Table Of ContentDepLing 2013
Proceedings of the Second International Conference
on Dependency Linguistics
August 27 – 30, 2013, Prague, Czech Republic
edited by
Eva Hajičová, Kim Gerdes, Leo Wanner
Vilém
Mathesius
Foundation
Prague
Editors and program committee co-chairs
Kim Gerdes, Université Sorbonne Nouvelle (co-chair, editor)
Eva Hajičová, Charles University in Prague (co-chair, editor)
Leo Wanner, Universitat Pompeu Fabra (co-chair, editor)
Jiří Mírovský (technical editor)
Eduard Bejček (technical editor)
Reviewers
Margarita Alonso-Ramos, Universidade da Henning Lobin, Justus-Liebig-University Giessen
Coruña Markéta Lopatková, Charles University in Prague
David Beck, University of Alberta Christopher Manning, Stanford University
Xavier Blanco, UAB Jasmina Milićević, Dalhousie University
Igor Boguslavsky, Universidad Politécnica de Henrik Høeg Müller, Copenhagen Business
Madrid School (CBS)
Bernd Bohnet, University Stuttgart Alexis Nasr, Université de la Méditerranée
Marie Candito, Université Paris 7 / INRIA Laboratoire d'Informatique Fondamentale
Silvie Cinková, Charles University in Prague Joakim Nivre, Uppsala University
Benoit Crabbé, Paris 7 et INRIA Kemal Oflazer, Carnegie Mellon University in
Eric De La Clergerie, INRIA Qatar
Denys Duchier, Université d'Orléans Martha Palmer, University of Colorado
Dina El Kassas, Minya University Jarmila Panevová, Charles University in Prague
Koldo Gojenola, University of the Basque Alain Polguère, Université Nancy 2
Country UPV/EHU Prokopis Prokopidis, Institute for Language and
Thomas Gross, Aichi University Speech Processing, Athena Research and
Barbora Hladká, Charles University in Prague Innovation Center
Richard Hudson, UCL Ines Rehbein, Potsdam University
Leonid Iomdin, Russian Academy of Sciences Dipti Sharma, IIIT
Sylvain Kahane Modyco, Université Paris Ouest Pavel Straňák, Charles University in Prague
& CNRS / Alpage, INRIA Gertjan van Noord, University of Groningen
Marco Kuhlmann, Uppsala University Daniel Zeman, Charles University in Prague
François Lareau, Macquarie University Zdeněk Žabokrtský, Charles University in
Alessandro Lenci, University of Pisa Prague
Haitao Liu, Zhejiang University
Published by
MATFYZPRESS Publishing House
of the Faculty of Mathematics and Physics
Charles University in Prague
Sokolovská 83, 186 75 Praha 8, Czech Republic
th
as the 434 publication
Designed and printed by Reprostředisko UK MFF
Sokolovská 83, 186 75 Praha 8, Czech Republic
First edition, Praha 2013
© Eva Hajičová, Kim Gerdes, Leo Wanner (editors), 2013
© MATFYZPRESS, Publishing House of the Faculty of Mathematics and Physics, Charles
University in Prague, 2013
Organized by the Charles University in Prague, Faculty of Mathematics and Physics,
Institute of Formal and Applied Linguistics (ÚFAL MFF UK).
ISBN 978-80-7378-240-5
FOREWORD
The DepLing 2013 conference is the second meeting in the newly established series
of international conferences on dependency linguistics started in 2011 by the first
DepLing in Barcelona. The response to the initiative to organize special meetings
devoted to the dependency linguistic theory (which nowadays seems to be in the
forefront of interests among both theoretical and computational linguists) was quite
supportive. We do hope that the present conference will manage to keep pace with
the high standards set at the Barcelona meeting.
To make all the accepted contributions available to the linguistic community and
beyond, we have decided to publish a full volume of Proceedings of both oral papers
and poster presentations. The coverage is rather broad: from the formal point of
view, the papers present different theoretical dependency models or compare the
dependency approach with the phrase structure based one. Issues pertaining to
different language layers range from morphology and morphosyntax to syntax
proper and even discourse, and language material is supplied for 10 languages both
modern and old or ancient. Several papers describe the application of dependency
analysis to the build-up of monolingual and multilingual treebanks.
We are proud that the invitation to give a plenary speech was accepted by two
prominent scholars, Richard Hudson as one of the main figures in dependency
linguistics and father of the Word Grammar, and Aravind Joshi, a prominent
representative of formal description of language and the original proponent of the
tree-adjoining grammar formalism.
Our sincere thanks go to the members of the Scientific Committee, who have
undertaken the task to read three papers each and have sent in – at least in majority
– detailed comments and suggestions.
We are also most grateful to our young colleagues from the Institute of Formal and
Applied Linguistics (ÚFAL), Charles University in Prague, who took care of the
conference management system (through EasyChair) and prepared the Proceedings
volume, first of all Filip Jurčíček, Jiří Mírovský, and Eduard Bejček. Our thanks also go
to Mrs. Anna Kotěšovcová, who was our link to the MatfyzPress Publishers.
