fine-tuned verbal derivation for more precise compounding

giellalt · May 14, 2024 · a41279f · a41279f
1 parent 92f7b70
commit a41279f
Show file tree

Hide file tree

Showing 8 changed files with 349 additions and 48 deletions.
diff --git a/src/fst/Makefile.am b/src/fst/Makefile.am
@@ -552,6 +552,7 @@ generator-raw-gt-desc.simple.hfst: generator-raw-gt-desc.simple.weightless.hfst
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/mus' -a 10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/ng' -a 10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/v' -a 10 -A \
+	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/vus' -a 10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/tav' -a 10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/nud' -a 10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/mata' -a 10 -A \
@@ -625,7 +626,7 @@ generator-raw-gt-desc.simple.hfst: generator-raw-gt-desc.simple.weightless.hfst
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Neg' -a 1 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Sup' -a 0 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Inf' -a 0 -A \
-	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Ger' -a 0 -A \
+	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Ger' -a 5 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Prc' -a 0 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Foc/gi' -a 0 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Emph' -a 0 -A \
@@ -748,6 +749,7 @@ guesser-raw.weighted.hfst: guesser-raw.simple.hfst
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/mus' -a -10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/ng' -a -10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/v' -a -10 -A \
+	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/vus' -a 10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/tav' -a -10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/nud' -a -10 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Der/mata' -a -10 -A \
@@ -821,7 +823,7 @@ guesser-raw.weighted.hfst: guesser-raw.simple.hfst
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Neg' -a 1 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Sup' -a 0 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Inf' -a 0 -A \
-	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Ger' -a 0 -A \
+	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Ger' -a 5 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Prc' -a 0 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Foc/gi' -a 0 -A \
 	| $(HFST_REWEIGHT) $(MORE_VERBOSITY) $(HFST_FLAGS) -S '+Emph' -a 0 -A \

diff --git a/src/fst/filters/block-derivations.est.xfscript b/src/fst/filters/block-derivations.est.xfscript
@@ -29,7 +29,8 @@ define BadDer1 [
     [~[?* ("+Guess") [["+N"] | ["+Num" "+Card"]]] "+Der/kond"] |
     [~[?* [[ ("+Guess") "+A"] | [("+Guess") "+A" "+Comp"] | [("+Guess") "+A" "+Superl"] | 
           [ ZZ {tama} ("+Guess") "+V"] | [ ZZ {dama} ("+Guess") "+V"] | 
-          [ {hooldama} ("+Guess") "+V"] | [ {soendama} ("+Guess") "+V"] | [ {tsema} ("+Guess") "+V"] | [ ? ? ? {lema} ("+Guess") "+V"]]] "+Der/us"] |
+          [ {hooldama} ("+Guess") "+V"] | [ {soendama} ("+Guess") "+V"] | 
+          [ {tsema} ("+Guess") "+V"] | [ ? ? ? {lema} ("+Guess") "+V"]]] "+Der/us"] |
     [~[?* [[ ZZ {tama} ("+Guess") "+V"] | [ ZZ {dama} ("+Guess") "+V"]]] "+Der/is"] |
     [~[?* [{uma} | {tellima}] ("+Guess") "+V"] "+Der/mus"] |             # devalveeruma - devalveerumus
     [~[?* {eerima} ("+Guess") "+V"] "+Der/ng"] |           # devalveerima - devalveering
@@ -65,7 +66,10 @@ define BadDer2 [
 
 # some suffix sequences are actually bad
 define BadDer3 [
-    [[ "+Der/nu" | "+Der/us" | "+Der/lane" ] "+N" "+Der/ti"]  
+    [[ "+Der/nu" | "+Der/us" | "+Der/lane" ] "+N" "+Der/ti"] | 
+    [[ "+Der/v" | "+Der/tav" | "+Der/nud" | "+Der/matu" | "+Der/tamatu" | "+Der/tu" ] "+A" "+Der/sti"] |  
+    [ "+Der/v" "+A" "+Der/us"] | 
+    [[ "+Der/lik" | "+Der/line" ] "+A" "+Der/sti"]  
     ] ;
 
