinguish it from other kinds of interpretation: a translation both refers to and paraphrases its source text. It is the target language into which one translates that ultimately constrains the process. An adequate target language must be functionally capable of self-interpetation through metalanguage. Cross-linguistic translation presupposes intralinguistic translation. Historical examples of languages changing through intertranslation abound in (post)colonial contexts in which authoritative texts in a dominant language are translated into a subordinated language. This process inevitably alters the semantics and pragmatics of the subordinate language. The direction, scope, and depth of change are historically variable. Examples are adduced from modern and colonial Yucatec Maya and Spanish.

Keywords: reflexive language, metalanguage, comparison, ethnographic methods, linguistic anthropology


Among the perennial challenges facing anthropologists, the obdurate difficulty of cross-cultural translation occupies a special place. For an ethnographer attuned to the subtlety of native concepts, the task of translating into the language of anthropology can be daunting and seems inevitably distorting. British social anthropology in the 1960s and 1970s; structural, symbolic, and interpetive anthropology; the critique of ethnographic writing—all engage the inherent difficulty of translating native concepts into our writings. More recent works on comparative ontologies (Descola [2005] 2013), perspectivism, and translation itself have placed a renewed focus on cultural difference, and with it the severe challenges to translation. Vivieros da Castro’s “controlled equivocation” represents one interesting response to this challenge (see Hanks and Severi, this issue).

Running parallel to anthropological approaches, there is a large literature bearing on translation in philosophy, linguistics, and semiotics. In these fields, the problem is usually approached through fine-grained, often technical analysis of language. Thought experiments, isolated example sentences, and typologies are standard fare. The use of formal notations is already a process of translation, and the sheer complexity of human speech in its semantic, pragmatic, and sociolinguistic aspects poses formidable problems for the would-be translator (Hallen and Sodipo 1997). This literature has much to say to anthropology and a great deal to gain from it as well. It contributes much-needed distinctions and an unparalleled level of precision and explicitness. It shows, among other things, that translation is a constant and unavoidable part of any single culture, and not only a problem of comparison. As soon as we recognize that translation is a family of social practices, it becomes an object of study and not only a means to an anthropological end. As we try to understand a cultural world, what is the relation between our translations of “them” and “their” translations of themselves?

From social anthropology, analytic approaches can learn about actual social worlds (not only experimental ones, which are systematically less interesting). This implies a wholesale recasting of the typically oversimplified ideas of context invoked in analytic work. More pointedly, actual social formations differ and are alike in ways far beyond what linguists and philosophers typically recognize. Here the empirical commitments of social anthropology reveal orders of social and historical embedding that change how we think of the elements and levels of translational practice.

The goal of this paper is therefore to articulate linguistic and semiotic aspects of translation with social and historical aspects of it. It is a first attempt to chart a thick boundary between the two broad traditions and to highlight some of their convergences and divergences. Translation as method and as practice is both too broad and too fine-grained to be encompassed by either alone. The paper therefore explores the space of translation both in the broadly anthropological sense of cross-cultural description and in the more narrow sense of rendering in one language what is expressed in another. We distinguish translation as a method of revealing difference and similarity, present in anthropology at least since Boas, from related but distinct practices aimed at interpretation. Drawing on the semiotics of Roman Jakobson, C. S. Peirce, and Charles Morris, the paper argues that intracultural translation plays a constitutive role in the social life of any human group, and not only in mediating between different groups and languages. This is evident in all varieties of what Urban (2001) called metaculture, including reported speech, paraphrase, commentary, and exegesis. Such processes are a key part of meaning production and circulation, and are in this sense generative. The ones just listed share with translation two features that distinguish it from other kinds of interpretation and reproduction: a translation both refers to and paraphrases its source text. While source texts and signs may be more or less difficult to translate, therefore, it is the target language into which one translates that ultimately constrains the process. In order for a semiotic system to serve as a medium of translation, it must be functionally capable of self-interpetation through metalanguage. As a shorthand, we can say that cross-linguistic translation presupposes intralinguistic translation. Moreover, just as the latter generates novel meaning statements, the former is also generative. Historical examples of languages changing through intertranslation abound, but the clearest are found in colonial contexts in which authoritative texts in a dominant language are translated into a subordinated language, for this process inevitably alters the semantics and pragmatics of the subordinate language. The variety of translation this entails is what I will call commensuration, a neologistic process. Cross-linguistic translation is therefore a metalinguistic process that takes place in a space of asymmetric difference and produces change in either or both of the languages. The direction, scope, and depth of change are historically variable. Examples are adduced from modern and colonial Yucatec Maya, Spanish, and English.

Translation as method

Translation has long been used as a method in both linguistics and anthropology, and is arguably in play in any comparison across cultures or languages. A classic example of this is Saussure’s Course in general linguistics ([1916] 2006), a foundational text for modern linguistics and what would become structuralism in anthropology. Saussure demonstrates that the link between signifier and signified, a perceptible sign form and its associated concept, is arbitrary in any language. He does so by juxtaposing translations of “the same” idea in two or more languages, as in French mouton as a translation of English “sheep,” or “mutton.” Saussure’s point in such examples is that languages differ in how they pair meanings with forms, and from this it follows that the pairings in any one language are a matter of convention—not of natural necessity or similarity between sign and object. The signifier, itself an “image acoustique,” pairs not with a thing, but with a concept.1 Ultimately, Saussure’s translations demonstrate the near impossibility of accurate cross-language translation. A pessimist would conclude that translation is impossible.

Boas (1911) and Sapir (1949) make essentially the same use of translation as does Saussure, but they draw the stronger conclusion that cross-linguistic differences are both profound and consequential for the ways that speakers of different languages grasp the world around them. This would become the relativity thesis, which continues to generate debate in the linguistic and psycholinguistic literature. Certainly one motivation for their relativistic view was that Boas and Sapir did extensive research on native languages of the North America, which presented varieties of structure and meaning hitherto unprecedented in linguistics based on analysis of Indo-European languages. It was their position that the grammatical models inherited from European linguistics were simply inadequate to describe New World languages, which required description “in their own terms.”2 As is typical in modern linguistics, cross-linguistic translation is a heuristic indicator, but the load-bearing evidence for any analysis must come from the language itself.

Some of Boas’ most striking comments on translation involve demonstratives like English “this, that,” which in many North American Indian languages encode a distinction between visible and invisible (Boas 1940: 229), as in (1):

1. Visibility as a distinctive dimension in Kwakiutl; demonstratives

T’e’semgya “this stone (visible, near me)”

T’e’semgya’ “this stone (invisible, near me)”

A similar point is made repeatedly in Boas (1911). His larger point is that demonstratives in many North American languages distinguish visibility of the referent to the speaker at the time and place of utterance, and that the speakers of such languages are in effect forced to attend to the visual access that they have to the objects they refer to. Speakers of European languages, which do not so clearly encode visibility in the demonstratives, are not so obliged.

Perceptual access to situated referents is widely attested in the world’s languages. In order to translate the simple English forms “this” and “that” into the languages listed below, we are forced to pay attention to the perceptual features shown, since they correspond to different deictics in the target language. Conversely, if we translate a demonstrative expression from one of these languages into English, we are forced to annotate the English or lose the distinctions.

