Interesting search procedure using GSE and LouwNida

[MGVH]

Logos3 has a new possibility using the OpenText texts to do sytactical searches. (If you haven't already, be sure to check out OpenText.org (http://www.opentext.org))
One can run a variety of different kind of searches, and it led me to ask the question how the preposition en is used to indicate temporal relationships in the NT, eg., en hêmera, en de tais hêmerais ekeinais, en ekeinô tô kairô, etc. One of the keys here is using the LouwNida domains to look for "time" words.

It requires learning a whole new set of grammatical concepts, and it still needs some matters worked, but it is an interesting new tool.

I tried constructing a similar search in BibleWorks7 using the Graphical Search engine. I could not do a syntax search in BW7, but the Logos results showed me that I could be fairly confident looking for an en followed by a dative noun/adjective 'time' word within 3 words of the en. Trying to do the search as one might first think will produce way too many hits, because there are 208 "Time" words including verbs and prepositions in the LouwNida domain.

Here now is the part that might help others interested in doing this kind of search.

After some experimenting, I finally figured out that what I first had to do was use the Word List Manager, generate a list using the LouwNida domains (again using 67=Time), selecting the whole list, then using Edit > Change morphology on selected words in order to specify nouns/adjectives in the dative (ie, applying > @[na]d* <), then save the list as an IEL file. (sorry, IEL files are not allowed to be attached)

I then went to the GSE and set up the attached query. (Note that I loaded up the IEL here.) When I run the query, I get an error noting that looking for a verb which I have set up as a noun/adj in the dative is an invalid form, but it is safe to mark the box to continue without asking about this and finish the search. I ended up with 217 results.

There were a few false hits (e.g., Matt 25.36,39,43), but some were due to other dative words within the range, and some were due to using the semantic domain list where a word like phulake can be used for being "in prison" but also used as in Luke 12.38 as a time reference: "in the second/third watch."

Comparing the BW7 results with the ones using the syntax search in Logos3, I found that I had missed a few because of how BGM tags nouns and adjectives differently. Note that I had tried to use @[na]d* to find all nouns or adjectives in the dative case. This doesn't really work, though. For nouns, *@n is followed by the case designator. For adjectives, *@a is followed by the type of pronoun designator then the case designator. For some reason, I missed all the ordinal adjectives though it picked up the rest. (So, for a complete search, I would really need to do the whole search twice--or double up using "or"--and look once for dative nouns and once for any kind of dative adjectives.) The GNM codiing has similar issues. The BYM coding, however, does work properly because different types of adjectives are not coded separately. (See for yourself by changing all the word and merge boxes to BGM, GNM, or BYM.)

SO, lessons learned:
1) Using Word List Manager with LouwNida to generate an IEL list that can have morphological specifications attached. This is a powerful way to enhance the GSE.
2) Note how the various morphological schemes treat coding. It makes a difference!

[Dunk]

Thanks Mark,
That was very helpful.
After reading your email I decided to use the same principle for searching for ek with timing words in the genitive, except i chose a maximum of 2 words inbetween.
I ended up with 112 verses with 116 hits.
I've enclosed the GSE (qf) file, and if it's at all possible I would appreciate if you could tell me if I've done this correctly as I have not been using the GSE for long.
Thanks
Dunk

[MGVH]

Looks good, Dunk, but there still are some interesting observations to make about how the search works.
1) It looks like you specified all the genitive cases individually. When you create the word list from LouwNida and then apply the morphology @[na]g* you will come up with a slightly different listing form. You get the same results, but it's slightly less work.
2) You will, however, miss a bunch of hits using your file and the BGM version. If you use my way of generating the IEL file and then use the BYM version for all the boxes, you will get 34 hits in the NT instead of 11.
a) Some are due to text critical issues. Cf Mt 19.20 in BYZ
b) Some are false hits. Cf. Rom 5.16 "from one" which is not a time ref.
c) Some are real hits due to encoding. BGM differentiates the various adjectives and misses numbers/ordinals, so you miss verses like Mat 26.42.

Thanks for sharing.

[Dunk]

Thanks for the help Mark.
I hadn't even considered textual variants, that's a good point.

I didn't get it before but I see now how you made your IEL list, and it appears to be a quicker way that I did.
What I did was to get the Louw-Nida domain (67 Time) list and then use Word list manager to compile a list of every genitive noun in BGM.
I can't remember exactly how i did this, but it involved using both the primary and secondary lists in Word List Manager - it didn't take long but it still wasn't as quick as your method.

I think one advantage of your method is that it gives you an IEL file that can be used in the GSE for both BGM and BYM tests.
In fact there should be a way in the GSE to analyse both texts and to not show repeated verses (but I haven't worked out how to do that yet.)

Peace
Dunk