Grammatical knowledge, or as some label it linguistic knowledge, entails the ability to produce certain sounds that have certain meanings and to understand the sounds made by others. Also, a further widely accepted definition of grammatical knowledge is the subconscious internalized knowledge of language structure and rules that help learners generate communicative utterances, momentarily analyze and comprehend received ones, and respond appropriately (Gass & Selinker, 1994). In other words, it's the ability to produce and comprehend proper communicative utterances in conversations.
Chomsky argued that children learn language not by habit formation but by acquisition a set of rules or grammar. This grammar will have a finite number of rules, but will be capable of generating an infinite number of well-formed sentences. Most of these sentences are new to our experience. This linguistic knowledge must have a generative capacity. In other words, children do not learn and reproduce a large set of sentences, but they routinely create new sentences that they have never learnt before. This is only possible because they internalize rules rather than strings of words; extremely common examples of utterances, such as "it breaked" or "mummy goed" Show clearly that children are not copying the language around them but applying rules. The task of the linguist, he claimed, is to describe this universal human ability, known as language competence, with a grammar from which the grammars of all languages could be derived. The linguist would develop this grammar by looking at the rules children use in hearing and speaking their first language. He termed the resulting model, or grammar, a transformational-generative grammar, referring to the transformations that generate language. (Chomsky, N. (1986)
When language use is considered as communication, the concepts of input, comprehensible input, and comprehensible output are appropriate metaphors because they call for images of messages (Swain & Lapkin, 1998). Corder, in (1976) made an important distinction between what is available to the learner to learn (input) and what has become part of his/her procedural knowledge (intake). What is available to the learner to learn does not count as part of his/her grammatical knowledge until it is integrated in the learner's current inter-language system. Thus, it's not enough to know about rules, lexemes, and sounds, but rather to be ready to use them whenever the learner is engaged in actual speech events. Furthermore, Sorace (1993a, 1993b, and Brad, Roebrtson, & Sorace, 1996), argued that there are two kinds of changes which occur learner's grammars: discontinuous and continuous. What Sorace interpreted from looking at data from learners of Italian was a differentiation in terms of input use with regard to auxiliary selection. She claims that it is possible for the input, or what she calls the evidence available to the learner, to have a varying effect depending on the part of grammar to be affected - more so for lexical semantics and les so for syntax.
Some grammatical structures can be learned explicitly while others may only be taught implicitly through interaction because even if they can be logically justified, they are still not used by him/her