A Tale of Two Translation Theories (Relevance Translation etc.)

Journal of Translation, Volume 1, Number 2 (2005) 35 A Tale of Two Translation Theories David J. Weber1 David Weber joined SIL in 1969 and, with his wife Diana, began fieldwork among the Quechua of Huánuco in central Peru. He holds a B.A. in Mathematics from the University of California at Santa Cruz. and an M.A. and Ph.D. in Linguistics from UCLA. He and Diana have been involved in linguistic research, Bible translation, and educational development. David has written or collaborated on a number of books related to linguistics and literacy and has developed computer programs for morphology and dialect adaptation. As an International Linguistic Consultant with SIL, he teaches periodically at SIL courses in U.S. universities and most recently in Lima, Peru. Abstract This paper considers two theories of communication and the theories of translation shaped by them. First, the Code Model of communication is characterized and reasons are presented for why it is an inadequate and misleading theory of communication. Then Source-Meaning- Receptor theories of translation that were shaped by the Code Model are characterized and their inadequacies surveyed. Second, the Relevance Theory of communication is introduced. Then Gutt’s Relevance Theory based theory of translation is sketched and its dramatic implications for translation are surveyed. Finally, broader implications of the shift from a Code Model to a Relevance Theory perspective are considered, including implications for the praxis of translation, the checking of translations, and the training of translators. 1. Introduction 1.1 A personal note In the early 80s, when I was in graduate school, I began reading work by Dan Sperber and Deirdre Wilson. Their collaboration resulted in the 1986 publication of Relevance: Communication and Cognition. I got it hot off the press. When I finished reading it I felt that translation theory would have to be rewritten. Although I did not undertake the task, Relevance Theory (RT) changed the way I translated. The task of rewriting translation theory in terms of Relevance Theory was undertaken by Ernst-August Gutt, resulting in various publications, among them the 1991 publication of his Translation and Relevance: Cognition and Context. Despite the enthusiasm Gutt and I, as well as others, have for this theory and how it applies to translation, it has not gained much ground in SIL circles; most have simply ignored it, being content with the theory they were taught in their pre-field training. There are some who have become outright hostile to it. This article is my attempt to introduce Relevance Theory and its implications for translation against the background of previous theories. The major sections following this brief introduction are: 2. The Code Model and its inadequacies; 3. The Source-Meaning-Receptor theories of translation; 4. RT-lite: A sketch of Relevance Theory; 5. Implications of RT for translation; and 6. Broader implications: 99 Theses; 7. Postscript: Reform or revolution (followed by infobase information and References). 1 This article is based on a mini-series I posted to the BT list (Bible translation e-mail discussion group), which was in turn based on a lecture I presented at SIL’s Americas Area meeting in Waxhaw in 2001. Since then I have presented it in various places and received feedback from many people: translators (some who are theorists, others who are “in the trenches”), administrators, students, some who are interested because their spiritual lives depend on translations, others simply curious as to what all the flap is about. To all I am grateful; the mistakes, of course, remain my own. 36 Journal of Translation, Volume 1, Number 2 (2005) 1.2 Disclaimer about theories Let me begin by saying a few words about theories/models and how they relate to praxis. A theory/model is framed in terms of various elements, making claims about them, about how they are related, and so forth. It provides a framework for thinking about a domain and a vocabulary with which to discuss it. But most significantly, a theory/model shapes how people think about the domain. As people work—having some theory in the back of their minds—they generally use the good sense with which God has blessed them to go beyond what the theory says. For example, the Code Model does not have a place for context. In spite of this, people who have thought about communication in terms of this model have recognized the importance of context and adjusted their practice accordingly. But still the model influenced how they thought about context, because it remained external to the model. (This will be discussed later on.) Attempting to characterize commonalities across diverse theories that span decades and to which innumerable minds have