Various metadata-related fixes and improvements.

[tzakharko]

• If a value list variable has no values (all missing), the json value list metadata is now serialized as an empty list [] for consistency
• NPStructurePresence is no longer classified as a PerLanguageSummaries dataset
• LID field was sometimes serialized as string, fixed
• Missing glottocodes were sometimes serialized as explicit "NA" string, fixed
• Multiple metadata fixes:
  • Added value list descriptions for PhonologicalFusion::FusionBinned6 and all variables that rely on it (such as GrammaticalMarkers::MarkerFusionBinned6)
  • Added value list descriptions for PositionalBehavior::MarkerBehaviorBinned4 and all variables that rely on it (such as GrammaticalMarkers::MarkerBehaviorBinned4)
  • Value list description for LocusOfMarking::LocusOfMarkingBinned5 was missing the value 'FloatingorClitic', fixed (this also fixes all the variables that rely on it, such as GrammaticalMarkers::LocusOfMarkingBinned5)
  • Fixed value list description for GrammaticalMarkers::MarkerPositionBinned4
  • Fixed value list description for GrammaticalMarkers::MarkerPositionBinned5
  • Fixed data type of GrammaticalMarkers::MarkerExpressesMultipleCategories to be logical
  • Added value list descriptions for ClauseLinkage::IntuitiveClassification, value "?" is now recoded as NA (missing)
  • Added value list descriptions for multiple fields in ClauseLinkage where they were missing. The fields are: AnticipatoryArgumentMarking, CataphoraConstraints, CategoricalSymmetry, ClauseLayer, ClausePosition, Embedding, ExtractionConstraints, FinitenessSimplified, FocusMarkingInDependent, FocusMarking, IllocutionaryMarking, IllocutionaryScope, InterpropositionalSemanticRelation, ReferenceTrackingSystem, TenseMarking and TenseScope
  • Fixed the value list description for ClauseWordOrder::WordOrderAPLex
  • Fixed the value list description for SemanticClass::SemanticClassBinned
  • Removed invalid values from GrammaticalRelationsRaw::SelectedArguments::SemanticCondition
  • Fixed the value list descriptionb for Register::OriginContinent
  • Computed variables in GrammaticalMarkersPerLanguage now have correct value list metadata
  • Computed variables in LocusOfMarkingPerLanguage now have correct value list metadata
  • Computed variables MorphologyPerLanguage::HasAny* are now correctly annotated as logical
  • Computed variables NPStructurePerLanguage::NPHas* are now correctly annotated as logical
  • NPStructurePerLanguage::NPStructureID is now correctly annotated as integer
  • Computed variables in VerbInflection* summary datasets now have correct value list metadata

[tzakharko]

Updated the glottocode for Ahom and Eastern Armenian in the database, will be reflected on the next data export. The only language with a missing glottocode is now Serbian Torlak, which I could not locate in Glottolog (there is a note mentioning this variety under standard Serbian though)

[xrotwang]

@tzakharko see https://github.com/glottolog/glottolog/issues/815#issuecomment-1048583147

[tzakharko]

@tzakharko see glottolog/glottolog#815 (comment)

Thanks, I forwarded this to the team, will update the glottocode once a decision is made.

[tzakharko]

@xrotwang All the issues you have found so far should be fixed now. If you don't see any other low-hanging issues, I would like to publish a bugfix release based on the contents of this branch.

[xrotwang]

It would be nice, if the issue(s) with the bibliography could be addressed, too. That would save me a couple of lines cleaning it up. Otherwise, yes, all issues addressed.

[tzakharko]

It would be nice, if the issue(s) with the bibliography could be addressed, too. That would save me a couple of lines cleaning it up.

Yes, of course! Adopted your version of the bibliography in latest commit.

[xrotwang]

So the bibliography isn't pulled out of AUTOTYP but only exists as this BibTeX file anyway?

[tzakharko]

It is pulled from the database, but that particular part of the database is currently on life support and is not actively updated. This is another part that will require an overhaul, as some sources are "hidden" in the comments or special fields of various data file and are not currently exported. It is a lot of legacy to deal with..

For now your curated file works great, and as we continue modernising the database structure we will put a new bibliogaphy-generating mechanism in place. Making use of the excellent Glottolog reference database sounds like an obvious choice here.