Children acquire parameters rapidly by exposing them to input and observing which parameter values produce grammatical sentences. Positive evidence (hearing grammatical sentences) can trigger parameter-setting, but negative evidence (hearing ungrammatical forms) is rare; parameter-setting is driven largely by exposure to positive input. Cross-linguistic studies show that children set parameters by stages, suggesting that some parameters depend on others or that certain parameter combinations require particular input.
Examine longitudinal acquisition data for children learning languages with different parameter settings, showing how acquisition schedules relate to parameter theory. Consider what input is necessary and sufficient for setting each parameter.
You already understand that Universal Grammar provides children with an innate language faculty that constrains the range of possible grammars, and that parameters are the binary dimensions on which languages vary within those constraints — settings like whether verbs precede or follow their objects, whether null subjects are permitted, or whether wh-words must move to the front of a clause. Parameter-setting acquisition asks a more specific question: *how* does a child, exposed to the specific input of a particular language, converge on the correct settings? The answer is more intricate than flipping switches, and it illuminates why human language acquisition is so remarkable.
The central puzzle is the Poverty of the Stimulus. Children converge on complex grammatical rules without explicit instruction and without encountering all the sentences that would directly demonstrate those rules. A child learning English never hears every grammatical sentence of English — the class is infinite — yet generalizes correctly to novel sentences. The parameter model explains this through the interaction of innate structure and positive evidence: the grammatical sentences a child hears trigger parameter settings because the parameters are already built into the language faculty. When a child learning Japanese consistently hears verbs at the end of sentences, this exposure activates the [+head-final] setting for the relevant parameter. No explicit teaching is needed; the input triggers a switch that was already there.
Negative evidence — explicit correction for ungrammatical production — turns out to play a surprisingly small role. Studies of child-directed speech consistently show that parents rarely correct grammar (they correct factual content far more often). Children also persist in systematic errors like "I goed" or "she runned" for months or years even when corrections do occur — the internal grammar resists direct modification. This points to a key principle: the grammar is not acquired by imitation and reinforcement but through an internal process triggered by exposure. The Subset Principle addresses a further logical problem: when two possible parameter settings are both consistent with the input a child has heard so far, the child defaults to the more restrictive grammar — the one that generates fewer sentences. This avoids the trap of overgenerating and needing negative evidence to retreat; the child starts small and expands only when positive evidence forces it.
The developmental timing of parameter-setting reveals additional structure beyond simple exposure. Children do not set all parameters simultaneously; they set them in sequence, and the orders are often consistent across children acquiring the same language and even across typologically different languages. Some parameters appear to depend on others: the pro-drop parameter (whether subjects can be omitted, as in Italian *parla* for "she speaks") may require a prior acquisition of rich verbal morphology, since it is that morphology that licenses the null subject. This parameter interaction suggests that the acquisition process is not merely bottom-up pattern extraction from input but a structured developmental sequence in which earlier grammatical knowledge scaffolds the acquisition of later knowledge. Cross-linguistic comparisons — tracking when children acquiring different languages master specific constructions — have become one of the most productive empirical tools for testing whether parametric theory correctly predicts acquisition orders.