ISMA-Vn

BACKGROUND

Nasalance and nasalization processes in languages are phenomena studied in various fields such as signal processing, acoustics, phonetics, phonology, dialectology, or historical linguistics. However, these disciplines do not always intersect, although specialists agree that nasalization is a dynamic and complex phenomenon whose linguistic, physical, and physiological mechanisms are still poorly understood. The acoustic structure of nasal vowels and consonants complicates the search for perceptual correlates of nasality, despite its importance in phonology: it is present in 97% of the world's consonant inventories and more than one in five languages distinguishes oral and nasal vowels.

Acoustic studies of nasal sounds are of major scientific interest. Current models involve acoustic coupling between the vocal and nasal tracts, generating a complex spectrum composed of poles and zeros. This coupling depends on both the shape of the tracts and the position of the velum, which can articulatorily connect the nasal and oral cavities without necessarily leaving an identifiable trace in the acoustic signal. Furthermore, the amplitude of the velum gesture is greater for nasal vowels than for nasal consonants in French. Despite these models, the complexity of the spectra produced makes it difficult to precisely explain the acoustic structure of nasal productions. Developing a relevant method to characterize these spectra is essential to better understand the emergence, realization, evolution, perception, and categorization of nasal vowels.

In a previous study, Feng and Kotenkoff (2006) used a device that separately isolated the mouth and nose outputs during the production of nasal sounds. This device showed that the oral output remained primarily "oral" even during nasal production, evolving with the lowering of the velum, while the nasal output was characterized by the physical properties of the pharyngonasal tract despite the coupling of the two tracts.

GOING FURTHER WITH THE SUPPORT OF LabEx CerCoG

The ISMA-Vn project builds on this context by aiming to use an improved version of Feng's device to record the nose and mouth outputs separately during nasal productions. A corpus of recordings will be collected from ten French-speaking speakers in a laboratory under controlled conditions. This data will make it possible to compare acoustic models with actual recordings by contrasting the modeled transfer function of the mouth output with the recording of the mouth output alone, and similarly for the nose output.

This approach will validate an acoustic model of nasal productions and shed light on the complex concept of acoustic coupling, which is still widely debated in the scientific community. In addition, a perceptual study involving about twenty native speakers will be conducted to analyze the phonological categorization of the separated nose and mouth signals, thereby validating the acoustico-perceptual correlates of nasal productions.

The support of LabEx CerCoG will be crucial in successfully carrying out this innovative project, which aims to advance the understanding of nasality phenomena through a rigorous methodology combining modeling, experimental recording, and perceptual analysis.