oji_timeline.txt

Timeline and responsibilities for Ojibwe finite state morphology

Primary development responsibility is currently assigned to Dustin Bowers

NOTE: DEVELOPMENT IN OJI HAS BEEN SUSPENDED (03/29/2016) TEMPORARILY IN FAVOR OF OTW. CURRENT MATERIALS ARE APPROPRIATE FOR NON-NORTHERN DIALECTS (LEXICON AND MORPHOLOGY), AND THERE HAS BEEN SOME WORK DEDICATED TO MAKING MORPHOPHONOLOGY FOR THE SYNCOPATED EASTERN DIALECTS. A "GLOSSING" APPROACH HAS BEEN TAKEN TOWARDS THE MORPHOLOGY, AND WORK HAS BEEN DONE TOWARDS TRANSLATING THAT TO INTO MORE ABSTRACT TAGS. OTW HAS FULLER VERSIONS OF PHONOLOGY, MORPHOLOGY, TAG ABSTRACTION, AND AN OTW-SPECIFIC LEXICON.

To do: 
    (not in Giella ... but must devote 2 days to an input converter for Nish dict)
    migrate oji to otw
    finalize the yaml files (more complex than it looks, see next section)
    write adjustment rules for morpho-ortho-phono system
    get full coverage of the inflectional system in lexc
    finish writing the converter for abstract morphological features
    derivational parser for keyword assignment/edification

Priorities:
    Pre-Nishnaabemwin teg:
        make adjustment rules so Alan has something to show
        to extent possible, increase inflectional coverage to boost above
        write input converter for Nish dict
    April:
        1: current yaml right side standardized
        2: glossing to abstract converter
        3: oji -> otw
        4: full inflectional coverage in yaml
        5: full inflectional coverage in lexc
        6: further adjustment rules
        7: derivational parser


###Some notes on how the work flow is set up:
YAML files are exhaustive and generated by script
    The paradigms in reference grammars are often missing or incomplete.
        Rand's Nish database might have some full pre-checked paradigms.
    Hence, there may not be a pre-existing standard to copy into a YAML file
    In many cases I need to just hypothesize what the form is.
    Some pretty disastrous phonology has also struck this language.
    To keep things simple, I'm pretending that it hasn't happened for now.
Lexc files are written by hand
    Often while writing Lexc files, I learn more about the morphology 
    This creates lag between the YAML files and Lexc
        (lag can occur on the right or left side of a line in YAML)
        (lag can also occur on whether a line should exist in YAML)

###oji->otw migration
move the current contents of oji
swap out stems/nouns.lexc, stems/verbs.lexc

###YAML finalizing (primarily right-side)
dependent V-initial nouns need to be correctly specified
dependent nouns need to require a prefix
double check all pre-syncope phonology is right (VAIs totally lack this)
create "canary" files, consisting of just core morphophonological types
VTA, VTI, VII YAML scripts must be prepared and run
VTA prep: process the comparison between Rand's book and lexc output 
!!!->make decision on what to list for post-syncope (current) language
duplicate all files into abstract and glossing versions

###Morpho-Ortho-Phono adjustments
convert adjustment rules in yaml scripts to xfst replace rules

###Inflectional lexc coverage:
Major aspects of the inflectional system (lexc files):
    X: Not done
    O: Done
    -: Partially done

NI  O
NA  O
    IND CNJ IMP 
VAI O   -   O (plurality of lexicon)
VTA O   -   X (largest infl system)
VTI X   X   X
VII X   X   * doesn't exist 

###Abstract Morphological Feature Conversion
Streamline rules into one stream, POS fork at end