Perceptual access to referents: Visibility in North America (after Hanks 2011)

  • Quileute has distinctive deictics for Visible vs. Invisible objects. The Visible category is split into Proximal vs. Medial vs. Distal.
  • Kwakwa’la and Chinook have distinct deictic series for Visible vs. Invisible.
  • Crow distal deictics are split into Visible vs. Invisible.
  • W. Greenlandic has a special morpheme marking Invisible objects.
  • Ute makes a three-way distinction between Proximal vs. Distal vs. Invisible.
  • Maya (Yucatec) distinguishes Tactile, Visual, and Peripheral sensory access.

Such examples illustrate the peculiar status of translation as an instrument used ultimately to reveal not equivalence but difference-within-sameness between languages. As many would do after them, Boas and Saussure both use it more to contrast systems than to align them.

At a very different level of description, ethnographers have also used the method of translation as a way of revealing and making sense of difference, and, like Boas, the objective for anthropologists has usually been to make sense of the foreign language in its foreignness. For example, Evans-Pritchard ([1937] 1976: Appendix 1) is scrupulous to make his translations into English strictly accountable to the coherence of Azande concepts in their own cultural context, a strategy also pursued in his classic study of Nuer religion ([1956] 1970). As a result, the English glosses are purely heuristic. Talal Asad’s (1987) well-known discussion of cultural translation in British social anthropology starts from a similar position, but introduces power asymmetry between the source languages and the (European) target languages. Asad is ultimately concerned not with cultural or linguistic differences as such, but with power relations between languages—reminding us that translatability is not only a question of interlinguistic relation, but also one of power, authority, and legitimacy. We will see in a moment some effects of power difference, although neither the original nor the target language is necessarily dominant.

We might say that any time an ethnographer or linguist attempts to explicate the coherence or meaning of a concept from a distant culture in the language of anthropology, translation is the mediating process. This is so even if word-for-word translations are abandoned, because analysis or comparison themselves translate. The intuition of this paper is that translation so understood is not merely a problem of redescribing a cultural form, but of understanding it in the first place. In other words, it has to do with our ability to gain knowledge of other cultures, a point made forcefully by Severi (this issue).

Intralingual translation and understanding

Whatever the problems and prospects of translation as a method of cross-cultural comparison, it is also a pervasive part of social life in any single language or culture. Speakers of any language routinely translate themselves and others in the same language. Rumsey (this issue) makes strong use of this in his comparison of bilingual interactions with tom yaya kange performers in Ku Waru. Both translate, even though the performers are speaking solely in Ku Waru. In general, any time a speaker reports the speech of another, paraphrases, glosses, overtly imitates, or renders in “prose” register a text in poetic register (verse or vice versa), translation is in play. Once we introduce the sociolinguistic truism that all languages have multiple registers, it becomes clear that intralingual translation is not only a fact of social life, but is, in effect, a design feature of language.

The classic statement of intralingual translation is Jakobson’s ([1959] 2004) article on linguistic aspects of translation. Whereas the sometimes profound differences between languages have led some to suggest that accuracy is ultimately impossible, or at least vanishingly rare, Jakobson takes the opposite position: not only is translation a ubiquitous feature of ordinary monolingual speech, but the intralingual translation of an expression quite simply is its meaning. If this is so, then intralingual translation is incorrigible, because one cannot compare the target text to the meaning of an erstwhile independent source. Once we move from cross-linguistic to language-internal translation, this circularity becomes unavoidable. This is all the more striking if the interpretive process is institutionalized in such a way that some interpretations are considered authoritative, a fact common to textual traditions like Christianity (Durston 2007; Hanks 2010) and Islam (Messick 1996).

Jakobson makes a three-way distinction, between three varieties of translation: (1) traditional cross-language translation; (2) the intralingual translation that occurs every time one speaker paraphrases, reports, or even understands another; (3) the cross-modal translation of speech into gesture or vice versa. The third of these implicates a problem that has become increasingly focal in linguistic anthropology and is relevant to any anthropologist who examines the relations between multiple media. Jakobson treats all three as instances of what Peirce called the interpetants of a sign. Every sign or representamen consists of a perceptible sign vehicle, an object stood for in some respect, and an interpretant. As Peirce puts it, the sign “addresses somebody, that is, creates in the mind of that person an equivalent sign or perhaps a more developed sign. That sign which it creates I call the interpretant of the first sign” (1940: 99). In his extensive corpus of writings, Peirce discusses interpretance in many places, distinguishing among kinds of interpretants, and suggesting that different classes of signs call forth different types of interpretant. Of particular relevance in the present context is his distinction between immediate interpretants, which manifest in the correct understanding of the sign, and dynamical interpretants, which are the direct result of the sign. Thus if I said to you, “I smell smoke, something’s burning,” the immediate interpretant would be your grasping the meaning of the utterance, and the dynamical interpretant might be the alarm you feel at the prospect of a fire, the gesture of sniffing or looking for the source of smoke, or calling for help. Both kinds of interpretant can be multiple and give rise, themselves, to further interpretants.

It is clear that if translation is equated with the process of generating interpretants, then it is at the very heart of understanding, and we can see why Jakobson says that translation is the meaning: to understand is to produce an immediate interpretant. Thus it is a crucial part of all semiotic processes, and not only those in which the first sign and the interpretant are in different languages. This way of formulating the question makes it self-evident that there are epistemological stakes in translation. How I translate Maya into English, or Maya culture into the language of anthropology, what I chose to compare it to, and so forth—all of these involve translating, and if the translations are inaccurate or full of spurious projections, then so is the knowledge they express. At the same time, to simply collapse translation into interpretance is far too general. Peirce never requires that an interpretant bear a specific relation of similarity to the sign it interprets—any further propagation of signs will do. The interpretant need not even overlap in reference with the first sign, as in (2):

2. A telephone exchange

A:   “Hi. Is Ben home?”

B:   “You’ve got the wrong number”

B’:   “He should be back in 5 minutes”

It seems to me unhelpful to say that B’s response is a translation of A’s question, but both B and B’ are perfectly good interpretants of it. Similarly, if some third person C is with B when the call occurs, and interupts B’s response, B can raise his hand, palm out, to signal “Please be quiet” or perhaps “Please stop.” This gesture is a fine dynamical interpretant to C’s interruption, but it is no translation of it. On the other hand, if in making the gesture B says “Please be quiet” or “Shh!” then the utterance and the gesture are arguably in a cross-modal translation relation.

Therefore only some interpretants are translations. We need constraints. In this paper I will follow Nelson Goodman (1978) in stipulating that one representation is a translation of another if (and only if) it both refers to and paraphrases the other. The reference constraint captures the fact that the translation stands for the source, which is its first object in the Peircean sense. Failing this, we may have two similar statements, but neither translates the other. The paraphrase constraint captures the fact that there must be some relation of similarity, analogy, or partial equivalence between source and target. This raises the key question of evaluation, which in turn requires that we be clear on the purpose of translation. It is one thing if the purpose is to capture truth-functional meaning, but quite another if one wants the translation to “paraphrase” style, tone, speech act force, and so forth.3 It is only relative to a frame of reference that partial equivalence can be judged.

