Chinese Intelligent Language Learning (CHILL) – Using Grammar Engineering Models for Automated Error Detection

Projektinformationen

CHILL

ID Finanzhilfevereinbarung: 101028782

DOI

10.3030/101028782

Das Projekt endete am 15 Januar 2023

EK-Unterschriftsdatum 24 März 2021

Startdatum 1 November 2021

Enddatum 31 Oktober 2023

Finanziert unter

• EXCELLENT SCIENCE - Marie Skłodowska-Curie Actions

Gesamtkosten

€ 144 980,64

EU-Beitrag

€ 144 980,64

Koordiniert durch

UNIVERZITA PALACKEHO V OLOMOUCI
Czechia

Periodic Reporting for period 1 - CHILL (Chinese Intelligent Language Learning (CHILL) – Using Grammar Engineering Models for Automated Error Detection)

Berichtszeitraum: 2021-11-01 bis 2023-10-31

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

This project addressed the disconnection between recent advances in Computer Assisted Language Learning and the actual needs of learners and teachers. At a time where digitization is increasingly becoming a primary necessity in many areas of society, education – and in particular language education – have faced challenges in the way technology has kept up with the actual needs of learners and educators.

This project addressed some of these problems by producing technology specifically targeting the pedagogical dimensions currently lacking in the existing infrastructure for online language learning and teaching. In particular, it worked towards building to build a high-precision error detection computational system for Mandarin Chinese, building on decades of work in linguistic theory, and implementing a computational grammar capable of not only producing high quality parses for Mandarin Chinese, but also exploit this implemented grammar to diagnose, with linguistic precision, the source of grammatical errors.

In particular, its overall objectives included:
Contributions in the syntactical/theoretical analysis Mandarin Noun Phrases;
Computational implementation of Mandarin Chinese grammar, with a particular emphasis in Noun Phrases;
Surveying common error classes made by Mandarin Chinese learners;
Adaptation of a Mandarin Chinese computational grammar into an error detection system, capable of detecting a variety of common grammatical errors, with particular emphasis in Noun Phrases

These objectives were to be met using well-grounded data-driven interdisciplinary methodologies, integrating standard methodologies for formal linguistic analysis and methodologies for software development from the field of Computer Science.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

This project was interrupted on month 15 (of 24), because the Research Fellow took a job offer for a tenure-track position in Computational Linguistics, which was in line with his overt plans to establish an academic career in Europe – defined as his goal career in the Career Development Plan, developed in the context of this action, and also well within the spirit behind the Reintegration Panel, to which this action belonged.

Nevertheless, the project execution was already in advanced stages of implementation, and even ahead of schedule in some areas.

The project had achieved two of its three milestones, signaling completion of:

Milestone 1: included an extensive literature review on the syntactic analysis of Mandarin Chinese, with particular focus on its NP structure; and data collection efforts to enable the data-driven methodology espoused by the project. Data collection sources included educational materials, data from linguistic papers, and from language consultants, and covered both proper/grammatical language as well as data illustrating common grammatical errors made by learners of Mandarin Chinese;

Milestone 2: included the formal syntactical analysis of the Mandarin Chinese NP, which culminated with a publication of a peer reviewed paper on the formal syntactical analysis of Mandarin Chinese in the context of grammatical error detection of common errors made by learners; In addition to this publication, this milestone also included the deployment of the online demo in which the computational implementation can be incrementally tested, and which also serves as an instrument for the dissemination of results;

In addition to research related results, the project was also very successful in the plans for its two-way transfer of knowledge (to the Research Fellow and to the host institution) as well as in furthering the professional maturity/independence of the Research Fellow during the fellowship.

Concerning dissemination efforts, even though the project was terminated early, it was ahead of schedule concerning its dissemination plans. The Research Fellow actively participated in 7 workshops/conferences/seminars, and published 8 peer-reviewed papers (3 of them, directly tied to the project, and 5 of them peripheral – setting new avenues to expand the project’s topic). The Research Fellow also co-taught 2 courses (Topics in Chinese Syntax and Semantics, Chinese Semantics and Lexicology) where parts of his research were featured as course content.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

Even though the project had an early termination, it is likely to have long-term impact in the formulation of future research projects and in direct benefits in Mandarin Chinese classrooms. Tied to its dissemination efforts, this project presents “computational grammars” and “learner treebanks” as worthy avenues to pursue with the goal of improving educational technology capable of providing direct feedback to students about common grammatical mistakes. This technology has been very well received by researchers and educators alike, and was presented as a main research agenda point that eventually led to the Research Fellow’s tenure-track position in Computational Linguistics. In particular, there is great interest in integrating the technology produced during this project into cognitive robots – opening new avenues to the way this technology can be delivered (i.e. going beyond the web-browser), and testing new grounds of robot-human interactions.

Concerning wider policy objectives and strategies, the work in this project has contributed to multiple goals set forth by Europe and NGOs. In particular, the main contributions of this project work towards the improvement of a practically non-existing infrastructure to sustain language education in the digital age. On the one hand, the conducted research can be seen as improving sustainability and accessibility goals, by making education depend less on expensive human resources and facilitating accessibility by delivering it through cheap and widespread digital mediums (i.e. internet). On the other hand, the research conducted through this project also contributes to wider goals of improvised quality of education, by building systems that can target students’ individual goals and weaknesses, making education more personalized and enjoyable, and built on sound principles of active learning (e.g. learning by guided correction of their own mistakes).

msca-upol-chill.jpg