Outline Grammar Formalisms Syntax in LTAG The derivation tree 1 Design principles for elementary trees 2 Laura Kallmeyer, Timm Lichte, Wolfgang Maier Sample derivations Universit¨ at T¨ ubingen 3 30.04.2007 Syntax in LTAG 1 Syntax in LTAG 2 Derivation trees (1) Derivation trees (2) Example: TAG derivations are described by derivation trees : derivation tree for the derivation of John sometimes laughs For each derivation in a TAG there is a corresponding derivation tree. This tree contains laughs nodes for all elementary trees used in the derivation, and 1 2 edges for all adjunctions and substitutions performed john sometimes throughout the derivation. ⇒ Semantic dependency graph? Whenever an elementary tree γ was attached to the node at address p in the elementary tree γ ′ , there is an edge from γ ′ to γ Derivation trees labeled with p . are context-free, and (For the tree addresses, Gorn addresses are used: The root has uniquely determine the derived tree. address ǫ (or 0), and the i th daughter of the node with address p has address pi .) ⇒ TAG is a linear context-free rewriting system, LCFRS Syntax in LTAG 3 Syntax in LTAG 4
Linguistic analyses with LTAG Lexicalization Each elementary tree has at least one non-empty lexical item, its lexical anchor . ⇒ All widely used grammar formalisms support some kind of lexicalization! Linguistic role of the adjunction/substitution dualism Design principles for elementary trees Reasons for lexicalization: Lexicalization The properties of a constituent depend on the lexical items Extended domain of locality occurring in the constituent: the structure of a VP depends Factoring of recursion on the subcategorization properties of its verb. Elementary tree minimality S S NP s VP NP s VP VP → V intrans VP → V trans NP V V NP o sleeps likes Syntax in LTAG 5 Syntax in LTAG 6 Elementary tree minimality Extended domain of locality All dependencies are represented in elementary trees. Elementary trees contain slots only for the complements of locality constraints (constraints for UDC, island constraints) their lexical head. → adjunction constraints Elementary trees are projections of lexical items. (2) a. who i did John tell Sam that Bill likes t i Factoring of recursion b. who i did John tell Sam that Mary said that Bill likes t i Example: to-infinitives S Example: S WH i S OA (1) John gives a book to Mary INFL NP VP COMP S S did V NP S ∗ NP ↓ VP that NP VP NP tell V NP ↓ PP WH i V NP John NP gives P NP ↓ who NP likes ǫ i Sam to Bill Syntax in LTAG 7 Syntax in LTAG 8
Principles related to semantics Functional elements Predicate argument cooccurrence: Besides lexical predicates, there are functional elements Each elementary tree associated with a predicate contains substitution nodes for each of its arguments. (complementizers, determiners, auxiliaries, negation) whose treatment in LTAG is less clear. They can be Semantic anchoring: either in separate elementary trees (e.g., XTAG grammar) Elementary trees are not semantically void ( to, that .) or in the elementary tree of the lexical item they are associated with. Compositional principle: An elementary tree corresponds to a single semantic unit. Syntax in LTAG 9 Syntax in LTAG 10 Sample derivations Sample derivations - Complements (1) (3) John buys Bill a book Elementary trees: Complements: NPs, PPs, adjectives, clauses (raising, S NP controlling) NP NP NP ↓ VP Adjuncts: adjectives, particles, relative clauses Det N V NP ↓ NP ↓ John Bill a book buys Complements ⇒ Substitution, Adjuncts ⇒ Adjunction ? buys Derivation tree: 1 22 23 John Bill a book Syntax in LTAG 11 Syntax in LTAG 12
Sample derivations - Complements (2) Sample derivations - Complements (2) (4) Bill hopes that John wins S (5) John seems to like Bill S Comp S S NP NP NP ↓ VP VP that NP ↓ VP NP ↓ VP V S ∗ Bill John V VP ∗ V VP NP ↓ hopes seems wins to like wins to like 1 ǫ Derivation tree: Derivation tree: hopes John 1 2 22 1 John seems Bill Bill Syntax in LTAG 13 Syntax in LTAG 14 Sample derivations - Complements (2) Sample derivations - Complements (2) (7) John expected [Mary to make a comment] (6) John expects [ Bill to win ] expected selects for a subject NP and an infinitival sentence: S S S S NP ↓ VP NP ↓ VP NP ↓ VP NP ↓ VP V S ∗ NP V S ∗ to win expects to make a comment expected John to win 1 ǫ The sentential object is realised as a foot node in order to allow Derivation tree: expects Bill extractions: 1 John (8) whom does John expect to come? Syntax in LTAG 15 Syntax in LTAG 16
Sample derivations - Complements (3) Sample derivations - Adjuncts (1) (9) the good student participated in every course during the semester N to make a comment : make and comment in the same elementary NP AP N ∗ tree since they form a light verb construction: Det N S A NP ↓ VP the student NP good V NP Det NP ∗ S to make N a VP NP ↓ VP VP ∗ PP comment V PP P NP ↓ participated P NP ↓ during in Syntax in LTAG 17 Syntax in LTAG 18 Sample derivations - Adjuncts (1) Derivation trees = Semantic dependency structure ? The derivation tree is not always the semantic dependency structure: (12) John claims Bill is likely to win (10) the dog [who ate the cake] to win Extraposed relative clauses: 1 2 ǫ (11) Somebody i lives nearby [who i has a CD-burner]. Bill claims is likely 1 John Syntax in LTAG 19 Syntax in LTAG 20
Symmary TAG derivations are described by derivation trees. In LTAG, elementary trees for lexical predicates contain slots for all arguments of these predicates, for nothing else. Recursion is factored away. The derived tree describes the constituent structure while the derivation tree is close to a semantic dependency graph. Syntax in LTAG 21
Recommend
More recommend