contributed requires a large measure of idealization or, to put it less politely, caricature. My caricatures are intended not to demean but to help us understand such matters as what we think, why we have come to think it, and the consequences of thinking it. (And I acknowledge that my caricatures are shaped more by my experience with the meaning-based theories of North America than by some more functionally oriented approaches originating in Europe.) I expect objections and I will welcome corrections and clarifications. These will be useful if they come with evidence (especially if accompanied by citation) and argumentation showing how a claim does or does not follow from a theory. But arguments like “Most Theory-X proponents believe Y so Theory X implies Y” and “I espouse Theory X and I never thought Y, so Y is not part of Theory X” will not be very helpful. To determine whether Y is a consequence of Theory X or the result of human intelligence we must consider the relationship between claim Y and Theory X without a human intermediary, for as noted above, people generally apply good sense and go beyond their theories. Valid criticisms of theories must be based either on internal problems (inconsistencies, contradictions) or on a failure to adequately relate the theory to the phenomena for which it should account: the theory fails to explain phenomena that fall in the domain of the theory, it fails to predict phenomena in its domain, it makes false predictions, and so forth. And, of course, such criticisms must be accompanied by evidence.2 1.3 Abbreviations The following abbreviations are used: BT Bible translation e-mail discussion group CM Code Model of communication (conduit metaphor) RL receptor language RT Relevance Theory SL source language SMR Source-Meaning-Receptor theory/theories of translation 2 I am indebted to Ernst-August Gutt for this last paragraph. A Tale of Two Translation Theories 37 2. The Code Model and its inadequacies Communication is a text-mediated relation between a speaker’s meaning and a hearer’s meaning. The Code Model (CM) of communication assumes that meanings are transmitted across a channel from speaker to hearer.3 This assumption is called the conduit metaphor. More precisely, the speaker encodes the meaning to be communicated into a linguistic form (sound, sign, writing) using his or her lexicon and grammar. The form is transmitted over some channel such as speech or writing to the hearer, who then recovers the meaning by decoding the form, using his or her lexicon and grammar. MEANINGspeaker — encode →FORM— transmit →FORM— decode →MEANINGhearer Or more simply: MEANING— encode →FORM— decode →MEANING Decoding is—at least to some extent—the opposite of encoding. Accurate communication depends on the identicalness (in all relevant respects) of the grammar and lexicon used for decoding and the grammar and lexicon used for encoding. For a tidy and fairly recent statement of the CM see Certo (1995, chapter 5). 2.1 Underwood’s assessment The CM has its roots in a 1948 paper by Claude Shannon (Shannon 1948, Shannon and Weaver 1949) in which the mathematical foundations of information transmission are laid. This model was appropriated as a model of human communication. For those with web access I recommend the overview at Mick Underwood’s website.4 The following is part of Underwood’s section criticizing the extension of transmission models to human communication: The Shannon and Weaver and Lasswell model are typical of so-called transmission models of communication. These two models also typically underlie many others in the American tradition of research, showing Source-Message/Channel-Receiver as the basic process of communication. In such models, communication is reduced to a question of transmitting information. Although transmission models have been highly influential in the study of human communication, it can be argued that, although Shannon’s and Weaver’s work was very fertile in fields such as information theory and cybernetics, it may actually be misleading in the study of human communication. Some criticisms that could be made of such models are: THE CONDUIT METAPHOR. Their model presents us with what has been called the ‘conduit metaphor’ of communication (Reddy 1979) The source puts ideas into words and sends the words to the receiver, who therefore receives the ideas. The whole notion of ‘sending’ and ‘receiving’ may be misleading, since, after all, once I’ve ‘sent’ a message, I still have it. The underlying metaphor is of putting objects into a container and sending them through some sort of conduit to the receiver who receives the containers and takes the objects out. The important question which is overlooked is: How do the ‘objects’ get into the ‘containers’? In other words, how do we succeed in putting meanings ‘into’ words and how does somebody else succeed in taking the meanings ‘out of’ words? Transmission models don’t deal with meaning.