DS10 - Défi de tous les savoirs

Effort and coordination in the production of stop consonants – StopNCo

StopNCo

What stop consonants can teach us about the control and coordination of speech gestures

Towards a functional model of stop consonant production in typical adult speakers and adults who stutter.

We often admire high-level athletes for their extraordinary performances and for precise gestural <br />coordination. We forget that we are all vocal athletes too! Speech, that we produce everyday in an <br />automatic way, actually requires a very precise coordination of breathing, laryngeal and articulatory <br />gestures, which takes some time for children to master, and which can dysfunction in some <br />speech disorders. The production of stop consonants (/p/, /b/, ...) is of particular interest for a better <br />understanding of speech control, as it requires a coordination of speech gestures not only in their <br />amplitude and force, but also in their timing. <br />The StopNCo project aims at improving our understanding of speech control by addressing three questions: <br />1. Which acoustic features are crucial for the intelligibility of stop consonants? <br />2. Which coordination of breathing, laryngeal and articulatory gestures enables the variation of these acoustic features, with what extent of physical constraints vs. speaker-specific control? <br />3. How the coordination of speech gestures can dysfunction in some speech disorders, like stuttering in particular?

Most phonetic studies on stop consonants have tried to characterize the acoustic and physiological differences between the different stop categories, produced in a «standard« manner. In this project, we are more interested in the acoustic and physiological variations within each sound category, by studying a wide variety of productions (increasing speech rate and effort, modal or whispered voice quality, clear hyper-articulated speech, percussive beatbox sounds, stuttered consonants, ...). To that goal, we recorded in laboratory conditions different types of speakers (typical, people who stutter, beatboxers) on very specific production tasks (production or imitation of logatoms) or by recreating natural face-to-face interaction conditions.
Previous studies have mainly focused on certain audible cues such as VOT (delay between the occlusion release and the onset of vocal folds vibration), occlusion duration or formant transitions, and on several physiological parameters considered separately, such as articulatory force or intra-oral pressure. Here we have focused on the variation and control of the bursts created at occlusion release. For this purpose, we first developed strain gauge force sensors and facial electromyography methods, then simultaneously acquired multiple acoustic and physiological signals (intraoral pressure, laryngeal vibration, articulatory displacements of the tongue and lips, inter-lip contact force, electromyographic activity of the lips), allowing us to study the relationship between these different parameters through multiparametric statistical analyses. In addition to physical models seeking to describe these relationships universally, we also sought, through hierarchical classification analyses, to identify certain relationships between parameters that are more subject to inter-individual variability, reflecting individual control (more or less precise, stable or efficient).

Three large multi-speaker, multi-signal databases have bee recorded during this project.
A first study explored the strategies adopted by speakers to correct stop consonants that were misperceived by their interlocutor in a face-to-face conversation. Preliminary analyses on 2 participants showed that these speakers reinforced several acoustic cues such as in particular the burst intensity and the occlusion duration, without these adaptations reinforcing contrasts between the target consonant and the misperceived one.
A doctoral thesis allowed methodological developments to measure the lip articulation force (force sensors; electromyography). It has also allowed detailed study of the control of bursts and the influence of various physiological parameters on the acoustic characteristics of these noises. Another work has identified, from endoscopic videos of the larynx, the different laryngeal structures involved in the production of stop consonants (epiglottis, ventricular bands, arytenoid cartilages) and studied their coordination as a function of vocal quality (modal vs. whispered), consonant (voicing, articulation site) and level of vocal effort.
A speech therapy dissertation and then a one-year post-doctoral fellowship explored the coordination of speech gestures in people who stutter (PWS). We observed lower inter-lip contact forces and lower levels of intra-oral pressure in these individuals, but on the contrary higher levels of electromyographic activity and more urgent glottic behaviour. Initial analyses showed that PWS were able to maintain a regular pulse even in the absence of external reference, but with greater variability than other individuals, suggesting a motor and/or temporal processing deficit that is not limited to speech.

The StopNCo project is a fundamental research project in phonetics, with applications in speech and language therapy, coordinated by Maëva Garnier and carried out within the GIPSA-lab in Grenoble. The project started in February 2015 and lasted 48 months. It benefited from an ANR grant of 214 k€.

