Institute For Deutsche Sprache
Institute For Deutsche Sprache
Arnold D.,University of Tübingen |
Arnold D.,Institute For Deutsche Sprache |
Tomaschek F.,University of Tübingen |
Sering K.,University of Tübingen |
And 2 more authors.
PLoS ONE | Year: 2017
Sound units play a pivotal role in cognitive models of auditory comprehension. The general consensus is that during perception listeners break down speech into auditory words and subsequently phones. Indeed, cognitive speech recognition is typically taken to be computationally intractable without phones. Here we present a computational model trained on 20 hours of conversational speech that recognizes word meanings within the range of human performance (model 25%, native speakers 20-44%), without making use of phone or word form representations. Our model also generates successfully predictions about the speed and accuracy of human auditory comprehension. At the heart of the model is a 'wide' yet sparse two-layer artificial neural network with some hundred thousand input units representing summaries of changes in acoustic frequency bands, and proxies for lexical meanings as output units. We believe that our model holds promise for resolving longstanding theoretical problems surrounding the notion of the phone in linguistic theory. © 2017 Arnold et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Hernandez Rizzardini R.,Galileo University |
Linares B.H.,Galileo University |
Mikroyannidis A.,Open University Milton Keynes |
Schmitz H.-C.,Institute For Deutsche Sprache
Journal of Universal Computer Science | Year: 2013
The ROLE project (Responsive Open Learning Environments, EU 7th Framework Programme, grant agreement no.: 231396, 2009-2013) was focused on the next generation of Personal Learning Environments (PLEs). A ROLE PLE is a bundle of interoperating widgets -often realised as cloud services - used for teaching and learning. In this paper, we first describe the creation of new ROLE widgets and widget bundles at Galileo University, Guatemala, within a cloud-based infrastructure. We introduce an initial architecture for cloud interoperability services including the means for collecting interaction data as needed for learning analytics. Furthermore, we describe the newly implemented widgets, namely a social networking tool, a mind-mapping tool and an online document editor, as well as the modification of existing widgets. The newly created and modified widgets have been combined in two different bundles that have been evaluated in two web-based courses at Galileo University, with participants from three different Latin-American countries. We measured emotional aspects, motivation, usability and attitudes towards the environment. The results demonstrated the readiness of cloud-based education solutions, and how ROLE can bring together such an environment from a PLE perspective. © J.UCS.
Nussbaumer A.,University of Graz |
Dahrendorf D.,Multimedia University |
Schmitz H.-C.,Institute For Deutsche Sprache |
Kravcik M.,RWTH Aachen |
And 2 more authors.
Computer Science and Information Systems | Year: 2014
This article presents an approach that supports the creation of personal learning environments (PLE) suitable for self-regulated learning (SRL). PLEs became very popular in recent years offering more personal freedom to learners than traditional learning environments. However, creating and configuring PLEs demand specific meta-skills that not all learners have. This situation leads to the challenge how learners can be supported to create PLEs that are useful to achieve their intended learning outcomes. The theory of SRL describes learners as self-regulated if they are capable of taking over control of the own learning process. Grounding on that theory, a model has been elaborated that offers guidance for the creation of PLEs containing tools for cognitive and meta-cognitive learning activities. The implementation of this approach has been done in the context of the ROLE infrastructure. A quantitative and qualitative evaluation with teachers describes advantages and ideas for improvement.
Prevot L.,Aix - Marseille University |
Gorisch J.,Institute For Deutsche Sprache |
Mukherjee S.,Italian Institute of Technology
29th Pacific Asia Conference on Language, Information and Computation, PACLIC 2015 | Year: 2015
Feedback utterances are among the most frequent in dialogue. Feedback is also a crucial aspect of all linguistic theories that take social interaction involving language into account. However, determining communicative functions is a notoriously difficult task both for human interpreters and systems. It involves an interpretative process that integrates various sources of information. Existing work on communicative function classification comes from either dialogue act tagging where it is generally coarse grained concerning the feedback phenomena or it is token-based and does not address the variety of forms that feedback utterances can take. This paper introduces an annotation framework, the dataset and the related annotation campaign (involving 7 raters to annotate nearly 6000 utterances). We present its evaluation not merely in terms of inter-rater agreement but also in terms of usability of the resulting reference dataset both from a linguistic research perspective and from a more applicative viewpoint.
Verstehen und Verständlichkeit von populärwissenschaftlichen Texten: Das Projekt PopSci-Understanding Science Comprehension and comprehensibility of popular science texts: the research project PopSci-Understanding Science La compréhension et la compréhensibilité de textes de vulgarisation scientifique: le projet PopSci-Understanding Science
Wolfer S.,Institute For Deutsche Sprache |
Hansen-Schirra S.,Johannes Gutenberg University Mainz |
Held U.,Hochmeisterstrasse 2
Information-Wissenschaft und Praxis | Year: 2015
The public accessibility and comprehension of scientific aims and results fundamentally influences the social acceptability and receptiveness of research. The contents of up-to-date research in the (natural) sciences are, however, not easily accessible to a lay audience because of many interfering factors. Aiming at the optimization of scientific publications in German print and online media, we investigate and validate present-day popular science discourse within our project PopSci-Understanding Science. For this purpose, stylistic features of German popular-science writing are identified through corpus-based research and their effects on the lay reader's processing of these texts are measured experimentally. The resulting recommendations will improve the linguistic style and knowledge representation of written and web-based publications. © 2015 by Walter de Gruyter Berlin Boston 2015.
