Thenceforward the sentence is the decisive element--we speak in sentences and not words. Attempts to define the two have been frequent, but never successful. According to F. N. Finck, word-formation is an analytical and sentence-formation a synthetical activity of the mind, the first preceding the second. It is demonstrable that the same actuality received as impression is variously understood, and words, therefore, are definable from very different points of view.[190] But according to the usual definition, a sentence is the verbal expression of a _thought_, a symbol (says H. Paul) for the connexion of several _ideas_ in the soul of the speaker. It seems to me quite impossible to settle the nature of the sentence from its contents. The fact is simply that we call the relatively largest mechanical units employed “sentences” and the relatively smallest “words.” Over this range extends the validity of grammatical _laws_. But as soon as we pass from theory to practice, we see that language as currently used is no longer such a mechanism; it obeys not laws, but _pulse_. Thus a race-character is involved, _a priori_, in the way in which the matter to be communicated is set in sentences. Sentences are not the same for Tacitus and Napoleon as for Cicero and Nietzsche. The Englishman orders his material syntactically in a different way from the German. Not the ideas and thoughts, but the thinking, the kind of life, _the blood_, determine in the primitive, Classical, Chinese, and Western speech-communities the type of the sentence-unit, and with it the _mechanical_ relation of the word to the sentence. The boundary between grammar and syntax should be placed at the point where the mechanical of speech ceases and the organic of speaking begins--usages, custom, the _physiognomy_ of the way that a man employs to express himself. The other boundary lies where the mechanical structure of the word passes into the organic factors of sound-formation and expression. Even the children of immigrants can often be recognized by the way in which the English “_th_” is pronounced--a race-trait of the land. Only that which lies between these limits is the “language,” properly so called, which has system, is a technical instrument, and can be invented, improved, changed, and worn out; enunciation and expression, on the contrary, adhere to the _race_. We recognize a person known to us, without seeing him, by his pronunciation, and not only that, but we can recognize a member of an alien race even if he speaks perfectly correct German. The great sound-modifications, like the Old High German in Carolingian times and the Middle High German in the Late Gothic, have territorial frontiers and affect only the speaking of the language, not the inner form of sentence and word.