Skip to content

Commit

Permalink
Extend contlexes
Browse files Browse the repository at this point in the history
  • Loading branch information
rueter committed Nov 20, 2024
1 parent cf854c7 commit ecdbca1
Show file tree
Hide file tree
Showing 5 changed files with 225 additions and 67 deletions.
13 changes: 12 additions & 1 deletion src/fst/morphology/affixes/adjectives.lexc
Original file line number Diff line number Diff line change
Expand Up @@ -18,6 +18,10 @@ LEXICON A_NAGRIZH !nagriž:nagr
N_NAGRIZH ;
+Comp:he%>mb N_POIG ;

LEXICON A_GOL1L1 ! golʼlʼ:gol
N_GOL1L1 ;
LEXICON A_VIL1L1 ! vilʼlʼ:vil
N_VIL1L1 ;
LEXICON A_RUSKED !!≈ * **@CODE@** = goes to NMN\_RUSKED
NMN_RUSKED ;

Expand All @@ -27,6 +31,13 @@ LEXICON A_MUNA !!≈ **@CODE@** muna:mun
N_MUNA ;
+Comp:e%>mb N_POIG ;

LEXICON A_MOENAE !!≈ **@CODE@** möhä:möh
! _ä/än/äd/Q/id
!Sg+Nom/Sg+Gen/Sg+Par/Sg+Ill/Pl+Par
N_MOEAE ;
+Comp:e%>mb N_POIG ;


!LEXICON A_NADO !!≈ * **@CODE@** = nado:nado
! N_NADO ;
!+Comp:%>mb N_POIG ;
Expand All @@ -53,7 +64,7 @@ LEXICON A_MARJ !!≈ * **@CODE@** = marj:marj
LEXICON A_JAUH !!≈ * **@CODE@** = jauh:jauh
!(-on\,\ -oid\)
N_JAUH ;
+Comp:%>mb N_POIG ;
+Comp:o%>mb N_POIG ;

LEXICON A_OIGED !!≈ * **@CODE@** = oiged:oig%{eØ%}d
!oi|ged (-ktan, -ktoid)
Expand Down
101 changes: 80 additions & 21 deletions src/fst/morphology/affixes/nouns.lexc
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,19 @@ LEXICON N_MUNA ! Stem vowel alternation ZERO-A-ZERO muna:mun
:%>a PL-ACC-SUF ;
: PL-OBLIQUE ;

LEXICON N_MOEHAE ! Stem vowel alternation ZERO-A-ZERO möhä:möh
!!€gt-norm: Noun ' / '
!!€ : +N+Sg+Nom
!!€ n: +N+Sg+Gen
!!€ d: +N+Pl+Par
!:ä ATTESTED-NOUN-STEM ;
:ä SG-NOM-SUF ;
:%>ä SG-OBLIQUE ;
! Plural
:%>ä PL-NOM-SUF ;
:%>ä PL-ACC-SUF ;
: PL-OBLIQUE ;


LEXICON N_SANA
! sana = pyrrhic <
Expand Down Expand Up @@ -121,17 +134,6 @@ R ; ! xxx check
:%>a PL-ACC-SUF ;
:o PL-OBLIQUE ;

LEXICON N_PAVEL
!: N_0/RmVow ;
: SG-NOM-SUF ;
+Sg+Ill:%^RmVow%>aha K ;
+Sg+Par:%^RmVow%>ad K ;
:%^RmVow%>a SG-OBLIQUE_NO_DERIV/ILL/PAR ;
! Plural
!:%>a PL-NOM-SUF ;
!:%>a PL-ACC-SUF ;
!:o PL-OBLIQUE ;

LEXICON N_FILIP
!: N_ ;
: SG-NOM-SUF ;
Expand Down Expand Up @@ -195,9 +197,9 @@ LEXICON N_LIND ! lind:lind
LEXICON N_0/un/ud/hu/uid ! lind:lind
! N_0/un/ud/hu/uid
!!€gt-norm: Noun ' / '
!!€ : +N+Sg+Nom
!!€ n: +N+Sg+Gen
!!€ d: +N+Pl+Par
!!€ lind: +N+Sg+Nom
!!€ lindun: +N+Sg+Gen
!!€ linduid: +N+Pl+Par
: SG-NOM-SUF ;
+Sg+Ill:%>hu K ;
:%>u SG-OBLIQUE ;
Expand Down Expand Up @@ -645,9 +647,9 @@ R ; ! xxx check

