New Directions in Logic, Language and Computation: ESSLLI 2010 and ESSLLI 2011 Student Sessions. Selected Papers PDF
Preview New Directions in Logic, Language and Computation: ESSLLI 2010 and ESSLLI 2011 Student Sessions. Selected Papers
Lecture Notes in Computer Science 7415 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 Daniel Lassiter Marija Slavkovik (Eds.) New Directions in Logic, Language and Computation ESSLLI 2010 and ESSLLI 2011 Student Sessions Selected Papers 1 3 VolumeEditors DanielLassiter StanfordUniversity DepartmentofPsychology 420JordanHall 450SerraMall Stanford,CA94305,USA E-mail:[email protected] MarijaSlavkovik UniversityofLuxembourg 6,RueRichardCoudenhoveKalergi 1359Luxembourg E-mail:[email protected] ISSN0302-9743 e-ISSN1611-3349 ISBN978-3-642-31466-7 e-ISBN978-3-642-31467-4 DOI10.1007/978-3-642-31467-4 SpringerHeidelbergDordrechtLondonNewYork LibraryofCongressControlNumber:2012940850 CRSubjectClassification(1998):F.4.1,I.2.3-4,I.2.7,D.3.2 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 We are happy to present you with the second volume of selected papers from the Student Sessions of the European Summer Schools in Logic, Language and Information (ESSLLI), in what we hope will develop into a biannual tradition. This collection contains revised and expanded version of papers presented at the Student Sessions of the 22nd and 23rd ESSLLI, held respectively in 2010 in Copenhagen, Denmark, and in 2011 in Ljubljana, Slovenia. ESSLLI has taken place every year since 1989 under the auspices of FoLLI, the Association for Logic,Languageand Information.Since 1996,the Student Sessionhas provided a forum in which promising young scholars can present their work in a friendly andsupportiveenvironmentandgetconstructivefeedbackfromexpertreviewers and audiences from diverse areas. Many papers from previous Student Sessions have represented original per- spectives and lasting insights from promising young researchers. In recognition of this, a volume of selected best papers from 2008 and 2009 appeared for the first time in early 2011 as a Lecture Notes in Computer Science volume (Inter- faces: Explorations in Logic, Language, and Computation,ed.ThomasIcardand ReinhardMuskens).Wehopethatwiththisfollow-upvolumewemayencourage promising young researchers in future years to submit high-quality work to the Student Session. Wereceived49submissionsinthe2010StudentSession,outofwhich16were selectedfororaland11for posterpresentations.Inthe 2011Student Sessionwe received 53 submissions out of which 16 were selected for oral and 6 for poster presentation.Outofthe32papersselectedfororalpresentations,weselected15 that were extended and reviewed again for inclusion in this volume. WewouldliketothankthosewithoutwhomtheStudentSessionandthisvol- ume would not have been possible: our hard-workingProgramCommittees, the helpful audiences,and especially the main ESSLLI organizersand the many ex- pertreviewerswhogenerouslytooktimetohelpselectthesepapersandsharpen the insights that they contain. April 2012 Daniel Lassiter Marija Slavkovik Organization Program Committee 2010 Jens Ulrik Hansen Roskilde University, Denmark Szymon Klarman VrijeUniversiteitAmsterdam,TheNetherlands Ekaterina Lebedeva INRIA Nancy, University Henri - Poincare, France Pierre Lison DFKI Saarbru¨cken,Germany Mingya Liu University of Go¨ttingen, Germany Marija Slavkovik University of Luxembourg Natalia Vinogradova Bordeaux-1 University, France Program Committee 2011 Anna Chernilovskaya Utrecht University, The Netherlands Hanna de Vries Utrecht University, The Netherlands Janez Kranjc Jozef Stefan Institute, Slovenia Daniel Lassiter New York University, USA Daphne Theijssen Radboud University Nijmegen, The Netherlands Sander Wubben Tilburg University, The Netherlands Fan Yang University of Helsinki, Finland Table of Contents Language and Computation Predicting the Position of Attributive Adjectives in the French NP ..... 1 Gwendoline Fox and Juliette Thuilier Abductive Reasoning for Continual Dialogue Understanding........... 16 Miroslav Jan´ıˇcek SCA: Phonetic Alignment Based on Sound Classes ................... 32 Johann-Mattis List Embodied Quantifiers ............................................ 52 Simon Pauw and Michael Spranger Ranked Multidimensional Dialogue Act Annotation .................. 67 Marcin Wl(cid:2)odarczak Logic and Computation The Good, the Bad, and the Odd: Cycles in Answer-Set Programs ..... 