Logical Aspects of Computational Linguistics: 7th International Conference, LACL 2012, Nantes, France, July 2-4, 2012. Proceedings PDF
Preview Logical Aspects of Computational Linguistics: 7th International Conference, LACL 2012, Nantes, France, July 2-4, 2012. Proceedings
Lecture Notes in Computer Science 7351 CommencedPublicationin1973 FoundingandFormerSeriesEditors: GerhardGoos,JurisHartmanis,andJanvanLeeuwen EditorialBoard DavidHutchison,UK TakeoKanade,USA JosefKittler,UK JonM.Kleinberg,USA AlfredKobsa,USA FriedemannMattern,Switzerland JohnC.Mitchell,USA MoniNaor,Israel OscarNierstrasz,Switzerland C.PanduRangan,India BernhardSteffen,Germany MadhuSudan,USA DemetriTerzopoulos,USA DougTygar,USA GerhardWeikum,Germany FoLLI Publications on Logic, Language and Information SublineofLecturesNotesinComputerScience SublineEditors-in-Chief ValentinGoranko,TechnicalUniversity,Lynbgy,Denmark ErichGrädel,RWTHAachenUniversity,Germany MichaelMoortgat,UtrechtUniversity,TheNetherlands SublineAreaEditors NickBezhanishvili,ImperialCollegeLondon,UK AnujDawar,UniversityofCambridge,UK PhilippedeGroote,Inria-Lorraine,Nancy,France GerhardJäger,UniversityofTübingen,Germany FenrongLiu,TsinghuaUniversity,Beijing,China EricPacuit,TilburgUniversity,TheNetherlands RuydeQueiroz,UniversidadeFederaldePernambuco,Brazil RamRamanujam,InstituteofMathematicalSciences,Chennai,India Denis Béchet Alexander Dikovsky (Eds.) Logical Aspects of Computational Linguistics 7th International Conference, LACL 2012 Nantes, France, July 2-4, 2012 Proceedings 1 3 VolumeEditors DenisBéchet AlexanderDikovsky UniversitédeNantes,LINAUMR6241 2,ruedelaHoussinière,BP92208,44322NantesCedex3,France E-mail:{denis.bechet;alexandre.dikovsky}@univ-nantes.fr ISSN0302-9743 e-ISSN1611-3349 ISBN978-3-642-31261-8 e-ISBN978-3-642-31262-5 DOI10.1007/978-3-642-31262-5 SpringerHeidelbergDordrechtLondonNewYork LibraryofCongressControlNumber:2012939839 CRSubjectClassification(1998):I.2,F.4.1,F.3 LNCSSublibrary:SL1–TheoreticalComputerScienceandGeneralIssues ©Springer-VerlagBerlinHeidelberg2012 Thisworkissubjecttocopyright.Allrightsarereserved,whetherthewholeorpartofthematerialis concerned,specificallytherightsoftranslation,reprinting,re-useofillustrations,recitation,broadcasting, reproductiononmicrofilmsorinanyotherway,andstorageindatabanks.Duplicationofthispublication orpartsthereofispermittedonlyundertheprovisionsoftheGermanCopyrightLawofSeptember9,1965, initscurrentversion,andpermissionforusemustalwaysbeobtainedfromSpringer.Violationsareliable toprosecutionundertheGermanCopyrightLaw. Theuseofgeneraldescriptivenames,registerednames,trademarks,etc.inthispublicationdoesnotimply, evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevantprotectivelaws andregulationsandthereforefreeforgeneraluse. Typesetting:Camera-readybyauthor,dataconversionbyScientificPublishingServices,Chennai,India Printedonacid-freepaper SpringerispartofSpringerScience+BusinessMedia(www.springer.com) Preface Welcome to the proceedings of the 7th International Conference on Logical AspectsofComputationalLinguistics,whichwasheldJuly2–4,2012,inNantes, France. The aim of LACL conferences is to bring together active researchers interested in all aspects concerning the use of logics in computational linguis- tics to discuss current research,new results, problems, and applications of both theoretical and practical nature. LACL 2012 focused on its traditional topics: – Typelogicalgrammars(LambekGrammars,AbstractCategorialGrammars, Combinatorial Categorial Grammars, Categorial Dependency Grammars) and other formal grammars closely related to them (Minimalist Grammars, extended TAG and other weakly context-sensitive grammars) – Formalsemanticsofnaturallanguage(type andprooftheoreticalsemantics, intensionalmodeltheoreticsemantics,dynamicsemantics,lexicalsemantics) – Logical models of discourse and dialogue (game theoretic models, ludics) Amongst24thoroughlytriplyrefereedsubmittedpapers,the ProgramCom- mittee consisting of 33 colleagues listed here selected 15 high-quality contri- butions by authors from China, France, Germany, Italy, Japan, Poland, Rus- sia, UK and USA. LACL 2012 included two invited talks: by M. Steedman (University of Edinburg) and A. Clark (Royal Holloway University of London) and a tutorial by C. Fouquer´e and M. Quatrini (University Paris 13 and Uni- versity of Aix Marseille II). Besides this, the technical program also included a System Demonstration session. We would like to thank all authors who submitted papers, the four invited speakers and all conference participants. We are grateful to the members of the Program Committee and of the Demo Session Committee for their thorough efforts in reviewing and discussing submitted contributions with expertise and carefulness. We are also grateful to our institutional sponsors and supporters: the Association for