Datasets:

COLING

/

COLING2024-papers

like 0

Follow

International Conference on Computational Linguistics 1

https://aclanthology.org/2024.rapid-1.6.bib

https://aclanthology.org/2024.rapid-1.6/

@inproceedings{themistocleous-2024-open, title = "Open Brain {AI}. Automatic Language Assessment", author = "Themistocleous, Charalambos", editor = "Kokkinakis, Dimitrios and Fraser, Kathleen C. and Themistocleous, Charalambos K. and Fors, Kristina Lundholm and Tsanas, Athanasios and Ohman, Fredrik", booktitle = "Proceedings of the Fifth Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric/developmental impairments @LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rapid-1.6", pages = "45--53", abstract = "Language assessment plays a crucial role in diagnosing and treating individuals with speech, language, and communication disorders caused by neurogenic conditions, whether developmental or acquired. To support clinical assessment and research, we developed Open Brain AI (https://openbrainai.com). This computational platform employs AI techniques, namely machine learning, natural language processing, large language models, and automatic speech-to-text transcription, to automatically analyze multilingual spoken and written productions. This paper discusses the development of Open Brain AI, the AI language processing modules, and the linguistic measurements of discourse macro-structure and micro-structure. The fast and automatic analysis of language alleviates the burden on clinicians, enabling them to streamline their workflow and allocate more time and resources to direct patient care. Open Brain AI is freely accessible, empowering clinicians to conduct critical data analyses and give more attention and resources to other critical aspects of therapy and treatment.", }

Language assessment plays a crucial role in diagnosing and treating individuals with speech, language, and communication disorders caused by neurogenic conditions, whether developmental or acquired. To support clinical assessment and research, we developed Open Brain AI (https://openbrainai.com). This computational platform employs AI techniques, namely machine learning, natural language processing, large language models, and automatic speech-to-text transcription, to automatically analyze multilingual spoken and written productions. This paper discusses the development of Open Brain AI, the AI language processing modules, and the linguistic measurements of discourse macro-structure and micro-structure. The fast and automatic analysis of language alleviates the burden on clinicians, enabling them to streamline their workflow and allocate more time and resources to direct patient care. Open Brain AI is freely accessible, empowering clinicians to conduct critical data analyses and give more attention and resources to other critical aspects of therapy and treatment.

Open Brain AI. Automatic Language Assessment

rapid-1.6

Poster

2306.06693

https://aclanthology.org/2024.rapid-1.7.bib

https://aclanthology.org/2024.rapid-1.7/

@inproceedings{mina-etal-2024-exploring, title = "Exploring the Relationship Between Intrinsic Stigma in Masked Language Models and Training Data Using the Stereotype Content Model", author = "Mina, Mario and Falc{\~a}o, J{\'u}lia and Gonzalez-Agirre, Aitor", editor = "Kokkinakis, Dimitrios and Fraser, Kathleen C. and Themistocleous, Charalambos K. and Fors, Kristina Lundholm and Tsanas, Athanasios and Ohman, Fredrik", booktitle = "Proceedings of the Fifth Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric/developmental impairments @LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rapid-1.7", pages = "54--67", abstract = "Much work has gone into developing language models of increasing size, but only recently have we begun to examine them for pernicious behaviour that could lead to harming marginalised groups. Following Lin et al. (2022) in rooting our work in psychological research, we prompt two masked language models (MLMs) of different specialisations in English and Spanish with statements from a questionnaire developed to measure stigma to determine if they treat physical and mental illnesses equally. In both models we find a statistically significant difference in the treatment of physical and mental illnesses across most if not all latent constructs as measured by the questionnaire, and thus they are more likely to associate mental illnesses with stigma. We then examine their training data or data retrieved from the same domain using a computational implementation of the Stereotype Content Model (SCM) (Fiske et al., 2002; Fraser et al., 2021) to interpret the questionnaire results based on the SCM values as reflected in the data. We observe that model behaviour can largely be explained by the distribution of the mentions of illnesses according to their SCM values.", }

Much work has gone into developing language models of increasing size, but only recently have we begun to examine them for pernicious behaviour that could lead to harming marginalised groups. Following Lin et al. (2022) in rooting our work in psychological research, we prompt two masked language models (MLMs) of different specialisations in English and Spanish with statements from a questionnaire developed to measure stigma to determine if they treat physical and mental illnesses equally. In both models we find a statistically significant difference in the treatment of physical and mental illnesses across most if not all latent constructs as measured by the questionnaire, and thus they are more likely to associate mental illnesses with stigma. We then examine their training data or data retrieved from the same domain using a computational implementation of the Stereotype Content Model (SCM) (Fiske et al., 2002; Fraser et al., 2021) to interpret the questionnaire results based on the SCM values as reflected in the data. We observe that model behaviour can largely be explained by the distribution of the mentions of illnesses according to their SCM values.

Exploring the Relationship Between Intrinsic Stigma in Masked Language Models and Training Data Using the Stereotype Content Model

rapid-1.7

Poster

https://aclanthology.org/2024.rapid-1.8.bib

https://aclanthology.org/2024.rapid-1.8/

@inproceedings{stamou-etal-2024-establishing, title = "Establishing Control Corpora for Depression Detection in {M}odern {G}reek: Methodological Insights", author = "Stamou, Vivian and Mikros, George and Markopoulos, George and Varlokosta, Spyridoula", editor = "Kokkinakis, Dimitrios and Fraser, Kathleen C. and Themistocleous, Charalambos K. and Fors, Kristina Lundholm and Tsanas, Athanasios and Ohman, Fredrik", booktitle = "Proceedings of the Fifth Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric/developmental impairments @LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rapid-1.8", pages = "68--76", abstract = "This paper presents a methodological approach for establishing control corpora in the context of depression detection in the Modern Greek language. We discuss various methods used to create control corpora, focusing on the challenge of selecting representative samples from the general population when the target reference is the depressed population. Our approach includes traditional random selection among Twitter users, as well as an innovative method for creating topic-oriented control corpora. Through this study, we provide insights into the development of control corpora, offering valuable considerations for researchers working on similar projects in linguistic analysis and mental health studies. In addition, we identify several dominant topics in the depressed population such as religion, sentiments, health and digestion, which seem to align with findings consistently reported in the literature", }

This paper presents a methodological approach for establishing control corpora in the context of depression detection in the Modern Greek language. We discuss various methods used to create control corpora, focusing on the challenge of selecting representative samples from the general population when the target reference is the depressed population. Our approach includes traditional random selection among Twitter users, as well as an innovative method for creating topic-oriented control corpora. Through this study, we provide insights into the development of control corpora, offering valuable considerations for researchers working on similar projects in linguistic analysis and mental health studies. In addition, we identify several dominant topics in the depressed population such as religion, sentiments, health and digestion, which seem to align with findings consistently reported in the literature

Establishing Control Corpora for Depression Detection in Modern Greek: Methodological Insights

rapid-1.8

Poster

https://aclanthology.org/2024.rapid-1.9.bib

https://aclanthology.org/2024.rapid-1.9/

@inproceedings{khanna-stark-2024-preliminary, title = "A Preliminary Evaluation of Semantic Coherence and Cohesion in Aphasic and Non-Aphasic Discourse Across Test and Retest", author = "Khanna, Snigdha and Stark, Brielle C.", editor = "Kokkinakis, Dimitrios and Fraser, Kathleen C. and Themistocleous, Charalambos K. and Fors, Kristina Lundholm and Tsanas, Athanasios and Ohman, Fredrik", booktitle = "Proceedings of the Fifth Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric/developmental impairments @LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rapid-1.9", pages = "77--86", abstract = "This paper evaluates global and local semantic coherence in aphasic and non-aphasic discourse tasks using the Tool for the Automatic Analysis of Cohesion (TAACO). The motivation for this paper stems from a lack of automatic methods to evaluate discourse-level phenomena, such as semantic cohesion, in transcripts derived from persons with aphasia. It leverages existing test-retest data to evaluate two main objectives: (1) Test-Retest Reliability, to identify if variables significantly differ across test and retest time points for either group (aphasia, control), and (2) Inter-Group Discourse Cohesion, where aphasic discourse is expected to be less cohesive than control discourse, resulting in lower cohesion scores for the aphasia group. Exploratory analysis examines correlations between variables for both groups, identifying any relationships between word-level and sentence-level semantic variables. Results verify that semantic cohesion and coherence are generally preserved in both groups, except for word-level and a few sentence-level semantic measures,w which are higher for the control group. Overall, variables tend to be reliable across time points for both groups. Notably, the aphasia group demonstrates more variability in cohesion than the control group, which is to be expected after brain injury. A close relationship between word-level indices and other indices is observed, suggesting a disconnection between word-level factors and sentence-level metrics.", }

This paper evaluates global and local semantic coherence in aphasic and non-aphasic discourse tasks using the Tool for the Automatic Analysis of Cohesion (TAACO). The motivation for this paper stems from a lack of automatic methods to evaluate discourse-level phenomena, such as semantic cohesion, in transcripts derived from persons with aphasia. It leverages existing test-retest data to evaluate two main objectives: (1) Test-Retest Reliability, to identify if variables significantly differ across test and retest time points for either group (aphasia, control), and (2) Inter-Group Discourse Cohesion, where aphasic discourse is expected to be less cohesive than control discourse, resulting in lower cohesion scores for the aphasia group. Exploratory analysis examines correlations between variables for both groups, identifying any relationships between word-level and sentence-level semantic variables. Results verify that semantic cohesion and coherence are generally preserved in both groups, except for word-level and a few sentence-level semantic measures,w which are higher for the control group. Overall, variables tend to be reliable across time points for both groups. Notably, the aphasia group demonstrates more variability in cohesion than the control group, which is to be expected after brain injury. A close relationship between word-level indices and other indices is observed, suggesting a disconnection between word-level factors and sentence-level metrics.

A Preliminary Evaluation of Semantic Coherence and Cohesion in Aphasic and Non-Aphasic Discourse Across Test and Retest

rapid-1.9

Poster

https://aclanthology.org/2024.rapid-1.10.bib

https://aclanthology.org/2024.rapid-1.10/

@inproceedings{cafiero-etal-2024-harnessing, title = "Harnessing Linguistic Analysis for {ADHD} Diagnosis Support: A Stylometric Approach to Self-Defining Memories", author = {Cafiero, Florian Rapha{\"e}l and Barrios Rudloff, Juan and Gabay, Simon}, editor = "Kokkinakis, Dimitrios and Fraser, Kathleen C. and Themistocleous, Charalambos K. and Fors, Kristina Lundholm and Tsanas, Athanasios and Ohman, Fredrik", booktitle = "Proceedings of the Fifth Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric/developmental impairments @LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rapid-1.10", pages = "87--94", abstract = "This study explores the potential of stylometric analysis in identifying Self-Defining Memories (SDMs) authored by individuals with Attention-Deficit/Hyperactivity Disorder (ADHD) versus a control group. A sample of 198 SDMs were written by 66 adolescents and were then analysed using Support Vector Classifiers (SVC). The analysis included a variety of linguistic features such as character 3-grams, function words, sentence length, or lexical richness among others. It also included metadata about the participants (gender, age) and their SDMs (self-reported sentiment after recalling their memories). The results reveal a promising ability of linguistic analysis to accurately classify SDMs, with perfect prediction (F1=1.0) in the contextually simpler setup of text-by-text prediction, and satisfactory levels of precision (F1 = 0.77) when predicting individual by individual. Such results highlight the significant role that linguistic characteristics play in reflecting the distinctive cognitive patterns associated with ADHD. While not a substitute for professional diagnosis, textual analysis offers a supportive avenue for early detection and a deeper understanding of ADHD.", }

This study explores the potential of stylometric analysis in identifying Self-Defining Memories (SDMs) authored by individuals with Attention-Deficit/Hyperactivity Disorder (ADHD) versus a control group. A sample of 198 SDMs were written by 66 adolescents and were then analysed using Support Vector Classifiers (SVC). The analysis included a variety of linguistic features such as character 3-grams, function words, sentence length, or lexical richness among others. It also included metadata about the participants (gender, age) and their SDMs (self-reported sentiment after recalling their memories). The results reveal a promising ability of linguistic analysis to accurately classify SDMs, with perfect prediction (F1=1.0) in the contextually simpler setup of text-by-text prediction, and satisfactory levels of precision (F1 = 0.77) when predicting individual by individual. Such results highlight the significant role that linguistic characteristics play in reflecting the distinctive cognitive patterns associated with ADHD. While not a substitute for professional diagnosis, textual analysis offers a supportive avenue for early detection and a deeper understanding of ADHD.

Harnessing Linguistic Analysis for ADHD Diagnosis Support: A Stylometric Approach to Self-Defining Memories

rapid-1.10

Poster

https://aclanthology.org/2024.rapid-1.11.bib

https://aclanthology.org/2024.rapid-1.11/

@inproceedings{choi-etal-2024-crosslinguistic, title = "Crosslinguistic Acoustic Feature-based Dementia Classification Using Advanced Learning Architectures", author = "Choi, Anna Seo Gyeong and Kim, Jin-seo and Kim, Seo-hee and Back, Min Seok and Cho, Sunghye", editor = "Kokkinakis, Dimitrios and Fraser, Kathleen C. and Themistocleous, Charalambos K. and Fors, Kristina Lundholm and Tsanas, Athanasios and Ohman, Fredrik", booktitle = "Proceedings of the Fifth Workshop on Resources and ProcessIng of linguistic, para-linguistic and extra-linguistic Data from people with various forms of cognitive/psychiatric/developmental impairments @LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rapid-1.11", pages = "95--100", abstract = "In this study, we rigorously evaluated eight machine learning and deep learning classifiers for identifying Alzheimer{'}s Disease (AD) patients using crosslinguistic acoustic features automatically extracted from one-minute oral picture descriptions produced by speakers of American English, Korean, and Mandarin Chinese. We employed eGeMAPSv2 and ComParE feature sets on segmented and non-segmented audio data. The Multilayer Perceptron model showed the highest performance, achieving an accuracy of 83.54{\%} and an AUC of 0.8 on the ComParE features extracted from non-segmented picture description data. Our findings suggest that classifiers trained with acoustic features extracted from one-minute picture description data in multiple languages are highly promising as a quick, language-universal, large-scale, remote screening tool for AD. However, the dataset included predominantly English-speaking participants, indicating the need for more balanced multilingual datasets in future research.", }

In this study, we rigorously evaluated eight machine learning and deep learning classifiers for identifying Alzheimer{'}s Disease (AD) patients using crosslinguistic acoustic features automatically extracted from one-minute oral picture descriptions produced by speakers of American English, Korean, and Mandarin Chinese. We employed eGeMAPSv2 and ComParE feature sets on segmented and non-segmented audio data. The Multilayer Perceptron model showed the highest performance, achieving an accuracy of 83.54{\%} and an AUC of 0.8 on the ComParE features extracted from non-segmented picture description data. Our findings suggest that classifiers trained with acoustic features extracted from one-minute picture description data in multiple languages are highly promising as a quick, language-universal, large-scale, remote screening tool for AD. However, the dataset included predominantly English-speaking participants, indicating the need for more balanced multilingual datasets in future research.

Crosslinguistic Acoustic Feature-based Dementia Classification Using Advanced Learning Architectures

rapid-1.11

Poster

https://aclanthology.org/2024.readi-1.1.bib

https://aclanthology.org/2024.readi-1.1/

@inproceedings{cripwell-etal-2024-evaluating, title = "Evaluating Document Simplification: On the Importance of Separately Assessing Simplicity and Meaning Preservation", author = {Cripwell, Liam and Legrand, Jo{\"e}l and Gardent, Claire}, editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.1", pages = "1--14", abstract = "Text simplification intends to make a text easier to read while preserving its core meaning. Intuitively and as shown in previous works, these two dimensions (simplification and meaning preservation) are often-times inversely correlated. An overly conservative text will fail to simplify sufficiently, whereas extreme simplification will degrade meaning preservation. Yet, popular evaluation metrics either aggregate meaning preservation and simplification into a single score (SARI, LENS), or target meaning preservation alone (BERTScore, QuestEval). Moreover, these metrics usually require a set of references and most previous work has only focused on sentence-level simplification. In this paper, we focus on the evaluation of document-level text simplification and compare existing models using distinct metrics for meaning preservation and simplification. We leverage existing metrics from similar tasks and introduce a reference-less metric variant for simplicity, showing that models are mostly biased towards either simplification or meaning preservation, seldom performing well on both dimensions. Making use of the fact that the metrics we use are all reference-less, we also investigate the performance of existing models when applied to unseen data (where reference simplifications are unavailable).", }

Text simplification intends to make a text easier to read while preserving its core meaning. Intuitively and as shown in previous works, these two dimensions (simplification and meaning preservation) are often-times inversely correlated. An overly conservative text will fail to simplify sufficiently, whereas extreme simplification will degrade meaning preservation. Yet, popular evaluation metrics either aggregate meaning preservation and simplification into a single score (SARI, LENS), or target meaning preservation alone (BERTScore, QuestEval). Moreover, these metrics usually require a set of references and most previous work has only focused on sentence-level simplification. In this paper, we focus on the evaluation of document-level text simplification and compare existing models using distinct metrics for meaning preservation and simplification. We leverage existing metrics from similar tasks and introduce a reference-less metric variant for simplicity, showing that models are mostly biased towards either simplification or meaning preservation, seldom performing well on both dimensions. Making use of the fact that the metrics we use are all reference-less, we also investigate the performance of existing models when applied to unseen data (where reference simplifications are unavailable).

