The Paradox of an Argument Losing Accuracy the Moment It Is Explained in Natural Language: What Percentage Does That Modal Auxiliary Express?

Scientific papers use uncertainty expressions such as may, might, could, and the Japanese -uru and kamoshirenai to avoid categorical claims. These expressions can refer to an event's probability, theoretical possibility, lack of investigation, dependence on conditions, an author's confidence, or rhetorical emphasis. Their functions can overlap, and readers may interpret them differently. Assigning a number such as 50% can therefore attribute to an author a probabilistic judgment they never made. This paper demonstrates the problem through a thought experiment that provisionally defines may as 50%: expressions of possibility recur when the definition's exceptions and scope are explained. It then examines a proposed "relay of checks" across an author, reviewers, an editor, and readers, in which an unchanged sentence may acquire different interpretations of certainty and institutional assurance. Peer review establishes a publication decision without guaranteeing agreement about each uncertainty expression. Drawing on research on verbal probabilities, hedging, and peer review, the paper distinguishes established findings from this untested relay hypothesis. It argues that numerical specification and institutional approval do not ensure the unchanged transmission of meaning.