Logic, Language and Information (FoLLI), CNRS, Ecole des Mines of Nantes, Laboratory of Informatics of Nantes (LINA), University of Nantes and its Faculty of Sciences and Technologies. We would also like to expressourgratitudetotheOrganizingCommitteeandallthepeoplewhomade this meeting possible. April 2012 Denis B´echet Alexander Dikovsky Organization Program Committee Michele Abrusci Universit`a di Roma Tre, Italy Nicholas Asher IRIT/CNRS, France Raffaella Bernardi University of Trento, Italy Philippe Blache LPL/CNRS, Aix-en-Provence,France Wojciech Buszkowski Poznan University, Poland Denis B´echet LINA - University of Nantes, France Phillipe De Groote LORIA/INRIA, Nancy Grand Est, France Michael Dekhtyar University of Tver, Russia Alexander Dikovsky LINA CNRS UMR 6241,Universit´ede Nantes, France Markus Egg Humboldt Universita¨t Berlin, Germany Annie Foret University of Rennes 1, France Nissim Francez Technion, Haifa, Israel Makoto Kanazawa NII, Japan Gregory M. Kobele University of Chicago, USA Marcus Kracht University of Bielefeld, Germany Alain Lecomte Universit´e Paris 8, France Hans Leiss Universit¨at Mu¨nchen, Germany Michael Moortgat Universiteit Utrecht, UiL OTS, The Netherlands Richard Moot LaBRI/CNRS, France Glyn Morrill Universitat Polit`ecnica de Catalunya, Barcelona, Spain Reinhard Muskens Tilburg University, The Netherlands Uwe M¨onnich Universita¨t Tu¨bingen, Germany Gerald Penn University of Toronto, Canada Mati Pentus Moscow State University, Russia Sylvain Pogodalla LORIA/INRIA, Nancy Grand Est, France Carl Pollard The Ohio State University, USA Anne Preller LIRMM/CNRS, France Christian Retor´e LaBRI/CNRS, France Sylvain Salvati INRIA, Bordeaux Sud-Ouest, France Chung-Chieh Shan Rutgers, The State University of New Jersey, USA Edward Stabler University of California, Los Angeles, USA Mark Steedman University of Edinburgh, UK Isabelle Tellier Universit´e Paris 3, France VIII Organization Demo Session Committee Denis B´echet LINA - University of Nantes Florian Boudin LINA - University of Nantes Alexander Dikovsky LINA - University of Nantes Nicolas Hernandez LINA - University of Nantes Organizing Committee Ramadan Alfared LINA - University of Nantes Denis B´echet LINA - University of Nantes Florian Boudin LINA - University of Nantes Alexander Dikovsky LINA - University of Nantes Anna Even LINA - University of Nantes Nicolas Hernandez LINA - University of Nantes Oph´elie Lacroix LINA - University of Nantes Annie Lardenois LINA Anne-Franc¸oise Quin LINA - CNRS Table of Contents Logical Grammars, Logical Theories................................ 1 Alexander Clark Ludics and Natural Language: First Approaches ..................... 21 Christophe Fouquer´e and Myriam Quatrini The Non Cooperative Basis of Implicatures.......................... 45 Nicholas Asher Movement-GeneralizedMinimalist Grammars........................ 58 Thomas Graf Toward the Formulation of Presupposition by Illative Combinatory Logic........................................................... 74 Yuri Ishishita and Daisuke Bekki Abstract Automata and a Normal Form for Categorial Dependency Grammars ...................................................... 86 Boris Karlov Importing MontagovianDynamics into Minimalism .................. 103 Gregory M. Kobele CoTAGs and ACGs .............................................. 119 Gregory M. Kobele and Jens Michaelis Gapping as Like-Category Coordination ............................ 135 Yusuke Kubota and Robert Levine L-Completeness of the Lambek Calculus with the Reversal Operation... 151 Stepan Kuznetsov Distributive Full Nonassociative Lambek Calculus with S4-Modalities Is Context-Free .................................................. 161 Zhe Lin Common Nouns as Types ......................................... 173 Zhaohui Luo Extractability as the Deduction Theorem in Subdirectional Combinatory Logic............................................... 186 Hiroko Ozaki and Daisuke Bekki X Table of Contents Agnostic Possible Worlds Semantics ................................ 201 Andrew Plummer and Carl Pollard Abstract Machines for Argumentation .............................. 213 Kurt Ranalter On the Completeness of Lambek Calculus with Respect to Cofinite Language Models ................................................ 229 Alexey Sorokin Dot-Types and Their Implementation .............................. 234 Tao Xue and Zhaohui Luo Author Index.................................................. 