28
enters, it has a certain probability of writing down a symbol. In the figure, the symbols are the
capital letters. The result is that the machine writes a string of symbols, a formula, e.g. NNSG.
FIGURE 1. Schematic illustration of a finite-state automaton. Numbers represent machine
states. Letters represent what the machine inscribes. The machine corresponds to a grammar
capable of producing an infinite number of formulae such as SXXSXG. This infinity is owing to
iterative recursion but not embedding, and hence cannot do justice to the syntax of any natural
language. Reproduced from Miller (1969: 131).
Chomsky devised the notion of a finite-state automaton or grammar to show how hopeless
it is for explaining natural-language syntax (1956). We can think of a finite-state automaton as
embodying rules for producing sentences. Such rules can be understood as instructions given to
an automaton governing its transition from one state to the next. If the automaton is in a state