TBT and Neurocognitive linguistics

Research and Theory
1

Hi! In a previous post I’ve commented about the idea of use Monty in linguistics tasks. This topic is just to clarify the basis of this idea, in case someone is interested.

In 1991 Sidney Lamb wrote a book titled “Pathways of the Brain”, about neurocognitive linguistics. Sidney Lamb’s neurocognitive linguistics is a theory that integrates linguistic structure with neurological processes. His Relational Network Theory (RNT) models language as a network of interconnected processing units (nections), rather than using traditional rule-based grammar.

The strong claim: “Understanding language, even at higher levels, is basically a perceptual process”.

Key Points of Lamb’s Theory:

  1. Relational Networks: Instead of hierarchical syntax trees, linguistic knowledge is represented as a vast, interconnected network of associations between linguistic elements.
  2. Nections: The fundamental unit in the network, nections act as nodes that process and store linguistic relationships. They are hypothesized to correspond to cortical columns in the brain.
  3. Distributed Representation: Language is not stored as fixed symbols but as activation patterns across a network, making processing more parallel and associative.
  4. Pathways and Activation: Understanding and producing language involve activating specific pathways within the network, similar to spreading activation in connectionist models.
  5. Learning and Change: Language learning involves modifying connections and strengthening or weakening pathways based on experience, akin to neural plasticity.
  6. Cognitive and Neurological Basis: The model aligns with findings in neuroscience, aiming to explain language processing in a biologically plausible way.
2 Likes
2

Representation of Concepts in the Relational Network

  1. Concepts as Patterns of Activation:

    • A concept (e.g., dog) is not a single entity but a pattern of activation across many nections, each contributing different aspects (e.g., phonological form, semantic attributes, syntactic roles).
    • The activation of a concept is context-dependent—different pathways can be activated depending on how the concept is used (e.g., dog as a pet vs. dog as a derogatory term).

  2. Hierarchical and Associative Structure:

    • A concept consists of lower-level nections encoding sensory-motor features, and higher-level nections encoding more abstract properties.
    • Concepts are linked by associative pathways, meaning related concepts (e.g., dog and animal) share overlapping networks of activation.
3

I think some reasons Lamb’s theory has not received much attention is the lack of knowledge about the neurological part, a strong opposition to the mainstream mindset in Linguistics and a lack of a computational implementation to test the ideas and claims.
Although it is a old site, the main ideas can be found here:

I’m not sure if the claims about cortex is updated (although he used Mountcastle as basis, as TBT).

4

Interesting. I can def see the simularities here. I think the biggest differences are going to be found in each frameworks approaches to cognition; embodied vs relational. Its kind of fun to envision how sensorimotor functioning leads to linguistic construction, though. Either way, very cool stuff!

3 Likes
5

Hi @HumbleTraveller
thanks for reading and commenting :slight_smile:
Not so directly related to TBT, but with useful ideas: