Tuesday, February 19, 2008

Review of "Relevance Theory and the Translation of Scripture"

Alright, so I'm going to throw this out there on the web. It's a review of an article in the Journal of the Evangelical Theological Society that interacts with contemporary linguistics and Bible translation (hence, up my alley). You can read an article by the same author, Karen Jobes, that covers some of the same ground here. WARNING: This is some inside baseball, even by my standards. If it bores you to tears, you can watch this. Or this:


Review of Karen H. Jobes, “Relevance Theory and the Translation of Scripture” Journal of the Evangelical Theological Society 50.4 (December 2007) 773-797.

Summary

Introduction

Jobes begins by pointing out the unique relationship between Christianity and translation. This embrace of translation is striking considering the long-discussed problems with translation and authority. In contrast to Islam and rabbinic Judaism, the first Christians embraced the translation of God’s authoritative Word into new languages. The apostles were content to quote the scriptures of the First Covenant in Greek translation, with every indication that they considered these quotes authoritative. Although Augustine considered the Old Greek translation the only authoritative Old Testament, evangelicals emphasize the authority of the autographs (original texts) and believe that translations are authoritative to the extent that they “faithfully communicate the same meaning as the original” (776).

The evangelical position requires us to interact with theories of language. Is it possible to translate “accurately into any and every other language?” (777). The many Christians engaged in Bible translation would say yes, but some linguistic scholars, both ancient (Iamblichus) and modern (Humboldt, Whorf, and Sapir) have said no. If accurate translation is possible, what makes for an accurate translation?

Controlling Beliefs

Jobes outlines 8 “controlling beliefs” about language that are “either explicitly stated in the Bible or . . . necessary entailment[s] of explicit biblical statements” (777). They are:

  1. “There is a fundamental correspondence between language used by human creatures and the world made by the Creator’s language—made independently of the human mind” (778). This is based on Gen 1:1-31; 2:7.
  2. God created human language so that we could communicate with him and each other (Gen 1:27; 2:16-23; 3:8-20).
  3. The relationship between language and reality among humans is no longer reliable, thanks to the fall. Humans believed the serpent’s false “reality” and perpetuate their own false “realities” with language. This is why God responds with verbal revelation.
  4. God is the originator of language diversity and the rules of each language (based on the story of Babel, Gen 11:1-9).
  5. An implication of #4 is that God’s message is not hindered by language diversity or the nature of any given language.
  6. “Meaning is sufficiently transferrable between languages for God’s purposes” (779).
  7. “There is no human language that is unsuitable for communicating God’s word” (779). Beliefs #6 and 7 are based on the story of Pentecost (Acts 2).
  8. “Bible translation is an essential part of God’s redemptive plan” (780). Jobes deduces this from the presence of speakers of every language in Rev 14:6.

Translation Theory

Regarding translation theory, Jobes contrasts preserving the form of a text in translation with preserving its meaning. She cites Luther to show that meaning takes priority over form. To answer the question of how to best transfer meaning, we must interact with psycholinguistic theory, in particular, a relatively new field called Relevance Theory.

Relevance Theory

The prevailing linguistic model of the 20th century envisioned meaning as encoded primarily in words. The most influential thinker in Bible translation, Eugene Nida, shows dependence on this model. In contrast, relevance theory (whose best-known proponents are Dan Sperber and Deirdre Wilson) focuses on “how what is only implied in a statement contributes to determining the meaning of what is explicitly said” (782). Jobes argues that the attention to pragmatic context makes relevance theory more cognitively and neurologically realistic: The model more closely reflects how the brain actually processes meaning. Four concepts of cognitive science that fit with relevance theory are “best-fit matching,” “spreading activation of neural networks,” “conceptual schemas,” and “cultural frames.”

Experimentation has shown that even in utero, the human brain begins categorizing some information as relevant (such as the speech sounds of the mother’s language) and other information as irrelevant (the sounds of other languages). In adults, the brain constantly disambiguates multiple possible meanings of words and sentences. Research has shown that the disambiguation process proceeds on the basis of which neural networks have already been activated in the hearer’s mind. For example, the sentence “She doesn’t play here anymore” will be interpreted differently if the context is a tennis court vs. a concert hall.

Sperber and Wilson theorize that three types of information are activated when a word is perceived: (1) the lexical entry, which contains the word’s sound-shape and syntactic information (e.g., whether it is a noun or a verb); (2) the set of deductive relationships that apply to the word (e.g., a dog is an animal); and (3) the encyclopedic entry, which is basically everything the person knows about the word (e.g., that dogs supposedly chase cats).

Feldman proposes that certain universal schemas govern the mental processing of language. The schemas develop upon earliest contact with the physical environment, e.g., an infant comes to understand the concept of physical support before she can use words like “support,” “on,” etc.

The sentence “He slipped on the wet floor and dropped the glass pitcher” illustrates how the brain processes explicit information in light of schemas (i.e., the nature of gravity) and encyclopedic knowledge (i.e., the nature of glass). The statement “communicates the proposition, ‘The pitcher is broken,’ even though those words are not explicitly stated” (788). Jobes continues, “if the speaker did not mean to communicate that the pitcher broke, she would have to go on to say, ‘And amazingly the pitcher didn’t break’ to disconfirm the strongly associated inference” (788).

The illustration and psychological research show that “what is implied by language is just as much a part of the meaning communicated as what is explicitly said” (789). In this light, it is important for Bible translators to know that “simply plugging in the equivalent words more often than not will fail to preserve the implicatures intended by the original language.” The translator must constantly decide what implicatures are intended by the original and whether/how to make them explicit in the translation.

A Wycliffe translator named Ernst-August Gutt gives contrasting examples of “direct” translation (where the implicatures are not made explicit) and “indirect” translation (where they are).

Matt 9:6 (RSV): “But that you may know that the Son of man has authority on earth to forgive sin”—he then said to the paralytic—“Rise, take up your bed and go home.”

Matt 9:6 (back-translation from Ifugao version): “But I will prove my speech to you. You know that it is God alone who removes sickness. You also know that it is God alone who forgives sin. And so, if I remove the sickness of this person and he walks, that’s the proof that I, the Elder sibling of all people, I also have the ability to forgive sin.” Jesus turned toward the paralytic and said, “Get up, take your stretcher and go to your (pl.) house!”

This illustration shows that the more implicatures a translation makes explicit, the more verbose it becomes (note the verbosity of The Message). Jobes then presents a simple comparison of some English versions on the basis of their verbosity. “If formal equivalence could be ideally achieved . . . the number of words in the translation would correspond closely to the number of words in the original” (795). But a comparison of word count “shows some surprising results (796):

Translation

% more words than Hebrew and Greek originals

NIV

33.18%

TNIV

33.60%

NLT

35.96%

ESV

38.93%

NKJV

43.31%

NASB

43.58%

ASV

43.93%

KJV

45.02%

NRSV

64.43%

Review

In my experience, biblical scholars often do a fairly mediocre job of interacting with contemporary linguistics. The field is a difficult one to appreciate, and outsiders tend to latch on to some piece of theory which they appreciate or abhor without understanding the whole argument. However, Jobes's interaction with linguistic and cognitive theory is possibly the best I have seen from a biblical scholar. She clearly understands Sperber and Wilson's arguments and presents them clearly. Jobes's case for the cognitive reality of relevance theory and its importance for Bible translators is convincing.

Areas for Development

There are a few underdeveloped themes in this essay, however, that I would love to see more work on. First, Jobes raises two fundamental (and related) questions about language: (1) Does language refer to things that are really "out there" in the world, or does it only refer to things "in our heads"? and (2) Is it possible to translate a message between all and any languages? Perhaps it is unfair to fault Jobes for not providing a complete answer to these complex questions, since to do so would take a completely different article (or even a few books). However, she answers them simply by deducing certain beliefs from Scripture, not by interacting with the linguistic or philosophical arguments for and against them. Namely, (1) because God created the universe by speech and spoke with human beings, there is a real connection between speech and the world "out there" and (2) because God spoke in multiple languages at Pentecost, all languages are suitable for God's revelation. I'm fairly sure I agree with Jobes's conclusions on both of these, but I don't see how her deductions would convince anyone who doesn't already agree. In particular, if I were an agnostic linguist who has made real arguments for language referring to things "in our heads", I think Jobes's argument from creation would simply annoy me. (As it is, I'm not sure what to believe on this question. Ray Jackendoff's argument for language referring to things "in our head" is compelling, and doesn't seem to be inimical to Christian faith. See Foundations of Language, chapter 10, "Reference and Truth".)

