Gross wishes to claim a greater importance for linguistic practice in science. Knowledge is of an approximate character. Our formalisms abstract and simplify. Each formalism is an idealization, often times approximating in its own DIFFERENT ways, each offering somewhat different coverage of the knowledge domain. (Changing what science is and how it's done, Bull. Am. Physical Soc., March 2009 meeting, Amer. Physical Soc., R. Jones)
We don't deal in "laws" but rather in arguments. Even mathematical proofs come in varying degrees of certainty. None are simply true or false. A formal grammar has the same power as a formal logic. We do not have a formal grammar which encompasses all of natural language, however, and the machinery of finite and continuous systems of mathematics DO make certain scientific reasoning easier.
Each "language" ( be it natural language, some formal language, some logic, etc.) makes it easy to say certain things but difficult to express others. One language can not necessarily be translated completely into some other arbitrary language. Concepts formalized in one logic/language may not exactly correspond to concepts formalized in another. Having MULTIPLE overlapping theories of a knowledge domain is then better than having just one theory (written in one language) If you have more languages at your command you can think more thoughts.