by Vern Sheridan Poythress
[A version of this essay is to appear in the Westminster Theological Journal 67/2 (2005). An earlier version was presented at the 56th annual meeting of the Evangelical Theological Society, Marriott Rivercenter Hotel and Riverwalk Hotel, San Antonio, TX, at 8:30 a.m., Nov. 18, 2004. Also to appear in a multi-author book on translation, to be published by Crossway Books.]
The last three centuries have put orthodox interpreters under pressure to show rigor and objectivity in biblical interpretation. We experience such pressure from Roman Catholic interpretation, from historical-critical interpretation, from wildly subjective and fanciful readings of the Bible, and—not least—from comparisons with the rigor, objectivity, and exactitude in modern science. But often rigor increases only by subtly ignoring or minimizing messy complexities. So let us think about complexity and richness in meaning.
Divine meaning, the meaning of God the primary author, is particularly complex,1 but complexities abound even at the level of human authors and human readers. Theologians know that some of the main topics of Scripture display rich meaning. Think of the biblical material concerning the nature of God, the image of God, sin, Christology, and eschatology. Think also of various literary phenomena such as metaphor, narrative, and poetry. Think of the role of the Holy Spirit in enabling readers to appropriate the message of Scripture. All of these point to mystery, complexity, and ultimately uncontrollable richness. In contrast to this richness, exegesis in its technical forms faces some reductionistic temptations.
We may conveniently focus on the whole area of the nature of language. What view do we hold about the nature of language? What is the nature of meaning in language? Do we allow richness here or not? Our assumptions about language will clearly influence our approach to word meanings, sentence meanings, exegesis, and Bible translation. If we have an impoverished view of language, we are likely to have an impoverished view of the Bible as well. For example, if we think that language is designed only to communicate literal propositions, we will probably end up minimizing the functions of metaphor and allusions. If we think that language is designed only to talk about this world, we will be suspicious of God-talk as an allegedly improper use.
Origins of language
Our challenges increase because of some unhealthy pressures deriving from the surrounding culture. To begin with, evolutionary modes of thinking would like to trace language back to animal cries and calls. According to this kind of thinking, just as man has gradually ascended from the slime, human language has gradually ascended from grunts. Modern language, like modern human anatomy, finds its original essence in providing for survival. This mode of thinking naturally throws suspicion on all use of human language for nonmaterial goals. The most material and simplest meaning is the most basic. Talk about God obviously stretches, perhaps to the breaking point, the original functions of language.
By contrast, the Bible shows that human language from the beginning included the function of serving for communication between God and man (Gen 1:28-30; 2:16-18). Speech about God and speech from God does not represent a stretch, but a normal function of human language.
For example, God is the first and principal ruler over the world. Human beings created in the image of God become subordinate rulers. The creation of human beings according to the plan of God produces an analogical relation between God’s rulership and human rule. God is king in the supreme sense, while human kings mirror his rule on a subordinate level. To call God king is not “mere” metaphor, in the sense of being unreal. It affirms a real analogy between God and man. It involves a normal function for human language.
Moreover, it is plain from Scripture that God designed language in such a way that there can be multi-dimensional, complex, nuanced communication between God and man. God can tell stories, both fictional (parables) and nonfictional. He can expound and reason theologically, as in Romans, and he can express the full range of human emotions, as in the Psalms. The Bible contains propositional truth, but can express it either in prose or poetry. It contains both short sayings, as in Proverbs, and multi-generational histories, as in Genesis. The meaning of one sentence in Genesis coheres with the meanings in the whole narrative. Meaning is not reducible to pellet-sized isolated sentences that are thrown together at random.
For example, Genesis 12:2 says, “I will make of you a great nation.” The meaning of that sentence can be determined only if we know who “I” and “you” are, which depends on the immediate context. And the full meaning of “great nation” can be seen only as the promise begins to find fulfillment near the end of Genesis and into Exodus. And what do we do with a more loaded term like “blessing”? “I will bless you and make your name great, so that you will be a blessing,” Genesis says (Gen 12:2). This promise contrasts subtly with the earlier arrogant attempt at Babel where people desired to “make a name for ourselves” (Gen 11:4). And it resonates with the later instances of blessing that run all the way through the Old Testament and into the New. All this is fairly obvious to a reasonably skilled reader. But we must now ask whether modern theories of meaning are adequate to capture this richness.
So let us look at three technical tools that have blossomed in the twentieth century: symbolic logic, structural linguistics, and translation theory. All three contribute to understanding language, but at the same time, when clumsily used, threaten to reduce meaning to one dimension.
Symbolic logic
Reflection about logic goes all the way back to Aristotle. But formal symbolic logic blossomed in the late nineteenth and early twentieth century with the work of Gottlob Frege, Bertrand Russell, and others.2 Symbolic logic made more rigorous the idea of a valid proof. And it proves useful in uncovering logical fallacies in informal reasoning. But what of its limitations? For the most part, the use of mathematical logic requires that we begin with isolated sentences. This step already involves a reduction of the full richness of human communication as it occurs in long discourses and social interaction. It also requires that a sentence be isolated from its situational context. It then treats the sentence almost wholly in terms of its truth value.
