Research

Revita spans multiple disciplines and research areas:
Research:

Revita tries to emulate a good tutor/teacher. A good teacher knows:

  1. the subject domain,
  2. the student,
  3. how to teach — given what we know about each learner, what are the best exercises to offer to them next, to optimize the learning outcomes.

In Educational Data Science, these are handled, respectively, by:

  1. the Domain Model: this component defines the set of "language constructs" — also known as "skills" — to be learned, for each language. There are  hundreds of skills the learner needs to master to reach advanced proficiency.
  2. the Student Model: this component performs continual assessment of the learner's skills. We need to assess learner skills quickly and accurately.
  3. the Instruction Model: this component is responsible for the personalization: based on what we know about the learner, and about the common learning paths — what are the next exercises, optimally suited to the learner's skills and needs.  The tutor guides the learner toward the correct answer by providing personalized feedback in case of incorrect answers, to help the learner arrive at the correct answer on her own (rather than simply giving away the correct answer).

     

We aim to address both written and oral skills. For oral skills, we develop components to analyze the students' ability to process spoken language. This will allow us to test hypotheses about the mechanisms for processing audio input by learners, and to train this ability through a personalized path for each student.

For additional information, please consult the Project's publications.

Also, please feel free to contact the Revita team directly!
Collaborators and Credits:

Revita builds on resources and tools developed by our many international colleagues and collaborators.

With our collaborators we develop components to support learning various languages.

Some important resources that Revita builds upon, and content providers, who have granted us permission to use their resources:

  • The GiellaTekno platform: language technology for Uralic languages, and endangered languages from other language families.
  • The Apertium platform for languages from the Uralic, Turkic and other language families.
  • Morphisto—the German morphological analyzer.
  • CrosslatorTagger for Russian, by Professor Eduard Klyshinskiy (Higher School of Economics, Moscow, Russia).
  • For Uralic and Turkic minority languages, CrosslatorTagger is used to detect code-switching into Russian in authentic texts.
  • SakhaTyla.Ru—portal for the Sakha (Yakut) language, providing analyzers and dictionaries.
  • ResponsiveVoice used under Non-Commercial License.
  • Stress library (Russian) by Rob Reynolds, Assistant Research Professor, Brigham Young University, Utah.
  • Icons by Madebyoliver from Flaticons and Rohan Gupta from the Noun Project.

The Revita Project is supported in part by:

  • Academy of Finland, Research Council for Culture and Society (Grant 267097)
  • Opetushallitus: The Finnish National Agency for Education (Grant OPH-1443-2020, TM-18-10846)
  • CIMO: Center for International Mobility (Grant TM-16-10082)
  • HIIT: Helsinki Institute for Information Technology
  • University of Pisa, Italy: ErasmusPlus Programme of the European Commission
  • Tulevaisuus Rahasto 2020,2022: Future Development Fund, Faculty of Arts, University of Helsinki