 # some words are not suitable for certain suffixes 

diff --git a/src/fst/morphology/affixes/verbs.lexc b/src/fst/morphology/affixes/verbs.lexc
@@ -414,7 +414,8 @@ LEXICON SUPINE_V
  @R.Part.One@           SUPINE_MA_FORMS ;    ! These can be 1st part of a compound
                       : MINE_DERIVATION ;    ! elamine
                       : JA_DERIVATION ;      ! elaja, õpetaja, õpetajanna, õpetajake, õpetajalik, ...
-                      : V_DERIVATION ;       ! õpetav, õpetavalt, õpetavus, õpetavam, ...
+                      : V_DERIVATION ;       ! õpetav, õpetavalt, õpetavam, ...
+                      : VUS_DERIVATION ;     ! õpetavus
                       : MATU_DERIVATION ;    ! segamatu, segamatult, segamatus, segamatum, ...
                       : MATA_DERIVATION ;     ! segamata
                       : MUS_DERIVATION ;     ! küllastumus, promoveerumus etc
@@ -461,11 +462,12 @@ LEXICON INFINITIVE    ! infinitive, gerund (des-form)
 
                       :%>D7 A_INFINITIVE ;
 
-LEXICON A_INFINITIVE   ! common part of an infinitive, gerund (des-form);
-                       ! also alone in a subparadigm of a few old words
+LEXICON A_INFINITIVE  ! common part of an infinitive, gerund (des-form);
+                      ! also alone in a subparadigm of a few old words
 
- @R.Part.One@             INF_FORMS ;     ! This can be 1st part of a compound
+ @R.Part.One@           INF_FORMS ;     ! This can be 1st part of a compound
                   +Ger:es NO_COMPOUND ;
+                        GER_DERIVATION ;      ! pool+magades, rõht+kirjutades;
 
 LEXICON INF_FORMS
                   +Inf:a GI ;
@@ -508,6 +510,14 @@ LEXICON PL3
 LEXICON NO_COMPOUND           ! this form cannot participate in a compound word
   @R.Part.One@@P.Part.Bad@ GI ;             
 
+! gerund is an exception in compounding and derivation ...
+LEXICON GER_DERIVATION    ! 2nd part of a compound
+ @R.Part.Two@@P.Part.Bad@@R.POS.Pref@  GER_DERIVATION_SUF ;    ! only 2 components: prefix + verb  
+
+LEXICON GER_DERIVATION_SUF       ! A derived adverb (but still tagged as +Verb+Ger)
+                  +Ger:es GI ;
+
+
 ! paradigmatic derivation, i.e. very regular
 ! @R.Part.One@ means that this is not a latter part of a compound 
 ! @R.Case.Par@ and @R.Case.Gen@ mean that this is a latter part of a compound 
@@ -516,8 +526,10 @@ LEXICON NO_COMPOUND           ! this form cannot participate in a compound word
 LEXICON NU_DERIVATION
   @R.Part.One@@P.POS.N@  NU_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Par@@P.POS.N@  NU_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.N@  NU_DERIVATION_SUF ;
   @R.Part.One@@P.POS.A@  NUD_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Par@@P.POS.A@  NUD_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.N@  NUD_DERIVATION_SUF ;
 
 LEXICON NU_DERIVATION_SUF
              +Der/nu+N:»nu AASTA ;             ! derivation: elanu etc
@@ -529,6 +541,7 @@ LEXICON NUD_DERIVATION_SUF
 LEXICON TU_DERIVATION  !    NB! the morpheme border + d/t is assigned before this lexicon
   @R.Part.One@@P.POS.A@  TU_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Gen@@P.POS.A@  TU_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.A@  TU_DERIVATION_SUF ;
 
