‚Informationstheoretisch basierte Modellierung sprachlicher Variation im Kontext‘ Workshop bei DGfS 2017

Das sprachliche System stellt uns die verschiedensten Möglichkeiten zur Kodierung von Inhalten zur Verfügung: So können wir mit „der von den Republikanern am wenigsten geliebte Präsidentschaftskandidat“ auf Donald Trump referieren, aber auch mit „der Präsidentschaftskandidat, der von den Republikanern am wenigsten geliebt wird“. Wir können mit „Einfach unglaublich!“ seinen Sieg in Florida kommentieren, aber auch mit „Das ist ja einfach unglaublich!“.
Dass für die Wahl einer konkreten sprachlichen Realisierung verschiedene kontextuelle Faktoren relevant sind, vom direkt vorangehenden Kontext bis hin zum Kontext der Situation und der Textsorte bzw. des Registers, dürfte nicht weiter kontrovers sein.  Unklar bleibt jedoch, wie diese  unterschiedlichen kontextuellen Faktoren Sprachprozesse beeinflussen.  In dieser AG gehen wir von der Hypothese aus, dass Sprache eine für die menschliche Kommunikation optimale Kodierung approximiert, indem sie  Anpassungen an kontextuelle Bedingungen ermöglicht, sei es an den unmittelbaren Kotext oder den Situationskontext. Zur Modellierung der Beziehung zwischen sprachlicher Kodierung und Kontext werden in neuerer Zeit informationstheoretische Ansätze verfolgt, die auf Shannons Informationsbegriff zurückgehen (Shannon, 1948) und ihn methodisch erweitern (siehe z. B. Informationsdichte/Surprisal).  Die Perspektive der Information erweist sich hier in Bezug auf die Modellierung ganz unterschiedlicher Sprachprozesse als sehr fruchtbar,  angefangen vom menschlichen Sprachverstehen über Sprachwandel und -evolution bis hin zur maschinellen Sprachverarbeitung, und zieht dabei alle sprachlichen Ebenen (Phonetik, Lexik, Syntax, Semantik, Diskurs) sowie ihre Interaktion in Betracht.
Wir bieten mit diesem Workshop eine Plattform für den Austausch derjenigen linguistischen Fachgebiete, die informationstheoretische Ansätze einsetzen bzw. weiterentwickeln, um die Beziehung von sprachlicher Kodierung und Kontext in unterschiedlichen Sprachprozessen zu modellieren. Die AG knüpft an die Ergebnisse der Kurz-AG ‚Information Density and Linguistic Variation‘ der DGfS 2012 in Frankfurt an und wird thematisch getragen durch den Saarbrücker SFB 1102 „Information Density and Linguistic Encoding“.

Shannon C., 1948. A Mathematical Theory of Communication. Bell System Technical Journal 27:379-423

Invited Speakers:

Tal Linzen (Ecole Normale Supérieure, Paris)
Hinrich Schütze (LMU, München)

List of accepted papers:

Robin Lemke, Ingo Reich & Eva Horch (Universität des Saarlandes)
Information density constrains article omission. An experimental approach

Rory Turnbull (Laboratoire de Sciences Cognitives et Psycholinguistique (ENS, EHESS, CNRS)) & Sharon Peperkamp (Ecole Normale Supérieure, Paris)
Lexical representation in phonological neighbourhood networks reveals limits of lexical organization

Fabian Tomaschek & Benjamin V. Tucker (Universität Tübingen)
Modeling segmental durations using the Naive Discriminative Learner

Jessie Nixon (Universität Potsdam) & Catherine Best (Western Sydney University)
High-variability distributions lower perceptual certainty during acoustic cue acquisition

Geertje van Bergen (Max Planck Institute for Psycholinguistik, Nijmegen)
Expectation management in interaction: Discourse particles signal surprisal of upcoming referents

Ekaterina Kravtchenko, Ashutosh Modi, Vera Demberg (Unviersität des Saarlandes), Ivan Titov (University of Amsterdam) & Manfred Pinkal (Universität des Saarlandes)
Does UID affect rate of pronominalization?

