forum_id
stringlengths 10
10
| forum_title
stringlengths 5
188
| forum_authors
sequencelengths 0
98
| forum_abstract
stringlengths 3
4.69k
| forum_keywords
sequencelengths 0
29
| forum_pdf_url
stringlengths 40
40
| note_id
stringlengths 10
10
| note_type
stringclasses 5
values | note_created
int64 1,695B
1,737B
| note_replyto
stringlengths 10
10
| note_readers
sequencelengths 1
6
| note_signatures
sequencelengths 1
1
| note_text
stringlengths 14
30.1k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
MicqkyVM16 | Avoiding 'generatese': the optimization of NLG systems through fit-for-purpose data collections | [] | Most linguistic research on the use and exploitation of Natural Language Generation (NLG) systems, whether through graphical interfaces (as in the case of ChatGPT or Gemini) or without them, has primarily focused on their ability to generate text on the basis of prompts. These systems have a wide range of applications, one of which is the interlingual translation of text. They are also able to generate text from a prompt, ei-ther in response to a question or a request to perform a linguistic task. Their apparent ability to generate coherent text from another text surpasses the functionalities of any previous linguistic resource.
A translated text often retains certain traces of the source text and language, a phe-nomenon known as "translationese" (Baker, 1993). With the widespread adoption of machine translation, especially in certain genres, there has been an observable intensi-fication of this phenomenon, which has been termed "post-editese" (Toral, 2019). This can be detected through measurements of specific linguistic aspects and comparisons of human and machine translations using parallel and reference corpora.
Recently, AI systems known as Large Language Models (LLMs) have begun to be used in both professional translation and translator training. The potential footprint such systems leave on translated texts could be called ‘generatese’ (Sánchez-Gijón, 2024). The principle of language agnosticism that underlies NLG systems can affect not only the form of discourse (the linguistic features of a text) but also its content (the con-cepts and ideas it contains and how they are developed) (Sánchez-Gijón, 2022; Imran et al., 2023). This paper aims to study the impact of using small, highly fit-for-purpose data collections to optimize NLG systems by reducing the randomness of their re-sponses and mitigating ‘generatese’. We will explore the creation and, in particular, the description of such data collections, along with their potential for enhancing the quality of translations produced by NLG systems. | [
"NLG systems",
"data collections",
"translation",
"generatese"
] | https://openreview.net/pdf?id=MicqkyVM16 | h7qjtNq0RG | decision | 1,717,402,741,334 | MicqkyVM16 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
LQ75lpOR6p | Radical and Word Entanglements: Take “女” as an Example | [] | The relationship between the part and the whole has been a concern of people since the emergence of logic. In linguistics, the relationship between radical and the word has been vaguely summarized as follows: as a part of the word, radical endows and affects semantics. To give a more scientific and rational explanation to the long-standing problem, this study collected 477 words containing radical “女” from the latest edition of Xinhua Dictionary. The clustering data is divided into 6 classes by meaning and corresponding extracted contextual word embeddings from a Chinese BERT model. This unsupervised machine learning observes the relationship between classifiers in terms of distribution, joint probability, and usage. In addition, the number distribution of 6 semantic classes, the position of radical, and the number of strokes in the word are also analyzed to help prove the research results: words with radical “女” can be divided into classes of “female”, “Quality”, “Movement”, “Name”, “Emotion”, and “Phenomenon”, and the word distribution should be affected by the frequency of use. The study also has broader implications for the language type distribution for computational research. | [
"radical",
"classifier",
"word distribution"
] | https://openreview.net/pdf?id=LQ75lpOR6p | NymXJtI8St | decision | 1,717,401,428,199 | LQ75lpOR6p | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author,
Congratulations. Our reviewers have rated your paper and, provided the necessary revisions are made, we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
L2oHwwRXpj | Manual data collection & qualitative analysis for social media data – “luddite” meme researcher insecurities in the age of AI | [] | Manual data collection & qualitative analysis for social media data – “luddite” meme researcher insecurities in the age of AI
Digital data and digital research methods have undoubtedly opened a whole new world for researchers. However, they have also raised fears and insecurities about research ethics and the researcher’s role. For some, AI and other innovative digital tools seem too scary to learn or take control of. This paper focuses these personal and professional insecurities about digital, automated research methods as researcher on memes who utilizes manual data collection and qualitative analysis in the age of AI, alongside with presenting the overall PhD project about representations of mental health and mental illness in Internet memes.
The data of this paper consists of approximately 900 manually collected mental health themed memes collected under two social media and Internet platforms: Instagram and Imgur.com. On Instagram, the data is collected manually by taking screenshots of memes under hashtags #mentalhealthmeme and #mentalillnessmemes, as well as three accounts focusing on mental health related themes. On Imgur.com, the data is detected and collected on the Most Viral page, especially under larger image cluster posts called Meme Dumps. The data is collected gradually throughout the research process, with more intensive data collection periods implemented. During the time of writing, the latest more intensive data collection phase was conducted in January 2024. Moreover, reverse snowball method (Särmä, 2014: 99) will be utilized in the data collection throughout the whole research process. The aim of this study is to examine different semiotic ways these memes represent mental health issues by analysing the data with multimodal critical discourse analysis. In this paper I present the insecurities about utilizing digital data collection and analysis tools in meme research, as well as the arguments for not utilizing them in relation to the overall PhD project. Fears of not learning AI or other data scraping tools include being left behinf in digital development, cherry-picking research data, and being labelled as a luddite not willing or not being able to learn new methods. However, arguments for more manual, researcher-oriented methods in collecting social media data, will also be discussed. Overall, the aim of this paper is to seek encouragement from peers, as well as new ideas and support in utilizing digital research methods, and most importantly, how to connect those methods to qualitative and critical research.
References
Särmä, S. (2014). Junk feminism and nuclear wannabes: Collaging parodies of Iran and North Korea. Tampere: Tampere University Press.Cambridge, MA: MIT Press. | [
"mental health",
"memes",
"researcher insecurities"
] | https://openreview.net/pdf?id=L2oHwwRXpj | 1UytFHE245 | decision | 1,717,402,799,593 | L2oHwwRXpj | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
KcrtFMyKWZ | Towns on the Eastern Border of the Swedish Great Power | [] | Digital Research Data and Human Sciences (DRDHum) 2024
Digital Research Data and Human Sciences in the Age of A.I., Joensuu December 10–12, 2024
Poster proposal
Towns on the Eastern Border of the Swedish Great Power
Power Structures and Commercial Networks in the Second Half of the 17th Century
The Towns on the Eastern Border of the Swedish Great Power project explores the power and trade networks of five cities in the late 17th century. Nyen, Kexholm, Sortavala, Brahea and Kajaani located closest to the eastern border. Despite their small size, they were important commercial centers of their regions. The project examines interaction within and between cities, but also connections with other regions of the Swedish Empire and Russia. The aim is to bring a comparative perspective to microhistorical research, enabled by new digital methods. In cooperation with the National Archives of Finland, HTR technology is used to produce a digitized corpus of sources that is analysed with digital network analysis tools. Network analysis, comparison of cities and a close reading of sources characteristic of microhistory are expected to reveal new features of the patterns of behaviour of urban communities in the early modern era.
The workflow of the project is as follows. The project has about 700 pages of machine-readable material ready for teaching the HTR-program's algorithm. The approximately 10,000 document pages in the audited cities are interpreted into an electronically readable format with the help of the Transkribus program in the National Archives of Finland. Coordinates can be added to the XML file format, for example, for place names where the detected point of text is located in the machine-scanned image. When creating the observation matrices, the PDF file format enables search functions based on OCR technology.
Observation matrices are created with Excel from the corpus for network analysis. Visone, which operates in a Java environment, has been chosen as the network analysis tool. The internal and external interactions of cities are primarily seen as explicit social networks. Through their descriptions, the systemic meanings of interaction and dependence are sought. Visual network descriptions produced with the Visone application may be accompanied by attributes that describe time and place (geographic coordinates).
The data can be combined with a digital map model, whereby the networks are also presented as distances to nodes that are connected to each other (links). The directions of interaction (knot, in-grade, or out-grade) and strength (intensity) may also be included in the review. From the city, place or group under consideration (hub), it is possible to determine its characteristics in the network, i.e. whether it is a "star" in the middle of the network, a “liaison”, a “bridge” or a so-called "gatekeeper". In this way, abstracted network descriptions of the interaction between places and people make it possible to investigate commercial systems between cities and bourgeois on the eastern border and compare networks within and between cities.
The project studies also the different aspects of segregation inside the cities. In segregation research, QGIS geoinformatics software is used for visual representations of the social dimensions of urban space. This is done by creating a digital map from town plan drawings to which various collected real estate data can be added, such as occupation derived from the titles of the plot owners, wealth based on tax information, ethnic background derived from the name (Swedish, Finnish, German, Karelian, Russian) and thus also an assessment of the person's religion. In this way, a picture can be obtained of the spatially ordered social characteristics of urban areas. | [
"Digital humanism",
"handwritten text recognition",
"network analysis"
] | https://openreview.net/pdf?id=KcrtFMyKWZ | RnA9dDDEUL | decision | 1,717,401,813,061 | KcrtFMyKWZ | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s
Congratulations. This is to let you know that we are happy to accept your proposed poster.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
JnAK1X05nF | Using deep learning to examine cross-linguistic similarities of registers | [] | This study examines the use of multilingual deep learning to analyze cross-linguistic similarities of registers – situationally defined text varieties such as news or reviews (Biber 1988). Register studies have repeatedly shown that differences in the situational context of a text are reflected in its linguistic characteristics. However, little is known about register variation across languages (see, however, Biber 2014; Li et al. 2023). One of the reasons for this is the lack of methods enabling the analysis of registers without the manual interpretation of register characteristics in each language. In this study, we apply multilingual deep learning to fill this gap.
We examine cross-linguistic similarities of registers using the deep learning model XLM-R (Conneau et al. 2020). Specifically, we target eight registers and eight languages: English, French, Swedish, Finnish, Turkish, Urdu, Chinese, and Farsi. First, using the multilingual CORE corpora (Laippala et al. 2022) and XLM-R, we train a multilingual register identification model. The model learns to classify documents to register classes and creates document vectors that represent the documents in one multilingual vector space. This allows us to examine registers and their similarities across languages by calculating document similarities in the vector space. Second, we extract keywords for the registers using the trained model and the model explanation method SACX (Rönnqvist et al. 2022). This enables the analysis of the linguistic motivation behind the learnt model. We group the keywords using semantic and grammatical criteria and analyze the registers and their similarities across languages based on these groupings. Furthermore, we compare our findings to previous studies based on more frequently applied statistical methods, such as multi-dimensional analysis.
Preliminary results show that the model learns to identify the registers at a nearly human-level performance. In the vector space, the documents are structured to language-independent and register-specific groupings. This shows that the model has learnt language-independent representations of the registers. Furthermore, the analysis of the keywords shows that the learning is based on linguistically motivated features. For instance, the keywords feature semantic properties such as stance and functional features such as reporting verbs that characterize registers across languages – and have been identified as register characteristics in previous studies focusing on individual languages. Thus, our findings support the existence of register universals (Biber 2014) and encourage the use of multilingual deep learning for cross-linguistic corpus analyses.
