May 12, 2008
Chapter 11
Feature structures, Unification
1
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Guidelines and tips
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
Feature Structures and Unification in Linguistic Theory: A Study of Pizza and Grammar, Study notes of Linguistics
The use of feature structures and unification in linguistic theory through the example of pizza. It covers the earley parser, problems with context-free grammars (cfg), and the solution of replacing atomic node labels with feature structures. The document also introduces some grammatical theories that use feature structures and unification, such as gpsg, hpsg, lfg, and construction grammar. It provides a pizza type hierarchy and explains how to represent pizza descriptions and models using unification.
Typology: Study notes
Pre 2010
1 / 56
This page cannot be seen from the preview
Don't miss anything!
Related documents
Partial preview of the text
Download Feature Structures and Unification in Linguistic Theory: A Study of Pizza and Grammar and more Study notes Linguistics in PDF only on Docsity!
May 12, 2008
Chapter 11
Feature structures, Unification
1
Overview
Leftovers: Earley parser
Problems with CFG
One solution: feature structure and unification
An aside on modeling language
Practice with unification (pizza)
Concepts relating to feature structures
Feature structures in the grammar
Using types to capture linguistic generalizations
2
The Earley Chart Parser, Features
One pass through the chart/sentence
Recognizer, which can be modified into a parser
Top-down, uni-directional, exhaustive (as parser)
4
Discussion: How does the Earley algorithm solve
these problems?
Ambiguity
Inefficient reparsing of subtrees
Infinite loops with left-recursive grammars
NP
NP PP
The cat on the mat by the door slept
5
Problems with CFG
Potentially arbitrary rules: D
S VP
return to this)Gets clunky quickly with cross-cutting properties (we’ll
(Not quite powerful enough for natural languages)
Solution: Replace atomic node labels with feature structures.
7
Some grammatical theories which explicitly use
feature structures and unification
GPSG: Generalized Phrase Structure Grammar
http://hpsg.stanford.edu/HPSG: Head-driven Phrase Structure Grammar
http://www-lfg.stanford.edu/lfg/LFG: Lexical Functional Grammar
http://www.constructiongrammar.org/Construction Grammar
Combinatory Categorial Grammar
8
Modeling language (HPSG)
Three things involved in modeling:
Real-world entities
Models
Descriptions of models
10
Feature Structure Descriptions
FEATURE
1
VALUE
1
FEATURE
2
VALUE
2
FEATURE
n
VALUE
n
11
TYPE
FEATURES/VALUES
IST
pizza pizza-thing
CRUST
thick, thin, stuffed
TOPPINGS
topping-set
pizza-thing
topping-set
OLIVES
ONIONS
MUSHROOMS
pizza-thing
vegetarian
topping-set
non-vegetarian
SAUSAGE
PEPPERONI
HAM
topping-set
13
Type Hierarchies
A type hierarchy...
types).... states what kinds of objects we claim exist (the
shared properties (the IST relations).... organizes the objects hierarchically into classes with
(the feature and feature value declarations).... states what general properties each kind of object has
14
Pizza Descriptions and Pizza Models
CRUST pizza
thick
TOPPINGS
OLIVES vegetarian
ONIONS
CRUST , thick
TOPPINGS ,
OLIVES ,
ONIONS ,
MUSHROOMS ,
CRUST , thick
TOPPINGS ,
OLIVES ,
ONIONS ,
MUSHROOMS ,
16
Pizza Descriptions and Pizza Models
CRUST pizza
thick
TOPPINGS
OLIVES vegetarian
ONIONS
‘type’/‘token’ distinction – applies to sentences as wellcorrespond to? How many pizzas-in-the-world do the pizza models
17
Unification
CRUST^ pizza
thick
TOPPINGS
OLIVES
ONIONS
HAM
19
Unification
CRUST pizza
thick
TOPPINGS
OLIVES
HAM
CRUST^ pizza
thin
TOPPINGS
OLIVES
ONIONS
20