Louw-Nida Annotated Texts and False Hits
The OpenText.org project is rendering an excellent service to students of the Greek New Testament. We will be referring to their syntactically analyzed GNT (currently available in Logos 3) in some detail soon. But today what I have in mind is one of its less talked-about features: the annotation (tagging) of each Greek word in the New Testament according to the semantic domains and subdomains set out by Louw and Nida in their Greek-English Lexicon of the New Testament Based on Semantic Damins.
Personally, I have always thought that Louw-Nida’s approach would be an extremely useful one for computer-based Bible study and research (cf. my article Advantages of Bible Software: Louw-Nida’s Lexicon as a Test Case). The first attempt to make use of these domains as part of Greek searches was carried out by BibleWork‘s GSE (then known as ASE – Advanced Search Engine). There was no Louw-Nida tagging involved, but domains could be used as inclusion-exclusion lists (IEL) and plugged into a graphical search query. This, of course, is still the case in BibleWorks 7. It is a useful approach, but not ideal.
With the recent partnership between Logos and OpenText.org, a whole new avenue of research has opened up, and the new Louw-Nida tagging is one of the added/enhanced levels of annotation one can find (along with the classical morphological tags and the brand new syntactical tags) in the electronic incarnation of the OpenText.org Syntactically Analyzed Greek New Testament. I immediately felt this would allow us to perform unprecedented powerful searches. And I was right… in part.
We now have the whole text annotated with Louw-Nida’s domains, but since each word is attached to all the domains it can belong to, quite frequently we are bound to get false hits in our searches. Let me give you an example. The Greek word φυλακή* is found across four different domains (7.24, 37.123, 67.196, and 85.85), some of which have widely different meanings, but how do we choose one and exclude the rest? That is the question!
OpenText.org is well aware of the challenge this poses to the user. In Andrew W. Pitts‘ own words, “a process of semantic domain disambiguation must be undertaken before the analysis is performed – at least if one is using annotated texts.”
So, it seems to me that we have made a lot of progress, but there is still work to do. Tagging will have to be refined, so that words are assigned to a single domain, attending to their context(s) (rather than leaving the door open to all possible meanings, and thus domains — see the current implementation here by simply passing the cursor over any Greek word of your choice), and/or search engines will need to be able to exclude a word when it is attached to a specific domain we are not interested in. In the meantime, as it is often the case, user will have to manually weed out false hits. Obviously, this will always be much better than not having the possibility to perform these types of searches at all. However, users need to be aware of certain shortcomings so that the accuracy of their research is not jeopardized.
* I am using this particular word because it was brought up in a recent thread on both BibleWorks User Forums and Logos Newsgroups.