last changed in 2002
In WG it is assumed that most of language, and indeed most of knowledge, is learned on the basis of experience rather than being innate (inborn).
This is possible because knowledge consists of a vast network of concepts, each of which is ultimately based on experience. Some concepts are based directly on experience; for example, the concept Fido (a particular dog) is obviously based on direct experience of Fido and includes all Fido's idiosyncratic characteristics. However when Fido is combined in a conceptual network with other individual dogs it is possible (and easy) to extract their shared features and to build a new concept Dog which has all these features - the prototypical dog. A more abstract concept can be extracted from dogs, cats and budgerigars: Pet, and similarly for other super-categories such as Mammal or even Animal. All of these concepts can be learned - i.e. built on the basis of experience, even if the relationship between the concept and the relevant experiences is quite indirect.
This kind of 'inductive' learning is made easier by the fact that all concepts are organised in isa hierarchies. If two existing concepts have features in common, the learner's mind can easily create a new super-concept which extracts this generalisation.
However it is of course likely that some concepts are innate. This must be true for those that correspond directly to sensory experiences (e.g. colours), and for the basic isa relationship, but it may also be true for some relationships such as Number and Vertical.
As far as language is concerned, there is very little that cannot be learned in this way. However one interesting suggestion is that we learn language on the basis of just one innate relationship: 'symbol' (the relationship between an observable and what it symbolises). This is what allows us to link a word to its sense and its referent. It is also clear that some of our physical apparatus (e.g. for speaking and for hearing speech) is innate - see chapter 10 of Deacon - for interesting evidence - so it is an open question how much of the rest of language is innate and how much is learned.
This is an article from the Encyclopedia of Word Grammar and English grammar. If you refer to it, please give the url as "http://tinyurl.com/wg-encyc".