Truncation is a search technique that lets you search for multiple forms of a word at once by entering only the root of that word followed by a special symbol, usually an asterisk (*). Instead of running separate searches for “child,” “children,” and “childhood,” you type child* and the database pulls results containing all of those variations. It’s one of the most useful tools for anyone searching academic databases, and it takes about 30 seconds to learn.
How Truncation Works
The idea is simple: you identify the unchanging root of a word, chop off the ending, and replace it with a truncation symbol. The database then matches every term in its index that begins with that root. For example, searching genetic* returns results containing “genetic,” “genetics,” and “genetically.” Searching therap* picks up “therapy,” “therapies,” “therapist,” and “therapeutic” in a single query.
This is sometimes called “stemming” because you’re working with the stem of a word. The key decision is where to cut. Cut too far into the word and you’ll get flooded with irrelevant results. Cut too little and you miss useful variations. If you searched cat* hoping to find articles about cats, you’d also get “catalyst,” “category,” “catastrophe,” and hundreds of other unrelated terms. A better choice would be to just search “cats” or use a longer root like catal* if catalysis is what you’re after.
Truncation Symbols by Database
The asterisk (*) is the most common truncation symbol. It’s the standard in PubMed, EBSCO databases (including CINAHL and SPORTDiscus), and many other major platforms. But not every database uses the same conventions.
Web of Science, for instance, uses three different symbols. The asterisk (*) represents any group of characters, including no character at all. The question mark (?) represents exactly one character, which is useful when you want to capture British and American spellings like “organi?ation” (matching both “organisation” and “organization”). The dollar sign ($) represents zero or one character, so colo$r matches both “color” and “colour.”
The question mark and dollar sign are technically called wildcard characters rather than truncation symbols, but they work on the same principle: substituting a symbol for unknown letters. The practical difference is that truncation goes at the end of a word to capture different endings, while wildcards typically go inside a word to capture spelling variations.
Right-Hand, Left-Hand, and Internal Truncation
Standard truncation, where you place the symbol at the end of a word root, is called right-hand truncation. This is the most widely supported type. Searching neuro* captures “neurology,” “neurological,” “neuroscience,” and so on.
Left-hand truncation places the symbol at the beginning: *glycemia would match “hyperglycemia” and “hypoglycemia.” Internal truncation places it in the middle: wom*n matches both “woman” and “women.” These options are less commonly available. Web of Science supports both internal wildcards and left-hand truncation in selected search fields, but many databases only support right-hand truncation. You’ll need at least one character before the symbol in most databases, and some fields require at least three characters before a truncation symbol.
The Precision vs. Recall Tradeoff
Truncation increases what researchers call “recall,” which is the proportion of relevant results you actually find. Without it, you might miss articles that use a different form of your search term. An article about “therapeutic interventions” won’t show up if you only searched “therapy.” Truncating to therap* solves that problem.
The cost is reduced “precision,” meaning a higher percentage of your results may be irrelevant. You’re casting a wider net, so more junk comes in with the good stuff. This tradeoff matters most in systematic reviews and other research where you need to be both comprehensive and efficient. For a quick literature scan, the extra noise is usually manageable. For a formal systematic review, you’ll want to test your truncated terms and review what they actually return before committing to them.
A good rule of thumb: keep your root word long enough to stay meaningful. Three or four letters is often too short unless the root is highly specific to your topic.
How Truncation Affects PubMed Searches
PubMed deserves special attention because it’s the most-used biomedical database and its truncation rules have quirks that catch people off guard. Normally, when you type a term into PubMed, the system runs it through automatic term mapping. It tries to match your term to standardized Medical Subject Headings (MeSH) and “explodes” those headings, meaning it also searches for all narrower terms underneath them.
Adding an asterisk turns all of that off. Truncation bypasses PubMed’s translation tables entirely and searches the All Fields index directly. So if you search breast neoplasm*, PubMed finds citations indexed with “Breast Neoplasms” but does not automatically include narrower terms like “Mammary Neoplasms” the way it would for an untruncated search. Similarly, truncating a “See Reference” term like breast cance* won’t map to the official MeSH heading “Breast Neoplasms” at all, because the mapping process never kicks in.
PubMed also caps truncation at 150 variations. If the root you enter generates more than 150 possible word endings in the index, PubMed uses only the first 150 and displays a warning. This rarely matters for long, specific roots, but it can be an issue with very short stems.
Truncation in Ovid Databases
Ovid-based databases, commonly used to search MEDLINE and Embase, have their own complication. By default, Ovid tries to map truncated terms to subject headings. If subject heading mapping is turned on when you use truncation, the system may throw an error message instead of running your search. Experienced searchers either turn off mapping before using truncation or use the database’s specific syntax to search keyword fields directly.
Practical Tips for Effective Truncation
- Test before committing. Run your truncated search and scan the first page or two of results. If you’re seeing a flood of unrelated terms, lengthen your root word.
- Combine with Boolean operators. Truncation works well alongside AND, OR, and NOT. Searching therap* AND adolescen* captures multiple word forms for both concepts while keeping results focused.
- Don’t truncate everything. Some words don’t have useful variant endings, or the database already searches for common variants automatically. PubMed, for instance, handles some plurals on its own when mapping is active.
- Know your database. Check whether your platform uses an asterisk, a question mark, or some other symbol. Look up whether left-hand or internal truncation is supported. A few minutes reading the help documentation saves hours of frustration.
- Watch for minimum character requirements. Some databases require at least three characters before a truncation symbol in certain fields. Searching an* may not work where anx* would.
Truncation is a small technique with outsized impact on search quality. When used thoughtfully, it closes the gaps that rigid keyword matching leaves open, helping you find relevant research that would otherwise slip through.