Modern evangelicalism has rightly insisted on propositional revelation in Scripture in response to liberal and neo-orthodox reductions of revelation to religious feeling and personal encounter. But in the process, we must beware of the reverse problem of reducing the discourse of Scripture merely to its truth value. It does have truth value. But the meaning of a whole discourse or of one sentence within it includes more than the fact that it is either true or false. It is related in meaning to many other parts of Scripture; it asks for application in our lives; it has the power to transform our hearts; and so on.
Symbolic logic is so obviously reductive in its approach to meaning that perhaps we do not need so much to remind ourselves of its reductive character. So let us pass on to the second great area of advance, structural linguistics.3
Simplifications in structural linguistics
As with symbolic logic, so with linguistics: we need to appreciate the value of linguistics, but also become aware of built-in limitations. Human language is so complicated and multi-dimensional that simplifications had to be made in order to get structural linguistics started. But it is easy along the way, in the excitement of discovery, to forget those simplifications and to make exaggerated or one-sided claims about the implications.
In considering the development of structural linguistics, I will have to make some simplifications myself, and confine myself to some high points illustrating the trends.4 I focus particularly on the issue of how linguistics treats meaning.
Ferdinand de Saussure, 1906-11
Many consider that structural linguistics had its origin in the lectures of Ferdinand de Saussure in 1906-07, 1908-09, and 1910-11, which were later compiled into the book Course in General Linguistics.5 After some historical observations, Saussure introduced the main body of discussion by delineating the object of linguistics. Linguistics will study language (langue) as a system, instead of studying speech (parole). That is, it will study the systematic regularities common to all native speakers, rather than the particularities of every individual speech by every individual speaker. For example, the verb tenses in English, the present tense (“I go”), the past tense (“I went”), the future tense (“I will go”), and so on, are one of the systematic regularities. An utterance using the past tense of the tense of the verb “to go” (“He went to town yesterday.”) is a particularity, an instance of parole.
In the light of hindsight this famous move toward focusing on the system of language decisively contributed to the delineation of linguistics as a subject distinct from textual analysis and exegesis. But the advance came with a cost. Any reasonable approach to the meaning of a specific communication (parole) must take into account the speaker, the audience, and the circumstances, since all three affect the nuances of a particular speech or text. The meaning of a particular utterance naturally depends on the particular words and their meanings. But it is not simply a mechanical product of word meanings, but includes a complex particular texture that varies with circumstance. The expression, “The door is open,” could be a simple assertion of fact. But a detective could be implying that it is a significant clue to solving a mystery. An angry person could use the expression to indicate indirectly that the person he addresses should get out. Saussure deliberately cut off the variations in order to study “the system.”
Second, Saussure largely cut off the influence of syntagmatic context (that is, textual context) by focusing on word-meanings.6 Like the earlier moves, this one flattens out the complexity of meaning. In later discussion he added context back in with the distinction between syntagmatic and associative (or paradigmatic) relations.7 But the damage has been done, since the consideration of syntagms still relies on words as its starting point. In many ways this reduction is quite understandable, perhaps in some sense necessary, because words are stable in relation to the surrounding speech (parole), and one must start with some simplifications if one is to get linguistics off the ground.
Third, Saussure introduced a model for linguistic signs with three parts: the “sound-image” or signifier, the “concept” or signified, and the “sign” that consists of both parts together. For example, the word arbor in Latin associates the concept of tree with the sound-image of a sequence a+r+b+o+r.8 The meaning we may associate with the concept, while the form consists in the sound-image. This move makes sense as a way of defining more rigorously the distinction between form and meaning. But it introduces a subtle reductionism in the thinking about meaning. Children learning a language often learn the meanings of words through their occurrences in social situations where there is reference to a real-world object. Words for milk and soup, cat and dog come to have meaning through the help of occurrences of milk and soup and cats and dogs in the environment. In the long run, referential functions have an indispensable role in meaning. Saussure has left out reference, and settled on “concept,” which suggest a purely mental phenomenon. This restriction is once again understandable, given his earlier decision to focus on the language system. The language system does not directly refer to objects in the world in the same way that specific speakers refer to such objects in specific speeches (parole). But one can never understand meaning in its fullness if one leaves out reference.
The omission of reference offers an open door for later reductionisms, as one can see with the case of certain forms of structuralism in which language is treated as a closed system of signs that refer only to other signs. In the hands of certain practitioners, the “meaning” of any one particular text got reduced to the central truth that meaning is a function of system.
Saussure proposed still another reduction when he shifted from “meaning” to “value.” By “value” he means the significance that a particular unit has by virtue of its oppositions or contrasts to neighboring units. He says, “Language is a system of interdependent terms in which the value of each term results solely from the simultaneous presence of the others.”9 The word “solely” signals the reduction. One will thereby ignore both reference and the historical accumulation of potential for literary allusion to earlier occurrences of the same expression. The benefits of focusing on the system of oppositions are now well known and undeniable. But we should not conceal from ourselves that these benefits derive partly from ignoring intractable complexities in what is left out.
Leonard Bloomfield, 1933
A second milestone in the development of structural linguistics occurred with Leonard Bloomfield’s publication of Language in 1933.10 Like Saussure, Bloomfield considered the correlation between sound and meaning to be fundamental.11 And initially he introduced meaning in connection with life situations in which language is used to accomplish practical tasks. But simplifications entered in as he focused on the concerns of linguistics. For one thing, Bloomfield used a simple stimulus-response model for understanding human behavior.12 He states bluntly, “… in all sciences like linguistics, which observe some specific type of human activity, the worker must proceed exactly as if he held the materialistic view.”13
Though Bloomfield in his early discussion equated meaning with the entire situation in which an utterance occurs, he soon reduced the task to “constant and definite meaning” for any one form.14 This move—again an understandable and convenient simplification to facilitate early progress in linguistics—ignores the influence of context. Meaning is effectively reduced to the meaning of an expression that is independent of the larger context.
