I only have a rudimentary ( or even less than rudimentary) knowledge of generative grammar.

But what strikes me is that the sentence formation rules are coinded using parts of speech. For example ( the most basic one ) :

                 Sentence  --> Noun Phrase + verb Phrase 

What is the motivation for this?

Is there behind this a hypothesis as to cognitive processes involved in speaking. I mean is there a hypothesis such as " subject, object, complement are artificial categories invented by linguists; the real syntactic categories mentally present in the mind of every speaker, the real natural syntactic categories are parts of speech: noun phrase, verb, adverb, preposition, etc. "

Does generative grammar reject traditional syntactic categories such as subject, object, circumstantial complement, agent complement, etc. ?

Why does generative grammar use parts of speech to formulate syntactic formation rules?

PS : I am french, maybe the labels I use are not traditional in english grammar.

  • 1
    In one place you write "S" and in another you write "Sentence". So far as I'm concerned, that is a stylistic choice you are free to make, and I don't see why anyone else should be concerned about it. Feel free. Correspondingly, in the formalized theory of phrase structure grammar, the symbols used for grammatical categories can be chosen arbitrarily. Feel free. – Greg Lee May 25 '19 at 23:31

To my understanding, it's the other way around.

According to generativists, syntactic categories are a fundamental part of the mental grammar of a language. When you learn a new lemma, like "purple", you also have to learn how it acts syntactically: in this case, it basically combines with an NP to make a new NP. (In practice it's a bit more complicated, but that's not important here.)

Ancient grammarians noticed that a lot of words tended to pattern in the same way, and gave these patterns arbitrary names. The terms "noun" and "verb", for instance, come from the Latin words for "name" and "word" (and are cognate with those two English words, actually).

But parts of speech describe syntactic categories, not the other way around. In many languages (Japanese, Swahili, Lingála, etc) there are two or more syntactic categories that English-speakers would group together as "adjectives". In other languages "adjectives" don't have a syntactic category at all, they're the same as stative verbs. So in these instances, the word "adjective" isn't useful, and we make up new terms and symbols instead.

TL;DR: The symbols like "NP" and "VP" are arbitrary and made up by syntacticians, representing mental structures we can't observe directly. In the distant past, ancient grammarians made up words to describe these same mental structures, so it's traditional for modern syntacticians to use these same words. But the names are less important than the structures they describe: you could replace "NP" with "Ξ" and "VP" with "Ш" and the model would still work just fine.

| improve this answer | |
  • Yes. "Parts of Speech" is a term left over from 4th-century science; and the Latin grammarians didn't talk about adjectives, either -- they behaved like nouns so they were called nouns. "Syntactic categories", on the other hand, refers precisely to those lexical categories that are relevant to the syntax of a language. Since syntax varies from one language or dialect to another, so do syntactic categories. – jlawler May 29 '19 at 2:09
  • The traditional term 'part of speech' applies to what are called categories of words or lexemes. In other words, the term are synonymous. – BillJ May 30 '19 at 11:08
  • Yes. In requiring the base of a generative grammar to be a CFG, all that is required is that there be a finite number of categories. Other than that, they may be anything you like. – Greg Lee Feb 27 at 19:06

Generative grammar emerged most directly from formalism (and see formalism in mathematics). I'm not saying that Noam Chomsky is a formalist, but in his early work from the 50s, it is clear that in establishing the Chomsky hierarchy, he uses formal methods. In a formal system, what identifies a system and its parts is its form alone. That's what makes it a formal system.

You may wish. for your own purposes, to supply histories and interpretations for the terms of the system. You're free to do that, but the system itself doesn't depend on any such specific interpretation, and the various users of the system need not agree about the interpretations of the symbols used in the system. Typically, actually, they don't agree. This can give formal systems great generality, because the parts are not tethered to specific interpretations.

The example rules given in the question resemble the productions (rules) of type-2 grammars in the Chomsky hierarchy, so let's assume that is what they are. I'll use CFG, standing for Context Free Grammar, as a synonym of type-2 grammar. You want to know about the history and interpretation of the non-terminal symbols that are typical in applications of CFG to the description of natural language and other similar systems (like computer language).

That's a natural question, but it really isn't possible to answer it, because those who have used CFG have had various ideas in mind when they used the symbols, and sometimes, presumably, haven't had anything in mind at all. Perhaps the symbols are derived from traditional antecedents, or perhaps they are not. In one offshoot of CFG, Generalized Phrase Structure Grammar, the symbols are interpreted in accordance with their names, which are finite sets of features and attributes of various sorts.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy