Refine
Year of publication
- 2020 (22) (remove)
Document Type
- Article (7)
- Part of a Book (7)
- Conference Proceeding (6)
- Book (1)
- Working Paper (1)
Is part of the Bibliography
- yes (22) (remove)
Keywords
- Gesprochene Sprache (22) (remove)
Publicationstate
- Veröffentlichungsversion (12)
- Zweitveröffentlichung (10)
- Postprint (2)
Reviewstate
- Peer-Review (16)
- (Verlags)-Lektorat (6)
Publisher
- European Language Resources Association (5)
- Association for Computational Linguistics (1)
- Benjamins (1)
- Buro van die WAT (1)
- Buske (1)
- CLARIN (1)
- Digital Curation Centre (1)
- Institut für Romanistik der Universität Wien & Institut für Schallforschung der Österreichischen Akademie der Wissenschaften (1)
- Institute of Croatian Language and Linguistics (1)
- Lang (1)
Sogenannte „Pragmatikalisierte Mehrworteinheiten“ sind im Deutschen hochfrequent und unterliegen bisweilen tiefgreifenden phonetischen Reduktionsprozessen. Diese können Realisierungsvarianten hervorbringen, die in der Rückschau auf mehr als eine lexematische Ursprungsform zurückführbar sind. Die vorliegende Studie untersucht mit [ˈzɐmɐ] einen besonders prägnanten Fall dieser Art anhand eines Perzeptionsexperimentes.
CLARIN contractual framework for sharing language data: the perspective of personal data protection
(2020)
The article analyses the responsibility for ensuring compliance with the General Data Protection Regulation (GDPR) in research settings. As a general rule, organisations are considered the data controller (responsible party for the GDPR compliance). Research constitutes a unique setting influenced by academic freedom. This raises the question of whether academics could be considered the controller as well. However, there are some court cases and policy documents on this issue. It is not settled yet. The analysis serves a preliminary analytical background for redesigning CLARIN contractual framework for sharing data.
We present web services which implement a workflow for transcripts of spoken language following the TEI guidelines, in particular ISO 24624:2016 “Language resource management – Transcription of spoken language”. The web services are available at our website and will be available via the CLARIN infrastructure, including the Virtual Language Observatory and WebLicht.
Im Beitrag steht das LeGeDe-Drittmittelprojekt und der im Laufe der Projektzeit entwickelte korpusbasierte lexikografische Prototyp zu Besonderheiten des gesprochenen Deutsch in der Interaktion im Zentrum der Betrachtung. Die Entwicklung einer lexikografischen Ressource dieser Art knüpft an die vielfältigen Erfahrungen in der Erstellung von korpusbasierten Onlinewörterbüchern (insbesondere am Leibniz-Institut für Deutsche Sprache, Mannheim) und an aktuelle Methoden der korpusbasierten Lexikologie sowie der Interaktionsanalyse an und nimmt als multimedialer Prototyp für die korpusbasierte lexikografische Behandlung von gesprochensprachlichen Phänomenen eine innovative Position in der modernen Onlinelexikografie ein. Der Beitrag befasst sich im Abschnitt zur LeGeDe-Projektpräsentation ausführlich mit projektrelevanten Forschungsfragen, Projektzielen, der empirischen Datengrundlage und empirisch erhobenen Erwartungshaltungen an eine Ressource zum gesprochenen Deutsch. Die Darstellung der komplexen Struktur des LeGeDe-Prototyps wird mit zahlreichen Beispielen illustriert. In Verbindung mit der zentralen Information zur Makro- und Mikrostruktur und den lexikografischen Umtexten werden die vielfältigen Vernetzungs- und Zugriffsstrukturen aufgezeigt. Ergänzend zum abschließenden Fazit liefert der Beitrag in einem Ausblick umfangreiche Vorschläge für die zukünftige lexikografische Arbeit mit gesprochensprachlichen Korpusdaten.
