Language Assessment for Children with a Migrant Background

Language is pivotal for educational success and, thus, lifetime careers. Educational systems, as a result, rely on various assessment tools to determine kids’ readiness and potential need for supportive measures before the kids enter school. However, a number of recent studies have found that the many existing tools in German-speaking countries largely lack the ability to assess the proper communicative potential of 4- to 5-year-olds (e.g. Neugebauer & Becker-Mrotzek 2013).

This R&D project aims at remedying the shortcomings produced by existing assessment tools by developing a new game-based language assessment procedure for children aged four to five years with German as a first and second language. The goal is to generate an accurate and fair assessment of the children’s communicative competencies with a special focus on detecting educational needs of, and on improving educational opportunities for, children with a migration background.

The new assessment is theoretically grounded in usage-based linguistics and incorporates language learning studies focusing on children’s authentic language production. Particularly, it aims at evaluating bilingual children’s language competencies in the L2 German in authentic communicative settings in the linguistically and conceptually challenging domain of spatial language. As a result, the assessment tool does not evaluate formal aspects of language. Instead, it relies on functional categories for evaluation. In order to avoid traditional examiner-examinee test situations an interactive app for mobile assessment following the principles of Serious Games was developed. The new assessment tool enables the kids to play the game on a tablet by themselves while the (practically invisible) examiner is nearby controlling the child’s tablet via a second device (ipod). Using a Serious Game environment, children are assigned the role of the expert whose prime job is to instruct his/her buddy (a cute but not always smart dog) to save the benign witch from various dangers. In order to manage this task, the child talks to the dog, a cartoon-like character on the tablet. In order to elicit children’s authentic linguistic competencies and potentials, test items are embedded in communicatively relevant situations applying a child-friendly background story.
For data analysis, the tablet records the children’s verbal reactions to the prompts. After being transmitted to the web-based editor OCTRA these audio snippets are transcribed using a crowdsourcing system and are analyzed in terms of spatio-semantic and functional categories. Analyzed transcripts allow the project team to determine to what extent the children’s educational needs may be assessed using semi-automatic analysis tools. Preliminary results show fewer than expected differences between kids with L1 and those with L2 German, even when contact time to German was as short as 16 months. Kids are vividly involved in developing the story and, thus, produce much more language than with traditional assessment tools. They often lack the same degree of precision produced by L1 speakers of German but, to a large degree, utterances are grammatically acceptable. The project is in the process of adding more domains (different discourse types, definiteness, possession), further developing the linguistic assessment procedures and conducting more pilot testing. The app is supposed to enter service in 2019, in conjunction with a larger standard-setting project.

Initiator: Wolfgang Klein, Max-Planck-Institut für Psycholinguistik, Nijmegen

Principle Investigator: Jörg Roche
Project Coordination: Nicole Weidinger
Staff LMU: Svenja Uth (since 12/2017), Christian Meyer (since 02/2017)

Partners:
Heike Behrens, Universität Basel
Stefanie Haberzettl, Universität des Saarlandes
Marcus Hasselhorn, DIPF Frankfurt
Dirk Ifenthaler, Universität Mannheim
Natalia Kapica, Universität Heidelberg
Gabriele Kecker, TestDaF-Institut Bochum
Karin Madlener, Universität Basel (until 10/2016)
Giulio Pagonis, Universität Heidelberg
Maike Schug, Universität des Saarlandes (until 02/2018)
Katrin Skoruppa, Universität Basel (until 09/2016)
Frank Thissen, Hochschule der Medien Stuttgart (until 10/2015)
Wolfgang Woerner, DIPF Frankfurt (since 06/2016)
Moiken Jessen, LMU München (until 07/2017)
Elisabetta Terrasi-Haufe, LMU München (until 12/2015)
Christoph Draxler, LMU München (since 12/2016)
Jan Delcker, Universität Mannheim (since 01/2016)

Students Assistents:
Sarah Faidt, LMU München
Agnieszka Kubacka-Mauer, LMU München