Language - Computational Explorations in Cognitive Neuroscience - page 316

Chapter 10

Language

10.7Sentence-LevelProcessing.............365
10.7.1 BasicPropertiesoftheModel........367
10.7.2 ExploringtheModel.............370
10.7.3 SummaryandDiscussion ..........375
10.8Summary ......................376
10.9FurtherReading ..................3 77

Contents

10.1Overview ......................323
10.2 The Biology and Basic Representations of Lan-
guage ........................325
10.2.1 Biology ...................325
10.2.2 Phonology ..................327
10.3 The Distributed Representation of Words and
Dyslexia.......................329
10.3.1 Comparison with Traditional Dual-Route
Models ...................330
10.3.2 The Interactive Model and Division of Labor 331
10.3.3 Dyslexia...................331
10.3.4 BasicPropertiesoftheModel........333
10.3.5 ExploringtheModel.............335
10.3.6 SummaryandDiscussion ..........341
10.4TheOrthographytoPhonologyMapping.....341
10.4.1 BasicPropertiesoftheModel........343
10.4.2 ExploringtheModel.............344
10.4.3 SummaryandDiscussion ..........349
10.5 Overregularization in Past-Tense Inflectional
Mappings......................350
10.5.1 BasicPropertiesoftheModel........352
10.5.2 ExploringtheModel.............353
10.5.3 SummaryandDiscussion ..........357
10.6 Semantic Representations from Word Co-
occurrencesandHebbianLearning........358
10.6.1 BasicPropertiesoftheModel........360
10.6.2 ExploringtheModel.............361
10.6.3 SummaryandDiscussion ..........365

10.1 Overview

Language plays an important role in many cognitive
phenomena, and represents a major point of divergence
between human cognition and that of even our clos-
est animal relatives. The nature of the relationship be-
tween cognition and language (e.g., does one require
the other?) has been much debated. Neural network
models can provide a useful perspective in this debate,
instantiating mechanistic principles that suggest a rich
and interesting relationship between cognition and lan-
guage.
Specifically, language can be viewed as a set of spe-
cialized but interacting processing pathways in percep-
tual, motor, and association areas. These pathways in-
teract with other pathways in these areas, via learned as-
sociations, to produce human cognition. For example,
verbal input can activate perceptual and nonverbal mo-
tor representations, and vice versa. Over time, learning
is shaped by influences both of language on cognition
and of cognition on language.
The issue of how words are represented in the brain
is obviously central to understanding language, and it
provides a central organizing theme of this chapter. The

323