Cognate Judgments and Alignments: Should I pre-analyze the current cognate sets automatically?

[LinguList]

Right now, we are ready to go for the next step by aligning cognate sets. The question is: should I do an alignment analysis first, just to make it easier to manually correct the alignments, or should we do everything manually?

Furthermore: Cognate IDs will need refinement. Some obvious cases are missing, some seem to cluster non-cognate words. What is needed here, if wrong assignments are found is the following:

• if two sets of words are marked as non-cognate (different cognate ids in the sam concept slot), but should be labelled as cognates, just assign all of the items one of the two different COGIDs.
• if a set of cognate words contains non-cognate items, the non-cognate items need to be assigned new COGIDs. Now, since it is not easy to know which ID can be used, I added the following functionality:
  • if you want to add a completely new COGID, just put "?" in the field, and press "ENTER". The next "free" ID will then be insterted into the COGID field.
  • having inserted the new COGID, press the "refresh" button on the right menu panel. This forces the app to re-calculate all cognate sets and recognize this in the alignment display mode.

Generally, I suggest we start with some obvious cases for alignment during the next time and test how the app works and what it is still missing.

[LinguList]

closing this, since we need a new workflow, as discussed today