hefin jones Posted March 27, 2014 Share Posted March 27, 2014 Hi Folks I was trying to do a pretty simple greek construct search... in fact I did it and a few others just fine, OR SO IT SEEMED, until I noticed that I was getting an odd result. So I'm trying to find occurrences of δια followed by εν within five words. I got a nice list including Romans 3.24. But on inspection of some other results I discovered that I hadn't got Romans 3.25 in my list which I think it should have captured. See the screenshot below. Can you tell me if this is sheer incompetence on my part (I'd be glad to know!!!) or if there's some odd gremlin at play here? BTW Rom 3.23-25 are the first three verses in the results. NB the highlighted prepositions in 24. Why not in 25 too? Link to comment Share on other sites More sharing options...
Λύχνις Δαν Posted March 27, 2014 Share Posted March 27, 2014 Hmmmm.... works on GNT28-T, but not on UBS4-T and apparently not on NA-28 ( I don't have it myself ). Sounds like it might be a module issue. Ignoring bracketed words didn't seem to help. If you increase the within to 7 it finds it in UBS4-T. There is a note attached to πιστεως which is perhaps throwing the search count off. It's hard for me to see in your image but it looks like there are bracketing ' signs around in the NA text. Guessing but maybe they are throwing the search off. Try a larger WITHIN value. Thx D Link to comment Share on other sites More sharing options...
hefin jones Posted March 27, 2014 Author Share Posted March 27, 2014 Thanks Daniel. Didn't realise that the sigla etc of NA28 could throw construct searching off a bit! That's really helpful to know. Will repeat the searches with GNT28 as suggested. Wonder if accordance staff want to chip in on expected behaviour of construct searching when using texts that include sigla etc... Link to comment Share on other sites More sharing options...
hefin jones Posted March 27, 2014 Author Share Posted March 27, 2014 OK - just done what Daniel suggested and I am getting different results from NA28 and GNT28 for this construct search. GNT28 returns more results including the one I was by now expecting!Is the lesson, "use GNT28 for construct searching"?Or is it, "HJ needs to know what counts as 'words' in construct searching, especially when using NA28"? Link to comment Share on other sites More sharing options...
Λύχνις Δαν Posted March 27, 2014 Share Posted March 27, 2014 I think there is a problem with how the words are being counted. Someone from Acc will need to explain what's going on - I don't really know. It doesn't seem, to me, to make sense to search sigla in the constructs but there may be divergence of opinion here. As to constructs I always use GNT28 but I do it because that's what the syntax module works with not because results are better. Thx D Link to comment Share on other sites More sharing options...
hefin jones Posted March 27, 2014 Author Share Posted March 27, 2014 Yep I can confirm that if I increase the number in the "within" parameter the search returns Rom 3.25 in NA28-T. Within 5 and it doesn't work. Within 6 and it does. There are 5 items (2 words + 2 brackets + 1 other siglum) in between the two prepositions in the NA28 text. διὰ [τῆς] πίστεως⸃ ἐν Is there a rationale for construct searching counting sigla? Is there a way to over-ride this behaviour? Or avoid text with sigla? Link to comment Share on other sites More sharing options...
Ken Simpson Posted March 27, 2014 Share Posted March 27, 2014 (edited) Hi Hefin and Daniel, while I am not a module developer or a programmer, this looks wrong to me. Both searches should return the same result IMHO. For the time being use GNT28-T, but it should work in NA28-T and UBS4-T (which needs the "within" term increased to 7 to find Rom 3:25. Joel? Rick? Thoughts? Edited March 27, 2014 by Ken Simpson Link to comment Share on other sites More sharing options...
Joel Brown Posted March 27, 2014 Share Posted March 27, 2014 Currently, punctuation, sigla, and some variant words are all counted as intervening words for the WITHIN command. We have plans somewhere in 10.4.x to improve this, but for now you'll need to take account of that in your searches. Link to comment Share on other sites More sharing options...
Abram K-J Posted March 27, 2014 Share Posted March 27, 2014 Nice find, Hefin! Glad for the fix, Joel. Link to comment Share on other sites More sharing options...
hefin jones Posted March 27, 2014 Author Share Posted March 27, 2014 Thanks Daniel, Ken, Joel and others who've commented. There's a few construct searches I've done in recent months that I may re-visit to double check whether they were done with NA28. Knowing now the behaviour of construct with a text with sigla means that some of my data might be a bit off. Varying the 'within' value is obvious but some of my searches involve tens to hundreds of results and keeping the value of 'within' low helps constrain the results. Link to comment Share on other sites More sharing options...
Ken Simpson Posted March 28, 2014 Share Posted March 28, 2014 Hey Hefin, why wouldn't you just use the GNT28-T? The data will be reliable and reproducible with no change to the construct except the text it links to.. Link to comment Share on other sites More sharing options...
hefin jones Posted March 28, 2014 Author Share Posted March 28, 2014 That's what I was intending to convey Ken - I'll just revisit some old construct searches and ensure that they weren't done with NA28, and/or re-do them using GNT28. Link to comment Share on other sites More sharing options...
Ken Simpson Posted March 28, 2014 Share Posted March 28, 2014 OK - cool. You only mentioned NA28 in the post. Just checking. Good to know your little discovery though. Link to comment Share on other sites More sharing options...
.png.c3123c857d1029f98e39c9394adc358d.png)
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now