Olga Seminck & Pascal Amsill (Université Paris Diderot-Paris 7)
Predicting processing cost of anaphora resolution

Ayush Jain, Vishal Singh, Sumeet Agarval & Rajakrishnan Raijkumar (Indian Institute of Technology, Delhi)
Uniform Information Density Models for Language Production: A Comparative Study of Hindi and English

Elli Tourtouri, Francesca Delogu & Matthew Crocker (Universität des Saarlandes)
Over-specifications efficiently manage referential entropy in situated communication

Eric Meinardt (USC, San Diego)
Non-stationarity and other critical mathematical problems for channel coding-based explanations of variation in language production

Maria Pinango, Ashwini Deo & Muye Zhang (Yale University)
Accessing locative interpretations in „have“ sentences: Context-sensitive variability

Aaron Steven White & Kyle Rawlins (Johns Hopkins University)
Entropy predicts uncertainty in subcategorization frame distributions

Stefania Degaetano-Ortlieb, Merel Scholmann, Stefan Fischer, Elke Teich & Vera Demberg (Universität des Saarlandes)
An Information-theoretic Account on the Diachronic Development of Discourse Connectors in Scientific Writing

David Howcroft (Universität des Saarlandes), Cynthia A. Johnson (Ghent University) & Rory Turnbull (Laboratoire de Sciences Cognitives et Psycholinguistique (ENS, EHESS, CNRS))
German morphosyntactic change is consistent with an optimal encoding hypothesis

Workshop „Informationstheoretisch basierte Modellierung sprachlicher Variation im Kontext“ bei DGfS 2017

Elke Teich, Bernd Möbius, Vera Demberg (Saarbrücken)

Sprache approximiert eine für die Kommunikation optimale Kodierung, indem sie  Anpassungen an kontextuelle Bedingungen (Kotext, Situationskontext) ermöglicht. Wir fokussieren informationstheoretisch-basierte Ansätze zur Modellierung von unterschiedlichen sprachlichen Prozessen (menschliches Sprachverstehen, maschinelle Sprachverarbeitung, Sprachwandel  u. –evolution)  auf allen linguistischen Ebenen (Phonetik, Lexik, Syntax, Semantik, Diskurs) unter besonderer Berücksichtigung der Beziehung von sprachlicher Variation im Kontext.

PhD Day

The First SFB 1102 PhD Day

(a colloquium organized by the PhD students of SFB 1102)

Location: Graduate Center (Campus, C9.3)

Date: 5 February 2016

Program

10:00 Welcome

10:15 Eva Horch (B3)

Article missing? Corpus and information-theoretic approaches to article omission in German

10:45 Merel Scholmann (B2)

Predictions of upcoming discourse structure based on “On the one hand”

11:15 Simon Ostermann (A3)

Project A3: Mapping texts to scripts

11:30 Lunch

13:00 Irina Stenger (C4)

Cyrillic script intelligibility in reading intercomprehension among Slavic languages

13:30 Klara Jagrova (C4)

Adaptation towards a target language: the potential of improving intercomprehension of Polish for Czech Readers

14:00 Coffee Break

14:15 Andrea Fischer (C4)

MDL-based modeling of regularity in parallel linguistic sequences

14:45 Robin Lemke (B3)

Sentential or not? Some experiments on the syntactic information encoded in fragments

15:15 Discussion

16:00 Wrap-up + Reception

First anniversary of SFB 1102

On the occasion of its first anniversary, SFB 1102 will be hosting a science slam and Halloween dance party for Saarland University’s PhD students on October 30.

PROGRAM

Science Slam

16:00    Welcome Slam from the “birthday child” – SFB 1102

Prof. Dr. Elke Teich (Speaker of SFB 1102)

16:10    SFB 1102: Information Density and Linguistic Encoding

Mirjana Sekicki & Christine Ankener (PhD students, Project A5: Distributing Referential Information across Modalities)
“Prediction is not crystal ball gazing”

Clayton Greenberg (PhD student, Project B4: Modelling and Measuring Information Density)
“A new language model architecture”

David Howcroft (PhD student, Project A4: Language Comprehension and Cognitive Control Demands)
“Steering while Hearing”

Alessandra Zarcone (Postdoc, Project A2: Script Knowledge for Modelling Semantic Expectation)
““How to feed a dog?” or the rules of script knowledge”

