The purpose of this thesis is two. Firstly, to criticize the nature of linguistic units in the present school grammar, secondly, to redefine the units according to the viewpoint which is suitable for language-in-use of native speaker. For this job, I highlightened the functionally working side of each linguistic units in language use. And moreover, I criticized that the present school grammar has overlooked the functional respect of grammatical units. All the units have a burden of function in some degree. But some of them have more portion than another. For example, the suprasegmental function as the main method of speech cohesion while the eojeol, which is equivalent to the sentence constituent, doesn't have much function in language use. The concept 'morpheme' has to be changed in order to focus on the word productivity. The concepts such as word, phrase, clause, sentence are important because they have the large portion of function in communication. Whenever the native speaker use a language, he or she certainly represent the proposition content as a clause. That is why the core unit in language is not a sentence but a clause. Up to now, the researchers majoring in school grammar have been stressing the discourse-level grammar rather than the sentence-level. But this kind of trend will bring about the exclusiveness between discourse and sentence, because of the difference of language view. I maintain that the sentence level grammar need to be reconsidered in the functional respect. The discourse concept should be also redefined for the purpose of 'one-grammar': i.e. the integration of the sentence level and the discourse. When it comes to the concept 'situation', that is the equivalent to the function or meaning in the functional view. But 'situation' has been one of the environmental variables all the while. That is, in the functional view, the meaning or function which each native speaker intends to refer is not all the variables but the specific one in the situation. The specific function in the situation corresponds to the Hallidayan concept 'register'. In conclusion, the grammatical knowledge concerning to discourse is 'the potential repertoire composed of linguistic items according to the register.' This research will be the foundation of the grammar education only if the grammatical knowledge are changed into the performative knowledge. Furthermore, this project will also strengthen the instrumental phase of grammar education.