I do not support the thesis that the scientific language (in the sense of a formalized or partially formalized framework or a more or less precise framework) has to be fixed up before scientists are able to get reasonable results. On the contrary the language of the respective scientific discipline is usually developed with the development of that discipline and flexible enough to allow for further revision on grounds of new knowledge. However there is a sense (or several senses) of language and coding-dependency which pertains to very specific problems and results in Logic and Mathematics. This is illustrated in this paper by some examples from Logic and Mathematics.