It scarcely seems surprising that learning to underline a modal verb, such as "can", and "may", does little to help students use them effectively in their own writing. These words are anyway grasped by tiny children without the need to know what they are called. This may tempt the conclusion that the teaching of grammar should be shelved altogether. But there are reasons to reform it rather than throw it away.
Understanding of language is part of a wider education in what makes human beings human. How concepts are turned into sounds, and how those sounds combine to form commands or questions, are issues that have occupied many language experts. What they reveal about the mind has exercised psychologists and cognitive scientists.
There are practical reasons to ask children to work hard at grammar, too. One is that a knowledge of it will make learning a foreign language easier. Even if you did know by nature how to make clauses in your native languages as a child — just without instruction — getting to grips with them in German or Russian in later years is simpler if you know how to define and spot them. As it is, many English-speakers come to understand grammar by studying a foreign language, rather than the other way round.
For grammarians keen on future jobs, the natural-language processing field is booming. After many years of poor results, technological wizards have developed programs for automated translation, speech recognition and other services that are actually usable, if far from perfect. These tools may rely more on knowledge of artificial intelligence than of the subjunctive, but linguistic expertise still matters, and may give beginners an edge over competitors whose best language is Python (一种编程语言).
Grammar could still be taught better. One small study showed improvement in some students when concepts are linked concretely to writing tasks. A cook does not need to know chemistry to make a delicious soup. But the science of how words combine to make meaning is fascinating and fundamental.