A NEW APPROACH TO SYNTAX 
By 
B. J. Dasher, Georgia Institute of Technology 
This paper describes a new method for syntactic analysis of 
English. Instead of the conventional subject-predicate structure 
as a basis for analysis, elementary sentence patterns are used. It 
is observed that there are two basic sentence formats in English. 
One, using a transitive verb, consists of the sequence noun, verb, 
noun, noun. The other, using an intransitive verb, consists of the 
sequence nou_.._nn, verb, adjective, noun. In each of these basic forms, 
syntax is specified by the word order. Since there are 64 ways to 
arrange four words when they are taken one, two, three, and four at 
a time, there are 12B elementary or cannonical sentences to be stud- 
ied. The central goal of analysis is to determine the particular 
c annonical sentence corresponding to a given statement. 
In order to show how any sentence can be reduced to its basic 
format, certain essentially algebraic operations are proposed, to- 
gether with certain rules for transforming one structure into another. 
Conversely, the same rules may be used to construct a generative gram- 
mar that permits a cannonical sentence to be expanded to an equiva- 
lent form in accord with prescribed requirements. In as much as the 
operations are essentially algebraic, the method is very advantageous 
for computer use. 
Through the use of various devices, word order can be changed 
without changing essential syntax. Also, the same basic structures 
can be used to express a variety of semantic relationships. For ex- 
ample, the two sentences, "Give the book to John" and "Save the book 
for John", have the same structure. The difference between the t__o_o 
and fo___~r relationship is semantic rather than grammatical. Moreover, 
the two statements, "Give John the money" and "Give the money to 
John", express the same relationship between~ive, Joh____~n, and money, 
and it would be the same if an inflected form for Joh_.__~n were used in- 
stead of the preposition, or if some other syntactic label were used. 
Thus, it should be possible to begin with a statement in one language, 
find its corresponding cannonical sentence, transform this cannonical 
sentence into a corresponding cannonical sentence in a new language, 
and then reconstruct the statement in the new language. Thus, the 
method is advantageous for machine translation. 
This paper describes the fund~ental concepts of the scheme and 
illustrates its potential. Many details remain to be supplied in 
order to obtain a working system. 
