Catégorie d'événements : SRPP

Three production and perception experiments were conducted to investigate the role of the lips in Anglo-English /r/. The results indicate that the presence of labiodental /r/ has resulted in auditory ambiguity with /w/ in Anglo-English. In order to maintain a perceptual contrast between /r/ and /w/, it is argued that Anglo-English speakers use their lips to enhance the perceptual saliency of /r/ in both the auditory and visual domains. The results indicate that visual cues of the speaker’s lips are more prominent than the auditory ones and that these visual cues dominate the perception of the contrast when the auditory and visual cues are mismatched. The results have theoretical implications for the nature of speech perception in general, as well as for the role of visual speech cues in diachronic sound change.

Voix et paroles atypiques : phonétique, rééducation et apprentissage

Dans cette présentation, je synthétise l’ensemble de mes travaux dont la finalité commune est la description phonétique des voix et parole atypiques, et ses retombées rééducatives, didactiques et pédagogiques. Partant de mes premières recherches décrivant principalement les voix pathologiques et chantées en langue maternelle sans en considérer les influences linguistiques, et d’autres travaux consacrés aux paroles silencieuse, et chantée toujours en langue maternelle, je me suis ensuite orientée vers la description des écarts de prononciation d’apprenants du Français Langue Étrangère sur les plans segmental et suprasegmental. Je me suis ensuite intéressée à réunir dans des travaux communs deux éléments d’ordinaire dissociés : la voix et la langue étrangère, montrant qu’une différence de qualité vocale en fonction de la langue parlée pouvait être ressentie et mesurée. Finalement, je me suis employée à synthétiser les conséquences de ces descriptions pour la prise en charge orthophonique de la voix chez des sujets mono et plurilingues et chez des enseignants, et pour l’optimisation de l’enseignement de la prononciation du Français Langue Étrangère.

Ces analyses montrent que parler, c’est vocaliser tout comme vocaliser, c’est, chez l’humain, parler : dans quasiment toutes les situations de communication orale, le locuteur utilise sa voix en même temps qu’il parle ; réciproquement, utiliser sa voix n’est presque jamais dissocié de la production d’une parole articulée. Les interrelations entre la voix et la parole par l’influence réciproque de l’une sur l’autre sont attestées dans la majorité de mes travaux, et nourrissent ma réflexion disciplinaire. Elles me permettent notamment de porter un nouveau regard sur la dysphonie, en en considérant la parole et la langue comme d’autres facteurs de variation. Parler, c’est aussi s’ajuster : la théorie de la Variabilité Adaptative (Lindblom 1990), me porte à réfléchir sur l’équilibre entre  effort, ampleur et quantité des gestes vocal et articulatoire à fournir par le locuteur, et leur perception par le ou les auditeurs, en fonction des divers contextes de production que j’ai étudiés. Parler, c’est aussi percevoir, produire, et finalement communiquer de façon multimodale : plusieurs modèles d’apprentissage de la parole étrangère et de production de la parole m’incitent à intégrer des canaux de perception élargis dans la boucle communicationnelle, et par extension les verbalisations et ressentis des acteurs de cette communication orale. Enfin, l’intégration de tous ces paramètres optimise une aide unifiée comprenant les sujets enseignant et apprenant, le savoir et la matière, et l’environnement d’apprentissage ou de rééducation.

Neurodegenerative diseases are one of the most common disorders of the central nervous system. They all share the existence of a progressive degeneration of all or part of the nervous system. Their frequency, the severity of their disorders and the disabilities to which they lead make them a major public health problem. Among these pathologies, Parkinson’s disease (PD) is particularly studied in movement disorders, in particular because of its prevalence (in Europe, 150 patients per 100,000 inhabitants, of the order of 140,000 patients in France). While the motor expression of symptoms is reflected mainly in the limbs, the musculature involved in speech production is also subject to characteristic dysfunctions. Speech disorders can therefore be developed by infdividuals suffering from PD or other movement disorders. All speech disorders in PD and movement disorders are grouped under the generic term dysarthria and to date, data regarding the pathophysiology of this disorder remains incomplete.

Prosodic prominence in typical and atypical speech

Prosody plays an essential role for conveying the meaning of an utterance and speakers use multiple cues in the phonetic domain to regulate prosodic marking. In intonation languages, such as in German, prominence marking requires changes in intonation and articulation. Speakers use laryngeal modifications such as the placement of a pitch accent and the choice between accent types to highlight important information in an utterance (Ladd 2014; Gussenhoven 2004). Furthermore, systematic changes in the supra-laryngeal system are observable, leading to a more distinct articulation of prosodic units such as syllables and words (Harrington, Fletcher & Beckman 2000; Georgeton & Fougeron 2014). However, intonational and articulatory marking of prominence were often treated separately (Mücke, Grice & Cho 2014).

In our project, we concentrated on the modelling of prosodic prominence in a dynamical systems approach (Browman & Goldstein 2000; Gafos 2006; Iskarous 2017; Mücke, Hermes & Tilsen 2020). We proposed a model that integrates intonation and articulation within a unified system (Roessig & Mücke 2019; Roessig, Mücke & Pagel 2019, Roessig 2021). Our analysis techniques were designed such that a large array of categorical (e.g. pitch accents) as well as continuous (e.g. magnitude of tonal movements) modulations of prosody could be captured. In terms of empirical evidence as well as modelling advances, we demonstrate that prosodic prominence entails a multi-dimensional bundle of cues that is used by speakers in a flexible, yet systematic way. We therefore recorded 27 native speakers of German. Articulatory and acoustic recordings were carried out simultaneously with a 3D Electromagnetic Articulograph. To obtain a controlled corpus that is suited for quantitative analyses of articulatory patterns and at the same time being ecologically valid we designed an interactive game-like experimental setting. In this setting, participants were involved in a story with two animated robots in a factory. The resulting large data set of 2160 items (27 speakers x 20 target words x 4 focus conditions) allowed for extensive analyses of categorical and continuous modulations in prosody and how they intertwine. We compare our results with the prosodic marking strategies of other speaker groups. Therefore, we investigate patients with Parkinson’s disease as well as older speakers (Thies et al. 2020; Mücke, Thies, Mertens & Hermes 2021). Our focus lies on the role of variability and compensation in groups with different abilities of the speech motor system.

(This is a joint work with James Kirby, as part of the ERC-funded EVOTONE project.)

Coarticulation constitutes one source of variation in speech, which may give rise to sound change. Beddor’s « coarticulatory path to sound change » model (Beddor 2009) postulates that a subset of sound changes arise from the reinterpretation of a cue as being associated to the effect of coarticulation rather than to the source of it. Crucially, prior to this reinterpretation, the coarticulatory source trades with its effect: The source and the effect (1) compensate each other in production, and (2) are perceived as equivalent. Beddor’s model was developed based on the ongoing phonologisation of vowel nasalisation in American English, but she suggested it may also apply to other sound changes including tonogenesis stemming from onset voicing.
Our project in progress investigates the synchronic pattern of coarticulation between onset voicing and F0 in French, where obstruent-intrinsic F0 perturbations (or CF0) are not phonologised. We conducted a series of acoustic and perception studies to explore the trading relation between the source (prevoicing) and the effect (CF0). In production, we failed to find such a trading relation. Rather, for most speakers, the length of voice lead correlates positively with the lowering of F0, rather than trading with it. In perception, we addressed the question of perceptual equivalence for lower-level vs. higher-level perception, using three tasks: AX discrimination, 2AFC identification, and identification on a continuous scale. Our interim findings do not support a default trading relation. Rather, across listeners, higher discriminability of VOT predicts higher discriminability of F0. However, listeners who show higher discriminability of F0 also show higher perceptual equivalence when VOT is ambiguous, but lower perceptual equivalence when VOT is unambiguous. Unlike vowel nasalisation, listeners’ grammars do not differ in how they weight VOT and F0 perceptually, but whether and how they integrate F0 as a cue to voicing and/or perceive it as a non-phonemic feature when VOT becomes unreliable. We speculate that CF0 phonologisation is not initiated by a listener-specific grammar based on cue weighting, but by a shift in attention to a secondary cue when the primary cue is ambiguous by listeners who have stronger cue integration.

Reference:
Beddor, P. S. (2009). A coarticulatory path to sound change. Language, 785-821.

La fluence de la parole est la capacité d’un locuteur à produire des énoncés sans heurts ni interruptions involontaires, avec des transitions continues entre les sons et les syllabes et cela, en maintenant un débit relativement rapide (Starkweather, 1987). Une parole fluente suppose donc des connaissances approfondies de la langue parlée, du point de vue de ses structures morphologiques, lexicales, syntaxiques, phonologiques… afin que le locuteur puisse exprimer ses pensées sans effort particulier. D’un point de vue physiologique, conserver une parole fluente nécessite un contrôle précis de l’appareil respiratoire, du larynx et de l’activité ayant lieu dans les cavités supraglottiques (Adams, 1974).

La fluence absolue n’existe pas. En effet, lorsque l’ensemble des conditions langagières et physiologiques évoquées ne sont pas remplies, des accidents de parole, nommées disfluences émergent. Les disfluences peuvent prendre les formes de pauses silencieuses, de pauses pleines telles que des « euh », d’allongements syllabiques, d’interjections, de répétitions de syllabes ou de mots, d’auto-corrections, des faux-départs etc. Ces accidents de parole peuvent être considérés comme normaux en production de la parole et du discours car ils laissent au locuteur le temps nécessaire pour une recherche lexicale, une correction du discours ou pour planifier la suite de l’énoncé.

