November 2024 - Corpora

ACL 2025 Call for Papers
by ACL Announcements 04 Nov '24

04 Nov '24

ACL 2025 Call for Papers MAIN CONFERENCE ACL 2025 Website: https://2025.aclweb.org/ [1] Submission Deadline: February 15, 2025 Conference Dates: July 27 to August 1, 2025 Location: Vienna, Austria Special Theme: "Generalization of NLP Models" Contact: * Roberto Navigli [2] (General Chair) * Wanxiang Che [3], Joyce Nabende [4], Mohammad Taher Pilehvar [5], Ekaterina Shutova [6] (Program Chairs): For questions related to paper submission, email: editors(a)aclrollingreview.org For all other questions, email: acl2025pcs(a)gmail.com OVERVIEW ACL 2025 invites the submission of long and short papers featuring substantial, original, and unpublished research in all aspects of Computational Linguistics and Natural Language Processing. ACL 2025 has a goal of a diverse technical program--in addition to traditional research results, papers may contribute negative findings, survey an area, announce the creation of a new resource, argue a position, report novel linguistic insights derived using existing computational techniques, and reproduce, or fail to reproduce, previous results. As in recent years, some of the presentations at the conference will be of papers accepted by the Transactions of the ACL (TACL) and by the Computational Linguistics (CL) journals. Papers submitted to ACL 2025, but not selected for the main conference, will also automatically be considered for publication in the Findings of the Association of Computational Linguistics. PAPER SUBMISSION INFORMATION Papers may be submitted to the ARR 2025 February cycle. Papers that have received reviews and a meta-review from ARR (whether from the ARR 2025 February cycle or an earlier ARR cycle) may be committed to ACL 2025 via the conference commitment site (TBA). SUBMISSION TOPICS ACL 2025 aims to have a broad technical program. Relevant topics for the conference include, but are not limited to, the following areas (in alphabetical order): * Computational Social Science and Cultural Analytics * Dialogue and Interactive Systems * Discourse and Pragmatics * Efficient/Low-Resource Methods for NLP * Ethics, Bias, and Fairness * Generation * Information Extraction * Information Retrieval and Text Mining * Interpretability and Analysis of Models for NLP * Language Modeling * Linguistic theories, Cognitive Modeling and Psycholinguistics * Machine Learning for NLP * Machine Translation * Multilinguality and Language Diversity * Multimodality and Language Grounding to Vision, Robotics and Beyond * NLP Applications * Phonology, Morphology and Word Segmentation * Question Answering * Resources and Evaluation * Semantics: Lexical and Sentence-Level * Sentiment Analysis, Stylistic Analysis, and Argument Mining * Speech recognition, text-to-speech and spoken language understanding * Summarization * Syntax: Tagging, Chunking and Parsing * Special Theme: Generalization of NLP Models ACL 2025 Theme Track: Generalization of NLP Models Following the success of the ACL 2020-2024 Theme tracks, we are happy to announce that ACL 2025 will have a new theme with the goal of reflecting and stimulating discussion about the current state of development of the field of NLP. Generalization is crucial for ensuring that models behave robustly, reliably, and fairly when making predictions on data different from their training data. Achieving good generalization is critically important for models used in real-world applications, as they should emulate human-like behavior. Humans are known for their ability to generalize well, and models should aspire to this standard. The theme track invites empirical and theoretical research and position and survey papers reflecting on the Generalization of NLP Models. The possible topics of discussion include (but are not limited to) the following: * How can we enhance the generalization of NLP models across various dimensions--compositional, structural, cross-task, cross-lingual, cross-domain, and robustness? * What factors affect the generalization of NLP models? * What are the most effective methods for evaluating the generalization capabilities of NLP models? * While Large Language Models (LLMs) significantly enhance the generalization of NLP models, what are the key limitations of LLMs in this regard? The theme track submissions can be either long or short. We anticipate having a special session for this theme at the conference and a Thematic Paper Award in addition to other categories of awards. TWO-STAGE REVIEW: SUBMISSION TO ARR, COMMITMENT TO ACL 2025 ACL 2025 will use ACL Rolling Review [7] (ARR) https://aclrollingreview.org/cfp as a reviewing system, but final decisions will be made by the conference. Both submissions of articles for review and commitment of reviewed articles to the conference will be performed via the Open Review [8] platform. Specifically, authors will follow a two-step process: * Authors submit articles to ARR, where submissions receive reviews and meta-reviews from ARR