Invited presentations
Invited presentations
Embedded Files
In March 2025, the Center for Groningen Language and Culture (CGTC) and the University of Groningen launched a Massive Open Online Course for Gronings, in which over 700 people have participated so far. The course takes the shape of a romantic comedy, in which learners follow the story of two young speakers of Gronings who make a road trip and learn about both the region's heritage sites and its language (and language variation). This talk will address the design of the course and the way it relates to other online tools for Gronings such as the language database WoordWaark. Furthermore, some preliminary results from a study on the motivations of the learners in this course will be discussed.
This talk provides an overview of recent developments in digital tools for Luxembourgish, a small language with growing technological support. It begins with a brief introduction to Luxembourgish as one of the official languages of Luxembourg, highlighting how the establishment of an official orthographic standard in recent decades has acted as a catalyst for progress. As more written material has become available, these resources have enabled the training of machine learning models. I will present our work on Automatic Speech Recognition, Speaker Recognition, Machine Translation, and Speech Synthesis, followed by case studies demonstrating how these tools are being deployed in Luxembourgish institutions such as public administration and the media.
Gilles, Peter, Nina Hosseini Kivanani & Léopold Edem Ayité Hillah. 2023. ASRLux: Automatic speech recognition for the low-resource language Luxembourgish. In International Conference of Phonetic Sciences (ICPhS), August 7-11, 2023, 3091–3095. Prague: Guarant International. https://hdl.handle.net/10993/55819.
Gilles, Peter, Nina Hosseini Kivanani & Léopold Edem Ayité Hillah. 2023. ASRLux: Automatic speech recognition for the low-resource language Luxembourgish. In International Conference of Phonetic Sciences (ICPhS), August 7-11, 2023, 3091–3095. Prague: Guarant International. https://hdl.handle.net/10993/55819.
Since 1997, Limburgish in the Netherlands is recognized as a regional language under part II of the European Charter for Regional and Minority Languages. Following from this, the Dutch State is obliged to act against any discriminatory measures and to actively pursue policy in support of the language.
In 2020, the Council of Europe concluded that the Dutch state does not take enough measures to support Limburgish. A number of (mostly voluntary) pro-Limburgish organizations decided that a professional workforce is needed to give the language the attention it deserves. The Hoes veur ‘t Limburgs was founded in 2023 and became an official cooperative in 2025.
In their policy document Same veur ‘t Limburgs - Uitvoeringsprogramma Limburgse taal 2025-2027 the Province of Limburg suggests promoting Limburgish along two policy lines:
1. Actively promote Limburgish in all levels of (pre-school) education, developing campaigns to increase Limburgish' citizens linguistic awareness, and investing in a professional workforce (the Hoes veur ‘t Limburgs) that has a main role in the aforementioned actions.
2. Investigate the possibility of acquiring a part III recognition under the European Charter in 2030. A part III recognition would open up new legal possibilities for supporting Limburgish and would lead to greater financial support by the Dutch government. A seperate committee has been formed (Actiecommissie Limburgse Taal) to address this policy line.
Taking a holistic and inclusive multilingual perspective, the Hoes veur ‘t Limburgs actively normalizes the multilingualism inherent to the Province of Limburg. Citizens are encouraged to use all their home languages and to actively discuss and question (outdated) norms around language use and politeness. In this talk we discuss our main initiatives, our successes as well as some challenges we encounter in the process of growing the impact of our young institute.
Unlike other Dutch regions such as Brabant, Holland, or Friesland, Limburg has no historical roots predating 1795. The province was artificially created under French rule and remained politically unstable throughout the nineteenth century. Limburg was only fully integrated into the Dutch state relatively late, which allowed many cultural and linguistic ‘phenomena’ to persist. As a result, Limburg dialects have maintained a remarkably strong social and cultural presence - around 40% of the population report speaking Limburgish most frequently at home.Since 1997, the Netherlands has officially recognized Limburgish as a regional or minority language under Part II (minor recognition) of the European Charter for Regional and Minority Languages (ECRML). Yet UNESCO classifies these dialects, particularly the Ripuarian varieties, as endangered. Although the province forms a single administrative unit, the dialects grouped under Limburgish are highly diverse and gradually merge with neighboring varieties.In this talk, I discuss both the syntactic features of Limburg dialects and the construction of regional identity through language practices.