[5] 3 Throughout this article I intend such statements to apply broadly to speaker, author, signer—and hearer, reader, viewer, etc. 4 www.cultsock.ndirect.co.uk/MUHome/cshtml/introductory/sw.html. 5 In this model, meaning is sharply distinguished from information. 38 Journal of Translation, Volume 1, Number 2 (2005) It’s probably worth saying that that’s not really a criticism of them, since they weren’t intended to deal with meaning, but rather a criticism of their (mis)application to human-to-human communication.… CONTEXT. Communication differs very greatly according to the context in which it takes place. I might give my partner a kiss when I leave her for work in the morning; if I give you a kiss when I leave work to go home in the evening, the same sign is decoded radically differently. Context is frequently not considered in transmission models. INTENTIONS. What do I hope to achieve by talking to you? What do you hope to get out of listening to me? Our intentions fundamentally affect the exchange.[6] CHANNEL OR MEDIUM. The information models pay no attention to the effect of the channel used. If I want to communicate with you, do I get someone else to pass on my message, phone you, send you a letter, send you a memo or seek you out to talk to you directly? The choice makes a difference. Obviously, there are also differences in the features of different media which make them more or less appropriate for saying what we want to say.… RELATIONSHIP BETWEEN SENDER AND RECEIVER I am a teacher and you are my student. There is an imbalance of power there, which must have a considerable influence on the way we communicate. Even if I appear open to criticism, you’re likely to be wary of being critical of my teaching, whereas I am probably quite prepared to be critical of your learning. 2.2 General inadequacies The CM reflects how people thought about communication at the time it was developed and took hold. But it also shaped—and continues to shape—how people think about communication. I will now briefly mention some of the concepts involved (in no particular order): 1. Language is a machine for encoding and decoding. It is primarily structural, consisting of a grammar and a lexicon. 2. Communication is quite mechanical. A person with an operational understanding of a language’s grammar and lexicon can encode and decode meanings and thereby communicate in that language.7 In fact, it is so mechanical computers ought to be able to do it! 3. Meaning, once encoded, is in the text. The text determines the meaning decoded from it. 4. Interpretation is like unpacking a container. Lecercle (1999) calls it the “tin-opener” theory by analogy to the task of opening a tin can to remove the contents. He says (ibid:43) that the view encapsulated in the conduit metaphor has established itself as doxa (dogma). He goes on to discuss how this is “deployed” in seven “maxims” applied to interpretation. 5. Little of what constitutes our humanness figures in the model. Communicators are simply encoders and decoders, making use of grammars and lexicons. (This is because the CM was an inappropriate extension of Shannon’s 1948 model of information transfer, as stated in sec. 2.1.) 6. The CM does not incorporate context. It does not say how context might influence the form a speaker utters nor what a hearer might infer from an utterance in a given context. That is not to say that context was ignored when the CM was dominant, only that the model said nothing about it. When people thought about context, the CM led them to think of it primarily as the surrounding 6 Underwood’s point is that unfortunately there is no place for intentions in transmission models. 7 The teaching of foreign languages is only recently freeing itself from this way of thinking. A Tale of Two Translation Theories 39 text, the co-text, not of the mind-mediated notion of context that has so advanced our understanding of the relationship between form and meaning (utterance interpretation). More modern conceptions of communication differ from the Code Model in many respects, among them the following: 1. Language is much more than an encoding-decoding machine. Little about human communication is mechanical. The role of grammar has been vastly overestimated and the lexicon is only the tip of an iceberg that includes our encyclopedic knowledge, worldview, attitudes, and so forth. 2. Texts do not contain meanings; meanings are in the minds of communicators. Texts do not determine meanings; along with context they guide interpretation. Interpretation is not like opening a tin and removing sardines. And context is absolutely crucial; its role must be accounted for in a model of communication. 