Which approach analyzes language as a network of pattern activations similar to neural networks?

Which approach analyzes language as a network of pattern activations similar to neural networks?

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

Answer: (B)

The main idea being tested is modeling language as distributed patterns of activation across many simple units, similar to how neural networks operate. This is the hallmark of the connectionist/parallel distributed processing approach. Instead of storing explicit grammatical rules, language knowledge emerges from patterns of activation spread through a network, with learning occurring by adjusting the strengths (weights) between units as exposure accumulates. This allows the system to generalize to new sentences, handle noisy input, and produce graded, probabilistic responses—key features of how a neural-network–inspired model processes language. This viewpoint contrasts with rule-focused theories that posit an innate set of grammatical rules or structures. For example, some theories emphasize an innate language faculty or universal grammar, or focus on transformations between deep and surface structures rather than distributed activations. By capturing language as a web of interacting activations shaped by experience, the connectionist/parallel distributed processing approach best fits the description.

The main idea being tested is modeling language as distributed patterns of activation across many simple units, similar to how neural networks operate. This is the hallmark of the connectionist/parallel distributed processing approach. Instead of storing explicit grammatical rules, language knowledge emerges from patterns of activation spread through a network, with learning occurring by adjusting the strengths (weights) between units as exposure accumulates. This allows the system to generalize to new sentences, handle noisy input, and produce graded, probabilistic responses—key features of how a neural-network–inspired model processes language.

This viewpoint contrasts with rule-focused theories that posit an innate set of grammatical rules or structures. For example, some theories emphasize an innate language faculty or universal grammar, or focus on transformations between deep and surface structures rather than distributed activations. By capturing language as a web of interacting activations shaped by experience, the connectionist/parallel distributed processing approach best fits the description.