Selected presentations
Selected presentations
We present a tool that addresses a weakness in FieldWorks Language Explorer (SIL FLEx), a software widely used for interlinear morphological glossing of texts (IGT) in the field of language documentation, in particular that of low-resource languages. The weakness springs from FLEx being unable to import IGT texts with glosses kept intact. As a result, it reduces the interoperability with other software solutions used for IGT purposes such as ELAN, EXMARaLDA, or Field Linguist’s Toolbox.
Our tool is a Python application with a simple GUI that reads .flextext XML files and appends their contents to the FieldWorks project database file, another flavor of XML with the .fwdata extension. The tool has been successfully put to use to integrate ca. 450 Tundra and Forest Enets (iso: enh and enf; Samoyedic < Uralic) texts (ca. 140,000 tokens) originally glossed in Toolbox into an existing .fwdata file, and to import 33 Kalmyk texts (iso: xal; Mongolic) with interlinear glossing into a new FLEx project.
Our tool is a Python application with a simple GUI that reads .flextext XML files and appends their contents to the FieldWorks project database file, another flavor of XML with the .fwdata extension. The tool has been successfully put to use to integrate ca. 450 Tundra and Forest Enets (iso: enh and enf; Samoyedic < Uralic) texts (ca. 140,000 tokens) originally glossed in Toolbox into an existing .fwdata file, and to import 33 Kalmyk texts (iso: xal; Mongolic) with interlinear glossing into a new FLEx project.
Large Language Models (LLMs) have demonstrated a remarkable capacity to achieve state-of-the-art performance across a wide range of natural language tasks. However, their size requires significant computational and memory resources. To address this challenge, we investigate a practical approach to reduce their size by converting them into Small Language Models (SLMs) while keeping most of their capabilities. In this work, we use Singular Value Decomposition (SVD) together with sensitivity analysis to guide this process. SVD helps break down the large weight matrices into smaller low-rank parts, removing repeated or less useful information. Sensitivity analysis then tells us which layers or parts of the model are more important for accuracy. Layers that are less sensitive can be compressed more, while sensitive layers are kept with higher detail. By combining these two ideas, we can shrink LLMs into smaller models that are faster, lighter, and easier to deploy, but still perform well on language tasks.
Project group and researchers: Laura Kotavuopio (SA), Kristiina Jomppanen (ULap), Berit-Ellen Juuso (SA), Ulla Richardson (UJY), Coppélie Cocq (UU) and Pigga Keskitalo (ULap)
Sámi education has long suffered from a shortage of suitable learning materials, with teachers often relying on outdated resources or creating their own. Lohkanlihkku, a three-year Interreg Aurora development project, addresses this gap by developing a digital learning game for five Sámi languages, aimed at strengthening pupils’ literacy skills. The project is a collaboration between the University of Lapland, University of Jyväskylä, Umeå University, and Sámi University of Applied Sciences. The development process actively involves Sámi teachers, pupils, guardians, and policy makers to ensure the game reflects Sámi culture and educational realities. Teacher interviews provide data on the current state of Sámi elementary education and literacy instruction methodologies, guiding the game’s pedagogical design. The project also offers training and workshops for teachers to support implementation. The Lohkanlihkku game builds on the GraphoLearn (Finnish: Lukukupla), itself based on the extensively researched GraphoGame (Ekapeli). GraphoGame adapts exercises to pupils’ individual literacy levels, a feature incorporated into Lohkanlihkku. The game will be freely available for computers and mobile devices. This presentation outlines the project’s objectives, development process, and current progress, with a particular focus on the first work package: mapping the current state and practices of Sámi literacy education.