Last but not least, we gratefully acknowledge the financial and moral support given
by the ÚFAL Management, by the LINDAT/CLARIN infrastructural project funded by
the Ministry of Education, Youth and Sports of the Czech Republic, by the 7th
framework EC-funded META-NET network and the Khresmoi integrated project, and
by the two Czech Grant Agency projects, namely P406/12/0658 (Coreference,
discourse relations and information structure in a contrastive perspective) and
P406/2010/0875 (Computational Linguistics: Explicit description of language and
annotated data focused on Czech).
Welcome to DepLing 2013 in Prague and have a good and rewarding time there!
Kim Gerdes Eva Hajičová Leo Wanner
DepLing 2013
the Second International Conference on Dependency Linguistics
August 27 – 30, 2013, Prague, Czech Republic
Organized by
Institute of Formal and Applied Linguistics (ÚFAL)
Faculty of Mathematics and Physics
Charles University in Prague
Czech Republic
(http://ufal.mff.cuni.cz)
and
CONFORG, s.r.o.
Czech Republic
(http://www.conforg.cz)
in the historic building at
Malostranské nám. 25
118 00 Prague 1
Czech Republic
Table of Contents
Invited talk: Dependency Structure and Cognition
Richard Hudson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Invited talk: Dependency Representations, Grammars, Folded Structures, among Other Things!
Aravind K. Joshi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
Exploring Morphosyntactic Annotation over a Spanish Corpus for Dependency Parsing
Miguel Ballesteros, Simon Mille and Alicia Burga . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Towards Joint Morphological Analysis and Dependency Parsing of Turkish
O¨ zlem C¸ etinog˘lu and Jonas Kuhn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Divergences in English-Hindi Parallel Dependency Treebanks
Himani Chaudhry, Himanshu Sharma and Dipti Misra Sharma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Dependency Network Syntax:
From Dependency Treebanks to a Classification of Chinese Function Words
Xinying Chen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Verb Cluster, Non-Projectivity, and Syntax-Topology Interface in Korean
Jihye Chun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Rule-Based Extraction of English Verb Collocates from a Dependency-Parsed Corpus
Silvie Cinkova´, Martin Holub, Ema Krejcˇova´ and Lenka Smejkalova´ . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
A Method to Generate Simplified Systemic Functional Parses from Dependency Parses
Eugeniu Costetchi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Dependency Distance and Bilingual Language Use:
Evidence from German/English and Chinese/English Data
Eva M. Duran Eppler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Collaborative Dependency Annotation
Kim Gerdes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Pragmatic Structures in Aymara
Petr Homola and Matt Coler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Towards a Psycholinguistically Motivated Dependency Grammar for Hindi
Samar Husain, Rajesh Bhatt and Shravan Vasishth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
The Syntax of Hungarian Auxiliaries: A Dependency Grammar Account
Andra´s Imre´nyi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .118
Subordinators with Elaborative Meanings in Czech and English
Pavl´ına J´ınova´, Lucie Pola´kova´ and Jirˇ´ı M´ırovsky´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
v
Predicative Adjunction in a Modular Dependency Grammar
Sylvain Kahane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
The Representation of Czech Light Verb Constructions in a Valency Lexicon
Va´clava Kettnerova´ and Marke´ta Lopatkova´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
A Deterministic Dependency Parser with Dynamic Programming for Sanskrit
Amba Kulkarni . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Reasoning with Dependency Structures and Lexicographic Definitions Using Unit Graphs
Maxime Lefranc¸ois and Fabien Gandon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Non-Projectivity in the Ancient Greek Dependency Treebank
Francesco Mambrini and Marco Passarotti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
More Constructions, More Genres: Extending Stanford Dependencies
Marie-Catherine de Marneffe, Miriam Connor, Natalia Silveira, Samuel R. Bowman,
Timothy Dozat and Christopher D. Manning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Why So Many Nodes?