Evaluating Document Simplification: On the Importance of Separately Assessing Simplicity and Meaning Preservation

readi-1.1

Poster

2404.03278

https://huggingface.co/papers/2404.03278

https://aclanthology.org/2024.readi-1.2.bib

https://aclanthology.org/2024.readi-1.2/

@inproceedings{hjartarson-fridriksdottir-2024-malmon, title = "Malmon: A Crowd-Sourcing Platform for Simple Language", author = {Hjartarson, Helgi Bj{\"o}rn and Fri{\dh}riksd{\'o}ttir, Steinunn Rut}, editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.2", pages = "15--21", abstract = "This paper presents a crowd-sourcing platform designed to address the need for parallel corpora in the field of Automatic Text Simplification (ATS). ATS aims to automatically reduce the linguistic complexity of text to aid individuals with reading difficulties, such as those with cognitive disorders, dyslexia, children, and non-native speakers. ATS does not only facilitate improved reading comprehension among these groups but can also enhance the preprocessing stage for various NLP tasks through summarization, contextual simplification, and paraphrasing. Our work introduces a language independent, openly accessible platform that crowdsources training data for ATS models, potentially benefiting low-resource languages where parallel data is scarce. The platform can efficiently aid in the collection of parallel corpora by providing a user-friendly data-collection environment. Furthermore, using human crowd-workers for the data collection process offers a potential resource for linguistic research on text simplification practices. The paper discusses the platform{'}s architecture, built with modern web technologies, and its user-friendly interface designed to encourage widespread participation. Through gamification and a robust admin panel, the platform incentivizes high-quality data collection and engagement from crowdworkers.", }

This paper presents a crowd-sourcing platform designed to address the need for parallel corpora in the field of Automatic Text Simplification (ATS). ATS aims to automatically reduce the linguistic complexity of text to aid individuals with reading difficulties, such as those with cognitive disorders, dyslexia, children, and non-native speakers. ATS does not only facilitate improved reading comprehension among these groups but can also enhance the preprocessing stage for various NLP tasks through summarization, contextual simplification, and paraphrasing. Our work introduces a language independent, openly accessible platform that crowdsources training data for ATS models, potentially benefiting low-resource languages where parallel data is scarce. The platform can efficiently aid in the collection of parallel corpora by providing a user-friendly data-collection environment. Furthermore, using human crowd-workers for the data collection process offers a potential resource for linguistic research on text simplification practices. The paper discusses the platform{'}s architecture, built with modern web technologies, and its user-friendly interface designed to encourage widespread participation. Through gamification and a robust admin panel, the platform incentivizes high-quality data collection and engagement from crowdworkers.

Malmon: A Crowd-Sourcing Platform for Simple Language

readi-1.2

Poster

https://aclanthology.org/2024.readi-1.3.bib

https://aclanthology.org/2024.readi-1.3/

@inproceedings{sauberli-clematide-2024-automatic, title = "Automatic Generation and Evaluation of Reading Comprehension Test Items with Large Language Models", author = {S{\"a}uberli, Andreas and Clematide, Simon}, editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.3", pages = "22--37", abstract = "Reading comprehension tests are used in a variety of applications, reaching from education to assessing the comprehensibility of simplified texts. However, creating such tests manually and ensuring their quality is difficult and time-consuming. In this paper, we explore how large language models (LLMs) can be used to generate and evaluate multiple-choice reading comprehension items. To this end, we compiled a dataset of German reading comprehension items and developed a new protocol for human and automatic evaluation, including a metric we call text informativity, which is based on guessability and answerability. We then used this protocol and the dataset to evaluate the quality of items generated by Llama 2 and GPT-4. Our results suggest that both models are capable of generating items of acceptable quality in a zero-shot setting, but GPT-4 clearly outperforms Llama 2. We also show that LLMs can be used for automatic evaluation by eliciting item reponses from them. In this scenario, evaluation results with GPT-4 were the most similar to human annotators. Overall, zero-shot generation with LLMs is a promising approach for generating and evaluating reading comprehension test items, in particular for languages without large amounts of available data.", }

Reading comprehension tests are used in a variety of applications, reaching from education to assessing the comprehensibility of simplified texts. However, creating such tests manually and ensuring their quality is difficult and time-consuming. In this paper, we explore how large language models (LLMs) can be used to generate and evaluate multiple-choice reading comprehension items. To this end, we compiled a dataset of German reading comprehension items and developed a new protocol for human and automatic evaluation, including a metric we call text informativity, which is based on guessability and answerability. We then used this protocol and the dataset to evaluate the quality of items generated by Llama 2 and GPT-4. Our results suggest that both models are capable of generating items of acceptable quality in a zero-shot setting, but GPT-4 clearly outperforms Llama 2. We also show that LLMs can be used for automatic evaluation by eliciting item reponses from them. In this scenario, evaluation results with GPT-4 were the most similar to human annotators. Overall, zero-shot generation with LLMs is a promising approach for generating and evaluating reading comprehension test items, in particular for languages without large amounts of available data.

Automatic Generation and Evaluation of Reading Comprehension Test Items with Large Language Models

readi-1.3

Poster

2404.07720

https://aclanthology.org/2024.readi-1.4.bib

https://aclanthology.org/2024.readi-1.4/

@inproceedings{shardlow-etal-2024-extensible, title = "An Extensible Massively Multilingual Lexical Simplification Pipeline Dataset using the {M}ulti{LS} Framework", author = {Shardlow, Matthew and Alva-Manchego, Fernando and Batista-Navarro, Riza and Bott, Stefan and Calderon Ramirez, Saul and Cardon, R{\'e}mi and Fran{\c{c}}ois, Thomas and Hayakawa, Akio and Horbach, Andrea and H{\"u}lsing, Anna and Ide, Yusuke and Imperial, Joseph Marvin and Nohejl, Adam and North, Kai and Occhipinti, Laura and Per{\'e}z Rojas, Nelson and Raihan, Nishat and Ranasinghe, Tharindu and Solis Salazar, Martin and Zampieri, Marcos and Saggion, Horacio}, editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.4", pages = "38--46", abstract = "We present preliminary findings on the MultiLS dataset, developed in support of the 2024 Multilingual Lexical Simplification Pipeline (MLSP) Shared Task. This dataset currently comprises of 300 instances of lexical complexity prediction and lexical simplification across 10 languages. In this paper, we (1) describe the annotation protocol in support of the contribution of future datasets and (2) present summary statistics on the existing data that we have gathered. Multilingual lexical simplification can be used to support low-ability readers to engage with otherwise difficult texts in their native, often low-resourced, languages.", }

We present preliminary findings on the MultiLS dataset, developed in support of the 2024 Multilingual Lexical Simplification Pipeline (MLSP) Shared Task. This dataset currently comprises of 300 instances of lexical complexity prediction and lexical simplification across 10 languages. In this paper, we (1) describe the annotation protocol in support of the contribution of future datasets and (2) present summary statistics on the existing data that we have gathered. Multilingual lexical simplification can be used to support low-ability readers to engage with otherwise difficult texts in their native, often low-resourced, languages.

An Extensible Massively Multilingual Lexical Simplification Pipeline Dataset using the MultiLS Framework

readi-1.4

Poster

https://aclanthology.org/2024.readi-1.5.bib

https://aclanthology.org/2024.readi-1.5/

@inproceedings{yamanaka-tokunaga-2024-siera, title = "{SIERA}: An Evaluation Metric for Text Simplification using the Ranking Model and Data Augmentation by Edit Operations", author = "Yamanaka, Hikaru and Tokunaga, Takenobu", editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.5", pages = "47--58", abstract = "Automatic evaluation metrics are indispensable for text simplification (TS) research. The past TS research adopts three evaluation aspects: fluency, meaning preservation and simplicity. However, there is little consensus on a metric to measure simplicity, a unique aspect of TS compared with other text generation tasks. In addition, many of the existing metrics require reference simplified texts for evaluation. Thus, the cost of collecting reference texts is also an issue. This study proposes a new automatic evaluation metric, SIERA, for sentence simplification. SIERA employs a ranking model for the order relation of simplicity, which is trained by pairs of the original and simplified sentences. It does not require reference sentences for either training or evaluation. The sentence pairs for training are further augmented by the proposed method that utlizes edit operations to generate intermediate sentences with the simplicity between the original and simplified sentences. Using three evaluation datasets for text simplification, we compare SIERA with other metrics by calculating the correlations between metric values and human ratings. The results showed SIERA{'}s superiority over other metrics with a reservation that the quality of evaluation sentences is consistent with that of the training data.", }

Automatic evaluation metrics are indispensable for text simplification (TS) research. The past TS research adopts three evaluation aspects: fluency, meaning preservation and simplicity. However, there is little consensus on a metric to measure simplicity, a unique aspect of TS compared with other text generation tasks. In addition, many of the existing metrics require reference simplified texts for evaluation. Thus, the cost of collecting reference texts is also an issue. This study proposes a new automatic evaluation metric, SIERA, for sentence simplification. SIERA employs a ranking model for the order relation of simplicity, which is trained by pairs of the original and simplified sentences. It does not require reference sentences for either training or evaluation. The sentence pairs for training are further augmented by the proposed method that utlizes edit operations to generate intermediate sentences with the simplicity between the original and simplified sentences. Using three evaluation datasets for text simplification, we compare SIERA with other metrics by calculating the correlations between metric values and human ratings. The results showed SIERA{'}s superiority over other metrics with a reservation that the quality of evaluation sentences is consistent with that of the training data.

SIERA: An Evaluation Metric for Text Simplification using the Ranking Model and Data Augmentation by Edit Operations

readi-1.5

Poster

https://aclanthology.org/2024.readi-1.6.bib

https://aclanthology.org/2024.readi-1.6/

@inproceedings{athugodage-etal-2024-transfer, title = "Transfer Learning for {R}ussian Legal Text Simplification", author = "Athugodage, Mark and Mitrofanove, Olga and Gudkov, Vadim", editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.6", pages = "59--69", abstract = "We present novel results in legal text simplification for Russian. We introduce the first dataset for such a task in Russian - a parallel corpus based on the data extracted from {``}Rossiyskaya Gazeta Legal Papers{''}. In this study we discuss three approaches for text simplification which involve T5 and GPT model architectures. We evaluate the proposed models on a set of metrics: ROUGE, SARI and BERTScore. We also analysed the models{'} results on such readability indices as Flesch-Kinkaid Grade Level and Gunning Fog Index. And, finally, we performed human evaluation of simplified texts generated by T5 and GPT models; expertise was carried out by native speakers of Russian and Russian lawyers. In this research we compared T5 and GPT architectures for text simplification task and found out that GPT handles better when it is fine-tuned on dataset of coped texts. Our research makes a big step in improving Russian legal text readability and accessibility for common people.", }

We present novel results in legal text simplification for Russian. We introduce the first dataset for such a task in Russian - a parallel corpus based on the data extracted from {``}Rossiyskaya Gazeta Legal Papers{''}. In this study we discuss three approaches for text simplification which involve T5 and GPT model architectures. We evaluate the proposed models on a set of metrics: ROUGE, SARI and BERTScore. We also analysed the models{'} results on such readability indices as Flesch-Kinkaid Grade Level and Gunning Fog Index. And, finally, we performed human evaluation of simplified texts generated by T5 and GPT models; expertise was carried out by native speakers of Russian and Russian lawyers. In this research we compared T5 and GPT architectures for text simplification task and found out that GPT handles better when it is fine-tuned on dataset of coped texts. Our research makes a big step in improving Russian legal text readability and accessibility for common people.

Transfer Learning for Russian Legal Text Simplification

readi-1.6

Poster

https://aclanthology.org/2024.readi-1.7.bib

https://aclanthology.org/2024.readi-1.7/

@inproceedings{deleanu-etal-2024-accessible, title = "Accessible Communication: a systematic review and comparative analysis of official {E}nglish Easy-to-Understand ({E}2{U}) language guidelines", author = "Deleanu, Andreea Maria and Orasan, Constantin and Braun, Sabine", editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.7", pages = "70--92", abstract = "Easy-to-Understand (E2U) language varieties have been recognized by the United Nation{'}s Convention on the Rights of Persons with Disabilities (2006) as a means to guarantee the fundamental right to Accessible Communication. Increased awareness has driven changes in European (European Commission, 2015, 2021; European Parliament, 2016) and International legislation (ODI, 2010), prompting public-sector and other institutions to offer domain-specific content into E2U language to prevent communicative exclusion of those facing cognitive barriers (COGA, 2017; Maa{\ss}, 2020; Perego, 2020). However, guidance on what it is that makes language actually {`}easier to understand{'} is still fragmented and vague. For this reason, we carried out a systematic review of official guidelines for English Plain Language and Easy Language to identify the most effective lexical, syntactic and adaptation strategies that can reduce complexity in verbal discourse according to official bodies. This article will present the methods and preliminary results of the guidelines analysis.", }

Easy-to-Understand (E2U) language varieties have been recognized by the United Nation{'}s Convention on the Rights of Persons with Disabilities (2006) as a means to guarantee the fundamental right to Accessible Communication. Increased awareness has driven changes in European (European Commission, 2015, 2021; European Parliament, 2016) and International legislation (ODI, 2010), prompting public-sector and other institutions to offer domain-specific content into E2U language to prevent communicative exclusion of those facing cognitive barriers (COGA, 2017; Maa{\ss}, 2020; Perego, 2020). However, guidance on what it is that makes language actually {`}easier to understand{'} is still fragmented and vague. For this reason, we carried out a systematic review of official guidelines for English Plain Language and Easy Language to identify the most effective lexical, syntactic and adaptation strategies that can reduce complexity in verbal discourse according to official bodies. This article will present the methods and preliminary results of the guidelines analysis.

Accessible Communication: a systematic review and comparative analysis of official English Easy-to-Understand (E2U) language guidelines

readi-1.7

Poster

https://aclanthology.org/2024.readi-1.8.bib

https://aclanthology.org/2024.readi-1.8/

@inproceedings{madina-etal-2024-languagetool, title = "{L}anguage{T}ool as a {CAT} tool for Easy-to-Read in {S}panish", author = "Madina, Margot and Gonzalez-Dios, Itziar and Siegel, Melanie", editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.8", pages = "93--101", abstract = "Easy-to-Read (E2R) is an approach to content creation that emphasizes simplicity and clarity in language to make texts more accessible to readers with cognitive challenges or learning disabilities. The Spanish version of E2R is called Lectura F{\'a}cil (LF). E2R and its variants, such as LF, focus on straightforward language and structure to enhance readability. The manual production of such texts is both time and resource expensive. In this work, we have developed LFWriteAssist, an authoring support tool that aligns with the guidelines of LF. It is underpinned by the functionalities of LanguageTool, a free and open source grammar, style and spelling checker. Our tool assists in ensuring compliance with LF standard, provides definitions for complex, polysemic, or infrequently used terms, and acronym extensions. The tool is primarily targeted at LF creators, as it serves as an authoring aid, identifying any rule infringements and assisting with language simplifications. However, it can be used by anyone who seek to enhance text readability and inclusivity. The tool{'}s code is made available as open source, thereby contributing to the wider effort of creating inclusive and comprehensible content.", }

Easy-to-Read (E2R) is an approach to content creation that emphasizes simplicity and clarity in language to make texts more accessible to readers with cognitive challenges or learning disabilities. The Spanish version of E2R is called Lectura F{\'a}cil (LF). E2R and its variants, such as LF, focus on straightforward language and structure to enhance readability. The manual production of such texts is both time and resource expensive. In this work, we have developed LFWriteAssist, an authoring support tool that aligns with the guidelines of LF. It is underpinned by the functionalities of LanguageTool, a free and open source grammar, style and spelling checker. Our tool assists in ensuring compliance with LF standard, provides definitions for complex, polysemic, or infrequently used terms, and acronym extensions. The tool is primarily targeted at LF creators, as it serves as an authoring aid, identifying any rule infringements and assisting with language simplifications. However, it can be used by anyone who seek to enhance text readability and inclusivity. The tool{'}s code is made available as open source, thereby contributing to the wider effort of creating inclusive and comprehensible content.

LanguageTool as a CAT tool for Easy-to-Read in Spanish

readi-1.8

Poster

https://aclanthology.org/2024.readi-1.9.bib

https://aclanthology.org/2024.readi-1.9/

@inproceedings{wilkens-etal-2024-paying, title = "Paying attention to the words: explaining readability prediction for {F}rench as a foreign language", author = "Wilkens, Rodrigo and Watrin, Patrick and Fran{\c{c}}ois, Thomas", editor = "Wilkens, Rodrigo and Cardon, R{\'e}mi and Todirascu, Amalia and Gala, N{\'u}ria", booktitle = "Proceedings of the 3rd Workshop on Tools and Resources for People with REAding DIfficulties (READI) @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.readi-1.9", pages = "102--115", abstract = "Automatic text Readability Assessment (ARA) has been seen as a way of helping people with reading difficulties. Recent advancements in Natural Language Processing have shifted ARA from linguistic-based models to more precise black-box models. However, this shift has weakened the alignment between ARA models and the reading literature, potentially leading to inaccurate predictions based on unintended factors. In this paper, we investigate the explainability of ARA models, inspecting the relationship between attention mechanism scores, ARA features, and CEFR level predictions made by the model. We propose a method for identifying features associated with the predictions made by a model through the use of the attention mechanism. Exploring three feature families (i.e., psycho-linguistic, work frequency and graded lexicon), we associated features with the model{'}s attention heads. Finally, while not fully explanatory of the model{'}s performance, the correlations of these associations surpass those between features and text readability levels.", }

Automatic text Readability Assessment (ARA) has been seen as a way of helping people with reading difficulties. Recent advancements in Natural Language Processing have shifted ARA from linguistic-based models to more precise black-box models. However, this shift has weakened the alignment between ARA models and the reading literature, potentially leading to inaccurate predictions based on unintended factors. In this paper, we investigate the explainability of ARA models, inspecting the relationship between attention mechanism scores, ARA features, and CEFR level predictions made by the model. We propose a method for identifying features associated with the predictions made by a model through the use of the attention mechanism. Exploring three feature families (i.e., psycho-linguistic, work frequency and graded lexicon), we associated features with the model{'}s attention heads. Finally, while not fully explanatory of the model{'}s performance, the correlations of these associations surpass those between features and text readability levels.

Paying attention to the words: explaining readability prediction for French as a foreign language

readi-1.9

Poster

https://aclanthology.org/2024.rfp-1.1.bib

https://aclanthology.org/2024.rfp-1.1/

@inproceedings{remijnse-etal-2024-tracking, title = "Tracking Perspectives on Event Participants: a Structural Analysis of the Framing of Real-World Events in Co-Referential Corpora", author = "Remijnse, Levi and Sommerauer, Pia and Fokkens, Antske and Vossen, Piek T.J.M.", editor = "Sommerauer, Pia and Caselli, Tommaso and Nissim, Malvina and Remijnse, Levi and Vossen, Piek", booktitle = "Proceedings of the First Workshop on Reference, Framing, and Perspective @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rfp-1.1", pages = "1--12", abstract = "In this paper, we present the outcome of a structural linguistic analysis performed on a referentially grounded FrameNet dataset. In this dataset, multiple Dutch events are referenced by multiple co-referential Dutch news texts. Mentions in those documents are annotated with respect to their referential grounding (i.e., links to structured Wikidata), and their conceptual representation (i.e., frames). Provided with each document{'}s temporal reporting distance, we selected documents for two events - the Utrecht shooting and MH17 - and performed an analysis in which we tracked the events{'} participants over time in both their focalization (number of mentions) and their framing (distribution of frame element labels). This way, we use the carefully collected and annotated data to schematize shifts in focalization and perspectivization of the participants as a result of the constantly developing narrative surrounding the events. This novel type of linguistic research involves reference to the real-world referents and takes into account storytelling in news streams.", }

In this paper, we present the outcome of a structural linguistic analysis performed on a referentially grounded FrameNet dataset. In this dataset, multiple Dutch events are referenced by multiple co-referential Dutch news texts. Mentions in those documents are annotated with respect to their referential grounding (i.e., links to structured Wikidata), and their conceptual representation (i.e., frames). Provided with each document{'}s temporal reporting distance, we selected documents for two events - the Utrecht shooting and MH17 - and performed an analysis in which we tracked the events{'} participants over time in both their focalization (number of mentions) and their framing (distribution of frame element labels). This way, we use the carefully collected and annotated data to schematize shifts in focalization and perspectivization of the participants as a result of the constantly developing narrative surrounding the events. This novel type of linguistic research involves reference to the real-world referents and takes into account storytelling in news streams.

