Vocabulary Networks: Program-9
This simulation illustrates how a vocabulary network
responds when some of the connections that make up
the network are re-linked in a constrained way.
It builds a 1000 word network and lets the network
settle into a stable attactor state.
nEv controls the number of events in the simulation.
Events activate a single word, and resets both its
input connections from a range limited by MaxN.
THINGS TO DO.
Find a network with about 500 active words. Set nEv to 20
and run the simulation with different values for rEv.
How does the network respond to events? How big are the
changes that he network exhibits? What is the biggest
change that you find? What is the average change?
Now repeat these observations with a network that has
about 800 activated words? How does this change affect
the patterns you observe? Do you get the same effect
when your network has only 200 activated words?
What happens when you vary the value of parameter rEv?
QUESTIONS TO ASK.
What is the probability of a large change in your network?
Can you predict the final level of activation in a network?
Would nom-random relinking produce different results?
What implications does this set of behaviours have
for theories of L2 vocabulary acquisition?