Dans le cas de certains troubles de la communication, le locuteur ne produit pas ces accidents de parole en vue de reprendre son discours. C’est le cas du bégaiement, où ces altérations de la fluence sont la conséquence de spasmes interrompant l’énoncé alors que le sujet parlant connait la suite du discours.

Les disfluences typiques du bégaiement, de même que la parole fluente produite par les personnes qui bégaient, ont donné lieu à un grand nombre d’études, qui se fondent généralement sur des données acoustiques. L’objectif de cette présentation est de se placer dans la continuité de ces recherches en employant cette fois des données articulatoires en vue de mieux comprendre les événements moteurs se déroulant pendant les disfluences typiques du bégaiement. Reposant sur 8 locuteurs (4 personnes qui bégaient vs. 4 personnes qui ne bégaient pas) ayant été enregistrés à l’aide d’un EMA, les études qui seront présentées proposeront une nouvelle classification des disfluences (Didirková et al., 2020 ; Didirková et al., 2019 ; Didirková, 2016) et aborderont la question de la coarticulation dans la parole bègue (Didirková & Hirsch, 2019).

Références :

Adams, M.R. (1974). A Physiologic and Aerodynamic Interpretation of Fluent and Stuttered Speech. Journal of Fluency Disorders, 1(1):35‑47.

Didirková I. (2016) Parole, langues et disfluences : une étude linguistique et phonétique du bégaiement. Thèse de Doctorat soutenue le 24 novembre 2016 à l’Université Paul Valéry Montpellier 3, 413 p.

Didirkova I., Le Maguer S., Hirsch F. (2020) An articulatory study of differences and similarities between stuttered disfluencies and non-pathological disfluencies. Clinical Linguistics & Phonetics, https://www.tandfonline.com/doi/full/10.1080/02699206.2020.1752803.

Didirková I., Le Maguer S., Gbedahou D., Hirsch F. (2019) Articulatory behaviour during disfluencies in stuttered speech. Actes des 19th International Congress of Phonetic Sciences 2019, 5-9 août, Melbourne (Australie), 2991-2995.

Starkweather, C.W. (1987). Fluency and Stuttering. N.J.: Pretince-Hall Inc, Englewood Cliffs.

It is well-established that in Tokyo Japanese, high vowels /i, u/ become devoiced between voiceless obstruents (e.g., /kuʧi/ → [ku̥ʧi] ‘mouth’). However, there is an ongoing debate on the exact phonetic/phonological status of these devoiced vowels. Some studies argue that devoiced vowels are completely deleted (Ogasawara 2013; Vance 2008) while others argue that they only lose their voicing while fully retaining their supralaryngeal gestures (Funatsu & Fujimoto 2011; Varden 2010). The disagreement seems to stem in large part from methodological differences. Studies that argue for deletion rely primarily on acoustic data of contexts in which only one of the two high vowels are phonotactically legal (e.g., [ʦuki] ‘moon’; *[ʦiki]), while studies that argue against deletion rely primarily on acoustic and articulatory data of [k]-initial contexts in which both high vowels are phonotactically legal in Japanese (e.g., [kiʧi] ‘good fortune’; [kuʧi] ‘mouth’). This talk presents a series of acoustic and articulatory studies on Japanese high vowels that control for different phonotactic predictability levels to tease apart methodological vs. phonetic/phonological effects. Results consistently show that complete deletion is possible, but also reveal an articulatory-acoustic disconnect, where devoiced high vowels are articulatory present but acoustically absent in certain contexts. The disconnect, together with perceptual results, could provide insight into different mechanisms behind production and perception grammars.

Recursion and the definition of universal prosodic categories

Lisa L.S. Cheng & Laura J. Downing

A body of recent work on the phonology-syntax interface within prosodic structure theory proposes thatthere are only two post-lexical phrasal categories in the Prosodic Hierarchy, Phonological Phrase and Intonation Phrase. These two categories are considered universal, as they are defined in terms of cross-linguistically valid syntactic categories, XP and clause:

(1)       Phrasal prosodic Hierarchy (Elfner 2012: 10; Itô & Mester 2013: 26; Selkirk 2011; Truckenbrodt 2007)

Intonation Phrase                 matches                     syntactic clause (CP)

|

Phonological Phrase              matches                     syntactic phrase (XP)

As Bennett & Elfner’s (2019) recent overview article makes clear, the claim of universality has the following consequences:

1. 1. all languages should show evidence that both of these prosodic categories actively condition phrasalphonology, not just one of them;
   2. only these two prosodic categories define domains for phrasal Even a phrasal prosodicdomain that does not straightforwardly match one of the syntactic domains in (1) must, nonetheless, bedefined as either a Phonological Phrase or Intonation Phrase.