Linguistic resources for translation

Any human language can be used as its own metalanguage. That is, one can define English words in English, French words in French, Maya words in Maya. Any competent speaker is capable of asking an addressee “What does that mean?” or its near equivalent. Similarly, anyone can in principle paraphrase their own speech in response to such a question. The monolingual dictionary or grammar is based on this, but the phenomenon is much more widespread in ordinary talk, regardless of whether the language is written. In the clearest cases, metalinguistic discourse refers to, and therefore objectifies, language, its parts, or its products in an utterance or text. The distinction between object language and metalanguage, and the recognition that one and the same language can function in both modes, has a considerable history in linguistic thought. In this paper, most relevant is the line of thought leading from Morris (1971), to Jakobson (1957, [1959] 2004), Silverstein (1976), myself (Hanks 1983, 1990, 1993), and Urban (2001). The first two were concerned with distinguishing “thing sentences” from metalinguistic sentences in order to rectify language as a medium of analysis. Unrecognized metalanguage resulted in “pseudo thing sentences,” which, because they stipulate the meaning of the words in which they are stated, are marred by circularity. Drawing on Peirce, Jakobson integrated the distinction into his famous article “Shifters, verbal categories and the Russian verb” (1957), which was adapted by Silverstein in his “Shifters, verbal categories and cultural description” (1976). The term “shifter” refers to certain linguistic categories which have in common that one must attend to the utterance context in which they occur, in order to fix the reference. In Jakobson’s (1957) terms, they illustrate “message referring to code” or “code referring to message.” Thus “here, now, there, this, that” and all other deictics are shifters because their reference is strictly context-sensitive; they illustrate code referring to message. For both Jakobson and Silverstein, this obligatory anchoring of meaning in utterance context illustrates the “metalinguistic function” of language, a topic I have examined in depth elsewhere (Hanks 1983, 1990, 1993, 1996, 2011). Urban (2001) argues that metasemiosis, encompassing both kinds, is among the most potent forces for the circulation of cultural form.4

This line of thought, then, distinguishes (1) metalinguistic function (as in deixis, where immediate interpretation of shifters requires reference to the utterance context in which the sign occurs: “I like it here”) from (2) metalinguistic discourse (in which explicit reference is made to language, for the purpose of glossing, paraphrasing, etc.: “here” designates a place close to you when you say it) from (3) the general self-interpreting capacity of any human language (“We are a plurilingual nation,” “You should be polite when speaking to someone older than you”), and from (4) canonical translation between two languages (“Here” veut dire “ici”). One insight of the pragmatist tradition is that these four are all intimatey related, and in effect, all three grow out of the first. Our ability to translate between languages is grounded in our ability to translate within our language, and this is in turn rooted in the metalinguistic function that underlies much of ordinary referring.

In “Language and human nature,” Taylor (1985) identifies this self-interpreting capacity as a fundamental feature of any human society. Taylor notes that native expression and self-description have a constitutive role in the social realities they ostensibly describe. While it is clear that much of what he is refering to is not metalinguistic in any close sense, still it is reflexive in the sense suggested in point (3) above, and developed by Lucy (1993). The implication is that intralingual translation may be part of what actually shapes any language, just as self-description is part of what shapes any cultural order. This capacity to shape makes self-description consequential, even if it is distorting. According to Taylor, social description that is limited by the self-interpretations of native members thereby runs the risk of incorrigibility. The reason is clear: if the self-description helps define the facts, then the facts cannot be confronted with the description in order to refute it.

Any human language provides multiple resources that make possible this act of reflexive translation. These include, for example, the following:

  1. The ability to mention or cite language forms, without actually using them. For instance “‘Sign’ is a four-letter word.”
  2. The ability to distinguish actual speech from hypothetical speech. For instance, “If you say ‘I give you my word,’ then you’ve made a promise,” in which the if-clause is hypothetical and the then-clause is a metalinguistic claim by the speaker.
  3. Various ways of reporting speech (as in “Ben said ‘Go!’,” “Ben said to go,” “Ben told them to start,” etc.). This points toward the lexicon of verbs of speaking and the grammar of complements of speaking. Prosody, deixis, evidentials, and other features may shift systematically to maintain recoverability of the original speaker from the report.
  4. Deictic shift: “I like it here” becomes “Bill said he liked it there.”
  5. Prosodic shifts to distinguish quotation, as in mocking repetition of a speaker’s utterance.

Depending upon the language, there are numerous other resources for glossing, reporting, and commenting on speech in the same language. The main point is that languages are rich in these resources, which reflect the fact that intralingual translation is a design feature of human language.

This then opens up a second path to translation as method, because this reflexive capacity is a very powerful resource for any student of a language, whether a child native learner, a second-language learner, or a researcher. In my own work in Yucatec Maya, I conducted all of my fieldwork in Maya, and for this, the metalinguistic resources of the language were crucial.5

In addition to a wealth of evidence about how Maya speakers objectify their own speech practices, metalinguistic discussion revealed their common-sense assumptions about which uses are more typical than others. They even went so far as to reject as unacceptable expressions that they themselves used on other occasions (Hanks 1993, 1996). This underscores the point that native language translations do not always define the rigorously linguistic meaning of expressions—they may actually distort facts of observable usage. Evaluated as rigorous claims of meaning or descriptions of use, therefore, they can only give clues, since they are, as Boas (1911) put it, “secondary interpretations.” Yet as testaments of native common sense, they are primary evidence. And what they tell us is how native speakers typify usage. Like Jakobson’s ([1959] 2004) translations, they simply are the relevant meanings. But we can recognize this root circularity in common sense while retaining the ground to distinguish native typification from linguistic or anthropological analysis. The two refer to different orders of social fact and, in particular, they do so from different vantage points. One can retain the ability to evaluate the translation relative to its object for analytic purposes, while recognizing that the two are confounded if the frame of reference is native common sense.

Metalanguage and translation of deictic expressions

Boas’ frequent observations on the salience of perceptual access in the meanings of deictics in North American languages foreshadowed what has become a major area of cross-linguistic research.6 At first blush, expressions like “here” and “there” appear far removed from the concerns of most anthropologists, and they have been poorly analyzed by traditional linguistics. One reason for this is the apparent lack of meaning conveyed by such expressions, since “here” and “there” tell us little about the places or objects to which they refer when uttered. They seem to have none of the social importance one associates with more standard examples of categorization (like kinship terms) or interpersonal address forms (like honorifics and titles). Moreover, there is an obviousness to the notion of “here”—what is close to me when I say “here”—and this apparent transparency has shielded it from scrutiny. Over the last three decades of research however, it has become clear that these received assumptions are both false. In fact, deictic systems vary widely in unexpected ways (Hanks 2011), and this variation reveals a wealth of social information about how cultures organize interactive context. By studying kinship, ethnobotanical, or ethnomedical terminologies, for instance, one can learn a great deal about how different peoples construe the domains of human relatedness, plants, or the living body. By contrast, a lifetime studying demonstratives like English “this and that” reveals next to nothing about how people categorize the objects that such expressions denote when uttered. This is because indexical expressions differ fundamentally in the kinds of information they encode.