• Slis, A., Garnier, M., DaFonseca, A., Savariaux, C. (2019) « Glottal Characteristics of People who Stutter and the Interactions with Syllable Complexity », Actes de ICPhS, Melbourne.
• Garnier, M., Da Fonseca, A., Savariaux, C. et Cattelain, T. (2018) «Efforts de production de parole chez les personnes qui bégaient«, Actes des Journées d’Etude sur la Parole, Aix en Provence, France.
• Garnier, M., Dohen, M., Buttiaux, L. et Gerber, S. (2018) «Clarification et correction d’indices segmentaux : une étude pilote sur les consonnes occlusives du français«, Actes des Journées d’Etude sur la Parole, Aix en Provence, France.
• Cattelain, T., Garnier, M., Gerber, S. et Perrier, P. (2018) «Analyse électromyographique de la production des plosives labiales : enjeux méthodologiques«, Actes des Journées d’Etude sur la Parole, Aix en Provence, France
• Garnier, M., Slis, A., Savariaux, C., DaFonseca, A. (2020) « Physiological continuum of atypicality between fluent and stuttered stop consonants», International Seminar on Speech Production (ISSP), Providence, USA.
• Slis, A., Savariaux, C., Gerber, S., Perrier, P., Garnier, M. (2020) « Complexity of rhythmic tapping task and stuttering”, International Seminar on Speech Production (ISSP), Providence, USA.
• Cattelain, T. Garnier, M., Savariaux, C., Perrier, P. (2017) «Time activation patterns of orofacial muscles in labial stop consonants«, International Seminar on Speech Production, Tianjin, China.
• Cattelain, T. Garnier, M., Savariaux, C., Perrier, P. (2017) «Muscle activation patterns in labial stop consonants«, Speech Motor Control, Groningen, The Netherlands.


We often admire high-level athletes for their extraordinary performances and for the precise coordination of their gestures. We forget that we are all vocal athletes too! Speech, that we produce everyday in an automatic way, actually requires a very precise coordination of breathing, laryngeal and articulatory gestures, which takes some time for children to master, and which can dysfunction in the case of voice or speech disorders. The production of stop consonants (/p/, /b/, ...) is of particular interest for a better understanding of speech control, as it requires a coordination of speech gestures not only in their amplitude and force, but also in their timing. Children with articulation disorders very often have troubles with this category of sounds, while dysphonic patients demonstrate excessive tensions and efforts.
The project StopNCo aims at improving our understanding of speech control by addressing 4 questions:

1. Which acoustic features are crucial for the intelligibility of stop consonants?
Instead of the traditional laboratory approach that examines the perceptual consequence of varying features in synthetic stimuli, we will here characterize how speakers enhance their speech in interactive situations requiring intelligibility, but where acoustic cues are altered or missing (speech produced in a noisy or reverberant environment, whispered speech, …).

2. Which coordination of breathing, laryngeal and articulatory gestures enables the variation of these acoustic features, with what extent of physical constraints vs. speaker-specific control?
We will collect simultaneously acoustic, aerodynamic (intra-oral pressure, airflow), laryngeal (electroglottography, endoscopy) and articulatory data (movement, force sensors, surface EMG). We will explore how the coordination of breathing, laryngeal and articulatory gestures varies with the speaking mode (murmured to shouted, whispered, fast, clear speech) and using speech perturbation paradigms (filtered auditory feedback, perturbed articulation, oral anesthetic, …).

3. How does this control develop normally in children and dysfunction in some of them?
We will characterize how the acoustic cues to stop consonants are refined with child age, or remain deviant in children with functional articulation disorders. Using non invasive methodologies, we will identify some aspects of speech coordination that differ from adult speech, and that some children with articulation disorders have troubles to develop.

4. How the coordination of speech gestures can vary in efficiency?
This will first require the development of methodologies to measure or estimate laryngeal and articulatory efforts. Production efforts will be compared between non pathological adult speakers and dysphonic patients, hypothesized to coordinate less efficiently their breathing, laryngeal and articulatory gestures.

A functional functional model of speech coordination in the production of stop consonants will be built, taking into account the different physiological levels and perceptual outcomes, the intra and inter-speaker.

The financial help requested mainly aims at recruiting a Ph.D student and a one-year post-doctoral researcher, so as to help me build a team around the questions of speech production efforts and stop consonants production that are not very developed in the GIPSA-lab yet. The consortium will be composed of a limited number of people, mainly from GIPSA-lab but also from the university of Lyon 1 and from the medical field, most of them beeing young researchers. Each collaborator will bring a specific expertise – in speech physiology and pathology, in biomechanics and in speech development in children – complementary to mine in vocal effort, speech adaptation and face to face interaction.

The project will bring fundamental knowledge on speech efforts and coordination, as well as new measurement methodologies and indices for the diagnosis and the rehabilitation of speech disorders.

Project coordinator

Madame Maëva GARNIER (Grenoble Images Parole Signal Automatique)

The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.

Partner

GIPSA-Lab Grenoble Images Parole Signal Automatique

Help of the ANR 214,257 euros
Beginning and duration of the scientific project: September 2014 - 48 Months

Useful links

Explorez notre base de projets financés

 

 

ANR makes available its datasets on funded projects, click here to find more.

Sign up for the latest news:
Subscribe to our newsletter