# -*- cg-pre-pipe: "$GTHOME/giella-core/scripts/preprocess --abbr=$GTHOME/langs/sma/tools/preprocess/abbr.txt | hfst-optimised-lookup $GTHOME/langs/sma/src/analyser-disamb-gt-desc.hfstol | $GTHOME/giella-core/scripts/lookup2cg" -*- # Distributed under the terms of the GNU General Public License version 2 # or any later version. # =================================================================== # # =================================================================== # #!! !!! S O U T H S Á M I D I S A M B I G U A T O R # =================================================================== # # =================================================================== # # =========================================== # # Short table of contents # # =========================================== # # Delimiters # Tags and sets # Mapping rules # Disambiguation rules # One-cohort disambiguation - cycle 0 # Local disambiguation - cycles 1 and 2 # Cycle 2: Other local disambiguation # Cycle 3: Global disambiguation # Cycle 4: Syntactic disambiguation # Cycle 5: Post-syntactic morphological disambiguation # ========== # #!! !!!Delimiters # ========== # DELIMITERS = "<.>" "" "" "<...>" "<¶>" sent ; #!! # ============= # #!! !!Tags and sets # ============= # #!! * BOS/EOS LIST BOS = (>>>) (~~); LIST EOS = (<<<) (~~); # vislcg and CG-2 together. #!! * Tags declared as single-membered LISTs # ====================================== LIST N = N ; LIST Sg = Sg ; LIST Pl = Pl ; LIST Nom = Nom ; LIST Acc = Acc ; LIST Gen = Gen ; LIST Ine = Ine ; LIST Ela = Ela ; LIST Ill = Ill ; LIST Com = Com ; LIST Ess = Ess ; LIST PxSg1 = PxSg1 ; LIST PxSg2 = PxSg2 ; LIST PxSg3 = PxSg3 ; LIST PxDu1 = PxDu1 ; LIST PxDu2 = PxDu2 ; LIST PxDu3 = PxDu3 ; LIST PxPl1 = PxPl1 ; LIST PxPl2 = PxPl2 ; LIST PxPl3 = PxPl3 ; LIST V = V ; LIST Prs = Prs ; LIST Prt = Prt ; LIST Sg1 = Sg1 ; LIST Sg2 = Sg2 ; LIST Sg3 = Sg3 ; LIST Du1 = Du1 ; LIST Du2 = Du2 ; LIST Du3 = Du3 ; LIST Pl1 = Pl1 ; LIST Pl2 = Pl2 ; LIST Pl3 = Pl3 ; LIST Inf = Inf ; LIST Neg = Neg ; LIST ConNeg = ConNeg ; LIST Foc/gan = Foc/gan ; LIST PrfPrc = PrfPrc ; LIST PrsPrc = PrsPrc ; LIST VGen = VGen ; LIST Ger = Ger ; LIST Ind = Ind ; LIST Imp = Imp ; LIST ImpII = ImpII ; LIST Pot = Pot ; LIST Cond = Cond ; LIST Imprt = Imprt ; LIST IV = IV ; LIST TV = TV ; LIST Opt = Opt ; LIST Actor = Actor ; LIST Clt = Clt ; LIST A = A ; LIST Ord = Ord ; LIST Coll = Coll ; LIST ABBR = ABBR ; LIST ACR = ACR ; LIST QMARK = """ ; # ADD " FOR SYNTAX COLOURING. LIST PPUNCT = PUNCT ; SET PUNCT = PPUNCT - QMARK ; LIST Comp = Comp ; LIST Superl = Superl ; LIST Attr = Attr ; LIST Pron = Pron ; LIST Pers = Pers ; LIST Dem = Dem ; LIST Interr = Interr ; LIST Rel = Rel ; #LIST Adp = Adp ; LIST Po = Po ; LIST Pr = Pr ; LIST Adv = Adv ; LIST Interj = Interj ; LIST Indef = Indef ; LIST Num = Num ; LIST Prop = Prop ; #!! !Derivation tags LIST Der/htalle = Der/htalle ; #!! Der/htalle LIST Der/lg = Der/lg ; #!! Der/lg LIST Der/ijes = Der/ijes ; #!! Der/ijes LIST Der/ihks = Der/ihks ; #!! Der/ihks LIST Der/les = Der/les ; #!! Der/les LIST Der/ldihkie = Der/ldihkie ; #!! Der/ldihkie LIST Der/ldahke = Der/ldahke ; #!! Der/ldahke LIST Der/ldh = Der/ldh ; #!! Der/ldh LIST Der/ht = Der/ht ; #!! Der/ht LIST Der/l = Der/l ; #!! Der/l LIST Der/st = Der/st ; #!! Der/st LIST Der/d = Der/d ; #!! Der/d LIST Der/Car = Der/Car ; #!! Der/Car LIST Der/htj = Der/htj ; #!! Der/htj LIST Der/Dimin = Der/Dimin ; #!! Der/Dimin LIST Der/Rec = Der/Rec ; #!! Der/Rec LIST Der/laakan = Der/laakan ; #!! Der/laakan LIST Der/vuota = Der/vuota ; #!! Der/vuota LIST Der/adte = Der/adte ; #!! Der/adte LIST Der/alla = Der/alla ; #!! Der/alla LIST Der/eds = Der/eds ; #!! Der/eds LIST Der/PassL = Der/PassL ; #!! Der/PassL LIST Der/NomAg = Der/NomAg ; #!! Der/NomAg LIST Der/NomAct = Der/NomAct ; #!! Der/NomAct LIST Der/ahtje = Der/ahtje ; #!! Der/ahtje LIST Der/InchL = Der/InchL ; #!! Der/InchL LIST Der/PassS = Der/PassS ; #!! Der/PassS LIST Der/A = Der/A ; #!! Der/A LIST = ; #!! LIST Err/Orth = Err/Orth Err/Orth-a/á Err/Orth-nom/gen Err/Orth-nom/acc Err/DerSub Err/CmpSub Err/UnspaceCmp Err/HyphSub Err/SpaceCmp Err/Spellrelax err_orth_mt ; LIST Err/Orth-spes = Err/Orth-a/á Err/Orth-nom/gen Err/Orth-nom/acc Err/DerSub Err/CmpSub Err/UnspaceCmp Err/HyphSub Err/SpaceCmp Err/Spellrelax err_orth_a_á_mt err_orth_nom_acc_mt err_orth_nom_gen_mt ; #!! * Sets for Semantic tags # ------------- LIST Sem/Act = Sem/Act Sem/Act_Group Sem/Act_Plc Sem/Act_Route; #Activity #Ollet guhká dainna X-Com (Veaigin bisánasttii Ristenge geahččat, man guhkás Lemet-gázzi juo ledje ollen dainna divvunbargguin.) # bargu LIST Sem/Amount = Sem/Amount Sem/Amount_Build Sem/Amount_Semcon; LIST Sem/Ani = Sem/Ani Sem/Ani_Body-abstr_Hum Sem/Body Sem/Body-abstr Sem/Body-abstr_Prod-audio_Semcon Sem/Body_Body-abstr Sem/Body_Clth Sem/Body_Food Sem/Body_Group_Hum Sem/Body_Hum Sem/Body_Mat Sem/Body_Measr Sem/Body_Obj_Tool-catch Sem/Body_Plc Sem/Body_Time Sem/Body_Group_Hum_Time ; #Animal #bivdit X-Loc veahki. LIST Sem/Aniprod = Sem/Aniprod Sem/Aniprod_Hum Sem/Aniprod_Obj-clo Sem/Aniprod_Perc-phys Sem/Aniprod_Plc ; LIST Sem/Body = Sem/Body Sem/Body_Body-abstr Sem/Body_Clth Sem/Body_Food Sem/Body_Group_Hum Sem/Body_Hum Sem/Body_Mat Sem/Body_Measr Sem/Body_Obj_Tool-catch Sem/Body_Plc Sem/Body_Time Sem/Body_Group_Hum_Time ; #Body #Buohkat ceggejedje gieđa. Mon gal attán dutnje bealji vuollái. Go duos dás speččohallá bahtii de gal buollá bargomokta! Čuolda basttii sus alemii. LIST Sem/Body-abstr = Sem/Ani_Body-abstr_Hum Sem/Body-abstr Sem/Body-abstr_Prod-audio_Semcon Sem/Body_Body-abstr ; # jierbmi, siellu, ... LIST Sem/Build = Sem/Amount_Build Sem/Ani_Build Sem/Ani_Build_Hum_Txt Sem/Build Sem/Build_Build-part Sem/Build_Clth-part Sem/Build_Edu_Org Sem/Build_Event_Org Sem/Build_Org Sem/Build_Route Sem/Build_Obj ; #Build #Sem #Syn: hukset X deike LIST Sem/Build-part = Sem/Ani_Build-part Sem/Build-part Sem/Build-part_Plc Sem/Build_Build-part ; #Building part, e.g. room #Sem:+con,+count,-mova #Syn: mannat X sisa LIST Sem/Cat = Sem/Cat ; LIST Sem/Clth = Sem/Body_Clth Sem/Clth Sem/Clth_Hum ; #Clothes #Sem #bidjat X ala/badjelii. LIST Sem/Clth-jewl = Sem/Clth-jewl Sem/Clth-jewl_Curr Sem/Clth-jewl_Money Sem/Clth-jewl_Plant Sem/Org_Clth-jewl Sem/Clth-jewl_Org ; #Clothes jewelry #Synt: bidjat X ala. *bidjat X badjelii. LIST Sem/Clth-part = Sem/Build_Clth-part Sem/Clth-part ; LIST Sem/Ctain = Sem/Ctain Sem/Ctain_Feat-phys Sem/Ctain_Furn Sem/Ctain_Tool Sem/Ctain_Tool-measr ; #Container #Sem:+mova #Syn: bidjat juoidá X sisa LIST Sem/Ctain-abstr = Sem/Ctain-abstr Sem/Ctain-abstr_Org ; LIST Sem/Ctain-clth = Sem/Ctain-clth Sem/Ctain-clth_Plant Sem/Ctain-clth_Veh ; LIST Sem/Curr = Sem/Clth-jewl_Curr Sem/Curr Sem/Curr_Org; #Syn: Mun mávssan dasa 100 X. Dat máksá 100 X. LIST Sem/Date = Sem/Date ; LIST Sem/Dance = Sem/Dance Sem/Dance_Org Sem/Dance_Prod-audio ; #Ollugat dánso swinga lávddi ovddabeale. LIST Sem/Dir = Sem/Dir ; LIST Sem/Domain = Sem/Domain Sem/Domain_Food-med Sem/Domain_Prod-audio; LIST Sem/Drink = Sem/Drink ; LIST Sem/Edu = Sem/Build_Edu_Org Sem/Edu Sem/Edu_Event Sem/Edu_Group_Hum Sem/Edu_Mat Sem/Edu_Org ; # vázzit X-Acc. addit X-Acc. X:s lea beroštupmi. LIST Sem/Event = Sem/Build_Event_Org Sem/Edu_Event Sem/Event Sem/Event_Food Sem/Event_Hum Sem/Event_Plc Sem/Event_Plc-elevate Sem/Event_Time ; # X maŋŋel mun ferten vuolgit. X:s leat ollu olbmot. *Lean leamaš doppe X:Gen. #LIST Sem/Feat-cogn = Sem/Feat-cogn ; # oaidnu, oamedovdu, viissisvuohta #LIST Sem/Feat-phys = Sem/Feat-phys ; # oaidnu LIST Sem/Feat = Sem/Feat Sem/Feat_Plant ; #many -vuohta and -dat nouns LIST Sem/Feat-measr = Sem/Feat-measr ; LIST Sem/Feat-phys = Sem/Ctain_Feat-phys Sem/Feat-phys Sem/Feat-phys_Tool-write Sem/Feat-phys_Veh Sem/Feat-phys_Wthr ; LIST Sem/Feat-psych = Sem/Feat-psych Sem/Feat-psych_Hum ; #not touchable, not countable ("beroštupmi" "gáđašvuohta" "mokta" "sáhkkiivuohta") - Mus lea X, Son lea X-adj persovdna # muitu LIST Sem/Fem = Sem/Fem (ant f) ; LIST Sem/Food = Sem/Body_Food Sem/Event_Food Sem/Food Sem/Food_Perc-phys Sem/Food_Plant ; #Syn: bassit/borrat/sohpat X-Acc. mon im læk borram dam guolest. mun boran iđitbiepmu. mus lea iđitbiebmu fárus. birgui borastuvvat LIST Sem/Food-med = Sem/Domain_Food-med