References:
Biber, D. (1988). Variation across speech and writing. Cambridge: Cambridge University Press.
Biber, D. (2013). In Gray, Bethany. (2013). Interview with Douglas Biber. Journal of English Linguistics 41(4), 359–379. https://doi.org/10.1177/0075424213502237.
Biber, D. (2014). Using multi-dimensional analysis to explore cross-linguistic universals of register variation. Languages in Contrast, 14(1), 7-34. https://doi.org/10.1075/lic.14.1.02bib
Conneau, A., Khandelwal, K., Goyal, N., et al. (2020). Unsupervised Cross-lingual Representation Learning at Scale. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 8440-8451. Association for Computational Linguistics.
Laippala, V., Salmela, A., Rönnqvist, S., Aji, A. F., Chang, L.-H., Dhifallah, A., Goulart, L., Kortelainen, H., Pàmies, M., Prina Dutra, D., Skantsi, V., Sutawika, L., & Pyysalo, S. (2022). Towards better structured and less noisy Web data: Oscar with Register annotations. In Proceedings of the Eighth Workshop on Noisy User-generated Text (W-NUT 2022), 215-221. Association for Computational Linguistics.
Li, H., Dunn, J. & Nini, A. (2022). Register variation remains stable across 60 languages. Corpus Linguistics and Linguistic Theory, 19(3), 397-426. https://doi.org/10.1515/cllt-2021-0090
Rönnqvist, S., Kyröläinen, A.-J., Myntti, A., & Laippala, V. (2022). Explaining Classes through Stable Word Attributions. In Findings of the Association for Computational Linguistics, 1063-1074. Association for Computational Linguistics. | [
"multilingual machine learning",
"web-as-corpus",
"web registers",
"register universals",
"keyword extraction"
] | https://openreview.net/pdf?id=JnAK1X05nF | LGnIsJel3b | decision | 1,717,404,008,462 | JnAK1X05nF | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
JjLcyjUkiP | Data-rich History for 19th Century Literature in Finland | [] | Data-rich History for 19th Century Literature in Finland
Kati Launis
University of Eastern Finland, Finland
kati.launis@uef.fi
Aino Mäkikalli
University of Turku, Finland
ainmak@utu.fi
Viola Parente-Čapková
University of Turku, Finland
viocap@utu.fi
Veli-Matti Pynttäri
University of Eastern Finland, Finland
veli-matti.pynttari@uef.fi
Osma Suominen
National Library of Finland, Finland
osma.suominen@helsinki.fi
Abstract
Consortium Digital History for Literature in Finland (Research Council of Finland, 2022–26) focuses on renewing current understanding of the literary history of the long 19th century – which, in Finland, covers the period of autonomy, 1809–1917 – by com-bining detailed historical analysis with a dedicated bibliographic data science framework, presented by Julia Matveeva, Osma Suominen and Leo Lahti in their abstract. Collabora-tion between researchers from the universities of Turku and Eastern Finland and the Na-tional Library of Finland makes it possible, for the first time, to build a data rich view of the history of Finland’s publishing landscape. We will systematically mine the biblio-graphic metadata for all available published fictional works that are included in the Na-tional Bibliography Fennica and complement this by close reading of the Finnish and Swedish language literary texts in either digital or traditional paper format.
In this presentation, we are focusing on the research done in the WP1 (Literary history in the 19th century Finland, UEF) and WP 3 (Digital resources, The National Library of Finland). Based on the curated corpus – which is currently under development – of about 2800 first editions of the published fictional works that are included in the National Bibli-ography Fennica, we ask, what kind of fiction was published in Finland during the 19th century. What kind of changes and patterns rise above others? Which genres can be con-sidered as prolific literary forms in the literary scene? The titles are of particular interest as paratexts (Genette 1987/1997), as they contain a wealth of genre terms: what do the titles tell us about the way contemporaries structured the genre?
This multidisciplinary project is framed in the field of digital humanities. It stems from the first wave of such initiatives in Finland; elsewhere, such digital or data rich literary history has been developed e. g. by Franco Moretti (2000/2013) and Katherine Bode (2018). The method used in the project is a combination of bibliographic data science (Lahti & al 2019), data-rich literary history (Bode 2018) as well as meso-analysis (Saint-Amour 2019, Parente-Capková, Launis & Westerlund 2023) or resourceful reading (Bode ja Dixon 2010), which combine the distant and close reading of literature. We use both new-empirical and eResearch techniques and put emphasis on empirical bibliographical study (Bode & Dixon 2010); our approach is defined by the use of digital methods and exper-tise in data science, bibliographic and library knowledge, and a solid knowledge of the literary history and the 19th century context.
Keywords: bibliographic data science, data-rich literary history, Finnish literature, meso-analysis, National Bibliography Fennica.
REFERENCES
Bode, K. (2018). World of Fiction. Digital Collections and the Future of Literary History. University of Michigan Press.
Bode, K. & Dixon, B. (2010). Resourceful Reading: A New Empiricism in the Digital Age. In K. Bode & R. Dixon (Eds.), Resourceful Reading: The New Empiricism, ERe-search and Australian Literary Culture, 1–27. Sydney University Press.
Genette, G. (1987/1997). Paratexts. Tresholds of Interpretation. Transl. Jane E. Lewin. Cambridge University Press.
Lahti L., Marjanen J., Roivainen H. & Tolonen, M. (2019). Bibliographic data science and the history of the book (c. 1500–1800). Cataloging & Classification Quarterly 57(1), 5–23. http://dx.doi.org/10.1080/01639374.2018.1543747
Moretti, F. (2000/2013). Distant Reading. Verso.
Parente-Čapková, V., Launis, K. & Westerlund, J. (2023). Digitaaliset metadata-arkistot ja mesoanalyysi kulttuurisen vaihdon kartoittamisessa. Avain - Kirjallisuudentutkimuksen aikakauslehti, vol 20, nro 1, 100-111. https://journal.fi/avain/article/view/127486/77833.
Saint-Amour, P. (2019). The Medial Humanities: Toward a Manifesto for Meso-Analysis. M/m, vol. 3, cycle 4. https://doi.org/10.26597/mod.0092. | [
"bibliographic data science",
"data-rich literary history",
"Finnish literature",
"meso-analysis",
"National Bibliography Fennica."
] | https://openreview.net/pdf?id=JjLcyjUkiP | jbTlGs20rD | decision | 1,717,140,853,763 | JjLcyjUkiP | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear Kati,
Very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit a pdf formatted according to the template (details for which will be send out in due course).
You can already register for the conference here:
Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
(Please chose the bottom option, being part of the organiser committee).
:-)
title: Paper Decision |
IlU2yTEYS8 | Grasping the 'Freedom of Speech' Argument on Social Media, Between Circulation and Escalation: Cross-Contributions of Digital Methods and Social Psychology | [] | I will present my ongoing doctoral research, which explores the role of freedom of speech as a catalyst for conflicts, examining its manifestations across various domains such as politics, culture, and digital regulation. Using digital methodologies and mixed approaches, I analyze how the argument of freedom of speech circulates across various media and platforms. This summary encapsulates the ongoing findings of my research, which will be showcased in the form of a poster. | [
"Freedom of speech",
"Polarization",
"Social network",
"Mixed-methods research",
"Transmedia circulation",
"Public discourse",
"Affect",
"Social psychology"
] | https://openreview.net/pdf?id=IlU2yTEYS8 | pMXfKt5JRX | decision | 1,717,158,704,104 | IlU2yTEYS8 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author/s
Congratulations. This is to let you know that we are happy to accept your proposed poster.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
Idywoazf8Y | Automatic Language Proficiency Assessment of Written Texts: Training a CEFR classifier in L2-Finnish | [] | The Common European Framework of Reference for Languages: Learning, Teaching, Assessment (CEFR), is a framework commonly used to assess the proficiency level of language learners (e.g. Martiyk & Noijons, 2007). It is also utilized for the assessment of language proficiency for citizenship purposes in Finland (Rocca et al.., 2020). To study the suitability of using a deep learning model for the CEFR classification task, we develop and present a language proficiency classifier for Finnish as a second lan-guage (F2) written texts. The classifier has been trained to recognize the six Common European Framework of Reference (CEFR) proficiency levels from A1 (basic user) to C2 (proficient user). During the development process, we seek answers to the following questions:
1. Is there enough Finnish learner language data for training deep learning models?
2. Training with existing datasets, how well can a deep learning model detect different CEFR levels?
3. How does the model compare to other CEFR-models?
The FinBERT (Virtanen et al., 2019) language model has been further trained with the datasets of (1) the International Corpus of Learner Finnish (ICLFI), (2) The Advanced Finnish Learner’s Corpus (LAS2), (3) subcorpus of young Finnish learners in the Ce-fling project, and (4) The Finnish Subcorpus of Topling - Paths in Second Language Acquisition. These datasets provide a volume of c. 8000 texts, combining to c. 1.5 mil-lion tokens.
After training of models, the best accuracy we obtain is a score of 76.8 %. Accuracy is calculated by dividing the number of correctly classified samples by the total number of samples (see, e.g., Tharwat, 2020), so for instance, 100 % accuracy would mean that the model classified all the data samples correctly. The results indicate that there is room for improvement in model performance and a need for more CEFR-annotated Finnish learner language training data. However, the wrong classifications were most-ly only off by one proficiency level (e.g., if the annotation should have been A2, the classifier should label it as A1 or B1). One should also keep in mind that even human assessors do not always agree, and a similar one-off phenomenon could also present itself (see, e.g. Yancey et al., 2023).
REFERENCES
Martinyuk, W. & Noijons, J. (2007). The use of the CEFR at national level in the Council of Europe Member States. The Common European Framework of Reference for Languges (CEFR) and the development of language policies: challenges and responsibilities, Strasbourg, 6-8 February 2007. Council of Europe.
Rocca, L., Carlsen, C. H, & Deygers, B. (2020). Linguistic Integration of adult migrants: requirements and learning opportunities. Report on the 2018 Council of Europe and ALTE survey on language and knowledge of society policies for migrants. Council of Europe.
Tharwat, A. (2020). Classification assessment methods. Applied computing and informatics 17(1),168-192. DOI:10.1016/j.aci.2018.08.003
Virtanen, A., Kanerva, J., Ilo, R., Luoma, J., Luotolahti, J., Salakoski, T., Ginter, F., & Pyysa-lo, S. (2019). Multilingual is not enough: BERT for Finnish. arXiv preprint arXiv:1912.07076.
Yancey, K., Laflair, P., Verardi, G. A., & Burstein, J. (2023). Rating Short L2 Essays on the CEFR Scale With GPT-4. In E. Kochmar, J. Burstein, A. Horbach, R. Laarmann- Quante, N. Madnani, A. Tack, V. Yaneva, Z. Yuan & T. Zesch (Eds.), Proceedings of the 18th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2023) (pp. 576-584). Association for Computational Linguistics. | [
"Automatic Writing Assessment",
"CEFR",
"Classifier",
"L2 Proficiency",
"LLM"
] | https://openreview.net/pdf?id=Idywoazf8Y | Zp7INKQY3P | decision | 1,717,402,467,461 | Idywoazf8Y | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
IbbT4lbX9G | Predicting Child Language Outcomes Across Diverse Longitudinal Cohorts: A Machine Learning Approach | [] | See pdf | [
"See pdf"
] | https://openreview.net/pdf?id=IbbT4lbX9G | Vmm8LkzlcR | decision | 1,717,404,052,286 | IbbT4lbX9G | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
HsuWRcui3t | Overheard in Therapy: An Analysis of a Language Aware Encounter | [] | This paper analyses a one-time therapy session with a bilingual and a monolingual client, and the therapist
doubling as the interpreter. The data come from a publicly available recording. Given the nature of
therapeutic work, the therapist’s office should be a language aware environment by default. Drawing from
Conversation Analysis, the study focuses on mitigating the linguistic insecurity of the monolingual client.
