Dynamic analysis of inter-words time intervals: a method to analyze the structure of communicative signals

Speech analyses are usually focused on words as signifiers ignoring inter-words time intervals (IWIs), which are related to the 'form' of speech, rather than to its 'content'. Applying the method of power spectrum analysis to inter-vocalizations time intervals of bird singing, underlying periodic processes were detected. In contrast, human IWIs revealed non-periodicity, which may be random or chaotic. To differentiate between these two possibilities, the non-linear dynamic methods of unstable periodic orbits and correlation dimension were applied to show that IWIs are characterized by a low dimensional chaotic attractor. Its correlation dimension of 3.2 +/- 1.1 suggests a minimum number of four variables underlying the system. The methods developed in the present communication can be further applied: (a) for the measurement of specific alterations in the processes underlying the form of speech in human disorders, i.e., schizophrenia, (b) for the assessment of normal and pathological developmental aspects of speech processes in children; (c) for comparing communicative signals between humans and other species.