Refine
Year of publication
- 2020 (60) (remove)
Document Type
- Part of a Book (26)
- Article (19)
- Book (6)
- Conference Proceeding (4)
- Other (3)
- Part of Periodical (1)
- Working Paper (1)
Is part of the Bibliography
- yes (60) (remove)
Keywords
- Deutsch (60) (remove)
Publicationstate
- Zweitveröffentlichung (30)
- Veröffentlichungsversion (22)
- Postprint (6)
Reviewstate
- (Verlags)-Lektorat (33)
- Peer-Review (16)
Publisher
We present recognizers for four very different types of speech, thought and writing representation (STWR) for German texts. The implementation is based on deep learning with two different customized contextual embeddings, namely FLAIR embeddings and BERT embeddings. This paper gives an evaluation of our recognizers with a particular focus on the differences in performance we observed between those two embeddings. FLAIR performed best for direct STWR (F1=0.85), BERT for indirect (F1=0.76) and free indirect (F1=0.59) STWR. For reported STWR, the comparison was inconclusive, but BERT gave the best average results and best individual model (F1=0.60). Our best recognizers, our customized language embeddings and most of our test and training data are freely available and can be found via www.redewiedergabe.de or at github.com/redewiedergabe.
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.
Aus diesem Grunde haben wir uns empirisch der Frage genähert, wie oder ob bestimmte Gruppen heute überhaupt noch Wörterbücher nutzen und ob sie sie bewusst von anderen sprachbezogenen Daten im Web unterscheiden. Es sollten empirische Daten gesammelt werden, um zu erfahren, wie DaF-Lernende tatsächlich arbeiten (und nicht was sie dazu retrospektiv sagen), vor allem um eine bessere empirische Basis für den Unterricht zur Verfügung zu haben. Zentrale Fragen dabei waren:
• Wie nutzen DaF-Lernende heutzutage lexikografische Ressourcen?
• Welche Suchstrategien wenden sie an?
• Differenzieren sie zwischen den unterschiedlichen Ressourcen?
• Welche Strategien erweisen sich als besonders erfolgreich?
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
Lexikonprojektion und Konstruktion: Experimentelle Studien zu Argumentalternationen im Deutschen
(2020)
Debates on lexicalist vs. constructionist modelling of argument alternations are typically based on data from single constructions, each including different types of verbs. Evidence from constructions with an identical set of verb types that systematically differ in their meaning is lacking, even though such evidence is imperative for specifically investigating the dependence of argument alternations on the interaction between construction and lexical meanings. We present two acceptability studies where verb lexeme meanings and constructions - specifically active voice, impersonal passive and the construction with man 'one' in German - vary systematically. Prima facie our results support a constructionist explanation, because each construction exhibits a unique acceptability cline. However, across constructions, an adequate explanation has to consider verb-based lexical meanings. The most plausible explanation is that the semantic features licensed by the construction are matched with the semantic features provided by the verb lexeme.
Objekte der Begeisterung
(2020)
We present a construction-based approach to German prepositional object (I’O) constructions occurring with the verb begeistern ,to thrill'. Traditionally, the preposition in such structures is analysed as a meaningless object marker that is lexically selected by the governing verb and not subject to variation. Drawing on a corpus study in the German reference corpus DeReKo, we show that our target verb occurs with four different PO prepositions (für ,lor‘,« ׳? ,at', von ,front' and über ,over‘) that can be analysed as markers o f schematic argument structure constructions in the Construction Grammar sense. We show that each construction comes with its own meaning and semantically coherent predicate restrictions. We argue that purely valency-based (lexical) approaches to argument structure fail to capture these generalisations. On the other hand, purely schema-based (constructionist) approaches to argument structure face the complcmentary problem o f accommodating item-specific restrictions and exceptions to the generalisations they embody. We suggest that the necessary synthesis can be formulated within an account that recognises both generalised constructions and item-specific valency properties.
Prosodische Morphologie
(2020)
The annual microcensus provides Germany’s most important official statistics. Unlike a census it does not cover the whole population, but a representative 1%-sample of it. In 2017, the German microcensus asked a question on the language of the population, i.e. ‘Which language is mainly spoken in your household?’ Unfortunately, the question, its design and its position within the whole microcensus’ questionnaire feature several shortcomings. The main shortcoming is that multilingual repertoires cannot be captured by it. Recommendations for the improvement of the microcensus’ language question: first and foremost the question (i.e. its wording, design, and answer options) should make it possible to count multilingual repertoires.