How grammar is acquired: a UG based dynamic model
Within the generative paradigm, several hypotheses have been developed to account for the acquisition of grammar in language learners. Beyond the principled question of how the acquisition of a language is possible at all given the complexity of the knowledge attained vis-a-vis the impoverished nature of the input provided (Logical Problem of Language Acquisition, Chomsky 1986), issues that need to be addressed in a model of language development concern the nature of learners grammars and the changes they undergo.
The Principles and Parameters model
Following the model of language acquisition implicit in the Principles and Parameters theory (Chomsky 1981) changes in learner grammars concerning parameter- ised grammatical properties are commonly conceived of as punctual or instantaneous events: language-specific properties of FCs are assumed to be triggered by positive data in the linguistic environment (input). Against the backdrop of the logical problem of language acquisition, this idealisation is necessary in order for linguistic theory to provide a principled account of how language acquisition is possible at all given the underdetermination by input data (Chomsky 1986: xxv). The developmental process and the question of how to account for the transition of one developmental stage to the next (developmental problem) are irrelevant to this model. A theory of language development, however, is faced with the task of providing an explanation of why there is something time-consuming in the development of grammars, why the succession of developmental stages takes the form it does, and how transitions between one stage and the next should be conceived of.
Several hypotheses have been proposed to account for these issues in the area of developmental linguistics, such as the Maturation hypothesis (according to which the developmental sequence is maturationally constrained), the Continuity hypothesis (which maintains that the full structure is available to the learner from the onset of the learning process), and the Structure-building hypothesis (which claims that structure is built progressively in accordance with the input) (cf. Pla- za-Pust 2000 for a detailed discussion). In an earlier work (Plaza-Pust 2000) we discussed the main claims and shortcomings of these hypotheses and concluded that the structure-building hypothesis accounts best for the development grammars because it does not relegate the developmental problem to a non-linguistic area (perception or maturation, as is the case of the continuity and maturation hypothesis respectively). This hypothesis is briefly sketched in the following.
-  Notice that the distinction between the notions of triggered and learned is crucial to the modelat hand: structural knowledge is assumed to be triggered, that is, guided by universally determined language-specific learning mechanisms. Idiosyncratic properties of the target language,in contrast, have to be learned.