Case Study: Understanding Natural Language - Thinking as Computation - page 175

Want to read more?
This chapter looked at the thinking behind parsing and interpreting expressions in a
natural language.
While there is a long tradition of studying grammar and parsing in classrooms, it
is primarily in the work of Noam Chomsky on what is called transformational grammar
starting in the late 1950s that the study of syntax took on a new mathematical preci-
sion and generality of description [41]. Chomsky is also well known for having first
proposed a solution to the puzzle of how young children could learn the language
of their parents after being exposed to so few sentences. His proposal is that we are
actually born with a universal grammar that allows us to adapt very quickly to the
actual grammar of the sentences we first hear.
The approach to semantics in this chapter, which seeks to query or update a sepa-
rate knowledge base of facts about the world, owes much to the work of the computer
scientist Terry Winograd. His doctoral thesis [43] at MIT in the early 1970s was a much
more elaborate version of what was considered here, in a world of blocks. To get a
glimpse of the major developments in computational linguistics since then, consult
the textbook by Jurafsky and Martin [42].
It needs to be emphasized that the discussion in this chapter only scratches the
surface of how natural language can be used and the kind of thinking that goes on
behind it. The few examples presented here may have given the impression that the
primary purpose of language is to provide information or to request it. In fact, even
a casual study of how people actually use language will reveal that this is but a small
part of the overall picture.
It was the philosopher John Austin in the 1950s who first observed that the
important thing about language is that people perform language actions (informing,
warning, lying, threatening, promising, and so on) to further their goals [40]. Some-
times these goals involve information, but often they are quite different. A speaker
might be interested, for example, in having a door closed. One way of achieving that
is to get up and close it; another way, in the right circumstances, is to utter some
words like, “Can you close the door?” (clearly not expecting a yes/no answer) or
even “Door, please.” This suggests that the thinking behind language actually has
much in common with the thinking about how to achieve goals using actions, which
is the topic of chapter 9.