Published on 15 July 2019
L’interdisciplinarité en SHS : faire germer les recherches de demain

At Université Paris-Saclay, many of the projects selected for development are in study areas conducive to interdisciplinarity in HSS. Often, the latter deal with environmental issues, the management of large databases or innovations.

The recent “Gilets Jaunes” citizens’ protest movement in France and the ensuing largescale national debate remind us that the humanities and social sciences (HSS) are necessary and vital in the construction of an enlightened, participative and inclusive society. “At Université Paris-Saclay, the HSS Department is home to about fifteen disciplines that fall into four main categories – namely, economics and management; law; sociology and political science; and the humanities and heritage sciences – as well as 36 research laboratories and about a thousand researchers,” reports Jean-Paul Markus, head of this Department at Université Paris-Saclay.

Interdisciplinarity enables each discipline to situate its particular study area within a broader context. “Today, society’s global problems are very complex and it’s increasingly improbable that any single scientific discipline will be able to devise solutions for them,” comments André Torre, director of the Paris-Saclay Humanities and Social Sciences Center (MSH Paris-Saclay). Created in 2015 to foster cross-pollination, MSH Paris-Saclay (See Focus opposite) “is specialized in research engineering for the purpose of facilitating collaboration between researchers,” says its director. It funds the organization of scientific events and the development of research projects in three categories: “environment, spatial planning, health”; “digital and humanities sciences”; and “transition and innovation”. “They often involve some of the most innovative interdisciplinary research.”

Massive amounts of data and the automatic exploration of language resources

One of these innovative projects is HistorIA.2, championed by linguist Ioana Vasilescu of the Computer Science Laboratory for Mechanics and Engineering Sciences, or LIMSI (CNRS). HistorIA.2 draws on conventional linguistics as well as artificial intelligence to study the evolution of languages, using automatic speech processing systems that can explore massive amounts of spoken language data. Assisted by fellow linguists and a LIMSI automatic processing specialist, Ioana Vasilescu is studying sound changes occurring over time in certain Romance languages (Spanish, Italian and Romanian) in particular regions.

With regard to the Spanish language, she is especially interested in the lenition of the consonants “b”, “d” and “g” placed between two vowels and the consonant “s” at the end of a word. “For instance, the word abogado, the Spanish word for lawyer, has come to be pronounced ‘aoao’,” says Ioana Vasilescu. “We wanted to know the extent to which Spanish was affected by this phenomenon.” By “hijacking” LIMSI’s automatic processing systems for her own purposes, she is able to quantify the sound variations detected in several hundred hours’ worth of recorded radio and television news programs. “In order to do this, we have to force the system to leave transcription errors that it would normally correct,” she explains.

Biodiversity and land use planning: fraternal adversaries?

Another MSH undertaking, IDÉES-BIO, deals with the project impact of environmental assessment in regional planning policies, mobilizing specialists in ecology, law, political science, geography and sociology. It bears on the planned metro line (Line 18) between Orly Airport and the Versailles Chantiers rail station and the consequences for the flora and fauna in the Saclay plateau. “With hindsight, we realized that the alignment coincided exactly with the movement paths of certain animal species,” observed team member Nathalie Frascaria-Lacoste, professor and researcher in ecology at the Ecology, Systematics and Evolution Laboratory (ESE – AgroParisTech/CNRS/Université Paris-Sud). “Today, impact assessments only go partway in identifying the biodiversity of a given territory.” In addition to analyzing eco-indicators, the team also seeks to understand the regulatory environment and develop new work methodologies. “In cases where environmental impairment cannot be avoided or reduced, planners find themselves constrained to take compensatory measures, which are often restrictive and very difficult to implement. Our purpose is to find a way to do things better, all together,” says Nathalie Frascaria- Lacoste. The goal is to work in concert with land use planners and, in three years, propose guidelines to help them more effectively reconcile land use planning and economic development with the protection of the environment, knowing that environmental assessment is a complex matter.

For non-discriminating intelligent algorithms

At the RITM economics and management laboratory (Université Paris-Sud), specialized in the study of networks, spatial issues and globalization, economists Serge Pajak, Matthieu Manant and a legal expert have tackled a project to evaluate the economic origin of discrimination bias generated by the intelligent algorithms used by social media or matchmaking platforms. “We were inspired by previous research showing that, on Facebook, the allocation of ad space by means of an auction mechanism and the difference in marketing advantage inherent to targeting a male or female cause disparities in display depending on the gender of the user. It is harder to ‘win’ an auction to display an ad to a woman user,” points out Serge Pajak. “Advertisers think they are not discriminating, but the behavior of the algorithm yields a result that is not neutral.”

In the summer of 2018, the team set out to learn more about this behavior. They supervised ad campaigns for an engineering school on Facebook and Snapchat. “We wanted to see if ad distribution varied according to type (feminine, masculine or neutral) or content,” said Matthieu Manant. The results, still being analyzed, seem to indicate that, on Snapchat, distribution in provincial France is similar to that of Paris, “as if the algorithm had learned from its experience in Paris how to distribute ads in cities for which it lacked data.”



∙ Vasilescu, I., et al. Exploring Temporal Reduction in Dialectal Spanish : A Large-scale Study of Lenition of Voiced Stops and Coda-s. INTERSPEECH 2018, 2728-2732.

∙ Colloque Gaié MSH 2018 « À quoi sert l’évaluation environnementale ? Pratiques, Enjeux et Perspectives » du 10 décembre 2018 :

∙ Manant M., et al. Can social media lead to labor market discrimination? Evidence from a field experiment. Journal of Economics & Management Strategy. 2018 ; 1-22.


Portrait : André Torre

“Now we work with nearly all of the teams in humanities and social sciences at Université Paris-Saclay.”

André TorreAn economist by training, André Torre is the director of the MSH Paris-Saclay, which fosters and supports research in humanities and social sciences. He is also a research director at INRA, involved in AgroParisTech, and is the director of research programs “for and about regional development”. His research focuses on analyzing proximity relations and their importance in processes of coordination between people. He is also President of the European Regional Science Association (ERSA) and managing editor of the journal Revue d’économie régionale et urbaine (RERU).


By Véronique Meder.

The original version of this article was published in L'Edition #10.