We have two possibilities to publish the repo with Zenodo:

1. transfer to the lexibank organization on Zenodo, we will take care of publication then (authorship etc. won't change, of course)
2. submit directly to zenodo, using Zendodo's possibility to submit repositories along with a new release. This means you have to follow zenodo's guideline and set this up yourself, and we make a release then, so it will be automatically linked on Zenodo.

#2

The current transcription with orthography.tsv is very good, but there are some mistakes that I don't see how to overcome without a rulebased transcriber, e.g.:

• line 1243 is the only non-BIPA conform line, since in gagáhaftjan > ^ g agá h a f t j a n $ > g a ɣ á h a ɸ t j a n the rule á > a never gets applied.
• in line 10 abrs > a β r s doesn't mark the syllabic <r̥> since the r gets clustered to <abr>, so the rule for <brs> doesn't apply.

Not only do these rules cancel each other out but also if I wanted to explicitly mention all these transformations in orthography.tsv, i.e. to hardcode every word (which I tried) then the clts-validation is failing since 1 word = 1 grapheme.

I think it would be nice if I could somehow plug in a third-party transcriber and check those results for clts-conformity. I've added the column IPA in a post-processing step for now. Those are generated with the epitran-library and I've included the transcription rules in the folder etc for more transparency. I like those transcriptions more, but of course that's not worth much if it can't be checked against the clts data.

Instructions can be found also specifically here: https://calc.hypotheses.org/2954

This explains the structure you need to convert data to CLDF. Note that conversion is done programmatically, transparently, not by having a dataset in converted form, but by using code to convert spreadsheet data to cldf.

@martino-vic, if you want to inspect the data through edictor, I recommend pyedictor (pip install pyedictor).

$ edictor wordlist --name streitberggothic

A tutorial on the library is pending (maybe next month as blog post). This will convert the data to the TSV you need for lingpy / edictor. Edictor is here: https://lingulist.de/edev/ (dev version), official version: https://digling.org/edictor . May be interesting to check the data in this form. I can explain another time how.

@martino-vic, do you prefer me to add the first code, or would you rather like to follow the workflow and make this Python script, as indicated by me in issue #1 ?

• files ending in *.bat
• files like cmd.txt

They also reveal your local desktop structure, which I consider a bit private ;)

The workflow is:

1. make a dictionary
2. make some code to map the senses to concepticon in a restricted concept list
3. make some precedence how the wordlist can be manually refined
4. make a combined lexibank script