Whereas standard descriptive terms reveal properties of the object they designate, deictics, including demonstratives, are notoriously sparse on such information. They seem to be so abstract as to be mere representatives of richer categories. Compare “mother’s brother’s daughter” with “that one” or just “her.” Yet what the indexicals precisely code is just what is missing in the kinship term: the relations between the speaker and the addressee of the utterance, between addressee and referent, and between speaker and referent. It is precisely these interactive relations that are delicately coded in the structure and use of deictics. When one spells out the spatial deictics in any language, for instance, one taps into schemas not of the objects referred to, but of the interactional relations in which referring is performed, and the situated perspectives from which participants in talk have access to the objects, places, and persons to which they refer. Moreover, the corporeal field in which such elementary referring occurs is a key part of the meaning of deictics, whereas standard descriptive terms reveal little about the body unless corporeality is the designation. This basic difference is linked to the metalinguistic function of deixis and raises two questions for translation. First, what happens when speakers attempt to translate deictics into paraphrases? And second, under what conditions is it possible to translate deictics across languages?

In (3) I have asked Don Ponso, an adult monlingual Maya speaker, the meaning of kó’oten té’ela,’ glossable as “Come (right) here.” His response puts the form in a scenario in which the would-be speaker is sitting in his own house, and offers a seat to a visitor. In order to do this, the speaker stands up, and offers a chair that he has in hand, saying “Come (over) here.” The gloss captures the pragmatic force of the utterance as an invitation, and cites a cross-modal translation relation between the words and the gesture of offering a seat. Given the functional density of the deictic, Don Ponso immediately glosses the expression by situating it as an utterance in context, thus engaging intralingual and cross-modal translation in a metapragmatic statement.

3. … bey xan hú tz’ik tech ump’ée bá’al akutale’,,

   … also he’ll give you a thing to sit (on),

   kulíik’le’, kyáaik teche’, kó’oten té’ela’. Eskeh ump’ée bá’al

   he gets up, he says to you, “Come here.” It’s that a thing

   umachk utz’áa teche’. Ká kulakechi.

   he takes in hand to give you. For you to sit on it. [1.A.25]

(3) also suggests that the form té’ela’ is used in addressing an addressee who is already close at hand and face to face with the speaker. This condition, which turned out to be pretty accurate, was explicitly stated by Don Ponso less than a minute later, as shown in (4).

4.  kó’oten té’ela’, k’abéet nàatz’ yàan techi’

    “Come (right) here” he has to be close to you (already). [1.A.061]

So how would a Yucatec speaker call an addressee who is at some remove, such as one on the other side of the market, in another part of the household, or out of sight in the woods? The answer is: kó’oten waye,’ “Come here.” The form waye’ is another kind of “here.” It designates an egocentric space around the Speaker, and to comply with this imperative, the Addressee need only follow the voice to its source. Don Ponso has tersely translated into words an unstated pragmatic constraint on typical usage of té’ela’: the interlocutors are already close, and, as illustrated in (3), the Speaker shows the referent to the Addressee.

(5) shows another example in which Don Ponso makes explicit background assumptions about context. I have asked him to comment on hé’eló’oba,’ “here they are” (a presentative, predicative deictic similar to French voici).

5.  hé’eló’oba’, amachmah [. . .] wá má amachma e’,

   “Here they are,” you’ve grabbed them [. . .] if you haven’t grabbed them,

   hé’ yàan hé’elo’.

   (you say) “There they are” [19.B.094]

Native metalinguistic glosses like these are an invaluable part of fieldwork on deixis in the language. By working in the native language, taking full advantage of the metalinguistic capacity of the language and its speakers, we can create a snowball effect whereby speakers’ commentaries reveal other features of the language, as well as translating into words their judgments of typical and proper usage. When we explore speakers’ typifications of usage, we are exploring their common-sense pragmatic schemas. It is on the basis of these that they can translate from pragmatic presuppositions to overt statements. We might say that any native speaker of a language is an interpreter of her or his language.7

I propose that it is precisely this native capacity for self-translation in any human language that underlies the ability to engage in cross-linguistic translation. In order to translate into a second language, that language must be self-interpreting. Any sign or collection of signs can be translated, but not any language can translate in the strong sense. The same semiotic and linguistic resources that permit self-interpretation in Maya, English, or Spanish are what make it possible to translate between the three.

One corollary of this is that, if we found a semiotic code incapable of self-interpretation, we would have a code into which translation would be severely limited, if not impossible. An example of this might be the well-known signs produced by bees, which are remarkably precise in indexing the direction and distance of pollen sources. Such signs can be approximately glossed into a human language, but no human signs can be translated into the gestural signs of bees, for these are not self-interpreting.

It is therefore not so much the source text that places limits on translatability, although this might pose specific problems. It is the target language that must meet the baseline requirement of the metalinguistic function of self-interpretation. Failing this, one cannot translate into the language.

As both Peirce and Jakobson observed, this process of translation interpretance within human languages generates new signs, new distinctions, and new ways of evaluating speech. The interpretant was proposed by Peirce to explain how signs beget signs, and thus the study of interpretance was, for him, part of rhetoric. In other words, irrespective of its ultimate “accuracy,” intralingual translation is productive and plays a crucial role in the social life of any language.

Does cross-language translation also generate new meaning and new usage, or is it merely a matter of accurately relaying meaning? History shows that it can be generative.

Translation and meaning production

Theories of translation differ in terms of the emphasis they place on the source text or the target text. Benjamin, for instance, judges the best translation to be the one that is most under the sway of the original (see Sammons and Sherzer 2000), while others have called for the transformation of the original according to the norms of expression in the target language. (Pym 2010 gives a useful discussion of both positions, and compare Mounin 1963.) The difference is essentially a matter of fixing the frame of reference for evaluation. From my perspective what is most important is that both ways of evaluating—privileging source over target or vice versa—assume a binary relation (source, target) and proceed by comparing the one to the other. The more different the two languages, the harder it is to balance fidelity to source with fidelity to target.

When we examine historical cases of translation, however, this picture shifts in a subtle but important way (Santamaría 1992; Durston 2007; Whalen 2003) . The target language may be altered in the process of translation. It may be incremented by neologisms, newly coined uses for existing forms, proper names, or portions of the source text left untranslated in the target. This is evident in all missionary translation as well as in literature like the usage manuals in products sold on the international market.

The importance of this fact is that the translation is no longer a simple binary relation between, say, Spanish and Maya. Rather, it becomes a three-part relation between Spanish, Maya, and the neologized version of Maya, which we can call Maya*. The neologized Maya* has elements of both languages, and serves as a medium of exchange between them (Burkhardt 1989; Bricker 2002).

The relation here is similar to a currency system into which value from incommensurable domains (say, labor and cattle, or Christianity and Post-Classic Maya religion) can be converted and hence compared. It was simply impossible to translate theologically freighted Spanish terms like “bautismo” directly into Maya, but it was entirely possible to create a medium of semantic exchange in which to commensurate between the two languages.

A simple illustration of this is provided by the translations of “baptism” into Maya by colonial missionaries in Yucatán. As with virtually all of the theologically loaded language of the missionaries, the idea of baptism had no equivalent in Maya:

6.   From Bocabulario de maya than (Acuña 1993: 141)

Figure 01

Note in (6) that there were two quite different translations of the Spanish baptismo, one focusing on the way the sacrament is performed (enter water) and the other on the sacramental effect (second birth). It is also worth noting that “sacramento” is untranslated—which disambiguates the Maya “enter water.” The same phrase was used in ordinary Maya to describe “leaky,” as in a leaky roof.8

Moreover the Maya “twice birth” renders explicit the theological backing of Catholic baptism, namely that the person is reborn in Christ. The same expression “twice birth” is used in Maya* for the resurrection of Christ. Thus a Maya speaker learning that “twice born” translates bautismo immediately knows something that a Spanish speaker will only learn through exegesis, namely that baptism is a form of resurrection. There are many scores of examples like this in the colonial corpus, where neologized Maya* essentially renders explicit elements of the conceptual or theological backing left implicit in the corresponding Spanish terms (for many more examples, see Hanks 1988; Laughlin and Haviland 1988; Smith-Stark 2007; Knowlton 2010).

