Prediction in speech and language processing

Current theories of perception assume that what we really see or hear derives from the integration of the available sensory evidence with prior assumptions.

This idea, which dates back to Helmholtz's suggestion that perceptions subconsciously draw on information from long-term memory (Helmholtz, 1909), constitutes the core of a predictive coding approach to speech and language processing.

The editors draw a principled distinction between pre-activating neural structures to process the identity of contextually specified linguistic units (knowing what next), and aligning neural time scales in phase to the units’ predicted onset time, an inherently predictive neural mechanism (knowing when). The contributions included in this Special Issue demonstrate the explanatory power of predictive processes at all levels of the linguistic hierarchy: speech sounds, words, sentences, and conversations.

The articles present experimental data from healthy and impaired human populations, acquired using cutting-edge brain imaging techniques, amongst them functional Magnetic Resonance Imaging (fMRI), electroencephalography (EEG) and magnetoencephalography (MEG), as well as interruptive methods such as repetitive transcranial magnetic stimulation (rTMS) and transcranial direct current stimulation (TDCS).

Original publication:

Tavano, A., & Scharinger, M. (2015). Prediction in speech and language processing. Cortex, 68(0), 1-7. doi:10.1016/j.cortex.2015.05.001  

Contact:

Dr. Alessandro Tavano
Max Planck Institute for Empirical Aesthetics, Frankfurt am Main
+49 69 8300479-321

Dr. Mathias Scharinger
Max Planck Institute for Empirical Aesthetics, Frankfurt am Main
+49 69 8300479-117