Tracking Perspectives on Event Participants: a Structural Analysis of the Framing of Real-World Events in Co-Referential Corpora

rfp-1.1

Poster

https://aclanthology.org/2024.rfp-1.2.bib

https://aclanthology.org/2024.rfp-1.2/

@inproceedings{lamorte-etal-2024-timeframe, title = "{T}ime{F}rame: Querying and Visualizing Event Semantic Frames in Time", author = "Lamorte, Davide and Rovera, Marco and Ferrara, Alfio and Tonelli, Sara", editor = "Sommerauer, Pia and Caselli, Tommaso and Nissim, Malvina and Remijnse, Levi and Vossen, Piek", booktitle = "Proceedings of the First Workshop on Reference, Framing, and Perspective @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rfp-1.2", pages = "13--17", abstract = "In this work we introduce TimeFrame, an online platform to easily query and visualize events and participants extracted from document collections in Italian following a frame-based approach. The system allows users to select one or more events (frames) or event categories and to display their occurrences on a timeline. Different query types, from coarse to fine-grained, are available through the interface, enabling a time-bound analysis of large historical corpora. We present three use cases based on the full archive of news published in 1948 by the newspaper {``}Corriere della Sera{''}. We show that different crucial events can be explored, providing interesting insights into the narratives around such events, the main participants and their points of view.", }

In this work we introduce TimeFrame, an online platform to easily query and visualize events and participants extracted from document collections in Italian following a frame-based approach. The system allows users to select one or more events (frames) or event categories and to display their occurrences on a timeline. Different query types, from coarse to fine-grained, are available through the interface, enabling a time-bound analysis of large historical corpora. We present three use cases based on the full archive of news published in 1948 by the newspaper {``}Corriere della Sera{''}. We show that different crucial events can be explored, providing interesting insights into the narratives around such events, the main participants and their points of view.

TimeFrame: Querying and Visualizing Event Semantic Frames in Time

rfp-1.2

Poster

https://aclanthology.org/2024.rfp-1.3.bib

https://aclanthology.org/2024.rfp-1.3/

@inproceedings{ivacic-etal-2024-comparing, title = "Comparing News Framing of Migration Crises using Zero-Shot Classification", author = "Iva{\v{c}}i{\v{c}}, Nikola and Purver, Matthew and Lind, Fabienne and Pollak, Senja and Boomgaarden, Hajo and Bajt, Veronika", editor = "Sommerauer, Pia and Caselli, Tommaso and Nissim, Malvina and Remijnse, Levi and Vossen, Piek", booktitle = "Proceedings of the First Workshop on Reference, Framing, and Perspective @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rfp-1.3", pages = "18--27", abstract = "We present an experiment on classifying news frames in a language unseen by the learner, using zero-shot cross-lingual transfer learning. We used two pre-trained multilingual Transformer Encoder neural network models and tested with four specific news frames, investigating two approaches to the resulting multi-label task: Binary Relevance (treating each frame independently) and Label Power-set (predicting each possible combination of frames). We train our classifiers on an available annotated multilingual migration news dataset and test on an unseen Slovene language migration news corpus, first evaluating performance and then using the classifiers to analyse how media framed the news during the periods of Syria and Ukraine conflict-related migrations.", }

We present an experiment on classifying news frames in a language unseen by the learner, using zero-shot cross-lingual transfer learning. We used two pre-trained multilingual Transformer Encoder neural network models and tested with four specific news frames, investigating two approaches to the resulting multi-label task: Binary Relevance (treating each frame independently) and Label Power-set (predicting each possible combination of frames). We train our classifiers on an available annotated multilingual migration news dataset and test on an unseen Slovene language migration news corpus, first evaluating performance and then using the classifiers to analyse how media framed the news during the periods of Syria and Ukraine conflict-related migrations.

Comparing News Framing of Migration Crises using Zero-Shot Classification

rfp-1.3

Poster

https://aclanthology.org/2024.rfp-1.4.bib

https://aclanthology.org/2024.rfp-1.4/

@inproceedings{gemelli-minnema-2024-manosphrames, title = "Manosphrames: exploring an {I}talian incel community through the lens of {NLP} and Frame Semantics", author = "Gemelli, Sara and Minnema, Gosse", editor = "Sommerauer, Pia and Caselli, Tommaso and Nissim, Malvina and Remijnse, Levi and Vossen, Piek", booktitle = "Proceedings of the First Workshop on Reference, Framing, and Perspective @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rfp-1.4", pages = "28--39", abstract = "We introduce a large corpus of comments extracted from an Italian online incel ({`}involuntary incelibate{'}) forum, a community of men who build a collective identity and anti-feminist ideology centered around their inability to find a sexual or romantic partner and who frequently use explicitly misogynistic language. Our corpus consists of 2.4K comments that have been manually collected, analyzed and annotated with topic labels, and a further 32K threads (300K comments) that have been automatically scraped and automatically annotated with FrameNet annotations. We show how large-scale frame semantic analysis can shed a light on what is discussed in the community, and introduce incel topic classification as a new NLP task and benchmark.", }

We introduce a large corpus of comments extracted from an Italian online incel ({`}involuntary incelibate{'}) forum, a community of men who build a collective identity and anti-feminist ideology centered around their inability to find a sexual or romantic partner and who frequently use explicitly misogynistic language. Our corpus consists of 2.4K comments that have been manually collected, analyzed and annotated with topic labels, and a further 32K threads (300K comments) that have been automatically scraped and automatically annotated with FrameNet annotations. We show how large-scale frame semantic analysis can shed a light on what is discussed in the community, and introduce incel topic classification as a new NLP task and benchmark.

Manosphrames: exploring an Italian incel community through the lens of NLP and Frame Semantics

rfp-1.4

Poster

https://aclanthology.org/2024.rfp-1.5.bib

https://aclanthology.org/2024.rfp-1.5/

@inproceedings{tan-bloem-2024-broadening, title = "Broadening the coverage of computational representations of metaphor through Dynamic Metaphor Theory", author = "Tan, Xiaojuan and Bloem, Jelke", editor = "Sommerauer, Pia and Caselli, Tommaso and Nissim, Malvina and Remijnse, Levi and Vossen, Piek", booktitle = "Proceedings of the First Workshop on Reference, Framing, and Perspective @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.rfp-1.5", pages = "40--50", abstract = "Current approaches to computational metaphor processing typically incorporate static representations of metaphor. We aim to show that this limits the coverage of such systems. We take insights from dynamic metaphor theory and discuss how existing computational models of metaphor might benefit from representing the dynamics of metaphor when applied to the analysis of conflicting discourse. We propose that a frame-based approach to metaphor representation based on the model of YinYang Dynamics of Metaphoricity (YYDM) would pave the way to more comprehensive modeling of metaphor. In particular, the metaphoricity cues of the YYDM model could be used to address the task of dynamic metaphor identification. Frame-based modeling of dynamic metaphor would facilitate the computational analysis of perspectives in conflicting discourse, with potential applications in analyzing political discourse.", }

Current approaches to computational metaphor processing typically incorporate static representations of metaphor. We aim to show that this limits the coverage of such systems. We take insights from dynamic metaphor theory and discuss how existing computational models of metaphor might benefit from representing the dynamics of metaphor when applied to the analysis of conflicting discourse. We propose that a frame-based approach to metaphor representation based on the model of YinYang Dynamics of Metaphoricity (YYDM) would pave the way to more comprehensive modeling of metaphor. In particular, the metaphoricity cues of the YYDM model could be used to address the task of dynamic metaphor identification. Frame-based modeling of dynamic metaphor would facilitate the computational analysis of perspectives in conflicting discourse, with potential applications in analyzing political discourse.

Broadening the coverage of computational representations of metaphor through Dynamic Metaphor Theory

rfp-1.5

Poster

https://aclanthology.org/2024.safety4convai-1.1.bib

https://aclanthology.org/2024.safety4convai-1.1/

@inproceedings{addlesee-2024-grounding, title = "Grounding {LLM}s to In-prompt Instructions: Reducing Hallucinations Caused by Static Pre-training Knowledge", author = "Addlesee, Angus", editor = "Dinkar, Tanvi and Attanasio, Giuseppe and Curry, Amanda Cercas and Konstas, Ioannis and Hovy, Dirk and Rieser, Verena", booktitle = "Proceedings of Safety4ConvAI: The Third Workshop on Safety for Conversational AI @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.safety4convai-1.1", pages = "1--7", abstract = "When deploying LLMs in certain commercial or research settings, domain specific knowledge must be explicitly provided within the prompt. This in-prompt knowledge can conflict with an LLM{'}s static world knowledge learned at pre-training, causing model hallucination (see examples in Table 1). In safety-critical settings, like healthcare and finance, these hallucinations can harm vulnerable users. We have curated a QA corpus containing information that LLMs could not have seen at pre-training. Using our corpus, we have probed various LLMs, manipulating both the prompt and the knowledge representation. We have found that our {`}Jodie{'} prompt consistently improves the model{'}s textual grounding to the given knowledge, and in-turn the overall answer accuracy. This is true in both the healthcare and finance domains - improving accuracy by up to 28{\%} (mean: 12{\%}). We have also identified that hierarchical and direct node-property graph structures could lead to more interpretable and controllable systems that provide a natural language interface with real-time in-domain knowledge. Our corpus will enable further work on this critical challenge.", }

When deploying LLMs in certain commercial or research settings, domain specific knowledge must be explicitly provided within the prompt. This in-prompt knowledge can conflict with an LLM{'}s static world knowledge learned at pre-training, causing model hallucination (see examples in Table 1). In safety-critical settings, like healthcare and finance, these hallucinations can harm vulnerable users. We have curated a QA corpus containing information that LLMs could not have seen at pre-training. Using our corpus, we have probed various LLMs, manipulating both the prompt and the knowledge representation. We have found that our {`}Jodie{'} prompt consistently improves the model{'}s textual grounding to the given knowledge, and in-turn the overall answer accuracy. This is true in both the healthcare and finance domains - improving accuracy by up to 28{\%} (mean: 12{\%}). We have also identified that hierarchical and direct node-property graph structures could lead to more interpretable and controllable systems that provide a natural language interface with real-time in-domain knowledge. Our corpus will enable further work on this critical challenge.

Grounding LLMs to In-prompt Instructions: Reducing Hallucinations Caused by Static Pre-training Knowledge

safety4convai-1.1

Poster

https://aclanthology.org/2024.safety4convai-1.2.bib

https://aclanthology.org/2024.safety4convai-1.2/

@inproceedings{parrish-etal-2024-diversity, title = "Diversity-Aware Annotation for Conversational {AI} Safety", author = "Parrish, Alicia and Prabhakaran, Vinodkumar and Aroyo, Lora and D{\'\i}az, Mark and Homan, Christopher M. and Serapio-Garc{\'\i}a, Greg and Taylor, Alex S. and Wang, Ding", editor = "Dinkar, Tanvi and Attanasio, Giuseppe and Curry, Amanda Cercas and Konstas, Ioannis and Hovy, Dirk and Rieser, Verena", booktitle = "Proceedings of Safety4ConvAI: The Third Workshop on Safety for Conversational AI @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.safety4convai-1.2", pages = "8--15", abstract = "How people interpret content is deeply influenced by their socio-cultural backgrounds and lived experiences. This is especially crucial when evaluating AI systems for safety, where accounting for such diversity in interpretations and potential impacts on human users will make them both more successful and inclusive. While recent work has demonstrated the importance of diversity in human ratings that underlie AI pipelines, effective and efficient ways to incorporate diverse perspectives in human data annotation pipelines is still largely elusive. In this paper, we discuss the primary challenges faced in incorporating diversity into model evaluations, and propose a practical diversity-aware annotation approach. Using an existing dataset with highly parallel safety annotations, we take as a test case a policy that prioritizes recall of safety issues, and demonstrate that our diversity-aware approach can efficiently obtain a higher recall of safety issues flagged by minoritized rater groups without hurting overall precision.", }

How people interpret content is deeply influenced by their socio-cultural backgrounds and lived experiences. This is especially crucial when evaluating AI systems for safety, where accounting for such diversity in interpretations and potential impacts on human users will make them both more successful and inclusive. While recent work has demonstrated the importance of diversity in human ratings that underlie AI pipelines, effective and efficient ways to incorporate diverse perspectives in human data annotation pipelines is still largely elusive. In this paper, we discuss the primary challenges faced in incorporating diversity into model evaluations, and propose a practical diversity-aware annotation approach. Using an existing dataset with highly parallel safety annotations, we take as a test case a policy that prioritizes recall of safety issues, and demonstrate that our diversity-aware approach can efficiently obtain a higher recall of safety issues flagged by minoritized rater groups without hurting overall precision.

Diversity-Aware Annotation for Conversational AI Safety

safety4convai-1.2

Poster

https://aclanthology.org/2024.safety4convai-1.3.bib

https://aclanthology.org/2024.safety4convai-1.3/

@inproceedings{ajayi-etal-2024-using, title = "Using Information Retrieval Techniques to Automatically Repurpose Existing Dialogue Datasets for Safe Chatbot Development", author = "Ajayi, Tunde Oluwaseyi and Negi, Gaurav and Arcan, Mihael and Buitelaar, Paul", editor = "Dinkar, Tanvi and Attanasio, Giuseppe and Curry, Amanda Cercas and Konstas, Ioannis and Hovy, Dirk and Rieser, Verena", booktitle = "Proceedings of Safety4ConvAI: The Third Workshop on Safety for Conversational AI @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.safety4convai-1.3", pages = "16--27", abstract = "There has been notable progress in the development of open-domain dialogue systems (chatbots) especially with the rapid advancement of the capabilities of Large Language Models. Chatbots excel at holding conversations in a manner that keeps a user interested and engaged. However, their responses can be unsafe, as they can respond in an offensive manner or offer harmful professional advice. As a way to mitigate this issue, recent work crowdsource datasets with exemplary responses or annotate dialogue safety datasets, which are relatively scarce compared to casual dialogues. Despite the quality of data obtained from crowdsourcing, it can be expensive and time consuming. This work proposes an effective pipeline, using information retrieval, to automatically repurpose existing dialogue datasets for safe chatbot development, as a way to address the aforementioned challenges. We select an existing dialogue dataset, revise its unsafe responses, as a way to obtain a dataset with safer responses to unsafe user inputs. We then fine-tune dialogue models on the original and revised datasets and generate responses to evaluate the safeness of the models.", }

There has been notable progress in the development of open-domain dialogue systems (chatbots) especially with the rapid advancement of the capabilities of Large Language Models. Chatbots excel at holding conversations in a manner that keeps a user interested and engaged. However, their responses can be unsafe, as they can respond in an offensive manner or offer harmful professional advice. As a way to mitigate this issue, recent work crowdsource datasets with exemplary responses or annotate dialogue safety datasets, which are relatively scarce compared to casual dialogues. Despite the quality of data obtained from crowdsourcing, it can be expensive and time consuming. This work proposes an effective pipeline, using information retrieval, to automatically repurpose existing dialogue datasets for safe chatbot development, as a way to address the aforementioned challenges. We select an existing dialogue dataset, revise its unsafe responses, as a way to obtain a dataset with safer responses to unsafe user inputs. We then fine-tune dialogue models on the original and revised datasets and generate responses to evaluate the safeness of the models.

Using Information Retrieval Techniques to Automatically Repurpose Existing Dialogue Datasets for Safe Chatbot Development

safety4convai-1.3

Poster

https://aclanthology.org/2024.safety4convai-1.4.bib

https://aclanthology.org/2024.safety4convai-1.4/

@inproceedings{dwivedi-yu-2024-fairpair, title = "{F}air{P}air: A Robust Evaluation of Biases in Language Models through Paired Perturbations", author = "Dwivedi-Yu, Jane", editor = "Dinkar, Tanvi and Attanasio, Giuseppe and Curry, Amanda Cercas and Konstas, Ioannis and Hovy, Dirk and Rieser, Verena", booktitle = "Proceedings of Safety4ConvAI: The Third Workshop on Safety for Conversational AI @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.safety4convai-1.4", pages = "28--39", abstract = "The accurate evaluation of differential treatment in language models to specific groups is critical to ensuring a positive and safe user experience. An ideal evaluation should have the properties of being robust, extendable to new groups or attributes, and being able to capture biases that appear in typical usage (rather than just extreme, rare cases). Relatedly, bias evaluation should surface not only egregious biases but also ones that are subtle and commonplace, such as a likelihood for talking about appearances with regard to women. We present FairPair, an evaluation framework for assessing differential treatment that occurs during ordinary usage. FairPair operates through counterfactual pairs, but crucially, the paired continuations are grounded in the same demographic group, which ensures equivalent comparison. Additionally, unlike prior work, our method factors in the inherent variability that comes from the generation process itself by measuring the sampling variability. We present an evaluation of several commonly used generative models and a qualitative analysis that indicates a preference for discussing family and hobbies with regard to women.", }

The accurate evaluation of differential treatment in language models to specific groups is critical to ensuring a positive and safe user experience. An ideal evaluation should have the properties of being robust, extendable to new groups or attributes, and being able to capture biases that appear in typical usage (rather than just extreme, rare cases). Relatedly, bias evaluation should surface not only egregious biases but also ones that are subtle and commonplace, such as a likelihood for talking about appearances with regard to women. We present FairPair, an evaluation framework for assessing differential treatment that occurs during ordinary usage. FairPair operates through counterfactual pairs, but crucially, the paired continuations are grounded in the same demographic group, which ensures equivalent comparison. Additionally, unlike prior work, our method factors in the inherent variability that comes from the generation process itself by measuring the sampling variability. We present an evaluation of several commonly used generative models and a qualitative analysis that indicates a preference for discussing family and hobbies with regard to women.