Sem/Food-med ; LIST Sem/Furn = Sem/Ctain_Furn Sem/Furn ; #Syn: bidjat X vissui: čohkkedit X ala LIST Sem/Game = Sem/Game Sem/Game_Obj-play ; LIST Sem/Geom = Sem/Geom Sem/Geom_Obj ; LIST Sem/Group = Sem/Act_Group Sem/Ani_Group Sem/Ani_Group_Hum Sem/Body_Group_Hum Sem/Edu_Group_Hum Sem/Group Sem/Group_Hum Sem/Group_Hum_Org Sem/Group_Hum_Plc Sem/Group_Hum_Prod-vis Sem/Group_Org Sem/Group_Sign Sem/Group_Txt Sem/Body_Group_Hum_Time ; #Bárdni lea riggámus sogas. -Plc LIST Sem/Hum = Sem/Ani_Body-abstr_Hum Sem/Ani_Build_Hum_Txt Sem/Ani_Group_Hum Sem/Ani_Hum Sem/Ani_Hum_Plc Sem/Ani_Hum_Time Sem/Aniprod_Hum Sem/Body_Group_Hum Sem/Body_Hum Sem/Clth_Hum Sem/Edu_Group_Hum Sem/Event_Hum Sem/Feat-psych_Hum Sem/Group_Hum Sem/Group_Hum_Org Sem/Group_Hum_Plc Sem/Group_Hum_Prod-vis Sem/Hum Sem/Hum_Lang Sem/Hum_Lang_Plc Sem/Hum_Lang_Time Sem/Hum_Obj Sem/Hum_Org Sem/Hum_Plant Sem/Hum_Plc Sem/Hum_Tool Sem/Hum_Veh Sem/Hum_Wthr Sem/Body_Group_Hum_Time ; #Human LIST Sem/Hum-kin = Sem/Hum-kin ; #Human kinship LIST Sem/Hum-abstr = Sem/Hum-abstr ; LIST Sem/Ideol = Sem/Ideol ; LIST Sem/Lang = Sem/Lang ; #Syn: hállat X:illatiiva. Son hállá X:Acc. Jorgalan dán X:lokatiiva Y:illatiiva. LIST Sem/Mal = Sem/Mal (ant m) ; LIST Sem/Mat = Sem/Body_Mat Sem/Edu_Mat Sem/Mat Sem/Mat_Plant Sem/Mat_Txt ; # Material #Sem: cannot pass through it without destroying it, ráhkadit juoidá X:s. Niibi lea X-Loc-Sg. #Dán báhkis šaddá čáppa gohppu. LIST Sem/Measr = Sem/Body_Measr Sem/Measr Sem/Measr_Time Sem/Measr_Sign ; # Lea lassánan 100 X-Loc. LIST Sem/Money = Sem/Clth-jewl_Money Sem/Money Sem/Money_Obj Sem/Money_Txt ; #Mun mávssán X. Mii ávžžuhit vuovdit alimus X-Ill (haddái). ruhtadit doaibmabušeahtain. vuovdit/bidjat ruđast - selge for penger LIST Sem/Obj = Sem/Body_Obj_Tool-catch Sem/Geom_Obj Sem/Hum_Obj Sem/Money_Obj Sem/Obj Sem/Obj_Semcon Sem/Obj_State Sem/Build_Obj Sem/Obj-com ; # concrete object, movable #Syn: Mun guoskkán X-Acc. Dá livčče dutnje X-Nom (sabehat). čievččastit geađggi johkii #LIST Sem/Obj-com = Sem/Obj-com ; #communication object #hálan telefuvnnas/e-poasttas #sádden X bokte LIST Sem/Obj-clo = Sem/Aniprod_Obj-clo Sem/Obj-clo ; #bidjat juoidá Obj-clo sisa LIST Sem/Obj-cogn = Sem/Obj-cogn ; LIST Sem/Obj-el = Sem/Obj-el ; # moveable, countable, touchable, electrical objects that can be turned on and of, mus lea X alde, mun bijan X ala LIST Sem/Obj-ling = Sem/Obj-ling ; LIST Sem/Obj-play = Sem/Obj-play Sem/Game_Obj-play Sem/Obj-play_Sport ; LIST Sem/Obj-rope = Sem/Obj-rope ; LIST Sem/Obj-surfc = Sem/Obj-surfc ; LIST Sem/Org = Sem/Build_Edu_Org Sem/Build_Event_Org Sem/Build_Org Sem/Ctain-abstr_Org Sem/Curr_Org Sem/Dance_Org Sem/Edu_Org Sem/Group_Hum_Org Sem/Group_Org Sem/Hum_Org Sem/Org Sem/Org_Clth-jewl Sem/Org_Rule Sem/Org_Txt Sem/Org_Veh Sem/Clth-jewl_Org ; LIST Sem/Part = Sem/Part Sem/Part_Prod-cogn Sem/Part_Substnc ; # bealli, oassi LIST Sem/Perc-cogn = Sem/Perc-cogn ; LIST Sem/Perc-emo = Sem/Perc-emo ; #Sem not touchable, not countable ("ballu" "suhttu" "vaši") - Mun dovddan garra X. Mun lean X+Loc # moraš LIST Sem/Perc-phys = Sem/Aniprod_Perc-phys Sem/Food_Perc-phys Sem/Perc-phys Sem/Perc-emo_Wthr ; # anistupmi, oaidnu LIST Sem/Perc-psych = Sem/Perc-psych ; LIST Sem/Phonenr = Sem/Phonenr ; LIST Sem/Plant = Sem/Clth-jewl_Plant Sem/Ctain-clth_Plant Sem/Feat_Plant Sem/Food_Plant Sem/Hum_Plant Sem/Mat_Plant Sem/Plant Sem/Plant_Tool Sem/Plant_Tool-measr ; #čáhccet X-Acc #X-Nom šaddet LIST Sem/Plant-part = Sem/Plant_Plant-part Sem/Plant-part ; #čáhccet X-Acc #X-Nom šaddet LIST Sem/Plc = Sem/Act_Plc Sem/Ani_Hum_Plc Sem/Ani_Plc Sem/Ani_Plc_Txt Sem/Aniprod_Plc Sem/Body_Plc Sem/Build-part_Plc Sem/Event_Plc Sem/Event_Plc-elevate Sem/Group_Hum_Plc Sem/Hum_Lang_Plc Sem/Hum_Plc Sem/Plc Sem/Plc_Pos Sem/Plc_Route Sem/Plc_Substnc Sem/Plc_Substnc_Wthr Sem/Plc_Time Sem/Plc_Tool-catch Sem/Plc_Wthr; LIST Sem/Plc-abstr = Sem/Plc-abstr Sem/Plc-abstr_Rel_State Sem/Plc-abstr_Route ; LIST Sem/Plc-elevate = Sem/Plc-elevate ; LIST Sem/Plc-line = Sem/Plc-line ; LIST Sem/Plc-water = Sem/Plc-water ; LIST Sem/Pos = Sem/Plc_Pos Sem/Pos ; #Social position (job etc.) LIST Sem/Process = Sem/Process ; LIST Sem/Prod = Sem/Prod ; #jáhkku LIST Sem/Prod-audio = Sem/Body-abstr_Prod-audio_Semcon Sem/Dance_Prod-audio Sem/Domain_Prod-audio Sem/Prod-audio Sem/Prod-audio_Txt ; #attáldat LIST Sem/Prod-cogn = Sem/Part_Prod-cogn Sem/Prod-cogn Sem/Prod-cogn_Txt ; # muittuhus, ipmárdus, jáhkku, osku, vuordámuš, viisodat? diehtu? LIST Sem/Prod-ling = Sem/Prod-ling ; LIST Sem/Prod-vis = Sem/Group_Hum_Prod-vis Sem/Prod-vis ; LIST Sem/Rel = Sem/Plc-abstr_Rel_State Sem/Rel ; LIST Sem/Route = Sem/Act_Route Sem/Build_Route Sem/Plc-abstr_Route Sem/Plc_Route Sem/Route ; LIST Sem/Rule = Sem/Org_Rule Sem/Rule ; LIST Sem/Semcon = Sem/Amount_Semcon Sem/Body-abstr_Prod-audio_Semcon Sem/Obj_Semcon Sem/Semcon Sem/Semcon_Txt ; # semantic concept LIST Sem/Sign = Sem/Group_Sign Sem/Sign Sem/Measr_Sign ; LIST Sem/Sport = Sem/Obj-play_Sport Sem/Sport ; LIST Sem/State = Sem/Plc-abstr_Rel_State Sem/State Sem/Obj_State ; # lihkku, ipmilbalolašvuohta, mokta, váivi, váttisvuohta LIST Sem/State-sick = Sem/State-sick ; LIST Sem/Substnc = Sem/Plc_Substnc Sem/Plc_Substnc_Wthr Sem/Substnc Sem/Substnc_Wthr Sem/Part_Substnc ; #Sem: can pass through it without destroying it LIST Sem/Sur = Sem/Sur cog ; LIST Sem/Symbol = Sem/Symbol ; LIST Sem/Time = Sem/Ani_Hum_Time Sem/Ani_Time Sem/Body_Time Sem/Event_Time Sem/Hum_Lang_Time Sem/Measr_Time Sem/Plc_Time Sem/Time Sem/Time_Wthr Sem/Body_Group_Hum_Time ; LIST Sem/Tool = Sem/Ctain_Tool Sem/Hum_Tool Sem/Lang_Tool Sem/Plant_Tool Sem/Tool ; LIST Sem/Tool-catch = Sem/Body_Obj_Tool-catch Sem/Plc_Tool-catch Sem/Tool-catch ; LIST Sem/Tool-clean = Sem/Tool-clean ; LIST Sem/Tool-it = Sem/Tool-it ; LIST Sem/Tool-measr = Sem/Ctain_Tool-measr Sem/Plant_Tool-measr Sem/Tool-measr ; LIST Sem/Tool-music = Sem/Tool-music ; LIST Sem/Tool-write = Sem/Feat-phys_Tool-write Sem/Tool-write ; LIST Sem/Txt = Sem/Ani_Build_Hum_Txt Sem/Ani_Plc_Txt Sem/Group_Txt Sem/Mat_Txt Sem/Money_Txt Sem/Org_Txt Sem/Prod-audio_Txt Sem/Prod-cogn_Txt Sem/Semcon_Txt Sem/Txt ; #Syn: Čále dearvvuođaidat guossegirjámet! # das čuožžu LIST Sem/Veh = Sem/Ani_Veh Sem/Ctain-clth_Veh Sem/Feat-phys_Veh Sem/Hum_Veh Sem/Org_Veh Sem/Veh Sem/Veh-cntain ; #Syn: Vuolggán X:Com LIST Sem/Veh-cntain = Sem/Veh-cntain ; #Manan X sisa. #vuolggán X:Com. LIST Sem/Wthr = Sem/Feat-phys_Wthr Sem/Hum_Wthr Sem/Perc-emo_Wthr Sem/Plc_Substnc_Wthr Sem/Plc_Wthr Sem/Substnc_Wthr Sem/Time_Wthr Sem/Wthr ; #Sem: Birget X:Ess. jietna gullui X-Loc (balvvas). Olgun/Odne lea + Wthr-Nom. ?Man ollu lea Wthr-Nom? Wthr-Gen maŋŋel boahtá X. Makkár lea dálki/siivu? #Ex: árvi, borga, muohtá, čoaskadálki, baján, bievla, bodnejiekŋa, bivdodálki, gasis, guoldu LIST Sem/Wpn = Sem/Wpn ; #oaggut oaggunstákkuin, goddit X+Com LIST Sem/Date = Sem/Date ; LIST Sem/Year = Sem/Year ; SET TIME-N-SET = N + Sem/Time ; LIST NOT-TIME = "dulvi" "lohpi" "vuorru" ; SET TIME-N = TIME-N-SET - NOT-TIME ; SET FIRSTNAME = Sem/Fem OR Sem/Mal ; LIST CS = CS ; LIST CC = CC ; LIST Pcle = Pcle ; LIST CLB = CLB ; LIST LEFT = LEFT ; LIST RIGHT = RIGHT ; LIST COMMA = "," ; LIST Arab = Arab ; #!! * Syntactic tags LIST @CNP = @CNP ; #!! @CNP LIST @CVP = @CVP ; #!! @CVP LIST @+FAUXV = @+FAUXV ; #!! @+FAUXV LIST @+FMAINV = @+FMAINV ; #!! @+FMAINV LIST @-FAUXV = @-FAUXV ; #!! @-FAUXV LIST @-FMAINV = @-FMAINV ; #!! @-FMAINV SET MAINV = @-FMAINV OR @+FMAINV ; #!! MAINV LIST @ADVL = @ADVL ; LIST @>ADVL = @>ADVL ; LIST @ADVL< = @ADVL< ; LIST @ = @ADVL> ; LIST @-FADVL = @-FADVL ; LIST @A< = @A< ; LIST @>A = @>A ; LIST @ActioN> = @ActioN> ; LIST @APP = @APP ; LIST @APP-N< = @APP-N< ; LIST @APP-Pron< = @APP-Pron< ; LIST @APP>Pron = @APP>Pron ; LIST @APP-Num< = @APP-Num< ; LIST @APP-ADVL< = @APP-ADVL< ; LIST @CMPND = @CMPND ; LIST @COMP-CS< = @COMP-CS< ; LIST @HAB = @HAB ; LIST @INTERJ = @INTERJ ; LIST @MEASURE = @MEASURE ; LIST @>N = @>N ; LIST @N< = @N< ; LIST @NNum> = @NNum>; LIST @NumN< = @NumN<; LIST @>Num = @>Num; LIST @Num< = @Num< ; LIST @NPron< = @NPron< ; LIST @NQ< = @NQ< ; LIST @NUM-PRON = @NUM-PRON ; #LIST @NUMBER = @NUMBER ; LIST @OBJ = @OBJ ; LIST @ = @OBJ> ; LIST @OPRED = @OPRED ; LIST @ = @OPRED> ; LIST @PCLE = @PCLE ; LIST @HNOUN = @HNOUN ; LIST @PrcN> = @PrcN> ; LIST @PronN< = @PronN< ; LIST @PronN> = @PronN> ; LIST @Pron< = @Pron< ; LIST @>Pron = @>Pron ; LIST @P< = @P< ; LIST @>P = @>P ; LIST @SPRED = @SPRED ; LIST @ = @SPRED> ; LIST @SUBJ = @SUBJ ; LIST @ = @SUBJ> ; LIST @SUBJ-QH = @SUBJ-QH ; LIST @TITLE = @TITLE ; LIST @VOC = @VOC ; LIST @X = @X ; #!! * More sets LIST REAL-TITLE = "prof" "cand.philol" ; LIST OFFICE = "gånka" ; SET TITLE = REAL-TITLE OR OFFICE ; LIST Guess = Guess ; # SET TITLE = REAL-TITLE OR TITLE-LIKE OR PROFESSION OR OFFICE ; LIST CASES = Nom Acc Gen Ill Ine Ela Com Ess ; LIST ADVLCASE = Ine Ill Com Ela ; LIST NUMBER = Sg Pl ; LIST NOMEN-NOT-PRON = N A Indef Num Ord ; # they can have Dem as attr LIST WORD = N A V Adv Pron Interj Num CS CC Pcle Pr Po ; SET REALWORD = WORD - Num - Ord ; SET REALWORD-NOTABBR = WORD - Num - Ord - ABBR ; SET NOT-QMARK = WORD OR CLB OR PUNCT - QMARK ; LIST NOT-ADV-INT-PCLE = N A Num Pron A V CC CS ; SET NOT-ADV = WORD - Adv ; SET NP-HEAD = Num OR N OR Pron ; SET PRE-NP-HEAD = (Prop Attr) OR (Prop @>N) OR (A Attr) OR (ABBR Attr) OR (Pron Pers Gen) OR (Pron Logo Gen) OR (N Gen) OR Num OR (Cmpnd) OR CC OR (Pron Dem) OR (Pron Refl Gen) OR (Pron Indef) OR (PrfPrc @>N) OR (PrfPrc @>N) OR (PrsPrc) OR (A Ord) OR Attr OR ("gånka") ; # The strict version of items that can only be premodifiers, not parts of the predicate - copied from sme-dis.rle SET NP-MEMBER = PRE-NP-HEAD OR N ; SET GUJMIE = (Pl Gen) OR (Pl1 Gen) OR (Pl2 Gen) OR (Pl3 Gen) ; SET NPNH = WORD - PRE-NP-HEAD OR (ABBR) ; # NPNH = "NOT-PRE-NP-HEAD" SET NPNHA = WORD - PRE-NP-HEAD OR (ABBR) OR Adv ; SET NPNHAI = WORD - PRE-NP-HEAD - Adv - Indef ; #!! * Noun sets SET PX = PxSg1 OR PxSg2 OR PxSg3 OR PxDu1 OR PxDu2 OR PxDu3 OR PxPl1 OR PxPl2 OR PxPl3 ; LIST FAMILY = "aahka" "aahkove" "aajja" "aajjove" "aajkoehkadtjh" "aajkohke" "aajnehke" "aarpije" "aehtjaahka" "aehtjaajja" "aehtjie" "baernie" "boelve" "daktere" "eejhtegh" "elkie" "fijrie" "fuelhkie" "gaalla" "geelle" "geellege" "gåeskie" "gåeskuve" "gåetehke" "gåmma" "gåmmebe" "ietnie" "jiekie" "jiekuve" "jijmie" "jijmuve" "jyöne" "jöömmetje" "krist’aehtjie" "krist’ietnie" "laahkoe" "laevie" "laeviehkadtjh" "maadteraahka" "maadteraajja" "maadtoe" "maake" "maaketje" "maana" "meata" "moehrele" "muahra" "månnja" "mænnja" "neapede" "neebne" "nååte" "nïejte" "onnevïelle" "onneåabpa" "seasa" "slïekte" "stoerrevïelle" "stoerreåabpa" "sïesele" "tjidtjaahka" "tjidtjaajja" "tjidtjie" "tjietsie" "tjietsuve" "voehpe" "voehpealmetjh" "vuanove" "vïelle" "vïjve" "åabpa" "åabpetjh" "åerpele" "åerpene" "åerpienadtjh" "vïelletjh" ; # those can both be buildings and represent humans # Duomma:lea LIST INSTITUTION = "jarnge" "kultuvrejarnge" "raerie" "kultuvreraerie" ; # the previous items can both represent abstract processes and humans LIST MEDIA = "media" "radio" "TV" ; # "oahpahus#rádio" "rádio" "televišuvdna" ; LIST ORGANIZATION = (Prop Org) "digkie" "digkiegåetie" "skuvlekontovre" ; LIST EDUCATION = "skuvle" ; LIST CURRENCY = "denaare" "kråvna" "kråvnoe" "kravna" "kr" "ru" "¢" "€" "$"; # suits to LANGUAGE Gen LIST LESSON = "kuvsje" ; LIST OTHER-INDUSTRY = "bálvalus#ealáhus" ; # Not related to certain political verbs # Duomma:lea LIST POLITICAL-INDUSTRY = "alkohol#politihkka" ; # Related to certain political verbs # Duomma:lea SET INDUSTRY = POLITICAL-INDUSTRY OR OTHER-INDUSTRY ; LIST NORGA = "Dánmárku-#Norga" "davvi#Norga" "Davvi-Norga" "gaska-#Norga" "Norga" "norga" "norgga#bealli" "Norgga-#ruoŧŧa" ; #!! * Verb sets # The set REALCOPULAS is smaller than COPULAS, made for verbs with PrfPrc complements: Seammás REALCOPULAS son dovdan iežas... LIST REALCOPULAS = "lea" ; # The set COPULAS is for predicative constructions LIST COPULAS = REALCOPULAS ; SET V-NOT-COP = V - COPULAS ; LIST MOD-ASP = "galkedh" "tjoeveridh" "tjoeredh" "sååjhtedh" ; # Verbs that never have arguments of their own LIST AUX-OR-MAIN = "aelkedh" "edtjedh" "maehtedh" "sïjhtedh" "sjïdtedh" "båetedh" "åadtjodh" ; # These verbs can take arguments, so they do not belong in the AUX group, # but they are nevertheless mapped to (@+FAUXV). SET AUX = COPULAS OR Neg OR MOD-ASP OR AUX-OR-MAIN ; SET VFIN = Ind OR Imprt ; SET VFIN-NOT-AUX = VFIN - AUX ; SET V-MAIN = V - COPULAS - Neg - MOD-ASP ; SET VFIN-NOT-NEG = VFIN - Neg ; # this might be to strict, besides, "iige" can be written "ii ge" SET NOT-NP = VFIN-NOT-NEG OR ConNeg OR Inf OR Pcle OR Interj OR CS ; # The set NP-BOUNDARY is bigger. LIST INFV = "aavoedidh" "aelkedh" "buektiehtidh" "provhkedh" ; # These are often followed by an infitive. LIST ILLV = "båetedh" "juhtedh" "ringkedh" ; LIST COMV = "budtedh" "pruvredh" ; LIST COM-ADJ = "latjkes" ; LIST STV = Der/ehte ; LIST ILLADV = "bæjjese" "davvese" "daase" "dahkoe" "diekie" "doekoe" "dohkoe" "jallese" "mænngese" "noerhtese" "olkese" "vueliebasse" "våålese" "åarjese" "åvtese" ; LIST INEADV = "daebpene" "debpene" "duebpene" "dubpene" "daagkoe" "dagkoe" "doegkoe" "dogkoe" "daajkoe" "dajkoe" "doejkoe" "dojkoe" ; LIST ELAADV = "daebpede" "debpede" "duebpede" "dubpede" "dåvvese" ; LIST GUKTIEGOSSE = ("gusnie" Adv) ("guktie" Adv) ("gosse" Adv) ("gåessie" Adv) ("gogkoe" Adv) ("gogka" Adv) ("gustie" Adv) ("gåabpelen" Adv) ("gåabpoeh" Adv) ("gåabpah" Adv) ("gåabph" Adv) ("gubpene" Adv) ("gubpede" Adv) ("gubpelde" Adv) ("gumhtie" Adv) ("mannemes" Adv) "magkeres" "magkere" "guktems" "gumhtems" "gumhteme" ("mijladtje" Adv) ("man" Adv) "mannasinie" ("mejnie" Adv) ("mennie" Adv) ("mestie" Adv); LIST DAESNIE = ("daesnie" Adv) ("åvtesne" Adv) ("sjisjnjielisnie" Adv) ("dusnie" Adv) ("miehtjielisnie" Adv) ("dubpielisnie" Adv) ("jilliebisnie" Adv) ("luvlielisnie" Adv) ("ålkoelisnie" Adv) ("gusnie" Adv) ("luvliegisnie" Adv) ("åarjalisnie" Adv) ("minngielisnie" Adv) ("bijjielisnie" Adv) ("sisnjielisnie" Adv) ("åarjielisnie" Adv) ("åerjielisnie" Adv) ("vuelielisnie" Adv) ("debpielisnie" Adv) ("duebpielisnie" Adv) ("åarjelisnie" Adv) ("noerhtelisnie" Adv) ("jillielisnie" Adv) ("luvliebisnie" Adv) ("vueliebisnie" Adv); LIST DAESTIE = ("daestie" Adv) ("åvteste" Adv) ("sjisjnjielistie" Adv) ("dustie" Adv) ("miehtjielistie" Adv) ("dubpielistie" Adv) ("jilliebistie" Adv) ("luvlielistie" Adv) ("ålkoelistie" Adv) ("gustie" Adv) ("luvliegistie" Adv) ("åarjalistie" Adv) ("minngielistie" Adv) ("bijjielistie" Adv) ("sisnjielistie" Adv) ("åarjielistie" Adv) ("åerjielistie" Adv) ("vuelielistie" Adv) ("debpielistie" Adv) ("duebpielistie" Adv) ("åarjelistie" Adv) ("noerhtelistie" Adv) ("jillielistie" Adv) ("luvliebistie" Adv) ("vueliebistie" Adv); LIST ADV-MOD-ADV = "aaptsah" "aaptsh" "aave" "gaaje" "gaajh" "joekoen" "ååpsen" "fer" "for" "dan" ("mij" Sg Gen) "øøvre" ; LIST ADV-MOD-NP = "ållesth" "allesth" "aave" "aejvie" "dovne" "eevre" "öövre" "ov" "ovve" "rikti" "ajve" "barre" "gaaje" "gaajh" "joekoen" "hujht" "hujhtskabph" "lïgke" "lïjhke" "gujht" "læjhkan" "van" "darhkan" "eantan" "man" "mahte" "ovrehte" "raakte" "varke" "varki" "aaptsah" "aaptsh" "medtie" ; LIST GRADE-ADV = "aaptsah" "aaptsh" "ållesth" "allesth" "ååpsen" "aave" "aejvie" "dovne" "eevre" "ajve" "barre" ("jienebe" Adv) ("jeenjebe" Adv) "ov" "ovve" "rikti" "gaaje" "gaajh" "joekoen" "åpmohts" "åpmehts" "aajmohts" "åpmoeh" ("man" Adv) "mahte" "ovrehte" "raakte" "varke" "varki" "seamma" "öövre" "uhtjienadtje" "sagke" ; #!! * BOUNDARY SETS SET S-BOUNDARY = (Pron Interr) OR (Pron Rel) OR ("men") OR ("mohte") OR ("\;") OR (":") OR ("-") OR ("–") OR CS OR GUKTIEGOSSE OR BOS OR EOS ; SET REALCLB = CLB - COMMA ; SET SV-BOUNDARY = S-BOUNDARY OR Inf OR V-MAIN - PrsPrc ; SET NP-BOUNDARY = BOS OR EOS OR REALCLB OR VFIN OR Inf OR (Actio Ess) OR ConNeg OR VGen OR Po OR Pr OR Pcle OR Interj OR CS ; # The following set contains the one-letter abbrs from abbr-sma-lex.txt LIST SOMESMALLABBR = "a" "r" "c" "d" "e" "f" "f" "g" "h" "i" "l" "m" "b" "j" "s" ; LIST SOMECAPABBR = "A" "R" "C" "D" "E" "F" "F" "G" "H" "I" "L" "M" "B" "J" "S" ; # Der sets LIST V-DER = Der/didh Der/htidh Der/alle Der/stidh Der/ehte Der/lidh Der/htalle Der/lgidh Der/htjidh Der/htovve Der/htoeve Der/htehte Der/ahtje Der/gåete Der/ehte Der/adte Der/eds Der/PassL; LIST V-DER-SUF = (".*didh"r) (".*htidh"r) (".*alledh"r) (".*stidh"r) (".*ehtedh"r) (".*lidh"r) (".*lgidh"r) (".*htjidh"r) (".*htovvedh"r) (".*htoevedh"r) (".*htehtedh"r) (".*ahtjedh"r) (".*gåetedh"r) (".