Spoken language is meaning conveyed by the modulation of a sound signal. It is not the modulation, but the meaning that is master.
'When I use a word,' Humpty Dumpty said in rather a scornful tone, 'it means just what I choose it to mean - neither more nor less.'
'The question is,' said Alice, 'whether you CAN make words mean so many different things.'
'The question is,' said Humpty Dumpty, 'which is to be master - that's all.'
Through the Looking-Glass - Lewis Carroll
The study of grammar.
The study of grammar, as that term is currently used, originated with the Indians, Persians and Greeks. They were primarily attempting to describe how language is used - descriptive grammar. Grammar was adopted in England's grammar schools as a pedagocical discipline - prescriptive grammar - at a time when Latin was still the language of the educated classes. It is interesting to note that Latin ceased to be a living language at about the time when its grammar was fossilised by academics. When other academics started to take an interest in the English language, during the 16th century, they took the categories and syntax of Latin grammar and attempted to force English into this mould. Had they succeeded, English might now be a dead language.
About four hundred years later, in the 1950's when I attended grammar school, much time was devoted to the teaching of "English" grammar. It was only in later years that I discovered the simple fact that I was being taught 'correct' English on the basis that English ought to conform to Latin grammar. I now use the derogatory term 'grammateecha' for people who insist on the 'correctness' of 'to whom do you wish to speak', the stilted phrase 'it is I', the forced location of 'to' next to a verb, etc.
The complications of grammar.
Computational linguists find that however many 'rules of grammar' they model, there will always be found an example of language which doesn't quite fit these rules. It appears that the number of rules of grammar will always equal N + 1, where N is the number of rules so far enumerated. This arises from the fact of language evolution and the fact that by the average language user the social norms of national language use adhered rigidly to are not.
Academic grammarians today insist that their grammars are not prescriptive. Yet pedagogic materials continue to show examples of ‘incorrect’ useage, whilst the academic literature frequently refers to ‘ill-formed’ constructs. 'Incorrect' and 'ill-formed' are terms which are at the same time prescriptive, derogatory and unscientific.
Descriptive grammarians supposedly formulate word, phrase and syntax categories in terms of collocations. Collocations are defined as words which are frequently found together in language constructs. What does not fall into the categories is defined as either ill-formed, slang, idiomatic or mere collocation. Collocations are then defined as words which are merely frequently found together. This gives rise to a reductio ad absurdum best expressed in the form of an argumentum ad circularem:
The rules of grammar:
Some collocations demonstrate the operation of rules of grammar.
Some collocations do not demonstrate the operation of rules of grammar.
Some constructs are well-formed: they follow the rules of grammar.
Some constructs are ill-formed: they do not follow the rules of grammar.
Any 'grammar' is simply a description of the most common patterns in language use. It can never be a sufficient and necessary description of a living language, or a dead one as used by ordinary people who left no record of their everyday speech. It is a common mistake to assume that, just because observers have failed to observe particular constructs in the field, it follows that the constructions are in some way 'wrong'.
For me, 'grammar' is somewhat like the Ptolemaic theory of planetary orbits - far too complex. Take two simple examples of noun-verb constructs using nonsense words: 'Jablot phranps.' 'Kribmods pharge.' Without any need to appeal to subject-predicate, active-passive, conjugation, tense, person or mood, we instantly know the following about these new words:
Jablot is a substance or phenomenon capable of phranping.
Kribmods are countable and are capable of pharging.
New meaning, new knowledge in the example is conveyed solely by the 'jumping S of English, as it moves between two words. In similar vein, other simple patterns convey implicit information. Lewis Carroll knew that.
Truth-values and language
In the literature of philosophy and linguistics, the notion of a truth value for every sentence is attacked with constructs such as 'The name of the current king of the USA is John.' Self-evidently, this 'grammatically impeccable' sentence is not true. It is only 'untrue' if the focus is on the whole idea expressed by the whole sentence. However, it is possible to show that this example sentence is simply a package of implied truth values. For example, kings exist, the USA exists, John is a name, etc.
We can determine a sentence to be in some sense explicitly false only because we can determine all of its implicit ideas to be in some sense true or false. Computers cannot yet analyse language in this fashion. This is a task for the computational pragmatist.