Examples like (6) are the product of a special kind of translation, which I call commensuration. The heart of the process lies in redescribing in grammatically correct Maya the objects or concepts stood for by the corresponding Spanish. The result is a generalized medium of semantic exchange in which the conceptual backing of the Spanish is paired with existing or newly formed signifiers in the Maya. In both cases, the resulting sign is a neologism.

Commensuration is a practical solution to the existential problem of incommensurability. When two languages or systems make distinctions sufficiently different as to make it impossible to intertranslate directly, then one translates via neologism and periphrastic description. Ultimately in a case like Maya, the neologos would bloom into a register that would in turn spread into the official discourse of the Maya republics.9 By the late colonial period, Maya* is in use in all genres of writing by native Maya writers (Hanks 2010).

Commensuration relies on precisely the same metalinguistic capacity as we saw in Don Ponso’s pithy translations of deictic utterances into descriptive statements. Don Ponso used Maya to refer to and paraphrase Maya. Four hundred years earlier, Franciscan missionaries and their Maya assistants used Maya* to refer to and paraphrase the Spanish of catechism and law. The difference is that in the colonial case, two cultural worlds are being commensurated for the purpose of exchanging meaning. The exchange was bidirectional, to be sure, but it was inevitably asymmetric, with power residing clearly in the European doctrine. As a result, it was Maya that underwent neologization under the pressure of Spanish, and not the other way around.

This may be a point on which Asad’s plea for power really strikes home. In cases of commensuration like this one, it is the subordinate language that is altered by neologism.10 By contrast, in ordinary intralingual metalanguage, speakers feel no need to create neologisms, because the gloss and what it glosses are in the same language and not divided by power asymmetry.

Under this account, what is special about commensuration is that it operates over incommensurable cultural worlds, and provides a “common denominator” by which to bring them into alignment. Ordinary intralingual glossing is more a matter of (partial) equivalence within a single lingua-cultural world. It renders as translation a process of interpretance already autochthonous to the language.

But if Spanish-Maya commensuration was improvisational, it was still highly constrained. The missionaries were translating sacred language that expressed Truth. Proper reference and Truth preservation were the sine qua non of adequate translation. Pernicious ambiguities or unwonted entailments in Maya* were a constant concern, and the translations were revised throughout the colonial period.

Judging by the entire colonial corpus, the missionaries were guided by five principles (or perhaps preferences):11

  1. Interpretance is the starting premise that for any expression in Spanish, it was in principle possible to find an adequate Maya interpretation.
  2. Economy dictated that translators use the minimum number of Maya roots to express the maximum number of distinct Spanish concepts. This was important for the register of Maya* to be learnable.
  3. Transparency dictated that translators craft Maya* neologisms whose morphosyntactic elements were clearly distinguishable and relatable to discrete aspects of the target meaning. “Enter water” for baptism is an example, as are “cast sin” for confession (sacrament of reconciliation), chochkeban “untie sin” for absolution, and so forth. Transparency required mastery of Maya grammar (especially verbal morphology, compounding, incorporation, transitivity in the verb, among others). Without such knowledge, economy would fail because different senses of the same root would be indistinguishable.
  4. Indexical grounding is the process whereby newly minted neologisms were bound to their canonical referents. Part of this process was the binding of the expressions into prayers and other texts, so that their meaning would be anchored in the cotextual elements.
  5. Beauty stipulates that, all other things being equal, an aesthetically pleasing translation was to be preferred because it would more effectively move the heart of the new Christians. The Franciscans displayed a preference for simplicity, directness, and “aptness” in translation. For instance, chochkeban, “untie sin,” is a very apt translation of absolution, because in canonical terms, absolution is the action whereby the priest unties sin.

One of the most striking features of commensuration in the colonial Maya case is that it altered the semantics of Maya language far beyond the confines of the missions. Just as translation was an ongoing process under revision, this transformation of Maya was a protracted historical process. Maya* was picked up by Maya speakers and writers, and became effectively the standard variety of the language. By the middle of the nineteenth century, this variety, Maya*, would become the language of rebellion against the colonial order under which it was born. This process, which I believe has analogues in many historical circumstances, raises a number of very productive questions, of which I mention two.

First, how did this appropriation occur? In broad outline, it took place through the local Maya governments, all of which functioned in Maya, and whose documentary archives are saturated with Maya*. Alphabetic writing was taught in the missions, and the local scribes were chosen and trained there.12 Maya* simply became the variety of Maya that was written in the colonial period. Being associated with the church and provincial governance, this variety also became the language of legitimacy and power. Catechism classes enforced verbatim repetition and regular prayer, thus further driving the neologos into the expressive habits of Mayas operating in the colonial world.13

The second question I want to raise is: What happens to a translation if the erstwhile source text is lost or otherwise “untied” from it? The missionaries always tied their labors back to the canonical texts, but for Christian Mayas, the Maya* versions of the prayers are effectively the originals. It is the Maya* version of the Our Father or the Credo that they repeated daily and from which lines were transposed into other genres. Regardless of the conditions under which the new variety was produced, as it became native, it ceased to be translation and became its own original. This process, which took about two centuries in Yucatán, signaled a veritable conversion of Maya language (Hanks 2013).

As a minor illustration of what was a pervasive spread of linguistic change, consider example (7), taken from one of the native histories known as Books of Chilam Balam. These books are usually taken to embody “classic” Maya, which may be true in some respects, but is thoroughly undercut by the ubiquitous presence of specifically Christian Maya* in the books (see Bricker 1989, 2002, 2007).14

Thus in (7), the references to hunab ku canal talane, “One God come from heaven” (7.1), to worship tuhahil auolah, “in the truth of your heart” (7.4), the monumental belief of oces tauol, literally, “cause it to enter your heart” (line 7.6), are all neologisms found in the catechism. There is not one single morpheme of Spanish in this passage, nor is it presented as a translation, but it is a recognizable commensuration in which the semantic backing is Spanish and the linguistic forms are Maya*. It is the product of translation, but in which there is no reference to the source, and therefore illustrates the ongoing conversion of Maya into Maya*.

Observe that this passage also displays exacting metalanguage in which the prophet quotes his own speech, and then (in 7.7–7.8) refers to it and paraphrases it as “weeping speech” and “explanation.” The term tzol was also (and still is) the term for cross-linguistic translation in Maya, thus introducing a highly apt ambiguity.

7.   Quoted speech in Codice Perez, page 73, lines 22–30 (Miram and Miram 1988: 3:67)

Figure 02

This one example is barely the tip of an iceberg, but will suffice to show that Maya language was fully capable of intralingual glossing, and Maya speakers deployed this functional capacity with astuteness. In the Classic Maya inscriptions studied by epigraphers, there is a great deal of metalanguage as well, showing that this capacity is in no way a product of colonization. Signs have been identified for “write, speak, tell,” and “signs” of various kinds. Throughout the colonial period, the prevalence of translation combined with the specific practices of commensuration to transform both the language and the consciousness of its speakers.