78 Johannes Klaus Fichte Distributed Knowledge with Justifications........................... 91 Meghdad Ghari Epistemic Logic, Relevant Alternatives, and the Dynamics of Context ...................................................... 109 Wesley H. Holliday Towards a Generalization of Modal Definability...................... 130 Tin Perkov Modeling Semantic Competence: A Critical Review of Frege’s Puzzle about Identity ................................................... 140 Rasmus K. Rendsvig Comparing Inconsistency Resolutions in Multi-Context Systems........ 158 Antonius Weinzierl Logic and Language Vague Determiner Phrases and Distributive Predication............... 175 Heather Burnett VIII Table of Contents The Syntax and Semantics of Evaluative Degree Modification.......... 195 Hanna de Vries A Kripkean Solution to Paradoxesof Denotation..................... 212 Casper Storm Hansen Trust Games as a Model for Requests .............................. 221 Jason Quinley Author Index.................................................. 235 Predicting the Position of Attributive Adjectives in the French NP Gwendoline Fox1 and Juliette Thuilier2 1 University of Paris 3 - SorbonneNouvelle (ILPGA) and EA 1483 2 UnivParis Diderot, SorbonneParis Cité, ALPAGE,UMR-I001 INRIA 1 Introduction French displays the possibility of both pre-nominal and post-nominal ordering of adjectives within the noun phrase (NP). (1) un magnifique tableau / un tableau magnifique a magnificent painting / a painting magnificient “a magnificient painting” While alladjectivesmayalternateinposition,the choicebetweenbothordersis not as free as suggested in (1): (2) a. un beau tableau / ??un tableau beau a nice painting / a painting nice b. un très beau tableau / un tableau très beau a very nice painting / a painting very nice “a (very) nice painting” c. *un beau à couper le souffle tableau / un tableau beau à a nice to cut the breath painting / a painting nice to couper le souffle cut the breath “a breathtakingly beautiful painting” The examples in (2) show that the positionning of attributive adjectives is a complex phenomenon: unlike magnifique, the adjective beau cannot be placed freely when it is the only element of the adjectival phrase (AP). It is strongly preferred in anteposition (2-a). The addition of the pre-adjectival adverb très gives more flexibility and equally allows both orders (2-b) whereas the use of a post-adjectival modifier constrains the placement to postposition (2-c). The phenomenon of adjective alternation has been widely studied in French linguistics ([1], [2], [3], [4], [5], [6], [7] among others). Many constraints were proposed on different dimensions of the language: phonology, morphology, syn- tax,semantics1,discourseandpragmatics.Onlyoneofthemiscategoricalinthe 1 In some cases, alternation leads to meaning differences for the adjective (see for instance [1], [3], [7]). The decision between the different possible accounts of how thesedifferencescouldbegeneratedisbeyondthescopeofthisarticle.Wethusleave aside these semantic considerations and focus here on the form of theadjective. D.LassiterandM.Slavkovik(Eds.):ESSLLIStudentSessions,LNCS7415,pp.1–15,2012. (cid:2)c Springer-VerlagBerlinHeidelberg2012 2 G. Fox and J. Thuilier sensethatitimposesaspecificpositiontoanyattributiveadjective:thepresence ofa post-adjectivalcomplement(3)ormodifier (2-c)only allowspostpositionof the adjective. (3) un homme fier de son fils / *un fier de son fils homme a man proud of his son / a proud of his son man “a man proud of his son” The other constraints participating in the alternationbetween antepositionand postposition are not categorical. For instance, as noted in the corpus studies of [2] and [3], length is a preferential constraint: short adjectives tend to be anteposed to the noun. The sequence “un magnifique tableau” in (1) illustrates that this rule can be violated whether one considers the length of the adjective alone: magnifique has 3 syllables, or the relative length between the adjective and the noun (3 > 2). Althoughtheabove-mentionedworkshaveenabledtoidentifytheconstraints playingaroleintheplacementofadjectives,mostarebasedonintrospectionand only examine a few of these constraints. It is thus very difficult to evaluate the actualimpactofeachoftheminusage,andthereforetoestimatetheirrespective weight in the speaker’s choice for one position over the other. This paper aims to get a better grasp of the general