Languages are dynamic and permeable, and they need to be for their functions as language. The notion of a language in which every word is exactly definable and used in exactly the same way all the time quickly breaks down in the face of real world problems. Mercantile markets are forced to create pidgin-versions, institutions create jargonized versions, scientific inquiry and communication requires regularized naming-extension practices suitable to the inquiry. And, of course, human beings are inveterate figurators; we analogize, make metaphors, speak obliquely, in almost everything we do.
One of the ways we keep language flexible enough to be practically useful is by making use of other languages. One way we do this is by using other languages as a neologistic reserve; this is commonly done with classical languages, and English primarily uses Latin and Greek for this. In this use we borrow words with some regard for their meaning in the original language, but loosely for our own purposes. But sometimes we just borrow them directly. Two common ways of doing this are calques and loanwords.
A calque is a loan-translation. Given a word in another language, we just translate it into ours and use it in the same sense. When computers began to be common, Spanish calqued the English word 'mouse', to indicate the device, by translating it as the word for 'mouse' in Spanish, ratón, and this has happened in multiple other languages. This happens very often with easily identifiable compound words. The English word 'skyscraper' has been calqued into literally dozens of languages, each of which translates it by translating 'sky' and 'scraper'.
A loanword is a direct borrowing. Déjà-vu is in English a loanword from French. When the word is borrowed, it is often done in a very specific way, so sometimes the words will only carry over part of their original spread of meaning. This is very common with food terms.
These two methods of borrowing have led to an amusing bit, which is actually why I've been thinking about this. 'Calque' is a loanword from French, not a calque. 'Loanword' is a calque from German (the original word was Lehnwort), not a loanword.
In any case, I've often thought that these points of dynamism in language should be taken more seriously by philosophers of language than they usually are. Philosophers of language have tended to focus on the relatively static features of language, but it's these dynamic features that arguably show how language actually works.