In this brief paper I have sketched out a line of thinking in which translation is not only productive but at the heart of language as a social form, and society as the dynamic product of self-interpretation. From an anthropological perspective, the examples I have adduced may appear narrow in their linguistic detail, but the processes involved are anything but narrow. As a method of comparative anthropology, translation reveals difference as much as similarity. This is an advantage. The very difficulty of translating terms as apparently simple as demonstratives casts the ethnographer onto the shoals of difference and clarifies the task of understanding a culture or society in terms of its own values (see Laughlin and Haviland 1988). The point is not to fetishize difference, but to situate the analytic task in the tension between partial similarity and partial difference. This is arguably the most subtle and difficult task for ethnographic description, and one with a long history in our discipline. It raises questions of commensurability, incorrigibility, and relativity. How can we recognize the constitutive role of self-interpretation in social life while avoiding the pitfall of collapsing analysis into native self-description? Native actors, including ourselves, are neither unconscious of the conditions of their lives nor are those conditions transparently available to them. They are neither caught in the blindness of sheer relativity, nor able to disembed their experiences and consider them dispassionately from afar. Language is neither the final arbiter of these questions, nor is it the mere projection of linguists. What is called for is a balanced recognition of the large mid-range in which these questions become not absolutes but variably blended elements.

Sooner or later, any understanding of translation, however generalized, must come to grips with language, which happens to be among the most delicate and finely calibrated institutions in any human society. And while we may focus selectively on key terms or concepts, no language is a collection of words, and to treat it as such is to distort it beyond recognition. In its variation, its generativity, its reflexive capacity for self-objectification, its ubiquity in social life, and its plurifunctionality, language is among the central features of social life. Boas and the Americanists were surely correct in attempting to discern the specificities of language in relation to consciousness, but they surely overstated its uniqueness. The sheer frequency with which speakers gloss, paraphrase, report, and justify linguistic practices demonstrates the robustness of their awareness. It is a truism that native metalanguage is not to be confused with the results of full-scale linguistic analysis: to know a language is not to be able to push an analysis to its logical limits, any more than knowing how to ride a bike implies the ability to spell out the physics implied in doing so. But this observation is merely the first step toward pluralizing ways of knowing and the knowledge to which they give rise.

I have insisted that the basis of cross-linguistic or cross-cultural translation lies in the self-interpretation that inheres in being a native actor. It is true that certain forms of translation pose formidable challenges, such as religious concepts, basic premises like perspectivism among some Amazonian groups, or banal assumptions as to the relatedness of objects and events which appear, from outside, as unrelated. It is equally true that in occupying any social world, one is always in the business of translating, transposing perspectives, transforming the implicit into the explicit, commensurating over difference, and shifting the figure-ground relations to leave implicit what is elsewhere explicit. It is always possible to gloss, but almost never possible to produce exact equivalence. We should therefore conceive of a family of operations instead of a static or monolithic relation between the translation and that which it purports to translate. This family would include cross-linguistic translation under carefully reasoned criteria, explanation, interpretation, and description of one lingua-cultural formation in the terms of another. But we must not forget that these operations are at work within any culture, and it is for that reason that we can even conceive of their operation across cultures. When a native Maya speaker glosses an indexical expression into a description of the force, meaning, or consequences of uttering that expression, (s)he is performing a metalinguistic act of objectification and interpretation, of rendering explicit pragmatic schemas that are at play implicitly in all speech. It makes no difference that their claims can be shown to be partial or contingent. This explicitation is the great advantage of such glossing, and it goes on when children are taught to speak and adult speakers are called upon to justify or interpret utterances by themselves or others. It is the stuff of understanding, and it is part of what it means to be a native actor. This is all the more true in that most societies are multilingual, and issues of translation and commensuration are inseparable from the ability to act and interact. We need not rush to the exotic in order to see this problem; it is a matter of daily practice.

However formidable the challenges of translating between truly different, maybe even incommensurable realities, such as colonizing religions and the beliefs or practices of those subjected to them, history demonstrates that people do manage to commensurate, however skewed and partial the results. It is ironic that exact translation is virtually impossible, but varieties of practically serviceable translation are everywhere. Here arises the question of evaluative criteria, for the impossibility is judged relative to a set of criteria usually foreign to those of people who actually translate on a daily basis. (This problem is one of the main issues addressed in Sammons and Sherzer 2000.) One way to state the challenge is to say we need to disembed the practice of translation from the institutionally defined concept of the translator as expert and arbiter of equivalence. We need to resituate it in the ordinary practices of native actors. Sometimes, as in the colonial Maya case, commensuration gives rise to a new register in the target language (Maya in this case). In less dramatic cases, sociolinguistic differentiation, blending, and “code switching” are virtually inevitable results of social plurality. Translation ceases to be a binary relation between two languages, and becomes a triadic relation between two languages mediated by a neologistic register in one (or both) of the languages. A neologized register, like Maya reducido (refered to above as Maya*), emerges in the manner of a generalized medium of exchange between the two starting languages. It commensurates between disjoint spheres, the way a money system commensurates between disjoint spheres of value. This is a historical process with consequences far beyond the language as a semantic system. When such a register spreads and is adopted by the speakers of the “target” language, it cuts loose from the source languages, ceasing to be translation and emerging as its own original. Here it is justified to speak of linguistic conversion and to spotlight the generative consequences of translation, so long as we keep in mind that what is emerging is no mere linguistic system, but a universe of practices, ways of self-objectifying, and schemes of interpretation—in short, if not a new world, then new ways of being in the world.


Acuña, René, ed. 1993. Bocabulario de maya than: Codex Vindobonensis N. S. 3833: facsimil y transcripción critica anotada. Mexico: Instituto Nacional Autónoma de México.

Agha, Asif. 2006. Language and social relations. Cambridge: Cambridge University Press.

Asad, Talal. 1987. “The concept of cultural translation in British anthropology.” In Writing culture: The poetics and politics of ethnography, edited by James Clifford and George E. Marcus, 141–64. Berkeley: University of California Press.

Benveniste, Émile. 1966. Problèmes de linguistique générale. Paris: Gallimard.

———. 1974. Problèmes de Linguistique Générale II. Paris: Gallimard.

———. 2012. Dernières leçons: Collège de France (1968–1969). Paris: Gallimard/Seuil/ EHESS.

Boas, Frans. 1911. “Introduction.” In Handbook of American Indian languages, Vol. 1, edited by Franz Boas, 1–83. Washington, DC: Smithsonian Institution.

———. 1940. Race, language and culture. Chicago: University of Chicago Press.

Bricker, Victoria R. 1989 . “The last gasp of Maya hieroglyphic writing in the Books of Chilam Balam of Chumayel and Chan Kan.” In Word and image in Maya culture: Explorations in language, writing, and representation, edited by William F. Hanks and Don S. Rice, 39–50. Salt Lake City: University of Utah Press.

———. 2002. An encounter of two worlds: The Book of Chilam Balam of Kaua. Tulane, LA: Middle American Research Institute.

———. 2007. “Literary continuities across the transformation from Maya hieroglyphic to alphabetic writing.” Proceedings of the American Philosophical Society 151 (1): 27–41.

Burkhart, Louise. 1989. The slippery earth, Nahua-Christian moral dialogue in sixteenth-century Mexico. Tucson: University of Arizona Press.