Although the group agree on ground rules, the interaction reveals the difficulties of implementing an ideal
language aware encounter. Although promising in the beginning, the interaction soon ruptures, as the
monolingual participant finds themselves excluded. The rapture in communication leads to adjusting the
practice of interpreting from treating the recipient as a completely monolingual, passive and separated from
the source language message to engaging them to interact with the source language and co-create the
interpretation. Despite being relatively short, the therapy session offers versatile material to evaluate the
success of language aware practices. Since the participants make active efforts to make their conversation
inclusive, one may extract from their strategies universal tips and guidelines for addressing linguistic
asymmetry elsewhere. | [
"Language Management Theory"
] | https://openreview.net/pdf?id=HsuWRcui3t | 9JMys4Nzdb | decision | 1,717,400,354,423 | HsuWRcui3t | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Reject
comment: Dear author,
Our reviewers have deemed the abstract submitted not sufficiently focussed for our conference and we regret to tell you that it has not been accepted this time.
Please log on to www.openreview.net to see the comments given.
We encourage you to think whether you might like to have a poster at DRDHum instead.
Thank you very much for the time and effort spent and we hope that we can welcome in you in December: Please follow the link to register : https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
https://sites.uef.fi/drd-hum-2024/
title: Paper Decision |
HSHmkmm5xf | Poster abstract: Automated tool for sharing experiential knowledge: the case of Human Science section in the Digital Citizen Science Center of University of Jyväskylä | [] | The Digital Citizen Science Centre is a multidisciplinary project that aims to develop mobile applications for the needs of different citizen science projects. It consists of four research groups from four faculties which work in collaboration with the digital services and the Centre of Open Knowledge of the University of Jyväskylä. The technology is based on a mobile application Research for JYU Mobile (RFJM).
This poster introduces the Human science section of the project which is focused on research of everyday experiences and meaningful places of nature. We will create a pilot model of an AI based and automated mobile tool for interviewing and collecting qualitative research data. We will analyze the everyday experiences and interactions with nature of different kinds, which could mean various elements and locations, such as forests, mires, plants, animals as well as urban environments such as parks etc. The application developed in the project will provide citizens with an opportunity to share their in-situ experiences of encountering nature and explore other users’ experiences of different natural sites.
Our aim in the human science project is to explore how an automated tool is accepted and used by citizens and how large quantities of qualitative data could be analyzed and used in research of affective everyday experiences. We will also consider the ethical challenges of creating such experiential data and motivating citizens in using the application to share their experiences.
Anna Kajander and Eerika Koskinen-Koivisto (University of Jyväskylä) | [
"citizen science",
"mobile tool",
"qualitative data",
"everyday experience"
] | https://openreview.net/pdf?id=HSHmkmm5xf | DiS8lO2aGP | decision | 1,717,158,739,621 | HSHmkmm5xf | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s
Congratulations. This is to let you know that we are happy to accept your proposed poster.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
HLA0HEU2FK | Data Sources for Automatic Classification and Analysis of Texts from Egyptian Antiquity | [] | In this poster, we present the aims and the current state of the research project "Automatic Classification and Analysis of Texts from Egyptian Antiquity", funded by the Kone Foundation. In short, the project aims to develop new state-of-the-art language technological methods for automatically processing textual documents from Egypt dating from the 8th century BCE to the Arab conquest in the 7th century CE. The project investigates the extensive textual evidence from the region as a whole, including the texts in both the Greek and the Egyptian languages.
A large part of the project is dedicated to collaboration between the project and various entities that own the copyright to the existing machine-readable texts within the focus of the research. We will identify sources for machine-readable texts pertinent to our study and, if they are not openly available, negotiate with the rightsholders for suitable access to the texts to use in the project.
We will create a database of all sources where relevant machine-readable text collections are available. The listing will be openly available on the project's website and updated throughout the project's lifespan. We will contact the entities and persons behind the text collections and aim to get the data as exports from their system instead of reverting to methods like web scraping. We have already identified several sources for texts that are usable by the project.
For the texts primarily written in Greek, we use all the transcribed texts available through the Papyri.info project as our data. Currently, the papyri.info collection contains metadata for over 100,000 texts, of which more than 50,000 are transcribed. In addition to the document data from papyri.info, we already have access to several thousand inscriptional texts from the Packard Humanities Institute's collection.
Thesaurus Linguae Aegyptiae (TLA), a digital publication platform, includes machine-readable texts written in Egyptian using either Hieroglyphic, Hieratic, or Demotic scripts. The TLA is the largest ongoing project collecting and publishing machine-readable ancient Egyptian texts, and their collection is continuously increasing. We expect the latest form of the logographic Egyptian writing, Demotic, to be most interesting regarding language contact, as it was used while the Greeks ruled Egypt. | [
"corpus",
"egyptian",
"greek"
] | https://openreview.net/pdf?id=HLA0HEU2FK | 3iwSisqPVf | decision | 1,717,159,072,128 | HLA0HEU2FK | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author/s
Congratulations. This is to let you know that we accept your proposed poster.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
G4vUTx5Cgy | The Trust Divide: Chatbots’ Superior Performance and Skeptical Students | [] | Chatbots have emerged as fundamental components of innovative teaching approaches to address the significant challenge of catering to students' diverse learning backgrounds with limited resources. However, their usefulness depends on their ability to provide high-quality, reliable outcomes and create trustworthiness. Against this background, the present study investigated the trustworthiness of chatbots in supporting educational tasks and the extent to which this trustworthiness is justified. The study involved 189 students from vocational nursing schools who created medical care plans with and without ChatGPT's (Model GPT-4) support. Additionally, ChatGPT was asked to solve the tasks without students being involved. Experts then evaluated all three sets of plans. This allowed us to compare the quality of the answers solved by the chatbot to the solutions by the students. To examine the trustworthiness of the chatbot, another independent care plan was evaluated by the students following experimentally manipulated feedback that it was created either by a human or a chatbot. Statistical analyses revealed that students' beliefs about the source of the care plan -explicitly informed as either chatbot-generated or human-created- significantly affected their perceptions of trustworthiness. Students who believed a chatbot solved the task perceived it as less trustworthy than those who thought a human created the plan. However, analyses of the expert ratings revealed that the solutions created by the chatbot were actually of higher quality than those created by humans. Therefore, the mistrust in the chatbots was not justified. This finding suggests that students underestimate chatbot performance in educational settings. | [
"chatbots",
"education",
"trustworthiness",
"quality",
"human computer interaction"
] | https://openreview.net/pdf?id=G4vUTx5Cgy | Fsnsz1ERPa | decision | 1,717,403,821,226 | G4vUTx5Cgy | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
FSSOco01K9 | Defining the core characters and events of a fictional narrative by two-mode social network analysis | [] | Inspired by the increased availability of digitized texts and the development of relevant technologies, as well as rise of the digital literary studies and the method of distant reading (Moretti, 2013), social network analysis has increasingly been used to analyze fictional narratives. The main lines of research focus on revealing patterns within a wider corpus and modelling interaction between characters within a single literary work (Chao et al., 2019). The present study adds to past research by analyzing two essential story elements in parallel, the plot and the characters. Plot is a structure connecting a series of events based on their logical relationships. Patterns of characters’ co-participation in the events allows readers to see how characters and plot define each other thus building a co-dependent relationship between these two elements of the story. Data for the article is drawn from the renowned Finnish war novel The Unknown Soldiers by Väinö Linna (2015 [1954]). A two-mode dataset was constructed to show how the 58 characters of the novel co-appear in the 88 sections of the book, each of which illustrates an episode en-countered by the troops. As a depiction of war, the novel describes various forms of interaction, including multiple characters sharing spaces and experiences together without any bilateral dialogue. The article contributes in part to the use of core-periphery analysis as an indicator of characters’ positions. Among the methods created for identifying the core-periphery structure, and whether such structure can be found, there are varying and often inconsistent assumptions about how the core and the periphery are connected to each other. A binary typology distinguishes between two-block models and k-cores (Gallagher R. J. et al., 2021). The former partitions a network into a binary hub-and-spoke layout while the k-core decomposition divides the network into a layered hierarchy. In the present article, the discussion is extended to two-mode networks and a qualitative verification of the methods. Both the two-block model (Borgatti et.al., 2018) and the k-core (Cerinšek & Batagelj, 2015) are implemented for two-mode networks. The article assesses how the selection of the core-periphery algorithm affects the ability to reveal relevant observations about the protagonists and key events of the novel. Finally, perspectives provided by the network measures are compared with those of past scholarship. | [
"two-mode networks",
"duality",
"core-periphery structure",
"Väinö Linna",
"Finnish war novel"
] | https://openreview.net/pdf?id=FSSOco01K9 | sU2IHaAxiW | decision | 1,717,403,898,759 | FSSOco01K9 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
EXvvMWkaiD | Emerging Indigenous Language Usage Practices in Digital Newspaper Readers’ Comments in Zimbabwe | [] | The advent of digital newspapers has ushered in a new era of participatory journalism whereby the stringent logic of the media is subverted. Editors no longer dictate the rules of engagement and readers are free to express themselves anyhow. Before the introduction of the digital newspapers the only available method of providing feedback to the newspapers was the “letter to the editor”, which needed to conform to certain editorial prescriptions such as sticking to the language used in the newspaper, stipulated length, disclosure of personal identity as well as adhering to ethical considerations. The new digital media environment has therefore ‘liberated’ newspaper readers from the shackles of editorial gatekeeping and the media gatekeepers’ tyranny as they now free to deliberate in a language of their choice thereby enabling them to mix the language of the media and indigenous languages. Deploying the concepts of hybridity and counter-hegemony, this paper qualitatively explores indigenous language usage practices in Zimbabwean digital newspapers. In particular, the paper examines indigenous language use practices among digital news commentators in two Zimbabwean dailies to shed light on the future of African indigenous knowledge in the context of globalization. It addresses two research questions, namely, (1) How do online news commentators use indigenous languages in online English medium newspapers? (2) How does the use of indigenous languages in online newspapers counter coloniality in the news? (3) What are the implications of using indigenous languages in digital news? Empirical data comprised of a corpus of 55 user-generated comments curated from two mainstream English medium newspapers, namely, The Herald and Newsday. The paper reveals that the digital news platform is a ‘liberated zone’ in which indigenous language speakers subvert conventional journalistic logics. Consequently, indigenous African languages become part and parcel of the global media culture, thus countering western hegemonic epistemologies. Understanding the language choices of digital news commentators helps in gaining deeper insights into the implications of mixing indigenous languages with colonial and indigenous African languages. The paper contributes new knowledge on the future digital humanities in general, and the future of indigenous languages indigenous in the context of digitality.
End/ | [
"Digital newspaper; Indigenous languages; hybridity; hegemony; counter-hegemony; media culture; globalization; User-Generated Comments"
] | https://openreview.net/pdf?id=EXvvMWkaiD | FMN5s97vuf | decision | 1,717,402,260,662 | EXvvMWkaiD | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
Please understand that we cannot provide further assistance for travela rrangements beyond what can be found on the website: https://sites.uef.fi/drd-hum-2024/practical-information/
title: Paper Decision |
DawO5FFN1W | Finding Patterns across Multiple Time Series Datasets: Democracy in the Twentieth-century Political Discourses in the United Kingdom, Sweden, and Finland | [] | This paper analyses the contextual variation of nouns and adjectives related to democracy in the United Kingdom, Sweden, and Finland in the twentieth century. We compare parliamentary data (Hansard, Riksdag, and Eduskunta) against press data (UK: Guardian and Times, Sweden: Dagens Nyheter and Svenska Dagbladet, Finland: Helsingin Sanomat and Suomen Kuvalehti). While our parliamentary datasets (Ihalainen et al. 2022) encompass several political ideologies simultaneously, the selected newspapers can broadly be categorized into conservative and liberal strands. By including both newspapers with diverse political leanings as well as parliamentary speeches, our study offers a fresh perspective on the relation between democratic discourses produced by politicians and journalists.
The approach includes visualizing the main similarities and differences in the use of democratic vocabulary between multiple historical time series datasets, as well as applying cross-correlation analysis to automatically find identical patterns between parliament and media or across different nations. The similarity of various word frequency time series charts is evaluated using the Pearson correlation coefficient (PCC), which can vary from -1 to 1. A value of 1 indicates a perfect positive correlation, where every increase in word frequency in dataset A is matched by a simultaneous increase in dataset B. Conversely, a value of -1 indicates a perfect negative correlation, where every increase in word frequency in dataset A corresponds to a simultaneous decrease in dataset B. The closer the PCC values are to 0, the weaker the relationship between the two variables (Derrick & Thomas 2004). The strengths of the PCC are its mathematical simplicity, easy interpretability, and tolerance for noise, while its main limitation is sensitivity to extreme outliers which can be mitigated by identifying and addressing outliers before conducting analysis.
Our findings indicate that the cross-correlation is strongest between similar political terms in the same dataset, e.g., the relative frequency of “democracy” and “democratic” over time in a national parliament (in Hansard 0.91, Riksdag 0.76, and Eduskunta 0.65). Another strong set of cross-correlations can be observed when the same political term appears in different datasets from the same country, e.g., the frequency of “democracy” in liberal and conservative press (in the UK 0.87, in Sweden 0.82, and 0.61 in Finland). The most important finding from a historical viewpoint is the statistically strong cross-correlation between media and parliamentary discourses, with values ranging from 0.55 to 0.76 for the term “democracy”. Transnational correlations of political terms were not as strong as intra-national correlations, but they were clearly evident in the PCC values, e.g., for the frequency of “democracy” they varied from 0.58 to 0.68 between three parliaments under investigation. The shared patterns between three parliamentary democracies include general increase in the use of “democracy” over time, with notable peaks in the 1930s as a reaction to totalitarianism, around the year 1968 related to the rise of social movements, and in the early 1990s with the fall of the Eastern bloc.
Keywords: newspapers, parliamentary speeches, text mining, time series
REFERENCES
Derrick, T., & Thomas, J. (2004). Time series analysis: The cross-correlation function. In N. Stergiou (Ed.), Innovative Analyses of Human Movement (pp. 189–205). Human Kine-tics Publishers.
Ihalainen, P., Janssen, B., Marjanen, J., & Vaara, V. (2022). Building and testing a com-parative interface on Northwest European historical parliamentary debates: Relative term frequency analysis of British representative democracy. In Digital Parliamentary Data in Action (pp. 52–68). CEUR Workshop Proceedings, Vol. 3133. http://ceur-ws.org/Vol-3133/paper04.pdf.
Wevers, M., Gao, J., & Nielbo, K. (2020). Tracking the consumption junction: Temporal dependencies between articles and advertisements in Dutch newspapers. Digital Humani-ties Quarterly, 14(2). http://www.digitalhumanities.org/dhq/vol/14/2/000445/000445.html | [
"text mining",
"time series",
"parliamentary speeches",
"newspapers"
] | https://openreview.net/pdf?id=DawO5FFN1W | HyNqrbEyo6 | decision | 1,717,404,035,637 | DawO5FFN1W | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
DUEWSrdFks | Multi-Dimensional Collocational Analysis of Discourses around COVID-19 Therapies | [] | The goal of this study is to describe the discourses related to the endorsement or opposition of alternative COVID-19 treatments during the pandemic. This was achieved by compiling a corpus of academic articles that either advocated for or criticized the use of treatments like hydroxychloroquine and azithromycin, in accordance with recommendations from the WHO and other health organizations. The dataset was selected to represent both perspectives equally. Our methodology employed Lexical Multi-Dimensional Analysis (LMDA; Berber Sardinha & Fitzsimmons-Doolan, 2024), an offshoot of Multi-Dimensional Analysis (Biber, 1988, 1995; Berber Sardinha & Veirano Pinto, 2014, 2019). This approach focuses on lexical features (e.g., lemmas) and applies multivariate statistical techniques, such as Factor Analysis, to identify correlated lexical features across the texts. Specifically, we examined the keyword collocations within two subgroups: pro-alternative treatment (PAT) and against-alternative treatment (AAT). Keywords for each subgroup were identified by using the other subgroup as a reference corpus. We then determined the collocations for each keyword within a four-word window on either side, comparing these within both the PAT and AAT datasets. For each keyword, this process yielded two sets of collocates, one for PAT and one for AAT. To manage differences in sample size, we selected the 500 most significant collocates from each subgroup based on their logDice scores. This enabled a direct comparison of how keyword collocation changes with differing treatment perspectives. For example, the collocation shift for "hydroxychloroquine" includes its association with 'treatment' in PAT texts, highlighting potential benefits and supportive guidelines, whereas in AAT texts, it is linked to discussions of mixed results concerning recovery times and side effects. Another example of collocation shift refers to ‘patients’: in the PAT texts, its collocates suggest a focus on high-risk individuals needing urgent care or facing greater health risks; conversely, in the AAT texts, 'patients' are depicted as participants in research, emphasizing a scientific evaluation of treatments' effects, efficacy, and safety. Through LMDA, we identified the major dimensions in keyword collocate use across the corpus. Initial results include a dimension contrasting discourses: one promotes the extensive use of repurposed drugs, focusing on potential benefits and minimizing risks, despite uncertain evidence of their efficacy and safety; the other advocates for a cautious evaluation of outcomes like mortality and clinical improvement, highlighting the importance of transparency and ethical considerations in research. The paper will introduce, discuss, illustrate, and compare the dimensions based on treatment stance. | [
"Lexical Multi-Dimensional Analysis",
"COVID-19",
"Collocations",
"Pseudo-Science"
] | https://openreview.net/pdf?id=DUEWSrdFks | vhA18q4rTP | decision | 1,717,403,696,221 | DUEWSrdFks | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
CBmIU75eoD | AI-based Personalized Feedback in Speech Therapy for People with Aphasia | [] | In the field of speech and language therapy, artificial intelligence has been used in di-agnostics, therapy, and assistive systems for people with aphasia (PWA) (Adikari et al., 2023; Azevedo et al., 2024; Pottinger & Kearns, 2024). AphaDIGITAL project (TDG, 2021) focuses on developing such a mobile application for German-speaking PWA that will provide personalized multilevel feedback with the help of Automatic Speech Recognition (ASR) and further text analysis. To build the corresponding pipe-line (Rykova & Walther, 2024), the following questions are addressed.
Which existing ASR solutions are suitable for the task-specific speech of Ger-man-speaking PWA?
More than 50 open-source ASR solutions were evaluated with the help of several speech recordings from different corpora (Rykova, Walther, & Zeuner, 2022). Thirteen models were selected and tested with atypical speech, including two small datasets of PWA’s speech (Rykova & Walther, in press - a). Based on Character Error Rate (CER), HITS (the number of precisely recognized words) and the number of empty outputs, four open-source ASR models were selected for the pipeline (Fleck, 2022; Grosman, 2022; Guhr, 2022; NVIDIA, 2022). These models are to a greater or lesser extent robust to speaker gender and age. The experiments suggest that for better sin-gle-word recognition the audio samples should be not too short and pronounced neither too slowly nor too fast (i.e. intentionally speeded up) (Rykova & Walther, in press - b).
How can selected ASR solutions be improved and/or adapted for the purposes of speech and language therapy?
In the absence of adequate data for ASR models’ (re)training, applying the knowledge about non-standard (aphasic and dialect) phonetic features post-hoc to ASR output was attempted. Aphasic features included recognition of syllables as separate words and vowel prolongation. Dialect features were selected from the Thuringia-Upper Saxon dialect group (Wallraff, 2007; Rocholl, 2015; B. Siebenhaar, personal communication, January, 2024). The method combined generating alternative pronunciations based on non-standard patterns (Masmoudi et al., 2014) and using alternatives for evaluation (Ali et al., 2017), and proved to work on the recordings of German aphasia test naming and repetition tasks (Huber, 1993).
How can a combination of selected ASR solutions and existing tools for seman-tic and grammatical analysis serve for speech production errors analysis?
If the answer of the speaker is not recognized as fully correct or containing phonet-ic/phonemic errors, it is subject to semantic analysis. It must be compared to the target in terms of their semantic relationship and distance. The current semantic analysis pipeline is built upon GermaNet – a semantic network for the German language (Hamp & Feldweg, 1997). It includes recognition of hyponymy/hypernymy, belonging to the same semantic (sub)category, and different lexical and conceptual relationships, de-rived from GermaNet. If the answer is not recognized as an existing word, a search for close orthographic matches is performed, and the match that is semantically the clos-est to the target is subject to the relationship analysis described above. This approach has been tested and described in detail in Rykova & Walther (2023). | [
"Aphasia",
"automatic speech recognition",
"speech and language therapy",
"digital health"
] | https://openreview.net/pdf?id=CBmIU75eoD | JyPxTEQdWA | decision | 1,717,159,181,082 | CBmIU75eoD | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author
Congratulations. This is to let you know that we are happy to accept your proposed paper.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
C40mUQsQ8Q | Encounters between the worlds of visual arts, easy language and AI | [] | Artificial intelligence produces images, but how does it recognize and interpret them? In my poster, I examine the encounters between the visual arts, easy language and artificial intelligence. How artificial intelligence can assist in producing text material, for different contexts and especially for audio description, a process of linguistic description of visual information in visual works for example for the visually impaired.
