Subject and Predicate Agreement in the Development of Generative Grammar. (Linguistics)
Michael Moss · 2000
1. Introduction In the following paper, I will attempt to assess the work undertaken until now to explain the phenomenon of agreement between the Subject and Predicate in a sentence. Some of the underlying questions that I will be asking are: When did generative linguistics break away from a string adjacent approach to agreement?; What kinds of mechanisms are proposed for communication between Subject and Predicate in absence of simple adjacency?; and finally, What questions still need to be answered in this area? I will go through the material chronologically in order to preserve the of the arguments as they were presented. Most of the material to be analyzed was produced by Noam Chomsky, although supporting and questioning materials have been introduced as appropriate. 2. The transformational approach to sentence construction With Syntactic structures in 1957 Chomsky led linguistics into the realm of generative grammar. He outlined a program by which languages could be broken down to a set of rules which, while being a finite set, would allow for the production of an infinite set of properly formed utterances. The set of rules would also limit the language to properly formed sentences. The heart of the in 1957 looked like this: (1) [SIGMA]: Sentence: [X.sub.l] [right arrow] [Y.sub.l] Phrase : [X.sub.n] [right arrow] [Y.sub.n] [T.sub.l] Transformational : [T.sub.j] [Z.sub.l] [right arrow] [W.sub.l] Morphophonemics : [Z.sub.m] [right arrow] [W.sub.m] (Chomsky 1957: 46). Before going further, let us note how these ideas were developed. The grammar was originally formalized in Chomsky (1957) as a finite number of rules of the form: XAY [right arrow] XYZ, where the arrow indicated 'is to be re-written as', and the following conditions were imposed: (2) Condition (1) X, Z, and Y are strings of symbols (X or Y or both possibly null) but A is a single symbol. Condition (2) Z is not null. Condition (3) A is not identical with Z (Postal 1964: 143). These conditions apply to both terminal and non-terminal constituents. Terminal constituents are those which can no longer be broken down by re-write rules. At this point, terminal constituents were labeled simply morphemes. The non-terminal constituents, thus, are variables or higher symbols, which can be broken down. The maximum non-terminal constituent was (and still is) the sentence, given the symbol S (Postal 1964: 143). Getting back to the model presented in (1), Chomsky presents the idea of the grammar with S as the highest non-terminal constituent as part of the model describing natural languages. Here, the constituent is made up of a group of unordered rules which re-write the constituents of a sentence. The constituents are broken down or parsed by the following kinds of rules: (3) a. Sentence [right arrow] NP + VP b. NP [right arrow] T + N c. VP [right arrow] Verb + NP d. T [right arrow] the e. N [right arrow] man, ball, etc. f. Verb [right arrow] hit, took, etc. (Chomsky 1957: 26) It is important to note that Chomsky here refutes the idea that phrase structure grammars suffice to formulate adequate grammars for the description of natural languages. The model given above is powerful. It is a great improvement over traditional Markovian finite state machines, (1) which analyze the sentence in a linear process. Yet due to some of structure's inherent similarities to a finite state machine, it quickly runs into difficulties itself. As such, it is still not powerful enough to adequately describe natural language on its own. The model analyzes a non-terminal string S by the constituent parts, which make up the whole. …