reviewers and action editors; * Authors commit their reviewed articles to a publication venue (e.g., ACL 2025), where Senior Area Chairs and Program Chairs make acceptance decisions from the ARR reviews and meta-reviews. ACL 2025 has chosen this approach in coordination with *CL 2024 conferences, which are adopting the same procedure and a coordinated submission plan to allow maximum flexibility during their submission periods for the authors. At each cycle, after a paper has been fully reviewed, authors have the option to commit their paper to a conference or revise and resubmit for another round of reviews. The reviewing process will continue to be double-blind. Reviewers will not see authors, nor will authors see reviewers, and reviews on ARR will not be made publicly visible. However, authors will be given the option through ARR to make their anonymized submitted articles publicly visible. MANDATORY REVIEWING WORKLOAD AS THE PACE OF RESEARCH IN THE FIELD CONTINUES TO INCREASE, WE NEED TO STRENGTHEN THE COMMITMENT TO REVIEWING FOR EACH PAPER SUBMISSION. DURING THE ARR SUBMISSION PROCESS, AUTHORS WILL BE REQUIRED TO SPECIFY WHICH CO-AUTHORS ARE COMMITTING TO COVER REVIEWING IN THIS REVIEWING CYCLE. PLEASE SEE THE NEW ARR POLICY REGARDING REVIEWING WORKLOAD HERE. AS THIS IS AN ARR-WIDE POLICY FOR ALL *CL CONFERENCES, QUESTIONS OR CLARIFICATIONS SHOULD BE ADDRESSED TO ARR DIRECTLY. IMPORTANT DATES Submission deadline (all papers are submitted to ARR) February 15, 2025 ARR reviews & meta-reviews available to authors of the February cycle April 15, 2025 Commitment deadline for ACL 2025 April 20, 2025 Notification of acceptance May 15, 2025 Withdrawal deadline May 30, 2025 Camera-ready papers due May 30, 2025 Tutorials July 27, 2025 Conference July 28 - 30, 2025 Workshops July 31 - August 1, 2025 Note: All deadlines are 11:59PM UTC-12:00 ("anywhere on Earth"). Paper Submission Details Both long and short paper submissions should follow all of the ARR submission requirements [9], including: * Long Papers [10] (8 pages) and Short Papers [11] (4 pages) * Instructions for Two-Way Anonymized Review [12] * Authorship [13] * Citation and Comparison [14] * Multiple Submission Policy [15], Resubmission Policy [16], and Withdrawal Policy [17] * Ethics Policy [18] including the responsible NLP research checklist [19] * Limitations [20] * Paper Submission and Templates [21] * Optional Supplementary Materials [22] Final versions of accepted papers will be given one additional page of content (up to 9 pages for long papers, up to 5 pages for short papers) to address reviewers' comments. Following the ACL and ARR policies (https://www.aclweb.org/portal/content/report-acl-committee-anonymity-policy) [23], there is no anonymity period requirement. At the time of submission to ARR, authors will be asked to select a preferred venue (e.g., ACL 2025). This is used only to calculate acceptance rates. Authors who selected ACL 2025 as a preferred venue when submitting to ARR may choose not to commit to ACL 2025 after receiving their reviews, and authors who selected a preferred venue other than ACL 2025 when submitting to ARR are still welcome to commit to ACL 2025. Presentation at the Conference All accepted papers must be presented at the conference to appear in the proceedings. The conference will include both in-person and virtual presentation options. Papers without at least one presenting author registered by the early registration deadline may be subject to desk rejection. Long and short papers will be presented orally or as posters as determined by the program committee. While short papers will be distinguished from long papers in the proceedings, there will be no distinction in the proceedings between papers presented orally and papers presented as posters. Links: ------ [1] https://2025.aclweb.org/ [2] https://www.diag.uniroma1.it/navigli/ [3] http://ir.hit.edu.cn/~car/ [4] https://sites.google.com/view/jnabende/home?authuser=0 [5] https://pilehvar.github.io/ [6] https://www.shutova.org/ [7] https://aclrollingreview.org/cfp [8] https://openreview.net/ [9] https://aclrollingreview.org/cfp#paper-submission-information [10] https://aclrollingreview.org/cfp#long-papers [11] https://aclrollingreview.org/cfp#short-papers [12] https://aclrollingreview.org/cfp#instructions-for-two-way-anonymized-review [13] https://aclrollingreview.org/cfp#authorship [14] https://aclrollingreview.org/cfp#citation-and-comparison [15] https://aclrollingreview.org/cfp#multiple-submission-policy [16] https://aclrollingreview.org/cfp#resubmission-policy [17] https://aclrollingreview.org/cfp#withdrawal-policy [18] https://aclrollingreview.org/cfp#ethics-policy [19] https://aclrollingreview.org/responsibleNLPresearch [20] https://aclrollingreview.org/cfp#limitations [21] https://aclrollingreview.org/cfp#paper-submission-and-templates [22] https://aclrollingreview.org/cfp#optional-supplementary-materials-appendice… [23] https://www.aclweb.org/portal/content/report-acl-committee-anonymity-policy

