meeting_080820

Meeting notes, Pedagogical programs, 080814

Participants

Biret Ánne, Ciprian, Lene, Saara, Trond.

Status quo

Saara:

  • Added conversion from smenob to nobsme lexica.
  • Added place names

Lene and Trond:

  • Improved translations in sme and fixed the order of them.
  • Looked at the outcome of nobsme.

Biret Anne:

  • Cleaned up mess in question-files

Vasta

Temporally divided according to POS.

    Numerals

    Numeral attributes

    They are not present for all numerals, only for certain long forms.

    Also for the numerals with attribute forms, we are unable to generate them.

    TODO:

    • Trond and Thomas to look at the lexc code.

    Leksa

    Concepts lexicalised in Sámi, not in Norwegian:

    • bártnáš = little boy (liten gutt) guttunge
    • gánddáš lunttaš
    • eanu = mors bror
    • guoika = stryk
    • guoikkaš = lite stryk
    • njavvi = lite stryk / stryk

    The leading principle will be to ask for lexicalised concepts. The family set (and other similar ones?) will be kept, and the phrases of the Norwegian translations are made into nob lemmas.

    The nobsme lexica

    In the nob nouns.xml, every entry has itself as an id.

    For the next conversion, translations of the same nob lemma will be unified under the same lemma.

    nob nouns:

    more than one nob lemma with the same id. we agreed to take the first translation from the sme nouns.xml but this gives ries to a problem:

    Picking dokke only and not dukke is ok but if the second translation is a lemma in other contexts, we have a problem:

    • lieđđi > blomst (i.e. blomst becomes lemma in nobsme)
    • rássi > plante, gress, blomst (i.e. plante becomes lemma in nobsme)

    The problem is then that we loose the blomst > lieđđi translation, when the user answers lieđđi, it is deemed wrong by the computer.

    So, we do not need dukke (which is a formal variant of dokke), but we desperately need the rássi = blomst link under blomst in nobsme.

    The id issue:

    sme:

    <lemma>ášši</lemma>

    <entry id="bassi_time"> <lemma>bassi</lemma>

    nob

    <entry id="hjørne"> <lemma>hjørne</lemma>

    In nob we don´t have to differ between lemmas in sámi with different inflecting, like:

    • vasker = bassi (in sme: bassi_actor)
    • høytid = bassi (in sme: bassi_time)

    Now, we go for the nob id policy (separate id-s all the time).

    Bug: too many å-s removed in the verb file: <tr xml: lang="nob">å gå på ski</tr>

    There are some (surprisingly few) homonyms across POS borders (N, V pleie a case in point. We are satisfied with unique ids modulo each file (i.e. POS), and leave it at that.

    TODO:

    • Lexicon conversion
      • Saara to convert again
      • All to test the result, before we do too much manual work
      • Lene, Trond to manually add the homonyms from the dehálasdubl.txt file.
      • Remove some entries (expressions), add more synonyms.

    The place names

      Today, we have a small placename list (389 entries), with the major names (the list might be too long rather than to short?)

      TODO:

      • Look through the sme propernouns.xml file and prune marginal

      Interface translation

      Run a new conversion of the translations, and keep it up-to-date.

      The documentation was found here: http: //giellatekno.uit.no/ped/OahpaTechnicalDocumentation.html 

      In order to create new strings:
Go to victorio:

cd ped/oahpa
/home/saara/django-trunk/django/bin/make-messages.py -a
   ... then 
cvs up 
   ... to locate the M files. Thereafter check them in:
cvs ci -m "new generated files" locale/fi/LC_MESSAGES/django.po
   ... etc. for the other files. 
After that, we work on the respective django.po files, and check them in again.

      Morfa

      Response

      The issue was discussed on the meeting on 26th of June.

      The feedback, the consonant gradation, diphthongs, rime...

      The lexicon contains this information: 

          <stem class="bisyllabic" diphthong="yes" gradation="yes" rime="u"/>

      TODO:

      • Lene, Trond and Biret Ánne (at least) to present a more concrete model for Morfa user feedback to implement to the next meeting.

      Forward

      We systematise our meeting memos into todo-lists the way the proofing meetings do it. The way forward is thus specified in the task list.

      Next meeting

      Tuesday 080902 at 10.00.

      Task list

      Biret Ánne

      • Clean up more mess in question-files
      • Work on dialogues
      • Look through the sme propernouns.xml file and prune marginal place names
      • Have a look at the wordorder - and suggest for Saara how to do
      • present a more concrete model for Morfa user feedback (with Lene, Trond)

      Ciprian

      • Testing the games : -)
      • Have a look at the ped lexica and compare with the general ones
      • Nothing developmental yet

      Lene

      • Look at the lexicon conversion: manually add the homonyms from the dehálasdubl.txt file.
      • Remove some entries (expressions), add more synonyms.
      • Have a look at the wordorder - and suggest for Saara how to do
      • Look through the sme propernouns.xml file and prune marginal place names
      • present a more concrete model for Morfa user feedback (with Biret Ánne, Trond)

      Saara

      • Run interface generation
        • done
      • Look at the missing feedback in
      • Convert nobsme again (fix å)
        • done
      • Fix bugs in vasta (Bug 714)
      • Away next week

      Trond

      • Play
      • Translate interface
      • Look at the lexicon conversion: manually add the homonyms from the dehálasdubl.txt file.
      • Discuss yr.no with Sjur and Børre, get name pairs
      • present a more concrete model for Morfa user feedback (with Biret Ánne, Lene)