David Knoll Posted November 18, 2014 Share Posted November 18, 2014 (edited) I am looking for all the occurrences of שמר without an object. I thought I'd search as in the screenshot attached, but the occurrences with the object are not filtered out. I must be doing something wrong. Can anyone point out to me how I should run this search? Edited November 18, 2014 by David Knoll Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 19, 2014 Share Posted November 19, 2014 David, Try it *without* the "Search both directions" box selected. Then it works (for comps following the verb). We have found an error in Acc 11 with syntax searching; I'll report it now. 1 Link to comment Share on other sites More sharing options...
David Knoll Posted November 19, 2014 Author Share Posted November 19, 2014 Thank you! Do I understand correctly that if I place the complement phrase first, the problem persists? Link to comment Share on other sites More sharing options...
Peter Bekins Posted November 19, 2014 Share Posted November 19, 2014 So, is this search logic indeed asking for all clauses involving the predicate שמר that lack a complement? Based on how NOT works elsewhere, I would most naturally read this as a search for a clause involving the predicate שמר and some other constituent that is not a complement, i.e. שמר [predicate] AND X where X IS NOT [complement]. Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 19, 2014 Share Posted November 19, 2014 (edited) Actually, Mr. Shades (your pic kills me -- it's almost as bad as my horrid studio photo), the search should simply have a negative Comp Phrase. But when I do that in Acc 11, it's gives me a warning. I think it's another bug. [Correction, now in 11.0.2 the warning no longer pops up, so that's fixed] David -- I forgot to try it by building it the other way. I just did and sadly it doesn't work. So, the positive build works, but negating the comp phrase does not. Thanks for finding this bug. Edited November 19, 2014 by Robert Holmstedt Link to comment Share on other sites More sharing options...
David Knoll Posted November 20, 2014 Author Share Posted November 20, 2014 Thank you prof Holmstedt. If I understand correctly the conversation between you and mr. Bekins there is another way to find what I want which still works. If that is the case, could you attach a screenshot of the construct? Link to comment Share on other sites More sharing options...
Peter Bekins Posted November 20, 2014 Share Posted November 20, 2014 My shades pick is much cooler than your yearbook pic, Robert. So if I understand correctly, you are saying NOT applied to a clause or phrase means -, so the search logic is "predicate phrase AND predicate IS שמר AND [- complement phrase]? David, my instinct is that this search should really be a search for a NULL complement, but I don't know if the NULL searching is working right yet and I do not remember how we treated the argument structure of שמר. Valency is probably something that needs to be given a second look once the initial tagging is complete. Link to comment Share on other sites More sharing options...
David Knoll Posted November 20, 2014 Author Share Posted November 20, 2014 My shades pick is much cooler than your yearbook pic, Robert. So if I understand correctly, you are saying NOT applied to a clause or phrase means -, so the search logic is "predicate phrase AND predicate IS שמר AND [- complement phrase]? David, my instinct is that this search should really be a search for a NULL complement, but I don't know if the NULL searching is working right yet and I do not remember how we treated the argument structure of שמר. Valency is probably something that needs to be given a second look once the initial tagging is complete. I wonder if it is wise to tag null complements. If that is done, the user needs to know what valency you decided each verb lexeme to have. Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 20, 2014 Share Posted November 20, 2014 Pete, Perhaps, but only slightly cooler. On valency, John Cook is working on finishing his valency lexicon and then we will indeed make another pass through the database for consistency. David -- Pete's correct. I apologize that I was answering the basic issue of the search syntax and not really paying attention to the verb you were using. שׁמר is always bivalent (i.e., has a complement, even if it is null). In fact, I'm using this search to catch errors and fix them in the tagging right now. 2 Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 20, 2014 Share Posted November 20, 2014 David, There is no theoretically consistent option other than doing so -- if we admit null items (which we do and I discuss why in the background paper that is available in the Accordance resources page, somewhere), then we have to be consistent in our use of them. If a verb has an overt NP comp 7 of every 10 occurrences and an implicit complement can be discerned from the discourse in the 2 more times, then it only stands to reason that the remaining 1 time includes a null comp, unless one can cogently argue that the verb in that case represents a distinct lexical entry (i.e., a monovalent entry). As for providing the user with a guide, that is precisely what John Cook's valency lexicon, which will match the final syntax product, will do. 1 Link to comment Share on other sites More sharing options...
David Knoll Posted November 20, 2014 Author Share Posted November 20, 2014 If that is sold in Accordance the problem is solved. Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 20, 2014 Share Posted November 20, 2014 That is the plan -- to coordinate it via active links. 1 Link to comment Share on other sites More sharing options...
David Knoll Posted November 20, 2014 Author Share Posted November 20, 2014 Could you perhaps post a screenshot with the null complement construct? I think I am doing something wrong here. Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 21, 2014 Share Posted November 21, 2014 Oh boy. It's supposed to look like this, but it's definitely not working. 1 Link to comment Share on other sites More sharing options...
David Knoll Posted November 21, 2014 Author Share Posted November 21, 2014 If any Accordance programmer is following... Link to comment Share on other sites More sharing options...
Robert Holmstedt Posted November 21, 2014 Share Posted November 21, 2014 Don't worry -- I've sent it on the primary syntax programmer. 3 Link to comment Share on other sites More sharing options...
David Knoll Posted November 22, 2014 Author Share Posted November 22, 2014 Thank you! Link to comment Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now