*ehtedh"r) (".*gåetedh"r) (".*adtedh"r) (".*edsedh"r) (".*sovvedh"r) ; LIST N-DER = Der/NomAct Der/NomAg Der/Dimin Der/voete ; LIST N-DER-SUF = (".*immie"r) (".*me"r) (".*æjja"r) (".*ije"r) (".*tje"r) (".*voete"r) ; LIST A-DER = Der/ldihkie Der/ijes Der/ihks Der/les Der/ldahke Der/ldh ; LIST A-DER-SUF = (".*ldihkie"r) (".*ijes"r) (".*ihks"r) (".*les"r) (".*ldahke"r) (".*ldh"r) ; LIST PASS = Der/PassS Der/PassL ; SET LEX-V = V - V-DER; SET LEX-N = N - N-DER; SET LEX-A = A - A-DER; LIST VERB-FORMS = Inf Sg1 Sg2 Sg3 ; SET FORMS = VERB-FORMS OR WORD ; LIST BUILDING = "gåetie" ; #LIST ROOM = "ladnja" "oadem#ladnja" ; LIST GEOGRAPHICAL-PLACE = "dajve" "deava" "gaedtie" "gádde" "jaevrie" "vaerie" "miehtsie"; LIST POLITICAL-PLACE = "rïjhke"; LIST GENERAL-PLACE = "sïjte"; #LIST SPEC-PLACE = "värro#áltár" ; # for Ill disamb LIST PLACE-P = "alkalisnie" "alkalistie" "avtelen" "avteli" "avtelisnie" "avtelistie" "avtem" "avtene" "baakta" "baaktoe" "baalte" "baaltelen" "baalteli" "baaltelisnie" "baaltelistie" "baaltes" "baaltese" "bealan" "bealese" "bealesne" "bealeste" "bielehth" "bielelen" "bieleli" "bielesne" "bieleste" "bielielisnie" "bielielistie" "bijjelen" "bijjeli" "bijjielisnie" "bijjielistie" "boehten" "boehth" "bïjre" "daebpiebisnie" "daebpelen" "daebpeli" "daebpielisnie" "daebpielisnie" "daebpielistie" "debpeli" "debpielisnie" "debpielistie" "duakan" "dubpelen" "dubpeli" "dubpielisnie" "dubpielistie" "duebpelen" "duebpeli" "duebpielisnie" "duebpielistie" "duekelen" "duekesne" "duekeste" "duekiem" "jillelen" "jilleli" "jillielisnie" "jillielistie" "luvleli" "luvlielisnie" "luvlielistie" "miehtjelde" "miehtjelen" "miehtjeli" "miehtjielasse" "miehtjielisnie" "miehtjielistie" "nïlle" "noerhtelen" "noerhteli" "noerhtelisnie" "noerhtelistie" "nualan" "nueleste" "nuelesne" "nuelie" "nueliem" "raajan" "raejeste" "rastah" "sisnie" "sisnjeli" "sisnjielisnie" "sisnjielistie" "sistie" "sjisjnjeli" "sjisjnjielisnie" "sjisjnjielistie" "sïjse" "tjïddh" "tjïrrh" "uvte" "vuelelen" "vueleli" "vuelielisnie" "vuelielistie" "vuestelen" "vuestie" "vuestielisnie" "vuestielistie" "vööste" "åarjalisnie" "åarjalistie" "åarjelen" "åarjeli" "åarjielisnie" "åarjielistie" "åelesne" "åerjielisnie" "ålkoelisnie" "ålkoelistie" "ålkolen" "ålkoli" "åvtelen" "åvteli" "åvtelisnie" "åvtelistie" "åvtem" "åvtene" "åvtese" "åvtesne" "åvteste" "åvtetji" "åvtine" "lïhkebe" "vuelege" ; LIST TIME = "biejjie" "b" "gïjre" "giesie" "tïjme" "dygne" "bodte" "bådte" "tjaktje" "tjaktjegiesie" "tjaktjedaelvie" "daelvie" "tsåahka" "jaepie" "jïjje" "jahke" "aske" "reakedsbiejjie" "våhkoe" "vahka" "måanta" "dæjsta" "aejkie" ; LIST CONCRETE-ROUTE = (".*raejkien"r) (".*raejkiem"r) "guhkies" ; LIST ABSTR-ROUTE = "mïjle" (".*mïjle"r) "lehkie#mïjletje" (".*meetere"r) ; SET ROUTE = CONCRETE-ROUTE OR ABSTR-ROUTE ; SET PLACE = GEOGRAPHICAL-PLACE OR POLITICAL-PLACE OR GENERAL-PLACE OR ROUTE OR BUILDING ; # ============== # #!! !!!Disambiguation # # ============== # BEFORE-SECTIONS #!! !!@CODE@ #!! Rule for adding Sem/Date as a tag to readings which looks like dates SUBSTITUTE:Date1 (Num Sg) (Num Sg Sem/Date) TARGET ("<[0-3][0-9][.][0-2][0-9][.][1-2][0-9][0-9][0-9]>"r Num) ; SUBSTITUTE:Date1 (Num Sg) (Num Sg Sem/Date) TARGET ("<[0-9][.][0-9][.][1-2][0-9][0-9][0-9]>"r Num) ; SUBSTITUTE:Date1 (Num Sg) (Num Sg Sem/Date) TARGET ("<[0-3][0-9][.][0-2][0-9][.][0-9][0-9]>"r Num) ; SUBSTITUTE:Date1 (Num Sg) (Num Sg Sem/Year) TARGET ("<[1-2][0-9][0-9][0-9]>"r Num) ; #!! !!Guessing: Rule for adding Adv Sem/Adr as a tag to readings which looks addresses SUBSTITUTE (?) (Adv Sem/Adr) TARGET (""r ?) OR ("http.*"r ?) OR ("<[A-Za-z][a-z].*[.][a-z][a-z].*"r ?) OR ("<[A-Za-z].*[@][a-z].*"r ?) OR ("<[A-Za-z].*[(][a][t][)][a-z].*"r ?) ; COPY:NNom (N Nom) EXCEPT Adv TARGET (Sem/Adr) ; ## www.samediggi.fi http://yle.fi/sapmi lene62@gmail.com anja.irina.ahola(at)gmail.com SECTION #!! !!@CODE@ #!! !!Cycle 0 #!! Removing non-lexicalised forms when lexicalised #REMOVE:PassS (Der/PassS) ; # ikke bra REMOVE:Vregex V-DER (0 V-DER-SUF) ; REMOVE:Nregex N-DER (0 N-DER-SUF) ; REMOVE:Aregex A-DER (0 A-DER-SUF) ; REMOVE:waitwithDis (2 V) (0 (1 V)) ; SELECT:ArabNom (Arab Nom) (-1 BOS)(NEGATE 1 Po) ; SELECT:Arab1 (Arab Nom)(0 ("1")) ; #$ a) Álgogursa 1 sámegielas SELECT:ArabNomPlc (Arab Nom) (-1 Sem/Plc)(NEGATE 1 Po) ; SELECT:Arab Arab ; # In Apertium we get analysis both with and without the tag. #!! Remove Num, ACR, ... REMOVE:R_Num Num (0 Neg); REMOVE:R_ACR ACR (0 CC); ## Im leah naan saemieskuvlesne vaadtseme. ## Jih daejtie gærjide edtjem aaj sjeakodh, – jijtje jeahta jih aalka sjeakodh. #foreløpig regel: IFF:S_Interj Interj (-1 BOS)(1 EOS OR CLB) ; #REMOVE:testA A-DER (0 LEX-A); #REMOVE:testV1 Der1 IF (0 Der3 OR Der2); #REMOVE:testV2 Der2 IF (0 Der3); #REMOVE:testV3 Der1 ; #REMOVE:testV4 V-DER (0 LEX-V) ; ## should be restriction #REMOVE:testN N-DER (0 LEX-N LINK NOT 0 PX) ; #!! Possessive suffix # Probably exists only for Refl and for kinship terms REMOVE:APx A + PX ; #Foreløpig revel, vi gjør unntak etterhvert REMOVE:PrfPrc_Not_PxSg1 PxSg1 IF (0 PrfPrc); ## Mijjieh saemieh daarah aaj aalkeme gærjah lohkedh. REMOVE:PxWhenAcc PxSg1 IF (0 Acc LINK *1 TV BARRIER WORD - ADVLCASE)(NEGATE *-1 Sg1) ; REMOVE:notFAMILY PX (0 N - FAMILY) ; #!! Short Pronouns #!! Proper nouns REMOVE:R_Nom_ABBR_Prop Nom IF (0 (ABBR Attr) OR (Prop Attr))(1 Prop OR ABBR); REMOVE:R_Attr_ABBR Attr IF (0 ABBR + Nom)(NOT 1 Prop OR ABBR); ## Trond Trosterud #REMOVE:CAPABBR ("<[a-zäöåøæï]>"r) IF (0 SOMECAPABBR); #NOTUSED does not work #REMOVE: ("j") IF (0 ("J")); #REMOVE: ("<[a-zäöåøæï]>"r) IF (0 ("<[A-ZÄÖÅØÆÏ]>"r)); ## J. A. #!! Trivialia REMOVE:prop_daate Prop IF (0 Pers); # too strong ## Altese voelph jeanoen nuelesne sealadamme. ## Dan eejhtegh nomme lea Laejsa gonnoeh Jåvvah Andersson. # agreement: # Dem Sg Nom + N Sg Nom Dem Pl Nom + N Pl Nom # Dem Sg Acc + N Sg Acc Dem Pl Acc + N Pl Acc # Dem Sg Gen + N Sg Gen Dem Pl Gen + N Pl Gen # Dem Sg Ill Attr + N Sg Ill Dem Pl Ill + N Pl Ill # Dem Sg Ine Attr + N Sg Ine Dem Pl Ine + N Pl Ine # Dem Sg Ela Attr + N Sg Ela Dem Pl Ela + N Pl Ela # Dem Sg Com + N Sg Com Dem Pl Com Attr + N Pl Com SELECT:DemIll (Dem Ill Attr)(*1 (N Sg Ill) BARRIER NPNHAI) (NEGATE *1 Po BARRIER NPNHAI) ; SELECT:DemCom (Dem Com) (*1 (N Pl Com) BARRIER NPNHAI) ; ## Månnoeh aaj dijjem gijhtien guhth tjaalegh jih guvvieh daan learoegærjese vadteme. ## Daaj baakoejgujmie edtjede jisrajelehkidie bueriesjugiehtidh. ## Akte mubpie lea slaameme jallh slaameminie, båeries povlijtigkerh jis vienhtieh edtja maehtedh gaajhkem dam orrestehtedh, jih dellie traakestieh, sinsitniem dej seamma baakoejgujmie diblieh mejtie leah daah minngemes tjijhtje-gaektsie- jallh uktsieluhkie jaepieh klajkehtamme barre ov goh dihte dle ihkuven saatna. SELECT:DemAttr (Dem Attr)(0 ("dagkeres") OR ("magkeres") LINK 1 N OR (A Attr)) ; SELECT:AttrNotNom (Pron Interr Attr) OR (A Attr) (0 (Pron Interr Sg Nom) OR (A Sg Nom)) ((*1C N BARRIER NPNH OR CC OR Pron LINK NOT -1 COMMA) OR (1 CC LINK 1 (A Attr) LINK 1 N) OR (1 ("laakan" Adv))) ; ##...dah almetji reaktah galka gaajhkh almetji bijre årrodh seamma magkeres naelie, naehkieklaerie,... SELECT:unifDem $$CASES (0C Dem OR Indef LINK *1 $$CASES BARRIER NPNH LINK 0 NOMEN-NOT-PRON) ; ## Klaasen tjirrh olkese goevli doj veelkes baeniej gåajkoe, ... REMOVE:NotPersWhenDem Pers IF (0 Dem + $$NUMBER + $$CASES LINK *1 N + $$NUMBER + $$CASES BARRIER NPNH); REMOVE:NotPersWhenDemSg Pers IF (0 Dem + Sg + $$CASES LINK *1 N + Sg + $$CASES BARRIER NPNH); REMOVE:NotPersWhenDemPl Pers IF (0 Dem + Pl + $$CASES LINK *1 N + Pl + $$CASES BARRIER NPNH); ## Dah minngemes jaepieh jïjtje jåhta. ## Dihte saemien nyjsenæjja, gie dihte dan baantese soptsesteminie, jijtse jih gaalleben bijre soptseste, sotnoeh såemies aejkien, gijredaelvien, edtjægan naan bijjies jaavratjisnie minnedh gøølestidh. REMOVE:NotDem Dem IF (0 Pers)(*1 NOMEN-NOT-PRON BARRIER NPNH)(NEGATE 0 $$CASES LINK *1 NOMEN-NOT-PRON LINK 0 $$CASES) (NEGATE *1 Po BARRIER NPNHAI); ## Dihte fievsesne jih mov govse. REMOVE:NotDem2 Dem IF (0 Pers)(NEGATE *1 NOMEN-NOT-PRON BARRIER NPNH) (NEGATE *1 COM-ADJ BARRIER S-BOUNDARY) (NEGATE 0 (Pron Dem Sg Gen)); ## Dihte joekoen lustes jalhts ajve akten jaepien båeries. SELECT:NGen (N Gen)(-1C (Dem