June 7th, 2008TAG+91 Binding Theory in LTAG Lucas Champollion University of Pennsylvania

Slides:



Advertisements
Similar presentations
Review of Homework 4 November 20, Exercise 1 (A) 1 pt Modify the DCG to accept the omission of the complementizer that for object relative clauses.
Advertisements

Lecture 2: Constraints on Movement.  Formal movement rules (called Transformations) were first introduced in the late 1950s  During the 1960s a lot.
 Christel Kemke 2007/08 COMP 4060 Natural Language Processing Feature Structures and Unification.
Recursion-06: 1 A Tale of Recursion (A very preliminary version) ARAVIND K. JOSHI April (revised May )
Notes on TAG (LTAG) and Feature Structures September AKJ.
07/05/2005CSA2050: DCG31 CSA2050 Introduction to Computational Linguistics Lecture DCG3 Handling Subcategorisation Handling Relative Clauses.
Lexical Functional Grammar : Grammar Formalisms Spring Term 2004.
Sub-constituents of NP in English September 12, 2007.
Dr. Abdullah S. Al-Dobaian1 Ch. 2: Phrase Structure Syntactic Structure (basic concepts) Syntactic Structure (basic concepts)  A tree diagram marks constituents.
Long Distance Dependencies (Filler-Gap Constructions) and Relative Clauses October 10, : Grammars and Lexicons Lori Levin (Examples from Kroeger.
1 Chapter Chapter 5 Grammars for Natural Language 5.1 Auxiliary Verbs and Verb Phrases 5.2 Movement Phenomena in Language 5.3 Handling Questions in Context-Free.
Detecting Complements and Adjuncts Rajat Kumar Mohanty Center for Indian Language Technology IIT Bombay.
Linguistic Theory Lecture 7 About Nothing. Nothing in grammar Language often contains irregular paradigms where one or more expected forms are absent.
Statistical NLP: Lecture 3
Lecture 11: Binding and Reflexivity.  Pronouns differ from nouns in that their reference is determined in context  The reference of the word dog is.
LTAG Semantics on the Derivation Tree Presented by Maria I. Tchalakova.
Reference Resolution #1 CSCI-GA.2590 Ralph Grishman NYU.
University of Alberta6/3/20151 Governing Category and Coreference Dekang Lin Department of Computing Science University of Alberta.
NLP and Speech 2004 Feature Structures Feature Structures and Unification.
Features & Unification Ling 571 Deep Processing Techniques for NLP January 26, 2011.
Week 5a. Binding theory CAS LX 522 Syntax I. Structural ambiguity John said that Bill slipped in the kitchen. John said that Bill slipped in the kitchen.
CIS 630 Slides for March Schedule We missed two classes. I will try to make up in April Spring break– Week of March 8 Your presentations I would.
Starting With Complex Primitives Pays Off: Complicate Locally, Simplify Globally ARAVIND K. JOSHI Department of Computer and Information Science and Institute.
 Christel Kemke 2007/08 COMP 4060 Natural Language Processing Feature Structures and Unification.
Natural Language Query Interface Mostafa Karkache & Bryce Wenninger.
MC-TAG, flexible composition, etc. ARAVIND K. JOSHI March
Bounding Theory Constraints on Wh-movement. NP islands What i did Bill claim [ CP that he read t i ?] *What did Bill make [ NP the claim [ CP that he.
Probabilistic Parsing Ling 571 Fei Xia Week 5: 10/25-10/27/05.
Notes on TAG (LTAG) and Feature Structures Aravind K. Joshi April
Syntax Nuha AlWadaani.
Linguistic Theory Lecture 3 Movement. A brief history of movement Movements as ‘special rules’ proposed to capture facts that phrase structure rules cannot.
VP: [VP[Vhelp[ [PRNyou]]
11 CS 388: Natural Language Processing: Syntactic Parsing Raymond J. Mooney University of Texas at Austin.
LING/C SC/PSYC 438/538 Lecture 19 Sandiway Fong 1.
LING 388: Language and Computers Sandiway Fong Lecture 17.
October 2008csa3180: Setence Parsing Algorithms 1 1 CSA350: NLP Algorithms Sentence Parsing I The Parsing Problem Parsing as Search Top Down/Bottom Up.
Tree-adjoining grammar (TAG) is a grammar formalism defined by Aravind Joshi and introduced in Tree-adjoining grammars are somewhat similar to context-free.
Continuous Discontinuity in It-Clefts Introduction Tension between the two approaches Our proposal: TAG analysis Equative it-cleft: It was Ohno who won.
1 Statistical Parsing Chapter 14 October 2012 Lecture #9.
Differential effects of constraints in the processing of Russian cataphora Kazanina and Phillips 2010.
Episode 4a. Binding Theory, NPIs, c- command. 4.3 CAS LX 522 Syntax I.
Binding Theory Describing Relationships between Nouns.
Intro to NLP - J. Eisner1 Tree-Adjoining Grammar (TAG) One of several formalisms that are actually more powerful than CFG Note: CFG with features.
A Cascaded Finite-State Parser for German Michael Schiehlen Institut für Maschinelle Sprachverarbeitung Universität Stuttgart
Markus Egg, Alexander Koller, Joachim Niehren The Constraint Language for Lambda Structures Ventsislav Zhechev SfS, Universität Tübingen
1 Introduction to Computational Linguistics Eleni Miltsakaki AUTH Fall 2005-Lecture 4.
An experimental investigation of referential/non-referential asymmetries in syntactic reconstruction akira omaki anastasia conroy jeffrey lidz Quantitative.
Semantic Construction lecture 2. Semantic Construction Is there a systematic way of constructing semantic representation from a sentence of English? This.
Rules, Movement, Ambiguity
Linguistic Theory Lecture 5 Filters. The Structure of the Grammar 1960s (Standard Theory) LexiconPhrase Structure Rules Deep Structure Transformations.
November 16, 2004 Lexicon (An Interacting Subsystem in UG) Part-II Rajat Kumar Mohanty IIT Bombay.
Supertagging CMSC Natural Language Processing January 31, 2006.
 Chapter 8 (Part 2) Transformations Transformational Grammar Engl 424 Hayfa Alhomaid.
Safir, Ken Vehicle Change and Reconstruction. Linguistic Inquiry. Vol. 30-4: Vehicle Change and Reconstruction in Ā-Chains Safir (1999)
November 9, Lexicon (An Interacting Subsystem in UG) Part-I Rajat Kumar Mohanty IIT Bombay.
Lecture 1: Trace Theory.  We have seen that things move :  Arguments move out of the VP into subject position  Wh-phrases move out of IP into CP 
Handling Unlike Coordinated Phrases in TAG by Mixing Syntactic Category and Grammatical Function Carlos A. Prolo Faculdade de Informática – PUCRS CELSUL,
Professor Ian Roberts having seen the two main types of rule systems (PS- rules/X’-theory and movement/transformational rules), we now.
LTAG Semantics for Questions Aleksandar Savkov. Contents Introduction Hamblin’s idea Karttunen’s upgrade Goals of the paper Scope properties of wh-phrases.
1 Principles & Parameters Approach in Linguistics II Bibhuti Bhusan Mahapatra.
Pauline Jacobson,  General introduction: compositionality, syntax/semantics interface, notation  The standard account  The variable-free account.
Coreferential Interpretations of Reflexives in Picture Noun Phrases: an Experimental Approach Micah Goldwater University of Texas at Austin Jeffrey T.
Roadmap Probabilistic CFGs –Handling ambiguity – more likely analyses –Adding probabilities Grammar Parsing: probabilistic CYK Learning probabilities:
Lecture 2: Categories and Subcategorisation
Embedded Clauses in TAG
CS 388: Natural Language Processing: Syntactic Parsing
Binding theory.
Tree-Adjoining Grammar (TAG)
Principles and Parameters (I)
Presentation transcript:

