T ree A djoining Gramma rs Natural Language Syntax with T A G Laura Kallmey er & Timm Li hte HHU D�sseldo rf WS 2012 17.10.2012 Natural Language Syntax with T A G 1
Outline 1 The derivation tree 2 Design p rin iples fo r elementa ry trees 3 Sample derivations Natural Language Syntax with T A G 2
Derivation trees (1): The ontext VP NP ↓ NP VP NP NP ↓ P eter V the fridge VP repaired AD V VP* easily derived tree derivation tree VP NP VP P eter AD V VP repaired easily V NP 1 2 22 repaired the fridge easily the_fridge Peter Natural Language Syntax with T A G 3
Derivation trees (2): T A G derivations a re uniquely des rib ed b y derivation trees . The derivation tree ontains: no des fo r all elementa ry trees used in the derivation, and edges fo r all adjun tions and substitutions p erfo rmed throughout the derivation, and edge lab els indi ating the ta rget no de of the rewriting op eration. tree γ Whenever an elementa ry rewrites the no de at Go rn address p tree γ ′ from γ ′ to γ in the elementa ry , there is an edge lab eled with p . Natural Language Syntax with T A G 4
Derivation trees (3): Go rn adresses F o r the no de addresses of elementa ry trees, Go rn addresses a re used: address ǫ The ro ot has (o r 0), and the i th daughter of the no de with address p has address pi . 0 1 2 3 21 22 31 311 312 Natural Language Syntax with T A G 5
Linguisti analyses with L T A G What is an elementa ry tree, and what is its shap e? ? ⇐ synta ti /semanti p rop erties = elementa ry trees of linguisti obje ts ⇒ Synta ti design p rin iples from [F rank, 2002℄: Lexi alization F undamental T A G Hyp othesis (FTH) Condition on Elementa ry T ree Minimalit y (CETM) θ -Criterion fo r T A G ⇒ Semanti design p rin iples [Ab eill� and Ramb o w, 2000 ℄ ⇒ Design p rin iple of e onomy Natural Language Syntax with T A G 6
Design p rin iples (1): Lexi alization Ea h elementa ry tree has at least one non-empt y lexi al item, its lexi al an ho r . ⇒ All widely used gramma r fo rmalisms supp o rt some kind of lexi alization! Reasons fo r lexi alization: F o rmal p rop erties: A �nite lexi alized gramma r p rovides �nitely many analyses fo r ea h string (�nitely ambiguous). Linguisti p rop erties: Synta ti p rop erties of lexi al items an b e a ounted fo r mo re dire tly . P a rsing: The sea r h spa e during pa rsing an b e delimited (gramma r �ltering). [S hab es and Joshi, 1990 , Joshi and S hab es, 1991 ℄ Natural Language Syntax with T A G 7
Design p rin iples (2): F undamental T A G Hyp othesis F undamental T A G Hyp othesis (FTH) Every synta ti dep enden y is exp ressed lo ally within an elementa ry tree. [F rank, 2002 ℄ �synta ti dep enden y� valen y/sub atego rization mo di� ation binding . . . �exp ressed within an elementa ry tree� terminal leaf (i.e. lexi al an ho r) nonterminal leaf (substitution no de and fo otno de) ma rking an inner no de fo r obligato ry adjun tion Natural Language Syntax with T A G 8
Design p rin iples (3): Cond. on Elementa ry T ree Minimalit y Condition on Elementa ry T ree Minimalit y (CETM) The synta ti heads in an elementa ry tree and their p roje tions must fo rm the extended p roje tion of a single lexi al head. [F rank, 2002 ℄ Note: W e only use simple, non-extended p roje tions! S | VP XP . . . VP . . . � X . . . V . . . head . . . sleeps . . . Natural Language Syntax with T A G 9
(4): θ Design p rin iples -Criterion fo r T A G θ -Criterion (T A G version) a. If H is the lexi al head of an elementa ry tree T, H assigns all its θ of -roles in T. b. If A is a frontier non-terminal of elementa ry tree T, A must b e a θ assigned -role in T. [F rank, 2002 ℄ ⇒ = V alen y/sub atego rization is exp ressed only with nonterminal leaves! S VP NP VP , V VP* V seems sleeps Natural Language Syntax with T A G 10