2.3 Specific inadequacies of the CM This said, let me catalog some of the misleading features of the CM. It should be clear that these result from the absence of context-based inference, and from underestimating the insubstantial (reflecting the bias that something is better than nothing). I will conclude with what I regard as the most significant: the CM’s failure to account for nonliteral meanings. 2.3.1 The CM’s view that silence can only mean nothing According to the CM, if nothing is encoded, transmitted, and decoded, then nothing should be communicated. Silence can only mean nothing. But think of the phenomenal expressive power of silence. By it we can communicate anger, assent, wonder, defeat, and disinterest. Aside from grammatically determined absences (null pronouns, certain types of ellipsis, and so forth), the interpretation of significant silence is directed entirely by the (mind-mediated) context. We need a model of communication that can account for this. You may feel that this argument is weak because significant silences are rather rare and exotic. Significant silences, however, pervade ordinary speech. Unfortunately the CM fosters the idea that something is better than nothing. (This, by the way, is particularly evident in highly edited texts.) 2.3.2 Overuse of connectives Some people assume that an explicit connective is more effective than no connective. However, consider the following example (due to Mann and Thompson): I’m hungry. Let’s go to the Fuji Gardens. We readily infer (with some cultural background) that the Fuji Gardens must be a nearby restaurant at which the speaker can acquire food to satisfy his hunger. Putting anything between the two sentences (e.g., therefore or for that reason) would sound pedantic and interrupt the tight problem-solution relationship of the parts. The CM cannot account for this because it is a matter of inference, not a matter of text-driven grammar. 2.3.3 Rare use of ellipsis The CM inhibits the use of ellipsis because meaning should result from decoding form, not from the absence of form. But ellipsis is frequent in natural discourse. Yesterday I heard a woman giving advice for selecting tile for a kitchen floor. She said, “You want something porous, but not very.” I don’t think anyone was left in doubt about what she meant. 40 Journal of Translation, Volume 1, Number 2 (2005) On a radio program8 a U.S. senator was interviewed about the high price of medicines for senior citizens. He said (and I quote from memory), “We are trying to foster competition in the pharmaceutical industry to degrade drugs used by seniors.” He communicatedas he intended–“bring down the cost of drugs.” A CM decoding of what the Senator actually said would lead to a very different meaning. Ellipsis keeps readers engaged by requiring them to fill in what has been left out. For example, the translation of Matt. 23:25 in Huallaga Quechua begins, Pucucunapa lätucunapa janallanta mayllanqui; rurintaga mana ‘You wash only the outsides of cups and plates; inside not’. To understand ‘inside not’ the reader must reconstruct ‘You do not wash the insides of the cups and plates’. That would be hard for someone who was not paying attention, for someone who was simply listening passively. 2.3.4 The position of modifiers Modifiers generally occur adjacent to the word they modify so, for example, adjectives generally occur adjacent to the noun they modify (possibly separated by another modifier). However, many languages allow a modifier to occur at a distance from the modified word, as illustrated by the following pair of examples (provided by Marvin Beachy): After his sauna, Hiromi ate a fish raw. After his sauna, Hiromi ate a fish naked. In the first we understand that raw modifies fish; in the second, that naked modifies Hiromi. This is the result of inference, not of syntactic rule. 2.3.5 Pronoun resolution How does one determine the referent of a pronoun? Decades of intense research trying to give an account strictly in terms of syntax has led to an appreciation for the role inference plays in pronoun resolution. Consider example 1 and the alternative continuations in 2a and 2b: 1. As John and Paul were walking along, Paul stubbed his toe. 2a. “Ouch,” he said. 2b. “Boy, that must have hurt,” he said. If 2a follows 1 then he refers to Paul, but if 2b follows 1 then he refers to John. This you infer quickly, effortlessly, and unconsciously. You do not compute it by syntactic rule; indeed, no syntactic rule(s) could be written to accurately resolve such cases. I was watching a TV program, a biography of Isaiah Thomas, star player for the Detroit Pistons. The narrator told how Isaiah had helped his family, especially one of his brothers. Then the narrator said, “After two decades of fighting drug addiction, he persuaded his brother to enter a rehabilitation program.” For a mome