The dataset consists of different artistic images that pose different challenges to both audio-desciption and interpretation. AI uses neural networks to identify elements in a work of art. It then uses these elements to describe the image linguistically. How does AI recognize semi-abstract paintings? Compositions, shapes, rhythm? Atmosphere? Light and shadow? Can a line represent hay or water? Thus, I am also interested in how AI is able to identify and interpret not only the visible but also the interpretive elements of the image. The generative language model, which is capable of discussion, creates new content based on the prompts. Especially in this context, as the paintings are not unambiguously representational, the connotations offered in the prompts to AI are important. In the poster, I present examples about conversations with AI representing its ability to interpret the connotative aspects of the painting and if the outcome can assist in creating reliable text for audio description. Moreover, examining the conversational processes along with the prompts I will be asking if the qualitative conversation analytical (CA) approach successful in examining AI's abilities to proceed in the conversation. Additionally, preliminary findings about AI’s ability to produce liguistically simplified (easy-to-read) text is examined in those settings. | [
"AI",
"audio-description",
"easy language",
"image",
"visual art",
"conversation analysis"
] | https://openreview.net/pdf?id=C40mUQsQ8Q | FP6JbcbzJN | decision | 1,717,158,906,631 | C40mUQsQ8Q | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author/s
Congratulations. This is to let you know that we do accept your proposed poster.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
Bbay6S4gCl | Regulation of AI? Comparing Czech and Portuguese Media Imaginaries with CADS | [] | The recent buzz around Artificial Intelligence (AI) has raised significant debate in the EU on how to regulate it. Like any emerging technology, AI development depends on media-shaped public perception (Chuan et al., 2019). This research aims to contribute to the study of this perception through an interdisciplinary comparative perspective. It analyses AI regulation representations in the Czech and Portuguese online mainstream media, using Corpus Approaches to Discourse Studies (CADS) (Baker et al., 2008; Partington et al., 2013).
Czechia and Portugal are similar in area size, population, or GDP (Eurostat, 2024) but differ significantly regarding their tech sectors. While news headlines labelled Czechia the “sick man of Europe” due to its stuttering industrial economy (Willoughby, 2023), Portugal, with its vivid tech scene, was referred to as “Europe’s Silicon Valley” (Böhnisch, 2021). Focusing on the EU debate, these countries’ membership differs by almost two decades.
CADS combines corpus linguistics with traditional CDA while reflecting on its critiques (Orpin, 2005). It mitigates issues of data representativeness and interpretative transparency. CADS allows the investigation of the aggregate effects of language, highlighting typical discursive patterns. Conceptually, this study approaches AI regulation through “sociotechnical imaginaries”, understood as “collectively held, institutionally stabilised, and publicly performed visions of desirable futures” (Jasanoff and Kim, 2015, p. 4).
Current studies using imaginaries are plagued by conceptual ambiguity (Richter et al., 2023). This research overcomes the issue by adopting a three-level imaginary concept (Sau, 2021) operationalised with CADS. It structures imaginary analysis by asking for representations of (1) social commentary, (2) vision of the future, and (3) means to achieve it. The analysis covers the period of discussions about the EU’s “AI Act” regulation (3/2018-12/2023).
Comparable corpora, collected from digitally available national media in each country in this period, are compiled and explored using Sketch Engine (Kilgarriff et al., 2014) in two steps. Firstly, distinct imaginary layers are investigated via collocations of the topic-related keywords and analysed via concordances in each corpus. Secondly, the analysis focuses on keywords of the sub-corpus after the ChatGPT onset to reveal specifics of the “ChatGPT-moment”. The results of both steps are then compared against each other.
Although not groundbreaking in terms of methodology, such research provides innovative, empirically rooted comparative insights into the current media debate on AI. It exhibits the usefulness of the imaginary concept for the CADS while providing a clearer perspective of sociotechnical imaginaries by grounding these to objective linguistic cues. | [
"Artificial Intelligence",
"Corpus Approaches to Discourse Studies",
"Media Discourse",
"Regulation of Digital Technologies",
"Sociotechnical Imaginaries"
] | https://openreview.net/pdf?id=Bbay6S4gCl | AgPniBcfvs | decision | 1,717,402,407,677 | Bbay6S4gCl | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
AKtmbBh8ID | Practical solutions for digitally administering and scoring of a children’s speechreading test | [] | (P)
Use of visual information about speech is pronounced in situations in which auditory information is degraded because of, for example, background noise or reverberation. In speechreading (often also called lip reading), information about lip, jaw and tongue movements but also the visual cues of facial expressions are used to perceive the message of a speaker. People with hearing loss try to make use of speechreading for complementing the insufficient auditory information caused by their hearing problem, but interindividual differences are large and reliable assessment methods are needed.
Aims of the research project Gaze on lips? (https://www.oulu.fi/en/projects/gaze-lips) are twofold: to construct, standardize and validate the pre-recorded Speechreading Test for Finnish Children (SPETFIC; Huttunen & Saalasti, 2023) with automatized scoring and to find out the developmental trajectory of primary school-age children’s speechreading skills. Data are currently being compiled from 8- to 11-year-old hearing children to have the age norms for the SPETFIC.
In the current presentation, we report the practical solutions of administering SPETFIC both in-person and remotely. Remotely collected samples are managed by REDCap tools hosted at the University of Oulu, Finland. REDCap (Research Electronic Data Capture; Harris et al. 2009; 2019) is a secure, web-based software platform designed to support data collection. We implemented remote testing by utilizing screen sharing of a Zoom meeting (Zoom, 2024) so that SPETFIC is run on a test administrator’s computer. For testing the stability and speed of the Internet connection and the capabilities of screen sharing of Zoom video call, a specific frame drop estimation test was constructed. If there were issues causing excessive frame dropping, a one-time direct access link to the REDCap running SPETFIC was conveyed to the child via chat channel of the Zoom.
SPETFIC includes the automatic scoring of the results, shown both as total and section specific score on screen after finishing the test. Additionally, the item-by-item and summary results of the test can be downloaded as comma-separated (CSV) files. After the validation phase, speech and language therapists testing the children at clinics can choose to separately administer either the section A (easier words), the section B (more difficult words) or the sentence section. Having these sections enables tapping of a fairly wide skill spectrum and following up of skill improvement along the child’s maturation and intervention. Automatic scoring rules out scoring errors in both research and clinical use of the test. | [
"Finnish language",
"lipreading",
"online testing",
"speechreading",
"visual speech processing"
] | https://openreview.net/pdf?id=AKtmbBh8ID | eQFhUDCQbG | decision | 1,717,401,252,927 | AKtmbBh8ID | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author/s
Congratulations. Our reviewers have rated your paper and, provided the necessary revisions are made, we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
AE2pgvLKzu | Ambiguous grammatical forms and power relations. A statistical analysis of Latvian corpora | [] | Qualitative and quantitative analysis of Latvian laws and policy documents has identified grammatical forms that make texts ambiguous (Kruk 2021; 2024). The reason is that some grammatical forms combine the characteristics of different parts of speech. It is up to the receiver to decide whether a particular word is to be interpreted as a noun, verb, adjective, or adverb. Meanwhile, the sender reserves the possibility to reject the interpretation as wrong. Ambiguity therefore can maintain power relations since it permits the sender to bear no responsibility for the content. The paper seeks a statistical foundation for such a claim in corpora of the Latvian language. If ambiguous grammatical forms (AGF) are helpful in power relations, then they must be less frequent in those specialized corpora and subcorpora whose primary topic is not the maintenance of power relations. The frequencies of four AGFs (nominalizations, indeclinable participles, cascades of genitive and subjunctive mood) were counted in 19 specialised sub/corpora using the Sketch Engine programme. The extracted data were normalised per million words and processed in SPSS. The two-step cluster analysis identified four clusters. The lowest incidence of AGF was found in fiction, poetry, literary and children’s magazines; the highest was found in legal documents and Ph.D. dissertations. Linear regression has found a strong dependence of indeclinable participles on nominalizations. Qualitative analysis of randomly selected fragments reveals that the overuse of AGF by legislators and doctoral students weakens the illocutive force of propositions. Creating an impression of ‘serious’ discourse, the authors, in fact, conceal the sense of propositions. They can blame critical readers for their inability to find the appropriate information in the text and for the lack of adequate expertise to grasp the complex language structures of the law or academic discipline.
The paper discusses the applicability of corpus-driven approach in critical discourse analysis.
References
Kruk, S. (2020). Uzticības, sadarbības un vienotības konceptu izpratne Nacionālajā attīstības plānā 2021.–2027. gadam. Akadēmiskā Dzīve 56: 131–147.
Kruk, S. (2024). Ambiguous grammars of legal discourse. Lettonica 53. | [
"Corpus linguistics",
"corpus statistics",
"corpus driven research",
"Latvian language",
"ambiguity in language"
] | https://openreview.net/pdf?id=AE2pgvLKzu | d4QP3ZrmN7 | decision | 1,717,403,522,214 | AE2pgvLKzu | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
9Y4Obbexyj | A Dialectometric Study of Low Saxon Syntactic Variation through Time | [] | We present a corpus-based dialectometric study of synchronic and diachronic syntactic variation in literary Low Saxon, where we focus on aggregate similarity on the one hand and the occurrence of particular structures on the other. These results are then compared to our previous studies targeting other levels of representation as well as to findings from traditional dialectology. Our two major research questions for this study are: 1) Does the overall similarity of the dialect groups change over time and, if yes, how? 2) Do certain structures considered characteristic for Low Saxon decrease in frequency in the written language, as found in studies on spoken language?
The major part of our Modern Low Saxon data comes from the LSDC dataset Siewert et al. (2020) and is divided into two time periods (1800–1939 and 1980–2022) and six major dialect groups. Our findings will be compared to the Reference Corpus Middle Low German / Low Rhenish ReN-Team (2019).
In addition to our own research, recent dialectometric studies of Low Saxon have appeared by, for instance, Buurke et al. (2022) and Bartelds and Wieling (2022). A slightly older study is by Lameli (2016) who re-analysed the Wenker atlas data and found a north-south split in German Low Saxon.
In previous experiments, we have compared aggregate distances in Modern Low Saxon, Standard Dutch and Standard German at the levels of characters, PoS (Part-of-Speech) tags and morphological features from whole corpora. Here, we have found different trends at the different levels of representation. Whereas Dutch Low Saxon seems to approach Standard Dutch at all levels, the picture for German Low Saxon is more diverse: While we find a comparable trend of German Low Saxon approaching Standard German at the PoS level, when adding morphological information, the northern dialects appear to approach Standard Dutch. Furthermore, similar to Lameli, we find a north-south division in German Low Saxon to be more prominent than the traditionally assumed east-west division (compare, e.g., Schröder, 2004).
To complement our previous studies, we make use of syntactic relations and lemmata to look at structures that PoS tags do not sufficiently differentiate. In addition to the aggregate similarity, we particularly want to investigate the occurrence of structures that according to Elmentaler and Borchert (2012) are often presented as characteristic for Low Saxon in textbooks and grammar books but which they have not found to be particularly frequent in the spoken language.
(The PDF file includes footnotes, a map and figures that could not be displayed here.)
Janine Siewert, Yves Scherrer, Martijn Wieling, and Jörg Tiedemann. LSDC - a comprehensive dataset for Low Saxon Dialect Classification. In Proceedings of the 7th Workshop on NLP for Similar Languages, Varieties and Dialects, page 25–35, Barcelona, Spain (Online), December 2020. International Committee on Computational Linguistics (ICCL). URL https://www.aclweb.org/anthology/2020.vardial-1.3.