Older adults are often exposed to elderspeak, a specialized speech register linked with negative outcomes. However, previous research has mainly been conducted in nursing homes without considering multiple contextual conditions. Based on a novel contextually-driven framework, we examined elderspeak in an acute general versus geriatric German hospital setting. Individuallevel information such as cognitive impairment (CI) and audio-recorded data from care interactions between 105 older patients (M = 83.2 years; 49% with severe CI) and 34 registered nurses (M = 38.9 years) were assessed. Psycholinguistic analyses were based on manual coding (k = .85 to k = .97) and computer-assisted procedures. First, diminutives (61%), collective pronouns (70%), and tag questions (97%) were detected. Second, patients’ functional impairment emerged as an important factor for elderspeak. Our study suggests that functional impairment may be a more salient trigger of stereotype activation than CI and that elderspeak deserves more attention in acute hospital settings.
We present a fine-grained NER annotations scheme with 30 labels and apply it to German data. Building on the OntoNotes 5.0 NER inventory, our scheme is adapted for a corpus of transcripts of biographic interviews by adding categories for AGE and LAN(guage) and also adding label classes for various numeric and temporal expressions. Applying the scheme to the spoken data as well as a collection of teaser tweets from newspaper sites, we can confirm its generality for both domains, also achieving good inter-annotator agreement. We also show empirically how our inventory relates to the well-established 4-category NER inventory by re-annotating a subset of the GermEval 2014 NER coarse-grained dataset with our fine label inventory. Finally, we use a BERT-based system to establish some baselines for NER tagging on our two new datasets. Global results in in-domain testing are quite high on the two datasets, near what was achieved for the coarse inventory on the CoNLLL2003 data. Cross-domain testing produces much lower results due to the severe domain differences.
This paper presents experiments on sentence boundary detection in transcripts of spoken dialogues. Segmenting spoken language into sentence-like units is a challenging task, due to disfluencies, ungrammatical or fragmented structures and the lack of punctuation. In addition, one of the main bottlenecks for many NLP applications for spoken language is the small size of the training data, as the transcription and annotation of spoken language is by far more time-consuming and labour-intensive than processing written language. We therefore investigate the benefits of data expansion and transfer learning and test different ML architectures for this task. Our results show that data expansion is not straightforward and even data from the same domain does not always improve results. They also highlight the importance of modelling, i.e. of finding the best architecture and data representation for the task at hand. For the detection of boundaries in spoken language transcripts, we achieve a substantial improvement when framing the boundary detection problem as a sentence pair classification task, as compared to a sequence tagging approach.
Interaktionale Semantik
(2020)
I’ve got a construction looks funny – representing and recovering non-standard constructions in UD
(2020)
The UD framework defines guidelines for a crosslingual syntactic analysis in the framework of dependency grammar, with the aim of providing a consistent treatment across languages that not only supports multilingual NLP applications but also facilitates typological studies. Until now, the UD framework has mostly focussed on bilexical grammatical relations. In the paper, we propose to add a constructional perspective and discuss several examples of spoken-language constructions that occur in multiple languages and challenge the current use of basic and enhanced UD relations. The examples include cases where the surface relations are deceptive, and syntactic amalgams that either involve unconnected subtrees or structures with multiply-headed dependents. We argue that a unified treatment of constructions across languages will increase the consistency of the UD annotations and thus the quality of the treebanks for linguistic analysis.
Lean syntax: how argument structure is adapted to its interactive, material, and temporal ecology
(2020)
It has often been argued that argument structure in spoken discourse is less complex than in written discourse. This paper argues that lean argument structure, in particular, argument omission, gives evidence of how the production and understanding of linguistic structures is adapted to the interactive, material, and temporal ecology of talk-in-interaction. It is shown how lean argument structure builds on participants' ongoing bodily conduct, joint perceptual salience, joint attention, and their Orientation to expectable next actions within a joint project. The phenomena discusscd in this paper are verb-derived discourse markers and tags, analepsis in responsive actions, and ellipsis in first actions, such as requests and instructions. The study draws from transcripts and audio- and video-recordings of naturally occurring interaction in German from the Research and Teaching Corpus of Spoken German (FOLK).
