What happens during stuttering-like disfluencies ? An EMA study

Stuttering is a communication disorder characterized by frequent and severe disruptions in the flow of speech. Several studies aimed to classify these stuttering-like disfluencies (henceforth SLD), mostly basing this classification on an acoustic or a perceptive analysis. Generally, researches dealing with stuttered speech noted three main types of disfluencies: blocks, sound/syllable prolongations or repetitions (e.g. Van Riper, 1982). However, studies which try to describe disfluencies by using physiological data are less frequent. Whereas several researches reveal laryngeal dysfunctions in speech produced by Persons Who Stutter (henceforth PWS) (Monfrais-Pfauwadel et al., 2005), data allowing to observe articulatory mechanisms explaining the emergence of disfluencies are rather rare. Moreover, most studies using articulatory data (electromagnetic articulometry, radiography etc.) have essentially described speech production in perceptually fluent speech in PWS (Zimmermann, 1980; McClean et al., 1995). Consequently, the aim of this study is to provide a description of articulatory behaviour during SLDs. More precisely, our objective is to analyse articulatory movements that occur during SLDs and propose a classification based on supraglottic articulatory gestures. Our hypothesis is that the nomenclature generally used to describe disfluencies does not reflect the articulatory behaviour. If a same percept can be a result of different articulatory gestures (Stevens et al., 2008), depending on speaker, phonetic environment etc., we postulate that the same articulatory patterns could be at the origin of several perceptual types of disfluencies (Didirková, 2016). Moreover, several articulatory configurations should result in the same perceptual kind of disfluency. Methodology Data acquisition & participants EMA data were collected by means of an electromagnetic articulograph Carstens AG501 3D at the Lorraine Research Laboratory in Computer Science and its Applications (LORIA, Nancy, France) with a sampling rate of 250 Hz and an accuracy of 0.3 mm. All data were stocked in a .pos file and synchronized with a sound recording (44,1 kHz, 16 bits, .wav). 10 sensors (2x3 mm) per subject were used: two were fixed on the lips of each subject (1 in the middle of the upper lip and another one in the middle of the lower lip). Three coils were situated on the tongue of each subject; one on the tongue tip, one on the tongue body and one on the tongue back. To track the mandible’s movements, another sensor was placed on the subjects’ jaw. The palate’s form was indicated by means of a tennth coil. Other sensors were used to control head’s movements. Two Adults Who Stutter (henceforth AWS) aged 25 and two control subjects (28-years-old) were recruited for this study. All subjects were matched according to gender, age and socio-professional category. They all were from Senegal and were bilingual french-wolof. Participants were recorded while reading the text of an Alphonse Daudet’s novel, La chèvre de Monsieur Seguin (Mister Seguin’s goat), in French. Recordings took place in a soundproof room. Data analysis All SLDs were identified in the production of PWS. A .wav file was extracted for each SLD in order to carry out a perception test with the freeware Perceval (Ghio et al., 2003) where listeners had to categorize SLDs as blocks, repetitions, prolongations or combined disfluencies. After this classification, the Visartico software (Ouni et al., 2012) was used to visualize and analyze the vertical movements (the z axis) of the upper and lower lip, the tongue tip, the tongue body and the mandible in segments that included the stuttered phone and its preceding and subsequent syllables. A classification of the SLDs, based on observations which have been made is proposed as result. Results Three categories of disfluencies have been revealed by EMA data: 1. Reiterations of series of movements which lead to sound(s) or syllables repetitions; 2. Global maintain of the articulatory posture with or without an acoustic output; 3. Anarchical movements with or without an acoustic output. A fourth category can be observed in our data, consisting in a combination of the above. While the first category only concerns repetitions, the two others can concern SLDs categorized as repetitions, prolongations or blocks, showing that the same articulatory pattern can be observed for the 3 types of disfluencies. As for the last category cited, we noticed that the disfluency is perceived either as a prolongation either as a repetition depending on the amplitude of movements (Fig 1). Indeed, when a succession of vertical movements is observed, the disfluency is considered as a prolongation as long as the articulator does not exceed a certain limit. If the articulator exceeds this limit before it returns in its initial position, a sound disruption, and therefore a repetition, is perceived. Discussion The instability of the vertical position of an articulator has an influence on perceptive categorization of a SLD. Moreover, and even if it is important to consider the subglottic and laryngeal configurations, articulatory movements have an implication on the classification of a SLD as a repetition or a prolongation. Consequently, the description of the articulatory movements during a disfluency could be considered as an indicator showing the limit of the gesture viability. Thus, the analysis of this type of speech disorder allows to study boundaries of the audibility of an articulatory gesture and hence brings evidence for the Quantal Theory (Stevens et al., 2008).