What this project is about
Pronunciation as a subject of research and teaching
Pronunciation is a central part of foreign language learning, but it is often treated less systematically than vocabulary, grammar, or textual competence in research, higher education, and school practice. For learners, however, pronunciation is not a marginal issue. It affects intelligibility, listening perception, confidence in speaking, and the question of how linguistic variation can be addressed meaningfully in the classroom.
This is where Pronunciation Matters begins. The project records, structures, and makes accessible speech data from learners so that it can be used specifically for research and teaching. The aim is not simply to evaluate pronunciation against a supposed native-speaker standard. What matters instead are intelligibility, systematic patterns, recurring difficulties, linguistic variation, and the question of how pronunciation can be addressed in foreign language teaching in ways that are transparent, comparable, and appropriate for learners.
The platform connects three levels: empirical research on learner pronunciation, research-based university teaching, and the development of materials for school contexts. Students do not only work with finished examples; they gain insight into how authentic speech data is created, processed, and analysed. Teachers and learners can benefit from materials that are developed out of this research process.
From a pilot project to a multilingual platform
The starting point for Pronunciation Matters was the Spanish pilot project MAR.ELE. Between 2024 and 2025, this project collected recordings of learners of Spanish at Philipps-Universität Marburg, processed them, and made them accessible through a web app. MAR.ELE showed that authentic learner pronunciation can be documented and also used productively for linguistic analysis, subject-specific didactic discussion, and teaching materials.
At the same time, MAR.ELE made clear where a small pilot project reaches its limits. The project initially connected with existing corpus-phonological research projects; in particular, it adopted the wordlist from the (I)FEC project in order to build on an established design for describing Spanish pronunciation variation. In the work with learners, however, it became clear that such materials cannot simply be transferred to a corpus of learner pronunciation. Some items were lexically unnecessarily difficult, some phenomena relevant to learner pronunciation were not covered often enough, and not every established task format fitted an elicitation design that needs to produce comparable data while remaining manageable for learners.
Pronunciation Matters builds on these experiences. The project uses existing research designs as important points of orientation, but adapts task formats and materials specifically for work with learners. This creates a platform structure in which data collection, processing, analysis, and didactic use are considered together from the outset.
The platform begins with the four areas represented in Marburg: Spanish, French, English, and German as a foreign or second language. The project follows a shared basic design: recordings are collected under comparable conditions, task formats are systematic, metadata is recorded transparently, and research data is pseudonymised and transferred into a common platform structure. At the same time, this design needs to be adapted for each language. Not every language requires the same materials, and not every task format is equally suitable for every learning context. For Spanish, French, and German, controlled sentence lists with easily comprehensible individual utterances may be appropriate, whereas in English an established connected text may be more suitable because English as a foreign language is learned, read, and taught under different conditions. The individual language corpora therefore remain comparable without artificially smoothing over their disciplinary differences.
Teaching, research, and transfer
Pronunciation Matters is designed as a teaching and research project. Since January 2026, the multilingual platform has been built in concrete terms; the first recordings have been carried out since early March 2026. In the summer semester of 2026, the project is embedded in four courses, each related to one of the project languages. In these courses, research, data collection, analysis, and material development are not separated from one another, but experienced as parts of one connected work process.
This is especially important for university teaching. Students can work with authentic research data without encountering it only as finished examples. They learn how recordings are made, how data is structured, which decisions are involved in task formats and annotation, and how didactic materials can be developed from analytical findings. The project therefore connects subject-specific academic training, digital methods, and didactic reflection.
Transfer to school contexts is built into the project from the beginning. Some teaching materials are designed and written together with students in university courses. This is not about transferring research data into classrooms without mediation. Rather, suitable excerpts, tasks, and explanatory formats are developed from the research process to support pronunciation awareness, listening perception, and reflection on linguistic variation.
Award and further development
The idea for Pronunciation Matters was recognised and funded as an innovative teaching idea in November 2025 through the Lehre@Philipp ideas competition at Philipps-Universität Marburg. The award came with funding of a little over 10,000 euros. These funds made it possible to expand the pilot approach into a multilingual platform within roughly ten months. This expansion was made possible by the involvement of the participating colleagues, who support the project in their respective languages and embed it in teaching, data collection, and material development.
The aim of the funded project is not to complete a research programme, but to build a solid starting structure. Pronunciation Matters creates the platform, establishes first language corpora, and develops first teaching materials. This creates a shared working space for research, university teaching, and school-based mediation that is intended to grow further. The platform is meant as a starting point: for the participating disciplines in Marburg as well as for interested researchers, teachers, and students who may want to work with the data, tools, and materials in the future.