FairPair: A Robust Evaluation of Biases in Language Models through Paired Perturbations

safety4convai-1.4

Poster

2404.06619

https://aclanthology.org/2024.safety4convai-1.5.bib

https://aclanthology.org/2024.safety4convai-1.5/

@inproceedings{pantazopoulos-etal-2024-learning, title = "Learning To See But Forgetting To Follow: Visual Instruction Tuning Makes {LLM}s More Prone To Jailbreak Attacks", author = "Pantazopoulos, Georgios and Parekh, Amit and Nikandrou, Malvina and Suglia, Alessandro", editor = "Dinkar, Tanvi and Attanasio, Giuseppe and Curry, Amanda Cercas and Konstas, Ioannis and Hovy, Dirk and Rieser, Verena", booktitle = "Proceedings of Safety4ConvAI: The Third Workshop on Safety for Conversational AI @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.safety4convai-1.5", pages = "40--51", abstract = "Augmenting Large Language Models (LLMs) with image-understanding capabilities has resulted in a boom of high-performing Vision-Language models (VLMs). While studying the alignment of LLMs to human values has received widespread attention, the safety of VLMs has not received the same attention. In this paper, we explore the impact of jailbreaking on three state-of-the-art VLMs, each using a distinct modeling approach. By comparing each VLM to their respective LLM backbone, we find that each VLM is more susceptible to jailbreaking. We consider this as an undesirable outcome from visual instruction-tuning, which imposes a forgetting effect on an LLM{'}s safety guardrails. Therefore, we provide recommendations for future work based on evaluation strategies that aim to highlight the weaknesses of a VLM, as well as take safety measures into account during visual instruction tuning.", }

Augmenting Large Language Models (LLMs) with image-understanding capabilities has resulted in a boom of high-performing Vision-Language models (VLMs). While studying the alignment of LLMs to human values has received widespread attention, the safety of VLMs has not received the same attention. In this paper, we explore the impact of jailbreaking on three state-of-the-art VLMs, each using a distinct modeling approach. By comparing each VLM to their respective LLM backbone, we find that each VLM is more susceptible to jailbreaking. We consider this as an undesirable outcome from visual instruction-tuning, which imposes a forgetting effect on an LLM{'}s safety guardrails. Therefore, we provide recommendations for future work based on evaluation strategies that aim to highlight the weaknesses of a VLM, as well as take safety measures into account during visual instruction tuning.

Learning To See But Forgetting To Follow: Visual Instruction Tuning Makes LLMs More Prone To Jailbreak Attacks

safety4convai-1.5

Poster

2405.04403

https://aclanthology.org/2024.signlang-1.1.bib

https://aclanthology.org/2024.signlang-1.1/

@inproceedings{battisti-etal-2024-advancing, title = "Advancing Annotation for Continuous Data in {S}wiss {G}erman Sign Language", author = "Battisti, Alessia and Tissi, Katja and Sidler-Miserez, Sandra and Ebling, Sarah", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.1", pages = "1--12", }

No abstract found

Advancing Annotation for Continuous Data in Swiss German Sign Language

signlang-1.1

Poster

https://aclanthology.org/2024.signlang-1.2.bib

https://aclanthology.org/2024.signlang-1.2/

@inproceedings{battisti-etal-2024-person, title = "Person Identification from Pose Estimates in Sign Language", author = {Battisti, Alessia and van den Bold, Emma and G{\"o}hring, Anne and Holzknecht, Franz and Ebling, Sarah}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.2", pages = "13--25", }

No abstract found

Person Identification from Pose Estimates in Sign Language

signlang-1.2

Poster

https://aclanthology.org/2024.signlang-1.3.bib

https://aclanthology.org/2024.signlang-1.3/

@inproceedings{bono-etal-2024-data, title = "Data Integration, Annotation, and Transcription Methods for Sign Language Dialogue with Latency in Videoconferencing", author = "Bono, Mayumi and Okada, Tomohiro and Skobov, Victor and Adam, Robert", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.3", pages = "26--35", }

No abstract found

Data Integration, Annotation, and Transcription Methods for Sign Language Dialogue with Latency in Videoconferencing

signlang-1.3

Poster

https://aclanthology.org/2024.signlang-1.4.bib

https://aclanthology.org/2024.signlang-1.4/

@inproceedings{borstell-2024-evaluating, title = "Evaluating the Alignment of Utterances in the {S}wedish {S}ign {L}anguage Corpus", author = {B{\"o}rstell, Carl}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.4", pages = "36--45", }

No abstract found

Evaluating the Alignment of Utterances in the Swedish Sign Language Corpus

signlang-1.4

Poster

https://aclanthology.org/2024.signlang-1.5.bib

https://aclanthology.org/2024.signlang-1.5/

@inproceedings{borstell-2024-approach, title = "How to Approach Lexical Variation in Sign Language Corpora", author = {B{\"o}rstell, Carl}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.5", pages = "46--53", }

No abstract found

How to Approach Lexical Variation in Sign Language Corpora

signlang-1.5

Poster

https://aclanthology.org/2024.signlang-1.6.bib

https://aclanthology.org/2024.signlang-1.6/

@inproceedings{desai-etal-2024-systemic, title = "Systemic Biases in Sign Language {AI} Research: A Deaf-Led Call to Reevaluate Research Agendas", author = "Desai, Aashaka and De Meulder, Maartje and Hochgesang, Julie A. and Kocab, Annemarie and Lu, Alex X.", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.6", pages = "54--65", }

No abstract found

Systemic Biases in Sign Language AI Research: A Deaf-Led Call to Reevaluate Research Agendas

signlang-1.6

Poster

2403.02563

https://aclanthology.org/2024.signlang-1.7.bib

https://aclanthology.org/2024.signlang-1.7/

@inproceedings{esselink-etal-2024-evaluating, title = "Evaluating Inter-Annotator Agreement for Non-Manual Markers in Sign Languages", author = "Esselink, Lyke D. and Oomen, Marloes and Roelofsen, Floris", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.7", pages = "66--76", }

No abstract found

Evaluating Inter-Annotator Agreement for Non-Manual Markers in Sign Languages

signlang-1.7

Poster

https://aclanthology.org/2024.signlang-1.8.bib

https://aclanthology.org/2024.signlang-1.8/

@inproceedings{filhol-von-ascheberg-2024-software, title = "A software editor for the {AZVD} graphical Sign Language representation system", author = "Filhol, Michael and von Ascheberg, Thomas", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.8", pages = "77--85", }

No abstract found

A software editor for the AZVD graphical Sign Language representation system

signlang-1.8

Poster

https://aclanthology.org/2024.signlang-1.9.bib

https://aclanthology.org/2024.signlang-1.9/

@inproceedings{gavrilescu-etal-2024-content, title = "Content Questions in Sign Language {--} From theory to language description via corpus, experiments, and fieldwork", author = "Gavrilescu, Robert and Geraci, Carlo and Mesch, Johanna", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.9", pages = "86--94", }

No abstract found

Content Questions in Sign Language – From theory to language description via corpus, experiments, and fieldwork

signlang-1.9

Poster

https://aclanthology.org/2024.signlang-1.10.bib

https://aclanthology.org/2024.signlang-1.10/

@inproceedings{halbout-etal-2024-matignon, title = "Matignon-{LSF}: a Large Corpus of Interpreted {F}rench {S}ign {L}anguage", author = "Halbout, Julie and Fabre, Diandra and Ouakrim, Yanis and Lascar, Julie and Braffort, Annelies and Gouiff{\`e}s, Mich{\`e}le and Beautemps, Denis", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.10", pages = "95--101", }

No abstract found

Matignon-LSF: a Large Corpus of Interpreted French Sign Language

signlang-1.10

Poster

https://aclanthology.org/2024.signlang-1.11.bib

https://aclanthology.org/2024.signlang-1.11/

@inproceedings{hall-etal-2024-phonological, title = "Phonological Transcription of the {C}anadian Dictionary of {ASL} as a Language Resource", author = "Hall, Kathleen Currie and Asthana, Anushka and Reid, Maggie and Gao, Yiran and Hobby, Grace and Tkachman, Oksana and Vesik, Kaili", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.11", pages = "102--110", }

No abstract found

Phonological Transcription of the Canadian Dictionary of ASL as a Language Resource

signlang-1.11

Poster

https://aclanthology.org/2024.signlang-1.12.bib

https://aclanthology.org/2024.signlang-1.12/

@inproceedings{imashev-etal-2024-retrospective, title = "Retrospective of {K}azakh-{R}ussian {S}ign {L}anguage Corpus Formation", author = "Imashev, Alfarabi and Kydyrbekova, Aigerim and Mukushev, Medet and Sandygulova, Anara and Islam, Shynggys and Israilov, Khassan and Makazhanov, Aibek and Yessenbayev, Zhandos", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.12", pages = "111--122", }

No abstract found

Retrospective of Kazakh-Russian Sign Language Corpus Formation

signlang-1.12

Poster

https://aclanthology.org/2024.signlang-1.13.bib

https://aclanthology.org/2024.signlang-1.13/

@inproceedings{inoue-etal-2024-enhancing, title = "Enhancing Syllabic Component Classification in {J}apanese {S}ign {L}anguage by Pre-training on Non-{J}apanese {S}ign {L}anguage Data", author = "Inoue, Jundai and Miwa, Makoto and Sasaki, Yutaka and Hara, Daisuke", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.13", pages = "123--130", }

No abstract found

Enhancing Syllabic Component Classification in Japanese Sign Language by Pre-training on Non-Japanese Sign Language Data

signlang-1.13

Poster

https://aclanthology.org/2024.signlang-1.14.bib

https://aclanthology.org/2024.signlang-1.14/

@inproceedings{isard-2024-building, title = "Building Your Query Step by Step: A Query Wizard for the {MY} {DGS} {--} {ANNIS} Portal of the {DGS} {C}orpus", author = "Isard, Amy", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.14", pages = "131--139", }

No abstract found

Building Your Query Step by Step: A Query Wizard for the MY DGS – ANNIS Portal of the DGS Corpus

signlang-1.14

Poster

https://aclanthology.org/2024.signlang-1.15.bib

https://aclanthology.org/2024.signlang-1.15/

@inproceedings{khan-etal-2024-investigating, title = "Investigating Motion History Images and Convolutional Neural Networks for Isolated {I}rish {S}ign {L}anguage Fingerspelling Recognition", author = "Khan, Sarmad and Murtagh, Irene and McLoughlin, Simon D.", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.15", pages = "140--146", }

No abstract found

Investigating Motion History Images and Convolutional Neural Networks for Isolated Irish Sign Language Fingerspelling Recognition

signlang-1.15

Poster

https://aclanthology.org/2024.signlang-1.16.bib

https://aclanthology.org/2024.signlang-1.16/

@inproceedings{kim-etal-2024-shedding, title = "Shedding Light on the Underexplored: Tackling the Minor Sign Language Research Topics", author = "Kim, Jung-Ho and Ko, Changyong and Huerta-Enochian, Mathew and Ko, Seung Yong", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.16", pages = "147--158", }

No abstract found

Shedding Light on the Underexplored: Tackling the Minor Sign Language Research Topics

signlang-1.16

Poster

https://aclanthology.org/2024.signlang-1.17.bib

https://aclanthology.org/2024.signlang-1.17/

@inproceedings{kimmelman-etal-2024-headshakes, title = "Headshakes in {NGT}: Relation between Phonetic Properties {\&} Linguistic Functions", author = "Kimmelman, Vadim and Oomen, Marloes and Pfau, Roland", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.17", pages = "159--167", }

No abstract found

Headshakes in NGT: Relation between Phonetic Properties & Linguistic Functions

signlang-1.17

Poster

https://aclanthology.org/2024.signlang-1.18.bib

https://aclanthology.org/2024.signlang-1.18/

@inproceedings{kimmelman-etal-2024-nonmanual, title = "Nonmanual Marking of Questions in {B}alinese Homesign Interactions: a Computer-Vision Assisted Analysis", author = "Kimmelman, Vadim and Price, Ari and Safar, Josefina and de Vos, Connie and Bulla, Jan", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.18", pages = "168--177", }

No abstract found

Nonmanual Marking of Questions in Balinese Homesign Interactions: a Computer-Vision Assisted Analysis

signlang-1.18

Poster

https://aclanthology.org/2024.signlang-1.19.bib

https://aclanthology.org/2024.signlang-1.19/

@inproceedings{klomp-etal-2024-extension, title = "An Extension of the {NGT} Dataset in {G}lobal {S}ignbank", author = "Klomp, Ulrika and Gierman, Lisa and Manders, Pieter and Nauta, Ellen and Otterspeer, Gom{\`e}r and Pelupessy, Ray and Stern, Galya and Venter, Dalene and Wubbolts, Casper and Oomen, Marloes and Roelofsen, Floris", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.19", pages = "178--183", }

No abstract found

An Extension of the NGT Dataset in Global Signbank

signlang-1.19

Poster

https://aclanthology.org/2024.signlang-1.20.bib

https://aclanthology.org/2024.signlang-1.20/

@inproceedings{konrad-etal-2024-corpus, title = "Corpus {\`a} la carte {--} Improving Access to the {P}ublic {DGS} {C}orpus", author = {Konrad, Reiner and Hanke, Thomas and Isard, Amy and Schulder, Marc and K{\"o}nig, Lutz and Bleicken, Julian and B{\"o}se, Oliver}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.20", pages = "184--193", }

No abstract found

Corpus à la carte – Improving Access to the Public DGS Corpus

signlang-1.20

Poster

https://aclanthology.org/2024.signlang-1.21.bib

https://aclanthology.org/2024.signlang-1.21/

@inproceedings{langer-etal-2024-introducing, title = "Introducing the {DW}-{DGS} {--} The Digital Dictionary of {DGS}", author = {Langer, Gabriele and M{\"u}ller, Anke and W{\"a}hl, Sabrina and Otte, Felicitas and Sepke, Lea and Hanke, Thomas}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.21", pages = "194--203", }

No abstract found

Introducing the DW-DGS – The Digital Dictionary of DGS

signlang-1.21

Poster

https://aclanthology.org/2024.signlang-1.22.bib

https://aclanthology.org/2024.signlang-1.22/

@inproceedings{lascar-etal-2024-annotation, title = "Annotation of {LSF} subtitled videos without a pre-existing dictionary", author = "Lascar, Julie and Gouiff{\`e}s, Mich{\`e}le and Braffort, Annelies and Danet, Claire", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.22", pages = "204--212", }

No abstract found

Annotation of LSF subtitled videos without a pre-existing dictionary

signlang-1.22

Poster

https://aclanthology.org/2024.signlang-1.23.bib

https://aclanthology.org/2024.signlang-1.23/

@inproceedings{malaia-etal-2024-capturing, title = "Capturing Motion: Using Radar to Build Better Sign Language Corpora", author = "Malaia, Evie and Borneman, Joshua and Gurbuz, Sevgi", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.23", pages = "213--218", }

No abstract found

Capturing Motion: Using Radar to Build Better Sign Language Corpora

signlang-1.23

Poster

https://aclanthology.org/2024.signlang-1.24.bib

https://aclanthology.org/2024.signlang-1.24/

@inproceedings{malmberg-etal-2024-exploring, title = "Exploring Latent Sign Language Representations with Isolated Signs, Sentences and In-the-Wild Data", author = "Malmberg, Fredrik and Klezovich, Anna and Mesch, Johanna and Beskow, Jonas", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.24", pages = "219--224", }

No abstract found

Exploring Latent Sign Language Representations with Isolated Signs, Sentences and In-the-Wild Data

signlang-1.24

Poster

https://aclanthology.org/2024.signlang-1.25.bib

https://aclanthology.org/2024.signlang-1.25/

@inproceedings{martinez-guevara-curiel-2024-quantitative, title = "Quantitative Analysis of Hand Locations in both Sign Language and Non-linguistic Gesture Videos", author = "Mart{\'\i}nez-Guevara, Niels and Curiel, Arturo", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.25", pages = "225--234", }

No abstract found

Quantitative Analysis of Hand Locations in both Sign Language and Non-linguistic Gesture Videos

signlang-1.25

Poster

https://aclanthology.org/2024.signlang-1.26.bib

https://aclanthology.org/2024.signlang-1.26/

@inproceedings{martinod-filhol-2024-formal, title = "Formal Representation of Interrogation in {F}rench {S}ign {L}anguage", author = "Martinod, Emmanuella and Filhol, Michael", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.26", pages = "235--243", }

No abstract found

Formal Representation of Interrogation in French Sign Language

signlang-1.26

Poster

https://aclanthology.org/2024.signlang-1.27.bib

https://aclanthology.org/2024.signlang-1.27/

@inproceedings{mcdonald-etal-2024-multilingual, title = "Multilingual Synthesis of Depictions through Structured Descriptions of Sign: An Initial Case Study", author = "McDonald, John and Efthimiou, Eleni and Fotinea, Stavroula-Evita and Wolfe, Rosalee", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.27", pages = "244--253", }

No abstract found

Multilingual Synthesis of Depictions through Structured Descriptions of Sign: An Initial Case Study

signlang-1.27

Poster

https://aclanthology.org/2024.signlang-1.28.bib

https://aclanthology.org/2024.signlang-1.28/

@inproceedings{mesch-etal-2024-swedish, title = "{S}wedish {S}ign {L}anguage Resources from a User{'}s Perspective", author = {Mesch, Johanna and Bj{\"o}rkstrand, Thomas and Balkstam, Eira and Hansson, Patrick and Riemer Kankkonen, Nikolaus}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.28", pages = "254--261", }

No abstract found

Swedish Sign Language Resources from a User's Perspective

signlang-1.28

Poster

https://aclanthology.org/2024.signlang-1.29.bib

https://aclanthology.org/2024.signlang-1.29/

@inproceedings{miyazaki-etal-2024-sign, title = "Sign Language Translation with Gloss Pair Encoding", author = "Miyazaki, Taro and Tan, Sihan and Uchida, Tsubasa and Kaneko, Hiroyuki", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.29", pages = "262--268", }

No abstract found

Sign Language Translation with Gloss Pair Encoding

signlang-1.29

Poster

https://aclanthology.org/2024.signlang-1.30.bib

https://aclanthology.org/2024.signlang-1.30/

@inproceedings{otterspeer-etal-2024-signcollect, title = "{S}ign{C}ollect: A {`}Touchless{'} Pipeline for Constructing Large-scale Sign Language Repositories", author = "Otterspeer, Gom{\`e}r and Klomp, Ulrika and Roelofsen, Floris", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.30", pages = "269--275", }

