Babies lead the way – a discovery with infants brings new insights to vowel perception

Linda Polka – linda.polka@mcgill.ca

School of Communication Sciences & Disorders, McGill University SCSD, 2001 McGill College Avenue, Montreal, Quebec, H3A 1G1, Canada

Matthew Masapollo, PhD
Motor Neuroscience Laboratory
Department of Psychology
McGill University

Popular version of 2ASC7 – What babies bring to our understanding of vowel perception
Presented at the 186th ASA Meeting
Read the abstract at https://eppro02.ativ.me/web/index.php?page=IntHtml&project=ASASPRING24&id=3672633

–The research described in this Acoustics Lay Language Paper may not have yet been peer reviewed–

From the early months of life, infants perceive and produce vowel sounds, which occupy a central role in speech communication across the lifespan. Infant research has typically focused on understanding how their vowel perception and production skills mature into an adult-like form. But infants, being genuine and notoriously unpredictable, often give us new insights that go beyond our study goals. In our lab, several findings initially discovered in infants are now directing novel research with adults. One such discovery is the focal vowel bias, a perceptual pattern we observed when we tested infants on their ability to discriminate two vowel sounds. For example, when testing infants (~4-12 months) to see if they could discriminate two vowel sounds such as “eh” (as in bed) and ‘ae” (as in bad), infants showed very good performance in detecting the change from ‘eh’ to ‘ae’, but very poor performance when the direction of change was reversed (detecting change from ‘ae’ to ‘eh”). Initially, these unexpected directional differences were puzzling because the sounds were identical. However, we soon realized that we could predict this pattern by considering the degree of articulatory movement required to produce each sound. Articulatory movement describes how fast and how far we have to move our tongue, lips, or jaw to produce a speech sound. We noticed that infants find it easier to discriminate vowels when the vowel that involves the most articulatory movement is presented second rather than first. In essence, this pattern shows us that vowels produced with more extreme articulatory movements are also more perceptually salient. Our scientific name for this pattern- the focal vowel bias – is a shorthand way to describe the acoustic signatures of the vowels produced with larger articulatory movements.

These infant findings led us to explore the focal vowel bias in adults. We ran experiments using the “oo” vowels in English and French, which are slightly different sounds. Compared to English “oo”, French “oo” has more articulatory movement due to enhanced lip rounding. Using these vowel sounds (produced by a bilingual speaker), we found that adults showed the pattern we observed in infants. They discriminated a change from English “oo” to French “oo” more easily than the reverse direction, consistent with the focal vowel bias. Adults did this regardless of whether they spoke English or French, showing that that the focal vowel bias is not related to language experience. We then ran many experiments using different versions of the French and English ‘oo” vowels, including natural and synthesized vowels, visual vowel signals (just a moving face with no sound), and animated dots and shapes that follow the lip movements of each vowel sound. We found that adults displayed the focal vowel bias for both visual and auditory vowel signals. Adults also showed the bias when tested with simple visual animations that retained the global shape, orientation, and dynamic movements of a mouth, even though subjects failed to perceive these animations as a mouth. No bias was found when movement and mouth orientation were disrupted (static images or animations rotated sideways). These findings show us that the focal vowel bias is related to how we process the speech movements in different sensory modalities.

These adult findings highlight our exquisite sensitivity to articulatory movement and suggest that the information we attend to in speech is multimodal and closely tied to how speech is produced. We now resume our infant research focused on a new question – as young infants begin learning to produce speech, do their speech movements also critically contribute to this perceptual bias and help them form vowel categories? We are eager to see where the next round of infant research will take us.