1 0

CfP: Second Forlì International Workshop on Corpus-based Interpreting Studies and Applications: At the Interface of Data and Technology
by Nannan Liu 04 Nov '24

04 Nov '24

Second Forlì International Workshop on Corpus-based Interpreting Studies and Applications: At the Interface of Data and Technology 8–10 May 2025, Forlì, Italy, and online Call for Papers The Department of Interpreting and Translation at the University of Bologna is organising the Second Forlì International Workshop on Corpus-based Interpreting Studies and Applications on 8–10 May 2025 in Forlì, Italy, and online. Background The Forlì International Workshop series was launched in 2015 to stimulate the creation of interpreting corpora and corpus-based research projects. The first workshop gathered more than 100 scholars from around the world, resulting in a volume of state-of-the-art research (Russo et al. 2018<https://link.springer.com/book/10.1007/978-981-10-6199-8>) and a special issue (Bendazzoli et al. 2018<https://www.intralinea.org/specials/cbis>). The tenth anniversary of the first workshop marks an opportune occasion to take stock of recent developments and chart new directions in light of corpora’s fundamental role in technological advancements. Theme Interpreting corpora serve as the descriptive foundation of research and the ground truth against which machine interpreting technologies are evaluated. Corpus-based interpreting studies, as envisaged by Shlesinger (1998<https://doi.org/10.7202/004136ar>), have developed into a highly productive line of inquiry with theoretical inputs from cognitive linguistics and sociology and methodological contributions from natural language processing, prosody research, and multimodality. Recently, large interpreting corpora have fuelled the deployment of machine interpreting technologies, together with deep learning algorithms that synthesise signing images and texts (e.g. Saunders et al. 2022<https://doi.org/10.1109/CVPR52688.2022.00508>). Amidst changing conceptual boundaries (Pöchhacker 2024<https://doi.org/10.1075/ts.23028.poc>), methodological developments, and technological landscape, a field-wide reflection on the role of corpora is necessary. In this context, we view the Second Forlì International Workshop as an opportunity to bring together researchers who create, analyse, and use corpora to study interpreting and develop tools and applications for corpus-based research, computer-assisted interpreting, machine interpreting, automated interpreting quality assessment, pedagogy, and other related domains. Submissions We particularly welcome abstracts addressing the following topics: * Theory testing and comparisons using corpora * Replication studies using corpora * Quantitative and qualitative approaches to corpus-based studies * Triangulating corpora with other data sources (e.g. user response, eye-tracking, EEG, and workplace ethnography) * Underrepresented corpora, e.g. those involving signed language interpreting and onsite versus remote interpreting * Corpora of constrained communication (Kotze & van Rooy 2024<https://doi.org/10.1075/coll.60.01kot>), e.g. those by untrained interpreters and learners * Multimodal and multiple interpreting corpora * Computer-assisted interpreting software and applications * Machine interpreting systems * Automated interpreting quality assessment based on corpora * Corpus design according to principles of scientific