Deductive syntax

Traditional Grammar developed a syntax based on the analysis of well-formed sentences given a priori. That is, it starts from a grammatical and acceptable sentence and analyzes it to illustrate the syntactic theory. There is nothing wrong with this method, as the traditional syntax is rich and successful. But one of the limitations of the analytical method of Traditional Grammar is that with it a large number of rules are not made explicit. In this article we will provide you the information about Deductive syntax.

By this method we study a possibility of phrase and not all similar possibilities of the chosen model. If we want a more exhaustive description of the syntactic rules of the language, it is better to resort to a deductive method. Transformational Generative Grammar, for example, is deductive in the sense that it does not start from examples of well-formed sentences, but from rules that should generate well-formed sentences.

The deductive method takes the opposite path of the analytical method of Traditional Grammar. While the traditional method goes from the given sentence to its syntactic structure, the deductive method derives sentences from the given syntactic structure.

semantic layer

The deductive method offers the advantage of being more rigorous and exhaustive, but it has a price. The definition of all the rules necessary to generate a certain type of phrase requires taking into account a very large volume of variables and the question that arises at this time is whether it is worth the effort to go down to the smallest details necessary to generate all the phrases possible for a model.

More than that, we have to question the possibility of creating such a complete and efficient set of rules that it generates all possible sentences in the language. A syntax that generates grammatical and acceptable sentences requires establishing semantic as well as syntactic rules. And in establishing the semantic rules of generation, the situation is considerably complicated.

For our part, we will not commit ourselves to exhaustiveness. We believe that it is possible to reach a satisfactory level of description if we restrict ourselves only to the syntactic aspect. The inclusion of semantic rules in the model certainly gives more power to the theory, but requires a description effort that exceeds our production resources. The definition of semantic rules will be left to a future lexicon usage dictionary. This dictionary will not limit itself to giving the possible meanings of a lexicon item, but will explain all the typical conditions of use of that item in the language.

We will exclude exclusively semantic rules from our proposal but we want to guarantee that the deductive syntax will generate grammatical sentences, although we cannot guarantee that these sentences are semantically valid. In other words, our syntactic-centric syntax will generate sentences like these:

• The boy read the book.
• The book read the boy.

Both sentences are grammatical, but only the first appears to be semantically valid. To prevent sentences like  the book the boy read from being generated by our deductive syntax, it would be necessary to add a layer of semantic rules to the sentence generation model. We will not deal with this for several reasons.

First, because our production capacity does not support such a large task. Second, because the relevance of seeking a description that descends to this level of detail is doubtful when the objective is to describe the general characteristics of the language. Furthermore, saying that sentences like  the book the boy read are semantically invalid is dangerous. An utterance is only semantically valid when immersed in a context. Perhaps against the background of a fable the books read.

The usefulness of a set of semantic rules that prevents the generation of sentences in which  book is the subject of the verb  to read is relative . Rules like that wouldn’t handle figurative language, fables, poetry. We do not want to fall into the error of thinking of language only in its ‘zero degree’, devoid of metaphors, metonyms, allegories, simply because ‘zero degree’ does not exist in the real world.

Scope of rules

A set of syntactic rules can have several levels of scope. Let’s look at some possibilities:

• Generates unacceptable utterances and no grammatical utterances . This type of rule is useless, totally disposable.
• It generates unacceptable utterances and part of possible grammatical utterances. This type of rule is at the lowest level of coverage.
• Generates unacceptable utterances and all possible grammatical utterances . This type of rules has a good level of coverage.
• It does not generate unacceptable utterances and generates part of the possible grammatical utterances . Rules of this type have a good level of coverage.
• It does not generate unacceptable utterances and generates all possible acceptable utterances. This is the highest level of coverage, which should be pursued unless there is reasonable justification for not pursuing it.

Real working conditions lead us to abdicate, in some cases, the ideal level of coverage in which the set of generated utterances is identical to the set of all possible grammatical utterances.

Let’s exemplify deviation from the ideal range with a rule about periods.

P = F ( [SCon F] _n )

The period (P) is formed by a phrase or by the concatenation of two or more phrases that are related two by two by a connecting phrase.

The given rule generates well-formed periods. Let’s see some of them:

• I think therefore I am.
• I came, I saw, I won.
• Ate and didn’t like it.

We can say that all well-formed periods satisfy the given rule. But from it, we can also get unacceptable periods like:

• * I think when I exist.
• * I came but I saw but I won.
• *Ate or didn’t like it.

The example rule generates all grammatical sentences but also generates unacceptable sentences. Its coverage level is good. In the example, unacceptable periods are generated because the rule does not establish restrictions on the simultaneous use of connectives, in addition to other peculiarities that need to be observed in the construction of periods. But this is not to say that the rule’s lack of comprehensiveness makes it useless. As it explains a characteristic common to all periods, we can consider it satisfactory in many contexts in which the level of coverage achieved is more than sufficient.

The choice of the level of coverage with which to operate is a decision to be taken on a case-by-case basis. There are cases in which only the general is sought. In other cases, the idea is to restrict yourself to the most common cases, leaving borderline and doubtful situations aside. It may also occur to seek a rigorous description of maximum coverage.