LEXICON N_OIGED !oiged:oiged
!!€gt-norm: Noun ' / '
!!€ : +N+Sg+Nom
!!€ n: +N+Sg+Gen
!!€ d: +N+Pl+Par
!!€ oiged: +N+Sg+Nom
!!€ oiktan: +N+Sg+Gen
!!€ oiktoid: +N+Pl+Par
!oi|ged (-ktan, -ktoid)
: SG-NOM-SUF ;
!^PEN^DEVOICE^RmVow^DEVOICE
Expand All @@ -658,6 +660,17 @@ R ; ! xxx check
:%^PEN%^DEVOICE%^RmVow%^DEVOICE%>a PL-ACC-SUF ;
:%^PEN%^DEVOICE%^RmVow%^DEVOICE%>o PL-OBLIQUE ;

LEXICON N_PAVEL
!: N_0/RmVow ;
: SG-NOM-SUF ;
+Sg+Ill:%^RmVow%>aha K ;
+Sg+Par:%^RmVow%>ad K ;
:%^RmVow%>a SG-OBLIQUE_NO_DERIV/ILL/PAR ;
! Plural
!:%>a PL-NOM-SUF ;
!:%>a PL-ACC-SUF ;
!:o PL-OBLIQUE ;

LEXICON N_IKUN !ikun:ikun
!!€gt-norm: Noun ' / '
!!€ ikun: ikun+N+Sg+Nom
Expand All @@ -673,11 +686,27 @@ R ; ! xxx check
:%^RmVow%>a PL-ACC-SUF ;
:%^RmVow%>o PL-OBLIQUE ;

LEXICON N_TUETAER !tütär:tütär
!!€gt-norm: Noun ' / '
!!€ tütär: tütär+N+Sg+Nom
!!€ tütren: tütär+N+Sg+Gen
!!€ tütrid: tütär+N+Pl+Par
!
: SG-NOM-SUF ;
: SG-PAR-SUF_t ;
!^RmVow
:%^RmVowe SG-OBLIQUE ;
R ; ! xxx check
! Plural
:%^RmVow%>e PL-NOM-SUF ;
:%^RmVow%>e PL-ACC-SUF ;
:%^RmVow%> PL-OBLIQUE ;

LEXICON N_SAMAL !samal:samal syncope
!!€gt-norm: Noun ' / '
!!€ : +N+Sg+Nom
!!€ n: +N+Sg+Gen
!!€ d: +N+Pl+Par
!!€ samal: +N+Sg+Nom
!!€ samlon: +N+Sg+Gen
!!€ samloid: +N+Pl+Par
: SG-NOM-SUF ;
: SG-PAR-SUF_t ;
:%^RmVow%^DEVOICE%>o SG-OBLIQUE ;
Expand Down Expand Up @@ -716,7 +745,7 @@ R ; ! xxx check
! Plural
:i PL-NOM-SUF ;
:i PL-ACC-SUF ;
:i PL-OBLIQUE ;
: PL-OBLIQUE ;

LEXICON N_1/en/ed/ehe/id ! apʼ:ap
!!€gt-norm: Noun ' / '
Expand Down Expand Up @@ -746,6 +775,34 @@ R ; ! xxx check
:de PL-ACC-SUF ;
:nz PL-OBLIQUE ;

LEXICON N_VIL1L1 ! vilʼlʼ:vil
!!€gt-norm: Noun ' / '
!!€ vilʼlʼ: vilʼlʼ+N+Sg+Nom
!!€ villän: vilʼlʼ+N+Sg+Gen
!!€ villid: vilʼlʼ+N+Pl+Par
:ʼlʼ SG-NOM-SUF ;
+Sg+Ill:läha K ;
:lä SG-OBLIQUE ;
:ʼlʼ R ; ! xxx check
! Plural
:lä PL-NOM-SUF ;
:lä PL-ACC-SUF ;
:l PL-OBLIQUE ;

LEXICON N_GOL1L1 ! golʼlʼ:gol
!!€gt-norm: Noun ' / '
!!€ golʼlʼ: golʼlʼ+N+Sg+Nom
!!€ golʼlʼan: golʼlʼ+N+Sg+Gen
!!€ gollid: golʼlʼ+N+Pl+Par
:ʼlʼ SG-NOM-SUF ;
:ʼlʼa SG-OBLIQUE ;
:ʼlʼ R ; ! xxx check
! Plural
:ʼlʼa PL-NOM-SUF ;
:ʼlʼa PL-ACC-SUF ;
:l PL-OBLIQUE ;


LEXICON N_KODI !kodi:kodi
!!€gt-norm: Noun ' / '
!!€ : +N+Sg+Nom
Expand Down Expand Up @@ -1359,6 +1416,8 @@ LEXICON POSS_SUF !based on kin terminology, reflexive pronoun, personal pronouns
+Sg+All+PxSg2:%>leiž K ;!Tataleiž
+Sg+All+PxSg3:%>leze K ;!mamaleze
+Sg+All+PxPl3:%>leze K ;!mamaleze
+Sg+Ade+PxSg3:%>laz K ;!mamalaz
+Sg+Ade+PxPl3:%>laz K ;!mamalaz
+Sg+Ela+PxSg1:%>sainpäi K ;
+Sg+Ela+PxSg2:%>saižpäi K ;
+Sg+Ela+PxSg3:%>sazepäi K ;
Expand Down
110 changes: 74 additions & 36 deletions src/fst/morphology/affixes/pronouns.lexc
Original file line number Diff line number Diff line change
Expand Up @@ -11,47 +11,58 @@ LEXICON PRONOUN-TYPES
! COLL-PRON ;
DEM-PRON ;
iče+Pron+Refl+Nom:iče K ;
ičein+Pron+Refl+Sg1+Gen:ičein K ;
ičein+Pron+Refl+Sg1+Ill:ičhein K ;
ičein+Pron+Refl+Sg1+Ine:ičesain K ;
ičein+Pron+Refl+Sg1+Ela:ičesainpäi K ;
ičein+Pron+Refl+Sg1+Par:ičtain K ;
ičein+Pron+Refl+Pl1+Gen:ičemoi K ;
ičein+Pron+Refl+Pl1+Ill:ičhemoi K ;
ičein+Pron+Refl+Pl1+Ine:ičesamoi K ;
ičein+Pron+Refl+Pl1+Ela:ičesamoipäi K ;
ičein+Pron+Refl+Pl1+Par:ičtamoi K ;
ičein+Pron+Refl+Pl1+All:ičelemoi K ;
ičeiž+Pron+Refl+Sg2+Gen:ičeiž K ;
ičeiž+Pron+Refl+Sg2+Ine:ičesaiž K ;
ičeiž+Pron+Refl+Sg2+Ela:ičesaižpäi K ;
ičeiž+Pron+Refl+Sg2+Ill:ičheiž K ;
ičeiž+Pron+Refl+Sg2+Par:ičtaiž K ;
ičeiž+Pron+Refl+Pl2+Gen:ičetoi K ;
ičeiž+Pron+Refl+Pl2+Ine:ičesatoi K ;
ičeiž+Pron+Refl+Pl2+Ela:ičesatoipäi K ;
ičeiž+Pron+Refl+Pl2+Ill:ičhetoi K ;
ičeiž+Pron+Refl+Pl2+All:ičeletoi K ;
ičeiž+Pron+Refl+Pl2+Par:ičtatoi K ;
ičeiž+Pron+Refl+Pl2+Apr2:ičetoinoks K ;
ičeze+Pron+Refl+Sg3+Gen:ičeze K ;
ičeze+Pron+Refl+Sg3+Ill:ičheze K ;
ičeze+Pron+Refl+Sg3+Com:ičezenke K ;
ičeze+Pron+Refl+Sg3+Ine:ičesaze K ;
ičeze+Pron+Refl+Sg3+Ela:ičesazepäi K ;
ičeze+v1+Pron+Refl+Sg3+Tra:ičekseze K ;
ičeze+v2+Pron+Refl+Sg3+Tra:ičeksaze K ;
ičeze+Pron+Refl+Sg3+All:ičeleze K ;
ičeze+Pron+Refl+Sg3+Par:ičtaze K ;
ičeze+Pron+Refl+Pl3+Gen:ičeze K ;
ičeze+Pron+Refl+Sg3+Apr2:ičezennoks K ;
ičeze+Pron+Refl+Sg3+Egr:ičezennopäi K ;
ičeze+Pron+Refl+Pl3+Apr2:ičezennoks K ;!Do we want + OR +SP3