Noam Chomsky, 1957
As our next milestone we may conveniently take Noam Chomsky’s Syntactic Structures in 1957.15 Chomsky’s book laid the foundation for what came to be known as generative grammar. Together with the later work Aspects of the Theory of Syntax,16 this book had enormous influence on the direction of linguistic research, because of its appeal to rigor and formalization, and because of the impressive conclusion that certain simple types of formal grammar were provably inadequate for the complexities of natural language.
But rigor and formalization came, as usual, with a price. Chomsky stipulated that a language was “a set (finite or infinite) of sentences, each finite in length and constructed out of a finite set of elements.”17 This definition, which allows for a language to be subjected to a rigorous mathematically-based analysis of syntax, ignores the role of context, both the context of a situation and the context of a discourse in paragraphs and larger sections. It is a vast simplification, but unfortunately Chomsky did not overtly acknowledge how much it simplifies. In the next sentence after this definition, he simply declared that “All natural languages in their spoken or written form are languages in this sense, ….” We also hear hints that grammaticality is independent of meaning, which is true only as a first approximation.18 In the long run grammatical categories make sense only in the service of meaningful communication.
Chomsky also introduced the significant distinction between kernel sentences and nonkernel sentences.19 Kernel sentences are simple, active-voice sentences like The boy fed the dog. These sentences arise within Chomsky’s formalism by the application of phrase structure rules and obligatory transformations. Nonkernel sentences include passive sentences, such as “The dog was fed by the boy,” and derived expressions like “It was the boy who fed the dog.” One must also consider expressions like “The boy’s feeding the dog.” The sentence “I was reassured by the boy’s feeding the dog” derives from two distinct kernel sentences, namely “The boy fed the dog” and “It reassured me.” All complex sentences, as well as other sentence types that derive from two or more kernel sentences, arise from applying optional transformational rules to the original set of kernel sentences.20
This schema opens the door to the possibility of a semantic analysis in which the meaning of a sentence is the sum of the meanings of the kernel sentences from which it is derived, plus the semantic relations between kernels that are specified by the grammatical links between them. Such an analysis is tempting precisely because in many cases it approximates the truth, and captures some of the core meaning or basic meaning that we associate with a sentence. But as a total account of meaning it is obviously reductionistic.
Linguistics has continued to develop since the Chomskyan revolution in 1957 and 1965. Chomsky’s generative grammar eventually mutated into the theory of government and binding, and then into the minimalist program.21 Though the detailed structure of the theories has changed markedly, the spirit of formalization and reductionism remains in place. But we also see challenges from competing theories. Cognitive linguistics with its meaning-centered approach challenges the grammar-centered approach of generative grammar and its successors.22 Other alternative linguistic theories continue to attract followers.23 Semantic theory has attracted continuing interest, sometimes without any strong dependence on a particular theory of grammar or phonology.24 The possibility of coherent alternative theoretical approaches suggests that any one approach is selective (and therefore potentially reductionistic) in its understanding.25
Translation Theory: Eugene Nida
In the twentieth century a theory of Bible translation developed in tandem with linguistics and tried to profit in a multitude of ways from the developments in structural linguistics.26 But while linguistics initially focused largely on issues of phonology and grammar, translation had to deal directly with meaning and all its complexities. Bible translators confronted the task of translating into thousands of third-world tribal languages. Eugene Nida, in consultation with other pioneers in the field, developed the theory of “dynamic equivalence” or “functional equivalence,” which stressed the importance of transferring meaning, not grammatical form.27 Nida discussed various kinds of complexity in meaning even at a comparatively early date, beginning with his 1947 publication of Bible Translating.28 He explicitly spoke about translating “fullest meaning” instead of a bare minimum.29
In 1964, Nida published the fuller and more theoretically advanced work, Toward a Science of Translating.30 By this time, he was aware of the formalistic approach in generative grammar, not only Chomsky’s Syntactic Structures, but also Katz and Fodor’s ground-breaking article, “The Structure of a Semantic Theory.”31 Nevertheless, in the first three chapters of his book he refused to be reductionistic. He spoke explicitly about many dimensions of meaning, and referred favorably to Roman Jakobson’s classification of meaning into emotive, conative, referential, poetic, phatic, and metalingual dimensions.32 He was so bold as to say:
… no word ever has precisely the same meaning twice, for each speech event is in a sense unique, involving participants who are constantly changing and referents which are never fixed. Bloomfield (1933, p. 407) describes this problem by saying that “every utterance of a speech form involves a minute semantic innovation.”33
And again:
In any discussion of communication and meaning, one must recognize at the start, each source and each receptor differs from all others, not only in the way the formal aspects of the language are handled, but also in the manner in which symbols are used to designate certain referents. If, as is obviously true, each person employs language on the basis of his background and no two individuals ever have precisely the same background, then it is also obvious that no two persons ever mean exactly the same thing by the use of the same language symbols. At the same time, however, there is an amazing degree of similarity in the use of language ….34
But Nida was also determined to use whatever insights he could obtain from Chomsky’s generative grammar. So in Chapter 4 he focused on what he called “linguistic meaning.” Here he looked at the meanings associated with distribution of a word within larger contexts and within grammatical structures.35 According to Nida, linguistic meaning often appears on “two levels”:
