more accurate BigLexEng (V3 still to do)

This commit is contained in:
aarne
2006-11-02 21:45:32 +00:00
parent 3396a13ee7
commit c922c0c488
5 changed files with 11569 additions and 11998 deletions

View File

@@ -1,4 +1,4 @@
--# -path=.:../abstract:../common:prelude
--# -path=.:alltenses:prelude
concrete BigEnglish of BigEnglishAbs =
English,

File diff suppressed because it is too large Load Diff

File diff suppressed because it is too large Load Diff

View File

@@ -1,3 +1,5 @@
import Char
infile = "mywordlist1"
main = do
@@ -6,21 +8,24 @@ main = do
mkOne s = case words s of
"--":_ -> ""
('(':_):w:cat:ws ->
let
(c,f) = mkCatf (nopar cat) (more ws)
in unwords $ [c, f, w]
('(':_):w:cat:ws -> unwords $ mkCatf (nopar cat) (more ws) w
_ -> "-- " ++ s
where
more ws = case ws of
_ | elem "(REG" ws -> "irreg"
_ -> "reg"
nopar = filter (flip notElem "()")
mkCatf c r = case c of
"Noun" -> ("N","regN")
"Adject" -> ("A","regA")
"Adject_LONG" -> ("A","longA")
"Verb" -> ("V","regV")
"PNoun" -> ("PN","regPN")
_ -> (c,"mk" ++ c)
mkCatf c r w = case c of
"Noun" -> ["N","regN",w]
"Adject" -> ["A","regA",w]
"Adject_LONG" -> ["A","longA",w]
"Verb" | r == "irreg" -> []
"Verb" -> ["V","regV",w]
"V2" | r == "irreg" -> ["V2","irreg", w, "_V"]
"V2" -> ["V2","regV2", w]
"PNoun" -> ["PN","regPN",toUpper (head w): tail w]
'V':'2':'_':prep | r == "irreg" ->
["V2","mkV2", w, "_V", map toLower prep]
x:'2':'_':prep -> [[x]++"2","prep" ++[x]++"2", w, map toLower prep]
_ -> [c,"mk" ++ c, w]

