Not long ago, when general linguistics tended to be identified with the philosophy of language, a linguist entitling a short article “The Word” would have been rightfully accused of unbearable presumptuousness. Any discussion of the word involved, in fact dealing with the relationship between thought and language, that is, penetrating into an area which the linguist neither dared nor desired to exclude from his researches but in which he felt ill-equipped to accomplish anything worthwhile alone. In the second place, it involved all the questions raised by the nature of the sign, in other words, semiology in its entirety. Finally it led to reconsidering the relationships between the “word” and the sentence on the one hand, the word and the “lower” elements of the sequence, syllables and “phonemes” on the other. The question, however, that the linguist could not ask was whether criteria existed that permitted, for all languages and all cases, the identification and delimitation of a segment of the sequence as a determined word. For this it would have been necessary for a linguist to feel that it was his duty to define precisely the terms he used. He would also have needed enough courage to foresee that the term “word” itself might have to be put aside if his researches showed that this term could not be given a universally applicable definition.