data management (e.g. FAIR principles; Wilkinson et al. 2016<https://doi.org/10.1038/sdata.2016.18>) The first page of the submission should contain each author’s name, affiliation (if any), and e-mail address. The abstract should be placed from the second page onwards, including the title, a text between 300 and 400 words (including examples, excluding references) that clearly states the research questions, methods, data, and (preliminary) results, and up to five keywords. The abstract should not contain any identifying information to allow for double-blind peer reviewing. There will be three types of presentations: * Full paper (20 minutes + 10 minutes Q & A) * Poster * Software demonstration Abstracts will be submitted to unic(a)dipintra.it<mailto:unic@dipintra.it>. Authors may indicate a full paper, a poster, or a software presentation upon submission. Posters will be displayed on Day 1 of the Workshop, and a time slot will be reserved in the programme for participants to discuss with poster presenters. Software demonstrations will be held on Day 2 in a dedicated time slot. A selection of papers will be published in an edited volume and a special issue in a scientific journal. Language of the conference English Important dates * Submission deadline: 1 February 2025 * Notification of acceptance and rejection: 1 March 2025 * Pre-workshop session on the Unified Interpreting Corpus (UNIC; https://unic.dipintra.it/) platform: 15:00–17:30, 8 May 2025 * Workshop: 9–10 May 2025 Pre-workshop session convenors * Nannan Liu * Mariachiara Russo Scientific committee members Claudia Angelelli (Herriot-Watt University) Alberto Barrón-Cedeño (University of Bologna) Claudio Bendazzoli (University of Verona) Silvia Bernardini (University of Bologna) Sabine Braun (University of Surrey) Agnieszka Chmiel (Adam Mickiewicz University) Elena Davitti (University of Surrey) Bart Defrancq (Ghent University) Adriano Ferraresi (University of Bologna) Min-hua Liu (Hong Kong Baptist University) Nannan Liu (University of Bologna) Bernd Meyer (Johannes Gutenberg University of Mainz) Maja Miličević Petrović (University of Bologna) Koen Plevoets (Ghent University) Bianca Prandi (University of Bologna) Mariachiara Russo (University of Bologna) Annalisa Sandrelli (University of the International Studies of Rome) Elisabet Tiselius (Stockholm University) Ira Torresi (University of Bologna) Kim Wallmach (Stellenbosch University) Organising committee members Alberto Barrón-Cedeño (University of Bologna) Michela Bertozzi (University of Bologna) Silvia Bernardini (University of Bologna) Francesca D’Angelo (University of Bologna) Bart Defrancq (Ghent University) Adriano Ferraresi (University of Bologna) Serena Ghiselli (University of Bologna) Nannan Liu (University of Bologna) Marco Lobascio (University of Bologna) Natacha Niemants (University of Bologna) Maja Miličević Petrović (University of Bologna) Bianca Prandi (University of Bologna) Mariachiara Russo (University of Bologna) Nicoletta Spinolo (University of Bologna) Han Wang (University of Bologna) Contact unic(a)dipintra.it<mailto:unic@dipintra.it> Workshop website Website coming soon. Dr Nannan Liu Marie Curie Fellow Project FAITH<https://cordis.europa.eu/project/id/101108651> Department of Interpreting and Translation University of Bologna

1 0

SwissText 2025: 1ST CALL FOR SHARED TASK, WORKSHOP AND TUTORIAL PROPOSALS
by Tuggener Don (tuge) 04 Nov '24