Mo di� ation and fun tional elements Ho w to insert mo di�ers ( easily ) and fun tional elements ( omplementizers, determiners, do-auxilia ries, ...)? Either b y sepa rate auxilia ry trees (e.g., XT A G gramma r), o r as o-an ho r in the elementa ry tree of the lexi al item they a re asso iated with. S S NP VP Comp S that NP VP V AP V sleeps A sleeps easily Natural Language Syntax with T A G 11
Mo di� ation and fun tional elements In XT A G, mo di�ers and fun tional elements a re generally rep resented b y auxilia ry trees. ⇒ F o otno des/A djun tions indi ate b oth omplementation and mo di� ation. ⇒ Enhan ement of the CETM: (see [Ab eill� and Ramb o w, 2000 ℄) o re tree (follo wing CETM) + spine VP S VP* AP Comp S* A that easily Natural Language Syntax with T A G 12
Prin iples related to semanti s See [Ab eill� and Ramb o w, 2000℄. Predi ate-a rgument o o urren e: Ea h elementa ry tree asso iated with a p redi ate ontains a non-terminal leaf fo r ea h of its a rguments. Semanti an ho ring: Elementa ry trees a re not semanti ally void ( to, that .) Comp ositional p rin iple: An elementa ry tree o rresp onds to a single semanti unit. Natural Language Syntax with T A G 13
Design p rin iple of e onomy Design p rin iple of e onomy The elementa ry trees a re shap ed in su h a w a y , that the size of the elementa ry trees and the size of the gramma r is minimal. Natural Language Syntax with T A G 14
Sample derivations Complementation with: NPs, PPs, adje tives, lauses (raising, ontrolling), ... Mo di� ation with: PPs, adje tives, pa rti les, temp o ral lauses, relative lauses, ... Natural Language Syntax with T A G 15
Sample derivations: NP omplements (1) John buys Bill a b o ok. Elementa ry trees: S NP NP NP NP NP ↓ VP N N Det NP* N NP ↓ NP ↓ V John Bill a b o ok buys Derivation tree: buys 1 22 23 John Bill book ǫ a Natural Language Syntax with T A G 16
Sample derivations: Sentential omplements (1) (2) Bill hop es that John wins. Elementa ry trees: S S NP S NP NP ↓ NP ↓ VP VP N Comp S* N S ∗ V V Bill that John hop es wins Derivation tree: wins ǫ 1 that John ǫ hopes 1 Bill Natural Language Syntax with T A G 17
Sample derivations: Sentential omplements (2) (3) John seems to lik e Bill. Elementa ry trees: S NP ↓ VP VP NP ↓ VP ∗ VP V seems V to lik e Derivation tree: to_like 1 2 22 seems John Bill Natural Language Syntax with T A G 18
Sample derivations: Sentential omplements (3) (4) John exp e ts [ Bill to win ℄. Elementa ry trees: S S NP ↓ NP ↓ VP VP S ∗ V V exp e ts to win Derivation tree: to_win ǫ 1 expe ts Bill 1 John Natural Language Syntax with T A G 19
Sample derivations: Sentential omplements (4) Question: Why is the sentential obje t rep resented as a fo otno de? The sentential obje t is realised as a fo ot no de in o rder to allo w extra tions: (5) Who do es John exp e t to win ? Elementa ry trees: S NP ↓ S VP S NP ↓ VP V VP* NP VP S ∗ V ǫ do es V exp e t to win Natural Language Syntax with T A G 20
Sample derivations: Multiple an ho rs Multiw o rd exp ressions and light verb onstru tions an b e rep resented b y elementa ry trees with multiple an ho rs: (6) John exp e ted [Ma ry to mak e a omment℄. S NP ↓ VP NP NP ∗ V NP Det to mak e N a omment Natural Language Syntax with T A G 21
Sample derivations: Mo di�ers (7) The go o d student pa rti ipated in every ourse during the semester. N NP NP N ∗ AP Det NP* N A the student go o d S NP ↓ VP VP VP ∗ PP V PP NP ↓ P NP ↓ pa rti ipated P during in Natural Language Syntax with T A G 22
Sample derivations: Relative lauses (8) The dog [who ate the ak e℄. NP NP NP NP* S NP ↓ VP Det NP* N NP ↓ V the dog ate Problem: Extrap osed relative lauses: (1) Someb o dy lives nea rb y [who has a CD-burner℄. i i Natural Language Syntax with T A G 23
Derivation trees = Semanti dep enden y stru ture ? The derivation tree is not alw a ys the semanti dep enden y stru ture, due to: indis ernibilit y of omplementation and mo di� ation in adjun tion, and missing links. Example fo r a missing link: (2) John laims [Bill seems to win℄ to_win ǫ 1 2 seems laims Bill 1 John Natural Language Syntax with T A G 24
Recommend
More recommend
