Computational Linguistics Overview

Principles

Definition

Computational linguistics is an interdisciplinary field concerned with the computational study of natural language, involving the application of computer science principles and methods to analyze, model, and process linguistic data. It investigates linguistic phenomena across various levels of analysis to develop algorithms and systems for understanding and generating human language.

Ontological type

Core Methods

Linguistic Levels

Key Applications

Key developments

Key Figures

Hybrid Formal-Statistical NLP era

Lotfi A. Zadeh [1] was a seminal figure in this era, with affiliations spanning the Massachusetts Institute of Technology [3] and the University of California, Berkeley [4]. In his 1975 papers The concept of a linguistic variable and its application to approximate reasoning—I [7] and The concept of a linguistic variable and its application to approximate reasoning—II [8], Zadeh articulated a framework that merges linguistic structure with approximate reasoning to handle uncertainty in interpretation. Lyn Frazier [2] was a central figure at Yale University [5] and South College [6] during this era. In her 1978 paper The sausage machine: A new two-stage parsing model [9], Frazier helped usher in parsing architectures that blend staged processing with empirical data, a development crucial for robust hybrid NLP parsing.

Probabilistic Structured Models era

Christiane Fellbaum [1] is a central figure in computational linguistics during this era, with positions at the University of Pennsylvania [3] and Princeton University [4]. Her key contribution during 1997–2009 includes coordinating WordNet: An Electronic Lexical Database [6], whose structured sense inventory and semantic relations anchored lexical resources used in sense modeling and multilingual pipelines. Jason Weston [2] contributed to the era through roles at the University of Washington [5] and Princeton University [4], culminating in the influential 2008 paper A unified architecture for natural language processing [7]. This work helped shape subsequent probabilistic and discriminative structured inference approaches across parsing and translation in this era.

Contextual Neural Pretraining era

Christopher D. Manning [1] is associated with Harvard University [3] and Stanford University [4] in this era. His key contributions include the development of Glove: Global Vectors for Word Representation [6], the Stanford CoreNLP Natural Language Processing Toolkit [7], and Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank [8], which advanced word representations, tooling, and hierarchical semantic modeling in contextual pretraining. Richard Socher [2] is associated with Stanford University [4] and University of Colorado Boulder [5] in this era. His key contributions include Glove: Global Vectors for Word Representation [6], Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank [8], and Better Word Representations with Recursive Neural Networks for Morphology [9], which advanced unsupervised pretraining and morphology-aware representation in contextual neural models.