Which approach emphasizes learning language through exposure to patterns and statistical regularities rather than relying on a prewired grammar?

Which approach emphasizes learning language through exposure to patterns and statistical regularities rather than relying on a prewired grammar?

• A. LAD
• B. Innateness Hypothesis
• C. Transformational Grammar
• D. Connectionist/Parallel Distributed Processing

Answer: (D)

This question focuses on learning language from patterns and statistical regularities in the input rather than depending on an already preformed grammar. The best fit is the connectionist/Parallel Distributed Processing approach. It uses neural-network-style systems that learn directly from lots of language data, adjusting connections based on how often and in what contexts words and morphemes co-occur. Over time, the network picks up patterns—like common word orders, suffixes, or phrase structures—without being told explicit rules, and it can generalize to new, unseen sentences. This emphasis on learning from exposure and distributional information is what makes it align with how we think language can emerge from experience rather than from an innate, prewired grammar. The other views rely more on built-in grammatical knowledge. A Language Acquisition Device proposes an innate system that guides learning, while the Innateness Hypothesis argues for a universal grammar already in place. Transformational Grammar centers on transforming underlying deep structures into surface forms through rules, implying predefined grammatical knowledge guiding syntax. None of these prioritize learning strictly from patterns and statistical regularities in the input in the same way as the connectionist perspective.

This question focuses on learning language from patterns and statistical regularities in the input rather than depending on an already preformed grammar. The best fit is the connectionist/Parallel Distributed Processing approach. It uses neural-network-style systems that learn directly from lots of language data, adjusting connections based on how often and in what contexts words and morphemes co-occur. Over time, the network picks up patterns—like common word orders, suffixes, or phrase structures—without being told explicit rules, and it can generalize to new, unseen sentences. This emphasis on learning from exposure and distributional information is what makes it align with how we think language can emerge from experience rather than from an innate, prewired grammar.

The other views rely more on built-in grammatical knowledge. A Language Acquisition Device proposes an innate system that guides learning, while the Innateness Hypothesis argues for a universal grammar already in place. Transformational Grammar centers on transforming underlying deep structures into surface forms through rules, implying predefined grammatical knowledge guiding syntax. None of these prioritize learning strictly from patterns and statistical regularities in the input in the same way as the connectionist perspective.