Verbosity

The chart of verbosity at the end is probably the part of the article that will grab the most attention, because it is concrete and it runs contrary to common perceptions about translations. The NLT comes off as less verbose than the ESV. This is not an argument for or against either translation, as Jobes makes clear. She makes the point that verbosity is a side effect of a legitimate endeavor, that is, making what is implicit in the text explicit. So the chart could be taken to show that the ESV is a "better" translation because it does more explaining.

The thing is, it's hard to believe that the NLT does less explaining of implicatures than the ESV. It seems like there has to be more to the difference in word count than explanatory expansions. One reason the ESV might have more words is that it translates the initial Ands while the NLT is more likely to leave them off in the interests of good English style. Let's look at an actual passage (Matt 9:1-3):

Matthew 9:1-3

Translation

Text

Word Count

Wooden Translation; one English word for one Greek word

And embarking into boat crossed and came to his own town. And behold brought him paralytic lying upon bed. And the Jesus seeing the their faith said to paralytic, "Courage, child; your sins are forgiven." And behold some of scribes said to themselves, "He blasphemes."

45 words

ESV

And getting into a boat he crossed over and came to his own city. And behold, some people brought to him a paralytic, lying on a bed. And when [...] Jesus saw [...] their faith, he said to the paralytic, "Take heart, my son; your sins are forgiven." And behold, some of the scribes said to themselves, "This man is blaspheming."

59 words; 13 words added (2 of which are making the implicit explicit--"some people"), 2 words omitted

NLT

Jesus climbed into a boat and went back across the lake [...] [...] to his own town. [...] [...] Some people brought to him a paralyzed man [...] on a mat. [...] Seeing [...] their faith, [...] Jesus said to the paralyzed man, “Be encouraged, my child! Your sins are forgiven.” But [...] some of the teachers of religious law said to themselves, “That’s blasphemy! Does he think he’s God?

61 words; 21 words added (13 of which are making the implicit explicit), 9 words omitted

The ESV and NLT both add words in some cases because English requires more words to express the same thought (e.g., the indefinite articles, subject pronouns, and expressions like "Take heart"). But the NLT clearly has more explanation than the ESV ("across the lake", "of religious law", "Does he think he's God?"). The NLT has roughly the same word count as the ESV because it also trims more words from the original to make up for the expansions, cutting words like "and", "behold", and "lying". Given that the ESV supposedly has 16,000 more words than NLT, my guess is that the NLT must trim these low-semantic-content words fairly aggressively throughout the Bible. I would also hypothesize that the NIV does a lot of the same trimming but doesn't make as many expansions, while the TNIV makes a few more expansions.

1 comment:

Richard A. Rhodes said...

Hey, don't blame me for the sins of the East Coast schools of linguistics.

You don't really get to say that it is the fault of linguists that there isn't much interaction. After all there are hundreds, if not thousands of linguists involved in Bible translation at this very moment. They just happen to be translating into minority languages.

Some of us linguists have been arguing for the relevance (no pun intended) of serious linguistics to English Bible translation for years. The people who won't listen are the theologians, many of whom stand to have their pet doctrinal ox gored by a serious, accurate, pre-theological 21st century translation. (See the end of this post of mine.

(And BTW, Sperber and Wilson's work builds on pragmatic work that has been around in outline more or less as we understand it now for 30 years. And most of the speech act thinking goes back even further. Pragmatics ain't new.)

There are linguists who have talking about this stuff in the blogosphere for some time now as well. See the Better Bibles Blog.