June 7th, 2008TAG+91 Binding Theory in LTAG Lucas Champollion University of Pennsylvania

June 7th, 2008 TAG+9 2 Overview Binding Theory (BT) and its local domains Previous work: Condition A This proposal: Conditions A, B, C Discussion

June 7th, 2008 TAG+9 3 Binding theory: A reminder Condition A: reflexives must be locally bound  John j thinks [ Bill b likes himself *j / b / *[other] ] Condition B: pronouns must be locally free  John j thinks [ Bill b likes him j / *b / [other] ] Condition C: full noun phrases must be free  *[ John j likes John j ]  *John j thinks [ Mary likes John j ]

June 7th, 2008 TAG+9 4 Binding theory in LTAG LTAG’s local domain = the verbal elementary tree and its arguments  (but not its adjuncts) Insight from previous work:  LTAG and BT have similar local domains This presentation’s central point:  Too many mismatches between local domains  We can’t reuse LTAG’s local domain for binding!

June 7th, 2008 TAG+9 5 Previous work reused LTAG’s local domain NP loves VNP VP S himself John NP VS* VP S thinks he Condition A

June 7th, 2008 TAG+9 6 Previous work reused LTAG’s local domain NP loves VNP VP S himself John NP VS* VP S thinks he Condition A

June 7th, 2008 TAG+9 7 Previous work reused LTAG’s local domain NP loves VNP VP S himself John NP VS* VP S thinks he Condition A

June 7th, 2008 TAG+9 8 Previous work reused LTAG’s local domain NP loves VNP VP S him John NP VS* VP S thinks he Condition B

June 7th, 2008 TAG+9 9 Ryant and Scheffler (2006) Only Condition A MCTAG set with a degenerate NP tree Tree-local MCTAG with flexible composition makes sure that antecedent and reflexive substitute into the same tree NP  loves VNP VP S himself NP i John NP NP* i {}

June 7th, 2008 TAG+9 10 Kallmeyer and Romero (2007) Only Condition A MCTAG set with a degenerate VP tree Tree-local MCTAG with flexible composition makes sure that antecedent and reflexive substitute into the same tree NP  loves VNP VP S himself NP i John NP VP* i {} (some features omitted)

