Merge pull request #280 from inariksit/somali

Somali

src/somali/CatSom.gf

@@ -98,7 +98,6 @@ concrete CatSom of Cat = CommonX - [Adv] ** open ResSom, Prelude in {
 -- additional lexicon modules.
 
     V,
-    VV,    -- verb-phrase-complement verb         e.g. "want" -- TODO: VPs that have sentential complement use waxa as stm? see Nilsson p. 68
     VS,    -- sentence-complement verb            e.g. "claim"
     -- TODO: eventually different lincats
     VQ,    -- question-complement verb            e.g. "wonder"
@@ -108,6 +107,8 @@ concrete CatSom of Cat = CommonX - [Adv] ** open ResSom, Prelude in {
     V2Q,   -- verb with NP and Q complement       e.g. "ask"
     V2A = ResSom.Verb ; -- verb with NP and AP complement      e.g. "paint"
 
+    VV = ResSom.VV ;    -- verb-phrase-complement verb         e.g. "want"
+
     V2 = ResSom.Verb2 ;
     V3 = ResSom.Verb3 ;
 
@@ -123,7 +124,7 @@ concrete CatSom of Cat = CommonX - [Adv] ** open ResSom, Prelude in {
 
 linref
     -- Cl = linCl ;
-    VP = linVP VInf ;
+    VP = infVP ;
     CN = linCN ;
     Prep = \prep -> prep.s ! P3_Prep ++ prep.sii ++ prep.dhex ++ prep.miscAdv ! Sg3 Masc ;
     S = \s -> linBaseCl (s.s ! False) ;

src/somali/IdiomSom.gf

@@ -40,7 +40,7 @@ concrete IdiomSom of Idiom = CatSom ** open Prelude, ResSom, VerbSom in {
     } ;
 
 
-  {- TODO: Saeed p. 92 optative
+  {- TODO: Saeed p. 92 and 207, optative
   -- : VP -> Utt ;       -- let's go
   ImpPl1 vp = { } ;
 

src/somali/LexiconSom.gf

@@ -312,7 +312,7 @@ lin see_V2 = mkV2 "ark" ;
 -- lin send_V3 = mkV3 "" ;
 -- lin sew_V = mkV "" ;
 -- lin sharp_A = mkA "" ;
--- lin sheep_N = mkN "" ;
+lin sheep_N = mkN "lax" fem ;
 -- lin ship_N = mkN "" ;
 -- lin shirt_N = mkN "" ;
 -- lin shoe_N = mkN "" ;

src/somali/NounSom.gf

@@ -23,12 +23,11 @@ concrete NounSom of Noun = CatSom ** open ResSom, Prelude in {
                 -- If cn has modifier, Nom ending attaches to the modifier
                 <_,Nom,True,_,_> => {nf=Def det.n ; c=Abs} ;
 
-                -- Definite
+                -- a Det with st=Indefinite uses Indef forms
-                <_,_,False,Definite,n> => {nf=Def n ; c=c} ;
+                <_,_,_,Indefinite,n>  => {nf=Indef n ; c=c} ;
 
-                <_,_,False,Indefinite,n>  => {nf=Indef n ; c=c} ;
+                -- All other determiners use the definite stem
-
+                _ => {nf=Def det.n ; c=c}
-                _ => {nf=Def det.n ; c=c} -- TODO check
              } ;
           art = gda2da cn.gda ! det.n ;
           num = case det.isNum of {True => Sg ; _ => det.n} ;
@@ -257,7 +256,7 @@ concrete NounSom of Noun = CatSom ** open ResSom, Prelude in {
     mod = \\st,n,c => --what to do with subject case if there's both adj and RS?
             cn.mod ! st ! n ! Abs
          ++ andConj st cn.hasMod
-         ++ rs.s ! st ! gennum cn n ! c ;
+         ++ rs.s ! st ! gennum cn Sg ! c ; -- gennum cn Sg, because plural form is only for 1st person plural
     hasMod = True ;
     } ;
 

src/somali/ParadigmsSom.gf

@@ -32,6 +32,11 @@ oper
   u  : Preposition ;
   noPrep : Preposition ;
 
+  -- TODO: add subjunctive too!
+  VVForm : Type ; -- Argument to give to mkVV
+  infinitive : VVForm ; -- Takes its complement in infinitive
+  waa_in : VVForm ; -- No explicit verb, just uses "waa in" construction
+
 
 --2 Nouns
 
@@ -85,6 +90,11 @@ oper
     mkV3 : V -> (_,_ : Preposition) -> V2 ; -- Already constructed verb with preposition
     } ;
 
+  mkVV : overload {
+    mkVV : (kar : Str) -> VV ; -- VV that takes its complement in infinitive.
+    mkVV : VVForm -> VV ; -- VV such as "waa in"
+   } ;
+
   -- TODO: actual constructors
   -- mkVA : Str -> VA = \s -> lin VA (mkVerb  s) ;
   --
@@ -161,6 +171,10 @@ oper
   la = ResSom.La ;
   u  = ResSom.U ;
   noPrep = ResSom.NoPrep ;
+
+  VVForm = Bool ;
+  infinitive = False ;
+  waa_in = True ;
   ------------------------
 
   mkN = overload {
@@ -217,6 +231,14 @@ oper
     mkV3 : V -> (_,_ : Preposition) -> V2 = \v,p,q -> lin V3 (v ** {c2 = p ; c3 = q}) ;
     } ;
 
+  mkVV = overload {
+    mkVV : (kar : Str) -> VV -- VV that takes its complement in infinitive.
+     = \kar -> lin VV ({isVS=False} ** mkV kar) ;
+    mkVV : VVForm -> VV -- VV such as "waa in"
+     = \b -> let dummyV : V = mkV "in"
+              in lin VV (dummyV ** {isVS=b ; s = \\_ => "in"})
+   } ;
+
   possPrep : N -> CatSom.Prep = \dhex -> emptyPrep ** {
     miscAdv = \\agr => 
         let qnt = PossPron (pronTable ! agr) ;

src/somali/ParamSom.gf

@@ -288,6 +288,7 @@ param
                       1/2SG/3SG M/2PL/3PL suga (VRel MascSg)
                       3 SG F sugta (VRel FemSg)
                       1PL sugna (VRel PlInv) -}
+    | VRelNeg     -- Saeed p. 211 have: lahayn, be: ahayni
     | VNegCond GenNum ;
 
   VAgr =
@@ -315,6 +316,14 @@ oper
     Pl1 _ => Pl1_ ; Pl2 => Pl2_ ; Pl3 => Pl3_
   } ;
 
+  isNeg : VForm -> Bool = \vf -> case vf of {
+    VNegPast _ => True ;
+    VNegCond _ => True ;
+    VRelNeg    => True ;
+    VImp _ Neg => True ;
+    VPres _ _ Neg => True ;
+    _ => False
+    } ;
 --------------------------------------------------------------------------------
 -- Clauses
 

src/somali/RelativeSom.gf

@@ -18,7 +18,8 @@ lin
         rclSubord = mergeRCl (cls.s ! True) ;
         rclStatement = mergeRCl (cls.s ! False) ;
     in rp.s ++ case <gn,c,t,a,p> of {
-        <g,Abs,Pres,Simul,Pos> => linVP (VRel g) vp ; -- reduced present only in absolutive
+        <g,Abs,Pres,Simul,Pos> => linVP (VRel g) Subord vp ; -- reduced present only in absolutive
+        <_,Abs,Pres,Simul,Neg> => linVP VRelNeg Subord vp ; -- special form for have and be
         <_,Nom,Pres,Simul,Pos> => rclStatement.s ! t ! a ! p ; -- the usual forms, not subordinate
         _ => rclSubord.s ! t ! a ! p } -- the rest is Subord because of negation.
     } ;

src/somali/ResSom.gf

@@ -471,6 +471,8 @@ oper
   Verb2 : Type = Verb ** {c2 : Preposition} ;
   Verb3 : Type = Verb2 ** {c3 : Preposition} ;
 
+  VV : Type = Verb ** {isVS : Bool} ;
+
   -- Saeed page 79:
   -- "… the reference form is the imperative singular form
   -- since it corresponds to the form of the basic root."
@@ -561,7 +563,8 @@ oper
           VInf          => arki ;
           VRel SgMasc   => qaat + "a" ;
           VRel SgFem    => arag + t + "a" ;
-          VRel PlInv    => arag + "na"
+          VRel PlInv    => arag + "na" ;
+          VRelNeg       => qaat + "o"  -- TODO check
 
            } ;
         sii, dhex = [] ;
@@ -630,8 +633,8 @@ oper
           VNegCond SgMasc => "ahaadeen" ; -- 1SG/3 SG M/3PL 
           VNegCond SgFem  => "ahaateen" ; -- 2SG/3 SG F/2PL 
           VNegCond PlInv  => "ahaanneen" ; -- 1PL
-          --VRelShort       => "ah" ;
+          VRel _          => "ah" ; -- All persons: see Nilsson p. 78. TODO check Saeed p. 103
-          VRel _          => "ah" ; -- TODO find right forms
+          VRelNeg         => "ahayni" ; -- Saeed 
           VInf            => "ahaan" ;
           VImp Sg pol     => if_then_Pol pol "ahaw" "ahaanin" ;
           VImp Pl pol     => if_then_Pol pol "ahaada" "ahaanina" ;
@@ -649,10 +652,19 @@ oper
           VPres _ Pl1_        Pos => "leenahay" ;
           VPres _ Pl2_        Pos => "leedihiin" ;
           VPres _ Pl3_        Pos => "leeyihiin" ;
+
+          VPres _ Sg1_Sg3Masc Neg => "lihi" ;
+          VPres _ Sg2_Sg3Fem  Neg => "lihid" ;
+          VPres _ Pl1_        Neg => "lihin" ;
+          VPres _ Pl2_        Neg => "lihidin" ;
+          VPres _ Pl3_        Neg => "laha" ;
+
+          VNegPast _              => "lahayn" ;
+
           VPast asp agr           => "l" + copula.s ! VPast asp agr ;
           VNegCond agr            => "l" + copula.s ! VNegCond agr ;
---          VRelShort                  => "leh" ;
+          VRel _                => "leh" ; -- All persons: see Nilsson p. 78
-          VRel _                => "leh" ; -- TODO find right forms
+          VRelNeg               => "lahayn" ;
           x                     => hold_V.s ! x }
     } ;
 
@@ -691,7 +703,7 @@ oper
     c2 : PrepCombination ; -- Prepositions can combine together and with object pronoun.
     obj2 : NPLite ; -- {s : Str ; a : PrepAgr}
     secObj : Str ; -- if two overt pronoun objects
-    vComp : Str ; -- VV complement
+    vComp : {pr,pst : Str} -- VV complement
     } ;
 
   VPSlash : Type = VerbPhrase ;
@@ -699,7 +711,8 @@ oper
   useV : Verb -> VerbPhrase = \v -> v ** {
     comp = \\_ => <[],[]> ;
     pred = case v.isCopula of {True => Copula ; _ => NoPred} ;
-    vComp,berri,miscAdv,refl = [] ;
+    vComp = {pr,pst = []} ;
+    berri,miscAdv,refl = [] ;
     c2 = Single NoPrep ;
     obj2 = {s = [] ; a = P3_Prep} ;
     secObj = []
@@ -839,26 +852,39 @@ oper
                 Statement => case <p,vp.pred,subj.a> of {
                                <Pos,Copula|NoCopula,Pl3|Sg3 _> => {p1 = "waa" ; p2 = []} ;
                                _ => stmarkerNoContr ! subj.a ! p }} ;
-      in (wordOrder subjnoun subjpron stm obj pred vp) ;
+      in wordOrder subjnoun stm obj pred vp cltyp ;
     } where {
         vp : VerbPhrase = case isPassive vps of {
                True => complSlash (insertComp vps np) ;
                _    => complSlash vps } ;
-        subj : NounPhrase = case isPassive vps of {True => impersNP ; _ => np}
+        subj : NounPhrase = case isPassive vps of {
+               True => impersNP ; 
+               _    => np }
       } ;
 
-  wordOrder : (sn,sp : Str) -> (stm,obj : {p1,p2 : Str}) -> {fin,inf : Str} -> VerbPhrase -> BaseCl =
+  wordOrder : (sn : Str) -> (stm,obj : {p1,p2 : Str}) -> {fin,inf : Str} -> VerbPhrase -> ClType -> BaseCl =
-    \subjnoun,subjpron,stm,obj,pred,vp -> {
+    \subjnoun,stm,obj,pred,vp,cltyp -> {
+   {- Saeed p. 210-211: "The relative clause resembles a main clause in syntax
+      except that the tendency for verb final order is much stronger. [..] Certain
+      elements such as subject clitic pronouns, and the negative word aan  'not' are
+      attracted to the head nominal and thus move away from the verbal group."
+    -}
         beforeSTM = vp.berri -- AdV
                   ++ subjnoun -- subject if it's a noun
-                  ++ obj.p1 ; -- object if it's a noun
+                  ++ case cltyp of {
-              stm = stm.p1 ;  -- sentence type marker
+                        Subord => [] ;
-         afterSTM = stm.p2    -- possible subj. pronoun
+                        _ => obj.p1 } ; -- noun object if it's a statement
+              stm = stm.p1 ; -- sentence type marker; empty if subordinate and positive
+         afterSTM = vp.vComp.pr -- "waa in" construction
+                  ++ stm.p2   -- possible subj. pronoun
+                  ++ case cltyp of {
+                        Subord => obj.p1 ; -- noun object if it's subordinate clause
+                        _      => [] } 
                   ++ obj.p2   -- object if it's a pronoun
                   ++ vp.sii   -- restricted set of particles
                   ++ vp.dhex  -- restricted set of nouns/adverbials
                   ++ vp.secObj   -- "second object"
-                  ++ vp.vComp    -- VV complement
+                  ++ vp.vComp.pst  -- VV complement
                   ++ pred.inf    -- potential infinitive/participle
                   ++ pred.fin    -- the verb inflected
                   ++ vp.miscAdv } ; ---- NB. Only used if there are several adverbs.
@@ -903,7 +929,7 @@ oper
       _ => vfStatement t ant p agr vp
       } ; -- TODO other relative forms
 
-  infVP : VerbPhrase -> Str = linVP VInf ;
+  infVP : VerbPhrase -> Str = linVP VInf Statement ;
 
   stmarkerContr : Agreement => Polarity => Str = \\a,b =>
     let stm = if_then_Pol b "w" "m"
@@ -926,11 +952,15 @@ oper
 -- linrefs
 
 oper
-  linVP : VForm -> VerbPhrase -> Str = \vf,vp ->
+  linVP : VForm -> ClType -> VerbPhrase -> Str = \vf,cltyp,vp ->
     let vp' = complSlash vp ;
         inf = {inf = vp.s ! vf ; fin=[]} ;
-        wo = wordOrder [] [] {p1,p2=[]} (vp'.comp ! pagr2agr vp.obj2.a) inf vp' ;
+        stm = case <cltyp,isNeg vf> of {
-     in wo.beforeSTM ++ wo.afterSTM ;
+                <Subord,True> => {p1 = "aan" ; p2 = []} ;
+                _             => {p1,p2 = []}
+               } ;
+        wo = wordOrder [] stm (vp'.comp ! pagr2agr vp.obj2.a) inf vp' cltyp ;
+     in wo.beforeSTM ++ wo.stm ++ wo.afterSTM ;
 
   linCN : CNoun -> Str = \cn -> cn.s ! Indef Sg ++ cn.mod ! Indefinite ! Sg ! Abs ;
   linAdv : Adverb -> Str = \adv ->

src/somali/StructuralSom.gf

@@ -176,11 +176,11 @@ lin yes_Utt = ss "haa" ;
 -------
 -- Verb
 
-lin have_V2 = mkV2 have_V noPrep ;
+lin have_V2 = mkV2 have_V noPrep ; -- TODO: check if {sii = "l" ++ BIND ; isCopula=True} makes sense for present tense negative
 lin can8know_VV = can_VV ; -- can (capacity)
-lin can_VV = mkV "kar" ;   -- can (possibility)
+lin can_VV = mkVV "kar" ;   -- can (possibility)
-{-lin must_VV = mkV "" ;
+lin must_VV = mkVV waa_in ;
-lin want_VV = mkV "" ;
+{-lin want_VV = mkV "" ;
 
 
 ------

src/somali/VerbSom.gf

@@ -15,10 +15,18 @@ lin
   ReflVP = ResSom.insertRefl ;
 
   -- : VV  -> VP -> VP ;
-  ComplVV vv vp = vp ** {  -- check Saeed p. 169
+  ComplVV vv vp = let vc = vp.vComp in case vv.isVS of {
-    s = vv.s ;
+    True => vp ** {
-    vComp = vp.vComp ++ vp.s ! VInf ;
+      vComp = vc ** {pr = vv.s ! VInf} ;
+      obj2 = vp.obj2 ** {s = []} ; -- word order hack to avoid more parameters: 
+      miscAdv = vp.miscAdv ++ vp.obj2.s -- dump it all to miscAdv
+      } ;
+
+    False => vp ** {
+      s = vv.s ; -- check Saeed p. 169
+      vComp = vc ** {pst = vc.pst ++ vp.s ! VInf} ;
       pred = NoPred ;
+      } 
     } ;
 
   -- : VS  -> S  -> VP ;

src/somali/unittest/cl.gftest

@@ -1,4 +1,5 @@
 -- Restrictive relative clauses (RelCN)
+-- Saeed p. 213
 
 -- LangEng: the books that the men buy
 LangSom: buugaag BIND ta niman BIND ku iibsadaan
@@ -14,7 +15,7 @@ LangSom: niman BIND ka buugaag BIND ta iibsada
 Lang: PhrUtt NoPConj (UttNP (DetCN (DetQuant DefArt NumPl) (RelCN (UseN man_N) (UseRCl (TTAnt TPres ASimul) PPos (RelVP IdRP (ComplSlash (SlashV2a buy_V2) (DetCN (DetQuant DefArt NumPl) (UseN book_N)))))))) NoVoc
 
 -- LangEng: the men that don't buy the books
-LangSom: niman BIND ka buugaag BIND ta aan iibsado
+LangSom: niman BIND ka aan buugaag BIND ta iibsado
 Lang: PhrUtt NoPConj (UttNP (DetCN (DetQuant DefArt NumPl) (RelCN (UseN man_N) (UseRCl (TTAnt TPres ASimul) PNeg (RelVP IdRP (ComplSlash (SlashV2a buy_V2) (DetCN (DetQuant DefArt NumPl) (UseN book_N)))))))) NoVoc
 
 
@@ -23,7 +24,7 @@ LangSom: niman BIND ka buugaag BIND ta iibsadaa waa kuwan
 Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP (DetCN (DetQuant DefArt NumPl) (RelCN (UseN man_N) (UseRCl (TTAnt TPres ASimul) PPos (RelVP IdRP (ComplSlash (SlashV2a buy_V2) (DetCN (DetQuant DefArt NumPl) (UseN book_N))))))) (UseComp (CompNP (DetNP (DetQuant this_Quant NumPl))))))) NoVoc
 
 -- LangEng: the men that don't buy the books are these
-LangSom: niman BIND ka buugaag BIND ta aan iibsado waa kuwan
+LangSom: niman BIND ka aan buugaag BIND ta iibsado waa kuwan
 Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP (DetCN (DetQuant DefArt NumPl) (RelCN (UseN man_N) (UseRCl (TTAnt TPres ASimul) PNeg (RelVP IdRP (ComplSlash (SlashV2a buy_V2) (DetCN (DetQuant DefArt NumPl) (UseN book_N))))))) (UseComp (CompNP (DetNP (DetQuant this_Quant NumPl))))))) NoVoc
 
 
@@ -45,6 +46,22 @@ Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP (DetCN (DetQ
 LangSom: naag BIND ta aad ku bartay
 Lang: PhrUtt NoPConj (UttNP (DetCN (DetQuant DefArt NumSg) (RelCN (UseN woman_N) (UseRCl (TTAnt TPast ASimul) PPos (RelSlash IdRP (SlashVP (UsePron youSg_Pron) (SlashV2a teach_V2))))))) NoVoc
 
+-- Combine with passive
+-- LangEng: the woman that was taught
+LangSom: naag BIND ta lagu baray
+Lang: PhrUtt NoPConj (UttNP (DetCN (DetQuant DefArt NumSg) (RelCN (UseN woman_N) (UseRCl (TTAnt TPast ASimul) PPos (RelVP IdRP (PassV2 teach_V2)))))) NoVoc
+
+
+-- Negative forms
+-- LangEng: sheep that aren't black don't see you
+LangSom: lax aan madow aha ku BIND ma arkaan
+Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PNeg (PredVP (DetCN (DetQuant IndefArt NumPl) (RelCN (UseN sheep_N) (UseRCl (TTAnt TPres ASimul) PNeg (RelVP IdRP (UseComp (CompAP (PositA black_A))))))) (ComplSlash (SlashV2a see_V2) (UsePron youPol_Pron))))) NoVoc
+
+-- LangEng: you don't see sheep that aren't black
+LangSom: laxo aan madow ahayni ma aragto
+Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PNeg (PredVP (UsePron youPol_Pron) (ComplSlash (SlashV2a see_V2) (DetCN (DetQuant IndefArt NumPl) (RelCN (UseN sheep_N) (UseRCl (TTAnt TPres ASimul) PNeg (RelVP IdRP (UseComp (CompAP (PositA black_A))))))))))) NoVoc
+
+
 -- Appositive relative clauses (RelNP)
 
 -- LangEng: he , that the men see

src/somali/unittest/vp.gftest

@@ -70,6 +70,14 @@ Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (GenericCl (ComplVV
 LangSom: in uu imanayo waa aynu ogaannay
 Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPast ASimul) PPos (PredVP (UsePron we_Pron) (ComplVS know_VS (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron he_Pron) (ProgrVP (UseV come_V)))))))) NoVoc
 
+-- LangEng: you must see mother of the cat
+LangSom: waa in aad aragtaa bisad BIND da hooya BIND deed
+Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron youPol_Pron) (ComplVV must_VV (ComplSlash (SlashV2a see_V2) (MassNP (ComplN2 mother_N2 (DetCN (DetQuant DefArt NumSg) (UseN cat_N))))))))) NoVoc
+
+-- LangEng: the mother of the cat must be seen
+LangSom: waa in la arkaa bisad BIND da hooya BIND deed
+Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (GenericCl (ComplVV must_VV (ComplSlash (SlashV2a see_V2) (MassNP (ComplN2 mother_N2 (DetCN (DetQuant DefArt NumSg) (UseN cat_N))))))))) NoVoc
+
 --------------------------------------------------------------------------------
 -- Adverbials