Differences between information in language and artificial communication systems - Speaker: Uriel Cohen Priva
Uriel Cohen Priva
Brown University, Department of Cognitive, Linguistic, and Psychological Sciences
Previous research has shown that language is efficient – everything else being equal, redundant elements are shorter and prone to be reduced. In my work I show that language differs from artificial communication systems in two fundamental ways. First, the efficiency of language is sub-optimal, in that language does not minimize linguistic elements only when they are contextually redundant but also when they are usually redundant. Thus, linguistic elements that are usually redundant are more likely to be reduced even when they are highly informative, and linguistic elements that are usually informative are less likely to be reduced even when they are redundant. Second, in artificial communication systems every symbol is as costly as any other symbol – the most efficient message is the shortest message that transmits a given amount of information. In contrast, linguistic elements have inherent cost – their markedness, confusability or articulatory properties – and languages minimize the cost of transmitting a message, rather than message length. In this talk I will discuss how these two principles interact, affecting both the distribution of sounds cross-linguistically, and license processes that lead to sound change.
If you would like to meet with the speaker, please contact Bernd Möbius.