BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Relational-Realizational Syntax: An Architecture for Describing an
 d Parsing Rich Morphosyntactic Descriptions - Reut Tsarfaty - Uppsala Univ
 ersity
DTSTART:20110311T120000Z
DTEND:20110311T130000Z
UID:TALK29967@talks.cam.ac.uk
CONTACT:Thomas Lippincott
DESCRIPTION:Precision grammars and treebank grammars present two alternati
 ves for obtaining an accurate\, consistent and maximally complete syntacti
 c analysis of natural language sentences. For a long time these two resear
 ch endeavors have been conducted in separate communities and optimized for
  disparate goals -- the former for rich and accurate descriptions of lingu
 istic structures\, and the latter for efficient and accurate statistical p
 arsing\,. Recently\, these disparate research efforts started to acknowled
 ge their usefulness for one another by borrowing terms\, theoretical const
 ructs and techniques from one research endeavor to the other. In this talk
  I take a step back to consider the morpho-syntactic analysis task from fi
 rst principles and develop a novel architecture which remains faithful to 
 both kinds of goals.\n\nIn this talk I present a novel architecture for sp
 ecifying rich morphosyntactic representations and learning the associated 
 grammars from annotated data. The key idea underlying the architecture is 
 the application of the traditional notion of a “paradigm” to the synta
 ctic domain. N-place predicates associated with paradigm cells are viewed 
 as relational networks that are realized recursively by combining and orde
 ring cells from other paradigms. The function of paradigm cells is mapped 
 to forms in a recursive fashion\, be means of realization rules that make 
 reference both to the morphological and to the syntactic domains. This arc
 hitecture\, called Relational-Realizational\, has a simple instantiation a
 s a generative probabilistic model of which parameters can be statisticall
 y learned from treebank data\, and which can be used for efficient parsing
 .\n\nAn application of the model to Hebrew and Swedish allows for accurate
  description of word-order and argument marking patterns of the different 
 language types. The associated treebank grammar can be used for statistica
 l parsing and is shown to improve state-of-the-art parsing results for the
  Semitic language Modern Hebrew. The availability of a simple\, formal\, r
 obust\, implementable and statistically interpretable working model opens 
 new horizons in computational linguistics — at least in principle\, we s
 hould now be able to quantify typological trends which have so far been st
 ated informally or only tacitly reflected in corpus statistics.
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR