MED granularity?

[HedvigS]

I finally updated my R scripts that cobble together a table from the glottolog data in a format that I prefer (here). As I was shifting from a langouds.csv and treedb approach to using almost only the glottolog cldf I noticed a few smaller changes. It all went fine, I think i understood most of the decisions.

I was just curious, will the cldf release ever contain more fine-grained categories for "med" below wordlist?

[xrotwang]

"wordlist" is the "lowest" type in the document types recognized by Glottolog. So providing more granularity "below" would only be possible, if the whole database would be annotated with new document types. I don't think this going to happen.

Why would you need this and which types would you suggest?

[HedvigS]

Right, okay. Will med ever include any more granularity, or is there another way of getting the old "doctype" from the cldf release?

I just found it a bit confusing to only work with these 5 med classes instead of the previous 16 doctypes. Is "specific_feature" lumped into med class "phonology/text" or "wordlist or less"? I take it "phonology/text" then if wordlist is the lowest?

[HedvigS]

Ah, right, Just found "From highest to lowest, the ranking is grammar, grammar sketch, dictionary/phonology/specific feature/text, wordlist, followed by the remaining document types.".

That means "phonology/text" includes "specific_feature". Right, okay. That makes the med scale a bit more clear. It makes me tempted to rename "phonology/text" into "dictionary/phonology/specific feature/text" and try and split "wordlist or less" into "wordlist" and the others. It also means that "Ethnographic Work" is under wordlist, correct?

[xrotwang]

Here's the code that computes med_type: https://github.com/glottolog/pyglottolog/blob/master/src/pyglottolog/references/bibfiles.py#L267-L278 and here's the doctypes: https://github.com/glottolog/glottolog/blob/master/config/document_types.ini with ordering defined by rank.

[xrotwang]

So yes, ethnographic work is ranked 3, below wordlist, ranked 10.

[xrotwang]

Getting the document type(s) for a MED from the CLDF release would work as follows:

1. You determine the BibTeX key of the MED, given in the Source column of the relevant row in values.csv, e.g. hh:s:Crevels:Canichana for cani1243, from

   cani1243-med,cani1243,med,grammar sketch,med-grammar_sketch,,hh:s:Crevels:Canichana,

2. You lookup the hhtype field of this reference in sources.bib:

   @incollection{hh:s:Crevels:Canichana,
      author = {Crevels, Mily},
      ...
      hhtype = {grammar_sketch},
      ...
      year = {2012}
   }

Note that a reference might be tagged with more than one document type.

[HedvigS]

Okay, so specifically "wordlist" is not the lowest type and the more granular hhtype is retrievable. Thanks!

[xrotwang]

Ah, sorry, I was wrong saying "wordlist is the lowest type". Will make sure the order at https://glottolog.org/meta/glossary#Doctype is by rank and not alphabetical. See https://github.com/clld/glottolog3/issues/125

[HedvigS]

No worries, thanks :)!