Dan Maxwell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Grammatical Markers and Grammatical Relations in the Simple Clause in Old French
Nicolas Mazziotta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
AnCora-UPF: A Multi-Level Annotation of Spanish
Simon Mille, Alicia Burga and Leo Wanner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Towards Building Parallel Dependency Treebanks:
Intra-Chunk Expansion and Alignment for English Dependency Treebank
Debanka Nandi, Maaz Nomani, Himanshu Sharma, Himani Chaudhary,
Sambhav Jain and Dipti Misra Sharma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Annotators’ Certainty and Disagreements in Coreference and Bridging Annotation
in Prague Dependency Treebank
Anna Nedoluzhko and Jiˇr´ı M´ırovsky´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .236
How Dependency Trees and Tectogrammatics Help Annotating Coreference
and Bridging Relations in Prague Dependency Treebank
Anna Nedoluzhko and Jiˇr´ı M´ırovsky´ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .244
Predicting Conjunct Propagation and Other Extended Stanford Dependencies
Jenna Nyblom, Samuel Kohonen, Katri Haverinen, Tapio Salakoski and Filip Ginter . . . . . . . . . . . . 252
´
A Look at Tesnie`re’s Ele´ments through the Lens of Modern Syntactic Theory
Timothy Osborne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
The Distribution of Floating Quantifiers: A Dependency Grammar Analysis
Timothy Osborne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
vi
Dependency and Constituency in Translation Shift Analysis
Manuela Sanguinetti, Cristina Bosco and Leonardo Lesmo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Managing a Multilingual Treebank Project
Milan Soucˇek, Timo Ja¨rvinen and Adam LaMontagne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
An Empirical Study of Differences between Conversion Schemes and Annotation Guidelines
Anders Søgaard . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 298
vii
Dependency Structure and Cognition
Invited talk
Richard Hudson
emeritus professor in the Department of Phonetics and Linguistics
University College London
Great Britain
would be; but we aren’t. We’re studying a part of 1983). Our generative colleagues are free to in
the human mind, and any human mind is the vent principles, parameters and structures at will,
product of a long and complicated experience; so unconstrained by anything but their basic formal
why should we believe that any mind is simple? assumptions and the purely ‘linguistic’ facts. As
As cognitive linguists argue, we learn our lan you can guess, I don’t think this is a good way to
guage from ‘usage’ (Barlow and Kemmer 2000) study language because I believe that language
– from the millions of examples of language that is, in fact, just like the rest of cognition in spite
we hear, each embedded in a very specific social of all the attempts to show the contrary.
context. And we interpret each example in terms
of the examples that went before, using a grow 2 Some things we know about cognition
ing system of concepts. Nothing there is simple:
We start with four very elementary findings
for any given language, thousands or millions of
which can be found in introductory textbooks on
speakers all follow different routes to a slightly
cognitive psychology such as Reisberg (2007),
different adult grammar, with numerous false
concerning networks, mental relations, complex
starts and detours on the way. It’s easy to under
ity and classification.
stand why linguists welcome the idea of a sim
Knowledge is a network of concepts in which
ple, perfect and uniform language as a way to es
each concept is associated with a number of
cape from this buzz of confusion and complexity.
other concepts. These ‘associations’ explain why
But, like the drunk looking for the keys that he
experiences evoke neighbouring memories –
has dropped, we have a choice: we can look un
memories that share links (in the network) to the
der the street lamp, where the light is good; or
same concepts; why we make mistakes (includ
we can look over in the dark corner, where we
ing speech errors) when we choose a neighbour
know that we actually dropped the keys – a
ing concept in place of the intended target; and
choice between esthetics and truth.
why an object in a psychological laboratory
In short, I believe we have to accept that lan
‘primes’ objects that are its neighbours (as when
guage is part of cognition. And with that accep
hearing the word doctor makes the word nurse
tance comes the principle that our theories of lan
easier to retrieve than it would otherwise be).
guage structure should be compatible with cogni
The notion of networks explains all these famil
tive science – in fact, our theories are part of
iar facts about cognition. But if knowledge in
cognitive science, and arguably a particularly im
general is a network, and if language is part of
portant part of cognitive science, given the rela
knowledge, then language itself must be a net
tive clarity and detail of the data found in lan
work. And that includes not only the whole of
guage. The reality that we are trying to capture in
language – the grammar and phonology as well
our theories is what is often called ‘psychologi
as the lexicon – but also the utterances that we
cal reality’.
interpret in terms of this network of knowledge.
But, you may object, how can we know what
But even though the notion of ‘association’ is
is psychologically real? It’s true that I can’t even
important, we can be sure that the links in our
look inside my own mind, let alone inside some
mental network are not merely associations, but
one else’s mind; but then, psychology has moved
relations of many different kinds. Just think of
a long way from the bad old days of introspec
all the words you know for kinship relations –
tion, and has findings which are supported by
words such as father, aunt and ancestor, each of
very robust experimental methods. The rest of
which names a relationship. Then think of all the
this paper is an attempt to develop some of the
other persontoperson relationships you can
consequences of taking these findings seriously
name, including ‘fatherinlaw’, ‘neighbour’ and
when building models of language. I shall pay
‘boss’? And then think of the prepositions and
special attention to their consequences for my
nouns you know for nonhuman relationships,
own theory, Word Grammar (WG, Hudson
such as beneath, opposite and consequence. The
1984, Hudson 1990, Hudson 2007, Hudson 2010,
point is that we seem to be able to freely create
Gisborne 2010, Eppler 2010).
and learn relational concepts, just as we do non
But before I go on to consider some of these
relational concepts such as ‘bird’ and
findings, I must admit that there is a way to
‘Londoner’. This conclusion takes us a long way
avoid my arguments. This is to claim that al
from theories in which our minds recognise only
though language is part of cognition, it is actu
a small, innate set of inbuilt relations called ‘syn
ally different from everything else – a unique
tactic functions’ or ‘semantic roles’.
‘module’ of the mind (Chomsky 1986, Fodor
2
