Beyond acoustics -- capacity limitations of linguistic levels

TL;DR

This paper reviews how speech signals are characterized at multiple timescales beyond acoustics, emphasizing their importance for understanding speech perception and neural processing.

Contribution

It provides an overview of experimental work on multi-timescale speech analysis and discusses its implications for neural models of speech perception.

Findings

Different linguistic levels are characterized at distinct temporal scales.

Multi-time scale analysis integrates diverse research areas.

Experimental results align with neural and dynamical models of speech perception.

Abstract

Speech is a multiplexed signal displaying levels of complexity, organizational principles and perceptual units of analysis at distinct timescales. This critical acoustic signal for human communication is thus characterized at distinct representational and temporal scales, related to distinct linguistic features, from acoustic to supra-lexical. This chapter presents an overview of experimental work devoted to the characterization of the speech signal at different timescales, beyond its acoustic properties. The functional relevance of these different levels of analysis for speech processing is discussed. We advocate that studying speech perception through the prism of multi-time scale representations effectively integrates work from various research areas into a coherent picture and contributes significantly to increase our knowledge on the topic. Finally, we discuss how these experimental results fit with neural data and current dynamical models of speech perception.