First, that meaning which is derived from the kernel construction by way of the transformations, and secondly that meaning which is supplied by the particular terminal construction (the end result in the process of transformation from the kernel to the resulting expression).36
In using the key terms kernel and transformations, Nida was clearly adopting the framework of Chomsky’s generative grammar. Meaning was now to be seen within this framework.37
The concentration on “linguistic meaning” involves a reduction. Nida was aware of this, and so in the following chapter he supplemented this account with a discussion of “referential and emotive meanings.”38 But someone less aware than Nida can easily use the schema reductionistically to think that all or almost all of the really significant meaning is linguistic meaning, and that this meaning comes to light exclusively through the Chomskyan framework. The temptation is all the stronger because Nida himself suggested that his scheme could serve as the basis for a translation procedure:
… it is most efficient for us to develop an approach to translation which takes these facts fully into consideration. Instead of attempting to set up transfers from one language to another by working out long series of equivalent formal structures which are presumably adequate to “translate” from one language into another, it is both scientifically and practically more efficient (1) to reduce the source text to its structurally simplest and most semantically evident kernels, (2) to transfer the meaning from source language to receptor language on a structurally simple level, and (3) to generate the stylistically and semantically equivalent expression in the receptor language.39
As Nida indicates in the surrounding discussion, an approach of this type looks promising particularly for languages whose formal (grammatical) structures do not match well with those of Indo-European languages such as English, German, Greek, and Latin. All languages show “remarkably similar kernel structures.”40 So if we can decompose meaning into these kernels, we can transfer it more easily from one language to another. In addition, the nonkernel structures do not necessarily reveal directly the underlying semantic relations. For example, the sentence “he hit the man with a stick”41 may mean either that he used the stick as an instrument, or that the man who received the blow had a stick in hand. Such ambiguous constructions often have to be translated differently depending on the underlying meaning. Nida therefore proposed a three-stage process, in which the first stage involves decomposition into underlying kernel meanings.
The three-stage process promises benefits. But it comes at the cost of leaving out much of the richness of meaning that Nida expounded in the immediately preceding chapter. We have a breath-taking reduction here. Let us list some of its features.
First, we engage in reduction by ignoring all the idiosyncrasies of an individual speaker.
Second, we reduce meaning to the meaning of sentences, and no longer consider the interaction with situational context or the larger textual context of discourse. It should be noted in Nida’s favor that elsewhere he explicitly called for attention to the larger contexts of paragraphs and discourse.42 But this sound advice of his is at odds with the transformational generative model of his day, which confined its analysis to the sentence and its constituents. The reduction to considering only sentence meaning, and to considering sentences one by one, leads to ignoring discourse cohesion, including cohesion achieved through the repetition of key words. This reduction then inhibits the reader from seeing meaning relations not only within individual books of the Bible, but in later allusions to earlier passages. The important theme of promise and fulfillment is damaged.
Third, we reduce all figurative expressions to a literal level, since the core formal structures in transformational generative grammar deal only with literal meanings.
Fourth, we reduce meaning from a richness including referential, emotive, expressive, and other dimensions to the single plane of “linguistic meaning.”
Fifth, we assume that meanings in the original are all clear and transparent. This assumption may be approximately true with some types of source texts on technical subjects or on mundane affairs; but it is far from being true with the Bible, which contains both obscurities and depths.43
Sixth, we reduce the meaning of a complex nonkernel sentence to its constituent kernels.44 This move is a genuine reduction, since meanings in fact do not reduce in a simple way to the meanings of kernel structures. Consider the expression, “God’s love.” Can we reduce this expression to the kernel structure “God loves you”? In many contexts, this involves a decided change of meaning, since the expression “God’s love” does not indicate the object of his love. Supplying an object such as “you” or “people,” as we must do in a kernel sentence, forces upon us greater definiteness than the original expression.45
A similar problem often occurs with passives. “Bill was overwhelmed” is less definite than “Something overwhelmed Bill.” For one thing, the passive expression does not indicate whether or not some one particular thing did the overwhelming. Maybe Bill felt overwhelmed, but there was no easily identifiable source for the feeling. Or maybe some other person, rather than some circumstance, overwhelmed Bill. The running back charged into him and overwhelmed him on the football field.
Similar problems occur when the back-transformation into a kernel requires us to supply an object. For example, the expression “Charlotte’s kiss” gets transformed into the kernel sentence “Charlotte kissed someone.” But did she kiss her dog? The “someone” in question may be an animal rather than a human being. The word someone does not then represent the possibilities quite adequately. Or did she throw a kiss to a large audience? Or did she just make a kissing sound, without directing her lips toward any particular someone? If we produce a kernel sentence to represent meaning, we expect it to have an object. But with any object we supply, such as “someone,” we change the meaning by introducing assumptions that are not contained in the vaguer expression, “Charlotte’s kiss.” (Generative grammar in 1965 could potentially handle some of this kind of complexity using so-called “subcategorization rules.” But such rules are still an abstraction that exists several steps away from the particular changes in meaning-nuances that one may observe in actual sentences in natural languages.)
