All over the world language data and technology are produced daily by universities, schools, GLAM institutions, news agencies, non-academic organizations, administrative bodies and more besides. While these resources typically serve individual needs, if made accessible in standardized formats they could stand to benefit much larger initiatives. The Institute for Applied Linguistics seeks to establish a language services research hub and observatory, the linguaXlab, to maximize access and (re)use of the linguistic data and technology produced by, and relating to, South Tyrol, and facilitate their exchange across the region between different user groups. Drawing from the expertise of the entire Institute, among other linguistic services the linguaXlab will offer support with data archiving, search functionality, terminological databases, machine learning, information extraction and data FAIRification to stakeholders from different backgrounds (e.g., the cultural heritage sector, public administration or the medical field) to collectively advance language research and development for South Tyrol and beyond.
Discover
See related experts, publications, and research topics.
Eurac Research is a private research center based in Bolzano (South Tyrol) with researchers from a wide variety of scientific fields who come from all over the globe. Together, through scientific knowledge and research, they share the goal of shaping the future.
Our research addresses the greatest challenges facing us in the future: people need health, energy, well-functioning political and social systems and an intact environment. These are complex questions, and we are seeking the answers in the interaction between many different disciplines. [About us](/en/about-us-eurac-research)
Eurac Research Headquarter
Viale Druso 1
39100 Bolzano
Italy
Eurac Research NOI Techpark
Via A.-Volta
39100 Bolzano
Italy
Except where otherwise noted, content on this site is licensed under a Creative Commons Attribution 4.0 International license.
My cookie preferencesDucceschi L, Franzini G
AUGUSTA (Artificial UnderstandinG of Utterances Spoken in Tyrolean Accents) is a family of Speech-to-Text language models fine-tuned for South Tyrolean dialect translation into Standard German.
DOI: 10.5281/zenodo.15553913
Spina S, Brasolin P, Franzini GH
This exploratory study investigates lexical change and innovation in contemporary Italian micro-blogging using a corpus of 5.32 million timestamped and geotagged tweets sampled from the 2022 Italian Twitter timeline. We develop a new method to identify 720 unattested forms (347 forms and 373 hashtags) as candidate neologisms. Our results show that orthographic variation, univerbation, suffixation, loanwords and portmanteaus are the most common categories of lexical creation in the data analysed, which appears to be driven by creativity, amusement and attention-seeking behaviour rather than a need for new words to define new objects, events or situations.
DOI: 10.32714/ricl.13.01.07
Ducceschi L, Franzini GH
This study presents the first fine-tuned Whisper model for the automatic translation of South Tyrolean dialectal speech into Standard German text. To address an unmet need for subtitling and translation, we introduce a small corpus of manually annotated and synthetic speech data compiled for this task. Through fine-tuning and hyperparameter optimisation, our model achieves a BLEU score of 86.18 significantly outperforming baseline error rates. Our findings highlight Whisper's effectiveness in handling dialectal speech, contributing to low-resource language research. The model is already being used in a heritage collaboration for large-scale translation of audiovisual archival material and is also being considered for application in news broadcasting and tourism promotion. Future directions include expanding the training data and extending hyperparameter optimisation to improve the model's performance and generalisation across South Tyrolean dialectal variations.
DOI: 10.21437/Interspeech.2025-1976
Franzini GH, Brasolin P, Spina S
This paper outlines ongoing research into lexical innovation in contemporary Italian in the context of social media. To date, the study has used a dataset of 5.32M timestamped and geotagged tweets extracted from the 2022 Italian timeline, yielding 720 emerging word forms. Here, we describe the reproducible pipeline developed to extract candidate neologisms from our dataset and introduce a custom tool to visualise the emergence and spread of candidate neologisms in the time-period under investigation.
DOI: 10.6092/unibo/amsacta/7927
Conference: First Workshop on Dialects in NLP — A Resource Perspective | Palma de Mallorca | 16.5.2026 - 16.5.2026
More information: poster
Conference: CLARIN Café - JOHD Special Collection on Language Dataset Reuse | online | 24.6.2026 - 24.6.2026