Gen)) ; ## Daam jijjem hov lea fealadamme, daan aereden varki bøøti, tjoevere åadtjodh åeriestidh. #SELECT:PronPers (Pron Pers Gen) (0 Pron LINK 1C Po) ; #dan åvteste is mostly Dem, repalced by the two following rules: ## Johan jih aahkam joekoen gaskese dan åvteste satnem veelji. REMOVE:PcleNotPron ("dån" Pcle) (0 (Pron Gen) LINK 1 Po) ; REMOVE:AttrNotGen Attr (0 (Pron Gen) LINK 1 Po) ; REMOVE:jallh (A Pl) (0 ("jallh" CC))(-1 N OR (Cmpnd))(*1 N BARRIER NPNHA) ; ## Mohte ohtsedibie amma treavkah jallh tjielhketjem jih minnibie doh deavabaahkoeh gierestallibie. REMOVE:jallh (A Pl) (0 ("jallh" CC) LINK 1 Neg LINK 1 S-BOUNDARY OR ConNeg LINK NOT 0 Pl3) ; ## Jeenjh gujht saemieh gieh man akth åvteste eah saemesth jallh eah doesth saemiestidh jallts saemien hïjven guarkah. REMOVE:jallh (A Pl) (0 ("jallh" CC))(-1 FORMS)(1 FORMS) ; ## Dagkeri tjoeverieh joekoen veaksehke juelkieh, mohte disse lissine tjoevere abpe almetje joekoen voerkes jallh fahkoes årrodh. SELECT:jallh (CC) (0 ("jallh"))(-1 BOS OR COMMA)(NEGATE 1 CC OR Pl3 OR Ine OR COMMA) ; ## Helena, jallh Aili man dan aaj nomme, lea saemien maadtoste. REMOVE:tjïrrh (N Pl) (0 ("tjïrrh" Po)) ; REMOVE:tjïrrh (V) (0 ("tjïrrh" Po) LINK *-1 Gen BARRIER NPNH LINK NOT 0 TIME) ; ## Bïegke vaarjoej tjïrrh bïegkedi. REMOVE:gujmie (N Sg Nom) (0 ("gujmie" Po) LINK -1 GUJMIE) ; #!! Verbs IFF:IsItConNeg ConNeg (*-1 Neg BARRIER CC OR COMMA LINK 1 ConNeg) (NEGATE 0 N - ("åerie") LINK *1 ConNeg BARRIER V) (NEGATE 0 Inf LINK *-1 AUX BARRIER V); ## Göökte ålmah ligan beetnegh lööneme ålmeste gie beetnegh löönehtalli. ## NEGATE: Laavloejigan jih tuhtjigan luste, jih gosse idtjigan åerieh jallh byøpmedh, die heehkigan jih beenghkigan dam baeniem guktie dihte gåetie edtja stoerre jih tjeekeds SJIDTEDH. REMOVE:PrtInNeg Prs IF (0 Neg)(*1C ConNeg + Prt BARRIER NOT-ADV); ## Ij lin gujht daarah altese vielie jienebh biejjieh daesnie. REMOVE:PrsInNeg Prt IF (0 Neg)(*1C ConNeg BARRIER NOT-ADV LINK NOT 0 Prt); LIST PERS-TAG = Sg1 Sg2 Sg3 Du1 Du2 Du3 Pl1 Pl2 Pl3 ; SELECT:PersonInNeg $$PERS-TAG (0 Neg + Prt)(*1C $$PERS-TAG + ConNeg BARRIER NOT-ADV); REMOVE:NegPrt (Neg Prt)(NEGATE 1 ("lea" ConNeg Prt)) ; ## this is for removing the 'ij' etc. ## Ij lin gujht daarah altese vielie jienebh biejjieh daesnie. SELECT:Du1NotVGen Du1 (0 VGen LINK -*1 ("månnoeh") BARRIER S-BOUNDARY) ; #!! Imperative #REMOVE:NotImpIfWord Imp IF (*-1 NOT-ADV-INT-PCLE BARRIER CS OR CC); #REMOVE:NotIndIfImpAndExcl Ind IF (NOT *-1 Neg)(0 Imp)(*1 ("excl") BARRIER Ind); ## # There can be Interj, VOC, SELECT:Imprta Imprt (-1 BOS OR (":") OR Interj)(0 (V Sg2))(NOT *1 (Pers Sg2 Nom) BARRIER S-BOUNDARY) ; ## Utnieh hijven. SELECT:Imprtb Imprt (-1 BOS OR (":") OR Interj)(1 COMMA) ; ## Jiehtieh, datne gien ij naan bovtsh, ih leah datne vielie saemie. REMOVE:Imprtc Imprt (-1 Interr OR Adv OR (Pers Sg2))(0 (Prs Sg2)) ; ## Maam leah datne darjoeminie? REMOVE:Imprtd Imprt (*-1 WORD BARRIER CC OR COMMA OR BOS)(NOT 0 Neg)(NEGATE -1 CC LINK *-1 Imprt BARRIER S-BOUNDARY OR V) ; ## Aejkie sjïdteme åarjese Voernesen vuestie jarredh. SELECT:Imprte (Imprt Sg2) (-1 CC LINK *-1C (Imprt Sg2) BARRIER S-BOUNDARY OR V) ; # Select the finite verb reading if this is the only finite verb reading within the clause SELECT:sjur VFIN IF ( NOT *0C VFIN BARRIER S-BOUNDARY OR CLB ) ; ## Saemieh leah akte åålmege jalhts joekehts laantine orre, gïele dïhte åålmegidie raasti dåaresth ektede. #!! !!CC- and CS-Mapping # ------------------- MAP (@COMP-CS<) TARGET Adv OR A IF (-1 ("goh" CS) LINK -1 A OR Adv)(1 COMMA LINK 1 ("die")) ; ## Daelie goh beetnehbïjle buerebe goh åvtelen, die bïjlen jienebh gaeptieh. MAP (@COMP-CS<) TARGET $$ADVLCASE IF (-1 ("goh" CS) LINK -1 $$ADVLCASE)(1 COMMA LINK 1 VFIN) ; # ## Juohkehaš guhte earrána áhkástis mange eará sivas go fuorrávuođas, dagaha su rihkkut náittosdili. # ## Kaarre, gie båata gujneste jeatjah bieleste goh vissjeste, edtja satnem vuartasjidh. # #ADD:r423 (@COMP-CS<) TARGET Nom IF (*-2 ("nu") BARRIER NOT-NP LINK 1 A OR Adv OR PrsPrc OR CC OR CS)(-1 ("go" CS)); # ## Nu guhkás go vejolaš galgá dat čuovvoluvvot. # ## Sin mearridanorgánain berre mearridanváldi leat nu viiddis go vejolaš ja nu čadni go vejolaš. # #MAP:r424 (@COMP-CS<) TARGET NP-MEMBER IF (*-1 ("go" CS) BARRIER NOT-NP LINK NOT -1 VFIN OR CC)(0 Nom OR ADVLCASE)(NEGATE 0 Attr OR TITLE LINK 1 N OR A)(NEGATE *1 VFIN BARRIER SV-BOUNDARY); # ## Ovdabarggut állanit seamma beallái go "láhkateavstta lunddolaš lohkan". # ## Ovdabarggut állanit seamma beallái go "láhkateavstta lohkan". # ## Sápmelaččat guđet orrot Norgga bealde Deanu, geavahit seammá gávtti go sápmelaččat Suoma bealde Deanu. # # #MAP:r425 (@COMP-CS<) TARGET Nom IF (*-1 ("go" CS) BARRIER NOT-NP LINK NOT -1 VFIN)(1 CC)(NEGATE *1 VFIN-NOT-IMP); # #MAP:r426 (@COMP-CS<) TARGET NP IF (*-1 ("go") BARRIER NOT-NP OR Nom LINK NOT -1 VFIN OR BOS)(1 CLB LINK NOT *1 VFIN BARRIER CC); # ## Muhto buođđogeažit ja giđđabuođut ráhkaduvvojit muđui seammá vugiid mielde go geassebuođut, ja danne heive buođđun birra obbalohkái juo muitaluvvot. # #MAP:r427 (@COMP-CS<) TARGET Loc IF (*-1 ("go" CS) LINK 1 Indef BARRIER NOT-NP OR Nom LINK NOT -1 VFIN OR BOS LINK *-1 Loc BARRIER SV-BOUNDARY)(1 CLB LINK NOT *1 VFIN BARRIER CC); # ## #MAP:r428 (@COMP-CS<) TARGET Com IF (*-1 ("go" CS) LINK 1 Indef BARRIER NOT-NP OR Nom LINK NOT -1 VFIN OR BOS LINK *-1 Com BARRIER SV-BOUNDARY)(1 CLB LINK NOT *1 VFIN BARRIER CC); #MAP:r429 (@COMP-CS<) TARGET Ill IF (*-1 ("go" CS) LINK 1 Indef BARRIER NOT-NP OR Nom LINK NOT -1 VFIN OR BOS LINK *-1 Ill BARRIER SV-BOUNDARY)(1 CLB LINK NOT *1 VFIN BARRIER CC); #MAP:r430 (@COMP-CS<) TARGET Ess IF (*-1 ("go" CS) LINK 1 Indef BARRIER NOT-NP OR Nom LINK NOT -1 VFIN OR BOS LINK *-1 Ess BARRIER SV-BOUNDARY)(1 CLB LINK NOT *1 VFIN BARRIER CC); # #MAP:r431 (@COMP-CS<) TARGET NP IF (*-1 ("go" CS) BARRIER NPNH LINK *-1 Nom BARRIER S-BOUNDARY)(*-1 ("go" CS) BARRIER NPNH LINK *-1 Comp BARRIER S-BOUNDARY)(NEGATE *-1 go BARRIER NPNH LINK -1 VFIN LINK *-1 VFIN BARRIER S-BOUNDARY)(0 Nom OR ADVLCASE)(1 EOS OR CLB) (NEGATE 1 CLB LINK *1 VFIN BARRIER CC) (NEGATE 1 CLB LINK *1 NP-HEAD-NOM BARRIER NPNH) (NEGATE 2 Interr LINK *1 VFIN LINK *1 VFIN BARRIER S-BOUNDARY); # ## ...muhto galjebut dat lea go guv-čalmmit. # #MAP:r432 (@COMP-CS<) TARGET NP IF (-1 ("dego") OR ("dugo"))(0 Nom OR ADVLCASE BARRIER Nom)(NEGATE *1 VFIN-NOT-IMP OR AUX BARRIER S-BOUNDARY); # ## Sii leat dego eŋgelat almmis. # #MAP:r433 (@COMP-CS<) TARGET NP IF (*-1 ("dego") OR ("dugo") BARRIER NOT-NP OR Nom)(0 Nom OR ADVLCASE)(1 EOS OR CLB); # ## Son oahppaladdá min allagasas dego beaivváža badjáneapmi. # #MAP:r434 (@COMP-CS<) TARGET NP IF (*-1 ("go" CS) BARRIER NPNH LINK -1 A)(0 Nom OR ADVLCASE BARRIER Nom)(NEGATE 0 TITLE LINK 1 Prop LINK 0 Sem/Fem OR Sem/Mal OR Sem/make Sur)(NEGATE 0 Attr LINK 1 N OR A)(NEGATE *1 VFIN BARRIER NOT-ADV-PCLE); # ## Máhttájeaddji ii leat stuorit go oahpaheaddjis. # #MAP:r435 (@COMP-CS<) TARGET Pron IF (*-1 ("go" CS) BARRIER NPNH LINK -1 A)(0 Nom OR ADVLCASE BARRIER Nom)(NEGATE 0 Attr LINK 1 N OR A)(NEGATE *1 VFIN BARRIER NOT-ADV-PCLE); # ## Dalle dat viežžá čieža eará vuoiŋŋa vel, bahábuid go son ieš. # #MAP:r436 (@COMP-CS<) TARGET Nom IF (*-2 ("dego") OR ("dugo") LINK *1 NP BARRIER NPNH LINK 1 Rel)(NEGATE 0 Attr LINK 1 N OR A)(*1 VFIN BARRIER S-BOUNDARY); # ## Dat lea dego olmmái gii manai olgoriikii. # #MAP:r437 (@COMP-CS<) TARGET NP IF (*-1 ("dego") OR ("dugo") BARRIER NOT-NP OR Nom) (NEGATE 0 Attr LINK *1 Nom BARRIER NPNH)(NEGATE *1 VFIN-NOT-IMP BARRIER S-BOUNDARY) (NEGATE 1 ("go") OR ("dego") OR ("dugo")); # ## Dego njamaheaddji eadni dikšumin mánáidis... # #MAP (@COMP-CS<) TARGET N (*-1 ("goh") BARRIER S-BOUNDARY ); #MAP (@COMP-CS<) TARGET NP-MEMBER IF (*-1 ("goh" CS) BARRIER NOT-NP LINK NOT -1 VFIN OR CC)(0 Nom OR ADVLCASE)(NEGATE 0 Attr OR TITLE LINK 1 N OR A)(NOT *-1 (@COMP-CS<) BARRIER S-BOUNDARY); ## #!! !!CNP mapping #!! Mapping CNP to CC and CS. MAP (@CNP) TARGET CC (0 ("gon")) ; MAP (@CNP) TARGET CS (0 ("goh") LINK *1 N OR Pers BARRIER NPNH LINK *1 S-BOUNDARY BARRIER V) ; MAP (@CNP) TARGET CS (0 ("goh") LINK *1 TITLE BARRIER NPNH LINK 1 N LINK *1 S-BOUNDARY BARRIER V) ; MAP (@CNP) TARGET CS (0 ("goh"))(NEGATE -1 CC OR CS OR (":"))(NEGATE *-1 VFIN OR AUX BARRIER S-BOUNDARY)(*-1 V OR N OR