The reduction arises partly from reductive moves that have already taken place within the theory of transformational generative grammar, which Nida was using as a model. But they also occur because somewhere along the line people may begin to assume that the transformations in question are meaning-preserving. They actually change meaning, as Nida admitted when he talked about “two levels of linguistic meaning,” the second of which is “supplied by the particular terminal construction.”46 Moreover, from a semantic point of view, the speaker does not necessarily start psychologically with a kernel sentence.47 The speaker may not know or may not be concerned to supply semantically absent information that would have to be supplied in order to construct a kernel structure.
In fact, generative grammar originated as an attempt to describe grammar, not meaning. It so happened that generative transformations connected sentences with analogous meanings. But no one could guarantee that the meanings would be identical. Sometimes differences in meaning are obvious. Compare the question, “Did you feed the dog?” with the analogous statement, “You fed the dog.” These two are transformationally related. But they differ in meaning because one is a question. By that very fact it has a different function in communication than the corresponding statement.
To insist that the meanings must be identical constitutes a reduction.48 It may still be a useful reduction. The linguist who uses the reductive process achieves rigor and insight of various kinds. But he also puts himself and his disciples in a position where they may forget the reduction, or refuse to acknowledge it. They then force meaning in human discourse to match their “scientific” results, rather than forcing their science to acknowledge the full reality of human communication.
Scientific rigor
The occurrence of the words “science” and “scientific” in the discussions can also signal a problem. Many have observed that in the twentieth century social scientists have often envied the rigor and prestige of natural sciences, and have struggled to achieve the same level of rigor within their own fields. But a field dealing with human beings contains innate complexities and multi-dimensional relationalities. In such a situation, rigor and fullness of meaning will often be like two ends of a seesaw. If one goes up, the other must go down.49
Nida’s 1964 book shows some telltale symptoms of this problem. He entitles the book, Toward a Science of Translating. Its title already introduces a tension: will we have “science,” so-called, with its ever-increasing rigor? If so, will we put ourselves at odds with the centuries-old philological and hermeneutical instinct that interpretation and translation alike are arts, not sciences?
Yes, we may have maxims for interpretation or translation. At points, we may have highly technical procedures for checking out our instincts, and for searching ever more minutely the meaning of particular words in particular contexts, and the meanings of various grammatical constructions. But in the end the process of translation is so complex and multi-dimensional that it must remain an art; it involves technique to be sure, as all good art does, but it is never reducible to a merely mechanical or formal process.50
Now Nida’s title does not say, “The Science of Translating,” but “Toward a Science of Translating.” The word “toward” signals that we are still feeling our way. We have not yet arrived at a full-fledged science. But the title nevertheless holds out as a goal the reduction of translation to science. And this, I would allege, contains a built-in bias in favor of formalism, and with it an invitation to move toward a reductionist approach toward meaning. It suggests in particular that all figurative, allusive, and metaphorical language must be reduced to the level of the literal, in order to be fit for processing by the scientific machinery.
Am I merely imagining sinister connotations that Nida did not intend? I do not propose to speculate about his inner intentions. Moreover, I have already indicated that Nida displays in his Chapter 3 a great deal of sensitivity and understanding concerning the multi-dimensional character of the meaning of texts. The problem, if you will, is not with Nida’s own personal awareness of meaning, but with the program he proposes to others—others who may be less aware of the complexities.
One can see the problem coming to life as one contemplates Nida’s description of translation after his discussion of generative grammar and kernel sentences:
Instead of attempting to set up transfers from one language to another by working out long series of equivalent formal structures which are presumably adequate to “translate” from one language into another, it is both scientifically and practically more efficient (1) to reduce the source text to its structurally simplest and most semantically evident kernels, (2) to transfer the meaning from source language to receptor language on a structurally simple level, and (3) to generate the stylistically and semantically equivalent expression in the receptor language.51
This key sentence contrasts two kinds of approach, both of which are utterly formalistic and mechanical about the translation process. The first approach would match surface grammatical structures between two languages, using an interminably long list. The second approach matches underlying kernels instead of surface structures.
But Nida has here presupposed that the only alternative to one formalistic approach is another formalistic one. He has not even mentioned the possibility of an art—the art of translation.52 What if we have someone translate who has a high level of comprehension of complex meanings in both languages? He does not translate merely by mechanical application of formal rules. Rather, using solid knowledge, artistry, and intuition together, he comes up with a translation that captures more fully the total meaning of the original. Is not this nonformal, nonmechanistic approach superior to both of Nida’s alternatives? Nida in his excitement over the potential of linguistics has lost sight of the complementary perspectives offered in the centuries-long traditions of hermeneutical theorists and literary theorists.53
The inclusion of the word “scientifically” in the middle of Nida’s sentence increases the problem. It biases readers to understand translation as a formal, mechanical process. It suggests that once the appropriate transformational rules are known for the two languages in question, one simply applies the mechanical process in order to produce the appropriate result.
I do not want to be too hard on Nida. Nida is partly thinking of the practical constraints on Bible translations into exotic languages. The professionally trained missionary Bible translator cannot hope to have the native speaker’s competence in Mazotec or Quechua. Given the translator’s limitations, thinking in terms of kernel sentences and transformations can provide genuine insights into differences between languages, and suggest ways in which the verses of Scripture may have to be re-expressed in a Bantu language.