 LEXICON TU_DERIVATION_SUF  !    NB! the morpheme border + d/t is assigned before this lexicon
              +Der/tu+A:u AASTA ;              ! derivation: elatu etc
@@ -537,48 +550,62 @@ LEXICON TU_DERIVATION_SUF  !    NB! the morpheme border + d/t is assigned before
 LEXICON MATA_DERIVATION
   @R.Part.One@@P.POS.A@  MATA_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Par@@P.POS.A@  MATA_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.A@  MATA_DERIVATION_SUF ;
 
 LEXICON MATA_DERIVATION_SUF
            +Der/mata+A:»mata GI ;            ! elamata 
 
 LEXICON MATU_DERIVATION
   @R.Part.One@@P.POS.N@  MATU_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Gen@@P.POS.N@  MATU_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.N@  MATU_DERIVATION_SUF ;
 
 LEXICON MATU_DERIVATION_SUF
            +Der/matu+A:»matu AASTA ;             ! segamatu etc
 
 LEXICON TAMATU_DERIVATION  !    NB! the morpheme border + d/t is assigned before this lexicon
   @R.Part.One@@P.POS.A@  TAMATU_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Par@@P.POS.A@  TAMATU_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.A@  TAMATU_DERIVATION_SUF ;
 
 LEXICON TAMATU_DERIVATION_SUF  !    NB! the morpheme border + d/t is assigned before this lexicon
            +Der/tamatu+A:amatu AASTA ;             ! segatamatu etc
 
 LEXICON V_DERIVATION
   @R.Part.One@@P.POS.A@  V_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Par@@P.POS.A@  V_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.A@  V_DERIVATION_SUF ;
 
 LEXICON V_DERIVATION_SUF
               +Der/v+A:»v MAGUS ;              ! elav, elava etc
 
+LEXICON VUS_DERIVATION
+  @R.Part.One@@P.POS.N@  VUS_DERIVATION_SUF ;   ! no previous part; no restrictions   
+  @R.Case.Gen@@P.POS.N@@P.Der.us@@C.NeedNoun@  VUS_DERIVATION_SUF ;
+
+LEXICON VUS_DERIVATION_SUF
+             +Der/vus+N:»vus OLULINE ;         ! elavus, ...
+
 LEXICON TAV_DERIVATION   !    NB! the morpheme border + d/t is assigned before this lexicon
   @R.Part.One@@P.POS.A@  TAV_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Par@@P.POS.A@  TAV_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.A@  TAV_DERIVATION_SUF ;
 
 LEXICON TAV_DERIVATION_SUF   !    NB! the morpheme border + d/t is assigned before this lexicon
             +Der/tav+A:av MAGUS ;              ! elatav, elatava etc
 
 LEXICON MINE_DERIVATION
   @R.Part.One@@P.POS.N@  MINE_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Gen@@P.POS.N@  MINE_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.N@  MINE_DERIVATION_SUF ;
 
 LEXICON MINE_DERIVATION_SUF
            +Der/mine+N:»mine OLULINE ;         ! elamine
 
 LEXICON JA_DERIVATION
   @R.Part.One@@P.POS.N@  JA_DERIVATION_SUF ;   ! no previous part; no restrictions   
   @R.Case.Gen@@P.POS.N@  JA_DERIVATION_SUF ;
+  @R.Case.Sem@@P.POS.N@  JA_DERIVATION_SUF ;
 
 LEXICON JA_DERIVATION_SUF
              +Der/ja+N:»ja AASTA ;             ! elaja, elajat etc
@@ -593,14 +620,15 @@ LEXICON IS_DERIVATION_SUF
 
 LEXICON MUS_DERIVATION
   @R.Part.One@@P.POS.N@  MUS_DERIVATION_SUF ;   ! no previous part; no restrictions   
-  @R.Case.Gen@@P.POS.N@  MUS_DERIVATION_SUF ;
+  @R.Case.Gen@@P.POS.N@@P.Der.us@@C.NeedNoun@  MUS_DERIVATION_SUF ;
 
 LEXICON MUS_DERIVATION_SUF
              +Der/mus+N:»mus OLULINE ;         ! promoveerumus, elamus, ...
 
