Parsing Bugs by Using Opinion, instead of morphology

[Enoch]

It has come to my attention that sometimes the Instant Details parsing uses opinion, instead of limiting itself to mere identification of forms by morphology. IMHO, if a form can be either imperative or indicative by morphology, the Instant Details should give both imperative & indicative. If a form can be either indicative ("Ye believe") or imperative ("believe!") then the parsing should be neutral; or if Accordance wants to give an opinion, it might underline its interpretation or something.

 

I find that great minds operate in the same way (LOL): and I found my POV posted elsewhere on the internet:

 

http://www.artsci.wustl.edu/~cwconrad/GrkVc.html says:

 

"1. Parsing: in my view it is not the business of a parser to determine whether a middle-passive form should be considered middle rather than passive or vice-versa; the parser should indicate the form only, so that all -μαι, -σαι, -ται, -μην, -σο, -το forms will be marked as Middle, all -θη- forms will be marked as Passive. . . . I know that the Bible programs (Bible Works, Accordance, Gramcord) base their parsing for voice on a judgment call regarding these matters, but I think they are wrong to do so; James Tauber's CCAT GNT Morph has got it right; I wish that the CCAT LXX were done the same way. "

[Timothy Jenney]

Hi, Enoch!

 

You are certainly correct that some forms are ambiguous. However, selecting choosing one form over another is rarely a simple matter of someone's "opinion." A good parser will take into account the context, literary style and date, even the habitual practices of the author and editor before s/he chooses between forms. That expertise and deep familiarity is not something most of our users can bring with them to the study of the text. Yes, I understand that language experts are all capable make their own judgement calls, but then those folks are not generally going to be relying on our Instant Details for parsing information. In cases where someone disagrees with a particular interpretation, I suggest making the correction in a User Note.

 

These parsing choices need not interfere with searching texts, either. Simply use an inflected search for a particular form, rather than a lemma search with a grammatical tag.