But, as Nida stresses elsewhere, there is no good substitute for testing a proposed translation with native speakers.54 One must take into account the full effects of connotative and affective meanings, of context, of previous enculturation, and so on. There can be no science of translation in the strict sense, and Nida’s own practical discussions are proof of it. The formalization of meaning constitutes a danger, because it can lead to a reductionistic approach to translation by those who do not see the partial and one-sided character of Nida’s proposed procedure. A wide human sensitivity and comprehension is needed, and this larger human involvement complements technical study of language and linguistics.55 And I should underline the complementarity here. The technical study of language and linguistics does have much to contribute. I am not advocating an ignorance of linguistics, or a minimization of its value, but an awareness of the specialized character of its foci, and the consequent limitations in the vision of any one linguistic approach.
In considering Nida’s approach and its subsequent development, one must also bear in mind the practical limitations that arise in many situations where the target for Bible translation is a tribal culture. Cultures with no previous knowledge of the Bible or Christianity, and sometimes with little or no previous knowledge of worldwide cultures, create special difficulties for communicating religious truths. The extra barriers put a heavy premium on making everything simple and clear. Without this simplicity—which itself constitutes a kind of reduction—the target readers, with minimal skills in literacy, may give up altogether and not read the Bible at all. One can sympathize with the goals of utmost simplicity and clarity in such cases without converting these goals into general standards for Bible translation or for discourse meaning and semantics.
Componential analysis of meaning
One can see a similar encroachment of reductionism in the componential analysis of meaning. In the approach called componential analysis, the meaning of a word gets reduced to a series of binary components. A “bachelor” is (1) human, (2) male, and (3) unmarried. We may express this result by providing a list of three binary components: [+ human], [+ male], and [- married]. Componential analysis has a considerable history in the area of phonology. Here it works reasonably and insightfully, because phonology deals with a small, limited system of sounds whose significance depends largely on contrasts with other elements in the system. Thus in English the phoneme /p/ is distinguished from /b/ by the role of the vocal chords, and from /f/ and /v/ by the fact that the air stream is at one point completely stopped. We say that /p/ is [- voiced] (unvoiced) and [+ stop]. In keeping with its formalistic and reductionistic program, generative grammar soon adopted the use of componential analysis in its study of meaning. By analogy with the procedure of decomposing phonemes into distinctive binary phonological features, we now decompose meanings into distinctive binary meaning components such as [+ male] or [- married].
When we deal with kinship terms and certain other well-defined, limited areas of meanings,56 an analysis into meaning components may yield significant insight. And it may be of value more broadly for the language learner who is trying to appreciate key meaning contrasts in a new language. Nida rightly saw the value and introduced “componential analysis” of meaning in connection with his instruction about translation.57 But Nida also indicated some limitations:
By analyzing only the minimal features of distinctiveness, many supplementary and connotative elements of meaning are disregarded, ….58
The danger here is that careless practitioners may later overlook the reductionistic character of componential analysis, and consider it to be the definitive statement about meaning.
The reductionism in componential analysis can get added to other reductionisms that we have observed in Nida’s use of kernel sentences. As a result, reductionistic approaches to meaning may enter the process of Bible translation. Anthony H. Nichols, in his extended analysis of dynamic equivalence translation, has shown that dangers of this kind are not merely hypothetical, but have had a baleful effect on some translations.59
Unfortunately, the formalistic, “scientific” cast of the theory may make it difficult to take criticism. We know, do we not, that science is superior to the rabble’s naïveté? Once we have a scientific theory, criticism from outside can easily be dismissed as uninformed, because it does not bow before the power and insight of the theory. Theorists have then discovered a means for self-protection. When an outside observer complains about losses of meaning in a sample translation,60 he may be told that he is not competent to judge, because he is not initiated into the mysteries of componential analysis and translation theory. What the translation theorist’s net does not catch is summarily judged not to be fish!
Decades ago, Bible translators learned the maxim that one must listen carefully to the judgments of native speakers about meaning. It would be ironic if now, as translation theory grows more mature, it were used in reverse to pronounce “expert” judgments about which kinds of meaning native speakers may be allowed to worry about.
Continued development
Linguistically-based translation theory has continued to develop since Nida wrote in 1964.61 Analysis of propositional relations and discourse has enriched the early model.62 Translators like Ernst-August Gutt have explicitly criticized over-simple approaches to meaning that characterized the early days of translation theory.63 Kenneth L. Pike early recognized the complexity of interlocking between form and meaning, and the embedding of language meaning in a larger human context.64 Textlinguistics emphasizes the role of a full discourse, including paragraphs and larger cohesive structures, rather than confining attention only to individual sentences in isolation.65
And above all, better translators have always known that translation is an art; Nida’s and others’ technical tools are only properly used as one dimension in the process of trying to do justice to total meaning.66
All this is good news. But the dangers of reductionism remain as long as linguists and translation theorists experience pressure from the prestige of scientific rigor. Rigor is possible in linguistics and in translation when we isolate a sufficiently small piece of language, or one dimension of language, and temporarily ignore the residue that does not cleanly fit into a formalized model. Such models offer insights, but the clumsy, the doltish, and the arrogant can still misuse them.
