Stephen Carlson wrote:
Jonathan Robie wrote:We're up to 295 now. But it may contain some false positives, I may need to restrict it further, I have only briefly scanned the results.
Three false positives (in the form of double counting) are where an articular δύνασθαι has an infinitive complement in 2 Cor 1:4, Eph 6:11, and Phil 3:21. Similarly, 2 Tim 3:15 governed by the articular participle τἀ δυνάμενα (the query should not select plural determiners); also in Mark 10:32 τὰ μέλλοντα αὐτῷ συμβαίνειν.
Same issue with τὸ δεῖν in Luke 18:1. Similarly, two more with δεῖ in 1 Thess 4:1, except that the syntactic structure is also misparsed (the interrogative is actually τὸ πῶς, see BDAG πῶς 1bα, 901b).
Another governed infinitive miscounted as an articular infinitive is χωρίζεσθαι in Acts 18:2 τὸ διατεταχέναι Κλαύδιον χωρίζεσθαι πάντας τοὺς Ἰουδαίους ἀπὸ τῆς Ῥώμης.
This is tremendously helpful to me. I have a shiny new toy to play with, and I want to figure out what it's good for, how best to use it, what the limitations are, and how it can be improved. I also want to understand the relative advantages of the three different models for New Testament Greek treebanks that are available as open data. This thread is a great opportunity to explore that (and after we're done here, maybe we can find other things to explore, one at a time and at a pace I can manage with my other responsibilities).
Do you like the format I used for reporting results? I'm slowly inching my way toward an environment that allows others do do queries and see results, and this is a format I can use fairly generically.
Except for the misparsing, these are all things I think I can fix in the query, probably later today if I can squeeze in some time during a work break, otherwise tomorrow. I have to work on these things opportunistically, in small blocks of time.
Incidentally, I'm more concerned about false negatives than false positives, because you can see the false positives and filter them out, but you don't see the false negatives. And I know there is at least one missing result, there could be others.
Stephen Carlson wrote:Rom 1:11 is misparsed: δέ should join its clause at a higher level than the preceding articular infinitive. Same with Rom 15:8. Something's going on with λατρεύειν in Luke 1:74, but I don't have the time to figure it out. These last three should be passed on to the maintainer of the syntax trees.
Thanks - I will enter bugs for these and the misparsing you mentioned above. I think all of these treebanks have been used more for analysis than for systematic queries at this point, and as we use them for queries, we will find - and fix - inconsistencies like this. The more we use them, the more we will fix.
ἐξίσταντο δὲ πάντες καὶ διηποροῦντο, ἄλλος πρὸς ἄλλον λέγοντες, τί θέλει τοῦτο εἶναι;