Monday, February 18, 2008

BibleWorks7, Logos3, Searching, Louw-Nida and Data Accuracy

In two previous posts (HERE and HERE) I looked at how you would BibleWorks7 or Logos3 to look for all the verbs in the imperative mood that occur in Louw-Nida domain 25 which deals with attitudes and emotions. It is pretty remarkable to have tools to conduct such searches. In this post I am first going to comment on how such searches are implemented in BW7 and Logos3, and then I will comment on the results that are generated.

As for the ease of use in conducting a search on Louw-Nida domains, both programs are quite easy to use. If, however, you want to do a LN domain search AND use morphological parameters, then BW7 required a somewhat obscure and mult-stepped process of using the Word List Manager to create a list of words, then apply morph coding, then save results as a file, and then open that file in the Graphical Search Engine. Everything is nicely handled in a single dialogue box in Logos3. On the other hand, while it perhaps took me nearly a minute to set up my search in BW7, it took less than 2 seconds for the entire search to run. In Logos3, it took only a few seconds to set up the search, but it took 45 seconds to start returning hits.
As for the results, in BW7 it is important to have your display versions set correctly so that you can see the results highlighted as shown here. If, however, you want to scan the results in English, switching to an English version will, of course, remove all the highlighting.

In Logos, the results appear like this:This is indeed a more attractive and helpful presentation. Not only is the Greek text present with hits highlighted, but by connecting with the NRSV Reverse Interlinear, you can also see the English with the corresponding words highlighted. (Note that one can choose a different display with just the references and English or references, hit word, and English.)
All in all, Logos3 works better in conducting this kind of search.

The search results, however, are only as good as the data, the links between the data, and the searching engine. (Skip to the bottom if you want my bottom line. A lot of picky details ahead...)

As I noted in the first post, even though this search is intended to include the sub-domain of "desire, wish, want," it does not necessarily include everything for which you may be looking. In Philippians, e.g., this search does not return 1.22 where Paul, reflecting on whether he ends up living or dying concludes by saying, "I do not know which I prefer." You would think that "prefer" would be included in that domain #25A of "desire, wish, will...," but in 1.22 the verb used is αιρεω, and in this situation, it has been grouped in domain #30F of "choose, select, prefer." This highlights the first of a number of important issues of which one should be aware when using software to conduct searches.

  • Louw-Nida has some categories that overlap, so a 'complete' search may require surveying more of the Louw-Nida categories.
  • Some of Louw-Nida's decisions regarding to which sub-domain a word is assigned can be questioned.
I conducted searches using BibleWorks7's Greek morphology (BNT/BNM) and Logos3's OpenText Greek morphology, but note that searches could similarly be conducted using BibleWorks7's Friberg morphology (GNT/GNM) or Logos3's Lexham Syntactic Greek NT database with its morphology. (Lexham is not yet complete for the whole NT.) Depending on which one you use, you will end up with four different sets of responses. Comparing the differences, here are some important issues to keep in mind.
  • The accuracy of the morphological coding: One must simply be aware that there are errors. In this particular search, OpenText identifies μεριμνατε as an imperative in Matthew 6.28, but this can't be right. (It does also identify it correctly as an indicative.)
  • Options in morphological coding: We were looking for imperative forms in this search, but using the Friberg morphology in BW7 or the Lexham in Logos3 return 1 Peter 3.14 which the other two do not. Why? Because Friberg and Lexham recognize in Greek that an aorist Greek imperative is not used with a negative and that the aorist subjunctive is used instead. (Note that Friberg actually is better in identifying as the subjunctive functioning as an imperative. Lexham somewhat incorrectly simply identifies it as an imperative.) Does μη with an aorist subjunctive count as an imperative or not? We simply need to be aware of how the morphological systems handle issues like this. (Other examples include
    -- there are quite a few instances where a verb can be taken as either an indicative or an imperative. E.g., ὑπομενετε in Hebrews 12.7 can be either. BW7 offers both possibilities, but Lexham only treats it as an indicative.
    -- the treatment of middle/passive/deponent
    -- whether participles are all lumped together or sorted (as Friberg does) into "participles" and "participles (imperative sense)"
    -- the handling of verbs with both -ω and -ομαι forms.
  • Coordination between datasets: Quite a few discrepancies between the various texts and the hits generated relate to the way the morphological coding hooks into the database. The LN lexicon can assign a word to a number of sub-domains. Sometimes the text made the link to all the sub-domains (and thus picked up false hits) and sometimes it only linked to specific sub-domains (and thus missed the desired hit). E.g., ἀσπαζομαι is regularly used in the imperative in the Pauline letters when he tells the recipients to "greet" others on his behalf. That verb is listed under LN25.130 with the meaning of "be happy about," but it is assigned to LN33.20 when it means "greet." Only the Lexham correctly handled all these instances.
  • Translation choices: As may be expected, when a word has shades of meanings that might cross a number of LN sub-domains, it becomes the choice of those setting up the links between the Greek word and LN as to which one they think is best. E.g., in 2 Corinthians 6.13, Paul encourages the recipients to "open wide your hearts" using the verb πλατυνω. That is a literal rendering, and as such it is assigned to LN 81.17. As an idiom, however, it means something like "show affection for," and there is a specific reference to such a meaning in LN25.53. Only Lexham made the connection to the LN25 listing.
  • As for other discrepancies between the lists of hits, sometimes I could not figure out why one of the texts missed a hit. As far as I could see, the text identified the word as an imperative from LN25, but simply did not find it. (E.g., ἀνεχεσθε is missed in OpenText and Lexham in 2 Cor 11.1. BW7 misses ἀνακυψατε in Luke 21.28.)
So, bottom line:
We are in a position to do far more with the biblical texts than ever was possible back in the day when I was hard-copy editions of BGAD and Moulton & Geden. It is, however, difficult to claim that an exhaustive and complete search has been conducted. There are errors and coding decisions made in the texts that 'hide' some of the results in which we are interested. The interactions between texts, lexicons, and translations is a subjective process that will often 'hide' some of the results we want.
From this simple exercise of searching for imperatives from LN25, I can make two general observations.
  1. It was easier working with Logos to conduct the search, and I think it does a better job of presenting the results.
  2. I was impressed with the Lexham database. It appears that closer attention is being paid to lexical, syntactical, and grammatical niceties. The specificity (i.e., it often makes a choice when options are possible) is usually a good thing, but I did disagree with it at times, so it is not perfect. I will, however, be eager for the Lexham Syntactic GNT text to be completed.

Post a Comment