How to Lie with Statistics: Bible Translation Edition
Zondervan has invited several bloggers who blog on the Bible (aka bibliobloggers) to discuss an article by Karen Jobes entitled “Bible Translation as Bilingual Quotation” (the article is available at Zondervan’s blog). John Hobbins has already written two excellent posts highlighting both the strengths and weaknesses of Jobes’ approach. In this post I will attempt to cover different ground than John and question some of the core assumptions about Jobes’ method.
To be brief, Jobes’ article focuses on translation as bilingual quotation (a la translation translation techniques used by UN translators) and on verbosity as a measure of translation. Based on the needs of the target language, as seen in UN translators, Jobes holds that a translation might need to be more verbose than the original language in order to translate adequately the statement in the original language. She quantifies this by dividing the number of words needed in the target language by the number of words in the original. (See the graph on Zondervan’s blog for a quick run down.)
The use of graphs and numbers always brings out the skeptic in me. When the subject is how accurate a translation is, this skepticism becomes even more heightened. In respect to the issue of verbosity as a useful measure of a translations fidelity, I think the entire discussion is misplaced. It doesn’t matter how many words a translation uses, what matters is which words are used. Let’s look at an example to see how this is the case.
The MT of Dan 9:25 has:
ותדע ותשכל מן מצא דבר להשיב ולבנות ירושלם עד משיח נגיד שבעים שבעה ושבעים ששים ושנים תשוב ונבנתה רחוב וחרוץ ובצוק העתים
According to my count, this verse contains thirty-three words (according to how Jobes would define them in her article). Lets look at how some of the translations Jobes uses translates this verse and what there verbosity is.
Know therefore and understand, that from the going forth of the commandment to restore and to build Jerusalem unto the Messiah the Prince shall be seven weeks, and threescore and two weeks: the street shall be built again, and the wall, even in troublous times. (KJV) 45 words. 1.36 Verbosity.
Know and understand this: From the issuing of the decree to restore and rebuild Jerusalem until the Anointed One, the ruler, comes, there will be seven ‘sevens,’ and sixty-two ‘sevens.’ It will be rebuilt with streets and a trench, but in times of trouble. (NIV) 45 words. 1.36 Verbosity.
Know and understand this: From the time the word goes out to restore and rebuild Jerusalem until the Anointed One, the ruler, comes, there will be seven ‘sevens,’ and sixty-two ‘sevens.’ It will be rebuilt with streets and a trench, but in times of trouble. (TNIV) 46 words. 1.39 Verbosity.
So you are to know and discern that from the issuing of a decree to restore and rebuild Jerusalem until Messiah the Prince there will be seven weeks and sixty-two weeks; it will be built again, with plaza and moat, even in times of distress. (NASB) 46 words. 1.39 Verbosity.
Know therefore and understand: from the time that the word went out to restore and rebuild Jerusalem until the time of an anointed prince, there shall be seven weeks; and for sixty-two weeks it shall be built again with streets and moat, but in a troubled time. (NRSV) 48 words. 1.45 verbosity.
Know therefore and understand that from the going out of the word to restore and build Jerusalem to the coming of an anointed one, a prince, there shall be seven weeks. Then for sixty-two weeks it shall be built again with squares and moat, but in a troubled time. (ESV) 50 words. 1.51 verbosity.
Know therefore and understand that from the going forth of the word to restore and build Jerusalem to the coming of an anointed one, a prince, there shall be seven weeks. Then for sixty-two weeks it shall be built again with squares and moat, but in a troubled time. (RSV) 50 words. 1.51 verbosity.
What is interesting in this discussion is that Jobes’ idea of verbosity seems to play out. I, at least, would hold that the NRSV, ESV and RSV are the most accurate of the translations listed, based especially on how they render עד משיח נגיד. However, notice that the number of words doesn’t play as important a role as the choice of those words. For example, if the NIV had “an anointed one, a ruler” rather than it’s obviously flawed “the Anointed One, the ruler” then it would be a much more faithful translation of the text. Yet, from the point of verbosity, the word count would not have changed at all. (While the NIV and TNIV do propose this translation in the notes, their choice of primary text is more my concern, since most laypeople and undergrads often fail to examine the footnotes.)
To be sure, I think that Jobes is on the right track: the idea that a strict, wooden translation is more accurate or faithful is an unexamined assumption that doesn’t play out when one looks at contemporary translation techniques in other field. However, the accuracy of a translation must always be about the words used not the number (or order) of those words.