Estonian Resource Grammar.
Project at the GF Summer School 2013

Inari Listenmaa, Kaarel Kaljurand

Third GF Summer School 2013, Frauenchiemsee, Bavaria

2013-08-30

Presenter Notes

Estonian resource grammar

Estonian
- spoken in Estonia by ~1 million people
- (Balto-)Finnic subgroup of Uralic languages
- related languages: Finnish (5M speakers), Võro (75k speakers)
work on the Estonian RG started at GF Summer School 2011
- modifying the Finnish RG
- progress has been slow
project pages:
- Estonian RG: https://github.com/GF-Estonian/GF-Estonian
- PhrasebookEst: https://github.com/Kaljurand/PhrasebookEst

Presenter Notes

Morphology

Nouns
- 14 cases * 2 numbers
Verbs
- Tense: present, imperfect
- Mood: indicative, imperative, conditional, quotative
- Voice: active, passive
- Nominal forms: {ma infinitive, da infinitive, past participle} * cases ; present participle
Adjectives
- Noun inflection + comparative and superlative

Presenter Notes

Morpho(phono)logy

Variable stress
- not shown in orthography
- kala ['ka.la] vs. banaan [ba.'naan] (mostly words of foreign origin)
No vowel harmony
3-way quantity system
- koli (I, junk), kooli (II, school (gen)), kooli (III, school (part, illat))
- link (nom), lingi (gen), linki (part)
Palatalization
- palk (salary), pa`lk (log)

Presenter Notes

Nouns

param
  Case = Nom | Gen | Part
     | Illat | Iness | Elat | Allat | Adess | Ablat
     | Transl | Ess | Termin | Abess | Comit;

  NForm = NCase Number Case ;

started implementing the noun morphology based on "Heiki-Jaan Kaalep. Eesti käänamissüsteemi seaduspärasused" (HJK EKS)
6 forms needed, others calculated from genitive
currently ignoring parallel forms
rules in HJK EKS depend on singular nominative, its stress, quantity, derivation from a verb, foreignness

Presenter Notes

Smart paradigms

13 templates for creating 6 forms from 1 (sg nom)

-- link -> link, lingi, linki, linki, linkide, linke
hjk_type_VI_link x =
    let
        x_n : Str = weaker x
    in
    hjk_nForms6 x (x_n+"i") (x+"i") (x+"i") (x+"ide") (x+"e") ;

Mapping singular nominative to the templates

/2 II V/               -> type 3
/2 I V/, foreign       -> type 3
/2 III V/              -> type 4a
/3 CV[lmnr]/           -> type 6 seminar
/e/, derived from verb -> type 7
...

TODO: genitive form as an additional argument

lakk, laka, lakka, lakka, lakkade, lakkasid
lakk, laki, lakki, lakki, lakkide, lakkisid

Presenter Notes

Adjectives

param
  AForm = AN NForm | AAdv ;

oper
  Adjective : Type = {s : Degree => AForm => Str} ;

derived from noun forms
comparative derived from genitive
superlative more complex but can be expressed also by kõige + Comparative
adverbial form derived from ablative

Presenter Notes

Verbs

param
    VForm = 
    Presn Number Person | Impf Number Person | Condit Number Person 
    | Imper Number | ImperP3 Number | ImperP1Pl | ImpNegPl 
    | PassPresn Bool | PassImpf Bool 
    | Inf InfForm | PresPart | PastPartAct AForm | PastPartPass AForm ;

    InfForm = 
    InfDa | InfDes | InfMa | InfMas | InfMast | InfMata | InfMaks ;

full conjugation tables from 4 forms (regular verbs) or 8 forms (25 irregular verbs)
- verbs be and go formed separately
- for comparison: Finnish worst case 12 forms
smart paradigms for 1, (2,) 4 and 8 arguments currently
TODO: add quotative, rethink the nominal forms (Inf)

Presenter Notes

Smart paradigms

15 templates for creating 8 forms from 1 forms
1-argument smart paradigms for some verbs
TODO better 1-arg paradigms, decide 2- and 3-arg paradigms

Presenter Notes

Testing morphology

TODO

main questions:
- percentage of words covered by the 1-arg smart paradigms?
- types of remaining problems?
using Filosoft's morph. synthesizer as the gold standard
test vocabulary e.g. from the Estonian WordNet (44k words in 29k synsets)

Presenter Notes

Syntax

(2 years old) copy of the Finnish RG
several changes
- "-ko" particle -> "kas"
- possessive endings -> "minu"/"tema"/"oma"
- reflexive possessive oma as a separate construction in ExtraEst
TODO
- trennbare verben
- also lacking from Finnish (Fin ylläpitää ~ pidän yllä; Est aru saama ~ saan aru)

Presenter Notes

Testing syntax

developing PhrasebookEst in parallel
- this has not highlighted any major problems
adding Estonian to ACE-in-GF
- this has not highlighted any major problems

Presenter Notes

Resources used

Noun morphology
- Heiki-Jaan Kaalep. Eesti käänamissüsteemi seaduspärasused http://kjk.eki.ee/ee/issues/2012/6/156
Verb morphology
- Eesti õigekeelsussõnaraamat (2006) classification system tüüpsõnad http://www.eki.ee/dict/qs/tyypsonad.html
- Eesti keele käsiraamat (2009) for internal paradigm building (VForms -> Verb) http://www.eki.ee/books/ekk09/index.php?p=3&p1=5&id=227
- Estonian WordNet http://www.cl.ut.ee/ressursid/teksaurus/

Presenter Notes

Future work

testing
cleanup
submission to the GF RGL
sharing of code with Finnish
Dict(Eng)Est, from WordNet?

Table of Contents	t
Exposé	ESC
Full screen slides	e
Presenter View	p
Source Files	s
Slide Numbers	n
Toggle screen blanking	b
Show/hide slide context	c
Notes	2
Help	h

Table of Contents

Help