Schnober C.,Institute For Deutsche Sprache
11th Conference on Natural Language Processing, KONVENS 2012: Empirical Methods in Natural Language Processing - Proceedings of the Conference on Natural Language Processing 2012 | Year: 2012
This paper describes a practical approach to use the information retrieval engine Lucene for the corpus analysis platform KorAP, currently being developed at the Institut für Deutsche Sprache (IDS Mannheim). It presents a method to use Lucene's indexing technique and to exploit it for linguistically annotated data, allowing full flexibility to handle multiple annotation layers. It uses multiple indexes and MapReduce techniques in order to keep KorAP scalable.
Deppermann A.,Institute For Deutsche Sprache |
Spranz-Fogasy T.,Institute For Deutsche Sprache
Communication and Medicine | Year: 2011
Based on German data from history-taking in doctor-patient interaction, the paper shows that the three basic syntactic types of questions (questions fronted by a question-word (w-questions), verb-first (V1) questions, and declarative questions) provide different opportunities for displaying understanding in medical interaction. Each syntactic question-format is predominantly used in a different stage of topical sequences in history taking: w-questions presuppose less knowledge and are thus used to open up topical sequences; declarative questions are used to check already achieved understandings and to close topical sequences. Still, the expected scope of answers to yes/no-questions and to declarative questions is less restricted than previously thought. The paper focuses in detail on the doctors' use of formulations as declarative questions, which are designed to make patients elaborate on already established topics, giving more details or accounting for a confirmation. Formulations often involve a shift to psychological aspects of the illness. Although patients confirm doctors' empathetic formulations, they, however, regularly do not align with this shift, returning to the description of symptoms and to biomedical accounts instead. The study shows how displays of understanding are responded to not only in terms of correctness, but also (and more importantly) in terms of their relevance for further action. Copyright © Equinox Publishing Ltd.
Brunner A.,Institute For Deutsche Sprache
Literary and Linguistic Computing | Year: 2013
This article presents the main results of a project, which explored ways to recognize and classify a narrative feature-speech, thought, and writing representation (ST&WR)-automatically, using surface information and methods of computational linguistics. The task was to detect and distinguish four types- direct, free indirect, indirect, and reported ST&WR-in a corpus of manually annotated German narrative texts. Rule-based as well as machine-learning methods were tested and compared. The results were best for recognizing direct ST&WR (best F1 score: 0.87), followed by indirect (0.71), reported (0.58), and finally free indirect ST&WR (0.40). The rule-based approach worked best for ST&WR types with clear patterns, like indirect and marked direct ST&WR, and often gave the most accurate results. Machine learning was most successful for types without clear indicators, like free indirect ST&WR, and proved more stable. When looking at the percentage of ST&WR in a text, the results of machinelearning methods always correlated best with the results of manual annotation. Creating a union or intersection of the results of the two approaches did not lead to striking improvements. A stricter definition of ST&WR, which excluded borderline cases, made the task harder and led to worse results for both approaches. © The Author 2013. Published by Oxford University Press on behalf of ALLC. All rights reserved.
Deppermann A.,Institute For Deutsche Sprache
Journal of Pragmatics | Year: 2013
Based on German speaking data from various activity types, the range of multimodal resources used to construct turn-beginnings is reviewed. It is claimed that participants in talk-in-interaction need to deal with four tasks in order to construct a turn which precisely fits the interactional moment of its production:. 1.Achieve joint orientation: The accomplishment of the socio-spatial prerequisites necessary for producing a turn which is to become part of the participants' common ground.2.Display uptake: Next speaker needs to display his/her understanding of the interaction so far as the backdrop on which the production of the upcoming turn is based.3.Deal with projections from prior talk: The speaker has to deal with projections which have been established by (the) previous turn(s) with respect to the upcoming turn.4.Project properties of turn-in-progress: The speaker needs to orient the recipient to properties of the turn s/he is about to produce.Turn-design thus can be seen to be informed by tasks related to the multimodal, embodied, and interactive contingencies of online-construction of turns. The four tasks are ordered in terms of prior tasks providing the prerequisite for accomplishing a later task. © 2012 Elsevier B.V.
Mahlow C.,Institute For Deutsche Sprache
ACM International Conference Proceeding Series | Year: 2015
Research on differencing and versioning in computer science, writing research, and scholarly editing are all concerned with the evolution of texts and documents through various drafts, versions, and changes between different stages. While the different disciplines all have their own perspectives, on closer inspection, there is a large overlap of concerns. However, there are currently no points of contacts between these disciplines. In this vision paper, we propose a working definition of "version" to reconcile the different views and open the way to closer collaboration. In particular, we propose to model the semantics of changes in what seems to be unstructured text as differences between syntactic parse trees. © 2015 ACM.