Canger, Una. 1997. “El arte de Horacio Caroche.” In La descripción de las Lenguas Amerindias en la época colonial, edited by Klaus Zimmermann, 59–74. Frankfurt: Verfuert Verlag.

Chuchiak, John. 2000. “The Indian Inquisition and the extirpation of idolatry: The process of punishment in the Provisorato de Indios of the diocese of Yucatán, 1563–1812.” Dissertation, Tulane University, UMI Dissertation Services, Ann Arbor, MI.

Descola, Philippe. (2005) 2013. Beyond nature and culture. Translated by Janet Lloyd. Chicago: University of Chicago Press.

Durston, Alan. 2007. Pastoral Quechua: The history of Christian translation in colonial Peru, 1550–1650. Notre Dame, IN: University of Notre Dame Press.

Edmonson, Munro S. 1970. “Metáfora maya en literatura y en arte.” In Verhandlungen des XXXVIII Internationalen Amerikanistenkongresses, Vol. 2, 37–50. Stuttgart/Munich.

Edmonson, Munro S. 1973. “Semantic universals and particulars in Quiche.” in Meaning in Mayan languages: Ethnolinguistic studies, edited by by Munro S. Edmonson, 235–46. The Hague: Mouton.

Evans-Pritchard, E. E. (1937) 1976. Witchcraft, oracles and magic among the Azande. London: Clarendon Press.

———. (1956) 1970. Nuer religion. Oxford: Clarendon Press.

Goodman, Nelson. 1978. Ways of worldmaking. Indianapolis, IN: Hackett.

Gossen, Gary. 1985. “Tzotzil literature.” In Supplement to the handbook of Middle American Indians, Vol. 3, edited by Munro S. Edmonson, 65–106. Austin: University of Texas Press.

Hallen, Barry, and J. Olubi Sodipo. 1997. Knowledge, belief and witchcraft: Analytic experiments in African philosophy. Stanford: Stanford University Press.

Hanks, William F. 1983. “Deixis and the organization of interactive context.” Ph.D. dissertation, Department of Linguistics and Department of Anthropology, University of Chicago.

———. 1988. “Grammar style and meaning in a Maya manuscript.” International Journal of American Linguistics 54 (3): 331–64.

———. 1990. Referential practice, language and lived space among the Maya. Chicago: University of Chicago Press.

———. 1993. “Metalanguage and pragmatics of deixis.” In Reflexive language: Reported speech and metapragmatics, edited by John Lucy, 127–58. Cambridge: Cambridge University Press.

———. 1996. “Language form and communicative practices.” In Rethinking relativity, edited by John J. Gumperz and Stephen C. Levinson, 232–70. Cambridge: Cambridge University Press.

———. 2010. Converting words: Maya in the age of the cross. Berkeley: University of California Press.

———. 2011. “Deixis and indexicality.” In Handbook of pragmatics, Vol. 1, edited by Wolfram Bublitz and Neal R. Norrick, 313–46. Berlin: Mouton de Gruyter.

———. 2013. “Language in Christian conversion.” In A companion to the anthropology of religion, edited by Janice Boddy and Michael Lambek, 387–406. Oxford: Wiley.

Jakobson, Roman. 1957. “Shifters, verbal categories and the Russian verb.” Department of Slavic Languages and Literatures, Harvard University.

———. (1959) 2004. “On linguistic aspects of translation.” In The translation studies reader, edited by Lawrence Venuti, 138–43. Second edition. London: Routledge.

Knowlton, Timothy W. 2010. Maya creation myths: Words and worlds of the Chilam Balam. Boulder: University Press of Colorado.

Laughlin, Robert M., and John B. Haviland. 1988. The great Tzotzil dictionary of Santo Domingo Zinacantán, with grammatical analysis and historical commentary. Volume I: Tzotzil-English. Washington, DC: Smithsonian Institution Press.

Lucy, John, ed. 1993. Reflexive language: Reported speech and metapragmatics. Cambridge: Cambridge University Press.

Messick, Brinkley. 1996. The calligraphic state: Textual domination and history in a Muslim Society. Berkeley: University of California Press.

Meyers-Scotton, Carol. 2003. “What lies beneath: Split (mixed) languages as contact phenomena.” In The mixed language debate: Theoretical and empirical advances, edited by Yaron Matras and Peter Bakker, 73–106. Berlin: Mouton de Gruyter.

Miram, Helga-Maria, and Wolfgang Miram. 1988. Konkordanz der Chilam Balames, Vols. 1–6, Transcriptions Vols. 1–4. Hamburg: Toto-Verlag.

Morris, Charles. 1971. Writings on the general theory of signs. The Hague: Mouton.

Mounin, Georges. 1963. Problèmes théoriques de la traduction. Paris: Éditions Gallimard.

Okoshi Harada, Tsubasa. 2006 . “Kax (monte) y luum (tierra): La transformación de los espacios mayas en el siglo XVI.” In El mundo Maya: Miradas Japonesas, edited by Kazuyasu Ochiai, 85–104. Mexico: Universidad Naciónal Autónoma de México.

Peirce, Charles Sanders. 1940. “Logic as semiotic.” In Philosophical writings of Peirce. Edited by Justus Buchler, 98–119. New York: Dover.

Pym, Anthony. 2010. Exploring theories of translation. London: Routledge.

Quine, Willard Van Orman. 1960. Word and object. Cambridge, MA: MIT Press.

Rafael, Vicente L. 1993. Contracting colonialism: Translation and Christian conversion in Tagalog society under early Spanish rule. Durham, NC: Duke University Press.

Restall, Matthew. 1997. The Maya world, Yucatec culture and society, 1550–1850. Stanford: Stanford University Press.

Rumsey, Alan. 2008. “Confession, anger and cross-cultural articulation in Papua New Guinea.” Special issue, “Anthropology and the opacity of other minds,” edited by Alan Rumsey and Joel Robbins, Anthropological Quarterly 81 (2): 455–72.

Sammons, Kay, and Joel Sherzer. 2000. Translating native Latin American verbal art: Ethnopoetics and ethnography of speaking. Washington, DC: Smithsonian Institution Press.

Santamaría, Francisco. 1992. Diccionario de Mejicanismos. Fifth. Mexico: Editorial Porrua, SA.

Sapir, Edward. 1949. Selected writings of Edward Sapir. Edited by David G. Mandelbaum. Berkeley: University of California Press.

Saussure, Ferdinand de. (1916) 2006. Course in general linguistics. Translated by Roy Harris. Chicago: Open Court.

Schieffelin, Bambi B. 2007. “Found in translating: Reflexive language across time and texts in Bosavi, Papua New Guinea.” In Consequences of contact: Language ideologies and sociocultural transformations in Pacific Societies, edited by Miki Makihara and Bambi B. Schieffelin, 140–65. New York: Oxford University Press.

Sherzer, Joel. 1990. Verbal art in San Blas: Kuna culture through its discourse. Cambridge: Cambridge University Press.

Silverstein, Michael. 1976. “Shifters, verbal categories and cultural description.” In Meaning in anthropology, edited by Keith H. Basso and Henry A. Selby, 11–55. Albuquerque: University of New Mexico Press.

