Even at 4 months, a labial is a good enough coronal, but not vice versa

Abstract

a b s t r a c tNumerous studies have revealed an asymmetry tied to the perception of coronal place of articulation: participants accept a labial mispronunciation of a coronal target, but not vice versa. Whether or not this asymmetry is based on language-general properties or arises from language-specific experience has been a matter of debate. The current study suggests a bias of the first type by documenting an early, cross-linguistic asymmetry related to coronal place of articulation. Japanese and Dutch 4-and 6-month-old infants showed evidence of discrimination if they were habituated to a labial and then tested on a coronal sequence, but not vice versa. This finding has important implications for both phonological theories and infant speech perception research.

The phonetic landscape in infant consonant perception is an uneven terrain

“…As outlined in the introduction, the language-independent biases documented here and in Tsuji et al (2015) align with asymmetries observed in word processing research with toddlers and neural processing research with adults. Clearly, these early perceptual biases are a precursor rather than a product of lexical development.…”

“…Thus, this finding shows that perception of coronal-labial consonant place is asymmetric before attunement to native language phonetics or phonotactics is underway. Even so, early attunement would not explain the asymmetric findings reported by Tsuji et al (2015) because coronal consonants occur more frequently than labials in Dutch whereas the reverse pattern is found in Japanese.…”

“…Thus, the same acoustic-phonetic property proposed to drive asymmetries in vowel perception (focalization) within the NRV framework may also account for the asymmetry observed for /r/ -/l/; notably, the /r/ and /l/ consonants are also classified as semi-vowels. To explain the labial-coronal (ompa-onta) place asymmetry, Tsuji et al (2015) offered a salience-based interpretation grounded in acoustic-phonetic properties of the stop release bursts based on research by Chang et al (2001); this account converges conceptually with NRV principles. The data reported by Segal et al (2016) suggests that discrimination of the /b/-/p/ voicing contrast is unbiased (symmetric) in young infants but become asymmetric once language attunement is clearly underway.…”

“…The authors suggest that discriminating the /ra/ to /la/ change may be more difficult because the /a/ vowel interferes (via forward masking) with the formant transitions for /la/ more than for /ra/. Tsuji, Mazuka, Cristia, and Fikkert (2015) reported another consonant discrimination asymmetry that also points to a language-independent effect. Using a habituation paradigm, they tested Dutch-learning and Japanese-learning 4-to 6-month-olds' discrimination of /ompa/ and /onta/; these non-word stimuli contain contrasting native consonants in both languages.…”

An order effect in English infants’ discrimination of an Urdu affricate contrast

“…Would infants fail to discriminate a prototypical exemplar from a subsequently presented less prototypical exemplar, but be able to discriminate the same two exemplars when they are presented in the reverse order? Note that there have been reports of asymmetry in discrimination between two consonants that are phonemically distinct in a child's language (e.g., Altvater-Mackensen & Fikkert, 2010;Tsuji et al, 2015;Nam & Polka, 2016). This kind of asymmetry within the native language is not what concerns us here; we aim to test possible asymmetry in non-native consonant perception, i.e., in perception of two segments which may have no straight-forward mapping onto distinct native language categories.…”