June 7th, 2008 TAG+9 11 Kallmeyer and Romero’s claim “Tree-local MCTAG display exactly the extended domain of locality needed to account for the locality of anaphora binding in a natural way.” -- Kallmeyer and Romero (2007)

June 7th, 2008 TAG+9 12 A counterexample VP* opposite PNP PP VP himself John NP V VP S imagined Cannot be handled by Kallmeyer and Romero (2007)  except by flexible composition (which they try to avoid) Bill

June 7th, 2008 TAG+9 13 ECM: another mismatch of localities NP to love VNP VP S Bill John NP VS* VP S expects him Can be handled with an extra feature No lexical ambiguity needed (unlike R&S 2006)

June 7th, 2008 TAG+9 14 Mismatches within Binding Theory Bill NP himself NP  V VP S found John NP NP* ’s Det NP  N’ NP picture N P NP* PP NP of Bill NP him NP  V VP S found John NP NP* ’s Det NP  N’ NP picture N P NP* PP NP of Judgments tested experimentally (Keller and Asudeh ‘01; Runner ‘03) A A B B

June 7th, 2008 TAG+9 15 Mismatches within Binding Theory VP* near P PP VP himself John NP V VP S saw a snake NP VP* near P PP VP him John NP V VP S saw a snake NP A A B B

June 7th, 2008 TAG+9 16 How to encode the other conditions? Condition A roughly corresponds to tree- locality Condition B = “enforced non-locality”? Condition C = ???  Need to propagate an unbounded number of potential antecedents

June 7th, 2008 TAG+9 17 This account in a nutshell Every NP receives three items from its environment:  a list “A” of local potential antecedents  a list “B” of local potential antecedents  a list “C” of nonlocal potential antecedents Every NP supplies its own individual variable to its environment The rest of the grammar is responsible for providing the correct lists to the NP substitution slots

June 7th, 2008 TAG+9 18 Technical innovation: List-valued features

June 7th, 2008 TAG+9 19 Elementary tree for “himself” (Condition A, simplified) “A reflexive must be locally bound.”

June 7th, 2008 TAG+9 20 Elementary tree for “he” (Condition B) “A pronoun must be locally free.”

June 7th, 2008 TAG+9 21 Elementary tree for “John” (Condition C) “A full noun phrase must be free.”

June 7th, 2008 TAG+9 22 Sample derivation

June 7th, 2008 TAG+9 23 Sample derivation

June 7th, 2008 TAG+9 24 Sample derivation

June 7th, 2008 TAG+9 25 Sample derivation

June 7th, 2008 TAG+9 26 Condition C: the default case Before...

June 7th, 2008 TAG+9 27 Condition C: the default case...and after unification of top/bottom features

June 7th, 2008 TAG+9 28 Condition C across clauses Before putting the trees together...

June 7th, 2008 TAG+9 29 Condition C across clauses The higher tree passes its subject down, then...

June 7th, 2008 TAG+9 30 Condition C across clauses...unification at the root node propagates the empty list

June 7th, 2008 TAG+9 31 Improvements over previous accounts...

June 7th, 2008 TAG+9 32 Binding into adjuncts Just propagate everything!

June 7th, 2008 TAG+9 33 Mismatches between domains easily encoded Non-complementary binding conditions easily handled with separate A and B list features No ad hoc trees needed for picture NPs (unlike K&R ‘07)

June 7th, 2008 TAG+9 34 C-command violations easily encoded No need for separate lexical entry Just extrapose subject NP along with its feature structure (he) (Himself) e.g. extraposition: “Himself i, he i likes.”

June 7th, 2008 TAG+9 35 Improvements at a glance All conditions are implemented Higher empirical accuracy No lexical ambiguity No flexible composition (K&R 2007) No syntactically unmotivated degenerate trees (Kallmeyer and Romero, 2008) Better integration with anaphora resolution (Branco, 2002) No explicit representation of c-command

June 7th, 2008 TAG+9 36 Issues / Future work Unknown complexity of list-valued features  Just a decoration on the trees though -- they do not rule out any sentences Lack of predictive power  How do we constrain possible feature values?  Metagrammar? Does TAG offer any insights into BT at all?

June 7th, 2008 TAG+9 37 Thank you. Lucas Champollion University of Pennsylvania

June 7th, 2008 TAG+9 38 Previous accounts do not interface well with anaphora resolution modules Previous accounts: parser delivers a forest of indexed trees  John i introduced Bill k to himself i vs. John i introduced Bill k to himself k  Problem: Anaphora resolution modules are not prepared to compare entire trees (Branco, 2002) Our solution outputs a compact set of constraints  Following Branco (2002)

June 7th, 2008 TAG+9 39 The grammar of picture NPs NP* ’s Det NP  N’ NP Bill NP himself NP  V VP S found picture N P NP* PP NP of John NP

June 7th, 2008 TAG+9 40 Missing link problem NP* ’s Det NP  N’ NP Bill NP himself NP  V VP S found picture N P NP* PP NP of John NP