Questions: Parameter Setting in Language Acquisition

Question 1

Studies show that parents rarely correct their children's grammatical errors, yet children acquiring English reliably stop producing forms like 'I goed' and 'she runned' over time. What does this reveal about parameter setting?

• A. Children learn grammar through reinforcement; their errors must be corrected by other speakers in the environment even if parents do not
• B. Grammar is not acquired through imitation and correction — the child's internal language faculty converges on correct settings through exposure to positive input, not through explicit feedback on errors
• C. Children outgrow these errors naturally due to general cognitive maturation, independent of linguistic input
• D. The errors correct themselves because children compare their output to memorized exemplar sentences

Answer: (B)

Children persist in systematic errors like 'I goed' for months or years even when corrections occur. The grammar is internally driven: children are not building rules by imitating what they hear (they produce forms they have never heard), nor by responding to corrections (which have little effect). The parameter model explains this as an innate language faculty that converges on correct settings through positive evidence — grammatical input triggers the appropriate parameter values.

Question 2

A child learning a language is exposed to input consistent with two possible parameter settings. According to the Subset Principle, the child will:

• A. Randomly select one of the two consistent settings and wait for disconfirming evidence
• B. Default to the more restrictive setting — the one that generates fewer sentences — to avoid overgenerating and needing negative evidence to retreat to a smaller grammar
• C. Default to the more permissive setting to maximize communicative range, then retreat if corrections occur
• D. Request clarification from caregivers until the ambiguity is resolved

Answer: (B)

The Subset Principle resolves an acquisition logic problem: if the child defaults to the more permissive grammar (which generates more sentences), it will produce utterances that are grammatical in one language but not the other — and negative evidence (correction) is rare and unreliable. By defaulting to the more restrictive grammar, the child never overproduces, and positive evidence (hearing sentences outside the restrictive grammar) can safely trigger the switch to the more permissive setting.

Question 3

Children learning different languages show consistent orderings in when they acquire specific constructions, suggesting some parameters depend on other parameters being set first.

• T. True
• F. False

Answer: True

Cross-linguistic acquisition data reveal parameter interaction effects. For example, the pro-drop parameter (licensing null subjects as in Italian 'parla' for 'she speaks') appears to require prior acquisition of rich verbal morphology — because it is that morphology that licenses the null subject. This means acquisition is not random or purely input-driven but follows a structured developmental sequence where earlier-acquired grammatical knowledge scaffolds later parameter settings.

Question 4

Parameter setting in language acquisition is driven primarily by negative evidence — when children hear corrections about ungrammatical utterances, they adjust their parameter settings accordingly.

• T. True
• F. False

Answer: False

Negative evidence plays a surprisingly minor role in parameter setting. Studies of child-directed speech show parents rarely correct grammar (they correct facts far more often). Children also persist in systematic errors for months after corrections, showing the internal grammar is resistant to direct modification. Parameter setting is driven by positive evidence — exposure to grammatical sentences of the target language triggers the appropriate settings. The Subset Principle is designed precisely to avoid dependence on negative evidence.

Question 5

What is the 'Poverty of the Stimulus' problem in language acquisition, and how does the parameter model offer a solution?

Model answer: The Poverty of the Stimulus problem: children converge on complex grammatical rules that go beyond the input they receive. They produce and correctly interpret sentence types they have rarely or never heard, generalize to novel sentences in ways that respect subtle grammatical constraints, and never make certain types of errors that would be predicted by simple pattern extraction. No general learning algorithm operating on raw input can explain this — the input is too sparse and too ambiguous. The parameter model's solution: children do not learn grammar from scratch. Universal Grammar provides an innate inventory of possible parameter settings, and exposure to the target language simply triggers the correct settings from this pre-specified menu. The child isn't inferring grammar; they're selecting among pre-built options.

Poverty of the Stimulus is a logical argument, not an empirical claim about input frequency. Even abundant input would underdetermine the grammar because the critical evidence (grammatical vs. ungrammatical forms) is not reliably present. The parameter model resolves this by making the relevant generalizations innate rather than learned.