No abstract found

SignCollect: A `Touchless' Pipeline for Constructing Large-scale Sign Language Repositories

signlang-1.30

Poster

https://aclanthology.org/2024.signlang-1.31.bib

https://aclanthology.org/2024.signlang-1.31/

@inproceedings{picron-etal-2024-easier, title = "The {EASIER} Mobile Application and Avatar End-User Evaluation Methodology", author = "Picron, Frankie and Van Landuyt, Davy and Omardeen, Rehana and Efthimiou, Eleni and Wolfe, Rosalee and Fotinea, Stavroula-Evita and Goulas, Theodore and Tismer, Christian and Kopf, Maria and Hanke, Thomas", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.31", pages = "276--281", }

No abstract found

The EASIER Mobile Application and Avatar End-User Evaluation Methodology

signlang-1.31

Poster

https://aclanthology.org/2024.signlang-1.32.bib

https://aclanthology.org/2024.signlang-1.32/

@inproceedings{rathmann-etal-2024-visuolab, title = "{V}isuo{L}ab: Building a sign language multilingual, multimodal and multifunctional platform", author = "Rathmann, Christian and de Quadros, Ronice Muller and Gei{\ss}ler, Thomas and Peters, Christian and Fernandes, Francisco and Loio, Milene Peixer and Fran{\c{c}}a, Diego", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.32", pages = "282--289", }

No abstract found

VisuoLab: Building a sign language multilingual, multimodal and multifunctional platform

signlang-1.32

Poster

https://aclanthology.org/2024.signlang-1.33.bib

https://aclanthology.org/2024.signlang-1.33/

@inproceedings{ranum-etal-2024-3d, title = "3{D}-{LEX} v1.0 {--} 3{D} Lexicons for {A}merican {S}ign {L}anguage and {S}ign {L}anguage of the {N}etherlands", author = "Ranum, Oline and Otterspeer, Gom{\`e}r and Andersen, Jari I. and Belleman, Robert G. and Roelofsen, Floris", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.33", pages = "290--301", }

No abstract found

3D-LEX v1.0 – 3D Lexicons for American Sign Language and Sign Language of the Netherlands

signlang-1.33

Poster

https://aclanthology.org/2024.signlang-1.34.bib

https://aclanthology.org/2024.signlang-1.34/

@inproceedings{de-quadros-etal-2024-signbank, title = "{S}ignbank 2.0 of Sign Languages: Easy to Administer, Easy to Use, Easy to Share", author = "de Quadros, Ronice Muller and Rathmann, Christian and Romanek, Peter Zal{\'a}n and Fernandes, Francisco and Cond{\'e}, Sther", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.34", pages = "302--314", }

No abstract found

Signbank 2.0 of Sign Languages: Easy to Administer, Easy to Use, Easy to Share

signlang-1.34

Poster

https://aclanthology.org/2024.signlang-1.35.bib

https://aclanthology.org/2024.signlang-1.35/

@inproceedings{reverdy-etal-2024-stk, title = "{STK} {LSF}: A Motion Capture Dataset in {LSF} for {S}ign{T}o{K}ids", author = "Reverdy, Cl{\'e}ment and Gibet, Sylvie and Le Naour, Thibaut", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.35", pages = "315--322", }

No abstract found

STK LSF: A Motion Capture Dataset in LSF for SignToKids

signlang-1.35

Poster

https://aclanthology.org/2024.signlang-1.36.bib

https://aclanthology.org/2024.signlang-1.36/

@inproceedings{roh-etal-2024-preprocessing, title = "Preprocessing Mediapipe Keypoints with Keypoint Reconstruction and Anchors for Isolated Sign Language Recognition", author = "Roh, Kyunggeun and Lee, Huije and Hwang, Eui Jun and Cho, Sukmin and Park, Jong C.", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.36", pages = "323--334", }

No abstract found

Preprocessing Mediapipe Keypoints with Keypoint Reconstruction and Anchors for Isolated Sign Language Recognition

signlang-1.36

Poster

https://aclanthology.org/2024.signlang-1.37.bib

https://aclanthology.org/2024.signlang-1.37/

@inproceedings{sahin-gokgoz-2024-decoding, title = "Decoding Sign Languages: The {SL}-{FE} Framework for Phonological Analysis and Automated Annotation", author = {{\c{S}}ahin, Karahan and G{\"o}kg{\"o}z, Kadir}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.37", pages = "335--342", }

No abstract found

Decoding Sign Languages: The SL-FE Framework for Phonological Analysis and Automated Annotation

signlang-1.37

Poster

https://aclanthology.org/2024.signlang-1.38.bib

https://aclanthology.org/2024.signlang-1.38/

@inproceedings{schulder-etal-2024-signs, title = "Signs and Synonymity: Continuing Development of the Multilingual Sign Language {W}ordnet", author = {Schulder, Marc and Bigeard, Sam and Kopf, Maria and Hanke, Thomas and Kuder, Anna and W{\'o}jcicka, Joanna and Mesch, Johanna and Bj{\"o}rkstrand, Thomas and Vacalopoulou, Anna and Vasilaki, Kyriaki and Goulas, Theodore and Fotinea, Stavroula-Evita and Efthimiou, Eleni}, editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.38", pages = "343--353", }

No abstract found

Signs and Synonymity: Continuing Development of the Multilingual Sign Language Wordnet

signlang-1.38

Poster

https://aclanthology.org/2024.signlang-1.39.bib

https://aclanthology.org/2024.signlang-1.39/

@inproceedings{sharma-etal-2024-facial, title = "Facial Expressions for Sign Language Synthesis using {FACSH}uman and {AZ}ee", author = "Sharma, Paritosh and Challant, Camille and Filhol, Michael", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.39", pages = "354--360", }

No abstract found

Facial Expressions for Sign Language Synthesis using FACSHuman and AZee

signlang-1.39

Poster

https://aclanthology.org/2024.signlang-1.40.bib

https://aclanthology.org/2024.signlang-1.40/

@inproceedings{susman-kimmelman-2024-eye, title = "Eye Blink Detection in Sign Language Data Using {CNN}s and Rule-Based Methods", author = "Susman, Margaux and Kimmelman, Vadim", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.40", pages = "361--369", }

No abstract found

Eye Blink Detection in Sign Language Data Using CNNs and Rule-Based Methods

signlang-1.40

Poster

https://aclanthology.org/2024.signlang-1.41.bib

https://aclanthology.org/2024.signlang-1.41/

@inproceedings{tan-etal-2024-seda, title = "{SEDA}: Simple and Effective Data Augmentation for Sign Language Understanding", author = "Tan, Sihan and Miyazaki, Taro and Itoyama, Katsutoshi and Nakadai, Kazuhiro", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.41", pages = "370--375", }

No abstract found

SEDA: Simple and Effective Data Augmentation for Sign Language Understanding

signlang-1.41

Poster

https://aclanthology.org/2024.signlang-1.42.bib

https://aclanthology.org/2024.signlang-1.42/

@inproceedings{uchida-etal-2024-hamnosys, title = "{H}am{N}o{S}ys-based Motion Editing Method for Sign Language", author = "Uchida, Tsubasa and Miyazaki, Taro and Kaneko, Hiroyuki", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.42", pages = "376--385", }

No abstract found

HamNoSys-based Motion Editing Method for Sign Language

signlang-1.42

Poster

https://aclanthology.org/2024.signlang-1.43.bib

https://aclanthology.org/2024.signlang-1.43/

@inproceedings{vazquez-enriquez-etal-2024-signamed, title = "{S}igna{M}ed: a Cooperative Bilingual {LSE}-{S}panish Dictionary in the Healthcare Domain", author = "V{\'a}zquez-Enr{\'\i}quez, Manuel and Alba-Castro, Jos{\'e} Luis and P{\'e}rez-P{\'e}rez, Ania and Cabeza-Pereiro, Carmen and Doc{\'\i}o-Fern{\'a}ndez, Laura", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.43", pages = "386--394", }

No abstract found

SignaMed: a Cooperative Bilingual LSE-Spanish Dictionary in the Healthcare Domain

signlang-1.43

Poster

https://aclanthology.org/2024.signlang-1.44.bib

https://aclanthology.org/2024.signlang-1.44/

@inproceedings{xia-etal-2024-diffusion, title = "Diffusion Models for Sign Language Video Anonymization", author = "Xia, Zhaoyang and Zhou, Yang and Han, Ligong and Neidle, Carol and Metaxas, Dimitris N.", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.44", pages = "395--407", }

No abstract found

Diffusion Models for Sign Language Video Anonymization

signlang-1.44

Poster

https://aclanthology.org/2024.signlang-1.45.bib

https://aclanthology.org/2024.signlang-1.45/

@inproceedings{zhou-etal-2024-multimodal, title = "A Multimodal Spatio-Temporal {GCN} Model with Enhancements for Isolated Sign Recognition", author = "Zhou, Yang and Xia, Zhaoyang and Chen, Yuxiao and Neidle, Carol and Metaxas, Dimitris N.", editor = "Efthimiou, Eleni and Fotinea, Stavroula-Evita and Hanke, Thomas and Hochgesang, Julie A. and Mesch, Johanna and Schulder, Marc", booktitle = "Proceedings of the LREC-COLING 2024 11th Workshop on the Representation and Processing of Sign Languages: Evaluation of Sign Language Resources", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.signlang-1.45", pages = "408--419", }

No abstract found

A Multimodal Spatio-Temporal GCN Model with Enhancements for Isolated Sign Recognition

signlang-1.45

Poster

https://aclanthology.org/2024.sigul-1.1.bib

https://aclanthology.org/2024.sigul-1.1/

@inproceedings{arnett-etal-2024-bit, title = "A Bit of a Problem: Measurement Disparities in Dataset Sizes across Languages", author = "Arnett, Catherine and Chang, Tyler A. and Bergen, Benjamin", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.1", pages = "1--9", abstract = "How should text dataset sizes be compared across languages? Even for content-matched (parallel) corpora, UTF-8 encoded text can require a dramatically different number of bytes for different languages. In our work, we define the byte premium between two languages as the ratio of bytes used to encode content-matched text in those languages. We compute byte premiums for 1155 languages, and we use linear regressions to estimate byte premiums for other languages. We release a tool to obtain byte premiums for any two languages, enabling comparisons of dataset sizes across languages for more equitable multilingual model development and data practices.", }

How should text dataset sizes be compared across languages? Even for content-matched (parallel) corpora, UTF-8 encoded text can require a dramatically different number of bytes for different languages. In our work, we define the byte premium between two languages as the ratio of bytes used to encode content-matched text in those languages. We compute byte premiums for 1155 languages, and we use linear regressions to estimate byte premiums for other languages. We release a tool to obtain byte premiums for any two languages, enabling comparisons of dataset sizes across languages for more equitable multilingual model development and data practices.

A Bit of a Problem: Measurement Disparities in Dataset Sizes across Languages

sigul-1.1

Poster

2403.00686

https://aclanthology.org/2024.sigul-1.2.bib

https://aclanthology.org/2024.sigul-1.2/

@inproceedings{mut-altin-saggion-2024-novel, title = "A Novel Corpus for Automated Sexism Identification on Social Media", author = "Mut Altin, Lutfiye Seda and Saggion, Horacio", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.2", pages = "10--15", abstract = "In this paper, we present a novel dataset for the study of automated sexism identification and categorization on social media in Turkish. For this purpose, we have collected, following a well established methodology, a set of Tweets and YouTube comments. Relying on expert organizations in the area of gender equality, each text has been annotated based on a two-level labelling schema derived from previous research. Our resulting dataset consists of around 7,000 annotated instances useful for the study of expressions of sexism and misogyny on the Web. To the best of our knowledge, this is the first two-level manually annotated comprehensive Turkish dataset for sexism identification. In order to fuel research in this relevant area, we also present the result of our benchmarking experiments in the area of sexism identification in Turkish.", }

In this paper, we present a novel dataset for the study of automated sexism identification and categorization on social media in Turkish. For this purpose, we have collected, following a well established methodology, a set of Tweets and YouTube comments. Relying on expert organizations in the area of gender equality, each text has been annotated based on a two-level labelling schema derived from previous research. Our resulting dataset consists of around 7,000 annotated instances useful for the study of expressions of sexism and misogyny on the Web. To the best of our knowledge, this is the first two-level manually annotated comprehensive Turkish dataset for sexism identification. In order to fuel research in this relevant area, we also present the result of our benchmarking experiments in the area of sexism identification in Turkish.

A Novel Corpus for Automated Sexism Identification on Social Media

sigul-1.2

Poster

https://aclanthology.org/2024.sigul-1.3.bib

https://aclanthology.org/2024.sigul-1.3/

@inproceedings{santos-etal-2024-advancing, title = "Advancing Generative {AI} for {P}ortuguese with Open Decoder Gerv{\'a}sio {PT}*", author = "Santos, Rodrigo and Silva, Jo{\~a}o Ricardo and Gomes, Lu{\'\i}s and Rodrigues, Jo{\~a}o and Branco, Ant{\'o}nio", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.3", pages = "16--26", abstract = "To advance the neural decoding of Portuguese, in this paper we present a fully open Transformer-based, instruction-tuned decoder model that sets a new state of the art in this respect. To develop this decoder, which we named Gerv{\'a}sio PT*, a strong LLaMA 2 7B model was used as a starting point, and its further improvement through additional training was done over language resources that include new instruction data sets of Portuguese prepared for this purpose, which are also contributed in this paper. All versions of Gerv{\'a}sio are open source and distributed for free under an open license, including for either research or commercial usage, and can be run on consumer-grade hardware, thus seeking to contribute to the advancement of research and innovation in language technology for Portuguese.", }

To advance the neural decoding of Portuguese, in this paper we present a fully open Transformer-based, instruction-tuned decoder model that sets a new state of the art in this respect. To develop this decoder, which we named Gerv{\'a}sio PT*, a strong LLaMA 2 7B model was used as a starting point, and its further improvement through additional training was done over language resources that include new instruction data sets of Portuguese prepared for this purpose, which are also contributed in this paper. All versions of Gerv{\'a}sio are open source and distributed for free under an open license, including for either research or commercial usage, and can be run on consumer-grade hardware, thus seeking to contribute to the advancement of research and innovation in language technology for Portuguese.

Advancing Generative AI for Portuguese with Open Decoder Gervásio PT*

sigul-1.3

Poster

2402.18766

https://aclanthology.org/2024.sigul-1.4.bib

https://aclanthology.org/2024.sigul-1.4/

@inproceedings{levow-2024-assessing, title = "Assessing Pre-Built Speaker Recognition Models for Endangered Language Data", author = "Levow, Gina-Anne", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.4", pages = "27--32", abstract = "Significant research has focused on speaker recognition, determining which speaker is speaking in a segment of audio. However, few experiments have investigated speaker recognition for very low-resource or endangered languages. Furthermore, speaker recognition has the potential to support language documentation and revitalization efforts, making recordings more accessible to researchers and communities. Since endangered language datasets are too small to build competitive speaker representations from scratch, we investigate the application of large-scale pre-built speaker recognition models to bridge this gap. This paper compares four speaker recognition models on six diverse endangered language data sets. Comparisons contrast three recent neural network-based x-vector models and an earlier baseline i-vector model. Experiments demonstrate significantly stronger performance for some of the studied models. Further analysis highlights differences in effectiveness tied to the lengths of test audio segments and amount of data used for speaker modeling.", }

Significant research has focused on speaker recognition, determining which speaker is speaking in a segment of audio. However, few experiments have investigated speaker recognition for very low-resource or endangered languages. Furthermore, speaker recognition has the potential to support language documentation and revitalization efforts, making recordings more accessible to researchers and communities. Since endangered language datasets are too small to build competitive speaker representations from scratch, we investigate the application of large-scale pre-built speaker recognition models to bridge this gap. This paper compares four speaker recognition models on six diverse endangered language data sets. Comparisons contrast three recent neural network-based x-vector models and an earlier baseline i-vector model. Experiments demonstrate significantly stronger performance for some of the studied models. Further analysis highlights differences in effectiveness tied to the lengths of test audio segments and amount of data used for speaker modeling.

Assessing Pre-Built Speaker Recognition Models for Endangered Language Data

sigul-1.4

Poster

https://aclanthology.org/2024.sigul-1.5.bib

https://aclanthology.org/2024.sigul-1.5/

@inproceedings{kuriyozov-etal-2024-bertbek, title = "{BERT}bek: A Pretrained Language Model for {U}zbek", author = "Kuriyozov, Elmurod and Vilares, David and G{\'o}mez-Rodr{\'\i}guez, Carlos", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.5", pages = "33--44", abstract = "Recent advances in neural networks based language representation made it possible for pretrained language models to outperform previous models in many downstream natural language processing (NLP) tasks. These pretrained language models have also shown that if large enough, they exhibit good few-shot abilities, which is especially beneficial for low-resource scenarios. In this respect, although there are some large-scale multilingual pretrained language models available, language-specific pretrained models have demonstrated to be more accurate for monolingual evaluation setups. In this work, we present BERTbek - pretrained language models based on the BERT (Bidirectional Encoder Representations from Transformers) architecture for the low-resource Uzbek language. We also provide a comprehensive evaluation of the models on a number of NLP tasks: sentiment analysis, multi-label topic classification, and named entity recognition, comparing the models with various machine learning methods as well as multilingual BERT (mBERT). Experimental results indicate that our models outperform mBERT and other task-specific baseline models in all three tasks. Additionally, we also show the impact of training data size and quality on the downstream performance of BERT models, by training three different models with different text sources and corpus sizes.", }

Recent advances in neural networks based language representation made it possible for pretrained language models to outperform previous models in many downstream natural language processing (NLP) tasks. These pretrained language models have also shown that if large enough, they exhibit good few-shot abilities, which is especially beneficial for low-resource scenarios. In this respect, although there are some large-scale multilingual pretrained language models available, language-specific pretrained models have demonstrated to be more accurate for monolingual evaluation setups. In this work, we present BERTbek - pretrained language models based on the BERT (Bidirectional Encoder Representations from Transformers) architecture for the low-resource Uzbek language. We also provide a comprehensive evaluation of the models on a number of NLP tasks: sentiment analysis, multi-label topic classification, and named entity recognition, comparing the models with various machine learning methods as well as multilingual BERT (mBERT). Experimental results indicate that our models outperform mBERT and other task-specific baseline models in all three tasks. Additionally, we also show the impact of training data size and quality on the downstream performance of BERT models, by training three different models with different text sources and corpus sizes.

