When you search a library catalogue for “heart attack” but the relevant books are all indexed under “myocardial infarction,” you walk away empty-handed even though the information exists. This mismatch between the words people use and the words a system stores is one of the oldest problems in information retrieval. The solution that librarians and indexers developed is called vocabulary control. It is the quiet machinery that decides which words a system will recognise, and it explains why a well-built catalogue feels almost effortless to search while a poorly built one feels like guesswork.
Table of Contents
- What is vocabulary control?
- Natural language versus controlled language
- Objectives of vocabulary control
- Ensuring consistency
- Reducing ambiguity
- Improving retrieval
- Methods of vocabulary control
- Controlling synonyms and near-synonyms
- Distinguishing among homographs
- Selecting and defining terms
- Examples of controlled vocabulary
- Thesauri
- Subject heading lists and authority lists
- Classification schemes
- Why vocabulary control still matters
What is vocabulary control?
Vocabulary control is the process of developing, maintaining, and using a controlled vocabulary. A controlled vocabulary is a carefully selected list of authorised terms that an indexer must use to describe the subject of a document, and that a searcher uses to retrieve it. Instead of letting indexers pick whatever words feel natural, the system fixes the choices in advance. In the vocabulary of an indexing language, one concept should be denoted by only one term, and that single term is applied consistently to every document on the topic.
The need for control comes directly from how natural language behaves. Human language is rich, flexible, and ambiguous, which is wonderful for poetry but disastrous for retrieval. According to the principles set out in standard thesaurus design literature, the need arises from two basic features of natural language: two or more different words can describe the same concept, and a single word can describe two or more different concepts. Left unmanaged, both features scatter related documents across the catalogue and pull unrelated ones together.
Natural language versus controlled language
To see the difference, consider the same idea expressed by three different authors: “old age homes,” “elder care facilities,” and “geriatric residential services.” A free-text search engine treats these as three separate things. A controlled vocabulary, by contrast, decides that one of them, say “Old age homes,” is the preferred term, and quietly redirects the other two to it. The searcher who types any of the three is steered to the same set of results. This is why controlled vocabularies are sometimes described as suppressing the anarchy of natural language in order to describe what a document is actually about.
It is worth being clear that control does not mean restriction for its own sake. The non-preferred terms are not thrown away. They are kept in the system as entry points that guide a user from the word they happened to think of toward the word the system uses.
Objectives of vocabulary control
The objectives of vocabulary control all serve a single larger goal: making information storage and retrieval reliable. Three objectives stand out as the most important.
Ensuring consistency
Consistency is the foundation. Without it, two indexers, or even the same indexer on two different days, might describe identical documents using different terms. The result is a catalogue where documents on one subject are split across several headings, and no single search ever gathers them all. Vocabulary control enforces a principle of uniform heading, where one term is selected from among the synonyms and applied to every document on that topic. The Wikipedia overview of controlled vocabularies frames this neatly: such systems reduce the ambiguity inherent in natural language and ensure consistency by creating a clear link between each concept and its preferred term.
Reducing ambiguity
The second objective is to remove the confusion caused by words that look the same but mean different things. The word “bank” can mean a financial institution or the side of a river. The word “pool” can mean a body of water or a game. If left alone, these words drag unrelated documents into the same search results. Vocabulary control resolves this by distinguishing among homographs, often by attaching a clarifying qualifier in parentheses, so that the system separates the concepts cleanly.
Improving retrieval
The third objective is the payoff for the first two. When terms are standardised, a retrieval system can match a search query against indexed records with far greater precision. The primary purpose of vocabulary control, as the established thesaurus design guidance puts it, is to achieve consistency in describing content and to facilitate retrieval. A controlled system filters out irrelevant results and, just as importantly, surfaces relevant results that a literal word-match would have missed. For a student doing project research, this is the difference between finding three sources and finding thirty.
Methods of vocabulary control
How is control actually achieved? The classic answer is that vocabulary control rests on three principal methods, all aimed at taming the quirks of natural language.
Controlling synonyms and near-synonyms
The first method handles the problem of many words for one idea. Synonyms (“car” and “automobile”), near-synonyms (“salary” and “wages”), variant spellings, and different word forms are all linked together through what is called the equivalence relationship. One term is chosen as the descriptor, or preferred term, and the rest become non-descriptors. In a thesaurus, this is recorded with standard abbreviations: USE points from a non-preferred term to the preferred one, and UF (Used For) points back the other way. So a thesaurus might record “Automobiles UF Cars,” telling the indexer to always choose “Automobiles” while still letting a searcher who typed “Cars” reach the right place.
Distinguishing among homographs
The second method tackles the opposite problem: one word, several meanings. Here the technique is to qualify the term so each preferred heading refers to only one concept. The standard example is the word “Cells,” which might mean biological cells, electric cells, or prison cells. Indexers resolve this by adding a parenthetical qualifier, producing distinct headings such as Cells (biology) and Cells (prison). A scope note can also be attached to a term, deliberately stating what the term does and does not cover, so that there is no doubt about how it should be applied.
