Challenges and Strategies in Cross-Cultural NLP
By Daniel Hershcovich, Stella Frank, Heather Lent, Miryam de Lhoneux, Mostafa Abdou, Stephanie Brandl, Emanuele Bugliarello, L. Piqueras, Ilias Chalkidis, Ruixiang Cui and 4 more. 2022
290 citations recorded by the source corpus.
Various efforts in the Natural Language Processing (NLP) community have been made to accommodate linguistic diversity and serve speakers of many different languages. However, it is important to acknowledge that speakers and the content they produce and require, vary not just by language, but also by culture. Although language and culture are tightly linked, there are important differences. Analogous to cross-lingual and multilingual NLP, cross-cultural and multicultural NLP considers these differences in order to better serve users of NLP systems. We propose a principled framework to frame these efforts, and survey existing and potential strategies.
Subfields: Artificial Intelligence, Natural Language Processing, Social Sciences & Humanities
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Vinodkumar Prabhakaran
- Ben Hutchinson
- Björn Ross
- Simone Conia
- Iason Gabriel
- Po-Sen Huang
- Jonathan Uesato
- William S. Isaac
In Natural Language Processing
- Björn Ross
- Simone Conia
- Ben Hutchinson
- Iason Gabriel
- John F. J. Mellor
- Roberto Navigli
- Johannes Welbl
- Geoffrey Irving
In Social Sciences & Humanities
Citations in the corpus
- Characteristics of Harmful Text: Towards Rigorous Benchmarking of Language Models (2022, background): “We note all benchmarks in Table 1 are designed to measure harms in English, indicating a lack of linguistic and cultural diversity that is well documented across other language tasks [51, 9, 10, 23].”
- Cultural Incongruencies in Artificial Intelligence (2022, background): “…interpretive tasks are inherently shaped by the culture within which they are embedded in, including not only the cultural-linguistic dependencies [16, 17] of tasks such as inferring emotion, sentiment, offensiveness, but also image and symbol interpretation—including gestures, facial…”
- Biases in Large Language Models: Origins, Inventory, and Discussion (2023, background): “Like with other biases, because model performance is often optimized for majority groups, cultural bias tends to afect minority classes, which is why it is recommended to update models for diferent groups and/or according to varied demographic attributes [63].”