! Present

FMT, VA, MD, MP

! Meeting agenda

* Draft the final work plan, including weekly targets

! Project objectives

* Write a finite-state morphological analyser for Chukchi that implements the verb system, including incorporation
* Get high (>90%) coverage over a medium sized text (~25-30,000 tokens incl. punctuation).

! Tasks 

[To be added to]

* Collect all available corpora in text format
* Implement nominal inflectional morphotactics
* Convert/categorise lexicon
* Add all non-inflecting words
* Implement nominal derivational morphotactics
* Implement morphophonology
* Make sure that N (<=10) sentences are fully covered per week.
* Implement the regular intransitive paradigm
* Implement copula stuff
* Implement the regular transitive paradigm
* Implement incorporation
** Basic version: (transitive verb prefixes) * noun lexicon * transitive verb suffixes, flags/constraints to allow only the "stem" to be incorporated
** Select known restrictions on incorporation and implement them
* Implement verbal derivation
* Define evaluation "deliverables" and weekly milestones
* Write a paper

! Questions

1) For the singulative, do we want to go with the lexicon/dictionary, or do we want to go with the stem ?
** FMT: If singulative is a limited-productivity, or non-productive derivation, go with the derived form in the lexicon.
** MD: There are minimal pairs, plain abs. + singul. abs.

2) Is the lexicon size likely to be big enough to cover all the forms in e.g. the folktale text ?
** FMT: Yes, it has basically all the stems in the paper dictionary.

3) Is it possible for Vasya to get all of the available machine readable corpora 
** MP: Will send the texts

4) We should choose one (20-30k word) text for using for the evaluation/annotation.
** Let's use the one we've been using

5) Is it possible to go simply with the noun classes in Michael's grammar ?

6) Github repo.


! Work plan


!! Community bonding
Week 0  (4th May -- 7th May) Noun stuffs

Week 1  (8th May -- 14th May) Noun stuffs

* Collect all available corpora in text format

Week 2  (15th May -- 21st May) Noun stuffs

* Convert/categorise lexicon
* Add all non-inflecting words

Week 3  (22nd May -- 28th May) Noun stuffs

* Implement nominal inflectional morphotactics

!!! Official start

Week 4  (29th May -- 4th June) 

* Implement morphophonology

[Bachelors thesis, 10-15 hours]

Week 5  (5th June -- 11th June)

* Implement the regular intransitive paradigm

[Bachelors thesis, 10-15 hours]

Week 6  (12th June -- 18th June)

* Implement the regular intransitive paradigm
* Implement copula stuff

Week 7  (19th June -- 25th June)
!!! Phase 1 evaluation

Deliverable: Full noun morphotactics + lexicon + intransitive paradigms (no incorp).

Week 8  (26th June -- 2nd July)

* Implement verbal derivation
* Implement the regular transitive paradigm

Week 9  (3rd July -- 9th July)

* Implement the regular transitive paradigm

Week 10 (10th July -- 16th July)

[Conference, 15-20 hours]

* Implement the regular transitive paradigm

Week 11 (17th July -- 23rd July)
!!! Phase 2 evaluation

Deliverable: Transitive paradigms + verbal derivation

* Implement incorporation

Week 12 (24th July -- 30th July)

* Implement incorporation

Week 13 (1st August -- 6th August)

* Implement incorporation

Week 14 (7th August -- 13th August)

* Implement incorporation

Week 15 (14th August -- 20th August)

* Implement incorporation

Week 16 (21th August -- 27th August)
!!! Final evaluation

Final deliverable: Full analyser.

* Evaluation 
* Write paper

Week 17 (28th August -- 3rd September)

* Write paper

Week 18 (4th September -- 6th September)

* Write paper

! Notes

1. http://wiki.apertium.org/wiki/User:Bandrandr/proposal


Lexicon size (in number of lexemes): 

V         2236
N         2719
A         626
ADV    592
Vbase 67
PRON 48
INTJ    32
PTCP  2



Lexicon examples:

аёпычьылгын (pl. эюпычьыт) колючка

* FMT: Does this have an absolutive that is not singulative?
пипикылгын пипикылгыт "mouse" -- "Singulative" is part of the stem. You can tell because of the i vowel. 
Otherwise it would have to be pepekylgyn

Tools

Here's my tool for converting between Cyrillic orthography and phonemic transcription:

https://github.com/evoling/chukchi-tools

This can also function as an approximate spell-checker, since typos are unlikely to be valid Chukchi.

Github usernames

ftyers
evoling



https://github.com/BasilisAndr/chkchn


https://ling.hse.ru/en/