Abstract:We argue that natural language can be usefully described as quasi-compositional and we suggest that deep learning-based neural language models bear long-term promise to capture how language conveys meaning. We also note that a successful account of human language processing should explain both the outcome of the comprehension process and the continuous internal processes underlying this performance. These points motivate our discussion of a neural network model of sentence comprehension, the Sentence Gestalt model, which we have used to account for the N400 component of the event-related brain potential (ERP), which tracks meaning processing as it happens in real time. The model, which shares features with recent deep learning-based language models, simulates N400 amplitude as the automatic update of a probabilistic representation of the situation or event described by the sentence, corresponding to a temporal difference learning signal at the level of meaning. We suggest that this process happens relatively automatically, and that sometimes a more-controlled attention-dependent process is necessary for successful comprehension, which may be reflected in the subsequent P600 ERP component. We relate this account to current deep learning models as well as classic linguistic theory, and use it to illustrate a domain general perspective on some specific linguistic operations postulated based on compositional analyses of natural language. This article is part of the theme issue ‘Towards mechanistic models of meaning composition’.

Computational Simulation of Language Comprehension

Language Cognition and Language Computation—human and Ma-Chine Language Understanding

Rule-Based and Word-Level Statistics-Based Processing of Language: Insights from Neuroscience

Neural Substrate Underlying the Learning of a Passage with Unfamiliar Vocabulary and Syntax.

Language Cognition and Language Computation -- Human and Machine Language Understanding

The Role of Linguistic Structure in Sentence and Text Comprehension: A Comparative Analysis of Depth-First Versus Breadth-First Models

Cognitive Computing on Chinese Sign Language Perception and Comprehension

Psycholinguistic Studies on Sentence Parsing in Language Comprehension

Towards a computational paradigm and technological approach to cognitive linguistics

Flexible and fine-grained simulation of speed in language processing

A Brief Introduction to Computational Linguistics

A review of contemporary theories of sentence comprehension

Connectionist Models of Speech Perception

A Review of Studies on the Mechanism of Memory in Language Comprehension

Machine Comprehension of Spoken Content: TOEFL Listening Test and Spoken SQuAD

A Computational Complexity Perspective on Segmentation as a Cognitive Subcomputation

An information-theoretic model of shallow and deep language comprehension

Neural correlates of semantic-driven syntactic parsing in sentence comprehension

Sentence comprehension: A parallel distributed processing approach

Quasi-compositional mapping from form to meaning: a neural network-based approach to capturing neural responses during human language comprehension

Computational Models to Study Language Processing in the Human Brain: A Survey