01 · The Question
What if the literature uses words you did not think to search?
You know your topic. You know the terms commonly used to describe it. You enter those terms into a database and retrieve relevant papers. The search appears to work.
But authors do not necessarily use your vocabulary. The same concept may be described with synonyms, spelling variants, abbreviations, older terminology, newer terminology, technical labels, everyday language, or terms inherited from another discipline. Databases may also assign controlled subject headings that differ from the words authors use in titles and abstracts.
This means that a search can look successful while quietly excluding relevant studies for a simple reason: you asked the database to recognize only one way of talking about the problem.
03 · What You Need to Know
How do you build search vocabulary that can recognize the literature?
Start with concepts, not a bag of keywords
A literature search should not begin by generating every vaguely related word you can imagine. First identify the concepts that the search genuinely needs to represent.
Suppose your question concerns the experiences of university students using generative artificial intelligence for academic writing. Potential concepts might include the population, the technology, and the activity of interest. Each concept can then be represented using alternative terminology where necessary.
The logic is important. Alternative terms usually broaden how a single concept can be expressed, while different concepts are combined to focus the search.
Concept
Identify what the search needs to represent.
Alternative terms
List plausible synonyms, abbreviations, spelling variants, related labels, and relevant controlled vocabulary.
Test
Run the search and inspect relevant records.
Learn
Look for terminology and indexing you did not anticipate.
Revise
Add useful terms, remove unproductive ones, and retest the strategy.
Authors can describe the same concept differently
Terminological variation is normal scholarly behavior, not an inconvenience invented by databases. Different disciplines develop their own vocabularies. Terminology changes historically. New technologies acquire names faster than indexing systems can always stabilize them. Authors may prefer a broad concept, a specific subtype, an abbreviation, or a theoretical label.
A search using only the phrase you personally use may therefore retrieve a coherent subset of the literature while missing another coherent subset written in a different vocabulary.
Source of variation
What to look for
Synonyms
Different words expressing the same or closely corresponding concept
Abbreviations and full forms
Acronyms, initialisms, and their expanded terminology
Spelling and word-form variation
Regional spellings, singular/plural forms, hyphenation, and related forms where relevant
Historical terminology
Older labels that may appear in earlier research
Emerging terminology
New labels appearing in recent research before vocabulary becomes standardized
Disciplinary terminology
Different labels used by fields examining similar phenomena
Controlled vocabulary
Database-assigned subject headings used to index concepts
Text words and controlled vocabulary do different jobs
Many bibliographic databases provide a controlled vocabulary: a standardized set of subject terms used to describe and index records. Other databases may rely more heavily on text searching or use different indexing systems.
Text-word searching looks for the language appearing in searchable record fields such as titles and abstracts. Controlled-vocabulary searching attempts to retrieve records assigned a standardized concept regardless of the exact words used by the authors.
Text words
Search the terminology authors actually use in searchable parts of the record.
Controlled vocabulary
Search standardized subject terms assigned according to the database's indexing system.
These approaches can complement one another. JBI guidance states that search strategies should use both subject headings and text words where applicable.
Controlled vocabulary alone can be insufficient for newly published or newly emerging concepts, while text words alone may miss records expressed or indexed differently. The precise balance depends on the database and topic.
Known relevant papers are vocabulary mines
One of the most productive places to discover search terminology is the literature you already know is relevant. Examine titles, abstracts, author keywords, and database-assigned subject terms. Ask what language those records use that your original strategy does not.
JBI's search methodology explicitly incorporates this iterative principle. Its three-step approach begins with an initial limited search, followed by analysis of text words in titles and abstracts and the index terms used to describe relevant articles. Those terms then inform database-specific searching.
This is a useful reminder that search development is empirical. You do not have to predict the entire vocabulary of a field from your desk before looking at the literature.
Search terms should evolve when the literature teaches you something
Imagine that you begin by searching for generative AI . Relevant records reveal frequent use of large language model , LLM , specific technology categories, and other terminology. Some of those terms may deserve incorporation into the strategy, depending on the scope of the question.
The same process can reveal obsolete terminology in older research or discipline-specific terminology that your initial search ignored. This becomes especially important when you are deliberately trying to identify important older research or capture important recent research , because vocabulary can change substantially over time.
More keywords can improve recall, but careless expansion creates noise
Adding alternative terms can increase the chance of retrieving relevant records. It can also retrieve enormous quantities of irrelevant material if the added terms are ambiguous or only loosely connected to the intended concept.
The objective is therefore not maximum vocabulary. It is useful vocabulary.
Potential Advantages
Captures relevant papers using alternative terminology.
Reduces dependence on one disciplinary or historical vocabulary.
Can improve retrieval of records not described using your preferred term.
Provides resilience when terminology is inconsistent or evolving.
Potential Limitations
Broad or ambiguous terms may greatly increase irrelevant retrieval.
Unnecessary concepts can make the strategy harder to manage and reproduce.
Complex queries can introduce syntax errors or unintended logic.
Terms useful in one database may not transfer directly to another.
A good search therefore balances sensitivity, the ability to retrieve relevant material, against the practical consequences of retrieving irrelevant material. The appropriate balance depends on the purpose of the search. Missing an eligible study matters differently in a systematic review than in a preliminary scan intended to understand a topic.
