Computational and Evolutionary Aspects of Language by 1rgrJJJ

VIEWS: 6 PAGES: 11

									Computational and Evolutionary
Aspects of Language


   Written by: Martin Nowak, Natalia
   Komarova, and Partha Niyogi
   Presentation by: Amerika Adams,
   Isaac Huang, and Margaret Kuwata
                              Purpose
   Attempt to formulate a synthesis of three theories.
   Addresses the importance of human language.
   Addresses the following questions:
    1. What is language?
     2. What is grammar?
     3. What is learning?
     4. How does a child learn language?
     5. How formal language theory and learning theory can be extended to study
    language as a biological phenomenon and as a product of evolution.
Formal Language Theory
   It is a mode of communication
   It is a crucial part of human behavior
   It is the sequencing of small units into
    bigger structures.
   There are rules for word grouping.
What is language?
   The alphabet is a set of symbols
   Sentences are strings of symbols
   Language is a set of sentences
       Example: Binary Language
What is grammar?
   It is a finite set of rules specifying a
    language.
   Expressed in terms of “rewrite rules”
       Languages, Grammars,
          and Machines.
   Correspondence between languages, grammars,
    and machines.
   Regular Languages are generated by finite-state
    grammars which is equal to finite-state automata.
   Finite-state automata have a start, a finite number
    of intermediate states, and a finish.
   A particular run from start to finish produces a
    sentence.
         Learning Theory
   Learning is the ability to generalize beyond
    one’s own experience to new
    circumstances.
   Learning theory describes the mathematics
    of learning with the aim of outlining
    conditions for successful generalization.
   Theory of Universal Grammar (UG)
What is special about language
acquisition?
   We learn the grammar of generative
    systems.
   The process occurs without being
    instructed about rules.
   The set of all computable languages is
    learnable by an algorithm that memorizes
    the rules.
Evolutionary Language Theory
   The evolution of language occurred in several
    incremental steps that were guided by natural
    selection.
   The theory states that by taking a population of
    individuals and with each individual using a
    particular language, the successful ones are able
    to communicate which results in a pay-off that
    attributes to fitness.
   Languages change overtime because the
    transmission from one generation to the next is
    not perfect.
                Agreements
   1. Language is a crucial part of human
    identity and behavior.
   2. Language has evolved through
    Darwinian dynamics of natural selection.
   3. That learning is the ability to generalize
    beyond one’s own experience. Not just
    modeling others.
            Disagreements
   1. The set of all computable languages is
    learnable by an algorithm that memorizes
    rules. (?)
   That children have a restricted set of innate
    languages they could learn correctly.

								
To top