How Language Could Have Evolved
Ken Ken Del Signore
February 2021
 

This paper develops a biologically inspired computational model of the Human language faculty and some associated thought processes. This model is developed starting from a simple proto-language, which humans are assumed to have inherited at speciation. This proto-language consists of single symbol exchange using a small set of symbols; similar to the observed dialogue systems in the existing Great Ape families. Computationally, the model is built using a single class with the form of a Markov graph node. Instances of this node class are used to symbolically represent words. The model is built iteratively in main() as a single graph. Nodes are added to this graph using a merge, or conjunctive join, operation between two existing nodes, notionally labeled as head and copy. A simple first order graph is developed which is hypothesized to be common to all Mammals and to generate shared mammal behaviours. This graph is then extended into an interactive c++ prototype that allows for more complex human language phenomena, including adverb periodicity, movement, illicit grammar detection, conjunction, 3rd person verb marking, irregular verbs, adjective ordering, and the progressive and perfected tenses.
Format: [ pdf ]
Reference: lingbuzz/005770
(please use that when you cite this article)
Published in: in preparation
keywords: merge, syntax
Downloaded:723 times

 

[ edit this article | back to article list ]