1 See Vern S. Poythress, “Divine Meaning of Scripture,” Westminster Theological Journal 48/1 (1986) 241-279; Vern S. Poythress, “What Does God Say Through Human Authors?” in Inerrancy and Hermeneutic: A Tradition, A Challenge, A Debate, ed. Harvie M. Conn (Grand Rapids: Baker, 1988), 81-99.
2 For an introduction, see Susanne K. K. Langer, An Introduction to Symbolic Logic (New York: Dover, 1953); Irving M. Copi, Introduction to Logic, 4th ed. (New York: Macmillan, 1972).
3 Structural linguistics, as a twentieth-century discipline, is here distinguished from historical linguistics (diachronic linguistics), which has a much longer history, and from earlier attempts at grammatical and phonetic analysis of a single language.
4 See the historical accounts in Leonard Bloomfield, Language (London: George Allen & Unwin, 1933), 4-19; Peter Matthews, A Short History of Structural Linguistics (Cambridge: Cambridge University Press, 2001).
5 Ferdinand de Saussure, Course in General Linguistics (New York/Toronto/London: McGraw-Hill, 1959).
9 Ibid. 114. On the same page Saussure explicitly distinguishes “value” from “signification.” Likewise he says, “In a language-state everything is based on relations” (p. 122).
15 Noam Chomsky, Syntactic Structures (The Hague: Mouton, 1957). The copy to which I have access is the second printing in 1962.
17 Chomsky, Syntactic Structures, 13. Chomsky also assumed that the sequences fall neatly into grammatical and ungrammatical types, which he acknowledged is an idealization (p. 14). To achieve the impressive formal result of showing that finite state grammars are inadequate for natural language, he also had to introduce the idealization that says that sentences may be indefinitely complex–though in fact the limitations of human memory disallow in practice sentences of a million words (p. 23).
20 Technically, the optional transformations are applied to “forms that underlie kernel sentences … or prior transforms” (ibid.).
21 See, for example, Noam Chomsky, Lectures on Government and Binding (Dordrecht/Cinnaminson, NJ: Foris Publications, 1981); Noam Chomsky, Some Concepts and Consequences of the Theory of Government and Binding (Cambridge: M.I.T. Press, 1982); Liliane Haegeman, Introduction to Government and Binding Theory (Oxford: Blackwell, 1991); Andrew Radford, Syntax: A Minimalist Introduction (Cambridge: Cambridge University Press, 1997); David Adger, Core Syntax: A Minimalist Approach(Oxford: Oxford University Press, 2003).
22 See, for example, David Lee, Cognitive Linguistics: An Introduction (Oxford: Oxford University Press, 2002).
23 See, for example, Mary Dalrymple, Lexical-Functional Grammar (San Diego: Academic Press, 2001); René Kager, Optimality Theory (Cambridge: Cambridge University Press, 1999). And I believe there is still value in the more nonformalized, discovery-oriented, antireductive approach of tagmemic theory (Kenneth L. Pike, Linguistic Concepts: An Introduction to Tagmemics [Lincoln, NB: University of Nebraska Press, 1982]).
24 See John Lyons, Semantics, 2 vols. (Cambridge: Cambridge University Press, 1977); D. Alan Cruse, Meaning in Language: An Introduction to Semantics and Pragmatics (Oxford: Oxford University Press, 2000).
26 For a broader context, see L. G. Kelly, The True Interpreter: A History of Translation Theory and Practice in the West (New York: St. Martin’s, 1979).
27 See the discussion in Vern S. Poythress and Wayne A. Grudem, The Gender-Neutral Bible Controversy: Muting the Masculinity of God’s Words (Nashville, TN: Broadman & Holman, 2000), 57-90.
30 Eugene Nida, Toward a Science of Translating: With Special Reference to Principles Involved in Bible Translating (Leiden: Brill, 1964).
31 Jerrold J. Katz and Jerry Fodor, “The Structure of a Semantic Theory,” Language 39 (1963) 170-210.
32 Nida, Toward a Science, 40-46. P. 45n3 refers to Roman Jakobson, “Linguistics and Poetics,” in Style in Language, ed. Thomas A. Sebeok (Cambridge: Technology Press, M.I.T., 1960), 350-377.
37 The terminology occurs also in Eugene A. Nida and Charles R. Taber, The Theory and Practice of Translation (Leiden: Brill, 1969), 39.
38 Nida, Toward a Science, 70-119; note the earlier delineation of kinds of meaning on pp. 41-43; see also Nida and Taber, Theory, 56-98.
39 Nida, Toward a Science, 68. One can see the three-stage process worked out more explicitly and practically in Nida and Taber, Theory, 104.
42 “… expert translators and linguists have been able to demonstrate that the individual sentence in turn is not enough. The focus should be on the paragraph, and to some extent on the total discourse” (Nida and Taber, Theory, 102).
43 The point about depth versus transparency is made eloquently by Stephen Prickett, Words and the Word (Cambridge: Cambridge University Press, 1986), 4-36.
45 Still another problem exists with an expression like “the love of God.” This expression may indicate either the love that God has towards someone, or the love that someone has toward God, depending on the context. And some contexts may deliberately play on the potential ambiguity.