04 Nov '24

Dear colleagues, We cordially invite submissions of proposals for shared tasks, workshops, and tutorials to be held at the SwissText 2025 conference. SwissText will take place from June 17-18, 2025 at ZHAW in Winterthur. ABOUT SwissText SwissText is an annual conference that brings together text analytics experts from industry and academia. It is organized by the Swiss Association for Natural Language Processing (SwissNLP) in collaboration with the Zurich University of Applied Sciences (ZHAW). SPECIAL EDITION This edition of SwissText will be special, since we celebrate its 10th anniversary! This is a great opportunity to look back. Hence, in addition to novel ideas for shared tasks, we also invite previous organizers of shared tasks to re-submit their ideas: What has changed since the last run of the shared task? Is the task still relevant? Do LLMs solve everything now? This offers us to see what progress has been made over the years. To give you some ideas, here is a list of previous shared tasks: * NLP for Sustainable Development Goals Monitoring * Swissdox Hackathon * Detecting greenwashing signals through a comparison of ESG reports and public media * Swiss German Speech to Standard German Text Shared Task * Low-Resource Speech-to-Text * The Sentence End and Punctuation Prediction in NLG text (SEPP-NLG) * German Text Summarization Challenge * .. and many more (see the SwissText website archive)! FORMAT FOR PROPOSALS Proposals for shared tasks should contain: * a title and a brief description of the topic of the task * a description of the data sets that will be used in the shared task and their readiness * a sketch of how the submitted systems will be evaluated * a tentative timeline Proposals for workshops should contain: * a title and a brief description of the topic * a description of the intended audience * workshop format (paper presentations, poster session, etc.) * a tentative timeline Proposals for tutorials should contain: * a title and a brief description of the topic and the goal of the tutorial * an introduction of the workshop speakers and their background and expertise * a description of the intended audience and the required level of expertise (beginners, experts, etc.) * a tentative outline of the tutorial schedule Note that the organization and running of the shared tasks, workshops, and tutorial is in the hands of the respective organizers. The SwissText organizers will provide infrastructure (rooms, paper submission platform) and assist where they can, of course. Interested? We are looking forward to your proposals. Submit your proposals by email to info(a)swisstext.org<mailto:info@swisstext.org> no later than November 30, 2024. Notifications will be sent out by December 15, 2024. Kind regards, Don ________________________________ ZHAW School of Engineering / CAI Dr. Don Tuggener Technikumstrasse 71 Postfach 8401 Winterthur Tel: +41 58 934 78 55 Web: https://www.zhaw.ch/de/ueber-uns/person/tuge/

1 0

Call for Applications: Georgetown Linguistics PhD and MS Programs
by Ethan Wilcox 03 Nov '24

03 Nov '24

Call for Applications: Georgetown University MS and Ph.D. programs in Computational Linguistics Georgetown University (Washington, DC) invites applications to our MS and Ph.D. programs for students wishing to study Computational Linguistics starting in Fall 2025. Georgetown is strategically located in the nation's capital with proximity to government institutions, a thriving tech industry, and other major universities. Our programs offer advanced training and cutting-edge research opportunities in topics such as natural language processing, syntactic/semantic analysis, corpora, contemporary machine learning/deep learning methods for language, and applications of Large Language Models and AI technologies. Current research priorities include computational psycholinguistics, computational models of discourse, and linguistics-based language modeling. Students have the opportunity to learn from faculty across a range of departments and to participate in our interdisciplinary research community (http://gucl.georgetown.edu/). Additionally, students who apply to the Ph.D. program in Computational Linguistics may elect to participate in the university’s Interdisciplinary Concentration in Cognitive Science. Admitted Ph.D. applicants are offered full funding. Admissions and funding decisions are made without regard to domestic vs. international status. Our programs include: - MS in Computational Linguistics <https://linguistics.georgetown.edu/graduate/master-degree-programs/master-o…> (Dept. of Linguistics, Jan. 15 deadline) - Ph.D. in Computational Linguistics <https://linguistics.georgetown.edu/graduate/phd-programs> (Dept. of Linguistics, Dec. 1 deadline) - Optional: Interdisciplinary Concentration in Cognitive Science <https://cogsci.georgetown.edu/concentration/> Requirements: A bachelor’s degree by August 2025 is required to enter either program. Both programs include core courses in Linguistics, so a prior degree in Linguistics is not required. Ph.D. applicants are expected to have some programming experience and a commitment to participating in computational linguistics research. See recommendations for the Ph.D. Statement of Purpose <https://medium.com/@nschneid/inside-ph-d-admissions-what-readers-look-for-i…> . Applications: https://linguistics.georgetown.edu/programs/apply/ Questions about application requirements and procedures should be directed to Erin Esch Pereira (eee8(a)georgetown.edu), Graduate Program Coordinator, Dept. of Linguistics. Sincerely, Nathan Schneider <http://nathan.cl/>, Depts. of Linguistics and Computer Science; concentration director Amir Zeldes <https://gucorpling.org/amir/>, Dept. of Linguistics, Computational Linguistics Ethan Gotlieb Wilcox <https://wilcoxeg.github.io/>, Dept. of Linguistics, Computational Linguistics -- Ethan Gotlieb Wilcox Assistant Professor, Computational Linguistics Georgetown University www.wilcoxeg.github.io