View File

@@ -351,9 +351,6 @@
(are are (Verb (FIN +) (PAST -) (AUX +) (SUBCAT PRED) (AGR PLUR)) BE NIL)
(area area Noun AREA NIL)
(arena arena Noun ARENA NIL)
("aren't" "aren't" (Verb (FIN +) (PAST -) (AUX +) (SUBCAT PRED) (AGR SING2) (NEG +)) BE NIL)
("aren't" "aren't" (Verb (FIN +) (PAST -) (AUX +) (SUBCAT PRED) (AGR PLUR) (NEG +)) BE NIL)
("aren't" "aren't" (Verb (FIN +) (PAST -) (AUX +) (SUBCAT PRED) (INV +) (AGR SING1) (NEG +)) BE NIL)
(argue argue (Verb (SUBCAT NULL)) ARGUE NIL)
(argument argument (N2_S_WITHABOUT)) ARGUMENT NIL)
(argument argument (Noun (SUBCAT STHAT)) ARGUMENT NIL)
@@ -799,7 +796,6 @@
(cannon cannon Noun CANNON NIL)
(cannot cannot (Verb (AUX +) (SUBCAT BASE_VP) (FIN +) (PAST -) (NEG +)) CAN NIL)
(canonical canonical (Adject_LONG) CANONICAL NIL)
("can't" "can't" (Verb (AUX +) (SUBCAT BASE_VP) (FIN +) (PAST -) (NEG +)) CAN NIL)
(cap cap Noun CAP NIL)
(capable capable (Adject_LONG (SUBCAT NULL)) CAPABLE NIL)
(capable capable (Adject_LONG (SUBCAT PPOF)) CAPABLE NIL)
@@ -1355,7 +1351,7 @@
(creator creator Noun CREATOR NIL)
(creature creature Noun CREATURE NIL)
(credit credit Noun CREDIT NIL)
(creep creep (V2_) (LAT -) (REG -)) CREEP NIL)
(creep creep V2 (LAT -) (REG -)) CREEP NIL)
(cricket cricket Noun CRICKET NIL)
(crime crime Noun CRIME NIL)
(criminal criminal (Adject_LONG) CRIMINAL NIL)
@@ -1436,9 +1432,6 @@
(dangle dangle (Verb (SUBCAT NULL)) DANGLE NIL)
(dare dare (Verb (SUBCAT BASE_VP)) DARE NIL)
(dare dare (Verb (SUBCAT SE1)) DARE NIL)
("daren't" "daren't" (Verb (FIN +) (PAST -) (SUBCAT BASE_VP) (AGR SING1) (NEG +)) DARE NIL)
("daren't" "daren't" (Verb (FIN +) (PAST -) (SUBCAT BASE_VP) (AGR SING2) (NEG +)) DARE NIL)
("daren't" "daren't" (Verb (FIN +) (PAST -) (SUBCAT BASE_VP) (AGR PLUR) (NEG +)) DARE NIL)
(dark dark Noun DARK NIL)
(dark dark ((V +) (N +) (LAT -)) DARK NIL)
(darken darken V2 DARKEN NIL)
@@ -1542,7 +1535,7 @@
(depth depth Noun DEPTH NIL)
(derivative derivative Noun DERIVATIVE NIL)
(derive derive (V2_FROM)) DERIVE NIL)
(descend descend (V2_)) DESCEND NIL)
(descend descend V2) DESCEND NIL)
(descendant descendant Noun DESCENDANT NIL)
(descent descent Noun DESCENT NIL)
(describe describe V2 DESCRIBE NIL)
@@ -1723,8 +1716,6 @@
(document document V2 DOCUMENT NIL)
(does does (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (AGR IT)) DO NIL)
(does does (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (AGR SING3)) DO NIL)
("doesn't" "doesn't" (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (NEG +) (AGR IT)) DO NIL)
("doesn't" "doesn't" (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (NEG +) (AGR SING3)) DO NIL)
(dog dog Noun DOG NIL)
(dollar dollar Noun DOLLAR NIL)
(dolt dolt Noun DOLT NIL)
@@ -1735,9 +1726,6 @@
(dominate dominate (Verb (AT -) (SUBCAT NP)) DOMINATE NIL)
(donate donate (Verb (SUBCAT NP_PPTO) (AT -)) DONATE NIL)
(donkey donkey Noun DONKEY NIL)
("don't" "don't" (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (AGR SING1) (NEG +)) DO NIL)
("don't" "don't" (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (AGR SING2) (NEG +)) DO NIL)
("don't" "don't" (Verb (AUX +) (SUBCAT DO_COMPL) (FIN +) (PAST -) (AGR PLUR) (NEG +)) DO NIL)
(doom doom Noun DOOM NIL)
(doomed doomed (Adject_LONG) DOOMED NIL)
(doomsday doomsday Noun DOOMSDAY NIL)
@@ -1818,7 +1806,7 @@
(duty duty Noun DUTY NIL)
(dwarf dwarf Noun DWARF NIL)
