HCMC Journal

Training for new RA

Set up workstations and did svn intro workshop for new RA CT, then sat in while CBR and ML walked through cognate encoding processes. CT will work mornings for a few weeks and will need help initially.

Dictionary rendering fixes

Met with ML and worked through a number of issues, implementing most of them:

• The <headword> element needs to be allowed twice to allow for other orthographies. I’ve made that change in the schema, but haven’t looked at processing yet, so the first usage of this will probably break the build. No decision on what to do with it yet, so the simplest fix will be to use only the first instance, until decisions are made.
• Switched out quotation marking for both languages so that definitions are in single quotes for English and in guillemets for French, with the latter having spaces after the opening one and before the closing one, as is the tradition (browsers don’t add the spaces automatically).
• Standardized our handling of cases where a form is not attested or cannot exist: we provide the table column in the aspect-mood table, but all cells contain only three dashes (a linguistics convention).
• Handled a very interesting scenario in which an entry has multiple verb classes due to having multiple senses. For the Verb Class column, instead of showing only the first, we now show the nearest common ancestor category for all users; then for the full view, we also enumerate the child categories. This was an interesting piece of XPath to figure out. Assuming you have the sequence of <category> elements in document order, then this will get you the nearest ancestor element:
  ($cats[1]/ancestor::* intersect $cats[last()]/ancestor::*)[last()] Although we’re dealing with ancestor nodes, the processor returns the intersection in document order, so it’s [last()] rather than [1].
• Fixed a missing value for pronominal prefixes in the aspect-mood table.