1 0

Edge Hill Corpus Research Group: Meeting #13
by Costas Gabrielatos 02 Nov '24

02 Nov '24

The next meeting of the Edge Hill Corpus Research Group will take place online (via MS Teams) on Friday 15 November 2024, 2-4 pm (GMT). Topic: Discourse-Oriented Corpus Studies 2-3 pm Katia Adimora (Edge Hill University) Mexican immigration/immigrants in American and Mexican newspapers 3-4 pm Dan Malone (Edge Hill University) When is the extreme also typical? Using prototypicality to investigate representations of the lone-wolf terrorist Attendance is free. The abstracts and registration link are here: https://sites.edgehill.ac.uk/crg/next Registration closes on Wednesday 13 November, 11 am (GMT). If you have any questions, please contact Costas Gabrielatos (gabrielc(a)edgehill.ac.uk<mailto:gabrielc@edgehill.ac.uk>). ________________________________ Edge Hill University<http://ehu.ac.uk/home/emailfooter> Modern University of the Year, The Times and Sunday Times Good University Guide 2022<http://ehu.ac.uk/tef/emailfooter> University of the Year, Educate North 2021/21 ________________________________ This message is private and confidential. If you have received this message in error, please notify the sender and remove it from your system. Any views or opinions presented are solely those of the author and do not necessarily represent those of Edge Hill or associated companies. Edge Hill University may monitor email traffic data and also the content of email for the purposes of security and business communications during staff absence.<http://ehu.ac.uk/itspolicies/emailfooter>

1 0

Final CFP : Workshop on Challenges in Processing South Asian Languages (CHiPSAL @ COLING 2025)
by Velayuthan Menan 01 Nov '24

01 Nov '24

Hello Everyone, Based on the emails received we have extended the submission deadline to November 7th 2024. Please read below for more information on the workshop and the updated timeline. ---- Don’t miss this unique opportunity to discuss key issues and contribute to the advancement of language processing in the South Asian region, home to 25% of the world’s population and rich in linguistic and cultural diversity. Submit your papers by October 30, 2024, and join us at the first Workshop on Challenges in Processing South Asian Languages (CHiPSAL), taking place at COLING 2025 on January 19, 2025. Please submit your papers via *https://softconf.com/coling2025/CHiPSAL25 https://softconf.com/coling2025/CHiPSAL25 * ---- *CHiPSAL 2025*, the First workshop on Challenges in Processing South Asian Languages (CHiPSAL), will be held as part of the 31st International Conference on Computational Linguistics (COLING 2025) in Abu Dhabi, UAE, on *January 19, 2025*. The workshop will be conducted in *virtual mode*. CHiPSAL 2025 invites the submission of original research papers, review/opinion papers, and system demonstration papers, in short or long forms, on topics that highlight the challenges related to South Asian languages, including but not limited to the following areas: - Encoding and Unicode Issues in South Asian Scripts - Orthographic Complexities and Their Impact on Language Technology - Morphological Analysis and Generation in South Asian Languages - Dialectal Variations and Language Standardisation - Code-Mixing and Multilingualism in South Asian Contexts - Building Linguistic Resources for South Asian Languages - Speech Recognition and Synthesis for South Asian Languages - Preserving Linguistic Heritage through Technology - Benchmarking Models for South Asian Languages *Important Dates* All deadlines are 11:59PM UTC-12:00 (“Anywhere on Earth”). The First CFP Monday, 15 July 2024 Submission Deadline October 30, 2024 *November 7, 2024* Notification of acceptance November 29, 2024 Camera-ready papers December 13, 2024 Pre-recorded video due January 5, 2024 Workshop (Virtual) January 19, 2025 *For more information: https://sites.google.com/view/chipsal https://sites.google.com/view/chipsal*