(dwarf dwarf V2 DWARF NIL)
(dwell dwell (V2_) (REG -)) DWELL NIL)
(dwell dwell V2 (REG -)) DWELL NIL)
(dwelling dwelling Noun DWELLING NIL)
(dynamic dynamic (Adject_LONG) DYNAMIC NIL)
(dynamically dynamically Adv DYNAMICALLY NIL)
@@ -2487,7 +2475,7 @@
(glow glow V2 GLOW NIL)
(gnat gnat Noun GNAT NIL)
(gnu gnu Noun GNU NIL)
(go go (V2_) (REG -)) GO NIL)
(go go V2 (REG -)) GO NIL)
(goat goat Noun GOAT NIL)
(god god Noun GOD NIL)
(goddess goddess Noun GODDESS NIL)
@@ -2628,9 +2616,6 @@
(has has (Verb (FIN +) (SUBCAT NP) (PAST -) (AGR SING3)) HAVE NIL)
(hash hash Noun HASH NIL)
(hash hash V2 HASH NIL)
("hasn't" "hasn't" (Verb (AUX +) (FIN +) (SUBCAT PSP) (PAST -) (AGR SING3) (NEG +)) HAVE NIL)
("hasn't" "hasn't" (Verb (AUX +) (FIN +) (SUBCAT PSP) (PAST -) (AGR S) (NEG +)) HAVE NIL)
("hasn't" "hasn't" (Verb (FIN +) (SUBCAT NP) (PAST -) (AGR SING3) (NEG +)) HAVE NIL)
(haste haste Noun HASTE NIL)
(hasty hasty (Adject_LONG) HASTY NIL)
(hat hat Noun HAT NIL)
@@ -2639,8 +2624,6 @@
(have have (V2 (REG -)) HAVE NIL)
(have have (Verb (SUBCAT PSP) (INFL -) (AUX +) (REG -)) HAVE NIL)
(haven haven Noun HAVEN NIL)
("haven't" "haven't" (V2 (NEG +)) HAVE NIL)
("haven't" "haven't" (Verb (SUBCAT PSP) (INFL -) (AUX +) (NEG +)) HAVE NIL)
(having having (Verb (VFORM ING) (PRD -) (AUX +) (SUBCAT PSP)) HAVING NIL)
(hazard hazard Noun HAZARD NIL)
(haze haze Noun HAZE NIL)
@@ -3100,8 +3083,6 @@
(is is (Verb (AUX +) (SUBCAT PRED) (FIN +) (PAST -) (AGR SING3)) BE NIL)
(is is (Verb (AUX +) (SUBCAT PRED) (FIN +) (PAST -) (AGR S)) BE NIL)
(island island Noun ISLAND NIL)
("isn't" "isn't" (Verb (AUX +) (SUBCAT PRED) (FIN +) (PAST -) (AGR SING3) (NEG +)) BE NIL)
("isn't" "isn't" (Verb (AUX +) (SUBCAT PRED) (FIN +) (PAST -) (AGR S) (NEG +)) BE NIL)
(isolable isolable (Adject_LONG) ISOLABLE NIL)
(isolate isolate (V2 (AT -)) ISOLATE NIL)
(isomorphic isomorphic (Adject_LONG) ISOMORPHIC NIL)
@@ -3378,7 +3359,7 @@
(luck luck Noun LUCK NIL)
(lump lump Noun LUMP NIL)
(lung lung Noun LUNG NIL)
(lunge lunge (V2_)) LUNGE NIL)
(lunge lunge V2) LUNGE NIL)
(lurk lurk (Verb (SUBCAT NULL)) LURK NIL)
(lush lush Adject LUSH NIL)
(luxury luxury Noun LUXURY NIL)
@@ -3525,7 +3506,6 @@
(middle middle (Adject_LONG) MIDDLE NIL)
(middle middle Noun MIDDLE NIL)
(might might (Verb (SUBCAT BASE_VP) (AUX +) (PAST -) (FIN +)) MIGHT NIL)
("mightn't" "mightn't" (Verb (SUBCAT BASE_VP) (AUX +) (PAST -) (FIN +) (NEG +)) MIGHT NIL)
(mighty mighty Adject MIGHTY NIL)
(migrate migrate (Verb (AT -) (SUBCAT NULL)) MIGRATE NIL)
(mild mild Adject MILD NIL)
@@ -3635,7 +3615,6 @@
(music music (Noun (COUNT -)) MUSIC NIL)
(musical musical Noun MUSICAL NIL)
(must must (Verb (SUBCAT BASE_VP) (FIN +) (PAST -) (AUX +)) MUST NIL)
("mustn't" "mustn't" (Verb (SUBCAT BASE_VP) (FIN +) (PAST -) (AUX +) (NEG +)) MUST NIL)
(mutter mutter (V2_ABOUT)) MUTTER NIL)
(mutual mutual (Adject_LONG) MUTUAL NIL)
(mystery mystery Noun MYSTERY NIL)
@@ -3670,7 +3649,6 @@
(need need Noun NEED NIL)
(need need (Verb (SUBCAT SR1)) NEED NIL)
(needle needle Noun NEEDLE NIL)
("needn't" "needn't" (Verb (SUBCAT BASE_VP) (NEG +)) NEED NIL)
(negate negate (Verb (AT -) (SUBCAT NP)) NEGATE NIL)
(negative negative (Adject_LONG) NEGATIVE NIL)
(negative negative Noun NEGATIVE NIL)
@@ -3886,7 +3864,6 @@
(otherwise otherwise (Adv OTHERWISE NIL)
(otherwise otherwise ((SUBCAT OTHERWISE)) OTHERWISE NIL)