Sámi education has long suffered from a shortage of suitable learning materials, with teachers often relying on outdated resources or creating their own. Lohkanlihkku, a three-year Interreg Aurora development project, addresses this gap by developing a digital learning game for five Sámi languages, aimed at strengthening pupils’ literacy skills. The project is a collaboration between the University of Lapland, University of Jyväskylä, Umeå University, and Sámi University of Applied Sciences. The development process actively involves Sámi teachers, pupils, guardians, and policy makers to ensure the game reflects Sámi culture and educational realities. Teacher interviews provide data on the current state of Sámi elementary education and literacy instruction methodologies, guiding the game’s pedagogical design. The project also offers training and workshops for teachers to support implementation. The Lohkanlihkku game builds on the GraphoLearn (Finnish: Lukukupla), itself based on the extensively researched GraphoGame (Ekapeli). GraphoGame adapts exercises to pupils’ individual literacy levels, a feature incorporated into Lohkanlihkku. The game will be freely available for computers and mobile devices. This presentation outlines the project’s objectives, development process, and current progress, with a particular focus on the first work package: mapping the current state and practices of Sámi literacy education.
Though LLMs typically struggle with dialect classification, fine-tuning with even a limited amount of data can drastically improve their performance (Faisal and Anastasopoulos, 2025; Gaman et al. 2020). This presentation introduces such a fine-tuned model for the Dutch dialects: Meertje.
Meertje, a fine-tuned version of Dutch BERTje (de Vries et al. 2019), was developed to identify and extract dialect material from Dutch corpora. The model is trained on a subset of the Dialect Novel Corpus, a collection of 1112 dialectal literary works from the Netherlands and Flanders. The training material comprised 10 texts from Drenthe, from which 23000 sentences were manually annotated as containing either Standard Dutch or dialect. BERTje was fine-tuned on this subcorpus to create MeerBERTje, or Meertje for short.
The model isolated Drents from Standard Dutch with 95% accuracy. Pertinently for dialect extraction, the model found 94% of all dialectal sentences in the test set. Evidence of its generalisability emerged during further testing: texts from Groningen, Twente, and Zeeuws-Vlaanderen produced recall scores between 86% and 99%. Meertje does underperform on noisy data (like OCR-mistakes), with uncommon Dutch (like the archaic goochemerd), or with sentences which are dialectal in their syntactic structure but Standard Dutch in their choice of lexical items. However, given the predictable nature of these limitations, Meertje could become a valuable tool for dialect extraction for a variety of Dutch texts.
To now apply this model towards linguistic analysis, we will perform the Leave-One-Out (LOO) method (Xie et al. 2024; Roemling et al. 2024). LOO processes each token in a sentence as a data point, rather than evaluating the model at sentence level. We can thereby identify which lexical items are most impactful in distinguishing dialect material from Standard Dutch. This, in turn, can illustrate which linguistic features are most characteristic of the Dutch dialects.
Meertje, a fine-tuned version of Dutch BERTje (de Vries et al. 2019), was developed to identify and extract dialect material from Dutch corpora. The model is trained on a subset of the Dialect Novel Corpus, a collection of 1112 dialectal literary works from the Netherlands and Flanders. The training material comprised 10 texts from Drenthe, from which 23000 sentences were manually annotated as containing either Standard Dutch or dialect. BERTje was fine-tuned on this subcorpus to create MeerBERTje, or Meertje for short.
The model isolated Drents from Standard Dutch with 95% accuracy. Pertinently for dialect extraction, the model found 94% of all dialectal sentences in the test set. Evidence of its generalisability emerged during further testing: texts from Groningen, Twente, and Zeeuws-Vlaanderen produced recall scores between 86% and 99%. Meertje does underperform on noisy data (like OCR-mistakes), with uncommon Dutch (like the archaic goochemerd), or with sentences which are dialectal in their syntactic structure but Standard Dutch in their choice of lexical items. However, given the predictable nature of these limitations, Meertje could become a valuable tool for dialect extraction for a variety of Dutch texts.
To now apply this model towards linguistic analysis, we will perform the Leave-One-Out (LOO) method (Xie et al. 2024; Roemling et al. 2024). LOO processes each token in a sentence as a data point, rather than evaluating the model at sentence level. We can thereby identify which lexical items are most impactful in distinguishing dialect material from Standard Dutch. This, in turn, can illustrate which linguistic features are most characteristic of the Dutch dialects.
In this presentation we will share our experiences with the development of Frisian digital language tools over the past decade. It consists of the following parts:
1. A brief sketch of the history of the digital Frisian language tools developed at Fryske Akademy, with a focus on the tools for language users and on financing.