Do not search every word in the research question
Researchers sometimes convert an entire research question into separate search concepts and require every concept to appear. This can make a search unexpectedly restrictive.
Some concepts may not be described consistently in titles, abstracts, or indexing. Others may be implicit in the study design or full text. Requiring an unstable concept can therefore exclude relevant records even when the central phenomenon is present.
The question is not “What words occur in my research question?” It is “Which concepts must the database reliably recognize for this search to work?”
Keyword diversity cannot solve database coverage
A beautifully constructed query still operates inside the information source being searched. If relevant journals or records are not represented there, vocabulary expansion cannot retrieve them.
That is why keyword development should be paired, where appropriate, with the separate question of whether you have searched beyond a single database . Search vocabulary determines what your query can recognize; database selection helps determine what literature is available to be recognized.
Do not let your preferred terminology predetermine the literature
Terminology can encode theoretical assumptions. Two research traditions may describe similar observations using different constructs because they understand the phenomenon differently. Searching only one vocabulary can therefore do more than miss papers. It can make one conceptual tradition appear to be the entire field.
This becomes especially important when trying to identify the literature that most matters to the question . Some of the most consequential sources may be precisely those that do not use the terminology you expected.
04 · A Practical Example
How one reasonable keyword can hide part of a literature
Hypothetical Example
Searching for student “AI dependence”
Suppose a researcher wants to investigate university students becoming overly dependent on generative AI for academic work. The initial search centers on phrases such as AI dependence and AI dependency .
The results are relevant but surprisingly sparse. While reading them, the researcher notices that other authors discuss closely related phenomena using terms such as overreliance , automation bias , or constructs associated with reliance on algorithmic recommendations. Some of these terms come from research traditions that predate widespread generative AI use.
The researcher does not indiscriminately add every related phrase. Instead, each candidate term is tested against the intended concept and eligibility criteria. Terms that retrieve conceptually relevant studies are incorporated where appropriate; terms that mostly retrieve unrelated forms of technological dependence are discarded.
Initial vocabulary
The search reflects the researcher's preferred wording.
Relevant records
Early papers reveal alternative terminology and related indexing terms.
Testing
Candidate terms are searched individually and in combinations to see what they actually retrieve.
Revision
Useful alternatives are incorporated while ambiguous terms are constrained or removed.
Result
The final strategy represents the concept through the language of the literature rather than through the researcher's vocabulary alone.
The lesson is not that every related term belongs in the search. It is that vocabulary should be tested against the literature rather than assumed in advance.
06 · What This Means for You
How should you develop and test your search terms?
Treat search vocabulary as a working model of how the literature talks about your concepts. Begin with what you know, test it against actual records, and allow relevant literature to correct your assumptions.
A simple decision framework
If a concept has obvious synonyms, abbreviations, or spelling variants
Test which alternatives retrieve relevant records and include useful variants where appropriate.
If the database uses controlled vocabulary
Inspect the subject headings assigned to known relevant records and consider combining appropriate headings with text words.
If relevant papers use terminology absent from your strategy
Test the newly discovered language and revise the strategy when it identifies relevant literature your existing terms miss.
If adding a term produces overwhelming irrelevant retrieval
Examine whether the term is too ambiguous, can be searched more precisely, or should be excluded.
If the topic spans disciplines or historical periods
Look deliberately for disciplinary and temporal changes in terminology rather than assuming current vocabulary is universal.
For formal evidence synthesis, document the final strategy sufficiently for others to understand what was searched. JBI guidance recommends recording search terms, databases, platforms, limits, languages, and search dates, while translating the strategy appropriately across databases.
The important habit is intellectual rather than technical: when the literature uses language you did not expect, do not merely learn the new term. Ask whether its absence from your search means you have also been missing the literature behind it.
07 · A Quick Checklist
Does your search recognize the different ways researchers describe your topic?
Before finalizing your search vocabulary, check:
I have identified the concepts that genuinely need to be represented in the search rather than mechanically searching every word in my research question.
I have considered relevant synonyms, abbreviations, spelling variants, and alternative labels for each major concept.
I have inspected titles, abstracts, author keywords, and indexing terms from known relevant papers for vocabulary I initially missed.
Where the database uses controlled vocabulary, I have considered appropriate subject headings as well as text words.
I have considered whether terminology differs across disciplines relevant to the question.
I have considered whether older and newer research use different terminology.
I have tested candidate terms rather than assuming every apparent synonym improves retrieval.
I have revised the search when relevant records revealed useful terminology that the original strategy did not contain.
09 · The Bottom Line
Do not make the literature speak only your language
The Bottom Line
A literature search should not depend on one preferred set of keywords because relevant researchers may describe, spell, abbreviate, conceptualize, or index the same phenomenon differently.
Build the search around concepts, learn vocabulary from relevant records, combine text words with controlled terms where appropriate, and revise the strategy when the literature exposes language you initially missed. The aim is not to collect the longest possible list of synonyms. It is to make sure that a difference in wording does not become an accidental reason for excluding important evidence.
11 · Cite this Guide
How to Cite This Guide
This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.
Recommended (Field Guide)
APA
MLA
Chicago
Copy Citation