(ought ought (Verb (SUBCAT INF) (FIN +) (PAST -) (AUX +)) OUGHT NIL)
("oughtn't" "oughtn't" (Verb (SUBCAT INF) (FIN +) (PAST -) (AUX +) (NEG +)) OUGHT NIL)
(out out ((V -) (N -) (SUBCAT PP)) OUT NIL)
(outdated outdated (Adject_LONG) OUTDATED NIL)
(outdoor outdoor (Adject_LONG) OUTDOOR NIL)
@@ -4665,7 +4642,7 @@
(rescue rescue V2 RESCUE NIL)
(resemble resemble V2 RESEMBLE NIL)
(reset reset V2 RESET NIL)
(reside reside (V2_)) RESIDE NIL)
(reside reside V2) RESIDE NIL)
(residence residence Noun RESIDENCE NIL)
(resident resident Noun RESIDENT NIL)
(residue residue Noun RESIDUE NIL)
@@ -4799,7 +4776,6 @@
(rush rush Noun RUSH NIL)
(russian russian (Adject_LONG) RUSSIAN NIL)
(russian russian Noun RUSSIAN NIL)
("'s" "'s" (Noun (POSS +)) S NIL)
(sabbatical sabbatical (Adject_LONG) SABBATICAL NIL)
(sad sad Adject SAD NIL)
(safe safe Noun SAFE NIL)
@@ -4823,7 +4799,7 @@
(sandy sandy (PNoun) SANDY NIL)
(sane sane Adject SANE NIL)
(sat sat (Verb (SUBCAT NULL)) SIT NIL)
(sat sat (V2_)) SIT NIL)
(sat sat V2) SIT NIL)
(satanic satanic (Adject_LONG) SATANIC NIL)
(satisfaction satisfaction Noun SATISFACTION NIL)
(satisfactory satisfactory (Adject_LONG) SATISFACTORY NIL)
@@ -4948,7 +4924,6 @@
(shall shall (Verb (AUX +) (SUBCAT BASE_VP) (FIN +) (PAST -)) SHALL NIL)
(shallow shallow Adject SHALLOW NIL)
(shame shame Noun SHAME NIL)
("shan't" "shan't" (Verb (AUX +) (SUBCAT BASE_VP) (FIN +) (PAST -) (NEG +)) SHALL NIL)
(shape shape V2 SHAPE NIL)
(share share Noun SHARE NIL)
(share share (Verb (SUBCAT NP_PPWITH)) SHARE NIL)
@@ -4985,7 +4960,6 @@
(should should (Verb (AUX +) (SUBCAT BASE_VP) (PAST -) (FIN +)) SHOULD NIL)
(shoulder shoulder Noun SHOULDER NIL)
(shoulder shoulder V2 SHOULDER NIL)
("shouldn't" "shouldn't" (Verb (AUX +) (SUBCAT BASE_VP) (PAST -) (FIN +) (NEG +)) SHOULD NIL)
(shout shout Noun SHOUT NIL)
(shout shout (Verb (SUBCAT NULL)) SHOUT NIL)
(show show Noun SHOW NIL)
@@ -5034,7 +5008,7 @@
(sir sir ((TITLE +)) SIR NIL)
(sister sister (N2_OF)) SISTER NIL)
(sit sit (Verb (SUBCAT NULL)) SIT NIL)
(sit sit (V2_)) SIT NIL)
(sit sit V2) SIT NIL)
(site site Noun SITE NIL)
(situation situation Noun SITUATION NIL)
(size size Noun SIZE NIL)
@@ -5188,7 +5162,7 @@
(spot spot V2 SPOT NIL)
(spread spread (Verb (SUBCAT NULL) (REG -)) SPREAD NIL)
(spring spring Noun SPRING NIL)
(spring spring (V2_) (REG -)) SPRING NIL)
(spring spring V2 (REG -)) SPRING NIL)
(spur spur Noun SPUR NIL)
(spurious spurious (Adject_LONG) SPURIOUS NIL)
(square square Adject SQUARE NIL)
@@ -5248,7 +5222,7 @@
(stem stem Noun STEM NIL)
(stem stem V2 STEM NIL)
(step step Noun STEP NIL)
(step step (V2_)) STEP NIL)
(step step V2) STEP NIL)
(sterile sterile (Adject_LONG) STERILE NIL)
(stick stick Noun STICK NIL)
(stick stick (V2_TO) (REG -)) STICK NIL)
@@ -5952,7 +5926,7 @@
(walk walk Noun WALK NIL)
(wall wall Noun WALL NIL)
(wan wan Adject WAN NIL)
(wander wander (V2_)) WANDER NIL)
(wander wander V2) WANDER NIL)
(want want (Verb (SUBCAT OR)) WANT NIL)
(want want (Verb (SUBCAT SE1)) WANT NIL)
(war war Noun WAR NIL)
@@ -6068,7 +6042,6 @@
(women women (Noun (PLU +)) WOMAN NIL)
(wonder wonder (Verb (SUBCAT Q)) WONDER NIL)
(wonder wonder Noun WONDER NIL)
("won't" "won't" (Verb (AUX +) (SUBCAT BASE_VP) (FIN +) (PAST -) (NEG +)) WILL NIL)
(wood wood Noun WOOD NIL)
(wooden wooden (Adject_LONG) WOODEN NIL)
(wool wool (Noun (COUNT -)) WOOL NIL)