Smith-Stark, Thomas C. 2007. “Lexicography in New Spain (1492–1611).” In Missionary linguistics IV/Lingüística misionera IV. Lexicography. Selected papers from the Fifth International Conference on Missionary Linguistics. Mérida/Yucatán.

Taylor, Charles. 1985. “Language and human nature.” In Human agency and language: Philosophical papers, Vol. 1, 215–47. Cambridge: Cambridge University Press..

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

Thiemer-Sachse, Ursula. 1997. “El vocabulario castellano-zapoteco y el arte en lengua zapoteca de Juan de Córdova—Intenciones y resultados (perspectiva antropológica).” In La descripción de las lenguas Amerindias en la época colonial, edited by Klaus Zimmerman, 147–74. Frankfurt: Vervuert Verlag

Urban, Greg. 1991. A discourse-centered approach to culture: Native South American myths and rituals. Austin: University of Texas Press.

———. 2001. Metaculture: How culture moves through the world. Minneapolis: University of Minnesota Press.

Whalen, Gretchen. 2003. “Annotated translation of a colonial Yucatec manuscript: On religious and cosmological topics by a native author.” Final report to the Foundation for the Advancement of Mesoamerican Studies.

L’espace de la traduction

Résumé : Cet article explore l’espace de la traduction s’étendant de la description transculturelle à l’acte verbal consistant à restituer dans une langue ce qui a été dit dans une autre. Nous effectuons une distinction ternaire entre la traduction comme méthode révélant des différences et des similarités, l’interprétation culturelle, qui s’en rapproche tout en s’en distinguant, et la traduction endogène qui a lieu au sein d’une même langue ou culture. La traduction intraculturelle joue un rôle constitutif dans la vie sociale de tout groupe humain, et pas seulement en traduisant entre différents groupes et langages. Ceci apparaît clairement dans la diversité des langages rapportés, des paraphrases, des commentaires et des exégèses. Ceuxci partagent avec la traduction deux éléments qui les distinguent d’autres formes d’interprétation : une traduction se réfère et paraphrase son texte source. C’est la langue ciblée par la traduction qui contraint le processus. Une langue cible adéquate doit être fonctionnellement capable d’une auto-interprétation rendue possible par un métalangage. La traduction translinguistique présuppose la traduction intra-linguistique. Les exemples historiques de langages modifiés par l’inter-traduction abondent dans les contextes (post-)coloniaux dans lesquels des textes faisant autorité et produits dans une langue dominante sont traduits dans une langue subordonnée. Ce processus modifie inévitablement la sémantique et la pragmatique du langage subordonné. La direction, l’ampleur et la profondeur du changement sont variables historiquement. Dans cet article, des exemples tirés du maya du Yucatan moderne et colonial et de l’espagnol sont analysés.

William F. HANKS received the Joint Ph.D. in Linguistics and Anthropology from the University of Chicago in 1983 and currently teaches in the Department of Anthropology, University of California, Berkeley, where he is also Director of Social Science Matrix, a cross-disciplinary research institute. He investigates the relation between grammar and communicative practices, colonial history, and contemporary shamanism among the Maya of Yucatán, Mexico.

William F. Hanks
Berkeley Distinguished Chair in Linguistic Anthropology
Professor of Anthropology
Affiliated Professor of Linguistics
Department of Anthropology
Kroeber Hall 232
University of California
Berkeley 94720


1. Benveniste (1966, 1974) correctly critiques Saussure for equivocating in this demonstration between the idea associated with the signifier, and the referent to which the signifier-signified pair refers. He also points out that for the native speaker, the pairing of form and meaning does not appear as mere convention, but as natural. It is under the perspective of cross-linguistic translation that the arbitrariness of any single language becomes visible.

2. In critiques of the relativity hypothesis, it is sometimes assumed that if it were accurate, then it should be impossible to learn or even understand foreign languages. But this reductio is absurd and finds no basis in the writings of either Boas, Sapir, or Benjamin Lee Whorf, all of whom also argued that languages have unversal properties. Boas and Sapir were polyglot and neither was naïve about the ability of speakers to learn even very different foreign languages, nor about the prevalence of bi-or multilingualism. The point is rather that through cross-linguistic translation, one can glimpse the uniqueness of structures and meanings in different languages, but it is only in relation to its own grammatical system that any expression can be ultimately analyzed.

3. There is a large literature on the problem of translating style, much of it produced by Americanists grappling with the difficulties of translation Native American oral traditions into written English. For classic statements bearing on Mesoamerica and South America, see Tedlock (1983), Gossen (1985), Sherzer (1990), Urban (1991), and Sammons and Sherzer (2000), and compare Rafael (1993) on Spanish and Tagalog in the Philippines, Shieffelin (2007) on Kaluli (PNG), and Rumsey (2008) on Ku Waru (PNG).

4. It is important to distinguish the functional capacity for reference to language from the social authority to exercise that capacity. All human languages are functionally designed to serve as their own metalanguage. The right to exercise that function, however, is far from universal; it is distributed over social persons and contexts.

5. Like Evans-Pritchard ([1937] 1976, [1956] 1970) or indeed Boas (1911), my attitude as a researcher was for many years that glosses in Spanish or English from the Maya were purely heuristic and had no evidence value in my analyses. It was only after spending many years studying the colonial history of Maya language and culture that I came to see translation as a key topic for research.

6. In what follows, I will use the term “shifter,” “deictics,” and “deixis” as general terms to cover deictic adverbs (place, time), demonstratives (nominal deictics), and pronouns (participant deictics).

7. This is a point on which Quine’s justly famous experiment in “radical translation” positively distorts not only the fieldwork practices of linguists, but what it means to know a language. Quine (1960) stipulates that the imaginary linguist confronts a hitherto unknown language without the aid of an interpreter. But the monolingual native speaker is in fact always already an interpreter of her or his own language (see Hanks and Severi, this issue).

8. The same phrase, och ha, occurs in Classic Maya hieroglyphic inscriptions, with the meaning “to die.” It is uncertain whether the missionaries knew this, but it further amplifies the aptness of the gloss for baptism, which marks the end of one person and the rebirth of a new person.

9. I use the term “register” in the sense developed by Agha (2006): that is, a variety of the language that is recognizable as distinct, backed by a discourse, and in a determinate relation to the standard. There is a substantial literature in linguistics on the concept of “mixed languages,” reopening questions raised in pidginization and creolistics. See Meyers-Scotten (2003) and papers in the same volume.

10. Though see Santamaría (1992) for ample evidence of how Mexican Spanish was also altered through its intermingling with indigenous languages.

11. I am summarizing a large body of evidence spelled out in Hanks (2010), where the concept of commensuration is first proposed. Compare Canger (1997), Thiemer-Sachse (1997), and Smith-Stark (2007).

12. Writing introduces a whole new order of questions regarding translation. See Benveniste (2012), which combines Peircean interpretance with Saussurian semiotics. And for fascinating case studies in Native America, see Tedlock (1983) and Sammons and Sherzer (2000). The Maya case makes a very productive comparison with Quechua as described in depth by Durston (2007).

13. For detailed study of an analogous process among Quechua people of Peru, see Durston (2007). For further background on the Yucatec case, see Restall (1997), Chuchiak (2000), Okoshi Harada (2006), and Hanks (2010).

14. Compare Edmonson (1970, 1973) and Knowlton’s (2010) excellent study and translation of the creation myth portions of the Books of Chilam Balam.