Selecting and defining terms
The third method is term selection, and it is more subtle than it first appears. The indexer must decide on the right level of specificity. Should a topic be represented as “Knowledge management software” as a single heading, or broken into separate terms? Charles Cutter’s old rule of specific entry still guides this work: a document is entered under the most specific term that describes it, not under a broader class that merely includes it. The Sears List, for instance, instructs catalogers to prefer a precise heading such as “Dinners” over a broader one like “Banquets” when the specific concept fits.
Term selection also has a procedural side. A controlled vocabulary can be built by the deductive method, where experts survey an entire field and list its terms in advance, or by the inductive method, where terms are added only as documents needing them arrive. In practice, as the ScienceDirect overview of controlled vocabularies notes, a combination of the two methods is usually applied. This is also why maintaining a controlled vocabulary is genuinely demanding work: every new term typically has to be reviewed by subject and technical experts before it joins the list.
Beyond equivalence and the handling of homographs, a mature controlled vocabulary also records two other kinds of links between terms: the hierarchical relationship, which connects broader terms (BT) and narrower terms (NT), and the associative relationship, which connects related terms (RT) that are neither broader nor narrower but conceptually close. These relationships turn a flat list of words into a navigable map of a subject.
Examples of controlled vocabulary
Vocabulary control is not an abstract theory; it is embodied in real tools that libraries and databases use every day. These tools fall into a few broad families.
Thesauri
A thesaurus is the most fully developed vocabulary control tool. It is an organised list of terms from a specialised field, arranged to help both the indexer and the searcher choose the same preferred terms. The international standard for information retrieval thesauri, ISO 25964, defines a thesaurus as a controlled and structured vocabulary in which concepts are represented by terms, relationships between concepts are made explicit, and preferred terms carry lead-in entries for their synonyms. A well-known example is MeSH (Medical Subject Headings), maintained by the United States National Library of Medicine and used to index the world’s biomedical literature, including the PubMed database that medical students and researchers in India rely on daily.
Subject heading lists and authority lists
Subject heading lists are the controlled vocabularies most familiar to library users. The Library of Congress Subject Headings (LCSH) is the most comprehensive, having become a de facto standard for cataloguing in academic and large research libraries. For smaller public and school libraries, the Sears List of Subject Headings, first published by Minnie Earl Sears in 1923, offers a simplified alternative to LCSH with shorter, more commonly understood headings. Both are used widely across Indian libraries.
Closely related is the authority file, a controlled vocabulary that records variants and synonyms for each term as cross-references guiding the user from a non-preferred form to the preferred one. As the literature on metadata best practices explains, an authority file functions as cross-references and is used especially for proper names. An author authority file, for example, ensures that works by a writer who has published under different spellings or initials are all gathered under one standardised name, so the catalogue does not fracture a single author into several.
Classification schemes
Classification schemes are a different species of controlled vocabulary, expressed in notation rather than words. The Dewey Decimal Classification (DDC) is the classic example, and the librarianship literature points out that the DDC is a notational vocabulary, with all notational vocabularies being controlled by definition. A classification number such as 385 for railroad transportation fixes both the meaning and the scope of a topic. Schemes built on faceted principles, an approach developed in Indian library science through the work of S. R. Ranganathan, take this further by recognising that a subject has multiple dimensions, or facets, that can be combined systematically.
Why vocabulary control still matters
It would be easy to assume that powerful full-text search engines have made vocabulary control obsolete. The opposite is closer to the truth. Free-text searching is fast and cheap, but it is at the mercy of the exact words an author chose, and it cannot tell a financial bank from a riverbank. Controlled vocabularies remain costly to build and maintain, because human experts must review terms and keep the list current as terminology shifts. Yet that human judgement is precisely what delivers the precision and the gathering power that automated matching alone cannot. In specialised fields such as medicine and law, where using the wrong term can mean missing critical information, controlled vocabularies are not a luxury but a necessity.
What do you think? If you were building a controlled vocabulary for a fast-changing field like artificial intelligence, how would you decide when a brand-new term has become stable enough to add as a preferred heading? And where would you draw the line between the precision of controlled vocabulary and the convenience of free-text search?
References
- https://egyankosh.ac.in/bitstream/123456789/35770/6/Unit-10.pdf
- https://marciazeng.metadataetc.org/Z3919/1need.htm
- https://www.librarianshipstudies.com/2020/03/controlled-vocabulary.html
- https://en.wikipedia.org/wiki/Controlled_vocabulary
- https://www.sciencedirect.com/topics/computer-science/controlled-vocabulary
- https://en.wikipedia.org/wiki/Thesaurus_(information_retrieval)
- https://libguides.ala.org/catalogingtools/subjects
- https://www.lisedunetwork.com/what-is-the-sears-list-of-subject-headings/
- https://link.springer.com/article/10.1057/dam.2010.29
- https://www.librarianshipstudies.com/2017/03/vocabulary-control.html

Leave a Reply