picture of the phenomenon. To do so, we present along the same lines as [8], [9] and [10], a quantitative study, based on two corpora: the French Tree Bank (henceforth FTB) and the Est-Républicain corpus (henceforth ER). We propose a regression model based on interpretable constraints and compare the prediction capacities of different subsets in order to determine what kind of informations are the most reliable to account for the placement of adjectives. The paper is organisedasfollows.We presentinsection2 the methodological aspectsofourstudy:constitutionofthedatatableandpresentationofthestatis- tical model. In section 3, we describe the variables derived from the constraints found in the literature. Section 4 is dedicated to the comparison of the models basedonthedifferentsubsetsofvariablesandtotheinterpretationoftheresults. 2 Methodology Building the Datatable. Thefirststepofthisworkistocollectthedatacon- cerningadjectivesandcapturetheconstraintsfoundintheliterature.Thestudy is based on the functionally annotated subset of the FTB corpus [11]2, which contains 12,351 sentences, 24,098 word types and 385,458 tokens. It is, for the moment,theonlyexistingtreebankforFrench.Weextractedalltheoccurrences of attributive adjectives from this corpus3, and filtered out numeraladjectives4, 2 This subset corresponds to thepart that was manually corrected. 3 We identified attributive adjectives using the following pattern in the treebank: an adjective occuring with a nominal head within a NPis an attributiveadjective. 4 Cardinal numerals such as trois ’three’, vingt ’twenty’, soixante ’sixty’... are some- times annotated as adjectives in theFTB. Predicting thePosition of AttributiveAdjectives in the French NP 3 adjectives appearing in dates5, abbreviations6 and incorrectly annotated occur- rences. We also discarded the 438 adjectives occurring with a post-adjectival dependent since postposition is obligatory in this case, regardless of the values of other constraints that we consider (see (2-c) and (3)). The remaining adjec- tives constitute the basis of the datatable, to which we have added information on the position of each adjective with respect to the noun it modifies, and 10 other variables that we describe in section 3. Three variables of our study are based on frequency counts: freq, Collo- cAnt and CollocPost. They were extracted from the ER corpus for more reliable counts. The raw corpus contains 147,934,722tokens,and is available on theATILFwebsite7.ItwastaggedandlemmatizedwiththeMorfette system[12] adapted for French. We used ER for these constraints because it is around 380 times larger than FTB. We therefore consider that frequency in ER is a better estimatorofthe probabilityofuseofanadjective.Also,weuse herea logtrans- formedvalueofthefrequencytoreducetherangeofvaluesofthisvariable.More precisely,thethreevariablestakethefollowingvalue:log(frequency in ER+1), in order to avoid a null value in case an adjectival lemma or noun-adjective combination is absent from ER. Presentation of the Datatable. The datatable contains 14,804 occurrences corresponding to 1,920 adjectival lemmas. 4,227 (28.6%) tokens appear in an- teposition,and10,577(71.4%)inpostposition.Table1showsthattheadjectival lemmas displaying occurrences in both positions representonly 9.5% of all lem- mas, yet these few lemmas correspond to 5,473 occurrences, i.e. 37.0% of the datatable, which means that very few adjectives actually alternate in usage but they are highly frequent. Note that amongthe alternatingadjectives (occurringin both positions),the ratio between anteposed and postposed occurrences is the reverse from that of all adjectives: there are 3,727 anteposed (68,1%) and 1,746 postposed (31,9%) adjectives. Alternating adjectives thus show a preference for anteposition. The generalpatternisthereforethatpostposedadjectivestendtobe infrequentlem- mas occurring only in postposition, whereas alternating adjectives tend to be frequent and to prefer anteposition. Table 1. Distribution of adjectival lemmas and tokensaccording to position anteposed postposed both positions Overall number of lemmas 125 1613 182 1920 6.5% 84.0% 9.5% 100% tokens 500 8831 5473 14804 3.4% 59.7% 37.0% 100% 5 Examples of dates containing adjectives: "[13]ADJ [mars]N", "[lundi]N [31]ADJ". 6 Nounsoradjectivesareviewedasabbreviationsiftheirlastletterisacapitalletter. 7 http://www.cnrtl.fr/corpus/estrepublicain/
The list of books you might like
