Difficult to hear but easy to see: Accounting for the evolution of an /r/ specific lip posture in Anglo‑English - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2021

Difficult to hear but easy to see: Accounting for the evolution of an /r/ specific lip posture in Anglo‑English

Résumé

Speech perception is influenced by what we see as well as what we hear. In face-to-face communication, the lips are a visible articulator, providing the listener with a complementary source of phonetic information to the auditory stream. We know that speech perception is enhanced when listeners are able to see the speaker as well as hear them. However, seeing speech may also alter what the listener hears as well as enhance it (e.g., McGurk Effect: McGurk & Macdonald, 1976). Given the significance of visual cues, some researchers maintain that multimodal speech is the primary mode of perception and that sound systems have evolved to be both heard and seen (Rosenblum, 2008). Visual speech may even constrain phonological inventories by ensuring that phonetic contrasts which are difficult to hear are easy to see. For example, the visual salience of [m] versus [n] may explain why, among nasals, almost all of the world’s languages contrast /m/ and /n/ despite their acoustic proximity (Dohen, 2009). We present perceptual evidence suggesting that listeners look to phonetic cues from the speaker’s lips to better disambiguate similar-sounding speech sounds. This may result in greater reliance on visual labial cues to compensate for a phonetically ambiguous phonological contrast. We propose that visual speech cues may thus be implicated in sound change. The sound change we consider involves the labial articulation of post-alveolar /r/ ([ɹ]) in Anglo-English. A change in progress is underway in England in which the lingual articulation for /r/ is dropped for a labiodental one (Foulkes & Docherty, 2000). Acoustically speaking, labiodental /r/ ([ʋ]) may be closer to [w] than to [ɹ] and perceptual confusion between [ʋ] and [w] is attested anecdotally (Foulkes & Docherty, 2000; Knight et al., 2008). Labiodental /r/ may be emerging due to the heavy visual prominence of the lips for [ɹ] (Docherty & Foulkes, 2001), although phonetic accounts of the lips are lacking (but see King & Ferragne, 2020a, 2020b). To test the visual salience of the lips, 40 native Anglo-English subjects were presented with words beginning with [ɹ], [w] and [l] produced by a native speaker, and were asked to identify the word they perceived from two word options. Stimuli were presented in noise in three presentation modalities: auditory-only (AO), visual-only (VO) and audio-visual (AV). The results follow previous research on multimodal speech perception in that the perception of the /r/-/w/ contrast is enhanced by the presence of visual cues: AV was significantly better than AO perception. However, the auditory contrast between /r/ and /w/ is ambiguous as Anglo-English listeners tend to perceive both AO /r/ and /w/ tokens as /r/. The auditory cues for the two other contrasts under study (/r/-/l/ and /w/-/l/) are less ambiguous and perceptual sensitivity was significantly better. We propose that the auditory bias for /r/ in the /r/-/w/ contrast is due to increased exposure to labiodental /r/ in Anglo-English, which has resulted in perceptual uncertainty due to the acoustic proximity of [ʋ] to [w]. Listeners have to tolerate such a high degree of acoustic variation for /r/ that even canonical productions of /w/ may be reconstructed as /r/ in perception. However, the results indicate that the visual cues for /r/ and /w/ are much less ambiguous than the auditory ones. Participants were highly sensitive to the /r/-/w/ contrast in the VO modality. They correctly identified 96% of the tokens despite not having access to sound at all. Contrary to the perception of the contrasts with /l/, where AV was better than VO, no significant difference was observed between VO and AV perception of /r/-/w/, which suggests that the perceptual advantage from visual speech cues does not require auditory input whatsoever. Visual cues are therefore more informative to Anglo-English listeners than the auditory ones for the /r/-/w/ contrast. We contend that a specific labial gesture is produced for Anglo-English /r/, which reinforces the phonological contrast with /w/. Although a [w]-like sound may be reconstructed as /r/ in perception, a [w]-like visual cue can only be interpreted as /w/ because /r/ is produced with a very different lip posture. We conclude that visual phonetic cues may evolve for speech sounds when pressure to maintain a phonological contrast is high.

Domaines

Linguistique
Fichier non déposé

Dates et versions

hal-04139013 , version 1 (23-06-2023)

Identifiants

  • HAL Id : hal-04139013 , version 1

Citer

Hannah King, Ioana Chitoran, Emmanuel Ferragne. Difficult to hear but easy to see: Accounting for the evolution of an /r/ specific lip posture in Anglo‑English. PAC 2021 – Spoken English varieties: redefining and representing realities, communities and norms, Sep 2021, Toulouse, France. ⟨hal-04139013⟩
11 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More