ReN-Team. Referenzkorpus Mittelniederdeutsch/Niederrheinisch (1200-1650), 2019. URL http://hdl.handle.net/11022/0000-0007-D829-8. Archived in Hamburger Zentrum für Sprachkorpora. Version 1.0. Publication date 2019-08-14.
Raoul Sergio Samuel Jan Buurke, Hedwig G Sekeres, Wilbert Heeringa, Remco Knooihuizen, and Martijn Wieling. Estimating the level and direction of aggregated sound change of dialects in the northern netherlands. Taal & Tongval, 74(2):183–214, 2022.
Martijn Bartelds and Martijn Wieling. Quantifying language variation acoustically with few resources, 2022.
Alfred Lameli. Raumstrukturen im Niederdeutschen. Eine Re-Analyse der Wenkerdaten. Niederdeutsches Jahrbuch. Jahrbuch des Vereins für niederdeutsche Sprachforschung, 139:131–152, 2016.
Ingrid Schröder. Niederdeutsch in der Gegenwart: Sprachgebiet – Grammatisches – Binnendifferenzierung. In Dieter Stellmacher, editor, Niederdeutsche Sprache und Literatur der Gegenwart, pages 35–97. Georg Olms Verlag, Hildesheim, Zürich and New York, 2004.
Michael Elmentaler and Felix Borchert. Niederdeutsche Syntax im Spannungsfeld von Kodex und Sprachpraxis. In Robert Langhanke, Kristian Berg, Michael Elmentaler, and Jörg Peters, editors, Germanistische Linguistik 220 – Niederdeutsche Syntax, page 101–135. Olms, 2012. | [
"Low Saxon",
"Low German",
"dialectometry",
"digital dialectology",
"diachronic variation",
"synchronic variation"
] | https://openreview.net/pdf?id=9Y4Obbexyj | MrEpqytaj5 | decision | 1,717,403,947,948 | 9Y4Obbexyj | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
9MlkHgoRbP | Analyzing Finnish Inflectional Classes through Discriminative Lexicon Models | [] | Descriptions of complex nominal systems make use of inflectional classes. Inflectional classes bring together nouns which have similar stem changes and use similar exponents in their paradigms. Establishing what the inflectional classes of a language are is far from trivial, and across grammatical descriptions, the number of classes distinguished can vary considerably. Although inflectional classes can be very useful for language teaching as well as for setting up finite state morphological systems, it is unclear whether inflectional classes are cognitively real, in the sense that native speakers would need to discover these classes in order to learn how to properly inflect the nouns of their language.
This study investigates whether the Discriminative Lexicon Model can understand Finnish inflected words without setting up inflectional classes, using a dataset with 55271 inflected nouns of 2000 high-frequency Finnish nouns of 49 inflectional classes. Two DLM models were constructed, one using Endstate of Learning (EOL), and the other using Frequency-Informed Learning (FIL). Both models were given the task of predicting Finnish FastText embeddings from words' forms, represented by trigram-based vectors. The models were trained on
40694 word tokens, and evaluated on 14577 held-out low-frequency tokens. Overall accuracy on test data was higher for EOL (78%) than for FIL (50%), which is to be expected since EOL represents optimal learning with infinite exposure. Importantly, for both models, accuracies increased for inflectional classes with more types, more lower-frequency words, and more hapax legomena. Importantly, the accuracy of the DLM models mirrors the productivity of the inflectional classes. The model struggles more with novel forms of unproductive and less productive classes, and performs far better for unseen forms belonging to productive classes.
These results demonstrate that the inflectional system of Finnish nouns can be learned without hand-crafting of inflectional classes. Crucially, the extent to which generalization is possible matches the productivity of the inflectional classes distinguished by linguistic analysis. We are currently investigating whether deep discriminative learning can provide even more accurate mappings, but we anticipate that with these even more powerful mappings, models will outperform what can be expected for human native speakers. | [
"Discriminative Lexicon Model",
"Finnish",
"FastText",
"Word embeddings",
"Inflectional morphology"
] | https://openreview.net/pdf?id=9MlkHgoRbP | jiqBQIsuGg | decision | 1,717,402,864,069 | 9MlkHgoRbP | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
9H3qrFUthT | How good is AI at Natural Language Understanding and Inferencing? | [] | This talk reports the results from two studies on testing how good AI is at Natural Language Understanding and Inferencing (NLU and NLI), which belong to crucial competencies of human communication. The first study tests the NLU and NLI of machines and humans in a semantic task of comparing the semantic similarity of sentences that address 20 questions about the recent covid-related pandemic crisis. We collected answers from 300 participants and asked four human annotators to annotate whether the answers semantically matched “expert” answers. Expert answers are defined as good answers suggested by medical experts. We used the embedding-based method S-BERT (Reimers & Gurevych 2020), a variant of BERT (Devlin et al. 2019), tailored to predict sentence similarity of sentences, in order to measure the semantic similarity of participants’ and experts’ answers. We then compared embedding-based semantic similarity measures with human annotations. The results from the first study show that the accuracy of the automatic method we tested shows significantly lower results than human annotations. We then used ChatGPT prompts for a similar semantic task that showed better accuracy results.
Discussion: Our question on how good machines at NLU and NLI are can be answered as follows. It depends on the machines. SBERT did not provide good accuracy results in comparison to humans in NLU and NLI tasks, but ChatGPT did. More research is needed in methods comparison in NLU and NLI. Our results of the first experiment are surprising given that BERT-based models achieve usually a good performance in NLI tasks if they are trained on NLI datasets (see Stanford NLI datasets and method comparison (https://nlp.stanford.edu/projects/snli/). We believe that this is due to the missing NLI datasets in the training of SBERT. We will discuss the importance of the datasets used for training the machines for the NLI and NLU tasks. | [
"AI",
"NLP",
"NLI",
"NLU"
] | https://openreview.net/pdf?id=9H3qrFUthT | ktjE4rIINh | decision | 1,717,402,817,159 | 9H3qrFUthT | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
9AjzBkD2rJ | Sentiment analysis for detecting suicidal youths’ positive and negative encounters with public service providers | [] | Despite a long-term downwards trend in suicide rates, Finland still experiences a higher prevalence of suicide compared to many other EU countries (Eurostat, 2024). Investigating the life courses of young people who have either attempted to or have committed suicide can yield new knowledge that can be used to influence policies and mitigate the effect of negative life trajectories or completely prevent them. Employing a broader framework of social autopsy, we examine the social and political conditions surrounding suicide incidents.
Our methodology involves 20 semi-structured interviews with young adults who have attempted suicide and family members of young persons who have committed suicide. As the data involves a wide spectrum of emotional and valuated expressions, we apply the novel method of automated sentiment analysis to detect youths’ encounters with service providers and classify them as positive, negative, and neutral. Sentiment analysis has widely been used in classifying product reviews and customer feedback, but more recently also for more varied tasks, such as detecting possible markers mental states such as depression or anxiety in social media posts (Tana, Shcherbakov & Espinosa-Leal, 2022). In our work, this method complements more traditional qualitative methods of close reading, thematic analysis and narrative analysis. Given the sensitive nature of our datasets, it is critical that machine learning models can be run locally in a secure environment. As the interviews were conducted in Finnish, we utilise the FinBERT model fine-tuned with FinnSentiment dataset. FinnSentiment consists of 27,000 polarity-annotated sentences drawn from the Suomi24 social media corpus (Lindén, Jauhiainen & Hardwick, 2023).
In this presentation, we will discuss the pilot results that demonstrate how sentiment analysis can complement the methodological toolkit of social autopsy. We will highlight the advantages and constraints of utilizing a machine learning model trained on social media corpus for the purpose of analysis of interview data. Our work is a part of the Young Despair research project that, in addition to qualitative interviews, utilize register data and official records including police reports, coroner’s autopsies, and forensic toxicology findings in studying young peoples’ suicidality. All these concerted efforts aim to enhance our understanding of the multifaceted determinants contributing to youths’ suicidal behaviors, thereby informing targeted intervention strategies and policy initiatives aimed at suicide prevention. | [
"suicide",
"young people",
"sentiment analysis",
"machine learning",
"research interviews",
"sensitive data"
] | https://openreview.net/pdf?id=9AjzBkD2rJ | skQjfdGut7 | decision | 1,717,402,839,024 | 9AjzBkD2rJ | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
7YfNd1Ru44 | Mitigating heterogeneity in the classroom? – Chatbots as support in nursing training | [] | Due to the shortage of qualified nursing staff in Germany (https://www.pflegenot-deutschland.de/ (11.0324)), attempts are being made to counteract this shortage by recruiting specialists from abroad (Böhlich et al., 2023). However, migrants already living in Germany also aspire to careers in nursing and therefore undertake nursing training, which leads to very heterogeneous classes in nursing schools. A significant challenge therefore is effectively addressing these students' heterogeneous learning backgrounds with limited resources. Chatbots have emerged as fundamental elements of innovative teaching approaches to address this issue (Kasneci, 2023). The technological advancement of Large Language Models (LLMs) makes interactions with chatbots increasingly indistinguishable from conversing with real people, contributing to their growing popularity. Chatbots could facilitate such interactions, enhance learners' motivation, personalize learning, and improve learning outcomes (Lazarides & Chevalère, 2021; Zhang & Aslan, 2021). “Language plays a special role in nursing training as communication is a nursing intervention.” (Daase/Fleiner, 2024) Researchers stress “the necessity of orienting nursing training towards the increasing linguistic heterogeneity in the group of trainees – not only due to migration.” (Daase/Fleiner, 2024)
In our study, we investigated whether students in nursing training can benefit from working with a chatbot, with a particular focus on students with German as a second language. The study involved 224 students from vocational nursing schools who created and completed medical educational tasks with and without a chatbot's support (ChatGPT, Model GPT-4). In our presentation, we will present the differences between the produced care plans, focusing on linguistic and specialized competence performed by the trainees in both sets of texts (with and without AI). Experts evaluated all two sets of answers. This allowed us to compare the quality of the answers solved by the chatbot to the solutions by the students. The evaluation focused firstly on the professional quality and secondly on the linguistic quality (morphosyntax, use of technical terms, syntax, and textuality). Statistical analyses highlighted that students' quality regarding both fields of interest was significantly higher when using the chatbot. However, they also revealed that students with advanced language skills benefited more compared to those with lower proficiency levels. The results indicate that using ChatGPT in nursing training could enhance the already existing heterogeneity in the classroom rather than mitigate it. We therefore suggest that – in order to make use of the technological advancements provided by chatbots – training in using AI is urgently required for teachers as well as students. | [
"chatbots",
"education",
"heterogeneity",
"German as a second language",
"quality",
"human computer interaction"
] | https://openreview.net/pdf?id=7YfNd1Ru44 | hPinbR2woL | decision | 1,717,402,987,906 | 7YfNd1Ru44 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author/s,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
6CKpzzFguB | Can the archives become as cool as a museum? - Data Circulation in the Budapest Time Machine | [] | Data is power and not only for present information. Users of all kinds turn to archives as a foundation nowadays, entitled to access the data and archival documents they need. The research room of Budapest City Archives (BCA) manages 3000 researchers per year. 56-58% of them are interested in architectural sources of various types, therefore this phenomenon needs to be addressed urgently.
