Using journalistic texts, this paper contrasts English norms for cohesion with Japanese norms. One major difference found is that English texts make much heavier use of anaphoric reference items than do Japanese texts. Second, in Japanese, predicates play an important role in giving a sequence of sentences a sense of unity. Third, English journalistic texts tend to have cohesive devices between paragraphs, more so than their Japanese equivalents. Fourth, the Japanese writer often repeats an identical word or phrase for a referent to maintain lexical cohesion, while the English writer has more options—the use of a synonym, metonymy, or superordinate—made available by the definite article. Fifth, translators putting English expository paragraphs into Japanese may have to use discourse markers that are implicit in the original, a difference created by the presence, or absence, of a code for paragraph organization. In either direction, translation done without regard for those differences creates texts that are unnatural at best, unintelligible at worst.