 LEXICON US_DERIVATION_VERB         ! us also changes A -> N; directing to that continuation class 
   @R.Part.One@   US_DERIVATION ;   ! no previous part; no restrictions   
   @R.Case.Gen@   US_DERIVATION ;
+  @R.Case.Sem@   US_DERIVATION ;
 
 LEXICON NG_DERIVATION
   @R.Part.One@@P.POS.N@  NG_DERIVATION_SUF ;   ! no previous part; no restrictions   

diff --git a/src/fst/morphology/root.lexc b/src/fst/morphology/root.lexc
@@ -456,6 +456,9 @@ Multichar_Symbols
 +Der/mus !!= * `@CODE@`
 !!€ küllastumus: küllastuma+V+Der/mus+N+Sg+Nom 
 
++Der/vus !!= * `@CODE@`
+!!€ elavus: elama+V+Der/vus+N+Sg+Nom 
+
 +Der/ng !!= * `@CODE@`
 !!€ devalveering: devalveerima+V+Der/ng+N+Sg+Nom 
 
@@ -713,9 +716,9 @@ Multichar_Symbols
  @D.Stem.topelt@
  @P.Stem.Guessed@   ! stem is guessed, it is not from the lexicon
  @R.Stem.Guessed@
- @D.Stem.Guessed@   ! if a lemma has it, then this lemma is restricted in its ability to be the last part of a compound 
+ @D.Stem.Guessed@   ! if a word has it, then this word cannot follow a guessed stem 
  @C.Stem@
-
+ 
 !! A special condition that is used for filtering derivations and compounds
  @P.NeedAdj.On@         ! +A or +Der/A
  @R.NeedAdj@
@@ -1224,16 +1227,17 @@ LEXICON Latter !!= * `@CODE@` the latter part of a compound
 
   @R.POS.ACRMinus@ AnyLatterVerb ;      
   @R.POS.Pref@ AnyLatterVerb ;  ! ebalugemine
+  @R.POS.Pref@ Verbs ;          ! gerund: poolmagades
 
   @R.POS.N@@R.Case.Gen@ LatterVerb ;   ! lauajooksja
   @R.POS.N@@R.Case.Par@ LatterVerb ;   ! laudajooksnud
 
   @R.POS.N@@R.Case.Nom@@R.Stem.Nom@ AnyLatterVerb ;  ! map orig. case flag for derivation needs
-  @R.POS.N@@R.Case.Sem@ AnyLatterVerb ;  ! map orig. case flag for derivation needs
+  @R.POS.N@@R.Case.Sem@ LatterVerb ;  ! 
   @R.POS.N@@R.Case.Short@ AnyLatterVerb ;  ! investeerimisnõustamine
 
   @R.POS.A@@R.Case.Short@ AnyLatterVerb ;  ! map orig. case flag for derivation needs
-  @R.POS.A@@R.Case.Sem@ AnyLatterVerb ;  ! map orig. case flag for derivation needs
+  @R.POS.A@@R.Case.Sem@ LatterVerb ;  ! 
   @R.POS.AComp@@R.Case.Sem@ AnyLatterVerb ;  ! map orig. case flag for derivation needs
 
   @R.POS.GA@@P.Case.Gen@ LatterVerb ;   ! eestivihkaja
@@ -1253,7 +1257,7 @@ LEXICON Latter !!= * `@CODE@` the latter part of a compound
 ! NB only some pronouns !
   @R.POS.Pron@@R.Case.Gen@ LatterVerb ;  
   @R.POS.Pron@@R.Case.Par@ LatterVerb ;
-  @R.POS.Pron@@R.Case.Sem@ AnyLatterVerb ;
+  @R.POS.Pron@@R.Case.Sem@ LatterVerb ;
 
   @R.POS.V@@D.Case@ AnyLatterVerb ;              ! infinitive
   @R.POS.V@@R.Case.Sem@ AnyLatterVerb ;          ! -ma, -mas etc

diff --git a/src/fst/morphology/stems/adverbs.lexc b/src/fst/morphology/stems/adverbs.lexc
@@ -2,8 +2,10 @@
 ! CompoundingAdverbs and NonCompoundingAdverbs ;
 
