If this question has gone unanswered for a while, that isn't because it is an uninteresting one. On the contrary! It raises a whole range of deep and difficult issues that have been the subject of a vast amount of discussion (from cognitive psychologists as well as arm-chair philosophers) for years. So I hesitate to plunge in. But still, since no one else has responded yet, let me get the ball rolling -- though these remarks are no more than a very preliminary sorting out of some of the issues. For we need to clarify what is meant here by (1) "language", (2) "thought", and (3) "tied up together". (1) What is meant by "language"? A shared natural language like English, or Welsh, or Sanskrit? Or might we more generously count as a language any system of representations which has a syntax (i.e. there are structural rules determining which arrays of elements from system are allowed) and a semantics (there are rules determining what these arrays mean )? Some have argued that we have an innate ...

Well, it is certainly true that introducing unambiguous, very carefully defined, agreed terminology and having a perspicuous notation can make reasoning easier and make us less prone to common reasoning errors. To take the obvious example, mathematicians aren't just being awkward when they use a lot of symbolism and make very careful distinctions wrapped up into technical terms (and borrow from the languages of formal logic to make clear, for example, the 'scope' of their quantifiers). If proofs all had to be written out in unaugmented English, then we'd get lost following them, even in elementary high school algebra: and proof-discovery would be orders of difficulty harder. I suppose we might say "mathematicians' English" -- meaning English augmented with their new definitions and notational devices -- is a new, better, language, more suited to (mathematical) reasoning than street English. But equally, we might say that it is just one part of a single inclusive language, modern English: it is just a...