2. An overview of the digital language tools that are available on frysker.nl, with special attention to:
a. Its architecture
b. the user-based presentation
c. its focus on writing
d. dealing with norm and variation (proscription).
3. We will illustrate our good and bad experiences in the development of Frisian language tools. It seems a success story in comparison with other low-resource languages with a similar status, thanks to the financial support from the provincial government, collaboration with other academic institutions and private companies, competent and motivated staff members, and enthousiatic end-users. However, administratitive procedures, political circumstances, the lack of a unified vision and roadmap, slow decission making, internal circumstances, competion between institutions and companies, personal agendas, etc. hamper and slow down the process of renewing the digital infrastructure for Frisian.
4. We will conclude with a recommendation to develop a Roadmap for Limburgish Language Technology, for which we have some – unsolicited – advice. This is based on our experiences with the Frisian case and with other minority and regional languages in Europe.
Other authors/people working on the project(s): Elvis Túquerres, Melissa Túquerres and Ernesto Farinango (Language Activists, Chirihuasi), Tom Brand (Computational Linguist, TNO), Sebastian Mooser, Charlotte Wennekes, Ainsley Graham and Linus Kemp (Students, Leiden University), Lisa van den Hoven (Linguist, Instituut voor de Nederlandse Taal), Ximena Buller Machado (Linguist), Dr. Martine Bruil (University Lecturer, Leiden University).
The Kichwa speaking community of Chirihuasi (Imbabura, Ecuador) has longly been a beacon of linguistic activism and fruitful collaboration between academics and language users. The community has been a driving force in the development of projects aimed at the preservation of the Kichwa ancestral language. In the last few years, this effort has been characterized by the creation of digital resources for the language. Buller Machado et al. (forthcoming) showcases one of these projects, an augmented reality interactive children book in Kichwa and English, and also highlights the success of bottom-up digital revitalization projects.
In the same spirit, our presentation focuses on showcasing the Chirihuasi experience and its community-lead design that gave birth to a broad ecosystem of revitalization efforts and collaborations with scholars from many academic institutions, among which Leiden, Ottawa University and the Pontificia Universidad Católica del Ecuador. We will present previous successful experiences like the aforementioned children book "Kichwawa” (Buller Machado et al.); and current projects, like the ongoing creation of a Kichwa language museum in the community, a project particularly promoted by the language activist Elvis Túquerres, the development of digital tools (an orthographic converter and the ongoing creation of an Imbabura Kichwa translator) and an online hub to group, host and disseminate the projects originating in Chirihuasi. A partial version of this website, called Kichwa Wasi, is already available on google sites.
In presenting this, we advocate for a collaborative grassroots approach to revitalization, we promote the role of community-lead projects and we propose further developments for the Chirihuasi ecosystem.
The reference for the forthcoming article cited in the abstract is: Buller Machado, Ximena, Túquerres, Elvis, Farinango, Ernesto and Bruil, Martine (to appear). La inclusión de elementos digitales en materiales para la revitalización de lenguas indígenas: Desarrollo del Proyecto colaborativo “Kichwawa” en la comunidad de Chirihuasi, sierra del Ecuador. In M. Françozo, R. v. Gijn, & M. Bruil, The many voices of indigenous Latin America: Reconstructing, Describing, and Preserving Cultural and Linguistic Diversity. Leiden: Brill.
The Kichwa speaking community of Chirihuasi (Imbabura, Ecuador) has longly been a beacon of linguistic activism and fruitful collaboration between academics and language users. The community has been a driving force in the development of projects aimed at the preservation of the Kichwa ancestral language. In the last few years, this effort has been characterized by the creation of digital resources for the language. Buller Machado et al. (forthcoming) showcases one of these projects, an augmented reality interactive children book in Kichwa and English, and also highlights the success of bottom-up digital revitalization projects.
In the same spirit, our presentation focuses on showcasing the Chirihuasi experience and its community-lead design that gave birth to a broad ecosystem of revitalization efforts and collaborations with scholars from many academic institutions, among which Leiden, Ottawa University and the Pontificia Universidad Católica del Ecuador. We will present previous successful experiences like the aforementioned children book "Kichwawa” (Buller Machado et al.); and current projects, like the ongoing creation of a Kichwa language museum in the community, a project particularly promoted by the language activist Elvis Túquerres, the development of digital tools (an orthographic converter and the ongoing creation of an Imbabura Kichwa translator) and an online hub to group, host and disseminate the projects originating in Chirihuasi. A partial version of this website, called Kichwa Wasi, is already available on google sites.