Our online databases and digitized materials give smooth access to the citizen scientists and average users to discover stories & histories of the city and its buildings. Many of them don’t even have to visit us in person. A huge joint database offers several Hungarian cultural institutions’ materials on a very user-friendly surface. Hungaricana is not only a home for databases and digitized cultural heritage but gives us an opportunity to synchronize and unify different layers of data on the same page at the same time with the help of georeferenced historical map layers. The idea of the Time Machine is to travel back to the past just as we do nowadays with the help of Google Street View. This requires a huge amount of work, but we have taken a few steps on the road already: switching between the different time layers offered by the historical maps presents a great opportunity to discover the changes in the structure of the city. We also added four types of georeferenced archival documents, so anyone can reveal the former inhabitants or shops of the building, or to check the original drawings & some vintage postcards, photographs of the neighborhood. Many of the latter have been annotated by AI.
3D reconstructions of some historical buildings are already available on the site, so we can compare different states of the city and the building density.
Offering easy access to our materials, some of our stakeholders were able to start building their own databases and create new documents based on our collection. The recurring cooperation with the Budapest100 (celebration of 100-year-old houses with volunteer researchers) and the ÓE YBL Architecture Faculty lead us to an agreement on offering the digital copies from our materials in exchange for the newborn documents they create. We will archive the house data sheets and the 3D building models and publish them as well, which creates “data circulation”.
In the framework of our joint project, “City Memories” we are working together with two city archives, Stockholm and Copenhagen, to bring the architectural archives closer to the wider audience, sharing our experiences int Best Parctice Guides. | [
"3D modelling",
"archives",
"database",
"architecture",
"urban history",
"data circulation",
"historical maps",
"university students",
"volunteer researchers",
"new born digital documents"
] | https://openreview.net/pdf?id=6CKpzzFguB | gppNRLvrr2 | decision | 1,717,401,475,105 | 6CKpzzFguB | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author,
Congratulations. Our reviewers have rated your paper and, provided the necessary revisions are made, we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
68aI3ddh2l | Artificial Melodies: Investigating the Limits of AI in Replicating Human Songwriting | [] | The emergence of Artificial Intelligence (AI) chatbots has ignited debates regarding the potential replacement of human-generated texts, particularly structured content such as weather forecasts or financial reports. However, its application in creative domains like poetry or songwriting is increasingly acknowledged. This study aims to investigate the capacity of AI to replicate creative human writing, specifically focusing on song lyric composition in English. To achieve this goal, we conducted two Lexical Multi-Dimensional analyses (LMDA; Berber Sardinha and Fitzsimmons-Doolan, 2024), employing a curated corpus of song lyrics spanning diverse musical genres (including country, pop, rap, rock, and soul), which encompassed both chart-topping and random lyrics. Additionally, we generated a comparative corpus of artificially produced lyrics using ChatGPT, Google’s Gemini, and Meta’s Llama. The corpus consisted of 4000 lyrics, evenly split between human-authored and AI-generated texts, with each subcorpus comprising 400 lyrics per style. The first analysis involved conducting an additive LMDA based on the dimensions of variation identified by Author (2023). These dimensions were derived from a large corpus of over 100,000 song lyrics, each tagged for semantic class using the USAS semantic tagger. The dimensions are the following: 1) Materialism and Superficiality, 2) Alterity and Interpersonal Dynamics, 3) Mysticism and Transcendence, and 4) Romanticism and Personal Quest. After scoring each of our lyrics on these dimensions, we ran a Discriminant Function Analysis (DFA) to classify the lyrics as either human-authored or AI-generated. The results showed a 64% accuracy rate in identifying AI-generated songs and an 84% accuracy rate for human-authored songs. The second LMDA used the actual vocabulary of the songs, rather than semantic classes. Key lemmas were extracted for each authorship condition, which were then subjected to factorial analysis, resulting in a five-dimensional model: 1) Social Justice versus Romance, 2) Reality versus Transcendence, 3) Rural versus Urban, 4) Individualism versus Collectivism, and 5) Extroversion and Physicality versus Introversion and Emotions. This model demonstrated efficacy in classifying lyrics, achieving a 69% success rate for AI-generated lyrics and 90% for human-authored songs. Overall, the findings indicate a significant discrepancy between AI and human songwriting capabilities. Only 30.90% of AI-generated songs closely resembled those written by humans, suggesting that while AI can replicate human songwriting vocabulary, it falls short in generating discourses that align with human musical expressions. Conversely, human-authored songs were accurately identified with a high degree of precision (90.05%), underscoring the distinct and irreplicable aspects of human creativity in songwriting. | [
"Corpus Linguistics",
"Popular music",
"Multi-Dimensional Analysis",
"Artificial Intelligence"
] | https://openreview.net/pdf?id=68aI3ddh2l | E2h66DyxcA | decision | 1,717,403,800,180 | 68aI3ddh2l | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
5l5Y9Usdpf | The Atlas of Finnish Literature 1870–1940 | [] | Atlas of Finnish Literature 1870–1940
The project Atlas of Finnish Literature 1870–1940 (the Alfred Kordelin Foundation – Major Cultural Projects 2022–2024) participates in the research tradition of literary cartography. Literary cartography, which emerged in the late 1990s as part of the spatial turn in the humanities, has, in the 2000s, adopted the methods of digital humanities. Our project applies these new methods for the first time to the study of Finnish fiction.
In the project Atlas of Finnish Literature 1870–1940, we have extracted geographical information from two corpora of literary texts (Project Lönnrot and Project Gutenberg) and developed an interactive web application where you can plot the spatial references from texts on a map. The texts were transformed from plain texts to TEI/XML and then processed with named entity recognition and linking tools. In this presentation, we introduce the process we have used: NER, geocoding and linking to external data sources.
We will also introduce the Atlas of Finnish Literature 1870–1940 web application, which is open to the public, and give research examples of the kind of information that can be found with the application. Our webpage contains 846 works from 1870 to 1944 that are geocoded. The application can be used for historical research, literary studies and geography. The database can be filtered and viewed by work, by author or by location. One can see, for example, where Santeri Ivalo’s Anna Fleming (1898) is situated, what is the spatial scope in Minna Canth’s works or which books mention Helsinki. Finally, we have added full-text search to the application, which brings interesting new possibilities to the study of Finnish fiction. | [
"Literary geography",
"digital humanities",
"maps",
"Finnish literature",
"cultural history",
"named entity recognition"
] | https://openreview.net/pdf?id=5l5Y9Usdpf | WjWQN5SVRf | decision | 1,717,403,491,392 | 5l5Y9Usdpf | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author/s
We are very happy to accept your paper. As you can see from the reviews, this is ranked amongst the top papers. Please log on to www.openreview.net to see the comments given.
You may want to take the reviewers’ comments into account and revise the paper but it is not obligatory.
Please resubmit, by 16/08/2024, a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
43bbqW7Agx | Quantitative and qualitative approach to Finnish Twitter during the Covid-19 pandemic: Topics, attitudes, and emotions | [] | Quantitative and qualitative approach to Finnish Twitter during the Covid-19 pandemic: Topics, attitudes, and emotions
The ways we discuss crises affect our understanding of major events and the world in general. These kinds of discussions can even change our behaviour (e.g., Mustafa-Awad & Kirner-Ludwig, 2017); thus the study of crisis communication from a linguistic perspective is essential. During the Covid-19 pandemic, social media became an effective arena for crisis communication, and it brought together different media actors from decision-makers and healthcare professionals to ordinary citizens through communication and interaction (Spencer, 2023). However, crises are often events in which people react strongly while they try to understand the situation (Bednarek et al., 2022). On social media platforms, discussions get easily heated when different emotions, experiences, and opinions collide. In this poster presentation, we describe how the global health crisis was represented on a popular microblogging site by addressing the following research questions: 1) what kind of topics are discussed in Finnish Twitter during the Covid-19 pandemic?; and 2) what kind of attitudes and emotions are attached to these topics? To answer these questions, we utilise a large corpus of 375,322 tweets in Finnish from January 2020 to August 2021. We adopt a multidisciplinary approach to the data as we use complementary quantitative and qualitative methods that allow us both to examine the data as a vast entity and to explore the linguistic meanings in more detail. First, we use the unsupervised machine learning method of topic modelling to automatically identify topics and keywords attached to them (Blei et al., 2003). Next, we study the attitudes and emotions attached to these topics with the framework of evaluative parameters (Bednarek, 2010). Based on the results, the topic model identified 35 pandemic-related topics that cover, for example, emotions and protective measures in healthcare, briefings and news broadcasts, associations offering support services, masks, and quarantine and infection rates. The preliminary analysis of the evaluative parameters suggests that expressions of emotivity, mental state, evidentiality and style were attached to these topics.
References
Bednarek, M. (2010). Evaluation in the news. A methodological framework for analysing evaluative language in journalism. Australian Journal of Communication, 37(2), 15–50.
Bednarek, M., Ross, S.A., Boichak, O., Doran, Y. J., Carr, G., Altmann, E. G., & Alexander, T. J. (2022). Winning the discursive struggle? The impact of a significant environmental crisis event on dominant climate discourses on Twitter. Discourse, Context & Media, 45, 1–13.
Blei, D., Ng, A., & Jordan, M. (2003). Latent dirichlet allocation. Journal of Machine Learning Research, 3, 993–1022.
Mustafa-Awad, Z., & Kirner-Ludwig, M. (2017). Arab women in news headlines during the Arab Spring: Image and perception in Germany. Discourse & Communication, 11(5), 515–538.
Spencer, A. (2023). International communication. In S. M. Croucher & A. Diers-Lawson (Eds.), Pandemic communication (pp. 215–230). Routledge. | [
"Topic modelling",
"evaluative parameters",
"crisis communication"
] | https://openreview.net/pdf?id=43bbqW7Agx | SJuQHvyrFm | decision | 1,717,158,995,297 | 43bbqW7Agx | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear author,
Congratulations. This is to let you know that we are happy to accept your proposed poster.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
3mg16j76VU | Quasi-Parallel Corpora for Less-Resourced Languages: Parallelized Translations of Plato´s Faidon in Basque and Finnish | [] | As at the time Director-General of UNESCO Irina Bokova put it, “Language loss en-tails an impoverishment of humanity in countless ways. Each language – large or small – captures and organizes reality in a distinctive manner; to lose even one closes off potential discoveries about human cognition and the mind” (Bokova, Irina, 2010). The Foreword by Jordi Solé (Rehm & Way, 2023) also reflects upon languages considered not only as pure communication tools or even as vectors of culture but also as factors of identity; multilingualism is an expression of the identity of Europe.
There are 24 EU official languages, 11 additional official languages, and 54 Regional and Minority Languages (RML), protected by the European Charter for Regional or Minority Languages (ECERML) since 1992, and the Charter of Fundamental Rights of the EU. Some are Indo-European languages and some are non-Indo-European, but both share the task of defining the identity of Europe.
Out of these 90 European languages, more than half have either poor or no technological support; for instance, consider Figure 1 comparing Finnish and Spanish, and keeping in mind that Spanish has half of the technological resources of English (only European context, in both cases), and compare this against Figure 2, technological re-sources for Finnish, Basque, and Karelian languages (created ad hoc from: Atlas of the World’s Languages in Danger - UNESCO Digital Library, n.d.).
