David Lewis’s account of conventions has received substantial criticism over the years, but one aspect of it has been less controversial and thus has been retained in various forms by other authors: his requirement that members of a group in which a convention obtains must know that they and others conform. I argue that knowledge of conformity requirements wrongly exclude certain paradigmatic conventions, including some central semantic conventions. Ruth Garrett Millikan’s account of conventions accommodates these cases, but it is marred by her (as I argue) mistaken claim that most linguistic conventions involve coordination, and by complications with her central notion of reproduction. Aiming to build upon Millikan’s insights while avoiding her account’s problems, I contend that doing A is conventional in groups in which there is widespread, interconnected copying of doing A as a way of doing something further, B, where there are other equally good and accessible ways to do B.

多年来，戴维·刘易斯对惯例的描述受到了实质性的批评，但它的一个方面争议较小，因此其他作者以各种形式保留了下来：他要求获得惯例的团体的成员必须知道他们和其他人一致。我认为关于一致性要求的知识错误地排除了某些范式约定，包括一些中心语义约定。露丝·加勒特·密立根 (Ruth Garrett Millikan) 对约定的描述适用于这些情况，但由于她（正如我所论证的）错误地声称大多数语言约定都涉及协调，以及与她的中心再生概念的复杂性，因此而受到损害。为了在避免她的帐户问题的同时建立在 Millikan 的见解之上，我认为做A在群体中是常规的，在这些群体中，做A的行为被广泛、相互关联地复制，作为进一步做某事B的方式，那里还有其他同样好的和可访问的方式来做B。