In presenting this, we advocate for a collaborative grassroots approach to revitalization, we promote the role of community-lead projects and we propose further developments for the Chirihuasi ecosystem.
The reference for the forthcoming article cited in the abstract is: Buller Machado, Ximena, Túquerres, Elvis, Farinango, Ernesto and Bruil, Martine (to appear). La inclusión de elementos digitales en materiales para la revitalización de lenguas indígenas: Desarrollo del Proyecto colaborativo “Kichwawa” en la comunidad de Chirihuasi, sierra del Ecuador. In M. Françozo, R. v. Gijn, & M. Bruil, The many voices of indigenous Latin America: Reconstructing, Describing, and Preserving Cultural and Linguistic Diversity. Leiden: Brill.
Computational approaches to dialectology offer possibility to work efficiently with large amounts of dialect data and identify main dialect groups in an objective way, without concentrating on a handful of manually selected features. However, while being successful in detecting the most important dialect groups, the aggregate approach gives us little information about linguistic features characteristic for each dialect group. This kind of information is, however, very important for theoretical linguistics, but also for applied tasks of dialect and speaker identification.
To overcome this problem, we have employed normalised Pointwise Mutual Information (PMI, Church and Hanks 1990) for dialect feature extraction, following Sung and Prokic (2024). nPMI is an association measure which calculates the tendency of a pair of categories to co-occur based on the probability of each category as well as the co-occurrence probability. In addition, we have calculated the Exclusivity (how often do we find a feature value only in a given cluster) and Representativeness (the proportion of the feature value found within the dialect cluster). These indications can tell us not only what the characteristic features are for different dialect groups, but also whether the dialect groups are homogenous or not.
For our case study, we have made use of a subset of the Goeman-Taeldeman-Van Reenen Project (GTRP) dataset of Dutch dialects (from Gabmap, Leinonen et al. 2016). This subset has been multi-string aligned (Prokic et al. 2009) for the feature extraction. Using a 6-cluster solution from cluster analysis (Ward’s method, Ward 1963), we have identified the top features associated to each dialect group using nPMI. While most dialect areas have their own characteristic features, the Exclusivity and Representativeness scores have shown that the Limburgish area has very unique, features, but at the same time, the entire group is very heterogenous. The application of nPMI is complementary to the cluster analysis, since it enhances our understanding of the features present in different areas, as well as their within-cluster homogeneity.
To overcome this problem, we have employed normalised Pointwise Mutual Information (PMI, Church and Hanks 1990) for dialect feature extraction, following Sung and Prokic (2024). nPMI is an association measure which calculates the tendency of a pair of categories to co-occur based on the probability of each category as well as the co-occurrence probability. In addition, we have calculated the Exclusivity (how often do we find a feature value only in a given cluster) and Representativeness (the proportion of the feature value found within the dialect cluster). These indications can tell us not only what the characteristic features are for different dialect groups, but also whether the dialect groups are homogenous or not.
For our case study, we have made use of a subset of the Goeman-Taeldeman-Van Reenen Project (GTRP) dataset of Dutch dialects (from Gabmap, Leinonen et al. 2016). This subset has been multi-string aligned (Prokic et al. 2009) for the feature extraction. Using a 6-cluster solution from cluster analysis (Ward’s method, Ward 1963), we have identified the top features associated to each dialect group using nPMI. While most dialect areas have their own characteristic features, the Exclusivity and Representativeness scores have shown that the Limburgish area has very unique, features, but at the same time, the entire group is very heterogenous. The application of nPMI is complementary to the cluster analysis, since it enhances our understanding of the features present in different areas, as well as their within-cluster homogeneity.
The development and adaptation of spelling systems in standard languages—Dutch included—has historically been complex and contested. These systems vary in their adherence to phonemic, morphological, and etymological principles. An additional dimension concerns the accommodation of regional linguistic variation. In this context, the organization Veldeke, which promotes dialect and cultural diversity in the Dutch province of Limburg, has played a decisive role in developing a unified spelling system that respects local pronunciation differences.