In diesem Beitrag werden exemplarisch verschiedene potenzielle Gebrauchsmuster mit dem deutschen Lemma wissen gesammelt und ihre in der Fachliteratur vorgelegten interaktionslinguistisch-funktionalen Beschreibungen für einen Strukturierungsversuch genutzt. Im Zentrum steht ein multifunktionaler handlungsorientierter Ansatz zur Beschreibung von Interaktion im Gespräch. Der Beitrag greift dabei Überlegungen auf, die im Rahmen des Forschungsprojekts Lexik des gesprochenen Deutsch (= LeGeDe) zur Erstellung einer korpusbasierten lexikogra- fischen Ressource lexikalischer Besonderheiten des gesprochenen Deutsch in der Interaktion thematisiert wurden.
Schlüsselwörter: Muster, Lexik des gesprochenen Deutsch, Interaktion, Internetlexikografie
Der Band leistet eine theoretisch begründete und empirisch validierte Entwicklung einer automatisierten Wortartenannotation (Part-of-Speech-Tagging) für Transkripte spontansprachlicher Daten des Forschungs- und Lehrkorpus Gesprochenes Deutsch (FOLK), das über die Datenbank für Gesprochenes Deutsch der Forschungsgemeinschaft öffentlich zugänglich ist. Dabei setzt er zwei Schwerpunkte: erstens die theoretische Aufarbeitung von Unterschieden von Transkripten gesprochener Sprache zu schriftsprachlichen Daten in Hinblick auf die Entwicklung eines Tagsets für das gesprochene Deutsch; zweitens die Darstellung der empirischen Arbeitsschritte zur Erstellung des automatisierten Part-of-Speech-Taggings, d. h. die Implementierung und Evaluierung für die Annotation des FOLK-Korpus. Der Band ist eine kritische Reflexion der Wortartentheorien im Spannungsfeld zwischen Theorie und datengeleiteter Arbeit. Er gibt Einblicke über die Korpusaufbereitung von Transkripten gesprochener Sprache und stellt diese in Bezug zu Theorien über die Eigenheiten gesprochener Sprache.
This paper studies practices of indexing discrepant assumptions accomplished by turn-constructional units with ich dachte ('I thought') in German talk-in-interaction. Building on the analysis of 141 instances from the corpus FOLK, we identify three sequential environments in which ich dachte is used to index that an assumption which a speaker (has) held contrasts with some other, contextually salient assumption. We show that practices which have been studied for English I thought are also routinely used in German: ich dachte is a means to manage epistemic incongruencies and to contrast an incorrect with a correct assumption in narratives. In addition, ich dachte is also used to account for the speaker's own prior actions which may have looked problematic because they built on misunderstandings which the speaker only discovered later. Moreover, ich dachte-practices may also be used to create comic effects by reporting an earlier, absurd assumption. The practices are discussed with regard to their role in regaining common ground, in managing relationships, in maintaining the identity of a rational actor, and in terms of their exploitation for other conversational interests. Special attention is paid to how co-occurring linguistic features, and sequential and pragmatic factors, account for local interpretations of ich dachte.
This article describes the development of the digital infrastructure at a research data centre for audio-visual linguistic research data, the Hamburg Centre for Language Corpora (HZSK) at the University of Hamburg in Germany, over the past ten years. The typical resource hosted in the HZSK Repository, the core component of the infrastructure, is a collection of recordings with time-aligned transcripts and additional contextual data, a spoken language corpus. Since the centre has a thematic focus on multilingualism and linguistic diversity and provides its service to researchers within linguistics and other disciplines, the development of the infrastructure was driven by diverse usage scenarios and user needs on the one hand, and by the common technical requirements for certified service centres of the CLARIN infrastructure on the other. Beyond the technical details, the article also aims to be a contribution to the discussion on responsibilities and services within emerging digital research data infrastructures and the fundamental issues in sustainability of research software engineering, concluding that in order to truly cater to user needs across the research data lifecycle, we still need to bridge the gap between discipline-specific research methods in the process of digitalisation and generic digital research data management approaches.
The lexicography of German
(2020)
This chapter discusses the main dictionaries of the German language as it is spoken and written in Germany, and also German as it is spoken and written in Austria, Switzerland, the eastern fringes of Belgium, and South Tyrol. It also briefly describes Pennsylvania German. Corpora and other language resources used in German dictionary-making are also presented. Finally, there is a discussion of some current issues in German lexicography, as well as future prospects.
