150813

Lene, Marja-Liisa, Trond, Ciprian, Erika

• Progress for MT-project

Progress for MT-project

Analyser has it's part in a bigger picture

1. Nouns soon ready
   1. 300 stems still waiting for classification (we will wait with those until we find them in the corpus)
2. Verbs, work is started
   1. even-syllable verbs will be modelled the same way as for even-syllable nouns
   2. look at the odd-syllable verbs
      1. how many types, example verbs
      2. add the correct lexicon to the verbstems (ML)
3. Propernouns
   1. Erika and Lene has started to make lexicons
4. Missing list, from corpus analysis
   1. look at other PoS
   2. correct PoS, both in FST and dict
   3. Erika and Lene will work with this
5. Morphophonology
6. Adjectives, not done yet
   1. Principle: VEESKIR and A_VEESKIR, etc. +A: --> X_VEESKIR (and Attr)
   2. VEESKIR --> +N: X_VEESKIR
   3. Have a look at fkv if you are curious
   4. Lene can have a look at them
7. Possessivesuffixes
   1. Usage more sme-like than fin-like
   2. Wait until they show up in text, look at it then
8. Derivation
   1. Do when other parts of FST are settled
   2. "... and when they show up in text (e.g. vuotâ)"

Bidix:

1. Fix errors (someone with smn-knowledge, Miina?)
2. Translate Inari Saami to North Saami (Erika?)
3. Ciprian will complete his work
   1. the bidix-data is planed to be done by the end of September at latest
   2. what is still to do:
      1. disambiguate the fin2smn data
      2. add pos values to fin and smn data (in order to make the intersection with the sme2fin easier, i.e., not too many ambiguities)
      3. intersect sme2fin with fin2smn
      4. transform the data into a bidix format

MT-system:

1. Kevin and Francis will get the tags etc. to function
2. We have a setup for testing (from smesma)
3. Transfer-rules for syntax and tags (linguistic work)
4. Collect parallel texts for testing and development