Merge pull request #12 from ayberkt/structural

Complete most of the linearizations in StructuralTur

src/turkish/CatTur.gf

@@ -11,6 +11,7 @@ concrete CatTur of Cat = CommonX ** open ResTur, Prelude in {
     NP = {s : Case => Str ; a : Agr} ;
     VP = Verb ;
     VPSlash = VP ** {c : Prep} ;
+    Conj = {s : Str ; s1 : Str ; s2 : Str ; ct : ConjType} ;
 
     Pron = ResTur.Pron ;
     Det = {s : Str; n : Number; useGen : UseGen} ;

src/turkish/ResTur.gf

@@ -23,13 +23,14 @@ resource ResTur = ParamX ** open Prelude, Predef, HarmonyTur in {
 
     agrP3 : Number -> Agr ;
     agrP3 n = {n = n; p = P3} ;
--- For $Adjective$
+    -- For $Adjective$
   oper
     Adjective = Noun ** { adv : Str } ;
--- For $Verb$.
+
+    -- For $Verb$.
 
   param
-    VForm = 
+    VForm =
        VProg      Agr
      | VPast      Agr
      | VFuture    Agr
@@ -39,6 +40,8 @@ resource ResTur = ParamX ** open Prelude, Predef, HarmonyTur in {
      | Gerund Number Case
      ;
 
+    param ConjType = Infix | Mixfix ;
+
     UseGen = NoGen | YesGen Agr | UseIndef ;
 
   oper
@@ -73,9 +76,37 @@ resource ResTur = ParamX ** open Prelude, Predef, HarmonyTur in {
     mkPrep : Str -> Case -> {s : Str; c : Case; lock_Prep : {}} =
       \s, c -> lin Prep {s=s; c=c};
 
+    mkNP : Noun -> Number -> Person -> {s : Case => Str; a : Agr} =
+      \noun, n, p -> {s = noun.s ! n; a = {n = n; p = p}} ;
+
     mkClause : Str -> Agr -> Verb -> {s : Str} =
       \np, a, v -> ss (np ++ v.s ! VProg a) ;
 
+    mkDet : Str -> Number -> UseGen -> {s : Str; n : Number; useGen : UseGen} =
+      \s, n, ug -> {s = s; n = n; useGen = ug} ;
+
+    mkConj : overload {
+      mkConj : Str -> {s : Str ; s1 : Str ; s2 : Str ; ct : ConjType} ;
+      mkConj : Str -> Str -> {s : Str ; s1 : Str ; s2 : Str ; ct : ConjType} ;
+    } ;
+
+    mkConj = overload {
+      mkConj : Str -> {s : Str ; s1 : Str ; s2 : Str ; ct : ConjType} =
+        \s -> {
+          s  = s  ;
+          s1 = s  ;
+          s2 = [] ;
+          ct = Infix
+        } ;
+      mkConj : Str -> Str -> {s : Str ; s1 : Str ; s2 : Str ; ct : ConjType} =
+        \s1, s2 -> {
+          s  = s1 ++ s2 ;
+          s1 = s1 ;
+          s2 = s2 ;
+          ct = Mixfix
+        } ;
+    } ;
+
     attachMe : Verb -> {s : Str} =
       \v ->
         let

src/turkish/StructuralTur.gf

@@ -11,7 +11,9 @@ concrete StructuralTur of Structural = CatTur **
       mkPron "o" "onu" "ona" "onun" "onda" "ondan" "onlu" "onsuz" Sg P3 ;
 
     i_Pron  =
-      mkPron "ben" "beni" "bana" "benim" "bende" "benden" "benli" "bensiz" Sg P1 ;
+      mkPron "ben" "beni" "bana" "benim"
+             "bende" "benden" "benli" "bensiz"
+             Sg P1 ;
 
     it_Pron =
       mkPron "o" "onu" "ona" "onun" "onda" "ondan" "onlu" "onsuz" Sg P3 ;
@@ -29,6 +31,9 @@ concrete StructuralTur of Structural = CatTur **
     this_Quant =
       mkQuant "bu" ;
 