BERTbek: A Pretrained Language Model for Uzbek

sigul-1.5

Poster

https://aclanthology.org/2024.sigul-1.6.bib

https://aclanthology.org/2024.sigul-1.6/

@inproceedings{arnardottir-etal-2024-beyond, title = "Beyond Error Categories: A Contextual Approach of Evaluating Emerging Spell and Grammar Checkers", author = "Arnard{\'o}ttir, {\TH}{\'o}runn and Ing{\'o}lfsd{\'o}ttir, Svanhv{\'\i}t Lilja and S{\'\i}monarson, Haukur Barri and Einarsson, Hafsteinn and Ingason, Anton Karl and {\TH}orsteinsson, Vilhj{\'a}lmur", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.6", pages = "45--52", abstract = "Automatic spell and grammar checking can be done using various system architectures, and large language models have recently been used to solve the task with promising results. Here we describe a new method of creating test data to measure the performance of spell and grammar checkers, including large language models. Three types of test data represent different approaches to evaluation, from basic error detection to error correction with natural language explanations of the corrections made and error severity scores, which is the main novelty of this approach. These additions are especially useful when evaluating large language models. We present a spell and grammar checking test set for Icelandic in which the described approach is applied. The data consists of whole texts instead of discrete sentences, which facilitates evaluating context awareness of models. The resulting test set can be used to compare different spell and grammar checkers and is published under permissive licenses.", }

Automatic spell and grammar checking can be done using various system architectures, and large language models have recently been used to solve the task with promising results. Here we describe a new method of creating test data to measure the performance of spell and grammar checkers, including large language models. Three types of test data represent different approaches to evaluation, from basic error detection to error correction with natural language explanations of the corrections made and error severity scores, which is the main novelty of this approach. These additions are especially useful when evaluating large language models. We present a spell and grammar checking test set for Icelandic in which the described approach is applied. The data consists of whole texts instead of discrete sentences, which facilitates evaluating context awareness of models. The resulting test set can be used to compare different spell and grammar checkers and is published under permissive licenses.

Beyond Error Categories: A Contextual Approach of Evaluating Emerging Spell and Grammar Checkers

sigul-1.6

Poster

https://aclanthology.org/2024.sigul-1.7.bib

https://aclanthology.org/2024.sigul-1.7/

@inproceedings{poudel-etal-2024-bidirectional, title = "Bidirectional {E}nglish-{N}epali Machine Translation({MT}) System for Legal Domain", author = "Poudel, Shabdapurush and Bal, Bal Krishna and Acharya, Praveen", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.7", pages = "53--58", abstract = "Nepali, a low-resource language belonging to the Indo-Aryan language family and spoken in Nepal, India, Sikkim, and Burma has comparatively very little digital content and resources, more particularly in the legal domain. However, the need to translate legal documents is ever-increasing in the context of growing volumes of legal cases and a large population seeking to go abroad for higher education or employment. This underscores the need for developing an English-Nepali Machine Translation for the legal domain. We attempt to address this problem by utilizing a Neural Machine Translation (NMT) System with an encoder-decoder architecture, specifically designed for legal Nepali-English translation. Leveraging a custom-built legal corpus of 125,000 parallel sentences, our system achieves encouraging BLEU scores of 7.98 in (Nepali → English) and 6.63 (English → Nepali) direction", }

Nepali, a low-resource language belonging to the Indo-Aryan language family and spoken in Nepal, India, Sikkim, and Burma has comparatively very little digital content and resources, more particularly in the legal domain. However, the need to translate legal documents is ever-increasing in the context of growing volumes of legal cases and a large population seeking to go abroad for higher education or employment. This underscores the need for developing an English-Nepali Machine Translation for the legal domain. We attempt to address this problem by utilizing a Neural Machine Translation (NMT) System with an encoder-decoder architecture, specifically designed for legal Nepali-English translation. Leveraging a custom-built legal corpus of 125,000 parallel sentences, our system achieves encouraging BLEU scores of 7.98 in (Nepali → English) and 6.63 (English → Nepali) direction

Bidirectional English-Nepali Machine Translation(MT) System for Legal Domain

sigul-1.7

Poster

https://aclanthology.org/2024.sigul-1.8.bib

https://aclanthology.org/2024.sigul-1.8/

@inproceedings{gonzales-etal-2024-bk3at, title = "{BK}3{AT}: Bangsamoro K-3 Children{'}s Speech Corpus for Developing Assessment Tools in the Bangsamoro Languages", author = "Gonzales, Kiel D. and Maranan, Jazzmin R. and Santelices, Francis Paolo D. and Renovalles, Edsel Jedd M. and Macale, Nissan D. and Palafox, Nicole Anne A. and Mendoza, Jose Marie A.", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.8", pages = "59--65", abstract = "Bangsamoro languages are among the under-resourced languages in the Mindanao region in the Philippines. Moreover, there is no currently publicly available data for children{'}s speech on most of these languages. BK3AT children{'}s speech corpus is a corpus designed for creating speech technologies that could help facilitators and teachers in K-3 education. The corpus consists of 122 hours of children speech data across 10 languages: Bahasa Sug, Chavacano, English, Filipino, Iranun, Maguindanaon, Meranaw, Sinama, Teduray, and Yakan. Preliminary experiments using Wav2Vec-XLSR architecture have been done in fine-tuning the Tagalog and L2 English corpus subsets to develop automatic speech recognition backend for literacy assessment. Results from the experiments show low word error rates (WERs) for small-vocabulary and targeted domains.", }

Bangsamoro languages are among the under-resourced languages in the Mindanao region in the Philippines. Moreover, there is no currently publicly available data for children{'}s speech on most of these languages. BK3AT children{'}s speech corpus is a corpus designed for creating speech technologies that could help facilitators and teachers in K-3 education. The corpus consists of 122 hours of children speech data across 10 languages: Bahasa Sug, Chavacano, English, Filipino, Iranun, Maguindanaon, Meranaw, Sinama, Teduray, and Yakan. Preliminary experiments using Wav2Vec-XLSR architecture have been done in fine-tuning the Tagalog and L2 English corpus subsets to develop automatic speech recognition backend for literacy assessment. Results from the experiments show low word error rates (WERs) for small-vocabulary and targeted domains.

BK3AT: Bangsamoro K-3 Children's Speech Corpus for Developing Assessment Tools in the Bangsamoro Languages

sigul-1.8

Poster

https://aclanthology.org/2024.sigul-1.9.bib

https://aclanthology.org/2024.sigul-1.9/

@inproceedings{poujade-etal-2024-corpusarieja, title = "{C}orpus{A}ri{\`e}ja: Building an Annotated Corpus with Variation in {O}ccitan", author = "Poujade, Clamenca and Bras, Myriam and Urieli, Assaf", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.9", pages = "66--71", abstract = "The Occitan language is a less resourced language and is classified as {`}in danger{'} by the UNESCO. Thereby, it is important to build resources and tools that can help to safeguard and develop the digitisation of the language. CorpusAri{\`e}ja is a collection of 72 texts (just over 41,000 tokens) in the Occitan language of the French department of Ari{\`e}ge. The majority of the texts needed to be digitised and pass within an Optical Character Recognition. This corpus contains dialectal and spelling variation, but is limited to prose, without diachronic variation or genre variation. It is an annotated corpus with two levels of lemmatisation, POS tags and verbal inflection. One of the main aims of the corpus is to enable the conception of tools that can automatically annotate all Occitan texts, regardless of the dialect or spelling used. The Ari{\`e}ge territory is interesting because it includes the two variations that we focus on, dialectal and spelling. It has plenty of authors that write in their native language, their variety of Occitan.", }

The Occitan language is a less resourced language and is classified as {`}in danger{'} by the UNESCO. Thereby, it is important to build resources and tools that can help to safeguard and develop the digitisation of the language. CorpusAri{\`e}ja is a collection of 72 texts (just over 41,000 tokens) in the Occitan language of the French department of Ari{\`e}ge. The majority of the texts needed to be digitised and pass within an Optical Character Recognition. This corpus contains dialectal and spelling variation, but is limited to prose, without diachronic variation or genre variation. It is an annotated corpus with two levels of lemmatisation, POS tags and verbal inflection. One of the main aims of the corpus is to enable the conception of tools that can automatically annotate all Occitan texts, regardless of the dialect or spelling used. The Ari{\`e}ge territory is interesting because it includes the two variations that we focus on, dialectal and spelling. It has plenty of authors that write in their native language, their variety of Occitan.

CorpusArièja: Building an Annotated Corpus with Variation in Occitan

sigul-1.9

Poster

https://aclanthology.org/2024.sigul-1.10.bib

https://aclanthology.org/2024.sigul-1.10/

@inproceedings{sekeres-etal-2024-developing, title = "Developing Infrastructure for Low-Resource Language Corpus Building", author = "Sekeres, Hedwig G. and Heeringa, Wilbert and de Vries, Wietse and Zwagers, Oscar Yde and Wieling, Martijn and Jensma, Goffe Th.", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.10", pages = "72--78", abstract = "For many of the world{'}s small languages, few resources are available. In this project, a written online accessible corpus was created for the minority language variant Gronings, which serves both researchers interested in language change and variation and a general audience of (new) speakers interested in finding real-life examples of language use. The corpus was created using a combination of volunteer work and automation, which together formed an efficient pipeline for converting printed text to Key Words in Context (KWICs), annotated with lemmas and part-of-speech tags. In the creation of the corpus, we have taken into account several of the challenges that can occur when creating resources for minority languages, such as a lack of standardisation and limited (financial) resources. As the solutions we offer are applicable to other small languages as well, each step of the corpus creation process is discussed and resources will be made available benefiting future projects on other low-resource languages.", }

For many of the world{'}s small languages, few resources are available. In this project, a written online accessible corpus was created for the minority language variant Gronings, which serves both researchers interested in language change and variation and a general audience of (new) speakers interested in finding real-life examples of language use. The corpus was created using a combination of volunteer work and automation, which together formed an efficient pipeline for converting printed text to Key Words in Context (KWICs), annotated with lemmas and part-of-speech tags. In the creation of the corpus, we have taken into account several of the challenges that can occur when creating resources for minority languages, such as a lack of standardisation and limited (financial) resources. As the solutions we offer are applicable to other small languages as well, each step of the corpus creation process is discussed and resources will be made available benefiting future projects on other low-resource languages.

Developing Infrastructure for Low-Resource Language Corpus Building

sigul-1.10

Poster

https://aclanthology.org/2024.sigul-1.11.bib

https://aclanthology.org/2024.sigul-1.11/

@inproceedings{johannsson-etal-2024-evaluating, title = "Evaluating {I}celandic Sentiment Analysis Models Trained on Translated Data", author = {J{\'o}hannsson, {\'O}lafur A. and Arndal, Birkir H. and J{\'o}nsson, Eysteinn {\"O}. and Olafsson, Stefan and Loftsson, Hrafn}, editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.11", pages = "79--89", abstract = "We experiment with sentiment classification models for Icelandic that leverage machine-translated data for training. Since no large sentiment dataset exists for Icelandic, we translate 50,000 English IMDb reviews, classified either as positive or negative, into Icelandic using two services: Google Translate and GreynirTranslate. After machine translation, we assess whether the sentiment of the source language text is retained in the target language. Moreover, we evaluate the accuracy of the sentiment classifiers on non-translated Icelandic text.The performance of three types of baseline classifiers is compared, i.e., Support Vector Machines, Logistic Regression and Naive Bayes, when trained on translated data generated by either translation service. Furthermore, we fine-tune and evaluate three pre-trained transformer-based models, RoBERTa, IceBERT and ELECTRA, on both the original English texts and the translated texts. Our results indicate that the transformer models perform better than the baseline classifiers on all datasets. Moreover, our evaluation shows that the transformer models trained on data translated from English reviews can be used to effectively classify sentiment on non-translated Icelandic movie reviews.", }

We experiment with sentiment classification models for Icelandic that leverage machine-translated data for training. Since no large sentiment dataset exists for Icelandic, we translate 50,000 English IMDb reviews, classified either as positive or negative, into Icelandic using two services: Google Translate and GreynirTranslate. After machine translation, we assess whether the sentiment of the source language text is retained in the target language. Moreover, we evaluate the accuracy of the sentiment classifiers on non-translated Icelandic text.The performance of three types of baseline classifiers is compared, i.e., Support Vector Machines, Logistic Regression and Naive Bayes, when trained on translated data generated by either translation service. Furthermore, we fine-tune and evaluate three pre-trained transformer-based models, RoBERTa, IceBERT and ELECTRA, on both the original English texts and the translated texts. Our results indicate that the transformer models perform better than the baseline classifiers on all datasets. Moreover, our evaluation shows that the transformer models trained on data translated from English reviews can be used to effectively classify sentiment on non-translated Icelandic movie reviews.

Evaluating Icelandic Sentiment Analysis Models Trained on Translated Data

sigul-1.11

Poster

https://aclanthology.org/2024.sigul-1.12.bib

https://aclanthology.org/2024.sigul-1.12/

@inproceedings{mc-cahill-etal-2024-exploring, title = "Exploring Text Classification for Enhancing Digital Game-Based Language Learning for {I}rish", author = "Mc Cahill, Leona and Baltazar, Thomas and Bruen, Sally and Xu, Liang and Ward, Monica and U{\'\i} Dhonnchadha, Elaine and Foster, Jennifer", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.12", pages = "90--96", abstract = "Digital game-based language learning (DGBLL) can help with the language learning process. DGBLL applications can make learning more enjoyable and engaging, but they are difficult to develop. A DBGLL app that relies on target language texts obviously needs to be able to use texts of the appropriate level for the individual learners. This implies that text classification tools should be available to DGBLL developers, who may not be familiar with the target language, in order to incorporate suitable texts into their games. While text difficulty classifiers exist for many of the most commonly spoken languages, this is not the case for under-resourced languages, such as Irish. In this paper, we explore approaches to the development of text classifiers for Irish. In the first approach to text analysis and grading, we apply linguistic analysis to assess text complexity. Features from this approach are then used in machine learning-based text classification, which explores the application of a number of machine learning algorithms to the problem. Although the development of these text classifiers is at an early stage, they show promise, particularly in a low-resourced scenario.", }

Digital game-based language learning (DGBLL) can help with the language learning process. DGBLL applications can make learning more enjoyable and engaging, but they are difficult to develop. A DBGLL app that relies on target language texts obviously needs to be able to use texts of the appropriate level for the individual learners. This implies that text classification tools should be available to DGBLL developers, who may not be familiar with the target language, in order to incorporate suitable texts into their games. While text difficulty classifiers exist for many of the most commonly spoken languages, this is not the case for under-resourced languages, such as Irish. In this paper, we explore approaches to the development of text classifiers for Irish. In the first approach to text analysis and grading, we apply linguistic analysis to assess text complexity. Features from this approach are then used in machine learning-based text classification, which explores the application of a number of machine learning algorithms to the problem. Although the development of these text classifiers is at an early stage, they show promise, particularly in a low-resourced scenario.

Exploring Text Classification for Enhancing Digital Game-Based Language Learning for Irish

sigul-1.12

Poster

https://aclanthology.org/2024.sigul-1.13.bib

https://aclanthology.org/2024.sigul-1.13/

@inproceedings{philippy-etal-2024-forget, title = "Forget {NLI}, Use a Dictionary: Zero-Shot Topic Classification for Low-Resource Languages with Application to {L}uxembourgish", author = "Philippy, Fred and Haddadan, Shohreh and Guo, Siwen", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.13", pages = "97--104", abstract = "In NLP, zero-shot classification (ZSC) is the task of assigning labels to textual data without any labeled examples for the target classes. A common method for ZSC is to fine-tune a language model on a Natural Language Inference (NLI) dataset and then use it to infer the entailment between the input document and the target labels. However, this approach faces certain challenges, particularly for languages with limited resources. In this paper, we propose an alternative solution that leverages dictionaries as a source of data for ZSC. We focus on Luxembourgish, a low-resource language spoken in Luxembourg, and construct two new topic relevance classification datasets based on a dictionary that provides various synonyms, word translations and example sentences. We evaluate the usability of our dataset and compare it with the NLI-based approach on two topic classification tasks in a zero-shot manner. Our results show that by using the dictionary-based dataset, the trained models outperform the ones following the NLI-based approach for ZSC. While we focus on a single low-resource language in this study, we believe that the efficacy of our approach can also transfer to other languages where such a dictionary is available.", }

In NLP, zero-shot classification (ZSC) is the task of assigning labels to textual data without any labeled examples for the target classes. A common method for ZSC is to fine-tune a language model on a Natural Language Inference (NLI) dataset and then use it to infer the entailment between the input document and the target labels. However, this approach faces certain challenges, particularly for languages with limited resources. In this paper, we propose an alternative solution that leverages dictionaries as a source of data for ZSC. We focus on Luxembourgish, a low-resource language spoken in Luxembourg, and construct two new topic relevance classification datasets based on a dictionary that provides various synonyms, word translations and example sentences. We evaluate the usability of our dataset and compare it with the NLI-based approach on two topic classification tasks in a zero-shot manner. Our results show that by using the dictionary-based dataset, the trained models outperform the ones following the NLI-based approach for ZSC. While we focus on a single low-resource language in this study, we believe that the efficacy of our approach can also transfer to other languages where such a dictionary is available.

Forget NLI, Use a Dictionary: Zero-Shot Topic Classification for Low-Resource Languages with Application to Luxembourgish

sigul-1.13

Poster

2404.03912

https://aclanthology.org/2024.sigul-1.14.bib

https://aclanthology.org/2024.sigul-1.14/

@inproceedings{santos-etal-2024-fostering, title = "Fostering the Ecosystem of Open Neural Encoders for {P}ortuguese with Albertina {PT}* Family", author = "Santos, Rodrigo and Rodrigues, Jo{\~a}o and Gomes, Lu{\'\i}s and Silva, Jo{\~a}o Ricardo and Branco, Ant{\'o}nio and Lopes Cardoso, Henrique and Os{\'o}rio, Tom{\'a}s Freitas and Leite, Bernardo", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.14", pages = "105--114", abstract = "To foster the neural encoding of Portuguese, this paper contributes foundation encoder models that represent an expansion of the still very scarce ecosystem of large language models specifically developed for this language that are fully open, in the sense that they are open source and openly distributed for free under an open license for any purpose, thus including research and commercial usages. Like most languages other than English, Portuguese is low-resourced in terms of these foundational language resources, there being the inaugural 900 million parameter Albertina and 335 million Bertimbau. Taking this couple of models as an inaugural set, we present the extension of the ecosystem of state-of-the-art open encoders for Portuguese with a larger, top performance-driven model with 1.5 billion parameters, and a smaller, efficiency-driven model with 100 million parameters. While achieving this primary goal, further results that are relevant for this ecosystem were obtained as well, namely new datasets for Portuguese based on the SuperGLUE benchmark, which we also distribute openly.", }

To foster the neural encoding of Portuguese, this paper contributes foundation encoder models that represent an expansion of the still very scarce ecosystem of large language models specifically developed for this language that are fully open, in the sense that they are open source and openly distributed for free under an open license for any purpose, thus including research and commercial usages. Like most languages other than English, Portuguese is low-resourced in terms of these foundational language resources, there being the inaugural 900 million parameter Albertina and 335 million Bertimbau. Taking this couple of models as an inaugural set, we present the extension of the ecosystem of state-of-the-art open encoders for Portuguese with a larger, top performance-driven model with 1.5 billion parameters, and a smaller, efficiency-driven model with 100 million parameters. While achieving this primary goal, further results that are relevant for this ecosystem were obtained as well, namely new datasets for Portuguese based on the SuperGLUE benchmark, which we also distribute openly.

