Vowel bias in French-learning 8-month-olds: measuring repetition perception with fNIRS

• In Event: Poster Session 11
  In Poster Session: PS 11 Topic - Language, Communication

Abstract

Syllable repetitions in the speech stream are salient and the newborn brain already automatically detects them. Using functional Near-InfraRed Spectroscopy (fNIRS), more neural activation was found for sequences containing syllabic repetition (e.g., ABB: mubaba, talulu... or AAB: babamu, luluta) than for random sequences (e.g., ABC: mubage, talupi...) (Gervain, Macagno, Cogoi, Pena & Mehler, 2008; Gervain, Berent & Werker, 2012). Again using fNIRS, the present study explores whether the infants' brain also detects repetitions carried by segments instead of syllables and whether consonants and vowels are differently processed in this respect. According to the "division of labor" hypothesis (Nespor, Pena & Mehler, 2003), vowels are more important than consonants for (morpho-)syntactic processing (V-bias) and should therefore be preferentially used for detecting structural regularities such as repetition-based patterns. Two behavioral rule-learning studies showed a V-bias in 6-month-old Italian-learning (Hochmann, Benavides‐Varela, Fló, Nespor & Mehler; 2018) and 11-month-old Spanish-learning infants (Pons & Toro, 2010). Here we test French-learning 8-month-olds to gather neural evidence for processing differences between consonants and vowels, and identify the neural bases of these processes. Following Gervain et al. (2008), French-learning infants were presented with blocks of trisyllabic ABC sequences (e.g., bukato, kidopa...) and either ABBc sequences (consonant repetition, e.g., bukaki, kidode..., Exp. 1) or ABBv sequences (vowel repetition, e.g., bukata, kidopo..., Exp. 2). Each block contained 10 sequences, constructed with 6 different consonants and 6 different vowels and synthetized with MBROLA with equal-duration consonants and vowels (120ms). No sequence was ever repeated. There were 6 blocks per sequence type, presented in interleaved order. Infants were tested with NIRx NIRScout 816 machine (source-detector separation: 3 cm; two wavelengths of 760 nm and 850 nm; sampling rate: 15.625Hz). The optodes were placed bilaterally over the temporal, frontal and parietal areas (10 channels per hemisphere, see Fig. 1). Changes in the concentration of oxygenated hemoglobin (oxyHb) were measured, submitted to a fully automatic preprocessing procedure, and analyzed with linear mixed models. A dynamic left-temporal increase in HbO to vowel repetitions was found (p =.004) with a higher HbO response to within-word vowel repetition at the start of the experiment (p =.018), a pattern which reverses by the end of the experiment (p=.013) (see Fig. 1). This response attests detection of phoneme repetition. Non-significant trends found for the perception of consonant repetition in the same ROI indicate different processing of vowels and consonants in this context, conform the division of labor hypothesis.

Authors

• Sandrien van Ommen, University of Geneva
  Presenting Author

• Anne van der Kant, Universität Potsdam
  Non-Presenting Author

• Irene de la Cruz-Pavía, University of the Basque Country UPV/EHU & Ikerbasque
  Non-Presenting Author

• Silvana Poltrock, Universität Potsdam
  Non-Presenting Author

• Thierry Nazzi, University of Paris
  Non-Presenting Author

• Judit Gervain, University of Paris
  Non-Presenting Author