LINGUIST List 32.261
Tue Jan 19 2021
Internships: Computational Linguistics; General Linguistics; Language Acquisition; Text/Corpus Linguistics; Translation; Call for 6 internships on Gender bias and Automatic Translation, Machine Translation Research Unit, FBK
Editor for this issue: Becca Morris <beccalinguistlist.org>
Beatrice Savoldi <beatrice.savoldi
Computational Linguistics; General Linguistics; Language Acquisition; Text/Corpus Linguistics; Translation; Call for 6 internships on Gender bias and Automatic Translation, Machine Translation Research Unit, FBK, Trento, Italy E-mail this message to a friend
University or Organization: Fondazione Bruno Kessler, Trento, Italy
Department: Machine Translation
Web Address: https://ict.fbk.eu/units/hlt-mt/
Linguistic Field(s): Computational Linguistics
Minimum Education Level: No Minimum
The Machine Translation (MT) group at Fondazione Bruno Kessler (Trento, Italy) is looking for proactive and dynamic young students, with good organizational, relational skills and an inclination for team-work, strongly motivated to work on innovative and interdisciplinary projects at the intersection of language and translation technology (and which are frequently addressed to third party companies).
Title: Gender Bias and Automatic Translation
Language technology has provided us with beneficial breakthroughs like predictive texts, automatic speech transcription and spam filtering. However, in the last years, these systems displayed unfair behaviors: recruiting tools dismissed women’s curricula and chatbots spitted out sexist slurs. In other words, such technologies can learn and reproduce existing social asymmetries, gender bias included.
These kinds of biases affect Automatic translation systems as well. In particular, this problem becomes evident when translating from English into languages with grammatical gender like Italian, Spanish or French. In fact, it has been shown that such systems mirror stereotypical gender roles - for instance systematically translating the English "doctor" as "dottore" - and that they suffer from a generic masculine default (e.g. I’ve been there → sono stato) although it is a woman talking. As a result, translation models encounter a performance loss whether they should generate feminine forms, but most importantly, they learn and amplify the under- and mis-representation of a demographic group.
We can offer an important training opportunity to deepen and experiment in the field of Gender Bias in Automatic Translation with the scope of advancing our understanding of the problem and the dynamics leading to its emergence.
In particular, candidates will carry out manual analysis of the (biased) outputs of translation models. Strong focus will be placed on contrastive variation across different languages, with qualitative and quantitative interpretations of the analysis outcomes.
- Ongoing academic background in Linguistics/Translation Studies/Computational Linguistics
- Good knowledge of written and spoken English
- Excellent knowledge of one or more of the following languages: French, Italian, Spanish.
- (Optional) Computational skills for textual analysis.
Motivated students are invited to submit their applications by 15.02.2021 through the form available at this link (https://forms.gle/xk8MzF1K3jMt5Bpx7
). The internship starting date is mid-March 2021 with a flexible duration of 1-3 months. Candidates will be asked to specify their preferred starting date and duration in the application form. Due to the current health situation, the internship will be done remotely. A minimum of two regular weekly meetings is planned: depending on the candidate’s needs and preference, such meetings can take place both in presence or remotely.
More information about the Machine Translation Research Unit is available here (https://ict.fbk.eu/units/hlt-mt/
Contacts: Luisa Bentivogli (FBK) and Beatrice Savoldi (FBK/UNITN)
Application Deadline: 15-Feb-2021
Web Address for Applications: https://forms.gle/xk8MzF1K3jMt5Bpx7
Page Updated: 19-Jan-2021