The Veldeke spelling 2003 is founded on five guiding principles: (1) phonemic representation, (2) word autonomy, (3) etymological origin, (4) local traditions, and (5) alignment with the principles of spelling Dutch. One notable consequence of this framework is the influence of Dutch orthography on the written representation of Limburg dialects. Furthermore, the prioritization of phonemic representation has led to considerable orthographic variation across Limburg dialects, reflecting phonetic diversity. While this variation supports the preservation of local differences, it can hinder readability and word recognition.
Efficient reading relies on consistent and easily processed word forms. This raises the question: can dialectal differentiation be reconciled with the cognitive demands of fluent reading? Should Limburg follow the example of Frisian, where orthographic standardization has had a significant impact despite internal dialectal variation? Alternatively, should the Veldeke principles be revised to reduce phonetic variation, or can recent advances in artificial intelligence—particularly in translation and text transformation—offer new solutions?Digital technologies may provide innovative approaches to managing dialectal diversity. In light of this, it is essential to consider how spelling in Limburg should evolve within the broader context of its status as a regional language, characterized by extensive morphological and lexical variation.
The Veldeke spelling 2003 is founded on five guiding principles: (1) phonemic representation, (2) word autonomy, (3) etymological origin, (4) local traditions, and (5) alignment with the principles of spelling Dutch. One notable consequence of this framework is the influence of Dutch orthography on the written representation of Limburg dialects. Furthermore, the prioritization of phonemic representation has led to considerable orthographic variation across Limburg dialects, reflecting phonetic diversity. While this variation supports the preservation of local differences, it can hinder readability and word recognition.
Efficient reading relies on consistent and easily processed word forms. This raises the question: can dialectal differentiation be reconciled with the cognitive demands of fluent reading? Should Limburg follow the example of Frisian, where orthographic standardization has had a significant impact despite internal dialectal variation? Alternatively, should the Veldeke principles be revised to reduce phonetic variation, or can recent advances in artificial intelligence—particularly in translation and text transformation—offer new solutions?Digital technologies may provide innovative approaches to managing dialectal diversity. In light of this, it is essential to consider how spelling in Limburg should evolve within the broader context of its status as a regional language, characterized by extensive morphological and lexical variation.
The Storyhouse research project addresses the urgent need for digital, linguistically inclusive learning resources in Sámi education. Its core aim is to translate and adapt the game-based Storyhouse learning environment into Sámi languages, supporting children and young people in developing multiliteracy skills while strengthening the digital accessibility and vitality of Sámi languages across Finland, Norway, and Sweden.
The need for new pedagogical approaches is underscored by recent concerns over declining literacy skills both nationally and internationally (Karvi, 2023; OECD, 2023). In Finland, national evaluations highlight decreased writing proficiency, especially at the end of basic education. International assessments, such as PISA, show similar trends. Simultaneously, students often describe school writing as mechanical, irrelevant, and disconnected from their everyday lives, which diminishes motivation and engagement (Freda et al., 2022).
Research suggests that game-based and playful learning environments can provide solutions by enhancing writing motivation, creativity, and text quality (Gee, 2008; Whitton, 2018). They also foster collaborative, multimodal, and participatory literacy practices aligned with multiliteracy pedagogy (New London Group, 1996; Cope & Kalantzis, 2009). In the Sámi context, where culturally responsive digital resources remain scarce, such innovations are especially critical (Kallionpää, 2017; Keskitalo & Olsen, 2024).
The project integrates three research perspectives: (1) writing motivation theory, emphasizing agency, self-efficacy, and meaningful engagement (Bruning & Horn, 2000; Graham et al., 2020); (2) multiliteracy pedagogy, highlighting multimodality, cultural sensitivity, and participatory practices (Cope & Kalantzis, 2009; Lankshear & Knobel, 2011); and (3) playful learning, which incorporates curiosity, exploration, and creativity into structured learning processes (Zosh et al., 2018; Kangas, 2010). Together, these strands create a framework for rethinking writing pedagogy in Sámi-language education.
