P ARSING WITH P ARADIGMS A Relational-Realizational Architecture for Specifying and Learning Morphosyntactic Descriptions Reut Tsarfaty The Department of Linguistics and Philology Uppsala University QUANTITATIVE MEASURES IN MORPHOLOGY AND MORPHOLOGICAL DEVELOPMENT University of California, San Diego January 15-16, 2011
70
Embed
Parsing with Paradigms 0.05inA Relational-Realizational ... 2011/QMMMDSlidesV2.pdf · PARSING WITH PARADIGMS A Relational-Realizational Architecture for Specifying and Learning Morphosyntactic
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
PARSING WITH PARADIGMSA Relational-Realizational Architecture for Specifying
and Learning Morphosyntactic Descriptions
Reut TsarfatyThe Department of Linguistics and Philology
Uppsala University
QUANTITATIVE MEASURES IN MORPHOLOGY AND MORPHOLOGICAL DEVELOPMENTUniversity of California, San Diego
January 15-16, 2011
Morphology and Syntax
“Morphology deviates in a number of important waysfrom the classical picture of word structure as simplythe combinatory syntax of morphemes. [As we haveseen,] morphology is best seen as a system thatdescribes relations among word structural types interms of the way the forms of words realize theproperties that compose their content.”
“In fact [...] much of what we normally think of asclearly part of syntax seems to have some of thissame character. [...] Rather than being exclusivelymatters of the construction and manipulation ofhierarchical constituent structure, a number of areasusually considered syntactic in character also turn outto be realizational, relational, and governed by asystem of constraints rather than (solely) by rules ofX-structure, displacement, and other manipulations ofphrasal structure.” (Anderson 2004)
Morphology and Syntax“Morphology deviates in a number of important waysfrom the classical picture of word structure as simplythe combinatory syntax of morphemes. [As we haveseen,] morphology is best seen as a system thatdescribes relations among word structural types interms of the way the forms of words realize theproperties that compose their content.”
“In fact [...] much of what we normally think of asclearly part of syntax seems to have some of thissame character. [...] Rather than being exclusivelymatters of the construction and manipulation ofhierarchical constituent structure, a number of areasusually considered syntactic in character also turn outto be realizational, relational, and governed by asystem of constraints rather than (solely) by rules ofX-structure, displacement, and other manipulations ofphrasal structure.” (Anderson 2004)
My Contribution
The IdeaApplying the principles underlying W&P models to syntax
The ProposalA Relational-Realizational (RR) modeling architecture
The OutcomeI Useful: Parsing less-configurational languagesI Interesting: Quantifying typological parameters
The Plan for Today
The Task:Statistical Parsing
The Challenge:Complex Form-Function Correspondence
The Method:Following the footsteps of Morphology
The Proposal:A Relational-Realizational Approach
⇒ A Stepping StoneTowards computational typology and statistical UG
Part 1: The Task
Statistical Parsing
Statistical Parsing
”This is easy”
Statistical Parsing
”This is easy”
Statistical Parsing
S
NP-SBJ
PRP
“This”
VP-PRD
VB
“is”
ADJP
ADJ
“easy”
Statistical Parsing
S
NP-SBJNP-SBJ
PRPNNP
“This”
VP-PRDVP-PRD
VBVB
“is”
ADJPADJP
ADJADJ
“easy”
Constituency-Based Supervised Statistical Parsing
S
NP-SBJ
PRP
“This”
VP-PRD
VB
“is”
ADJP
ADJ
“easy”
Model Study F-Score
Treebank Charniak 75Grammar 1996
Head- Collins 88.6Driven 1997
Discriminative Collins 89.7Reranking 2000
Discriminative- Johnson &Reranking Charniak 2005 91.0
Self- McClosky 92.1Training 2006
State- Petrov et al 90.1Splits 2007
Forest Liang Huang 91.7Reranking 2008
Constituency-Based Supervised Statistical Parsing
And what about this?
And this?
And this?
And? ...
Language Parser F-Score
Rafferty & 79.2German Manning 2008
Collins 79.3Czech et al. 1999
Levy & 78.8Chinese Manning 2003
Maamouri, Bies & 78.1Arabic Kulick 2008
Tsarfaty & 74.4Hebrew Sima’an 2007
So What Is Going On?
Often ConsideredI Corpora Size
E.g., For Chinese (Bikel & Chiang 2000)
I Annotation IdiosyncrasiesE.g., For Arabic (Maamouri, Bies & Kulick 2008, 2009)
I Evaluation MattersE.g., For German (Rehiben & van Genabith 2007, Kubler 2008)
Not so often..I Parsers’ ParameterizationI Language Variation
So What Is Going On?
Often ConsideredI Corpora Size
E.g., For Chinese (Bikel & Chiang 2000)
I Annotation IdiosyncrasiesE.g., For Arabic (Maamouri, Bies & Kulick 2008, 2009)
I Evaluation MattersE.g., For German (Rehiben & van Genabith 2007, Kubler 2008)
Not so often..I Parsers’ ParameterizationI Language Variation
We presented a Relational-Realizational Architecture forSpecifying and Learning Morphosyntactic DescriptionsI SimpleI FormalI RobustI ImplementableI InterpretableI Explanatory⇒ Paradigms augmented with realization rules constitute auseful and powerful modeling strategy also for (Morpho)Syntax.
Thank You!
Questions?
For more InformationRelational-Realizational ParsingReut Tsarfaty, University of AmsterdamPhD Thesis, 2010