Fostering the Ecosystem of Open Neural Encoders for Portuguese with Albertina PT* Family

sigul-1.14

Poster

2403.01897

https://aclanthology.org/2024.sigul-1.15.bib

https://aclanthology.org/2024.sigul-1.15/

@inproceedings{jauhiainen-linden-2024-improving, title = "Improving Language Coverage on {H}e{LI}-{OTS}", author = "Jauhiainen, Tommi and Lind{\'e}n, Krister", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.15", pages = "115--125", abstract = "In this paper, we add under-resourced languages into the language repertoire of an existing off-the-shelf language identifier, HeLI-OTS. Adding more languages to a language identifier often comes with the drawback of lessened accuracy for the languages already part of the repertoire. We aim to minimize this effect. As sources for training and development data in the new languages, we use the OpenLID and FLORES-200 datasets. They are openly available high-quality datasets that are especially well-suited for language identifier development. By carefully inspecting the effect of each added language and the quality of their training and development data, we managed to add support for 20 new under-resourced languages to HeLI-OTS without affecting the performance of any existing languages to a noticeable extent.", }

In this paper, we add under-resourced languages into the language repertoire of an existing off-the-shelf language identifier, HeLI-OTS. Adding more languages to a language identifier often comes with the drawback of lessened accuracy for the languages already part of the repertoire. We aim to minimize this effect. As sources for training and development data in the new languages, we use the OpenLID and FLORES-200 datasets. They are openly available high-quality datasets that are especially well-suited for language identifier development. By carefully inspecting the effect of each added language and the quality of their training and development data, we managed to add support for 20 new under-resourced languages to HeLI-OTS without affecting the performance of any existing languages to a noticeable extent.

Improving Language Coverage on HeLI-OTS

sigul-1.15

Poster

https://aclanthology.org/2024.sigul-1.16.bib

https://aclanthology.org/2024.sigul-1.16/

@inproceedings{masala-etal-2024-improving, title = "Improving Legal Judgement Prediction in {R}omanian with Long Text Encoders", author = "Masala, Mihai and Rebedea, Traian and Velicu, Horia", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.16", pages = "126--132", abstract = "In recent years,the entire field of Natural Language Processing (NLP) has enjoyed amazing novel results achieving almost human-like performance on a variety of tasks. Legal NLP domain has also been part of this process, as it has seen an impressive growth. However, general-purpose models are not readily applicable for legal domain. Due to the nature of the domain (e.g. specialized vocabulary, long documents) specific models and methods are often needed for Legal NLP. In this work we investigate both specialized and general models for predicting the final ruling of a legal case, task known as Legal Judgment Prediction (LJP). We particularly focus on methods to extend to sequence length of Transformer-based models to better understand the long documents present in legal corpora. Extensive experiments on 4 LJP datasets in Romanian, originating from 2 sources with significantly different sizes and document lengths, show that specialized models and handling long texts are critical for a good performance.", }

In recent years,the entire field of Natural Language Processing (NLP) has enjoyed amazing novel results achieving almost human-like performance on a variety of tasks. Legal NLP domain has also been part of this process, as it has seen an impressive growth. However, general-purpose models are not readily applicable for legal domain. Due to the nature of the domain (e.g. specialized vocabulary, long documents) specific models and methods are often needed for Legal NLP. In this work we investigate both specialized and general models for predicting the final ruling of a legal case, task known as Legal Judgment Prediction (LJP). We particularly focus on methods to extend to sequence length of Transformer-based models to better understand the long documents present in legal corpora. Extensive experiments on 4 LJP datasets in Romanian, originating from 2 sources with significantly different sizes and document lengths, show that specialized models and handling long texts are critical for a good performance.

Improving Legal Judgement Prediction in Romanian with Long Text Encoders

sigul-1.16

Poster

2402.19170

https://aclanthology.org/2024.sigul-1.17.bib

https://aclanthology.org/2024.sigul-1.17/

@inproceedings{li-vu-2024-improving, title = "Improving Noisy Student Training for Low-resource Languages in End-to-End {ASR} Using {C}ycle{GAN} and Inter-domain Losses", author = "Li, Chia-Yu and Vu, Ngoc Thang", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.17", pages = "133--142", abstract = "Training a semi-supervised end-to-end speech recognition system using noisy student training has significantly improved performance. However, this approach requires a substantial amount of paired speech-text and unlabeled speech, which is costly for low-resource languages. Therefore, this paper considers a more extreme case of semi-supervised end-to-end automatic speech recognition where there are limited paired speech-text, unlabeled speech (less than five hours), and abundant external text. Firstly, we observe improved performance by training the model using our previous work on semi-supervised learning {``}CycleGAN and inter-domain losses{''} solely with external text. Secondly, we enhance {``}CycleGAN and inter-domain losses{''} by incorporating automatic hyperparameter tuning, calling {``}enhanced CycleGAN inter-domain losses.{''} Thirdly, we integrate it into the noisy student training approach pipeline for low-resource scenarios. Our experimental results, conducted on six non-English languages from Voxforge and Common Voice, show a 20{\%} word error rate reduction compared to the baseline teacher model and a 10{\%} word error rate reduction compared to the baseline best student model, highlighting the significant improvements achieved through our proposed method.", }

Training a semi-supervised end-to-end speech recognition system using noisy student training has significantly improved performance. However, this approach requires a substantial amount of paired speech-text and unlabeled speech, which is costly for low-resource languages. Therefore, this paper considers a more extreme case of semi-supervised end-to-end automatic speech recognition where there are limited paired speech-text, unlabeled speech (less than five hours), and abundant external text. Firstly, we observe improved performance by training the model using our previous work on semi-supervised learning {``}CycleGAN and inter-domain losses{''} solely with external text. Secondly, we enhance {``}CycleGAN and inter-domain losses{''} by incorporating automatic hyperparameter tuning, calling {``}enhanced CycleGAN inter-domain losses.{''} Thirdly, we integrate it into the noisy student training approach pipeline for low-resource scenarios. Our experimental results, conducted on six non-English languages from Voxforge and Common Voice, show a 20{\%} word error rate reduction compared to the baseline teacher model and a 10{\%} word error rate reduction compared to the baseline best student model, highlighting the significant improvements achieved through our proposed method.

Improving Noisy Student Training for Low-resource Languages in End-to-End ASR Using CycleGAN and Inter-domain Losses

sigul-1.17

Poster

2407.21061

https://aclanthology.org/2024.sigul-1.18.bib

https://aclanthology.org/2024.sigul-1.18/

@inproceedings{tazakka-etal-2024-indonesian, title = "{I}ndonesian-{E}nglish Code-Switching Speech Recognition Using the Machine Speech Chain Based Semi-Supervised Learning", author = "Tazakka, Rais Vaza Man and Lestari, Dessi and Purwarianti, Ayu and Tanaya, Dipta and Azizah, Kurniawati and Sakti, Sakriani", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.18", pages = "143--148", abstract = "Indonesia is home to a diverse linguistic landscape, where individuals seamlessly transition between Indonesian, English, and local dialects in their everyday conversations{---}a phenomenon known as code-switching. Understanding and accommodating this linguistic fluidity is essential, particularly in the development of accurate speech recognition systems. However, tackling code-switching in Indonesian poses a challenge due to the scarcity of paired code-switching data. Thus, this study endeavors to address Indonesian-English code-switching in speech recognition, leveraging unlabeled data and employing a semi-supervised technique known as the machine speech chain. Our findings demonstrate that the machine speech chain method effectively enhances Automatic Speech Recognition (ASR) performance in recognizing code-switching between Indonesian and English, utilizing previously untapped resources of unlabeled data.", }

Indonesia is home to a diverse linguistic landscape, where individuals seamlessly transition between Indonesian, English, and local dialects in their everyday conversations{---}a phenomenon known as code-switching. Understanding and accommodating this linguistic fluidity is essential, particularly in the development of accurate speech recognition systems. However, tackling code-switching in Indonesian poses a challenge due to the scarcity of paired code-switching data. Thus, this study endeavors to address Indonesian-English code-switching in speech recognition, leveraging unlabeled data and employing a semi-supervised technique known as the machine speech chain. Our findings demonstrate that the machine speech chain method effectively enhances Automatic Speech Recognition (ASR) performance in recognizing code-switching between Indonesian and English, utilizing previously untapped resources of unlabeled data.

Indonesian-English Code-Switching Speech Recognition Using the Machine Speech Chain Based Semi-Supervised Learning

sigul-1.18

Poster

https://aclanthology.org/2024.sigul-1.19.bib

https://aclanthology.org/2024.sigul-1.19/

@inproceedings{kondo-tamura-2024-inter, title = "Inter-language Transfer Learning for Visual Speech Recognition toward Under-resourced Environments", author = "Kondo, Fumiya and Tamura, Satoshi", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.19", pages = "149--154", abstract = "In this study, we introduce a method of inter-language transfer learning for under-resourced visual speech recognition. Deploying speech-related technology to all languages is a quite important activity. However, applying state-of-the-art deep-learning techniques requires huge-size labeled corpora, which makes it hard for under-resourced languages. Our approach leverages a small amount of labeled video data of the target language, and employs inter-language transfer learning using a pre-trained English lip-reading model. By applying the proposed scheme, we build a Japanese lip-reading model, using the ROHAN corpus, the size of which is about one 450th of the size of English datasets. The front-end encoder part of the pre-trained model is fine-tuned to improve the acquisition of pronunciation and lip movement patterns unique to Japanese. On the other hand, the back-end encoder and the decoder are built using the Japanese dataset. Although English and Japanese have different language structures, evaluation experiments show that it is possible to build the Japanese lip-reading model efficiently. Comparison with competitive schemes demonstrates the effectiveness of our method.", }

In this study, we introduce a method of inter-language transfer learning for under-resourced visual speech recognition. Deploying speech-related technology to all languages is a quite important activity. However, applying state-of-the-art deep-learning techniques requires huge-size labeled corpora, which makes it hard for under-resourced languages. Our approach leverages a small amount of labeled video data of the target language, and employs inter-language transfer learning using a pre-trained English lip-reading model. By applying the proposed scheme, we build a Japanese lip-reading model, using the ROHAN corpus, the size of which is about one 450th of the size of English datasets. The front-end encoder part of the pre-trained model is fine-tuned to improve the acquisition of pronunciation and lip movement patterns unique to Japanese. On the other hand, the back-end encoder and the decoder are built using the Japanese dataset. Although English and Japanese have different language structures, evaluation experiments show that it is possible to build the Japanese lip-reading model efficiently. Comparison with competitive schemes demonstrates the effectiveness of our method.

Inter-language Transfer Learning for Visual Speech Recognition toward Under-resourced Environments

sigul-1.19

Poster

https://aclanthology.org/2024.sigul-1.20.bib

https://aclanthology.org/2024.sigul-1.20/

@inproceedings{her-kruschwitz-2024-investigating, title = "Investigating Neural Machine Translation for Low-Resource Languages: Using {B}avarian as a Case Study", author = "Her, Wan-hua and Kruschwitz, Udo", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.20", pages = "155--167", abstract = "Machine Translation has made impressive progress in recent years offering close to human-level performance on many languages, but studies have primarily focused on high-resource languages with broad online presence and resources. With the help of growing Large Language Models, more and more low-resource languages achieve better results through the presence of other languages. However, studies have shown that not all low-resource languages can benefit from multilingual systems, especially those with insufficient training and evaluation data. In this paper, we revisit state-of-the-art Neural Machine Translation techniques to develop automatic translation systems between German and Bavarian. We investigate conditions of low-resource languages such as data scarcity and parameter sensitivity and focus on refined solutions that combat low-resource difficulties and creative solutions such as harnessing language similarity. Our experiment entails applying Back-translation and Transfer Learning to automatically generate more training data and achieve higher translation performance. We demonstrate noisiness in the data and present our approach to carry out text preprocessing extensively. Evaluation was conducted using combined metrics: BLEU, chrF and TER. Statistical significance results with Bonferroni correction show surprisingly high baseline systems, and that Back-translation leads to significant improvement. Furthermore, we present a qualitative analysis of translation errors and system limitations.", }

Machine Translation has made impressive progress in recent years offering close to human-level performance on many languages, but studies have primarily focused on high-resource languages with broad online presence and resources. With the help of growing Large Language Models, more and more low-resource languages achieve better results through the presence of other languages. However, studies have shown that not all low-resource languages can benefit from multilingual systems, especially those with insufficient training and evaluation data. In this paper, we revisit state-of-the-art Neural Machine Translation techniques to develop automatic translation systems between German and Bavarian. We investigate conditions of low-resource languages such as data scarcity and parameter sensitivity and focus on refined solutions that combat low-resource difficulties and creative solutions such as harnessing language similarity. Our experiment entails applying Back-translation and Transfer Learning to automatically generate more training data and achieve higher translation performance. We demonstrate noisiness in the data and present our approach to carry out text preprocessing extensively. Evaluation was conducted using combined metrics: BLEU, chrF and TER. Statistical significance results with Bonferroni correction show surprisingly high baseline systems, and that Back-translation leads to significant improvement. Furthermore, we present a qualitative analysis of translation errors and system limitations.

Investigating Neural Machine Translation for Low-Resource Languages: Using Bavarian as a Case Study

sigul-1.20

Poster

2404.08259

https://aclanthology.org/2024.sigul-1.21.bib

https://aclanthology.org/2024.sigul-1.21/

@inproceedings{haberland-etal-2024-italian, title = "{I}talian-{L}igurian Machine Translation in Its Cultural Context", author = "Haberland, Christopher R. and Maillard, Jean and Lusito, Stefano", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.21", pages = "168--176", abstract = "Large multilingual machine translation efforts are driving improved access and performance for under-resourced languages, but often fail to translate culturally specific and local concepts. Additionally, translation from practically relevant input languages may flag behind those that are comparatively over-represented in the training dataset. In this work, we release a new corpus, ZenaMT, containing 7,561 parallel Ligurian-Italian sentences, nearly a fifth of which are also translated in English. This corpus spans five domains: local and international news, Ligurian literature, Genoese Ligurian linguistics concepts, traditional card game rules, and Ligurian geographic expressions. We find that a translation model augmented with ZenaMT improves a baseline by 20{\%}, and by over 25{\%} (BLEU) compared to NLLB-3.3B, which is over 50 times the size. Our results demonstrate the utility of creating data sets for MT that are specifically tailored for the cultural context of Ligurian speakers. We freely release ZenaMT and expect to periodically update the corpus to improve MT performance and domain coverage.", }

Large multilingual machine translation efforts are driving improved access and performance for under-resourced languages, but often fail to translate culturally specific and local concepts. Additionally, translation from practically relevant input languages may flag behind those that are comparatively over-represented in the training dataset. In this work, we release a new corpus, ZenaMT, containing 7,561 parallel Ligurian-Italian sentences, nearly a fifth of which are also translated in English. This corpus spans five domains: local and international news, Ligurian literature, Genoese Ligurian linguistics concepts, traditional card game rules, and Ligurian geographic expressions. We find that a translation model augmented with ZenaMT improves a baseline by 20{\%}, and by over 25{\%} (BLEU) compared to NLLB-3.3B, which is over 50 times the size. Our results demonstrate the utility of creating data sets for MT that are specifically tailored for the cultural context of Ligurian speakers. We freely release ZenaMT and expect to periodically update the corpus to improve MT performance and domain coverage.

Italian-Ligurian Machine Translation in Its Cultural Context

sigul-1.21

Poster

https://aclanthology.org/2024.sigul-1.22.bib

https://aclanthology.org/2024.sigul-1.22/

@inproceedings{de-jesus-nunes-2024-labadain, title = "Labadain-30k+: A Monolingual Tetun Document-Level Audited Dataset", author = "de Jesus, Gabriel and Nunes, S{\'e}rgio", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.22", pages = "177--188", abstract = "This paper introduces Labadain-30k+, a monolingual dataset comprising 33.6k documents in Tetun, a low-resource language spoken in Timor-Leste. The dataset was acquired through web crawling and augmented with Wikipedia documents released by Wikimedia. Both sets of documents underwent thorough manual audits at the document level by native Tetun speakers, resulting in the construction of a Tetun text dataset well-suited for a variety of natural language processing and information retrieval tasks. This dataset was employed to conduct a comprehensive content analysis aimed at providing a nuanced understanding of document composition and the evolution of Tetun documents on the web. The analysis revealed that news articles constitute the predominant documents within the dataset, accounting for 89.87{\%} of the total, followed by Wikipedia documents at 4.34{\%}, and legal and governmental documents at 3.65{\%}, among others. Notably, there was a substantial increase in the number of documents in 2020, indicating 11.75 percentage points rise in document quantity, compared to an average of 4.76 percentage points per year from 2001 to 2023. Moreover, the year 2017, marked by the increased popularity of online news in Tetun, served as a threshold for analyzing the evolution of document writing on the web pre- and post-2017, specifically regarding vocabulary usage. Surprisingly, this analysis showed a significant increase of 6.12 percentage points in the Tetun written adhering to the Tetun official standard. Additionally, the persistence of Portuguese loanwords in that trajectory remained evident, reflecting an increase of 5.09 percentage points.", }

This paper introduces Labadain-30k+, a monolingual dataset comprising 33.6k documents in Tetun, a low-resource language spoken in Timor-Leste. The dataset was acquired through web crawling and augmented with Wikipedia documents released by Wikimedia. Both sets of documents underwent thorough manual audits at the document level by native Tetun speakers, resulting in the construction of a Tetun text dataset well-suited for a variety of natural language processing and information retrieval tasks. This dataset was employed to conduct a comprehensive content analysis aimed at providing a nuanced understanding of document composition and the evolution of Tetun documents on the web. The analysis revealed that news articles constitute the predominant documents within the dataset, accounting for 89.87{\%} of the total, followed by Wikipedia documents at 4.34{\%}, and legal and governmental documents at 3.65{\%}, among others. Notably, there was a substantial increase in the number of documents in 2020, indicating 11.75 percentage points rise in document quantity, compared to an average of 4.76 percentage points per year from 2001 to 2023. Moreover, the year 2017, marked by the increased popularity of online news in Tetun, served as a threshold for analyzing the evolution of document writing on the web pre- and post-2017, specifically regarding vocabulary usage. Surprisingly, this analysis showed a significant increase of 6.12 percentage points in the Tetun written adhering to the Tetun official standard. Additionally, the persistence of Portuguese loanwords in that trajectory remained evident, reflecting an increase of 5.09 percentage points.