Methodologically, the project involves translating and adapting The Storyhouse into multiple Sámi languages and cultures, followed by school-based pilots across Sámi regions. These pilots are paired with research on how playful and game-based pedagogies support Sámi students’ literacy development, motivation, and writer identity. The project also investigates how culturally grounded digital environments can enhance the vitality of endangered languages and strengthen students’ sense of belonging.
The expected contributions are threefold. First, the project develops innovative Sámi-language digital learning resources to address a persistent gap in education. Second, it advances literacy research by combining writing motivation theory with playful multiliteracy pedagogy in a minority language context. Third, it provides practical, scalable models that can support language revitalization through playful and digital pedagogies.
In sum, the project responds to both local and global challenges by creating equitable opportunities for Sámi children and youth. It demonstrates how playful multiliteracy learning can bridge the gap between school writing and everyday media practices while contributing to the vitality and future of Sámi languages and cultures.
The need for new pedagogical approaches is underscored by recent concerns over declining literacy skills both nationally and internationally (Karvi, 2023; OECD, 2023). In Finland, national evaluations highlight decreased writing proficiency, especially at the end of basic education. International assessments, such as PISA, show similar trends. Simultaneously, students often describe school writing as mechanical, irrelevant, and disconnected from their everyday lives, which diminishes motivation and engagement (Freda et al., 2022).
Research suggests that game-based and playful learning environments can provide solutions by enhancing writing motivation, creativity, and text quality (Gee, 2008; Whitton, 2018). They also foster collaborative, multimodal, and participatory literacy practices aligned with multiliteracy pedagogy (New London Group, 1996; Cope & Kalantzis, 2009). In the Sámi context, where culturally responsive digital resources remain scarce, such innovations are especially critical (Kallionpää, 2017; Keskitalo & Olsen, 2024).
The project integrates three research perspectives: (1) writing motivation theory, emphasizing agency, self-efficacy, and meaningful engagement (Bruning & Horn, 2000; Graham et al., 2020); (2) multiliteracy pedagogy, highlighting multimodality, cultural sensitivity, and participatory practices (Cope & Kalantzis, 2009; Lankshear & Knobel, 2011); and (3) playful learning, which incorporates curiosity, exploration, and creativity into structured learning processes (Zosh et al., 2018; Kangas, 2010). Together, these strands create a framework for rethinking writing pedagogy in Sámi-language education.
Methodologically, the project involves translating and adapting The Storyhouse into multiple Sámi languages and cultures, followed by school-based pilots across Sámi regions. These pilots are paired with research on how playful and game-based pedagogies support Sámi students’ literacy development, motivation, and writer identity. The project also investigates how culturally grounded digital environments can enhance the vitality of endangered languages and strengthen students’ sense of belonging.
The expected contributions are threefold. First, the project develops innovative Sámi-language digital learning resources to address a persistent gap in education. Second, it advances literacy research by combining writing motivation theory with playful multiliteracy pedagogy in a minority language context. Third, it provides practical, scalable models that can support language revitalization through playful and digital pedagogies.
In sum, the project responds to both local and global challenges by creating equitable opportunities for Sámi children and youth. It demonstrates how playful multiliteracy learning can bridge the gap between school writing and everyday media practices while contributing to the vitality and future of Sámi languages and cultures.
Transcribing minority languages presents both technical and linguistic challenges. As Bucholtz (2000) points out, transcriptions can never be entirely neutral. Even if numerous programmes and the rise of digital humanities have facilitated and automated the transcription of spoken material, this only holds true for codified languages. Therefore, minority languages, in particular, pose a significant challenge: not only are they often neither codified nor standardised but additionally they only have a few speakers and tend to be geographically and socially isolated. This presentation attempts to show how minority languages—despite the lack of standardisation—can be transcribed. Pennsylvania German, a North American language shaped by immigrants from southwestern Germany in the 18th century (Louden 2017), serves as the reference variety. Drawing on data collected within the Ontario Dialects Project (2018–2019), this presentation highlights the limitations of using the International Phonetic Alphabet, especially given that involving the speech community offers potential for insights (Neuhausen 2023). Furthermore, this talk shows how archaisms and the influence of German dialects complicate transcription. Finally, I argue for a hybrid transcription using Latin letters to keep the transcription accessible to the speech community and reflective of the variation of Pennsylvania German to allow future research.
Page updated
Google Sites
Report abuse