The following code seems to produce incorrect or incomplete results:

from greek_inflexion import GreekInflexion
inflexion = GreekInflexion('stemming.yaml', 'STEM_DATA/homer_lexicon.yaml')
print(inflexion.generate('λύω', 'AAI.3S').keys())
# ... fails
print(inflexion.generate('βάλλω', 'AAI.3S').keys())
# ... fails
print(inflexion.generate('βαίνω', 'AAD.2S').keys())
# ... only gives βῆθι, should also give βῆσον (Iliad 8.285)

I am using this excellent library to generate paradigms for various verbs. I am using the lexicon in STEM_DATA/morphgnt_lexicon.yaml, and stemming.yaml from the root of the repo.
I find that generating the form of δίδωμι with parsing PAP.DPF generates this:

διδο{athematic}ούσαις

It would obviously generate the wrong form is I just removed the {athematic} substring.

Looking in stemming.yaml, I find this rule commented out for PAP.DPF:

• "|ο{athematic}>ού<_|σαις"

If I uncomment this in, then the form becomes:

διδούσαις

which is, of course, correct.

I am not sure what other verbs would be affected by this change -- surely some others. I have not yet checked whether this can be applied without negative effects.

Hope this helps, and many thanks to all contributors for all your hard work.

Best wishes,

Ulrik Sandborg-Petersen

Hi,

In STEM_DATA/morphgnt_lexicon.yaml, I found this line:

3-: θησ/θε{athematic}

I am guessing this is intended to say that there are two stems for certain aorist forms, delimited by the slash.

I haven't found any aorist forms in the Greek New Testament (Nestle 1904, to be exact) where either stem is attested. Am I misinterprething this line?

If not, should it be deleted from the morphgnt_lexicon.yaml?

Many thanks.

Best wishes,

Ulrik Sandborg-Petersen

This looks really nice -- thanks for making it open source!

I would suggest adding the following basic installation instructions to the README:

git clone https://github.com/jtauber/greek-inflexion
pipenv install inflexion [... or something like pipenv --python /bin/python3 if that doesn't work]
pipenv shell
./data_test.py
exit

If I'm understanding correctly, then the inflexion and greek-accentuation libraries are automagically located and pulled in from git or something (although it's not obvious to me how it knows) -- it might be helpful to say this in the README, because from reading it, I was under the impression that I would have to go to some other repo and separately install these dependencies.

For some forms, inflexion.conjugate() includes the string "{root}". I'm not sure, but this looks like it may happen when there is no attested root in the lexicon.

For instance:

inflexion.conjugate("ἀνίστημι", "AAN", "PAN", "PAD", "AMD", "PAI", "AAI", "PAP", "AAP", tags={"final-nu-aai.3s"})

is correct for most forms, but has {root} for some of the AMD forms:

AMD.2S: ἀνάστα{root}αι/ἀνάστησαι
AMD.3S: ἀναστα{root}άσθω/ἀναστησάσθω
AMD.2P: ἀναστά{root}ασθε/ἀναστήσασθε
AMD.3P: ἀναστα{root}άσθων/ἀναστησάσθων

I suspect that this is simply that I'm not using the library correctly but when I run

inflexion.parse("ἀπεχόμενος")

I get:

set()

I thought it was a normalisation issue but it doesn't seem to be. What am I doing wrong?

Thanks a lot for writing this software. I was getting to write something similar. I'm glad you beat me to the punch. However, so far I've noticed one bug

in line 109 of the fileformat.py

when the partnum is noun, there is no key in partnum_to_key_regex called noun.

I've been reading over this code for about 5 hours now and I can't figure out what I'm supposed to put in for the second argument

inflexion.conjugate("λύω", "PAI", "AAI", tags={"final-nu-aai.3s"})

It goes into this function

def conjugate(self, lemma, *TVMs, tags=None):

Where is a list of all the possible TVMs?

You never state what that is.