---
title: 'Natural language processing (NLP)'
description: 'The field of building computational systems that analyze, retrieve, translate, or generate human language in text or speech.'
canonical_url: 'https://darkfactory.dev/glossary/natural-language-processing'
markdown_url: 'https://darkfactory.dev/glossary/natural-language-processing.md'
collection: glossary
date_published: '2026-08-05T00:00:00-04:00'
date_modified: '2026-08-26T00:00:00-04:00'
---

# Natural language processing (NLP)


## Definition

Natural language processing is the field of building computational systems that work with human language in text or speech. Its tasks include tokenization, search, classification, information extraction, translation, summarization, question answering, speech recognition, and text generation.

Language has structure at several levels. A useful system may need to handle sounds or characters, word forms, syntax, meaning, references across sentences, conversational context, or some combination of them. The task determines which distinctions matter. A sentiment classifier and a translation system both belong to NLP, but they solve different problems and require different evaluations.

## How the field developed

NLP grew across computational linguistics, machine translation, speech processing, information retrieval, and artificial intelligence. The sources reviewed do not support crediting the name to one person. Earlier systems often relied on hand-written rules and dictionaries. Later work added statistical models learned from corpora, followed by neural networks, transformers, and large language models.

That history matters because NLP is broader than its current dominant model family. Rules remain useful for controlled formats. Retrieval may be the core of a search product. A smaller classifier can be a better fit than an LLM when the labels and operating conditions are stable.

## Distinguish it from nearby terms

- **Computational linguistics** studies language with computational methods and overlaps heavily with NLP. The label often signals more attention to linguistic analysis, while NLP often emphasizes building and evaluating systems.
- A **large language model** is one model class used for many NLP tasks. It does not define the field.
- A **chatbot** is an application. It may use an LLM, retrieval, rules, speech components, or several of them together.
- Fluent output does not establish factual grounding, reasoning, or human-like understanding. Those claims need their own evidence.

## Operational significance

"Uses NLP" is too broad to serve as a requirement. Name the language, task, input conditions, output contract, and evaluation target. Accuracy can vary with dialect, domain, document length, noise, and language. A system that performs well on polished English may fail on customer shorthand, legal text, code-switching, or a lower-resource language.

## Check your understanding

A vendor says its product "understands documents." Ask whether it classifies, retrieves, extracts fields, answers questions, or generates summaries, then test that specific task on representative documents.

## Also called

NLP, natural language processing

## Related terms

- [Large language model (LLM)](https://darkfactory.dev/glossary/large-language-model)
- [Transformer](https://darkfactory.dev/glossary/transformer)
- [Embedding](https://darkfactory.dev/glossary/embedding)
- [Causal language model](https://darkfactory.dev/glossary/causal-language-model)

## Evidence and further reading

- [Speech and Language Processing](https://web.stanford.edu/~jurafsky/slp3/)
- [Andreessen Horowitz AI Glossary](https://a16z.com/ai-glossary/)
- [Stanford HAI Artificial Intelligence Glossary](https://hai.stanford.edu/ai-definitions)
- [MIT Sloan Generative AI Basics Glossary](https://mitsloanedtech.mit.edu/ai/basics/glossary/)