251 Logical Grammars, Logical Theories Alexander Clark Department of Computer Science RoyalHolloway, University of London Egham, TW20 0EX United Kingdom [email protected] Abstract. Residuatedlattices form oneofthetheoretical backbonesof the Lambek Calculus as the standard free models. They also appear in grammaticalinferenceasthesyntacticconceptlattice,analgebraicstruc- ture canonically defined for every language L based on the lattice of all distributionallydefinablesubsetsofstrings.Recentresultsshowthatitis possible to build representations, such as context-free grammars, based ontheselattices, andthattheserepresentationswill beefficientlylearn- able using distributional learning. In this paper we discuss the use of thesesyntacticconceptlatticesasmodelsofLambekgrammars,anduse thetools of algebraic logic totry tolink theproof theoretic ideas of the Lambekcalculuswiththemorealgebraicapproachtakeningrammatical inference. We can then reconceive grammars of various types as equa- tional theories of the syntacticconcept lattice of thelanguage. Wethen extend this naturally from models based on concatenation of strings, to ones based on concatenations of discontinuous strings, which takes us from context-free formalisms to mildly context sensitive formalisms (multiple context-freegrammars) and Morrill’s displacement calculus. 1 Introduction Logicis concernedwith proof;in the logicalgrammartraditionthe roleofproof is central and well understood [1]. But logic is also concerned with truth – and whatdoesitmeanforagrammar,suchasaLambekgrammar,tobetrue?What canitbetrueof?Thisisaproblemnotofprooftheorybutofmodeltheory,and whiletherehasbeenagreatdealofworkonmodelsforthe Lambekcalculus[2], thesearemodelsforthe calculus,notforthe grammars.Thesemodels arefree– the only statements that are true in these models are things that are true of all languages – but this seems to be inappropriate – there are things that are true of some languages and not of others; true about English but not about French or Dyirbal. A Lambek grammar consists only of a finite set of type assignments: (w,T) wherewisanonemptysequence,typicallyoflength1,andT isatype.Therefore the questionofthe truth of the grammarreduces to the question ofthe truth of the type assignments. The metatheory of Lambek grammar is explained in [3] and [4]. Several things are clear: first the types in the grammar are intended to D.B´echetandA.Dikovsky(Eds.):LACL2012,LNCS7351,pp.1–20,2012. (cid:2)c Springer-VerlagBerlinHeidelberg2012 2 A. Clark refertosetsofstringsinagivenlanguage,andthe compositetypesA•B, C/B, A\C and the associated calculus are motivated and justified by the algebraic properties of the associated operations on these sets. The following quotes from [4] lay out the metatheoretical assumptions. First Sets of strings of English words will be called (syntactic) types. Secondly, that the operations are ‘free’: Whenapplyingthisnotationtoanaturallanguage...wearethinking ofthe ...system...freely generatedbythe wordsofthe languageunder concatenation. Finally, thatthe startingpointforconstructingagrammaristo model anexist- ing language. Lambek does not conceive of this as a learning problem, but as a modelingprobleminthestructuralisttradition.Ratherthandefiningagrammar G in vacuo and then showing how that grammar defines a language L — the generative stance — he starts with a language L∗, and constructs a grammar G(L∗), — the descriptive approach— using a non-algorithmic method: One can continue playing this game until every English word has been assigned a finite number of types. One hopes that the grammardefined in this way will define a languagewhich is equal to the original language L∗, a condition Lambek states as follows: Onemayconsiderthecategorialgrammartobeadequate providedit assigns type S to w if and only if the latter is a well-formed declarative sentence according to some other standard. Lambek’s specific proposals within this metatheoretical program include what are now called Lambek grammars, and later, pregroup grammars. Others have studiedtherelationshipbetweentypesandsetsofstrings;specificallyBuszkowski in [5], and the possibilities of developing the informal process of construction of a grammar into an algorithmically well-defined learning procedure [6,7]. In this paper we will take a different tack, motivated again by a focus on learnability. We are interested in identifying formalisms that are not only de- scriptively adequate but that also can be learned from the data in a principled way – in Chomskyan terms, in achieving explanatory adequacy. Defining the types as sets of strings seems a reasonable first step. We can in a similar vein consider nonterminal symbols in a context-free grammar to be sets of strings: namely the set of strings that can be derived from that nonterminal. For a cfg G, with a nonterminal N we define as usual L(G,N)={w|N ⇒∗ w}. A sym- G bol NP which represents noun phrases can be considered as the set of strings that can be noun phrases; the start symbol S corresponds to the language L itself. In Lambek grammars and other categorial grammars, we have the same cor- respondence,butitismorecomplex.We canassociatewitheachtype,primitive
The list of books you might like