In recent analyses, like Ito & Mester (2013), Elfner (2015) and Elordieta (2015), the number of prosodic categories appealed to has been kept to a minimum by crucially using recursion to parse strings into variants ofPhonological Phrase or Intonation Phrase. For example, Ito & Mester (2013) proposes that the accent phrase orminor phrase that has often been invoked in analyzing Japanese prosody is best recast as a minimal Phonological Phrase. Elfner (2012) and Elordieta (2015) propose that, in Irish and Basque respectively, CP isparsed as a recursion of Phonological Phrase, rather than Intonation Phrase, in apparent violation of principle 1,above.

In this paper, we discuss this use of recursion by critically examining the prosodic categories chosen to parseadjunction in syntactic structure. In particular, we re-examine how the recent literature handles the prosodicparse of adjectives, adverbs, topics and relative clauses. Our main empirical focus will be Bantu languages,building on work like that found in Cheng & Downing (2016), Downing & Rialland (2017) and Truckenbrodt (1995).

We show that the unrestricted use of recursion leads to a number of undesirable consequences: (a) differentprosodic categories are chosen to parse the same structure;

(b) different prosodic correlates are associated with the same recursive prosodic phrases; and (c) there is agreater mismatch between prosodic and syntactic structure. These problems obviously conflict with the goal of having a universal set of prosodic categories, with a cross-linguistically consistent definition. A final issuewe take up is whether factors related to information structure that interfere with the default phrasing motivatean additional category in the Prosodic Hierarchy.

Selected references:

Bennett, Ryan & Emily Elfner. 2019. The syntax-prosody interface. Annual Review of Linguistics 5, 151-171.

Cheng, Lisa L-S & Laura J. Downing. 2016. Phrasal syntax = phrasal phonology? Syntax 19, 159-191. Downing, Laura J. &Annie Rialland, eds. 2017. Intonation in African Tone Languages. Berlin: Mouton de Gruyter.

Elfner, Emily. 2012. Syntax-Prosody Interactions in Irish. Ph.D. dissertation, University of Massachusetts, Amherst.

Elfner, Emily. 2015. Recursion in prosodic phrasing: evidence from Connemara Irish. NLLT 33, 1169- 1208.

Elordieta, Gorka. 2015. Recursive phonological phrasing in Basque. Phonology 32, 49-78. Ito, Junko & ArminMester. 2013. Prosodic subcategories in Japanese. Lingua 124, 20-40.

Selkirk, Elisabeth. 2011. The syntax-phonology interface. In John Goldsmith, Jason Riggle & Alan C. Yu (eds.), The Handbook of Phonological Theory, 2nd edition. Oxford: Wiley-Blackwell, 435- 484.

Truckenbrodt, Hubert. 1995. Phonological phrases: their relation to syntax, focus, and prominence. Doctoral dissertation, Massachusetts Institute of Technology.

Truckenbrodt, Hubert. 2007. The syntax-phonology interface. In Paul de Lacy (ed.), The Cambridge Handbook ofphonology. Cambridge University Press, 435-456.

Listeners are highly proficient at adapting to the speech of a novel talker. Various mechanisms may underlie such adaptation, and at differing time courses. These may include phonetic cue calibration (e.g., McMurray & Jongman, 2012), phonetic covariation (Chodroff & Wilson, 2018), or general auditory contrast, particularly at short timescales (e.g., Lotto & Kluender, 1998). Under the cue calibration account, listeners adapt by estimating a talker-specific average for each phonetic cue or dimension; under the cue covariation account, listeners adapt by exploiting consistencies in how the realization of speech sounds varies across talkers; under the auditory contrast account, adaptation is not talker-specific, but results instead from (partial) masking of spectral components shared by adjacent stimuli. To distinguish between these mechanisms, we investigate perceptual adaptation to talker-specific sibilant fricatives at short and long timescales. In rapid adaptation, our findings indicate a strong role of general auditory mechanisms; however, a longer period between talker-specific exposure and test could reveal secondary, phonetic-based learning mechanisms.

The “bouba–kiki” effect is the perceptual correspondence between the nonce word “bouba” with a rounded and “kiki” with a spiky shape. This phenomenon – popularized by Ramachandran and Hubbard (2001), themselves inspired by Köhler’s classic work (1929) – has since been replicated numerous times in various experiments, serving as a test bed for understanding the psychology of crossmodal correspondence between sound and vision. In this talk, I will present the results of an extensive web-based study with 917 speakers of 25 different languages exploring the “bouba–kiki” effect across the globe. Controlling for genealogy and orthography, we demonstrate that the effect holds robustly across most of the investigated languages. We argue that such widespread relations between auditory signals and visual shapes might arguably have bootstrapped the origins and evolution of spoken languages by helping our ancestors form connections between speech sounds and their referents (Cuskley & Kirby, 2013; Ramachandran & Hubbard, 2001).