Adv OR A OR Pron OR Num); ## Ussjedh barre - gosse jijnjesh daagkoe sjidtebe, gosse gåetieh gaajhkine baenine sjidtieh - die Karijuse, die manne goh jeatjah gånka daennie mov gåetesne maahtam tjahkasjidh jih abpe staarem vuartasjidh MAP (@>CC) TARGET CC (0 ("dovne")) ; MAP (@CNP) TARGET CC (-1 $$CASES LINK 2 $$CASES) ; MAP (@CNP) TARGET CC (-1C Inf LINK 2C Inf) ; MAP (@CNP) TARGET CC (*-1C (Pl Gen) BARRIER WORD)(1 (Pl Com)) ; #!! !!CVP Mapping #!! Mapping @CVP to all CS MAP (@CVP) TARGET CS ; #$ Ussjedh barre - gosse jijnjesh daagkoe sjidtebe, gosse gåetieh gaajhkine baenine sjidtieh - die Karijuse, die manne goh jeatjah gånka daennie mov gåetesne maahtam tjahkasjidh jih abpe staarem vuartasjidh MAP (@CVP) TARGET CC (0 ("mohte") OR ("men")) ; MAP (@CVP) TARGET CC (*-1 MAINV BARRIER S-BOUNDARY)(*1 MAINV BARRIER S-BOUNDARY); # MAP (@+FAUXV) TARGET VFIN (0 Neg OR MOD-ASP) (*1 V BARRIER S-BOUNDARY); MAP (@+FAUXV) TARGET VFIN (0 COPULAS) ((*1 PrfPrc OR Ger BARRIER S-BOUNDARY OR CC) OR (-1 PrfPrc)); MAP (@+FMAINV) TARGET VFIN (NOT 0 COPULAS) (1 VGen); MAP (@+FAUXV) TARGET VFIN (0 AUX LINK NOT 0 COPULAS) (*1 V BARRIER S-BOUNDARY OR CC); MAP (@+FMAINV) TARGET VFIN (0 COPULAS)(NOT *1 PrfPrc OR Ger BARRIER S-BOUNDARY); MAP (@+FAUXV) TARGET VFIN (0 MOD-ASP OR Neg); MAP (@-FAUXV) TARGET AUX (0 Inf OR ConNeg)(*1 Inf BARRIER S-BOUNDARY OR CC); MAP (@+FMAINV) TARGET VFIN IF (NOT 0 AUX) ; MAP (@+FMAINV) TARGET VFIN IF (0 AUX LINK NOT *1 Inf OR PrfPrc BARRIER S-BOUNDARY) ; MAP (@-FMAINV) TARGET Ger OR ConNeg OR PrfPrc OR Inf IF (*1 VFIN BARRIER S-BOUNDARY LINK 0 AUX) ; MAP (@-FMAINV) TARGET Ger OR ConNeg OR PrfPrc OR Inf IF (*-1 VFIN BARRIER S-BOUNDARY LINK 0 AUX) ; #$ Luste tjoejkedh. # #!! !!PrfPrc #!! Select PrfPrc if DerNomAct #SELECT: PrfPrc (*-1 ("lea") BARRIER V OR S-BOUNDARY) ; SELECT:PrfPrs_DerNomAct PrfPrc (0 (Der/NomAct) LINK *-1 S-BOUNDARY BARRIER V)(NEGATE *1 V BARRIER S-BOUNDARY) ; ## Disse maahtam tjaeledh gosse sijhtem, jih aaj lohkedh gosse båarasåbpoe sjidteme. ## Manne aktem gärjam man nomme jupmele-vaerie lohkeme. #!! !!Person #!! leah Prs Sg2 = Pl3 SELECT:VSg2IfPersSg2a (V Sg2)(NOT 0 N) ((*-1 (Pers Nom Sg2) OR (Rel Sg Nom) BARRIER V OR Nom) OR (1 (Pers Nom Sg2))); SELECT:Pl3WhenPlSubj (V Pl3) IF ((*-1 (N Pl Nom) OR (Num Sg Nom) OR (Pers Nom Pl3) BARRIER Sg2 OR S-BOUNDARY) OR (-1 (N Sg Nom) LINK -1 CC))(0 ("lea" V)); #SELECT:Pl3WhenSubjCC (V Pl3) IF (-1 (N Sg Nom) LINK -1 CC)(0 ("lea" V)); REMOVE:Pl3 (V Pl3) (NOT *0 Nom BARRIER S-BOUNDARY) ; SELECT:Sg3WhenRelSg (V Sg3) ((-1 (Rel Sg Nom)) OR (-1 CC LINK *-1C (V Sg3) BARRIER S-BOUNDARY OR V)); #SELECT:Sg3WhenSg3toLeft (V Sg3) (-1 CC LINK *-1C (V Sg3) BARRIER S-BOUNDARY OR V); #!! Select Inf If Infv SELECT:InfIfInfv (V Inf) IF (*-1 INFV OR AUX BARRIER V) ; # Manne båatam sagke buerebe datneste jieledh, Karijuse. SELECT:AUXIfInf AUX IF (*1 Inf BARRIER V) ; ##Klaasen tjirrh olkese goevli doj veelkes baeniej gåajkoe, jih dellie maam akt bøøti ussjedidh. SELECT:PrfPrcIfLea PrfPrc (*-1 ("lea") OR (",") OR CC BARRIER V); ## Gïjre lea sjïdteme daelvien männgan goh lea nåake gåatome jïh jïjnjh juvrh orreme. SELECT:PlIfPlSubj Pl3 IF (*-1 (Pl Nom) BARRIER NOT-ADV OR COMMA); ## Jijnjesh leah sijhteminie edtjibie gieline barkedh, maehtibie tjåahkoeh, seeminarh øørnedh gusnie åadtjoejibie gaavnedidh jih rååresjadtedh. REMOVE:FocGan Foc/gan (0 (V Du3)) ; SELECT:Ger Ger (0 (Der/NomAct) LINK *-1 REALCOPULAS BARRIER S-BOUNDARY) ; ## Maam leah datne darjoeminie? # Contract or not -- removed, we have removed the sma contract verb analysis # REMOVE:NotContrIfEven (".*ijidh"r) IF (0 (".*[oe]dh"r)); # This rule might be problematic in light of the empirical work. #!! !!Span sentences #"" REMOVE:lea Pl3 IF (<*-1 Sg1 OR Sg2)(NOT *-1 (N Pl Nom)); #!! !Nomen #!! REmove Px if not family REMOVE:No_Px PX IF (0 N)(NOT 0 FAMILY OR Pron); ## Bïepmehtehteme jåarhka goh gïjrese vuertieminie. #!! Remove Prop Attr if not 1 Prop #SELECT: Attr IF (0 Prop)(1 Prop); # Too hard. REMOVE:PropAttr (Prop Attr) IF (NOT 1 Prop); ## Åvla gihtjie, vaadtsa lihkebe jih geelkie dam naehpiem. #SELECT:GenToAdp Gen IF (1 Adp); #NOTUSED # check this #REMOVE:NotNomToAdp Nom IF (1 Adp); #NOTUSED # check this SELECT:GenIfLeaNom Gen (1 ("lea" V) LINK 0 Pl3 OR Sg3 LINK *1 (N Nom) BARRIER NPNH) ; # Dan lin baenieh njaelmesne. SELECT:AccIfSVOV (Acc Sg) IF (*-1C INFV BARRIER NPNH)(1C V); # consider VFIN for INFV ## Jis sååjhtoe ålkoelaantesne gogka akth dienesjidh dellie barkosne tjuara aelkedh dam laanten gïelem soptsestidh. SELECT:IneIfEss/Com (N Pl Ine) (0 (N Ess) OR (N Sg Com)) (1 ("viedtedh" V* TV Der1 Der/ldahke N Sg Ine) OR ("vïedteldahke" N Sg Ine)) ; # Njolkedassh ålmine veadtaldahkesne REMOVE:EssIfCom/Ine (N Ess) (0 (N Sg Com) OR (N Pl Ine)) (*-1 (Pron Sg Com) OR (Pron Pl Ine) BARRIER NPNH OR S-BOUNDARY) ; # dajnie barkoeplaanine SELECT:NPlIll (N Pl Ill) (0 (N Pl Acc)) (NOT *0 (V TV) BARRIER S-BOUNDARY) ; # ..., mejtie juridigkesne veadtjaldahkesne staatide, jih deklarasjovnide. #!! !Verb or Noun SELECT:PlNomObj (N Pl Nom) IF (0 (V Pl3)) (NOT *0 VFIN BARRIER NPNH OR S-BOUNDARY) ; #...Henny Bergsland, mij vuelieh tjeeli mejtie Jonetta joejki REMOVE:NotNomIfInf (N Pl Nom) IF (0 Inf)(*-1 Ind or Imp LINK *-1 Nom); ## jieledh. #gïelen: Gen vs. VGen gïelem: Sg1 vs. Acc gïelh: Pl vs. Sg2 REMOVE:gïele VFIN (0 ("gïele") LINK 0 Acc OR Nom)((*1 (V TV) OR Inf BARRIER S-BOUNDARY OR MAINV) OR (*-1 (V TV) BARRIER NPNH OR S-BOUNDARY)) ; ## Gaajhkesh mih maaje daejribie man vihkele lea daam mijjen gielem vaarjelidh. REMOVE:gïele Sg2 (0 ("gïele") LINK -1 Num OR Indef) ; ## Nøørjesne aaj jeatjah gïelh juktie leah jeenjh sïjsejuhtijh gieh jïjtsh ietniengïelem soptsestieh. REMOVE:gïele VGen (0 ("gïele") LINK NEGATE -1 MAINV) ; ## Jih dihte mij, destie jienehke guessie gielesne, guktie vihth dan tseahkan båetibie, tjoeveribie barre vaajtelidh, mijjen noerh åadtjoeh jih sijhtieh guessie gielh lieredh guktie buektiehtieh ohtsedidh dah gielen roehtsh jih åadtjodh daejredh guktie dam aaj almetjh aerebi soptsestin. # Dem #!! !!CC and CS or Adv IFF:GoNotCs ("goh" Pcle) (-1 Neg OR ConNeg OR ("sån")); ## Ih goh måjhtijh ahte månnoeh daesnie sagke seadtoesåbpoe årroen. # Pcle ## Dallah goh Jeesuse tjaetseste tjuedtjele, dellie vuajna Elmie rihpesåvva jih Voejkene altasasse goh ledtie suaja. # CS REMOVE:CSNotVfin CS IF (NOT *-1 VFIN OR ("jis" CS))(0 ("dellie") OR ("dle")); ## REMOVE:NotNomIfCs (N Nom) (0 CS)(1 (N Nom)) ; # Guktie vielle Baktuse, giehtelh maa! SELECT:CCdovne CC IF (0 ("dovne"))(*1 ("jïh" CC) BARRIER VFIN); ## … almetjistie gieh maehtieh saemien dovne tjaaleldh jïh njaalmeldh barkoesijjide tjïeltesne. #REMOVE:CsIfNotFuncword Pcle (NEGATE *-1 NOT-ADV-INT-PCLE)(0 CS) ; #NOTUSED # Jaa, jis damth seekere ..... # GoNotCs (above) takes this sentence instead. REMOVE:NotCsIfCc CS ((-1 CC OR ("mij" Nom)) OR (NEGATE -1 WORD))(0 Adv) ; # Jih dellie jeehti... #REMOVE:NotCsIfBOS CS (NEGATE -1 WORD)(0 Adv) ; SELECT:JisCS ("jis" CS) (-1 BOS) ; ## Jis manne båatam satne mænna gåatan. IFF:JisPcle ("jis" Pcle) (-1 N OR Pron OR TIME) (NEGATE -1 ("buerie") OR ("luste") OR ("seamma") OR ("juktie") OR ("nuelie") OR ("gaevhtie") OR ("dovne")) ; ## Daah gärhkoeh hov seamma laaketje sijjine jis vuartasje guktie byjreske dej bïjre lea. Daate akte vihkeles dirrege jis edtja saemien kultuvrem, gielem jih ektievuekiejielemem nænnoestidh jih vååjnesisnie darjodh. Tjaaleme nännoestimmie nimhtems ierine saemiestimmie galka saemien gielesne årrodh ahte nännoestimmiem maehtien saemien gielese njaelmeste jarkoestidh reerenasseste jis dihte aajne daam säjhta. Tjilkeme dovne raereste akth vielie luhpiedimmiem buerie ektievuekie ståvroen jih pleenumen gaskoeh seamma goh tjelten laakesne nuepiem bååstide vaeltedh lihtsegijstie jis majoriteete jeatjhligkie pleenumisnie sjädta.