1
0
forked from GitHub/gf-core

treebank creating script; duplicated consonants in LexiconEng

This commit is contained in:
aarne
2006-03-04 14:58:11 +00:00
parent 277a333a02
commit ee26cf955b
9 changed files with 70 additions and 16 deletions

View File

@@ -23,6 +23,9 @@ multimodal:
compiled:
cd .. ; tar cvfz compiled.tgz alltenses/ mathematical/ multimodal/ present/
treebank:
gf <mkTreebank.gfs
stat:
wc */*.gfc

View File

@@ -13,13 +13,13 @@ lin
art_N = regN "art" ;
ask_V2Q = mkV2Q (regV "ask") [] ;
baby_N = regN "baby" ;
bad_A = regADeg "bad" ;
bad_A = mkADeg "bad" "badly" "worse" "worst" ;
bank_N = regN "bank" ;
beautiful_A = regADeg "beautiful" ;
become_VA = mkVA (irregV "become" "became" "become") ;
beer_N = regN "beer" ;
beg_V2V = mkV2V (regDuplV "beg") [] "to" ;
big_A = regADeg "big" ;
big_A = duplADeg "big" ;
bike_N = regN "bike" ;
bird_N = regN "bird" ;
black_A = regADeg "black" ;
@@ -74,7 +74,7 @@ lin
find_V2 = dirV2 (irregV "find" "found" "found") ;
fish_N = mk2N "fish" "fish" ;
floor_N = regN "floor" ;
forget_V2 = dirV2 (irregV "forget" "forgot" "forgotten") ;
forget_V2 = dirV2 (irregDuplV "forget" "forgot" "forgotten") ;
fridge_N = regN "fridge" ;
friend_N = regN "friend" ;
fruit_N = regN "fruit" ;
@@ -94,7 +94,7 @@ lin
hill_N = regN "hill" ;
hope_VS = mkVS (regV "hope") ;
horse_N = regN "horse" ;
hot_A = regADeg "hot" ;
hot_A = duplADeg "hot" ;
house_N = regN "house" ;
important_A = compoundADeg (regA "important") ;
industry_N = regN "industry" ;
@@ -142,7 +142,7 @@ lin
radio_N = regN "radio" ;
rain_V0 = mkV0 (regV "rain") ;
read_V2 = dirV2 (irregV "read" "read" "read") ;
red_A = regADeg "red" ;
red_A = duplADeg "red" ;
religion_N = regN "religion" ;
restaurant_N = regN "restaurant" ;
river_N = regN "river" ;
@@ -186,7 +186,7 @@ lin
teach_V2 = dirV2 (irregV "teach" "taught" "taught") ;
television_N = regN "television" ;
thick_A = regADeg "thick" ;
thin_A = regADeg "thin" ;
thin_A = duplADeg "thin" ;
train_N = regN "train" ;
travel_V = (regDuplV "travel") ;
tree_N = regN "tree" ;
@@ -204,7 +204,7 @@ lin
white_A = regADeg "white" ;
window_N = regN "window" ;
wine_N = regN "wine" ;
win_V2 = dirV2 (irregV "win" "won" "won") ;
win_V2 = dirV2 (irregDuplV "win" "won" "won") ;
woman_N = mk2N "woman" "women" ;
wonder_VQ = mkVQ (regV "wonder") ;
wood_N = regN "wood" ;

View File

@@ -18,6 +18,7 @@ abstract MinStructural = Cat ** {
if_Subj : Subj ;
in_Prep : Prep ;
that_NP : NP ;
this_NP : NP ;
we_Pron : Pron ;
whichPl_IDet : IDet ;
whichSg_IDet : IDet ;

View File

@@ -13,6 +13,7 @@ incomplete concrete MinStructuralI of MinStructural = open Structural in {
if_Subj = if_Subj ;
in_Prep = in_Prep ;
that_NP = that_NP ;
this_NP = this_NP ;
we_Pron = we_Pron ;
whichPl_IDet = whichPl_IDet ;
whichSg_IDet = whichSg_IDet ;

View File

@@ -0,0 +1,10 @@
-- to create a treebank with 3066 trees, for all languages
i minimal/MinimalEng.gf
gt -depth=4 -cat=S | tb -xml | grep -v "/treebank>" | wf min.xml
gt -depth=4 -cat=QS | tb -xml | grep -v "treebank>" | af min.xml
gt -depth=3 (UttImpSg ? ?) | tb -xml | grep -v "treebank>" | af min.xml
gt -depth=4 -cat=NP | tb -xml | grep -v "<treebank" | af min.xml
e
i -nocf langs.gfcm
rf min.xml | tb -trees | tb -xml | wf langs.xml