greek dialect nlp talk

Author: verenablaschke

_events/2025-08-14-athena-rc.md

@@ -0,0 +1,17 @@
+---
+title: "AI Meets Linguistic Diversity: Documenting and Modeling Under-Resourced Varieties  (of Greek origins or in contact with Greek)"
+abstract: "<p>Οver the last 7 years, our team has been enthusiastically engaged in an effort to use AI tools and knowledge transfer (where applicable) for the detailed documentation of endangered non-Greek language varieties spoken in the present Greek territory and Greek varieties independently of geographical location. For most of the Greek varieties we relied on knowledge transfer from STT and morphosyntactic models for Standard Modern Greek. We have also experimented with the dialect-to-standard normalization of Greek dialects, in order to expand the coverage of Natural Language Understanding systems. Since all the language varieties we treated are primarily -or only- oral, we had to make decisions about their orthography both for STT and morphosyntactic model purposes. In addition, we had to handle challenges of working at the intersection of detailed language documentation, the “universals-oriented” design of annotation schemes like UD, and the limitations of available open-source tools for model development. We will attempt to discuss our experience with Plank (2015) as a background.</p> 
+<p>We have collected oral material and developed STT and morphosyntactic models drawing on the transcripts and the textual legacy, if any, of Pomak (a Slavic language variety spoken in Thrace/Greece, Bulgaria and Turkey) and Cretan, Messinian and Lesbian (all Modern Greek dialects). We are currently working on a morphosyntactic model of Griko (a severely endangered Greek dialect of Southern Italy) and on STT and morphosyntactic models of Cypriot Greek (a living Greek dialect). We are conducting more fieldwork on Cretan,  Aperathitika (a living Greek dialect of Central Aegean) and Arvanitika (a severely endangered  Albanian dialect of Central Greece).  Our team has developed GUD, a treebank for Standard Modern Greek, which we use for knowledge transfer. We have also conducted a study on dialectal Greek proverbs, measuring how normalizing them to Standard Modern Greek can help us derive previously obscured cultural insights. The resources and the models for Modern Greek dialects are openly available to the research community. The Pomak treebank is available on the UD treebank repository.</p>"
+speaker: Stella Markantonatou, Stavros Bompolas, Antonis Dimakis <br/>(Archimedes, Athena Research Center, Greece)
+bio: '<p><a href="https://archimedesai.gr/en/researchers/stella-markantonatou">Stella Markantonatou</a> is a Research Director at the Institute for Language and Speech Processing (ILSP)/ATHENA RC. She has served as a Lecturer (University of Essex 1992), a Visiting Assistant Professor (University of Cyprus 1998) and an Associate Professor (University of Crete 2002). She is a member of the Management Committee of the MSc programme “Language Technology” (co-organised by the NKUA and ATHENA RC). She has been a member of the Scientific Advisory Board of ILSP (2011-2016), the Board of the National Documentation Center of Greece (2017-2019) and the SIGLEX Standing Committee for Multiword Expressions (2017-2019). She has coordinated or had the scientific responsibility of 20 national and European R&D projects. She has published more than 90 papers in national and international refereed journals and conferences, and has edited 5 volumes for international publishers and the Academy of Athens.</p>
+<p><a href="https://archimedesai.gr/en/researchers/stavros-bompolas">Stavros Bompolas</a> obtained his PhD from the Philology Department (Linguistics Section) at the University of Patras in 2023, specializing in Computational Dialectology. He received the “Andreas Mentzelopoulos Scholarships for Postgraduate Studies at the University of Patras” (2020-2022) for his doctoral research. He holds a BA Honours (Rank: 5th out of 298) and a Master’s degree (Rank: 1st out of 28) from the same Department. Since 2014, he has been a research member at the Laboratory of Modern Greek Dialects (University of Patras), contributing to four research projects. His research has been disseminated through presentations at conferences and publications in proceedings, scientific volumes, and journals. Notably, a part of his postgraduate dissertation, titled “Lexical Self-Organization in the Acquisition and Processing of Modern Greek Conjugation: An Artificial Neural Network Approach”, was awarded the “Distinguished Young Paper Award” at the “3rd Italian Conference on Computational Linguistics” (CLiC-it 2016).<br/> His research interests are centered around computational linguistics and dialectology, particularly emphasizing the computational analysis and quantification of linguistic (dis)similarities among Greek dialects that have developed in contact with typologically diverse languages, including Italo-Romance and Turkish.</p>
+<p><a href="https://archimedesai.gr/en/researchers/antonis-dimakis">Antonis Dimakis</a> received his bachelor’s degree from the Department of Informatics and Telecommunications of the National and Kapodistrian University of Athens, with specializations in the Foundations of Computer Science and Data and Knowledge Management. He has experience in designing and developing machine learning models, especially large language models (LLMs). He has won awards in various student contests, including the EUSO Experimental Natural Science competition, the “Aristotelis” Physics competition, and the Greek Mathematical Society’s Panhellenic Mathematics Competition, in 2017 and 2018.<br/> Antonis’ research interests include machine learning, linguistics and cognitive science. Specifically, his research will focus on areas such as neural multilingual language models and NLP for low-resource languages.</p>'
+website: https://archimedesai.gr/en/researchers/stella-markantonatou
+time: August 14, 2025; 14:45–15:45
+location: Akademiestr. 7, room 218A (meeting room)
+roomfinder: https://mainlp.github.io/contact/
+img: 
+imgalt: 
+imgside: left
+anchor: 2025-08-14-athena-rc
+---