This paper presents the corpus-based lexicographical prototype that was developed within the framework of the project Lexik des gesprochenen Deutsch (=LeGeDe) as a thirdparty funded project. Research results regarding the information offered in dictionaries have shown that there is a necessity for information on spoken lexis and its interactional functions. The resulting LeGeDe-prototype is based on these needs and desiderata and is thus an innovative example for the adequate representation of spoken language in online dictionaries. It is available online since September 2019 (https://www.owid.de/legede/). In the following sections, after first focusing on the presentation of the project’s goals, the data basis, the intended end user, and the applied methods, we will illustrate the microstructure of the prototype and the information provided in a dictionary entry based on the lemma eben. Finally, we will summarize innovative aspects that are important for the implementation of such a resource.
This paper describes the development of a systematic approach to the creation, management and curation of linguistic resources, particularly spoken language corpora. It also presents first steps towards a framework for continuous quality control to be used within external research projects by non-technical users, and discuss various domain and discipline specific problems and individual solutions. The creation of spoken language corpora is not only a time-consuming and costly process, but the created resources often represent intangible cultural heritage, containing recordings of, for example, extinct languages or historical events. Since high quality resources are needed to enable re-use in as many future contexts as possible, researchers need to be provided with the necessary means for quality control. We believe that this includes methods and tools adapted to Humanities researchers as non-technical users, and that these methods and tools need to be developed to support existing tasks and goals of research projects.
The newest generation of speech technology caused a huge increase of audio-visual data nowadays being enhanced with orthographic transcripts such as in automatic subtitling in online platforms. Research data centers and archives contain a range of new and historical data, which are currently only partially transcribed and therefore only partially accessible for systematic querying. Automatic Speech Recognition (ASR) is one option of making that data accessible. This paper tests the usability of a state-of-the-art ASR-System on a historical (from the 1960s), but regionally balanced corpus of spoken German, and a relatively new corpus (from 2012) recorded in a narrow area. We observed a regional bias of the ASR-System with higher recognition scores for the north of Germany vs. lower scores for the south. A detailed analysis of the narrow region data revealed – despite relatively high ASR-confidence – some specific word errors due to a lack of regional adaptation. These findings need to be considered in decisions on further data processing and the curation of corpora, e.g. correcting transcripts or transcribing from scratch. Such geography-dependent analyses can also have the potential for ASR-development to make targeted data selection for training/adaptation and to increase the sensitivity towards varieties of pluricentric languages.
The newest generation of speech technology caused a huge increase of audio-visual data nowadays being enhanced with orthographic transcripts such as in automatic subtitling in online platforms. Research data centers and archives contain a range of new and historical data, which are currently only partially transcribed and therefore only partially accessible for systematic querying. Automatic Speech Recognition (ASR) is one option of making that data accessible. This paper tests the usability of a state-of-the-art ASR-System on a historical (from the 1960s), but regionally balanced corpus of spoken German, and a relatively new corpus (from 2012) recorded in a narrow area. We observed a regional bias of the ASR-System with higher recognition scores for the north of Germany vs. lower scores for the south. A detailed analysis of the narrow region data revealed – despite relatively high ASR-confidence – some specific word errors due to a lack of regional adaptation. These findings need to be considered in decisions on further data processing and the curation of corpora, e.g. correcting transcripts or transcribing from scratch. Such geography-dependent analyses can also have the potential for ASR-development to make targeted data selection for training/adaptation and to increase the sensitivity towards varieties of pluricentric languages.
As a part of the ZuMult-project, we are currently modelling a backend architecture that should provide query access to corpora from the Archive of Spoken German (AGD) at the Leibniz-Institute for the German Language (IDS). We are exploring how to reuse existing search engine frameworks providing full text indices and allowing to query corpora by one of the corpus query languages (QLs) established and actively used in the corpus research community. For this purpose, we tested MTAS - an open source Lucene-based search engine for querying on text with multilevel annotations. We applied MTAS on three oral corpora stored in the TEI-based ISO standard for transcriptions of spoken language (ISO 24624:2016). These corpora differ from the corpus data that MTAS was developed for, because they include interactions with two and more speakers and are enriched, inter alia, with timeline-based annotations. In this contribution, we report our test results and address issues that arise when search frameworks originally developed for querying written corpora are being transferred into the field of spoken language.