Introduction to Lexical Functional Grammar. Wellformedness conditions on f- structures. Constraints on f-structures

Similar documents
Grammar Development with LFG and XLE. Miriam Butt University of Konstanz

Grammar Development with LFG and XLE. Miriam Butt University of Konstanz

Lexical entries & clauses

Unification in Unification-based Grammar

Clausal Architecture and Verb Movement

AUTOMATIC LFG GENERATION

Computational Linguistics: Feature Agreement

The uniform treatment of constraints, coherency and completeness in a Lexical Functional Grammar compiler

Elementary Operations, Clausal Architecture, and Verb Movement

Wh-questions. Ling 567 May 9, 2017

Dependency grammar and dependency parsing

Getting Started With Syntax October 15, 2015

Context-Free Grammars

The anatomy of a syntax paper

Indefinites and Sluicing. A type logical approach

Context-Free Grammars

Supervised Models for Coreference Resolution [Rahman & Ng, EMNLP09] Running Example. Mention Pair Model. Mention Pair Example

Syntax 380L December 4, Wh-Movement 2. For notational convenience, I have used traces (t i,t j etc.) to indicate copies throughout this handout.

Context-Free Grammars. Carl Pollard Ohio State University. Linguistics 680 Formal Foundations Tuesday, November 10, 2009

Goals of the talk: of linking theory into the projection architecture of Integration LFG Choice of correct linkings by preference rankings 2

A high-level approach to language description

Proseminar on Semantic Theory Fall 2013 Ling 720 An Algebraic Perspective on the Syntax of First Order Logic (Without Quantification) 1

An Efficient Implementation of PATR for Categorial Unification Grammar

Lecture 5. Logic I. Statement Logic

Dependency grammar and dependency parsing

Dependency grammar and dependency parsing

Syntax and Grammars 1 / 21

Grammar Knowledge Transfer for Building RMRSs over Dependency Parses in Bulgarian

CSE 20 DISCRETE MATH. Fall

The Verb. From Probabilities to Internal Categories. Cem Bozşahin. Cognitive Science Department, The Informatics Institute, ODTÜ

L322 Syntax. Chapter 3: Structural Relations. Linguistics 322 D E F G H. Another representation is in the form of labelled brackets:

1 Introduction. 2 Set-Theory Formalisms. Formal Semantics -W2: Limitations of a Set-Theoretic Model SWU LI713 Meagan Louie August 2015

CS 261 Fall Mike Lam, Professor Integer Encodings

SynAF, ISO NWI. Thierry Declerck DFKI

Compilers and Interpreters

COMPUTATIONAL SEMANTICS WITH FUNCTIONAL PROGRAMMING JAN VAN EIJCK AND CHRISTINA UNGER. lg Cambridge UNIVERSITY PRESS

Attempto Controlled English: Language, Tools and Applications. Getting Started

Dependency and (R)MRS

Einführung in die Computerlinguistik

MINIMALIST INQUIRIES AND DERIVATION BY PHASE MAIN POINTS

LING 510, Lab 3 September 23, 2013

LING 253: SYNTAX I SECTION. (Specifier Rule) (Adjunct Rule) (Complement Rule)

Topics in Parsing: Context and Markovization; Dependency Parsing. COMP-599 Oct 17, 2016

1 Ellipsis resolution as a semantic relation between source and target. Dalrymple et al. (1991), Shieber et al. (1996): The equational analysis

COMP 181 Compilers. Administrative. Last time. Prelude. Compilation strategy. Translation strategy. Lecture 2 Overview

2 Ambiguity in Analyses of Idiomatic Phrases

Ling/CSE 472: Introduction to Computational Linguistics. 5/9/17 Feature structures and unification

English Understanding: From Annotations to AMRs

Appendix 1. Description Logic Terminology

Computational Linguistics (INF2820 TFSs)

Appendix 1. Description Logic Terminology

User-Defined Nonmonotonicity in Unification-Based Formalisms

A Logical Approach to Structure Sharing in TAGs

Introduction to Semantics. Expanding Our Formalism, Part 2 1

Propositional Logic. Part I

LDLS Syntax Crash Course

Lecture 2 Finite Automata

CS152 Programming Language Paradigms Prof. Tom Austin, Fall Syntax & Semantics, and Language Design Criteria

A Minimal GB Parser. Marwan Shaban. October 26, Boston University

Intro to Haskell Notes: Part 5

More Theories, Formal semantics

Programming Language Concepts, cs2104 Lecture 04 ( )

14 What VP Ellipsis Can Do, and What it Can t, but not Why

Optional Arguments in Abstract Categorial Grammar

Ling 571: Deep Processing for Natural Language Processing

Chapter 3. Describing Syntax and Semantics

A Method for. Disjunctive Constraint Satisfaction. Xerox Palo Alto Research Center

1. He considers himself to be a genius. 2. He considered dieting to be unnecessary. 3. She considered that the waffle iron was broken. 4.

1 Informal Motivation

Sustainability of Text-Technological Resources

Semantic Pattern Classification

CSE 20 DISCRETE MATH. Winter

The CKY Parsing Algorithm and PCFGs. COMP-550 Oct 12, 2017

structure of the presentation Frame Semantics knowledge-representation in larger-scale structures the concept of frame

Specifying Syntax. An English Grammar. Components of a Grammar. Language Specification. Types of Grammars. 1. Terminal symbols or terminals, Σ

564 Lecture 21 Nov. 4, 1999

Chapter 3: Describing Syntax and Semantics. Introduction Formal methods of describing syntax (BNF)

The CKY algorithm part 1: Recognition

CMSC 201 Fall 2016 Lab 09 Advanced Debugging

1. Logic as a subject matter is the study of logics, understood as models of acceptable inference patterns

Computational Linguistics: Syntax-Semantics Interface

Hidden Markov Models. Natural Language Processing: Jordan Boyd-Graber. University of Colorado Boulder LECTURE 20. Adapted from material by Ray Mooney

F08: Intro to Composition

Compilers: Principles, Techniques, and Tools. Jing-Shin Chang Department of Computer Science & Information Engineering National Chi-Nan University

SURVEY PAPER ON WEB PAGE CONTENT VISUALIZATION

CSE450 Translation of Programming Languages. Lecture 4: Syntax Analysis

Compilers: Principles, Techniques, and Tools. Jing-Shin Chang Department of Computer Science & Information Engineering National Chi-Nan University

CAS LX 522 Syntax I. Case. [ucase:acc] [ucase:nom] Pronouns. NPs need case

Semantics and Generative Grammar. Expanding Our Formalism, Part 2 1. These more complex functions are very awkward to write in our current notation

Computational Linguistics: Syntax-Semantics

COSC252: Programming Languages: Semantic Specification. Jeremy Bolton, PhD Adjunct Professor

Evaluation of Predicate Calculus By Arve Meisingset, retired research scientist from Telenor Research Oslo Norway

Defining Program Syntax. Chapter Two Modern Programming Languages, 2nd ed. 1

Homework: Simple functions and conditionals

Lecture 1 Contracts : Principles of Imperative Computation (Fall 2018) Frank Pfenning

Chapter 4 Fuzzy Logic

Syntax-semantics interface and the non-trivial computation of meaning

COMP4418 Knowledge Representation and Reasoning

Question Answering Using XML-Tagged Documents

The CKY algorithm part 2: Probabilistic parsing

Transcription:

Introduction to Lexical Functional Grammar Session 8 f(unctional)-structure & c-structure/f-structure Mapping II & Wrap-up Summary of last week s lecture LFG-specific grammar rules (i.e. PS-rules annotated with functional schemata) LFG-specific lexicon entries (i.e. annotated with functional schemata) Mapping between f-description (from c-structure) and f-structure Constraints on f- In LFG, a sentence is grammatically correct only if it satisfies 2 criteria The grammar must be able to assign a correct c- structure The grammar must be able to assign a correct wellformed f-structure Moreover the f-structure must match the c- structure and it has to satisfy 3 formal constraints Uniqueness (or Consistency) Condition Coherence Condition Completeness Condition Uniqueness Condition (also called Consistency Condition) Attribute names must be unique and every attribute has a unique value Ex. A clause cannot have two different SUBJs or two different tenses: Inconsistent f-structure SUBJ PRED Veit` SUBJ PRED Tom` PRED TENSE TENSE sleep ( SUBJ) ` PAST FUT Completeness Condition All functions specified in the value of a PRED feature must be present in the local f-structure. Complete f-structure PRED like ( SUBJ) ( OBJ) ` OBJ PRED Mary` Incomplete f-structure PRED like ( SUBJ) ( OBJ) `? Coherence Condition All argument functions in an f-structure must be selected by the local PRED feature. Complete f-structure PRED fall ( SUBJ) ` Incoherent f-structure PRED fall ( SUBJ) `? OBJ PRED Mary` 1