1 1

Opening of the Faetar Low-Resource ASR Challenge 2025
by Ewan Dunbar 01 Nov '24

01 Nov '24

Opening of the Faetar Low-Resource ASR Challenge 2025 We are pleased to officially announce the opening of the Faetar Low-Resource ASR Challenge 2025. While we were not able to secure a special session dedicated to the challenge at the conference, we strongly encourage submission of papers describing your systems to Interspeech 2025. As such, we plan to adhere to a timeline that will allow us to return test results and announce winners in time for participants to prepare Interspeech papers (see below). Challenge website: https://perceptimatic.github.io/faetarspeech/ The Faetar Low-Resource ASR Challenge aims to focus researchers’ attention on several issues which are common to many archival collections of speech data: - noisy field recordings - lack of standard orthography, leading to noise in the transcriptions in the form of transcriber inconsistencies - only a few hours of transcribed data - a larger collection of untranscribed data - no additional data in the language (textual or speech) that is easily available - “dirty” transcriptions in documents, which contain matter that needs to be filtered out By focusing multiple research groups on a single corpus of this kind, we aim to gain deeper insights into these problems than can be achieved otherwise. The challenge uses the Faetar ASR Benchmark Corpus. Faetar (pronounced [fajdar]) is a variety of the Franco-Provençal language which developed in isolation in Italy, far from other speakers of Franco-Provençal, and in close contact with Italian. Faetar has less than 1000 speakers around the world, in Italy and in the diaspora. It is endangered, and preservation, learning, and documentation are a priority for many community members. The benchmark data represents the majority of all archived speech recordings of Faetar in existence, and it is not available from any other source. We propose four tracks: - Constrained ASR. Participants should focus on the challenge of improving ASR architectures to work with small, poor-quality sets. Participants may not use any resources to train / fine-tune their models beyond the files contained in the provided train set. No external pre-trained acoustic models or language models are allowed, and the use of the unlabelled portion of the Faetar challenge data set is not allowed either. Three other “thematic tracks” can be explored, and should not be considered mutually exclusive: - Using pre-trained acoustic models or language models. Participants focus on the most effective way to make use of models pre-trained on other languages. - Using unlabelled data. The challenge data also includes ~20 hrs of unlabelled data. Participants focus on finding the most effective way to make use of it. - Dirty data. The training data was extracted and automatically aligned from long-form audio and partial transcriptions in “cluttered” word processor files, relying on (error-prone) VAD, scraping, and alignment. Participants focus on improving the pipeline for extracting useful training data, with the ultimate goal of improving performance. Submissions will be evaluated on phone error rate (PER) on the test set. Participants are provided with a dev kit allowing them to calculate the PER on dev and train, as well as reproduce the baselines. For more information, and to register and obtain the data and the dev kit, please visit the challenge website: https://perceptimatic.github.io/faetarspeech/ For more information, or for questions, please contact us by writing to faetarasrchallenge(a)gmail.com.

1 0

2024

2023

2022

Corpora November 2024