47 In fact, Chomsky warned against understanding generative grammar as a psychological theory (Chomsky, Aspects, 9).
48 There are complexities about how we might treat transformations. In Syntactic Structures (1957) Chomsky postulated a simple system of phrase structure rules leading to a relatively simple set of kernel sentences. Under this schema, questions were to be derived from statements by applying the optional transformation “Tq” (p. 63). But by 1965 Chomsky had incorporated the question marker into the base structure, and the question transformation became obligatory, so that a transformation analogous to Tqcould preserve the additional meaning involved in asking a question (see Chomsky, Aspects, 132).
Obviously over a period of time one can incorporate more and more previously neglected meaning aspects into the base grammar, in hopes of achieving a more adequate account of meaning. But the cost is increasing complexity in the base. In the limit, one might imagine a situation where all the lost meaning has been reintegrated; but the cost would likely be a horrendous complexity. In fact, for the sake of rigorous testability, generative grammar chooses in spirit to seek reduction rather than fullness of meaning.
In 1964 Nida did not fully endorse Chomsky’s later (1965) view that transformations must be meaning-preserving. Whether because he was working with the Chomsky’s 1957 view in Syntactic Structures, or because he saw the reductionism inherent in generative grammar, he affirmed that some extra meaning is contributed by the “particular terminal construction” (Toward a Science 65). But if so, it vitiates the attempt to translate by reducing meaning to the underlying kernel structures (as Nida proposed inToward a Science, 68).
49 Kelly delineates the problem: “Linguists’ models assume that translation is essentially transmission of data, while hermeneutic theorists take it to be an interpretative re-creation of text. It is hardly surprising then, that each group, sure that it has the whole truth, lives in isolation from the other” (Kelly, True Interpreter, 34).
50 Note the duality that Kelly sees in theories of translation: “For the majority, translation is a literary craft …. In contrast, linguists and grammarians have identified theory with analysis of semantic and grammatical operations” (Kelly, True Interpreter, 2).
52 Further down on the same page (ibid.) Nida mentions “the really competent translator,” by which he presumably means someone who knows both languages intimately. But Nida uses this temporary tip of the hat towards competence only as evidence that restructuring is sometimes legitimate; he does not consider whether the existence of this competent translator also shows the limitations in the reductionism and formalism that Nida proposes everywhere else on the page.
53 See Kelly, True Interpreter, 2-4, 36. “In the polemic between these three groups of theorists, only a few individuals have perceived that their approaches are complementary” (pp. 3-4). “Where linguistics concentrates on the means of expression, the complementary hermeneutic approach analyses the goal of linguistic interactions. The focus here is anti-empiricist: the central reality is not the observable expression, but the understanding of the cognitive and affective levels of language through which communication takes place” (p. 7).
56 Modern linguistic theory speaks of “semantic domains” or “semantic fields.” See, for example, John Lyons, Semantics (Cambridge: University of Cambridge Press, 1977) 1:250-269. For an application to Greek lexicography, see Johannes P. Louw and Eugene A. Nida, Greek-English Lexicon of the New Testament Based on Semantic Domains, 2 vols. (New York: United Bible Societies, 1988). For an example with kinship terms, see Nida, Toward a Science, 90-93.
59 Anthony Howard Nichols, “Translating the Bible: A Critical Analysis of E. A. Nida’s Theory of Dynamic Equivalence and Its Impact Upon Recent Bible Translations,” Ph.D. dissertation, University of Sheffield, 1996. As one might have guessed from the nature of Nida’s dynamic equivalence model, one of the effects is a flattening or elimination of figurative speech. Figurative speech poses a genuine challenge for translation, because a word-for-word rendering of a figure into another language may be difficult to understand or may invite misunderstanding. But this is not to say that we must go to the opposite extreme and systematically eliminate figurative expressions because of an aversion to anything that is not transparently clear.
60 For an eloquent complaint by such an “outsider,” see Leland Ryken, The Word of God in English: Criteria for Excellence in Bible Translation (Wheaton, IL: Crossway, 2002). Note the Appendix (ibid. 295-327) by C. John Collins, who has more of an “insider’s” understanding of the issues.
62 John Beekman and John Callow, Translating the Word of God: With Scripture and Topical Indexes (Grand Rapids: Zondervan, 1974), especially 267-342; Kathleen Callow, Discourse Considerations in Translating the Word of God (Grand Rapids: Zondervan, 1974). For a framework that acknowledges still more dimensions of meaning, see Vern S. Poythress, “A Framework for Discourse Analysis: The Components of a Discourse, from a Tagmemic Viewpoint,” Semiotica 38-3/4 (1982) 277-298; Vern S. Poythress, “Hierarchy in Discourse Analysis: A Revision of Tagmemics,” Semiotica 40-1/2 (1982)107-137.
63 Ernst-August Gutt, Relevance Theory: A Guide to Successful Communication in Translation (Dallas: Summer Institute of Linguistics, 1992); Translation and Relevance: Cognition and Context (Oxford/Cambridge: Blackwell, 1991).
64 Kenneth L. Pike, Language in Relation to a Unified Theory of the Structure of Human Behavior (The Hague/Paris: Mouton, 1967), especially 62-63; I illustrate in Vern S. Poythress, “Gender and Generic Pronouns in English Bible Translation,” in Language and Life: Essays in Memory of Kenneth L. Pike, ed. Mary Ruth Wise, Thomas N. Headland, and Ruth M. Brend (Dallas: SIL International and The University of Texas at Arlington, 2003), 371-380.