 
+
 LEXICON CompoundingAdverbs
 
+
 @P.Stem.topelt@järel+Adv:@P.Stem.topelt@järel GI  "weight: 5 " ; 
 @D.Stem.Guessed@@P.Stem.topelt@koos+Adv:@D.Stem.Guessed@@P.Stem.topelt@k˘oos GI  "weight: 4 " ; 
 @P.Stem.topelt@kõrval+Adv:@P.Stem.topelt@kõrval GI  "weight: 5 " ; 
@@ -449,11 +451,10 @@ pooleli+Adv:pooleli GI  "weight: 7 " ;
 poolt+Adv:p˘oolt GI  "weight: 4 " ; 
 praokil+Adv:pr˘aokil GI  "weight: 11 " ; 
 praokile+Adv:pr˘aokile GI  "weight: 11 " ; 
-@D.Stem.Guessed@@P.Stem.vähe@puht+Adv:@D.Stem.Guessed@@P.Stem.vähe@p˘uht GI  "weight: 9 " ; 
-@D.Stem.Guessed@@P.Stem.vähe@vähe+Adv:@D.Stem.Guessed@@P.Stem.vähe@vähe GI  "weight: 5 " ; 
 puhevil+Adv:puhevil GI  "weight: 11 " ; 
 puhevile+Adv:puhevile GI  "weight: 11 " ; 
 puhta+Adv:p˘uhta GI  "weight: 11 " ; 
+@D.Stem.Guessed@@P.Stem.vähe@puht+Adv:@D.Stem.Guessed@@P.Stem.vähe@p˘uht GI  "weight: 9 " ; 
 puhvi+Adv:p˘uhvi GI  "weight: 11 " ; 
 pungi+Adv:p˘ungi GI  "weight: 10 " ; 
 pungil+Adv:pungil GI  "weight: 9 " ; 
@@ -714,6 +715,7 @@ võlgu+Adv:v˘õlgu GI  "weight: 8 " ;
 võõriti+Adv:võõriti GI  "weight: 11 " ; 
 võõrsil+Adv:v˘õõrsil GI  "weight: 8 " ; 
 võõrsile+Adv:v˘õõrsile GI  "weight: 11 " ; 
+@D.Stem.Guessed@@P.Stem.vähe@vähe+Adv:@D.Stem.Guessed@@P.Stem.vähe@vähe GI  "weight: 5 " ; 
 vähem+Adv:vähem GI  "weight: 5 " ; 
 välja+Adv:v˘älʲja GI  "weight: 3 " ; 
 vääri+Adv:vääri GI  "weight: 11 " ; 
@@ -752,8 +754,10 @@ väärt+Adv:v˘äärt GI  "weight: 7 " ;
 @D.Stem.Guessed@ülle+Adv:@D.Stem.Guessed@˘ülle GI  "weight: 10 " ; 
 ümber+Adv:˘ümber GI  "weight: 5 " ; 
 
+
 LEXICON NonCompoundingAdverbs
 
+
 aasta-aastalt+Adv:˘aasta-˘aastalt GI  "weight: 9 " ; 
 aastaringselt+Adv:˘aasta#r˘ingselt GI  "weight: 10 " ; 
 aastates+Adv:˘aastates GI  "weight: 11 " ; 

diff --git a/src/fst/morphology/stems/prefixes.lexc b/src/fst/morphology/stems/prefixes.lexc
@@ -97,6 +97,7 @@ piko+Pref:piko»- # ;
 pisi+Pref:pisi»- # ;
 polaar+Pref:pol˘aar»- # ;
 poliit+Pref:pol˘iit»- # ;
+pool+Pref:p˘ool»- # ;
 pop+Pref:p˘op»- # ;
 pseudo+Pref:pseudo»- # ;
 psühho+Pref:psühho»- # ;