iče+Pron+Refl+Ade:ičel K ;
iče+Pron+Refl+Sg1+Gen:ičein K ;
iče+Pron+Refl+Sg1+Ade:ičelain K ;
iče+Pron+Refl+Sg1+All:ičelein K ;
iče+Pron+Refl+Sg1+Ill:ičhein K ;
iče+Pron+Refl+Sg1+Ine:ičesain K ;
iče+Pron+Refl+Sg1+Ela:ičesainpäi K ;
iče+Pron+Refl+Sg1+Par:ičtain K ;
iče+Pron+Refl+Sg1+Egr:ičeinnopäi K ;
iče+Pron+Refl+Pl1+Gen:ičemoi K ;
iče+Pron+Refl+Pl1+Ill:ičhemoi K ;
iče+Pron+Refl+Pl1+Ine:ičesamoi K ;
iče+Pron+Refl+Pl1+Ela:ičesamoipäi K ;
iče+Pron+Refl+Pl1+Par:ičtamoi K ;
iče+Pron+Refl+Pl1+Ade:ičelamoi K ;
iče+Pron+Refl+Pl1+All:ičelemoi K ;
iče+Pron+Refl+Sg2+Gen:ičeiž K ;
iče+Pron+Refl+Sg2+Ade:ičelaiž K ;
iče+Pron+Refl+Sg2+All:ičeleiž K ;
iče+Pron+Refl+Sg2+Ine:ičesaiž K ;
iče+Pron+Refl+Sg2+Ela:ičesaižpäi K ;
iče+Pron+Refl+Sg2+Ill:ičheiž K ;
iče+Pron+Refl+Sg2+Par:ičtaiž K ;
iče+Pron+Refl+Pl2+Gen:ičetoi K ;
iče+Pron+Refl+Pl2+Ine:ičesatoi K ;
iče+Pron+Refl+Pl2+Ela:ičesatoipäi K ;
iče+Pron+Refl+Pl2+Ill:ičhetoi K ;
iče+Pron+Refl+Pl2+Ade:ičelatoi K ;
iče+Pron+Refl+Pl2+All:ičeletoi K ;
iče+Pron+Refl+Pl2+Par:ičtatoi K ;
iče+Pron+Refl+Pl2+Apr2:ičetoinoks K ;
iče+Pron+Refl+Sg3+Gen:ičeze K ;
iče+Pron+Refl+Sg3+Ill:ičheze K ;
iče+Pron+Refl+Sg3+Com:ičezenke K ;
iče+Pron+Refl+Sg3+Ine:ičesaze K ;
iče+Pron+Refl+Sg3+Ela:ičesazepäi K ;
iče+v1+Pron+Refl+Sg3+Tra:ičekseze K ;
iče+v2+Pron+Refl+Sg3+Tra:ičeksaze K ;
iče+Pron+Refl+Sg3+All:ičeleze K ;
iče+Pron+Refl+Sg3+Ade:ičelaze K ;
iče+Pron+Refl+Sg3+Par:ičtaze K ;
iče+Pron+Refl+Pl3+Gen:ičeze K ;
iče+Pron+Refl+Sg3+Apr2:ičezennoks K ;
iče+Pron+Refl+Sg3+Egr:ičezennopäi K ;
iče+Pron+Refl+Pl3+Apr2:ičezennoks K ;!Do we want + OR +SP3
nece+Det:ne DET_NECE ;! occurs also as a determiner
nece+Pron:ne DET_NECE ;
kaikutte+Pron:kaiku A_IN_TTE ;
mitte+Pron:mi PRON_IN_TTE ;
nimitte+Pron:nimi PRON_IN_TTE ;
mitte-se+Pron:mi PRON_IN_TTE_SE ;
mitte-ni+Pron:mi PRON_IN_TTE_NI ;
ken-ni+Pron+Indef+Sg+Nom:ken-ni K ;
ken-ni+Pron+Indef+Sg+Gen:kenen-ni K ;
ken-ni+Pron+Indef+Sg+Com:kenenke-ni K ;
Expand All @@ -68,7 +79,33 @@ kuna+Pron+Rel:kuna K ;
LEXICON PRON_IN_TTE !mitte
N_IN_TTE ;
! 2024-11-15
LEXICON PRON_IN_TTE_SE !mitte-se
+Sg+Nom:tte-se K ;
+Sg+Ill:ččehe-se K ;
+Sg+Par:ttušt-se K ;
+Sg+Ade:ččel-se K ;
+Sg+All:ččele-se K ;
+Sg+Gen:ččen-se K ;
+Sg+Tra:ččeks-se K ;
+Sg+Ine:ččes-se K ;
+Pl+Nom:ččed-se K ;
+Pl+Par:ččid-se K ;
+Pl+Ade:ččil-se K ;
+Pl+Ine:ččiš-se K ;

