Analyzing Authentic Texts for Language Learning. Web-based Technology for Input Enrichment and Question Generation.

Autor/in	Chinkina, Maria
Titel	Analyzing Authentic Texts for Language Learning. Web-based Technology for Input Enrichment and Question Generation.
Quelle	Tübingen: Universitätsbibliothek Tübingen (2018), xxiii, 198 S. PDF als Volltext kostenfreie Datei (1); PDF als Volltext kostenfreie Datei (2) Link als defekt meldenVerfügbarkeit Dissertation, Eberhard Karls Universität Tübingen, 2018.
Sprache	englisch
Dokumenttyp	online; Monographie
URN	urn:nbn:de:bsz:21-dspace-833918
Schlagwörter	Lernen; Sprachverarbeitung; Computerlinguistik; Technologie; Dissertation
Abstract	Acquisition of a language largely depends on the learner's exposure to and interaction with it. Our research goal is to explore and implement automatic techniques that help create a richer grammatical intake from a given text input and engage learners in making form-meaning connections during reading. A starting point for addressing this issue is the automatic input enrichment method, which aims to ensure that a target structure is richly represented in a given text. We demonstrate the high performance of our rule-based algorithm, which is able to detect 87 linguistic forms contained in an official curriculum for the English language. Showcasing the algorithm's capability to differentiate between the various functions of the same linguistic form, we establish the task of tense sense disambiguation, which we approach by leveraging machine learning and rule-based methods. Using the aforementioned technology, we develop an online information retrieval system FLAIR that prioritizes texts with a rich representation of selected linguistic forms. It is implemented as a web search engine for language teachers and learners and provides effective input enrichment in a real-life teaching setting. It can also serve as a foundation for empirical research on input enrichment and input enhancement. The input enrichment component of the FLAIR system is evaluated in a web-based study that demonstrates that English teachers prefer automatic input enrichment to standard web search when selecting reading material for class. We then explore automatic question generation for facilitating and testing reading comprehension as well as linguistic knowledge. We give an overview of the types of questions that are usually asked and can be automatically generated from text in the language learning context. We argue that questions can facilitate the acquisition of different linguistic forms by providing functionally driven input enhancement, i.e., by ensuring that the learner notices and processes the form. The generation of well-established and novel types of questions is discussed and examples are provided; moreover, the results from a crowdsourcing study show that automatically generated questions are comparable to human-written ones. (Orig.).
Erfasst von	Deutsche Nationalbibliothek, Frankfurt am Main
Update	2020/2