Top Down or Bottom Up...or both?

I have an idea for how a translation process can achieve a better translation. It assumes the truth of two linguistic fundamentals:

1. The meaning of a text is greater than the sum of its parts (bottom-up)
2. The context of each part selects the meaning of each part (top-down)

In other words, there is a top-down cognitive mechanism to both understanding and constructing a text as well as a bottom-up one. Both processes (top-down and bottom-up) happen simultaneously.

Here is the translation process in a nutshell.

1. Draft or use a "as literal as possible, as free as necessary" translation. Since it is quite difficult to think in ancient Koine Greek (we have no one to converse with), it is necessary for the next step to use an English translation that gets you most of the way there. I don't think one needs to be created; however, it would be beneficial to have a analytic translation along side of the one being produced by this process. See point 6 below.
   
2. Write up a structured outline of the book. The outline is linguistically based and fairly high level. The elements of the outline refer to the book, sections and subsections, and paragraphs. It is not verse oriented (poetry excepted). This outline is developed from existing English translations since people synthesize information best in their own language. However, this still won't be easy since the majority of English translations (and commentaries) don't flow well to begin with (For example, NT Wright is the only person I know that ties Romans 9-11 in with the rest of the Pauline letter! In fact, some commentators think 9-11 was redacted in at a later date). Note, however, that what we are trying to accomplish is a synthetic translation that allows for people to simply read the text. That requires the text to just flow. The primary purpose of this step and the next is to get a good feel for how the original author develops the point (or points) of the book. There is a lot of work to be done in this step and the next. And it will not be easy.
   
3. Associate a precis with each line of the outline. For example, the precis for a paragraph would look very much like a topic sentence. Larger lingusitic constituents will require larger than sentence precis. This part of the process helps force the coherence.
4. The outline is peer-reviewed for coherence. In other words, it should notread like a list of unsorted daily devotionals. Nor should it look like a list of headings. It should, pretty much, read like a synopsis (or abstract) of the entire book and it should be clear and natural. There should be logical/rhetorical transitions between each element. For pragmatic as well as authentic reasons, multiple outlines would be allowed at this point. There will be quality control feedback loops feeding back to this step so the outlines will gravitate toward greater accuracy. Someone who is familiar with the underlying book should understand the flow as given in the outline. Scholars might (very likely would) hold strong reservations regarding agreeing with it without further research; but, the flow would "make sense." The quality control for the outline--that is, does it authentically reproduce the author's original intent--would be hammered out in the steps outlined next. It is very important to understand that the metric of success of this step is this: Is the annotated outline coherent?
   
5. Each paragraph is translated. This is fundamentally a bottom-up procedure. The precis would be adjusted and frequently reconsidered as the paragraph is translated. Note, however, that the flow of the overall document must be upheld--coherence must be maintained. That is, the coherence of the outline must be maintained even if the outline itself is modified. The precis become extremely valuable in this step since they define the context within which the cognitive processes disambiguate linguistic and translation choices. Each of the multiple outlines would be considered against the word-level choices in order to better assess the value of each outline. A complete rewrite of the precis would be allowed, and changes to the outlines would be allowed, but those changes require redoing step 4 above and maintaining the metric of coherence.
   
6. If a "literal as possible, as free as necessary" translation was produced in step one, it would also be adjusted. That translation is meant to serve a more analytic audience than the synthetic translation being developed by the processed outlined here. However, the analytic translation would also serve as support for the more synthetic translation.
7. Field test the language of the synthetic translation. Note: This is not a field test of the content, certainly not a field test of the accuracy or authenticity of the content. It is to test the communicative accuracy. In other words, whether the test taker agrees with the content is irrelevant. Does the average person "get the point" of the text? That's the question. This step would result in appropriately revisiting any or all the steps above.

Lastly, each of the above steps would produce various kinds of supporting documentation. These articfacts would be openly available so the translators remain ultimately accountable to those being served by the translation.

Categories: coherence, Bible translation, precis, outline, translation process, peer-review, field testing, discourse flow