16:30    GRK 2021 European Dream-Cultures

““The Dreamers”– Who we are and what we do (besides dreaming)”

Prof. Dr. Christiane Solte-Gresser (Speaker of GRK 2021)
Jonas Nesselhauf (associated PhD student)
Mert Akbal (associated PhD student)

16:45    Coffee break

17:15    IGK 1864: Diversity. Mediating Difference in Transcultural Spaces

Svetlana Seibel (PhD student, North American Literary and Cultural Studies)
““Personal Totems”: Indigenous Popular Culture in North America”

17:20    SFB 894: Ca2+signals: Molecular Mechanisms and Integrative Functions

Prof. Dr. Jens Rettig (Speaker of SFB 894)
“Ca2+signals: Molecular Mechanisms and Integrative Functions”

Hawraa Bzeih (PhD student, Project A10: Molecular mechanisms of maturation, Ca2+-dependent fusion and endocytosis of lytic granules)
“Cytotoxic T lymphocytes, the killers of our immune system”

Viktoria Götz (PhD student, Project A18: GnRH-induced Ca2+ signals in pituitary go- nadotrope cells)
“Central control of reproduction: from the brain to the gonads – and back”

17:35    Helmholtz Institute for Pharmaceutical Research Saarland

Hanzey Yasar (PhD student, Drug Delivery)
“Trojan horses as needle-free vaccine delivery against influenza”

17:40    SFB 1027: Physical Modelling of Non-Equilibrium Processes in Biological Systems

Prof. Dr. Dr. Karsten Kruse (Theoretical Biological Physics)
“The zombie principle”

Nicolas Thewes (PhD student, Experimental Physics)
“Biology meets physics to study bacterial adhesion”

Sarah Klein (PhD student, Theoretical Physics)
“Tug-of-war game: When small motors pull fat cargos”

17:55    SFB/TRR 14: Automatic Verification and Analysis of Complex Systems

Prof. Dr. Holger Hermanns (Dependable Systems and Software)
“When pigs are sweating”

Halloween Party

Location: Graduate Center (Campus, Geb. C9.3) Time: 19:30 – open end

Entrance free of charge!

Snacks & Drinks will be provided for a small price.

Feel free to bring your spouse, children and best friend with you! A costume is highly recommended!

Newly published overview article on IDeaL

Abstract

We introduce IDeaL (Information Density and Linguistic Encoding), a collaborative research center that investigates the hypothesis that language use may be driven by the optimal use of the communication channel. From the point of view of linguistics, our approach promises to shed light on selected aspects of language variation that are hitherto not sufficiently explained. Applications of our research can be envisaged in various areas of natural language processing and AI, including machine translation, text generation, speech synthesis and multimodal interfaces.

The full article can be found here.

New SFB Research Project approved

The German Research Foundation (DFG) has approved the Research Project B6 – Aspects of Information Density in Human and Machine Translation as of August 1st 2015. The Project will be coordinated by Prof. Josef van Genabith.

More details coming soon.

UdS Graduate Centre inaugurated

On Monday, Saarland University opened its new Graduate Centre as a building in which young scientists will be able to work and organise meetings or small conferences. The opening ceremony for the Graduate Centre, which brought together researchers from all disciplines under the heading „Science and Humanities in the Digital World„,  included a talk by SFB speaker Elke Teich entitled „Language – Text – Information„.

For further information on the new Graduate Centre, please click here.

Radio interview with Elke Teich

In an interview with radio station SR 2 aired on April 16, Elke Teich gave an overview of the research conducted in the SFB 1102.

Elke Teich nominated by DFG to join AcademiaNet

AcademiaNet is an international expert database for outstanding female academics run by Robert Bosch Stiftung. It was launched by Chancellor Dr. Angela Merkel in 2010 to gather the profiles of the most highly qualified women in their fields. Membership is by invitation only.

http://www.academia-net.org/

DiSpoL 2015

The workshop Identification and Annotation of Discourse Relations in Spoken Language (DiSpoL 2015) is part of the COST Action TextLink „Structuring Discourse in Multilingual Europe“.

The overall goal of the workshop is the development of a unified annotation framework for discourse-relational devices (DRDs) that is applicable to different languages and text types, including spoken and written dimensions.

For more information, please click here.

Successfully