this paper deals with the first lines of Aristotle’s De interpretatione. Starting from lo Piparo’s (2003) authoritative reading, i try to show that sym- bolon does not mean either ‘symbol’ or ‘sign’ in the sense commonly ascribed to these words. On etimological grounds (see symballo), i suggest that sym- bolon means something as ‘mark’, ‘countresign’, i.e., a double-faced unity. in any case, symbolon establishes a symmetric relation between its relata: if A is symbolon of b, b is symbolon of A. On the contrary, semeion (i.e. ‘sign’, for example a tombstone on a grave) establishes an asymmetric relation between its relata: if A is semeion of b, b cannot be semeion of A. lo Piparo thinks that Aristotle’s use of the word homoioma looks like Euclides’s (and Wittgentein’s) use of ‘likeness’. but ‘likeness’ in our languages establishes a symmetric rela- tion between its relata (if A is like b, b is like A) while Greek homoioma does not acts so (if A is homoioma of b, b cannot be homoioma of A). therefore, i see a strict analogy between Aristotle’s use of homoioma in De interpreta- tione and Plato’s use of cognate words in Republic Vi, 510 a 9-10. What about pragma? With De rijk (2002 i, 106), and differently from most of scholars, i think that «pragma must stand not for actual thing, but for the content of an expression». So we would have a four-terms ratio in De interpretatione, as well as in Plato’s Republic Vi. So: 1. written types (graphomena, grammata) behave to voices as pictures to their pattern; 2. the whole phonetic level of a language behaves to the semantic level as a picture to its pattern; 3. on the phonetic level of a language, the pathemata tes psyches stand for words, names (onomata); 4. on the semantic level, pragmata (i.e. ‘facts’, ‘states of affairs’) stand for logoi (‘proposition’). Hence, pragmata are contents of (complete) expressions, i.e. propositions (logoi). [ABSTRACT FROM AUTHOR]