Menu

How learning to read changes and doesn’t change our perception of speech

Event: How learning to read changes and doesn’t change our perception of speech

Date: Wednesday 25 November 2026

Time: 12:15-13:15

Venue: MKS414, MAKS Building

It is estimated that around 5% of the population have trouble learning to read. From the perspective of cognitive neuroscience, it is in fact surprising that this number is so low.

Reading has been around for less than 5,000 years, and the idea that reading is something everybody can achieve is less than 300 years old. This means that our brains are completely unprepared for the task of understanding language in written form.

Research has shown how learning to read co-opts brain structures for object recognition and even carves out its own area in the brain, a region previously involved in recognising objects and faces. Learning to read also changes our appreciation of speech. For us, it seems natural that speech can be captured in a series of letters.

However, in contrast to what this might suggest, this alphabetic principle appears to have been discovered only once in human history, and non-readers do not think of speech as a sequence of segments.

This raises the question of whether learning to read merely creates meta-linguistic awareness of speech—thinking of it as a sequence of phonemes—or whether it fundamentally restructures speech processing itself.

In particular, researchers working on visual word recognition have argued that literacy changes pre-lexical representations, making them multimodal and activating them bottom-up during spoken word recognition. Here, I challenge this view, proposing instead that phonemic awareness is a post-lexical phenomenon that may actually reduce sensitivity to phonetic reality for speech sounds without orthographic coding.

The empirical basis for this challenge comes from data on awareness of glottal stops, which have explicit graphemes in Maltese (q) and German (*), but also occur without orthographic coding, where they mark prosodic boundaries. If literacy creates multimodal pre-lexical units, listeners should identify glottal stops as segments even in these uncoded instances, because pre-lexical units are, by their nature, activated by bottom-up input. However, our results from over 200 German and Maltese participants show the opposite.

Together with the finding that participants report the presence of deleted segments in spontaneous speech despite their phonetic absence, these results suggest that literacy creates meta-linguistic knowledge while simultaneously rendering listeners 'deaf' to phonetic reality.

This event will be held at the Faculty of Media and Knowledge Sciences.

Author Profile:

Prof. Holger Mitterer obtained his PhD from Maastricht University before moving to the Max Planck Institute for Psycholinguistics in Nijmegen.

During his ten-year stay, he established himself as a leading researcher in spoken-word recognition and speech perception. His work brings together experimental psychology, linguistics and cognitive neuroscience to shed light on one of our most fundamental human abilities: understanding each other through speech. He has held visiting professorships at Hanyang University (Seoul, South Korea), Ghent University (Belgium), and the University of Tübingen (Germany).

He is currently Editor-in-Chief of the journal Language and Speech (SAGE Publications).


Categories