Better Bibles Blog has moved. Read our last post, below, and then
click here if you are not redirected to our new location within 60 seconds.
Please Bookmark our new location and update blogrolls.

Monday, July 04, 2005

Flesch-Kincaid and Bible reading levels

I just came across a blog post which discusses use of the Flesch-Kincaid reading level indicator as a method to improve a writer's prose. (Click on the title for this post to get to that post.) A number of English Bible version teams these days indicate the reading level of their translation. As I had indicated briefly in some previous posts, these indicators of reading levels are just approximations. The tests are really too simplistic, especially when they are assessing translated material where it is all too easy to use translationese, English which is not natural English, which may have relatively short words and sentences and so score fairly well on a reading level test. But that translationese English may have non-English syntax (the reading level tests to not test for that) and obsolete vocabulary (again, not tested for), which would make a Bible version much more difficult to read, but one would not know that just from noting the reading level given that version by these simple tests.

In any case, there is some value is using these tests. They do allow us to get a relatively good idea of which versions will be more difficult to read and which ones easier. I, personally, agree with the author of the post to which I am linking that use of a test, such as Flesch-Kincaid, can be useful as a heuristic during the translation process, to monitor the translation to see if it is staying within the reading grade level that the translation team intends for it.

There is greater heuristic value in using these tests if one is thinking and writing directly in English, not involving another language, as happens when translation occurs. So for the authors who read this blog, you might want to take note of ideas in the bloggers post.

Finally, I personally believe that it would be of great value if more English Bible translators approached the translation task as if they were composing in English, rather than translating. I recall one of an elder translation consultant telling us many years ago that if he could do his translation over again, he wishes that he could have the MTT (mother tongue translator) of the language read over an entire section of Scripture in English (or whatever language the native speaker will translate from). Then the MTT would close the source text and narrate the translation in his own words. This should bring a greater degree of naturalness to the translation. Along with naturalness comes greater communicative accuracy (as long as the exegete on the team also ensures exegetical accuracy), since what sounds good in our own language will communicate better to us. What sounds choppy, unnatural, or obsolete creates a significant barrier to understanding for most "ordinary" users of the Bible. Many who read this blog are "non-standard" users of the Bible, able to read and understand the unnatural English which occurs in many English Bible versions. But the Bible was not written for non-standard language speakers. It was written for people who were ordinary speakers of the original Hebrew, Aramaic, and Greek languages, in most cases people who did not have much advanced education. The more accurately and naturally a Bible version communicates to standard speakers of a language, the better that Bible version will be. I realize that this is counter to much of what we hear from conservative Christian pulpits and seminaries these days, because our pastors and Bible teachers have been trained by seminary professors who have been trained by their professors to understand non-standard classroom English translation of the biblical languages.

Let us strive to make better Bibles. Let us encourage our denominational leaders to make better Bibles. Let us distribute better Bibles to those who need to hear God's Word in their language.

Better Bibles bring greater independence to Bible readers, spiritual independence from those things which hinder us from living the kinds of lives God desires for us, and independence from the need for as many extra-biblical helps to translate to better English the poor English which is found in many Bible versions. Today is Independence Day in the U.S. May each day become a greater day of independence for Bible users around the world as we become more attuned to how we actually speak and write our languages, and how the Bible can speak accurately and clearly and naturally in our languages.

Happy Independence days to all!

Categories: , , ,


At Tue Jul 05, 07:19:00 AM, Blogger Trevor Jenkins said...

Ah yes reading levels. :-) The problem with them isn't so much that they are devised for texts written in an author's native language but they only consider the lexical content of the word stream. We could feed one of these analysers nonsense such as Lewis Carroll's "Twas brillig" and it would still report a reading level.

These scores can be useful in Bible translation when some metric is applied to the same passage taken from different translations. Then we can make statements about the relative differences between them. What we can't do is make absolute statements about a specific translation as being "grade 10" or "grade 5", which is what many Bible publishers do.

Also some consideration needs to be given to which implementation of these metrics is being used. The Flesch-Kincaid measure as implemented within Microsoft Word is capped to scoring no higher than grade 14. When run over big enough samples of the KJV the result is never more than grade 14 --- surprise surprise. However, when a different implementation that has not been artificially capped is used then a very much higher and (IMO) more realistic score, equivalent to grade 20 or post-doctoral study, is obtained. When conducting my own reabability studies of Bible translations I have used the diction program from the Free Software Foundation's GNU project. This is a name-sake re-implementaton of the tool in the Programmer's Workbench edition of Bell Lab's UNIX operating system. An second advantage of this version of diction is that it includes many other readability metrics --- some which are suitable than Flesch-Kincaid.

The metrics themselves are not really tuned for arbitrary texts. Many of them were created as a regression analysis of either school text books or military manuals.

Even when we done all this analysis of the text there still remains one issue that these numbers can't help with ... the low level of reading skills amongst adults. It has been suggested that the average reading age of American adults is equivalent to that of a schoolchild in grade 6! More substantive is a recent statistic that 25% of American adults are illiterate. Sadly there are similar recent statistics published that indicate the situation in the United Kingdom (especially England itself) is no better. The UNESCO studies on literacy don't use these metrics at all. They follow an approach similar to that of "field testing" where comprehension of text is measure and, for the UK the results are dismal. Or the suggestion that at least 10% of UK adults are dyslexic. Anecdotal evidence has been gathered that upwards of 80% of adults are a-literate --- they can read but choose not to. These are sizable proportions of the general populace and are also of church-goers/Bible readers. To date I've seen no attempts by national Bible societies or Bible publishers to address any of these statistics.

At Wed Jul 06, 09:09:00 PM, Anonymous Anonymous said...

The Flesch-Kincaid formula has been abused by the King James Only people to "prove" the readability of their favourite translation. The Waite father-son team give the King James Version at 5th-grade readability rating based solely on a mechanical application of the formula ( I think Trevor's point is pertinent: theirs a desperate case of making Jabbaworky appear readable.

Joseph Ng


Post a Comment

Links to this post:

Create a Link

Subscribe to Post Comments [Atom]

<< Home