Labadain-30k+: A Monolingual Tetun Document-Level Audited Dataset

sigul-1.22

Poster

https://aclanthology.org/2024.sigul-1.23.bib

https://aclanthology.org/2024.sigul-1.23/

@inproceedings{ljubesic-etal-2024-language, title = "Language Models on a Diet: Cost-Efficient Development of Encoders for Closely-Related Languages via Additional Pretraining", author = "Ljube{\v{s}}i{\'c}, Nikola and Suchomel, V{\'\i}t and Rupnik, Peter and Kuzman, Taja and van Noord, Rik", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.23", pages = "189--203", abstract = "The world of language models is going through turbulent times, better and ever larger models are coming out at an unprecedented speed. However, we argue that, especially for the scientific community, encoder models of up to 1 billion parameters are still very much needed, their primary usage being in enriching large collections of data with metadata necessary for downstream research. We investigate the best way to ensure the existence of such encoder models on the set of very closely related languages - Croatian, Serbian, Bosnian and Montenegrin, by setting up a diverse benchmark for these languages, and comparing the trained-from-scratch models with the new models constructed via additional pretraining of existing multilingual models. We show that comparable performance to dedicated from-scratch models can be obtained by additionally pretraining available multilingual models even with a limited amount of computation. We also show that neighboring languages, in our case Slovenian, can be included in the additional pretraining with little to no loss in the performance of the final model.", }

The world of language models is going through turbulent times, better and ever larger models are coming out at an unprecedented speed. However, we argue that, especially for the scientific community, encoder models of up to 1 billion parameters are still very much needed, their primary usage being in enriching large collections of data with metadata necessary for downstream research. We investigate the best way to ensure the existence of such encoder models on the set of very closely related languages - Croatian, Serbian, Bosnian and Montenegrin, by setting up a diverse benchmark for these languages, and comparing the trained-from-scratch models with the new models constructed via additional pretraining of existing multilingual models. We show that comparable performance to dedicated from-scratch models can be obtained by additionally pretraining available multilingual models even with a limited amount of computation. We also show that neighboring languages, in our case Slovenian, can be included in the additional pretraining with little to no loss in the performance of the final model.

Language Models on a Diet: Cost-Efficient Development of Encoders for Closely-Related Languages via Additional Pretraining

sigul-1.23

Poster

2404.05428

https://huggingface.co/papers/2404.05428

https://aclanthology.org/2024.sigul-1.24.bib

https://aclanthology.org/2024.sigul-1.24/

@inproceedings{bick-etal-2024-man, title = "Man or Machine: Evaluating Spelling Error Detection in {D}anish Newspaper Corpora", author = "Bick, Eckhard and Blom, Jonas Nygaard and Rathje, Marianne and Schack, J{\o}rgen", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.24", pages = "204--211", abstract = "This paper evaluates frequency and detection performance for both spelling and grammatical errors in a corpus of published Danish newspaper texts, comparing the results of three human proofreaders with those of an automatic system, DanProof. Adopting the error categorization scheme of the latter, we look at the accuracy of individual error types and their relative distribution over time, as well as the adequacy of suggested corrections. Finally, we discuss so-called artefact errors introduced by corpus processing, and the potential of DanProof as a corpus cleaning tool for identifying and correcting format conversion, OCR or other compilation errors. In the evaluation, with balanced F1-scores of 77.6 and 67.6 for 1999 texts and 2019 texts, respectively, DanProof achieved a higher recall and accuracy than the individual human annotators, and contributed the largest share of errors not detected by others (16.4{\%} for 1999 and 23.6{\%} for 2019). However, the human annotators had a significantly higher precision. Not counting artifacts, the overall error frequency in the corpus was low ( 0.5{\%}), and less than half in the newer texts compared to the older ones, a change that mostly concerned orthographical errors, with a correspondingly higher relative share of grammatical errors.", }

This paper evaluates frequency and detection performance for both spelling and grammatical errors in a corpus of published Danish newspaper texts, comparing the results of three human proofreaders with those of an automatic system, DanProof. Adopting the error categorization scheme of the latter, we look at the accuracy of individual error types and their relative distribution over time, as well as the adequacy of suggested corrections. Finally, we discuss so-called artefact errors introduced by corpus processing, and the potential of DanProof as a corpus cleaning tool for identifying and correcting format conversion, OCR or other compilation errors. In the evaluation, with balanced F1-scores of 77.6 and 67.6 for 1999 texts and 2019 texts, respectively, DanProof achieved a higher recall and accuracy than the individual human annotators, and contributed the largest share of errors not detected by others (16.4{\%} for 1999 and 23.6{\%} for 2019). However, the human annotators had a significantly higher precision. Not counting artifacts, the overall error frequency in the corpus was low ( 0.5{\%}), and less than half in the newer texts compared to the older ones, a change that mostly concerned orthographical errors, with a correspondingly higher relative share of grammatical errors.

Man or Machine: Evaluating Spelling Error Detection in Danish Newspaper Corpora

sigul-1.24

Poster

https://aclanthology.org/2024.sigul-1.25.bib

https://aclanthology.org/2024.sigul-1.25/

@inproceedings{vergez-couret-etal-2024-managing, title = "Managing Fine-grained Metadata for Text Bases in Extremely Low Resource Languages: The Cases of Two Regional Languages of {F}rance", author = "Vergez-Couret, Marianne and Bernhard, Delphine and Nauge, Michael and Bras, Myriam and Ruiz Fabo, Pablo and Werner, Carole", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.25", pages = "212--221", abstract = "Metadata are key components of language resources and facilitate their exploitation and re-use. Their creation is a labour intensive process and requires a modeling step, which identifies resource-specific information as well as standards and controlled vocabularies that can be reused. In this article, we focus on metadata for documenting text bases for regional languages of France characterised by several levels of variation (space, time, usage, social status), based on a survey of existing metadata schema. Moreover, we implement our metadata model as a database structure for the Heurist data management system, which combines both the ease of use of spreadsheets and the ability to model complex relationships between entities of relational databases. The Heurist template is made freely available and was used to describe metadata for text bases in Alsatian and Poitevin-Santongeais. We also propose tools to automatically generate XML metadata headers files from the database.", }

Metadata are key components of language resources and facilitate their exploitation and re-use. Their creation is a labour intensive process and requires a modeling step, which identifies resource-specific information as well as standards and controlled vocabularies that can be reused. In this article, we focus on metadata for documenting text bases for regional languages of France characterised by several levels of variation (space, time, usage, social status), based on a survey of existing metadata schema. Moreover, we implement our metadata model as a database structure for the Heurist data management system, which combines both the ease of use of spreadsheets and the ability to model complex relationships between entities of relational databases. The Heurist template is made freely available and was used to describe metadata for text bases in Alsatian and Poitevin-Santongeais. We also propose tools to automatically generate XML metadata headers files from the database.

Managing Fine-grained Metadata for Text Bases in Extremely Low Resource Languages: The Cases of Two Regional Languages of France

sigul-1.25

Poster

https://aclanthology.org/2024.sigul-1.26.bib

https://aclanthology.org/2024.sigul-1.26/

@inproceedings{al-ali-aldarmaki-2024-mixat, title = "Mixat: A Data Set of Bilingual Emirati-{E}nglish Speech", author = "Al Ali, Maryam Khalifa and Aldarmaki, Hanan", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.26", pages = "222--226", abstract = "This paper introduces Mixat: a dataset of Emirati speech code-mixed with English. Mixat was developed to address the shortcomings of current speech recognition resources when applied to Emirati speech, and in particular, to bilignual Emirati speakers who often mix and switch between their local dialect and English. The data set consists of 15 hours of speech derived from two public podcasts featuring native Emirati speakers, one of which is in the form of conversations between the host and a guest. Therefore, the collection contains examples of Emirati-English code-switching in both formal and natural conversational contexts. In this paper, we describe the process of data collection and annotation, and describe some of the features and statistics of the resulting data set. In addition, we evaluate the performance of pre-trained Arabic and multi-lingual ASR systems on our dataset, demonstrating the shortcomings of existing models on this low-resource dialectal Arabic, and the additional challenge of recognizing code-switching in ASR. The dataset will be made publicly available for research use.", }

This paper introduces Mixat: a dataset of Emirati speech code-mixed with English. Mixat was developed to address the shortcomings of current speech recognition resources when applied to Emirati speech, and in particular, to bilignual Emirati speakers who often mix and switch between their local dialect and English. The data set consists of 15 hours of speech derived from two public podcasts featuring native Emirati speakers, one of which is in the form of conversations between the host and a guest. Therefore, the collection contains examples of Emirati-English code-switching in both formal and natural conversational contexts. In this paper, we describe the process of data collection and annotation, and describe some of the features and statistics of the resulting data set. In addition, we evaluate the performance of pre-trained Arabic and multi-lingual ASR systems on our dataset, demonstrating the shortcomings of existing models on this low-resource dialectal Arabic, and the additional challenge of recognizing code-switching in ASR. The dataset will be made publicly available for research use.

Mixat: A Data Set of Bilingual Emirati-English Speech

sigul-1.26

Poster

2405.02578

https://aclanthology.org/2024.sigul-1.27.bib

https://aclanthology.org/2024.sigul-1.27/

@inproceedings{arthur-etal-2024-multi, title = "Bi-dialectal {ASR} of {A}rmenian from Naturalistic and Read Speech", author = "Arthur, Malajyan and Khurshudyan, Victoria and Avetisyan, Karen and Dolatian, Hossep and Nouvel, Damien", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.27", pages = "227--236", abstract = "The paper explores the development of Automatic Speech Recognition (ASR) models for Armenian, by using data from two standard dialects (Eastern Armenian and Western Armenian). The goal is to develop a joint bi-variational model. We achieve state-of-the-art results. Results from our ASR experiments demonstrate the impact of dataset selection and data volume on model performance. The study reveals limited transferability between dialects, although integrating datasets from both dialects enhances overall performance. The paper underscores the importance of dataset diversity and volume in ASR model training for under-resourced languages like Armenian.", }

The paper explores the development of Automatic Speech Recognition (ASR) models for Armenian, by using data from two standard dialects (Eastern Armenian and Western Armenian). The goal is to develop a joint bi-variational model. We achieve state-of-the-art results. Results from our ASR experiments demonstrate the impact of dataset selection and data volume on model performance. The study reveals limited transferability between dialects, although integrating datasets from both dialects enhances overall performance. The paper underscores the importance of dataset diversity and volume in ASR model training for under-resourced languages like Armenian.

Bi-dialectal ASR of Armenian from Naturalistic and Read Speech

sigul-1.27

Poster

https://aclanthology.org/2024.sigul-1.28.bib

https://aclanthology.org/2024.sigul-1.28/

@inproceedings{nguyen-sakti-2024-multilingual, title = "Multilingual Self-supervised Visually Grounded Speech Models", author = "Nguyen, Huynh Phuong Thanh and Sakti, Sakriani", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.28", pages = "237--243", abstract = "Developing a multilingual speech-to-speech translation system poses challenges due to the scarcity of paired speech data in various languages, particularly when dealing with unknown and untranscribed languages. However, the shared semantic representation across multiple languages presents an opportunity to build a translation system based on images. Recently, researchers have explored methods for aligning bilingual speech as a novel approach to discovering speech pairs using semantic images from unknown and untranscribed speech. These aligned speech pairs can then be utilized to train speech-to-speech translation systems. Our research builds upon these approaches by expanding into multiple languages and focusing on achieving multimodal multilingual pairs alignment, with a key component being multilingual visually grounded speech models. The objectives of our research are twofold: (1) to create visually grounded speech datasets for English, Japanese, Indonesian, and Vietnamese, and (2) to develop self-supervised visually grounded speech models for these languages. Our experiments have demonstrated the feasibility of this approach, showcasing the ability to retrieve associations between speeches and images. The results indicate that our multilingual visually grounded speech models yield promising outcomes in representing speeches using semantic images across multiple languages.", }

Developing a multilingual speech-to-speech translation system poses challenges due to the scarcity of paired speech data in various languages, particularly when dealing with unknown and untranscribed languages. However, the shared semantic representation across multiple languages presents an opportunity to build a translation system based on images. Recently, researchers have explored methods for aligning bilingual speech as a novel approach to discovering speech pairs using semantic images from unknown and untranscribed speech. These aligned speech pairs can then be utilized to train speech-to-speech translation systems. Our research builds upon these approaches by expanding into multiple languages and focusing on achieving multimodal multilingual pairs alignment, with a key component being multilingual visually grounded speech models. The objectives of our research are twofold: (1) to create visually grounded speech datasets for English, Japanese, Indonesian, and Vietnamese, and (2) to develop self-supervised visually grounded speech models for these languages. Our experiments have demonstrated the feasibility of this approach, showcasing the ability to retrieve associations between speeches and images. The results indicate that our multilingual visually grounded speech models yield promising outcomes in representing speeches using semantic images across multiple languages.

Multilingual Self-supervised Visually Grounded Speech Models

sigul-1.28

Poster

https://aclanthology.org/2024.sigul-1.29.bib

https://aclanthology.org/2024.sigul-1.29/

@inproceedings{nakarmi-etal-2024-nepal, title = "{N}epal Script Text Recognition Using {CRNN} {CTC} Architecture", author = "Nakarmi, Swornim and Sthapit, Sarin and Shakya, Arya and Chulyadyo, Rajani and Bal, Bal Krishna", editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.29", pages = "244--251", abstract = "Nepal Script (also known as Prachalit Script) is the widely used script of Nepal Bhasa, the native language of the Kathmandu Valley in Nepal. Derived from the Brahmi Script, the Nepal Script was developed in the 9th century and was extensively used till the 20th century, before being replaced by the Devanagari script. Numerous ancient manuscripts, inscriptions, and documents written in the Nepal Script are still available containing immense knowledge on architecture, arts, astrology, ayurveda, literature, music, tantrism, etc. To preserve and revive Nepal Bhasa, digitizing such documents plays a crucial role. This paper presents our work on text recognition for the Nepal Script. The implementation includes the Nepal Script text recognizer based on CRNN CTC architecture aided by line and word segmentations. Leveraging a carefully curated dataset that encompasses handwritten and printed texts in the Nepal Script, our work has achieved CER of 6.65{\%} and WER of 13.11{\%}. The dataset used for this work is available as Nepal Script Text Dataset on Kaggle. The paper further explores the associated challenges due to the complex nature of the script such as conjuncts, modifiers and variations; and the current state of the script.", }

Nepal Script (also known as Prachalit Script) is the widely used script of Nepal Bhasa, the native language of the Kathmandu Valley in Nepal. Derived from the Brahmi Script, the Nepal Script was developed in the 9th century and was extensively used till the 20th century, before being replaced by the Devanagari script. Numerous ancient manuscripts, inscriptions, and documents written in the Nepal Script are still available containing immense knowledge on architecture, arts, astrology, ayurveda, literature, music, tantrism, etc. To preserve and revive Nepal Bhasa, digitizing such documents plays a crucial role. This paper presents our work on text recognition for the Nepal Script. The implementation includes the Nepal Script text recognizer based on CRNN CTC architecture aided by line and word segmentations. Leveraging a carefully curated dataset that encompasses handwritten and printed texts in the Nepal Script, our work has achieved CER of 6.65{\%} and WER of 13.11{\%}. The dataset used for this work is available as Nepal Script Text Dataset on Kaggle. The paper further explores the associated challenges due to the complex nature of the script such as conjuncts, modifiers and variations; and the current state of the script.

Nepal Script Text Recognition Using CRNN CTC Architecture

sigul-1.29

Poster

https://aclanthology.org/2024.sigul-1.30.bib

https://aclanthology.org/2024.sigul-1.30/

@inproceedings{cusenza-coltekin-2024-nlp, title = {{NLP} for Arb{\"e}resh: How an Endangered Language Learns to Write in the 21st Century}, author = {Cusenza, Giulio and {\c{C}}{\"o}ltekin, {\c{C}}a{\u{g}}r{\i}}, editor = "Melero, Maite and Sakti, Sakriani and Soria, Claudia", booktitle = "Proceedings of the 3rd Annual Meeting of the Special Interest Group on Under-resourced Languages @ LREC-COLING 2024", month = may, year = "2024", address = "Torino, Italia", publisher = "ELRA and ICCL", url = "https://aclanthology.org/2024.sigul-1.30", pages = "252--256", abstract = {Societies are becoming more and more connected, and minority languages often find themselves helpless against the advent of the digital age, with their speakers having to regularly turn to other languages for written communication. This work introduces the case of Arb{\"e}resh, a southern Italian language related to Albanian. It presents the very first machine-readable Arb{\"e}resh data, collected through a web campaign, and describes a set of tools developed to enable the Arb{\"e}resh people to learn how to write their language, including a spellchecker, a conjugator, a numeral generator, and an interactive platform to learn Arb{\"e}resh spelling. A comprehensive web application was set up to make these tools available to the public, as well as to collect further data through them. This method can be replicated to help revive other minority languages in a situation similar to Arb{\"e}resh{'}s. The main challenges of the process were the extremely low-resource setting and the variability of Arb{\"e}resh dialects.}, }

Societies are becoming more and more connected, and minority languages often find themselves helpless against the advent of the digital age, with their speakers having to regularly turn to other languages for written communication. This work introduces the case of Arb{\"e}resh, a southern Italian language related to Albanian. It presents the very first machine-readable Arb{\"e}resh data, collected through a web campaign, and describes a set of tools developed to enable the Arb{\"e}resh people to learn how to write their language, including a spellchecker, a conjugator, a numeral generator, and an interactive platform to learn Arb{\"e}resh spelling. A comprehensive web application was set up to make these tools available to the public, as well as to collect further data through them. This method can be replicated to help revive other minority languages in a situation similar to Arb{\"e}resh{'}s. The main challenges of the process were the extremely low-resource setting and the variability of Arb{\"e}resh dialects.

NLP for Arbëresh: How an Endangered Language Learns to Write in the 21st Century

sigul-1.30

Poster