+    no_Quant =
+      mkQuant "hiç" ;
+
     we_Pron =
       mkPron "biz" "bizi" "bize" "bizim" "bizde" "bizden" "bizli" "bizsiz" Pl P1 ;
 
@@ -68,6 +73,8 @@ concrete StructuralTur of Structural = CatTur **
     in_Prep =
       variants {mkPrep "içinde" Gen; mkPrep "içerisinde" Gen} ;
 
+    except_Prep = mkPrep "dışında" Nom | mkPrep "dışında" Gen ;
+
     -- ... sırasında
     during_Prep =
       mkPrep "sırasında" Nom ;
@@ -76,9 +83,9 @@ concrete StructuralTur of Structural = CatTur **
     between_Prep =
       mkPrep "arasındaki" Gen ;
 
-    and_Conj = ss "ile" ;
+    and_Conj = mkConj "ile" ;
 
-    or_Conj = ss "veya" ;
+    or_Conj = mkConj "veya" ;
 
     yes_Utt = ss "evet" ;
     no_Utt  = ss "hayır" ;
@@ -86,11 +93,187 @@ concrete StructuralTur of Structural = CatTur **
     always_AdV = {s = "her zaman"} ;
 
     but_PConj = ss "ama" ;
+    therefore_PConj = ss "dolayısıyla" ;
 
-    at_most_AdN = ss "en fazla" ;
+    everybody_NP  = mkNP (mkN "herkes")     Sg P3 ;
+    everything_NP = mkNP (mkN "herşey")     Sg P3 ;
+    nothing_NP    = mkNP (mkN "hiçbir şey") Sg P3 ;
+    somebody_NP   = mkNP (mkN "biri")       Sg P3 ;
+    something_NP  = mkNP (mkN "bir şey")    Sg P3 ;
 
-    at_least_AdN = ss "en az" ;
+    -- The sentence (PredVP (UsePron he_Pron) (UseV sing_V)) would be
+    -- linearized as
+    --   > nobody sings
+    -- in English, whereas in Turkish it would literally translate to
+    --   > nobody doesn't sing
+    -- Linearizing (PredVP (UsePron he_Pron) (UseV sing_V)) will yield
+    -- nobody sings regardless. The double negation will be implemented
+    -- when `UseCl` is implemented eventually.
+    nobody_NP      = mkNP (mkN "hiç kimse") Sg P3 ;
 
-    as_CAdv = {s = "kadar"; p = "kadar"} ;
+    many_Det       = mkDet "birçok" Sg NoGen ;
+    every_Det      = mkDet "her"    Sg NoGen ;
+    all_Predet     = {s = "her"} ;
+    almost_AdA     = {s = "neredeyse"} ;
+    almost_AdN     = {s = "neredeyse"} ;
+
+    by8agent_Prep  = mkPrep "tarafından" Gen ;
+    by8means_Prep  = mkPrep "tarafından" Gen ;
+
+    although_Subj  = {s = "buna rağmen"} ;
+
+    that_Subj = {s = "o"} ;
+
+    -- TODO: this does not straightforwardly translate to Turkish which does
+    -- not have a "when" as a subordinating conjunction.
+    when_Subj = {s = "[TODO]"} ;
+
+    because_Subj   = {s = "çünkü"} ;
+
+    here_Adv       = mkAdv "burada" ;
+
+    everywhere_Adv = mkAdv "her yerde" ;
+
+    if_Subj        = {s = "eğer"} ;
+
+    both7and_DConj  = mkConj "hem" "hem de" ;
+    either7or_DConj = mkConj "ya"  "ya da"  ;
+
+    few_Det = mkDet "birkaç" Sg NoGen ;
+
+    for_Prep = mkPrep "için" Nom ;
+
+    under_Prep = mkPrep "altında" Gen ;
+
+    to_Prep   = mkPrep "" Dat ;
+    from_Prep = mkPrep "" Ablat ;
+
+    -- TODO: this is probably not correct. There is no straightforward
+    -- translation for `possess_Prep` in Turkish.
+    possess_Prep = mkPrep "" Gen ;
+
+    -- There are four senses of "through" as a preposition in English:
+    --
+    --   1. from one side of an opening to the other ("go through the window"),
+    --   2. entering, then later leaving ("drive through the town"),
+    --   3. by means of ("win through intimidation"), and
+    --   4. to or up to ("1945 to 1991").
+    --
+    -- All four of these would be translated in a different way to Turkish
+    -- so I don't know what's the best way to implement `through_Prep`. The
+    -- best thing to do is to just probably implement the most common use.
+    -- TODO: implement linearization for through_Prep.
+
+    -- TODO: there is really no have_V2 in Turkish.
+    -- have_V2
+
+    -- This is really just `here_Adv` in ablative form.
+    here7from_Adv = mkAdv "buradan" ;
+
+    -- This is really just `here_Adv` in dative form.
+    here7to_Adv = mkAdv "buraya" ;
+
+    somewhere_Adv = mkAdv "bir yere" ;
+    there_Adv     = mkAdv "oraya" ;
+
+    how8many_IDet = {s = "kaç tane"} ;
+
+    how8much_IAdv = {s = "ne kadar"} ;
+
+    how_IAdv = {s = "nasıl"} ;
+
+    -- Conditionals in Turkish are handled through inflections.
+    -- I will decide what to do with this later.
+    if_then_Conj = mkConj "foo" "bar" ;
+
+    -- TODO: in8front_Prep
+    in8front_Prep = mkPrep "önünde" Gen ;
+
+    language_title_Utt = {s = "Türkçe"} ;
+
+    more_CAdv = {s = "daha"; p = "daha"} ;
+
+    most_Predet = {s = "en çok"} ;
+
+    much_Det = mkDet "çok" Pl NoGen ;
+
+    without_Prep = mkPrep "" (Abess Neg) ;
+
+    please_Voc = {s = "lütfen"} ;
+
+    very_AdA = {s = "çok"} ;
+
+    why_IAdv   = {s = "neden"} ;
+    where_IAdv = {s = "nerede"} ;
+
+    can8know_VV = {
+      s = "(TODO: can8know_VV)"
+    } ;
+
+    can_VV = {
+      s = "(TODO: can_VV)"
+    } ;
+
+    must_VV = {
+      s = "(TODO: must_VV)"
+    } ;
+
+    not_Predet = {
+      s = "(TODO: not_Predet)"
+    } ;
+
+    quite_Adv = {
+      s = "bayağı"
+    } ;
+
+    -- TODO: there is probably a better translation for this and it is quite
+    -- tricky.
+    so_AdA = { s = "çok" } ;
+
+    -- TODO: not tested, probably wrong.
+    somePl_Det = { s = "bazı" ; n = Pl ; useGen = NoGen  } ;
+
+    -- TODO: not tested, probably wrong.
+    someSg_Det = { s = "bazı" ; n = Sg ; useGen = NoGen } ;
+
+    there7from_Adv = { s = "oradan" } ;
+
+    there7to_Adv = { s = "oraya" } ;
+
+    too_AdA = { s = "fazla" } ;
+
+    -- TODO: this depends on the linearization for `ComplVV` and is really a
+    -- morphological construct so it might be a bit tricky to implement.
+    want_VV = { s = "(TODO: want_VV)" } ;
+
+    whatPl_IP = { s = "neler" } ;
+
+    whatSg_IP = { s = "ne" } ;
+
+    -- Not sure what this is for given that we have separate functions for the
+    -- plural "what" case and the singular "what" case.
+    what_IP = { s = "ne" } ;
+
+    when_IAdv = { s = "ne zaman" } ;
+
+    which_IQuant = { s = "hangi" } ;
+
+    whoPl_IP = { s = "kimler" } ;
+
+    whoSg_IP = { s = "kim" } ;
+
+    -- TODO: depends on `PredetNP`; test after the work in `NounTur.gf` has
+    -- been merged.
+    only_Predet = { s = "sadece" } ;
+
+    otherwise_PConj = { s = "aksi takdirde" } ;
+
+    part_Prep = { s = "(TODO: part_Prep)" ; c = Nom } ;
+
+    at_most_AdN = ss "en fazla" ; 
+
+    at_least_AdN = ss "en az" ; 
+
+    as_CAdv = {s = "kadar"; p = "kadar"} ; 
 
 }