Tip:
Highlight text to annotate it
X
[ Music ]
>> Professor James Stanford: Agent-based modeling is a technique
where you have a screen full of hundreds of little creatures we call agents,
and the idea is that they interact with each other on the screen,
and each agent is independent in the sense that it has its own dialect features
and language abilities, but then as it comes into contact with another agent, it can change.
Its dialect may change.
It's a really powerful tool that's been used in medicine and social science,
and so we're applying it to dialect and language research as a way to model how languages change
and even possibly predict the future of what direction a language might take.
What we're excited about in this computational modeling project is
that we're able to model human interactions.
The great majority of language variation and change is happening
as individuals interact with each other.
So in that way, our model is able to show a great amount
of the ways that languages can change.
So our first model was a model of Chicago and St. Louis,
and so on the screen there are hundreds of little agents in Chicago,
hundreds of little agents in St. Louis, and there's even an interstate between the two.
It's like Interstate 55 that the agents can go up and down to communicate with each other.
In real life, there's a vowel change shift happening in Chicago,
it's around the Great Lakes area, and what's happening is that when people pronounce a word
like sock, like a sock that you would wear, it sounds a bit more like sack.
So sociolinguists have been examining this, and trying to see what's causing the change,
how does it shift between generations, and then also how is it diffused
from Chicago down into other cities.
So what our model is looking at is how it's diffusing from Chicago down to St. Louis
and also how it's going from parent to child.
So within our model, we actually have children being born.
So there's birth and death and there's shifts as people, as the agents move from city to city.
Some of the vowel change shift diffuses down to St. Louis, but not all of it,
and that's the [inaudible] in real life.
So we're able to test some of those principles, but do it on the screen in a way
that we can model generations in just a couple hours of the run of a model.
So in that sense, we can almost predict what might happen in the future in that area.
[ Talking in Foreign Language ]
>> Professor James Stanford: So our second project that we're doing
with the Newcomb grant is looking at endangered languages and seeing
if agent-based modeling can provide useful information
about how communities might revitalize their language.
>> I think this one's funny.
[ Talking in Foreign Language ]
>> Professor James Stanford: There are about 7,000 languages in the world,
and scholars estimate that we may lose about half of those languages in the current century.
So one thing that our linguistic program here
at Dartmouth emphasizes is describing these languages.
So for one thing, going out in the field and collecting, recording of these languages
and trying to see exactly the sound system, the words, and the sentence structure
of these languages, and analyze them, you know, for posterity to have that record, but also,
we're working with these communities and looking at ways
that they may be able to revitalize their language.
With our model, one thing we can do is set up a situation where there are endangered languages
within a larger community of a majority language.
So the example would be something like English as a large majority language, which is socially
and economically powerful, and then you have several indigenous languages within that world
that are trying to propagate their language to their children and continue the process,
but there's this constant social and economic pressure from the larger language.
So we model that in our agent-based system, and we're looking at issues
such as what factors are involved.
When there's more than one indigenous language in the system, how does that affect them,
but also we're testing some of the techniques that have been used
in real life to help revitalize languages.
And in the same way that meteorologists may model the way a hurricane moves
across the continent within certain air bars, one of our hopes is that we might be able
to predict what may happen with some of these endangered languages over several generations.
Within a certain error range, we could predict what might happen in the future.
[ Music ]