and Louw-Nida Semantic Domains

I’ve mentioned in the past that the Syntactically Analyzed Greek New Testament will have Louw-Nida domain information available at the word level. This means that one can combine syntax with Louw-Nida semantic domains and do some interesting stuff when searching.

This is much easier to show you than to write and tell you. So I fired up the video capture software and threw together a quick search. Where, I wonder, in the Greek New Testament does something like James 2.19 occur? (“Even the demons believe — and shudder!”) This, translated into a search query relying on semantic domains instead of words, could be stated like:

Find a subject with a head term in semantic domain 12 (Supernatural Beings and Powers) preceding a predicator (verb) with semantic domain 31 (Hold a View, Believe, Trust)

(Flash Presentation, approx. 4 megs, 1024×768).

The video is a single take, no edits. Pardon some of the mouse jitters.

This isn’t searching on words, it is searching on domains. It finds clause subjects that contain a word (a “head term”, meaning is the primary word in the word group) that are also tagged as having to do with “Supernatural Beings and Powers” that have a clause predicator (verb or predicate) that contains a word (again, a “head term”) that is tagged as having to do with belief or trust.

You know, sort of like James 2.19: “Even the demons believe — and shudder!”. Only without words, so you can find instances where supernatural beings are said to trust or believe.

With a few more clicks (note the “Copy” button in the Syntax Query dialog, which can “clone” the currently selected structure) we could add an “OR” to search for where the predicator precedes the subject, just to cover all of our bases.

Note especially all of the different ways in which the search results are shown. You can view them with the clausal breakdown, as a syntax graph, or in a reverse interlinear (I have the ESV specified, but I could’ve specified the NRSV through preferred Bible settings). Click and view. With the English and/or Greek highlighted.

There is a whole lot more going on. Did you see the glossary popup on “Predicator” when the mouse cursor hovered? Did you see the entries from BDAG pop up on hover when hovering Greek text in the clause breakdown? The same thing in the syntax graph? And in the reverse interlinear? The actions captured by the video were all done with the mouse, either via point/click (specifying the query) or hover (glossary information, lexicon information).

This capability (BDAG assuming you have purchased it) should be available with the next beta release of Logos Bible Software v3.0.

We’re interested in knowing what you think of this sort of stuff, so please feel free to leave us feedback in the comments to this post. Thanks!


  1. Philip Gons says:

    I was a little surprised at some of the verbs that apparently fall within the semantic domain of belief or trust. The first example was that angels came to Jesus. I’m not quite sure how coming is semantically related to believing. I know that sometimes coming carries that meaning, such as in “coming to Jesus,” but that is clearly not what is going on in Mat 4:1. Can you help me understand this?

  2. Hi Philip.
    The semantic domain tagging encodes all possible domains that a given word may occur in; it doesn’t disambiguate each occurrence in the NT. So using the semantic domains inside of material is like a short-cut to a topic-sensitive wordlist.
    In this case (Mt 4.11) the word PROSERXOMAI is the verb. Louw-Nida classify it in domains 15, 31, and 34. What the analysis has done is to tag each instance of PROSERXOMAI with each of these domains. I searched on domain 31, so I found this instance. Note Louw-Nida list 1Ti 6.3 as an instance of PROSERXOMAI in domain 31.
    This tagging allows a search to use a domain as, essentially, a word list: Find instances of words that Louw-Nida classify as occurring in a given domain.
    A further step would be to disambiguate the domains at each word instance in the NT, to positively say “this word occurs in this domain”; but that work is not available. We are considering what it might take to do such work so that this sort of search can be refined even more in the future.
    But still — pretty cool, no?
    - Rick

  3. Philip Gons says:

    Thank you for the clarification. Yes, it still is pretty cool. :) I guess I’ve just been getting spoiled with all this new stuff and just assumed that this would be context sensitive. It’s not a big deal to weed through a few extra occurrences. Tagging each verb and verbal individually would be an improvement, though I’m sure a time-consuming one! Even without such an improvement, there’s still a great amount of benefit in this!

  4. David Phillips says:

    This is a great feature, Rick. Thanks for giving us more options for how to search. Will this be available in the next beta to try out?

  5. The ability to search in this way looks great. This syntactical searching could really set Logos apart.

  6. Jonathan says:

    This is FABULOUS. This saves hours of time in exegetical analysis. I check the blog almost everyday to see the new features you guys are developing.
    A couple questions: When do you anticipate the release of 3.0? What is the cost to current Logos users?

  7. Scott Stocking says:

    I was trying to do a couple syntactic searches and couldn’t figure out how get consistent results. For example, I was searching Anderson-Forbes (Hebrew) for occurrences of yom with a quantifier (“third day”, “three days,” etc.) trying to confirm or deny Henry Morris’s (??) conclusion that yom with a number indicates a 24-hour day. I used the following:
    Phrase 2: Phrase-Type = {Noun, Cardinal Number, Ordinal Number} AND Semantics = Time
    Segment 1: Lexeme = “???”
    Segment 3: Any Descendant — Semantics = Quantity
    I got 68 hits, but the odd thing was, I got days one through five of creation, but it did not return day six as a hit. I tried making the day phrase definite (by default it is when a noun is qualified, especially with an ordinal number, but I wanted to see what would happen) and I got the same number of results. When I tried broadening the search (I eliminated the “quantity” segment), I got over 2000 hits, but I got phrases that did not have cardinal or ordinal numbers in them. So where’s the middle ground that helps assure I’ve gotten all occurences I’m looking for?
    On the NT side, in the SAGNT, I tried searching for all occurences of a genitive pronoun that comes after a verb but before the noun it modifies. When I simply specified “verb,” I got 198 hits. When I specified “active verb,” I got 112 hits. When I specified “either middle or passive” (and each individually) I got zero hits! I even tried the eimi verb and only got 28 hits. I think I tried participles, too, but those should have been included in active verbs, right? I have a theory about the significance of that construction, but I’m keeping it to myself.
    Any help or references would be appreciated.