Figure: 1. Technological Factors in Spanish and Finnish
Figure: 2. Technological factors in Finnish, Basque, and Karelian languages
In the spirit of ELE, we present the first aligned Basque-Finnish corpus, both non-Indo-European languages. On the one hand, it is a finished project with the four steps for building a text-aligned corpus, and the description of the procedure can be used as a best practices manual for further prospects (Garai, 2024). On the other hand, it could be seen as a forerunner of a larger desideratum project of building a multilingual aligned corpus comprising all the European non-Indo-European languages to be used for both contrastive linguistic studies and a testbed for shared strategies and approaches to Language Technologies, given some typological convergences such as their postpositional nature or their rich morphology.
Whilst comparable corpora are made of comparable texts following some given criteria, be they from the same language or different languages, parallel refers to translations of a given text (McEnery & Xiao, 2018). Rather, here we coin the term “quasi-parallel” because one is not the direct translation of the other, but both are translations of the same omega text; in this case, Plato’s Faidon (Plato, 2006 & Plato, 1978), one translated by Calamnius and the other by Zaitegi. Using already extant translations, and parallelizing them is the cheap path we are proposing for creating linguistic technologies for less-resourced languages.
As a finished study, this work travels through all four stages of building a corpus: (a) from printed text to machine-readable, (b) the standardization of the Basque text to erase graphemic idiosyncrasies to facilitate the next two steps, (c) the alignment, and (d) the automatic annotation following the Universal Dependencies (de Marneffe et al., 2021). The access to the actual outcomes will be shortly available in a repository to be announced. | [
"Annotation Universal Dependencies",
"Less-resourced languages",
"Parallel corpora",
"Plato Finnish-Basque",
"Text alignment"
] | https://openreview.net/pdf?id=3mg16j76VU | 6QrLhpLq8Q | decision | 1,717,401,375,805 | 3mg16j76VU | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author,
Congratulations. Our reviewers have rated your paper and, provided the necessary revisions are made, we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
39h6uH70O8 | Assessing the Linguistic Characteristics of AI-Generated Texts Across Different Registers | [] | Recent studies that compare AI-generated texts to those authored by humans predominantly focus on lexical characteristics. This has resulted in a limited understanding of the ability of AI to mimic human writing from a lexicogrammatical perspective. Furthermore, this body of research often overlooks the role of register variation, neglecting to examine the degree to which AI-generated texts reflect the register-specific features found in human-authored texts. The premise of this study is that an evaluation of AI-generated text quality necessitates consideration of register. Prior corpus-based analyses of human-authored texts have convincingly shown that register significantly influences linguistic variation (Biber, 2012). Hence, for assessments aiming to determine equivalency between human-authored and AI-generated texts, register must be taken into account. It is postulated that the majority of training data for Large Language Models lacks explicit register labels, leading to a predominance of inferred over directly learned register distinctions by AI, which raises concerns about the precision and dependability of register knowledge in AI models. In this paper, we employ Multi-Dimensional (MD) Analysis (Biber, 1988, 1995; Berber Sardinha & Veirano Pinto, 2014, 2019) to assess the similarity between AI-generated and human-authored texts. This involves a detailed MD analysis of a corpus comprising register-specific texts produced by humans in natural settings and texts generated by ChatGPT 3.5. The comparison is grounded in the five principal dimensions of register variation identified by Biber (1988), which are determined by sets of co-occurring lexicogrammatical features. Both AI and human subcorpora include four distinct registers: news reports, research articles (in Chemistry and Applied Linguistics), student compositions, and conversations, with each category containing 100 texts, for a total of 800 texts (546,568 words). The human-authored subcorpus was compiled from verified sources that predate the public availability of AI to avoid any AI-generated content. The MD analysis indicated notable differences between AI-generated and human-authored texts across the individual registers and the five dimensions, with AI-generated texts generally not mirroring their human counterparts accurately. Additionally, a linear discriminant analysis, conducted to evaluate the capability of dimension scores to predict text authorship, showed that AI-generated texts could be distinguished with relative ease based on their multidimensional profiles. The findings highlight the existing challenges AI faces in replicating natural human communication effectively. The specifics of the register-based comparisons will be elaborated in the full paper. | [
"Multi-Dimensional Analysis",
"Register Variation",
"Artificial Intelligence"
] | https://openreview.net/pdf?id=39h6uH70O8 | HA9DsPpgMi | decision | 1,717,403,215,539 | 39h6uH70O8 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept (Best Paper)
comment: Dear Tony,
Congratulations. Our reviewers have rated your paper well, and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
159bIOEQ70 | Linguistic Practices and Identity Construction on Social Media: Italian Americans on Instagram | [] | Between the 19th and the 20th centuries, Italian migrants who arrived in the United States built their identity through the cultural encounter with their foster society (Michaud, 2011). Nowadays, a significant number of younger Italian Americans have shown greater interest in their heritage (De Fina, 2014), and many have turned to social media to engage with it. Since social media facilitate exchanges within ethnic communities (Dekker et al. 2015), this work aims to test the hypothesis that Italian Americans deploy linguistic practices on social media to engage with their heritage and shape their ethnic identity (Longo, 2023). In this initial phase of this qualitative research, to deal with a manageable yet representative pilot dataset, I gathered and examined 250 comments posted by Italian American Instagram users under posts shared by accounts discussing Italian American culture. To construct my dataset, starting from March 31st, 2024, I clicked on the first five accounts listed after having typed “Italian American” into the Instagram search bar. I, then, opened the ten most recent posts for each account; finally, I selected the first five comments for each post and ended up with a total of 250 comments, which were examined by means of a linguistic-pragmatic approach. At this stage of my research project, being a pilot study, I preferred focusing on methodology prioritising each step of the analysis over quantity reliance, therefore I selected and worked on 250 comments only. The analysis of the data has shown three main aspects, which will be corroborated (or not) by the analysis of a larger dataset. Firstly, all the Italian Americans whose comments were examined are native speakers of English as L1, thus showing the effects of the language shift that characterised the Italian American experience between the first and the second generation and that caused members of the latter to lose proficiency in their heritage languages (Haller, 2011). Secondly, in spite of their limited skills in either Italian or their parents’ and grandparents’ regional dialects, many Italian American Instagram users proudly defend their heritage, and some even share their families’ experiences after their arrival in the U.S., thus shaping their own identity via their interactional linguistic practices (Bucholtz & Hall, 2005) and through their ancestors’ experiences. Thirdly, some features typical of online language (Yus, 2011) are used by Italian Americans on Instagram not only to communicate with one another but also to signal allegiance to their roots.
References
Bucholtz, M., & Hall, K. (2005). Identity and interaction: A sociocultural linguistic approach. Discourse Studies, 7(4-5), 585-614. https://doi.org/10.1177/1461445605054407
De Fina, A. (2014). Language and identities in US communities of Italian origin. Forum Italicum, 48(2), 253-267. http://dx.doi.org/10.1177/0014585814529227
Dekker, R., Belabas, W., & Scholten, P. (2015). Interethnic Contact Online: Contextualising the implications of social media use by second-generation migrant youth. Journal of Intercultural Studies, 36(4), 450-467. https://doi.org/10.1080/07256868.2015.1049981
Haller, H. W. (2011). Varieties, use and attitudes of Italians in the U.S.: The dynamics of an immigrant language through time, in T. Stehl (Ed.), Sprachen in mobilisierten Kulturen: Aspekte der Migrationsinguistik (pp. 57-70). Universitätsverlag Potsdam. https://d- nb.info/1218860448/34
Longo, S. (2023). eEthnicity: Social media, Italian Americans, and Cultural Identity. Proceedings of The World Conference on Social Sciences, 2(1), 24-44. https://doi.org/10.33422/worldcss.v2i1.98
Michaud, M. C. (2011). The Italians in America, from transculturation to identity renegotiation. Diasporas, 19, 41-51. https://doi.org/10.4000/diasporas.1788
Yus, B. (2011). Cyberpragmatics. Internet-mediated communication in context. John Benjamins. https://doi.org/10.1075/pbns.213 | [
"digital ethnography",
"ethnic identity",
"Italian Americans",
"social media language",
"sociolinguistics"
] | https://openreview.net/pdf?id=159bIOEQ70 | e8TGfuMhvV | decision | 1,717,401,330,153 | 159bIOEQ70 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept with Revisions
comment: Dear author,
Congratulations. Our reviewers have rated your paper and, provided the necessary revisions are made, we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
0HmlpEh8h6 | How to identify ‘umbrella’ concepts not spoken out? Exploring German and Finnish plenary debates on 'Democracy' (1990-2020) with a TNA-based method | [] | A very common way to analyse issue-specific political debates is to use keyword-based queries to identify relevant documents. Although proven to provide a rather solid material base for analysis, when measured against known contextual attributes like temporal distribution, the representativeness of the sample is far more difficult to be evaluated. With very concrete and unambiguous concepts (e.g. climate change, debt) a standard word-based query most probably returns reliable results.
But how about ‘umbrella’ concepts, whose definition often integrates a bunch of lower-level, yet powerful and ‘loaded’ concepts and which are used as discursive frames even if not verbally mentioned? For example, someone can speak about costs and revenues, products and competition so that it becomes clear that she also talks about ‘market economy’, even if she never say ‘market economy’.
My paper will introduce a powerful, yet easy-to-use, word list (a.k.a. dictionary) based method for the identification of discursive frames from a document corpus. The method itself integrates Text Network Analysis (TNA) with both tf-idf and collocational analysis in a snowballing method to explore contextual frames. The tool allows a researcher to explore and browse her document corpus as a clustered concept collocation network, in which a cluster represents a discursive frame based on concept collocations. As the concept network can be enriched e.g. with time-related information, it allows a researcher to explore how discursive frames surface, evolve and vanish in a certain period of time. Being wholly aware of quantum leaps made in both semantic and AI based search engines, my approach defends is usefulness and relevance by integrating the power of computational methods with a corpus- and context-aware approach to textual data.
My paper will exemplify and evidence the power of the underlying method through a comparative analysis of Finnish and German plenary debates on ‘democracy’ from 1990 to 2020. | [
"Plenary debates",
"Exploratory Data Analysis",
"Data mining",
"Conceptual history",
"Text Network Analysis",
"Democracy",
"Finland",
"Germany"
] | https://openreview.net/pdf?id=0HmlpEh8h6 | 4QxAZpfRn7 | decision | 1,717,401,671,619 | 0HmlpEh8h6 | [
"everyone"
] | [
"uef.fi/University_of_Eastern_Finland/DRDHum/2024/Conference/Program_Chairs"
] | decision: Accept
comment: Dear author,
Congratulations. Our reviewers have rated your paper and we are happy to accept your talk.
Please log on to www.openreview.net to see the comments given.
Please take the reviewers’ comments into account and revise the paper accordingly before resubmission by 16/08/2024. Please resubmit a photo-ready document using the template (details for which will appearing on the https://sites.uef.fi/drd-hum-2024/call-for-papers/ page). You can already register for the conference here: Please follow the link to register: https://registration.contio.fi/uef/Registration/Login?id=7500-T_7500-8717
Please note that there will be a Pre-Conference Workshop FIN-CLARIAH tools to make sense of web data, open to all attendees, on Tuesday morning, 10:15 – 12:30. If you like to attend, please tick the relevant box on the registration form.
title: Paper Decision |
Subsets and Splits