Extended Coherence Condition All argument functions in an f-structure must be selected by the local PRED feature. Non-argument functions have to bear an appropriate relation to a PRED: ADJUNCTS must be in f- containing a PRED TOPIC or FOCUS must be identified with, or anaphorically linked to, arguments or adjuncts. Functional equations can be devided into 2 groups Defining equations: The functional equations we have seen so far. They define an f-structure attribute as existing and having a particular value. : do not create a feature-value pair, but require a particular feature value to be present They are notationally distinguished by subscripting the letter c to the equal sign ` = c Ex. bewundern V ( PRED)= bewundern ( SUBJ) ( OBJ) ` ( OBJ CASE)= c ACC The constraining equation does not provide information directly, but constrains (i.e. checks the appropriateness of) information coming from somewhere else The constraining equation will be violated either if the feature is not specified at all somewhere else, or if it is specified differently from what the pronoun (in our example) requires Inequality (negation) It constrains a feature not to have a particular value Represented either ( TENSE) = PRES or ( TENSE) PRES Note: there is no way to interpret the inequality as defining. Therefore, the c subscript is not used in the equations. Existential constraint It requires a feature to be present without requiring that it have a particular value The notation: (f a) Ex. The complementizer that requires its clause to have a finite verb, i.e. a verb with the attribute TENSE. The value of TENSE, however, is irrelevant. Conversely, the complementizer to cannot occur with the feature TENSE: Conditional equations Useful in testing for the presence of a feature as a condition for functional specification Represented by the conditional ` Ex. Malayalam determines the functions of NPs on the basis of their case attributes (K.P. Mohanan, 1982). A nominative case-marked NP can be either a subject or (if inanimated) an object; animate objects are accusative: that to ( TENSE) ( TENSE) a) ( CASE) = NOM ( SUBJ) = b) [( CASE) = ACC & ( ANIM) = +] ( OBJ) = c) [( CASE) = NOM & ( ANIM) = ] ( OBJ) = 2

Summary of types of constraining equations a) A = c B positive b) (A = B), A B negative c) (f a) existential d) A B conditional (also defining) Designators like ( SUBJ) ` or ( SUBJ NUM) `areoutside-in: They define a path into the f- structure from a point specified by Inside-out impose a constraint on the larger f-structure. They define, starting at, a path outward past a particular function As notation, the is put at the end Ex. (DF ) Primarily used in the analyses of long-distance dependencies ( wh-movement ) and anaphora Example for inside-out designator What is a wh-word, but in contrast to other wh-words it can only introduce questions in standard English but no relative clauses a) [Who did you see]? the syntactician [who you saw] b) [Where did you read that]? the newspaper [where you read that] c) [What did you eat]? *the falafel [what you ate] Assuming the feature TYPE for clause type with values such as Q, REL, etc. The f-structure in which what appears (DF is some discourse function): TYPE DF Q PRED PRON PRO` WH The lexical item what must lexically specify that the clause in which it occurs has the feature [TYPE Q]: what D ( PRED) = PRO` ((DF ) TYPE = Q ( PRON) = WH Functional uncertainty For example ( SUBJ) or ( SUBJ NUM), etc. represent a fixed path of attributes For some phenomena, e.g. such as anaphora and longdistance dependencies, however, the path cannot be specified in advance. It might be potentially infinite. Functional uncertainty extends functional equations to allow a path of attributes of variable length LFG schemata can contain the metavariables and in addition to and. We will see examples of this in the next sessions f-structure: more examples Expletive pronouns Ex. a) It snows. b) It seems that we are having fun. c) I take it that the world is flat. These are meaningless elements, i.e. they do not refer to a discourse referent, they have no θ-role, but they have a syntactic function Since they are meaningless, they lack the PRED feature; instead they have the FORM feature Since they are non-thematic arguments, they are placed outside the angle brackets 3

f-structure: more examples The f-structure for It snows. PRED snow ( SUBJ) ` SUBJ FORM it non-thematic arguments outside the angle brackets no PRED feature, FORM feature instead The most basic grammatical functions are argument functions: SUBJ (subject), OBJ (object), OBJ2 (secondary object), OBL θ (oblique) family POSS (possessor), COMP (complement) The PRED for seem PRED seem < ( COMP) > ( SUBJ) ` There are also non-argument functions: - ADJ (adjunct) - FOCUS, TOPIC Further distinction among argument functions: Core functions: SUBJ, OBJ, OBJ2 Are typically realized as DPs (e.g. English) or nominative and accusative Case (in languages with morphological Case) Are more strictly grammatical functions Non-core functions: OBL θ family Are typically marked with prepositions or Cases expressing θ- roles Are more tied to semantics Argument functions are arranged in a relational hierarchy (Keenan & Comrie 1977) SUBJ > OBJ > OBJ2 > OBL θ (indicates relative accessibility to grammatical processes such as relativization, antecedence of anaphora, etc.) Overlay functions relate a syntactic element to a larger syntactic or discourse structure SUBJ (partially) connects clauses within a sentence TOPIC and FOCUS relate a sentence to the larger discourse (grammaticized) discourse functions (Bresnan 2001) Non-overlay argument functions are also called complement functions argument core noncore SUBJ OBJ OBL θ ADJ OBJ2 COMP POSS overlay nonoverlay nonargument FOCUS TOPIC etc. overlay 4

Wrap-up See separate file Wrap-up on our web-site Homework This is a bibliographic work meant as preparation for the sessions in the new year (where you are supposed to write LFG grammar fragments for a particular phenomenon in your (or another) language) Find out until 09 Monday (9 am) which linguistic phenomena in your language seem to be interesting for linguists working in LFG (consult the websites at Stanford and Essex) Write a short summary of no more than half a page (in prose!) and indicate already, if possible, which of the phenomena might interest you 5