Language studies
Nathalie Henrich Bernardoni, Univ. Grenoble Alpes, CNRS, Grenoble INP, GIPSA-lab, 38000 Grenoble, France
Have you ever wondered where the sounds coming out of your mouth come from and how they are produced? Why do some people sing in a very sonorous way, and others with a lot of breath in their voice? Why, when listening to a person's voice on the phone, are you able to recognise them? And when you do not know the voice, why are you able to make so many inferences on the basis of that voice alone – its gender, age, emotional state, or why not its linguistic origins? These questions have personally challenged me and led me to work in language sciences as a multidisciplinary scientist. As a physicist by training, my research subject, the human voice in all its forms of expression, has led me to develop skills in many disciplinary fields, such as musical acoustics, acoustic and clinical phonetics, physiology, signal and image processing, musical perception and ethnomusicology.
We are thinking beings. We apprehend the world around us in the light of our perceptions. As a scientist, I wanted to go beyond simple observation, to explain and reproduce sensitive phenomena. Modelling is our gateway to understanding the physical phenomena that surround us, and translating them into a language that makes sense to us, that is transmissible and evolutive. This is a very common approach in physics, a discipline in which I am well versed or qualified. Modelling a phenomenon involves developing or using concepts, theories and tools to describe, understand and predict a real-world phenomenon.
A model is, by definition, an abstraction of the physical world, limited by the theoretical contours on which it is based, the hypothesis from which it extends, the observable factors by which it is fed. It is likely to evolve, complexify itself, or sometimes be reformulated, rethought in the light of new knowledge, new experimental tools, new mathematical approaches.
Modelling a phenomenon means constantly going back and forth between what we can observe and what we predict. In physics as in language or life sciences, modelling is fed by data – experimental data, linguistic data.
In speech sciences, there are different approaches to modelling, depending on the purpose of the model and its use. If we wish to reproduce the phenomenon in the simplest, most manageable way – for the purpose of real-time synthesis, as an example – we will go towards mathematical or statistical modelling. If we wish to understand the physics of the phenomenon and reproduce its behavior, we will go towards physical modelling and numerical simulation based on physics equations.
Modelling human voice production allows us to better understand it. It can lead to an adjustment, even a complete change in vocal practice. While some vocal artists are not curious about how their instrument works, others will use this knowledge to improve their vocal performance.
Cite this item: Nathalie Henrich Bernardoni, “Modelling”, translated by Lauren Fabrizio, Performascope: Interdisciplinary Lexicon of Performance and Research-Creation, Grenoble: Université Grenoble Alpes, 2021, [online]: http://performascope.univ-grenoble-alpes.fr/en/detail/177853
