Personal tools
You are here: Home Events ANC/DTC Seminar: Marcelo Montemurro, Manchester University (Host: Matthias Hennig)

ANC/DTC Seminar: Marcelo Montemurro, Manchester University (Host: Matthias Hennig)

— filed under:

Entropy, semantics and a novel statistical universal in language

  • ANC/DTC Seminar
When Feb 18, 2014
from 11:00 AM to 12:00 PM
Where IF 4.31/4.33
Add event to calendar vCal

Language is arguably the highest-level signal produced by the human brain, and has been shaped by evolution to efficiently transfer complex information across different brains.  As other complex information structures, language shows a delicate balance between order and disorder, which derives from the interplay between the diversity of basic symbols and their correlations.  In the talk, I will discuss two approaches that use information theory to quantify the degree of complexity in linguistic sequences from large text corpora.  First, I will show how a direct implementation of Shannon information is able to quantify and extract semantic information from language sequences; thus providing a link between Shannon information and meaning.  IN the second part, I will introduce a measure of relative entropy that can quantify the degree of order of words taking into account the full correlation structure of language.  When this measure is applied to languages from diverse linguistic families an almost constant value emerges that can be interpreted as novel statistical universal language, whose origins are likely due to cognitive constraints.