Ekaterina Vylomova

Ekaterina Vylomova

Auto-generated profile, based on publicly available information

Ekaterina Vylomova

The University of Melbourne, Melbourne, VIC, Australia

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2020 in European Review of Social Psychology
Verified
Journal Article
Closed access
NHNick Haslam
Nick Haslam
BDBrodie C. Dakin
Brodie C. Dakin
FFFabian Fabiano
Fabian Fabiano
MMMelanie J. McGrath
Melanie J. McGrath
JRJoshua Rhee
Joshua Rhee
MWMelissa A. Wheeler
Melissa A. Wheeler
“Concept creep” is the gradual semantic expansion of harm-related concepts such as bullying, mental disorder, prejudice, and trauma. This review presents a synopsis of relevant theoretical advances and empirical research findings on the phenomenon. It addresses three fundamental questions. First, it clarifies the characterisation of concept creep by refining its theoretical and historical dimensions and presenting studies investigating the change in harm-related concepts using computational linguistics. Second, it examines factors that have caused concept creep, including cultural shifts in sensitivity to harm, societal changes in the prevalence of harm, and intentional meaning changes engineered for political ends. Third, the paper develops an account of the consequences of concept creep, including social conflict, political polarisation, speech restrictions, victim identities, and progressive social change. This extended analysis of concept creep helps to understand its mixed implications and sets a multi-pronged agenda for future research on the topic.

29 June 2023 in PLOS ONE
Verified
Journal Article
Open access
PDF file
YXYu Xiao
Yu Xiao
NBNaomi Baes
Naomi Baes
EVEkaterina Vylomova
Ekaterina Vylomova
NHNick Haslam
Nick Haslam
Research on concept creep indicates that the meanings of some psychological concepts have broadened in recent decades. Some mental health-related concepts such as ‘trauma’, for example, have acquired more expansive meanings and come to refer to a wider range of events and experiences. ‘Anxiety’ and ‘depression’ may have undergone similar semantic inflation, driven by rising public attention and awareness. Critics have argued that everyday emotional experiences are increasingly pathologized, so that ‘depression’ and ‘anxiety’ have broadened to include sub-clinical experiences of sadness and worry. The possibility that these concepts have expanded to include less severe phenomena (vertical concept creep) was tested by examining changes in the emotional intensity of words in their vicinity (collocates) using two large historical text corpora, one academic and one general. The academic corpus contained >133 million words from psychology article abstracts published 1970–2018, and the general corpus (>500 million words) consisted of diverse text sources from the USA for the same period. We hypothesized that collocates of ‘anxiety’ and ‘depression’ would decline in average emotional severity over the study period. Contrary to prediction, the average severity of collocates for both words increased in both corpora, possibly due to growing clinical framing of the two concepts. The study findings therefore do not support a historical decline in the severity of ‘anxiety’ and ‘depression’ but do provide evidence for a rise in their pathologization.

1 January 2021
Verified
Conference Paper
Open access
PDF file
ESElizabeth Salesky
Elizabeth Salesky
BABadr Abdullah
Badr Abdullah
SMSabrina Mielke
Sabrina Mielke
EKElena Klyachko
Elena Klyachko
OSOleg Serikov
Oleg Serikov
EVEkaterina Vylomova
Ekaterina Vylomova
Elizabeth Salesky, Badr M. Abdullah, Sabrina Mielke, Elena Klyachko, Oleg Serikov, Edoardo Maria Ponti, Ritesh Kumar, Ryan Cotterell, Ekaterina Vylomova. Proceedings of the Third Workshop on Computational Typology and Multilingual NLP. 2021.

10 April 2025 in Proceedings of the National Academy of Sciences
Verified
Journal Article
Open access
TKTemuulen Khishigsuren
Temuulen Khishigsuren
TRTerry Regier
Terry Regier
EVEkaterina Vylomova
Ekaterina Vylomova
CKCharles Kemp
Charles Kemp
Claims about lexical elaboration (e.g. Mongolian has many horse-related terms) are widespread in the scholarly and popular literature. Here, we show that computational analyses of bilingual dictionaries can be used to test claims about lexical elaboration at scale. We validate our approach by introducing BILA, a dataset including 1,574 bilingual dictionaries, and showing that it confirms 147 out of 163 previous claims from the literature. We then identify previously unreported examples of lexical elaboration, and analyze how lexical elaboration is influenced by ecological and cultural variables. Claims about lexical elaboration are sometimes dismissed as either obvious or fanciful, but our work suggests that large-scale computational approaches to the topic can produce nonobvious and well-grounded insights into language and culture.

15 June 2022 in arXiv
Verified
Preprint
Open access
PDF file
KBKhuyagbaatar Batsuren
Khuyagbaatar Batsuren
GBGábor Bella
Gábor Bella
AAAryaman Arora
Aryaman Arora
VMViktor Martinović
Viktor Martinović
KGKyle Gorman
Kyle Gorman
EVEkaterina Vylomova
Ekaterina Vylomova
The SIGMORPHON 2022 shared task on morpheme segmentation challenged systems to decompose a word into a sequence of morphemes and covered most types of morphology: compounds, derivations, and inflections. Subtask 1, word-level morpheme segmentation, covered 5 million words in 9 languages (Czech, English, Spanish, Hungarian, French, Italian, Russian, Latin, Mongolian) and received 13 system submissions from 7 teams and the best system averaged 97.29% F1 score across all languages, ranging English (93.84%) to Latin (99.38%). Subtask 2, sentence-level morpheme segmentation, covered 18,735 sentences in 3 languages (Czech, English, Mongolian), received 10 system submissions from 3 teams, and the best systems outperformed all three state-of-the-art subword tokenization methods (BPE, ULM, Morfessor2) by 30.71% absolute. To facilitate error analysis and support any type of future studies, we released all system predictions, the evaluation script, and all gold standard datasets.

19 July 2026
Verified
Conference Paper
Open access
RMRaphael Merx
Raphael Merx
EVEkaterina Vylomova
Ekaterina Vylomova
Machine translation for low-resource languages suffers from domain-imbalanced corpora, causing quality degradation on technical text. However, in-context learning opens the possibility to rely on limited in-domain corpora to inform translation. We present lessons learned from Tulun, a retrieval-augmented system combining neural MT with LLM post-editing, guided by user-configurable translation memories and glossaries. Deployed for medical translation in Timor-Leste (Tetun) and disaster relief translation in Vanuatu (Bislama), the system achieves accuracy improvements over baseline MT by 16.90-22.41 ChrF++ points, while offering rapid adaptability and transparency to end-users. Key recommendations include: domain granularity matters more than broad categories; translation target audience should inform retrieval; and RAG-augmented MT is most effective for languages that lack domain corpora but remain within LLM pretraining distributions.