(does NOT work here) # select A Attr if the word is directly followed by another noun. Probably needs a barrier to avoid merging two N's that just happen to be colocated, and where the first one is ambiguous with an A Attr reading. SELECT:AdjAttrBeforeNoun (A Attr) IF ( 1C N ) ; # Select the Inessive reading of an unambiguous noun in front of a location verb: #!! !!Adj or Adv REMOVE:AdvNotA A (0 Adv)((1C V OR (A Sg Nom)) OR (1 Ela LINK *1 V BARRIER NOT-ADV)); ## Minngemes aejkien goh dam heannedi lij 1988. ## Manne båatam sagke buerebe datneste jieledh, Karijuse. REMOVE:AdvNotA_Cop A (0 Adv)(-1C V LINK NOT 0 COPULAS)(NEGATE 1 A OR N) ; ## Maanah utnieh guhkiem nuekies vuarteme jih daelie mahte vissjehtovveminie. SELECT:AAttr (A Attr) ((*1C N BARRIER NPNH OR CC OR Pron LINK NOT -1 COMMA) OR (1 CC LINK 1 (A Attr) LINK 1 N) OR (1 ("laakan" Adv)))(NOT 0 N) ; ## Gïjre lea sjïdteme daelvien männgan goh lea nåake gåatome jïh jïjnjh juvrh orreme. REMOVE:NotAAttr (A Attr) (NOT 1 N OR A LINK NEGATE 0 CC OR COMMA LINK 1 Attr) ; ## Laedtieh gietskiebasse saemielaantese båetieh bene Aslak tuhtjie dajve gujht stoerre gåabpatjahkide. ## Dam maam dorje lea gaerhtelesvoeten jïh riektesvoeten mietie, jïh altese stillemh leah stynkehke jïh ihkuvasse tjåadtjoeh. ## Datne edtjh dov elkiejgujmie, dov måarine jïh dov mænnjajgujmie vïnhtsese tjaangedh, jïh fïerhten jielijijstie, urries jïh minngels, edtjh akth paarrh vïnhtsese vaedtsiehtidh, båetieh datnine ektine jieledh. #!! !!Grammatisk ord eller N eller A # "" N Sg Nom Pron Pers Pl3 Gen Pron Pers Sg Gen PxSg3 Pron Pers Sg3 Gen # Remove the "skotunge" reading if not in typical N position REMOVE:altese_Noun Nom IF (0 ("altese"))(0 Gen LINK *1 N BARRIER NPNH); ## Reerenasse tseegkie altese politihkem ... # "" Adv N Sg Nom REMOVE:daelie N (0 ("daelie" Nom) LINK 1 Nom) ; ## Daelie Tåamma riejries dåakteren luvnie. SELECT:CCCS CC OR CS (*-1 VFIN BARRIER S-BOUNDARY)(*1 VFIN BARRIER S-BOUNDARY); ## Tåamma gylje juktie bååktjehke dorje. REMOVE:IllNotVfin VFIN (0 ILLADV)(1 VFIN) ; REMOVE:IllNotNom Nom (0 ILLADV)(1 VFIN) ; ## Dah aaj naan aejkien diekie båetieh, mohte eeremes tjidtjie dihte dahkoe vualka. ## Dahkoe lin aaj sijhteme edtjin naan noere saemieh aaj båetedh. #!! !!N or V # "" N Sg Gen V IV Ind Prs Du1 @+FMAINV REMOVE:aejkie1 VFIN (0 ("aejkie" Gen))((-1 ("naan") OR (Indef Gen) OR ("aerebi") OR (Num Gen) OR Ord) OR (1 VFIN)); ## Guktie don øvteben aejkien soptsestim, gielemoenehtasse lea ussjedeminie daan jaepien barkedh guktie dihte gielebarkoe edtja juhtedh. #REMOVE:aejkie2 VFIN (0 ("aejkie" Gen) LINK 1 VFIN); ## Naan aejkien leah kruana jih jeatjebh aejkien provne. SELECT:Vfin VFIN (0 (N Gen) LINK 1 S-BOUNDARY)(NEGATE -1 Pr)(NEGATE *-1 VFIN BARRIER S-BOUNDARY)(NEGATE -1 Attr); ## Tjidtjie diekie juhtieji goh aehtjine pruvri. SELECT:N Gen (0 (Du1) LINK -1 Attr) ; SELECT:golme (Num Gen)(0 ("golme"))(1 (N Gen)) ; ## Jih die libriem dijpebe, gulhkine tjaeliestibie golmen aejkien jih doeltehtibie. #!! Ger or Der/NomAct SELECT:GerNotNomAct Ger (NEGATE *-1 V BARRIER S-BOUNDARY)(0 (Der/NomAct) LINK 1 S-BOUNDARY) ; ## Mov leah aajkohkh, muahrah jih jyønh gieh desnie årroeminie. #!! Adj or Indef SELECT:AdjNotIndef (A Pl) (0 (Indef))(*1 (N Pl) BARRIER S-BOUNDARY) ; ## Gïjre lea sjïdteme daelvien männgan goh lea nåake gåatome jïh jïjnjh juvrh orreme. REMOVE:AdjBeforeV A (NEGATE 0 Nom)(1 MAINV) ; ## Mov aehtjie jijnjem maahta. #!! Num SELECT:NumAndNum Num IF (-1 CC)(*-2C Num BARRIER V); SELECT:NumAndNumIne (Num Ine) (0 ("akte") LINK *1 (N Ine) BARRIER NPNH); SELECT:NumAndOrdNum (Ord Nom)(0 Num LINK 1 N); ## Manne skuvlesne vijhtede klaassesne. # Putative contract verb #!! Rel or Interr SELECT:InterrNotRel Interr (0 Rel LINK NEGATE *-1 Interr)(*1 ("?")) ; REMOVE:RelNotInterr Interr (0 Rel LINK -1 WORD LINK NOT 0 Interj OR CC OR CS) ; REMOVE:RelNotInterrQ Interr (NEGATE *1 ("?")) ; SELECT:RelGen (Rel Gen) (*-1 VFIN BARRIER S-BOUNDARY)(*1 N BARRIER NPNH); ## Jaepien 1877 die pruvri aktine man nomme Jonas Mortensen. # Interj #!! Po or Pr SELECT:PoGena Po (0 Pr)(-1 Gen) ; REMOVE:Pr Pr (NEGATE *1 N BARRIER V OR S-BOUNDARY) ; REMOVE:PoGenb Po (NEGATE *-1 Gen BARRIER V OR S-BOUNDARY) ; #!! Adv or Po/Pr SELECT:AdvNotPo Adv (0 Po)(NEGATE -1 Gen) ; ## Desnie lim dam mietskem jaepien 1941, vaaksjoeminie ektine mov gåmmine, Henny Bergsland, mij vuelieh tjeeli mejtie Jonetta joejki. REMOVE:PoNotAdv Adv (0 Po)(-1 Gen) ; SELECT:PcleNotPron ("dån" Pcle) (0 (Pron Gen) LINK 1 A OR V OR Pcle OR Coll OR Adv) (NEGATE 1 Po OR Ord) ; ## Mujhtedh vielle, gosse månnoeh dan jijnjh gåahtah jih karamellh biejjegi byøpmedien... REMOVE:"ektine" (Pr) (*1 Com BARRIER NPNH) ; #!! Illative or genetive SELECT:IllNotGena Ill (0 Gen)(*1 ILLV BARRIER NOT-ADV); REMOVE:IllNotGenb Gen (0 Ill)(NOT *-1 Pr BARRIER NPNH)(1 EOS OR S-BOUNDARY); ## Tåamma fuehpesne gåatan. SELECT:Ill Ill (*0 ILLV BARRIER S-BOUNDARY OR ILLADV) ; #!! Com SELECT:ektine (N Com) OR (Pron Com) OR (A Com) (*-1 ("ektine") BARRIER NPNH) ; ## Desnie lim dam mietskem jaepien 1941, vaaksjoeminie ektine mov gåmmine, Henny Bergsland, mij vuelieh tjeeli mejtie Jonetta joejki. SELECT:ektine (N Com) OR (Pron Com) OR (A Com) (*1 ("ektine") BARRIER S-BOUNDARY) ; ## Giejnie leah skuvlesne ektine? SELECT:ComIfComV Com (*0 COMV BARRIER S-BOUNDARY) ; SELECT:ComNotIne Com (0 Ine LINK *1 COM-ADJ BARRIER S-BOUNDARY) ; #!! Accusative or illative SELECT:AccNotIll (N Acc) (0 (N Ill) LINK *1 STV BARRIER NOT-ADV) ; ## Mov aaj aehtjie gie båanta. Dihte fievsesne fierhten biejjien, govside båhtja, jih gaajhkide juvride biepmedahta. #!! Indef or Adv SELECT:IndefNotAdv (Pron Indef) (0 Adv) ((*1C A OR N BARRIER NPNH OR CC OR Pron LINK NOT -1 COMMA) OR (1 CC LINK 1 (A Attr) LINK 1 N) OR (1 ("laakan" Adv))) ; ## ..., gaajhkh dovnesh seamma vihkeles jih vihkeles ektesne barkedh abpevoetesne. SELECT:AdvNotIndef Adv IF (0 (Pron Indef)) (1 S-BOUNDARY); ## ..., reerenassen ulmie dajvesne seamma goh daaletje aalkoebarkoe jih proposisjovnh... #!! special lemmas SELECT:daelie Adv IF (0 ("daelie")); # meaning "dal (mest i stedsnavn)" discarded. SELECT:nomme N (0 ("nomme") LINK 1 Prop) ; ## Sara jaepien 1889 Röörosen dajvese bööti jih jaepien 1893 die pruvri aktine man nomme Anders Paulsen SELECT:giesie ("giesie" Gen)(1 Num); REMOVE:almetje1 ("alma") IF (0 ("almetje")); REMOVE:almetje2 ("elmie") IF (0 ("almetje")); REMOVE:gyhtjelasse ("gæhtjodh") IF (0 ("gyhtjelasse")); SELECT:nomme ("nomme" N) IF (0 ("nomme" Num)) ((*-1 (A Attr) OR (Pron Attr) BARRIER NPNH OR CC OR Pron LINK NOT -1 COMMA) OR (1 CC LINK 1 (A Attr) LINK 1 N) OR (1 ("laakan" Adv))); ## Die maa onterligksh nommh, ... SELECT:Jupmele ("Jupmele") IF (0 ("jupmele")); #!! Adverb context prefers Adv SELECT:vihth Adv IF (-1 V)(0 ("vihte"))(1 NPNH OR N); # Adverb context prefers Adv SELECT:meatan Adv IF (0 ("meatan"))(NOT 1 Po OR N OR A); #!! Verb person vs. Inf -- moved here in order to have the pronouns disambiguated first. REMOVE:Inf_Not_Sg2 (V TV Der1 Der/didh Ind Prs Sg2) IF (*-1 VFIN BARRIER Pron + Sg2 + Nom); REMOVE:Inf_Not_Pl2 (V TV Ind Prt Pl2) IF (*-1 VFIN BARRIER Pron + Pl2 + Nom); # These rules are too weak. I flip them around: REMOVE:Sg2_Not_Inf Inf IF (0 (V TV Der1 Der/didh Ind Prs Sg2)) (*-1 Pron + Sg2 + Nom BARRIER VFIN); REMOVE:Pl2_Not_Inf Inf IF (0 (V TV Ind Prt Pl2))(*-1 Pron + Pl2 + Nom BARRIER VFIN); # Even stronger. No pro-drop. ## WARNING! This is probably too strong. REMOVE:Inf_Not_Sg2_ProDrop (V TV Der1 Der/didh Ind Prs Sg2) IF (NOT *-1 Pron + Sg2 + Nom); REMOVE:Inf_Not_Pl2_ProDrop (V TV Ind Prt Pl2) IF (NOT *-1 Pron + Pl2 + Nom); #!! !!!Proper nouns # ============ #!! Rule set taken from sme # Proper nouns # ============ #!! Substituting Prop tags # ------------------------- SUBSTITUTE:Sur (Sem/Plc) (Sem/Sur) TARGET Prop IF (-1 (Sem/Mal Attr) OR (Sem/Fem Attr))(NEGATE 0 Sem/Sur); ## Linda England. SUBSTITUTE:Sur (Sem/Plc) (Sem/Sur) TARGET Prop IF (-2 (Sem/Mal Attr) OR (Sem/Fem Attr))(-1 ABBR); ## Linda B. England. # Prop or not # ----------- SELECT:Prop Prop IF (NEGATE *-1 BOS OR (":") OR (PUNCT LEFT) OR (Num Nom) BARRIER REALWORD); ## Áirras Egil Olli nammaduvvui Sámedikki ovddasteaddjin. ## 1.1. Sámi doaibmaplána darbu ja ulbmilat. # Capitalised nouns in the middle of a sentence are names - (not always). REMOVE:PropSg (Prop Pl) IF (0 (Prop Sg))(NEGATE 0 ("Saemiedigkie")); ## Marit lea dál. # Marit N Sg, not Mari+t Pl Nom SELECT:PropGen (Prop Sg Gen) (0 (Prop Sg Nom) LINK *1 N BARRIER V OR S-BOUNDARY) ; ## Daennie JENSEN njaelmesne nov maa månnoeh daesnie seadtosne åadtjoen årrodh! # Not this one… LIST PROP-OR-NOT2 = "Bergene" "Jeesus" "Kristus"; # But this one! LIST PROP-OR-NOT1 = "Bergen" "Jeesuse" "Kristuse" ; SELECT:Prop PROP-OR-NOT1 IF (0 PROP-OR-NOT2) ; #(NEGATE -1 BOS LINK 2 Prop); REMOVE:Prop (Prop Pl Nom) IF (-1 BOS)(NEGATE *1 V + Pl3 BARRIER S-BOUNDARY); # SV-BOUNDARY ## Unnit nuppástusaiguin. #SELECT:Prop Prop IF (-1 BOS)(0 ("Saemie" Prop))(1 ORGANIZATION OR Org); ## Sámi allaskuvla láhčá diliid dulkoma joatkkaoahppofáldahkii, mii oaivvilduvvo olbmuide, geat dovdet dearvvašvuođa- ja sosiálasuorggi, ja maiddái MBD ovddasvástádussuorggi. # Removing proper nouns that are lookalikes # - - - - - - - - - - - - - - - - - - - - - SELECT:r1894 ("Da") OR ("Dan") IF (NEGATE -1 BOS OR (":") OR PUNCT)(1 (Prop Sem/Sur) OR (Prop Sem/Mal)); SELECT:r1895 ("Da") OR ("dan") IF (-1 BOS OR (":") OR PUNCT); REMOVE:Lea ("Lea") IF (-1 BOS OR PUNCT)(NEGATE 1 V + Sg3); REMOVE:Lea ("Lea") IF (-1 BOS OR PUNCT OR (":"))(1 A); ## Lea dušše okta Ipmil. Lea idja. ## Lea buolaš. Son dajai: Leage gierdávaš. #REMOVE:r1902 ("Man") IF (-1 BOS OR PUNCT OR (":"))(0 (V Imprt)); ## De son celkkii sutnje: "Mana basadit Siloaláddos." # Cannot think of sensible conditions. # Particular proper nouns # ----------------------- LIST SAPMI-PROP-HEAD = "Sijte" "skuvle" "skovle" "misjovne" "learoeviertie" "parlamentarije" "statistihke" "Saervie" "Institudte" "Gielemoenehte" ; # Those combine with the propernoun Sápmi instead of sápmi LIST SAPMI-N-HEAD = "gïele" "lidterature" "voestes" "sijjienomme" "tjaalaldahke" (PUNCT LEFT) ; # These should probably be unified later. # Sámi + these words = common noun: (this is the open group, this should be default). #REMOVE:r1908 Prop IF (0 ("Saemie" Gen))(1 SAPMI-N-HEAD); ## Sámi perspektiiva lea dehálaš. # Sámi + these words = proper noun: (this is the closed set, the set of institutions) #SELECT:r1909 Prop IF (0 ("Saemie"))(*1 SAPMI-PROP-HEAD BARRIER NPNH); ## Sámi dearvvašvuođadutkama guovddáš. ## Sámi Ovdanahttinfoanda ferte vuordit árra čakčii ovdal ruhtadilli čielga. #REMOVE:r1910 Prop IF (0 (N Gen))(1 CC)(2 (N Gen))(NEGATE 2 Prop); ## Sámi ja dáža kultuvra. #REMOVE:r1911 Prop IF (0 ("Saemie"))(NEGATE *1 SAPMI-PROP-HEAD BARRIER NPNH); # We remove the last Prop-analysis. # SAFE RULES # ---------- MAP (@SUBJ) TARGET Inf IF (*-1 (A Nom) BARRIER NOT-ADV); ## Luste tjoejkedh. #MAP (@SUBJ) TARGET Inf IF (*-1 (N Nom) BARRIER NOT-ADV LINK -1 (A Attr)); # ## Lustes aate tjoejkedh. #MAP (@HNOUN) TARGET Nom IF (*-1 BOS BARRIER NPNH)(1 EOS); # ## Tjåetskeme. #MAP (@HNOUN) TARGET Po IF (*-1 BOS BARRIER NPNH)(1 EOS); # ## Bijlen nuelesne. #MAP (@HNOUN) TARGET ADVLCASE IF (*-1 BOS BARRIER NPNH)(1 EOS); # ## Jaevresne. # # SECTION # # # # subject rules and spred rules # # ----------------------------- # #MAP (@SUBJ>) TARGET Nom IF (*-1 BOS BARRIER NPNH)(*1 VFIN BARRIER S-BOUNDARY) (NEGATE *1 (@SUBJ)); # ## Dov orre gaamegh bijlen nuelesne. #MAP (@SUBJ) TARGET Nom IF (-1 GUKTIEGOSSE)(NOT *1 VFIN BARRIER S-BOUNDARY); # ## Gusnie @ADVL> gaamegh? @SUBJ #MAP (@SUBJ) TARGET Nom IF (*-1 BOS BARRIER NPNH)(NOT 0 A LINK *1 (N Nom) OR ADVLCASE + PLACE OR DAESNIE OR PLACE-P BARRIER NPNH)(NOT *1 VFIN BARRIER S-BOUNDARY); # ## Dov orre gaamegh @SUBJ seangkoen nuelesne <@ADVL. # # for ellipsis #MAP (@) TARGET Nom IF (*-1 BOS BARRIER NPNH)(*1 @SUBJ BARRIER NOT-ADV LINK 0 Inf); # ## Luste tjoejkedh. # #$ Lustes aate tjoejkedh. # # # adverbial regler # # ---------------- # #LIST OLLU = "gellie" ; # ##!! adverbs modifying another adverb #MAP:adv-modif-adv (@>ADVL) TARGET Adv IF (0 ("seamma"))(*1 OLLU BARRIER NOT-QMARK); # ## seamma gellie skuvlh vaadtseme goh dah # #MAP:adv-modif-adv (@>ADVL) TARGET ADV-MOD-ADV IF (1 Adv LINK NOT 0 A)(NEGATE *1 ("hov") OR ("aaj") OR ("guktie" Adv) OR ("gumhtie") BARRIER NOT-QMARK) (NEGATE 0 ("mij" Sg Gen) LINK *1 Po BARRIER NOT-QMARK)(NEGATE 0 ("jienebe") OR ("jeenjebe") LINK 1 ("unnebe") OR ("uhtjebe")); # ## Man hijvenlaakan manne maahtam voejedidh # #MAP (@ADVL>) TARGET DAESNIE OR DAESTIE IF (0 Adv) ; ##MAP (@ADVL>) TARGET OLLU IF (*-1 ("seamma") OR ("seammá") BARRIER NOT-QMARK)(0 Adv)(NEGATE *1 Pl BARRIER NOT-QMARK)(*1 FMAINV BARRIER S-BOUNDARY2); ##MAP (@N) TARGET OLLU IF (*-1 ("seamma") OR ("seammá") BARRIER NOT-QMARK)(0 Pron); # ##!! Adv modifying a pronoun or noun #MAP (@>Pron) TARGET Adv IF (0 ADV-MOD-NP)(1 Pron); # ## # #MAP (@ADVL>CS) TARGET Adv (0 ("mænngan")) (1 CS LINK 0 ("goh")) ; # #MAP (@>N) TARGET ADV-MOD-NP (0 Adv)(1 N) ; # #MAP (@>A @>ADVL @>N) TARGET GRADE-ADV IF (*1 A BARRIER NOT-QMARK LINK NOT 0 Po OR Inf OR PrfPrc) (NEGATE 0 ("jienebe") OR ("jeenjebe") LINK *1 ("unnebe") OR ("uhtjebe") BARRIER NOT-QMARK); # ## Dah ligan dan øøvre onne guktie idtjigan dejtie maehtieh vuejnedh barre stueriedehtije klaasen tjirrh. # ##MAP (@>N) TARGET ADV-MOD-NP IF (NEGATE 0 N)(*1 (N Acc) OR @MEASURE + Acc OR CURRENCY + Acc OR TIME + Gen BARRIER NP-BOUNDARY OR N OR Num OR VERB LINK NOT 1 (A Nom) OR ("minngelen") OR ("åvtesne") OR ("avtesne") OR ("bijjelen"); # #MAP (@>Num) TARGET Adv IF (0 ADV-MOD-NP)(*1 Num BARRIER NOT-QMARK); # #MAP (@ADVL>) TARGET (Adv) (*1 VFIN OR PrfPrc OR (Der/NomAct) BARRIER S-BOUNDARY) ; # # #MAP (@ADVL) TARGET Adv IF (0 ADV-MOD-NP OR ("rååktesth") OR ("ryöktesth") OR ("easkah"))(*1 @ADVL BARRIER NPNH); # ## easkah dæjman # # #MAP:modif-advl (@>ADVL) TARGET (Ine) (0 Attr OR Pl LINK 0 $$ADVLCASE LINK *1 $$ADVLCASE BARRIER NPNH); #MAP:modif-advl (@>ADVL) TARGET (Com) (0 Attr OR Pl LINK 0 $$ADVLCASE LINK *1 $$ADVLCASE BARRIER NPNH); #MAP:modif-advl (@>ADVL) TARGET (Ela) (0 Attr OR Pl LINK 0 $$ADVLCASE LINK *1 $$ADVLCASE BARRIER NPNH); #MAP:modif-advl (@>ADVL) TARGET (Ill) (0 Attr OR Pl LINK 0 $$ADVLCASE LINK *1 $$ADVLCASE BARRIER NPNH); # #MAP (@>ADVL) TARGET (A Ord) (0 ($$ADVLCASE+Attr) LINK *1 $$ADVLCASE OR A BARRIER NPNH); # #MAP (@ADVL>) TARGET (Ine) (0 NP-HEAD OR A) (NOT 0 Attr) (*1 VFIN OR PrfPrc OR (Der/NomAct) BARRIER S-BOUNDARY OR CLB) ; # #$ Baenieh njaelmesne lij luste. @N< ? evtl. lage en ADD regel etter hvert # #MAP (@) TARGET (Com) (0 NP-HEAD OR A) (NOT 0 Attr) (*1 VFIN OR PrfPrc OR (Der/NomAct) BARRIER S-BOUNDARY) ; # #MAP (@) TARGET (Ela) (0 NP-HEAD OR A) (NOT 0 Attr) (*1 VFIN OR PrfPrc OR (Der/NomAct) BARRIER S-BOUNDARY) ; # #MAP (@) TARGET (Ill) (0 NP-HEAD OR A) (NOT 0 Attr) (*1 VFIN OR PrfPrc OR (Der/NomAct) BARRIER S-BOUNDARY) ; # #MAP (@) TARGET (Ess) IF (0 )(*-1 VFIN) ; # # # ##MAP (@) TARGET (Ill) IF (*1 FMAINV BARRIER S-BOUNDARY) ; # ## Gellie våhkoeh nåhkeme dan mænngan. # #MAP (@) TARGET Gen IF (0 TIME OR ROUTE)(*1 VFIN BARRIER S-BOUNDARY); # # #MAP (@) TARGET ADVLCASE IF (*1C @SPRED BARRIER NPNH) ; # ## Jaevresne tjåetskeme. #MAP (@OBJ>) TARGET Acc (NOT 0 TIME OR ROUTE)(*1 MAINV + TV BARRIER S-BOUNDARY OR COMMA) ; #MAP (@) TARGET Acc (NOT 1 EOS) ; #"guhkies" A Sg Acc @OBJ> MAP:952 #MAP (@ MAP:950 # ##Manne SOVMEN barre kruepies-laejpieh dah båetieh. "sovme" N Sg Acc @OBJ> MAP:952 # #MAP (@OBJ>) TARGET (Pl Nom) (NOT 0 TIME OR ROUTE)(*1 MAINV + TV BARRIER S-BOUNDARY OR COMMA) ; #MAP (@ MAP:958 # ##Men Jense, dihte guhten dah BAENIEH ij lij dihte ov geerjene dejtie desnie tjoeredh utnedh. "baenie" N Pl Nom @OBJ> MAP:958 AFTER-SECTIONS # # ======================================== # LIST WORDLEMMA = (".*"r) ; REMOVE SUB:1 (cmp_sggen) (0/* (cmp_sgnom)) ; #This for Apertium REMOVE:errsub $$WORDLEMMA + Err/Orth (0 $$WORDLEMMA - Err/Orth ) ; # Apertium REMOVE:errsub2 $$WORDLEMMA + Err/Orth-spes (0 $$WORDLEMMA - Err/Orth-spes ) ; # Apertium #Disse kommenteres inn når analysen skal brukes i functions.cg3 SUBSTITUTE (A) (A ) WORD ; #RemoveFromApertium SUBSTITUTE (N) (N ) WORD ; #RemoveFromApertium SUBSTITUTE (Adv) (Adv ) WORD ; #RemoveFromApertium SUBSTITUTE (V) (V ) WORD ; #RemoveFromApertium SUBSTITUTE (Num) (Num ) WORD ; #RemoveFromApertium SUBSTITUTE (Interj) (Interj ) WORD ; #RemoveFromApertium SUBSTITUTE (Po) (Po ) WORD ; #RemoveFromApertium SUBSTITUTE (Pr) (Pr ) WORD ; #RemoveFromApertium SUBSTITUTE (Pron) (Pron ) WORD ; #RemoveFromApertium SUBSTITUTE (CC) (CC ) WORD ; #RemoveFromApertium SUBSTITUTE (CS) (CS ) WORD ; #RemoveFromApertium SUBSTITUTE (Pcle) (Pcle ) WORD ; #RemoveFromApertium # Not in use: #"" - ikke reell #IFF:GoNotCs - fungerer for eksempelsetninga #IFF:IsItConNeg - fungerer for eksempelsetninga #IFF:S_Interj - fungerer - hvis setninga består bare av interj. #REMOVE:r1901 - rm Lea - tatt fra sme #SELECT:r1894 - redd Dan #SELECT:r1895 - redd Dan