LEXICON PRON_IN_TTE_NI !mitte-ni
+Sg+Nom:tte-ni K ;
+Sg+Ill:ččehe-ni K ;
+Sg+Par:ttušt-ni K ;
+Sg+Ade:ččel-ni K ;
+Sg+All:ččele-ni K ;
+Sg+Gen:ččen-ni K ;
+Sg+Tra:ččeks-ni K ;
+Sg+Ine:ččes-ni K ;
+Pl+Nom:ččed-ni K ;
+Pl+Par:ččid-ni K ;
+Pl+Ade:ččil-ni K ;
+Pl+Ine:ččiš-ni K ;

LEXICON PERS
!! @LEXNAME@ split in persons
Expand Down Expand Up @@ -96,6 +133,7 @@ LEXICON PersSg2
+Pers+Sg2:u ACC-SUF ;
+Pers+Sg2+Ade:ai K ;! sinai
+Pers+Sg2+All:ei K ;
+Pers+Sg2+Ill:uhuiž K ;
+Pers+Sg2:u OBLIQUE_CASES_FOR_PRONOUNS ; ! sinunnoks..

LEXICON PersSg3
Expand Down
12 changes: 9 additions & 3 deletions src/fst/morphology/affixes/quantifiers.lexc
Original file line number Diff line number Diff line change
Expand Up @@ -3,17 +3,23 @@
LEXICON NUM_KAKS1
!! **LEXICON @LEXNAME@**
+Sg+Nom:ksʼ K ;
+Sg:hte OBLIQUE_CASES_FOR_PRONOUNS ;
+Sg+Ill:hthe K ;
:hte SG-OBLIQUE ;

LEXICON NUM_YKS1
!! **LEXICON @LEXNAME@**
+Sg+Nom:ksʼ K ;
+Sg:hte OBLIQUE_CASES_FOR_PRONOUNS ;
+Sg+Ill:hthe K ;
:hte SG-OBLIQUE ;

LEXICON NUM_UEHESA
!! **LEXICON @LEXNAME@**
+Sg+Nom: K ;
+Sg: OBLIQUE_CASES_FOR_PRONOUNS ;
: SG-OBLIQUE ;

LEXICON NUM_VIL1L1
!! **LEXICON @LEXNAME@**
: N_VIL1L1 ;

LEXICON NUM_SEICHEME !seičeme:seičem
:e SG-NOM-SUF ;
Expand Down
Loading

0 comments on commit ecdbca1

Please sign in to comment.