a lot more,there's only one other thing the night.
spent a lot of time on,and maybe less a little love,and around to something to play with.and that was.
then the use of finite state devices.
compositional english.
interestingly,that came from the same place,in a sense that unification came from.
um.
we wanted to be able to talk about the way more themes were organized,to find words and the way they were modified when they came.jason.
two other,more.
and what of phonology?
of.
graphology is that,would that computational english?
so the way describing this thing.
so it would be reversible.
and if you look at the way a lot of that was done,the best mum we had for.
more local phonology at the time.
was.
seven pattern of english.
but i truly.
and it has this set of ordinary writing rules.
started to talk to you what different talks about flying,whatever will be applied,some rules can be.
optional,others.
like you guys,when that was a thing about the cycle,so it was mistake to implement that.
actually worked in the states,but let me the mistake.and,and so the question was,how shall we.
no.
deal with these rules in such ways to make them reversible,and then about the reversible thing you can imagine,the way they come.
then describe strictly,as i say it rules about them.so if you have this is input,you know exactly how to get there.but now,you perverse person.that have.
clear you could think about was doing,it would probably say quite a lot of time.
and we noticed,run out,and i really,really remarkable thing about the real.
if you write down the formula for everybody is somehow that is rewritten to beta.
in the environment number,it is on the left and on the right that one looks like.
now people will look at that and say.
that's a month of process that's equivalent to turning when everybody knows that.
and of course,you don't know what it is equivalent to.it's not as if you can't make a very much like that until a specified great,just exactly.
you and your.
these two.
when we made it very,very small change from the usual one about general rewriting moves.
the charity was that once a given rule.
has removed his piece of string and replaced it by the piece of string.that rule,come.
that piece of the three,it can look at it.
use it doesn't environment to changing something else,but he can't change that string again.
once you do that,the thing is.
to finite state machine.from one end of the sky cleared down,almost to the other changes in.
we should,how you could come back,how you could compile one of these rules into a finite sequence.user.
we discovered that.
the mathematicians all about finite state automaton,nothing about finance excuses.so we had to invest a lot of that.and then you could take the final difference uses and compose them into one machine.
and you work very hard of doing that.but by early,once you've got it,it wasn't like a.
extremely good,i generating all.
and i didn't manually.nothing is completely symmetrical for every point of view.
so that was,that was a really big.
i'm really going on,i ve now really interesting very well.
i always do.
but you,all the time.i.
i have only one other little too much to me.
but i will do very quick,one little piece of the controversy surrounding a field that,having given us certain to have like this.
i can't believe it without saying something about it.
i feel this comes back to where it started.
we had computers.if we had language,we work quite sure what to do when two of them got together in the same room.
the thought from machine conversion,because the one thing we can do is we could count.
we.
things that you can never have before,i mean,we could care of how many things with this is the way,how many letters,the word,how many words,the words.i mean,you can count.so surely.
but we didn't have much.
statistical.
statistical had added.
but it wasn't sophistical,either.
stuff to bring to bear on it.we didn't have much in the way of corporate.the worldwide web was a long way.
but on that side of.
but now we do know that we do that.we do that,the world will we turn him,we can be receiving.
we can build beijing.we can do know about noisy channels.that was.
and a lot of this.
big english.
a lot of it will get some stuff out of the lead.it's going to go to the way they went last time.it's just a question of.
notice that the way use the way it is supposed of those these days.
this was going to be a real simple way.
let me in,simple formula.it.
the way you do it is you ever say a person.six.
the structures that they ought to have,and then you learn the grammar automatically from those.
this a great deal.
no.
there are lots of things wrong with this.what is the motivation?
yes,you would like to know you do not like your car if you get not just all the possible lives is not only possible romantic interpretation,but the right one.
so you would like you to know more than where as new before this is,this is is to n p,pp with numbers on them and help.
and to help you get this right,you going to get you to use the just conceptions.should.
and what that means is that in those numbers,there is hidden a lot of stuff which has got nothing to do with language.owing with its.
the statistic stands as a surrogate for knowledge of the world a lot of the time.this is not knowledge about.
and if we try to cover it with statistics,this may be good injury.
and in there,i'm not say,shouldn't say it shouldn't be done.maybe we can actually get some practical things.in the meanwhile,i was putting off the day when we can try to understand.
one of those things do need to understand.
because what the numbers are a cover for our.
and we have to do something about that ignorant.learn what's going on there.my belief is that there is nobody in the linguistic centers of our brain.
so it's tossing coins,or rolling any dice.
i didn't believe that that that's happening.
um.
and there was another thing that was wrong.instances.
but then going to tell you about.
but i think they won't tell you about it.
because it really is getting late,and i forgot what it was anyway.
ah.
okay.