Grammatical gender associations outweigh topical gender bias in crosslinguistic word embeddings

By Kate McCurdy, Oguz Serbetci. 2020

22 citations recorded by the source corpus.

Recent research has demonstrated that vector space models of semantics can reflect undesirable biases in human culture. Our investigation of crosslinguistic word embeddings reveals that topical gender bias interacts with, and is surpassed in magnitude by, the effect of grammatical gender associations, and both may be attenuated by corpus lemmatization. This finding has implications for downstream applications such as machine translation.

Subfields: Artificial Intelligence, Cultural Studies

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

1. Goran Glavas
2. Ivan Vulic
3. Anne Lauscher
4. Simone Paolo Ponzetto
5. Soumya Barikeri
6. Adam Lopez
7. Rafik Takieddin
8. Seraphina Goldfarb-Tarrant

In Cultural Studies

1. Ivan Vulic
2. Soumya Barikeri
3. Goran Glavas
4. Anne Lauscher

Citations in the corpus

• RedditBias: A Real-World Resource for Bias Evaluation and Debiasing of Conversational Language Models (2021, methodology): “Subsequent work proposed extensions to further embedding models (Liang et al., 2020a,b) and languages (e.g., McCurdy and Serbetci, 2020; Lauscher and Glavaˇs, 2019; Lauscher et al., 2020b; May et al., 2019), analyses of the proposed measures (e.g., Gonen and Goldberg, 2019; Ethayarajh et al.,…”
• Intrinsic Bias Metrics Do Not Correlate with Application Bias (2020, background): “It is important to take a language with pervasive gender-marking (Spanish) into account, as previous work has shown that grammatical gender-marking has a strong effect on gender bias (McCurdy and Serbetci, 2017; Gonen et al., 2019; Zhou et al., 2019).”
• AraWEAT: Multidimensional Analysis of Biases in Arabic Word Embeddings (2020, background): “g Association Test (WEAT), measures associations between words in an embedding space in terms of cosine similarity between the vectors. They propose 10stimuli sets, which we adapt in our work. Later, McCurdy and Serbetci (2017) extend the analysis to three more languages, Dutch, German, and Spanish,”
• Theories of “Gender” in NLP Bias Research (2022, unknown): “[94] Katherine McCurdy and Oguz Serbetci.”