Choose manuscript keywords that describe the work you actually did
Build a manuscript keyword shortlist from your actual research, test its meaning in search results, and reconcile it with the journal's instructions.
About 4 min read · Allow extra time for the practical checks
Choose keywords from the manuscript’s actual question, subject and approach, then test whether those terms lead to work a reader would reasonably compare with yours. Keep the target journal’s instructions beside the list. A fashionable term is a poor choice when the paper does not study it.
This is a guide to author-supplied manuscript keywords. If you are collecting literature, the guide to building a search string addresses a different task.
Start with the manuscript and the journal
Read the journal’s current instructions for the permitted number, wording and submission fields. Taylor & Francis recommends checking journal-specific requirements before drawing up and testing a shortlist. Do not turn one publisher’s example or a recent paper’s list into a universal keyword count.
Then write a plain sentence explaining what your own paper examines. Identify its main subject, the central relationship or process and any setting or approach needed to distinguish it. Keep the sentence beside the candidate list so you can test every proposed term against something the manuscript actually contains.
An informative research-paper title and a keyword list should point toward the same work. They have different jobs, however: a title makes a readable statement, while the keyword field holds a limited set of useful labels. Do not force every candidate into the title.
Keep a reason for each candidate
The following example is entirely fictional. Imagine a manuscript about staff accounts of appointment scheduling in community libraries, based on interviews. It does not evaluate an algorithm or measure the effects of an intervention.
| Candidate | Decision for this fictional manuscript | Reason to record |
|---|---|---|
| Appointment scheduling | Keep for the shortlist | Names the process examined |
| Community libraries | Keep for the shortlist | Identifies the setting |
| Staff perspectives | Test wording | Describes whose accounts the paper examines; check the language used in relevant literature |
| Interviews | Consider alongside more specific alternatives | Names the approach, but may be less discriminating than the main subject terms |
| Artificial intelligence | Exclude | The manuscript does not examine an AI system |
| Improved efficiency | Exclude unless the manuscript supports that framing | An attractive outcome label would imply an evaluation absent from this example |
This is a decision exercise, not a recommended final keyword list. No searches have been performed for this fictional paper, and no measured demand is implied. The point is to make inclusion accountable: “popular topic” is not a substitute for a connection to the work.
Elsevier’s guidance suggests balancing broader and more specific terms and testing their relevance through searches. A specific phrase can identify your subject well; an extremely local label may need a more recognisable companion. Decide from the actual manuscript and the journal’s requirements rather than treating specificity as either always good or always bad.
Test the meaning of the search results
Search promising candidates in a suitable scholarly discovery service. Read enough of the results to decide whether they concern a comparable subject and use the words in the sense you intend. Record the service and your observation if it changes the decision.
For the fictional library paper, a search dominated by hospital scheduling would prompt a setting check. Results about software optimisation would prompt a check of whether the phrase wrongly implies a technical evaluation. Neither observation would prove that a keyword is forbidden; it would tell you what ambiguity to investigate.
Do not select a term merely because its results are numerous or because the authors of unrelated highly cited papers use it. A test search helps you examine fit. It does not establish search volume, promise database indexing or predict citations.
Reconcile the final list before submission
Read the final keywords against the title, abstract and manuscript. Confirm that the same concepts have stable names and that no label makes the study broader, more causal or more technical than it is. Remove redundant candidates when they consume space without adding a useful distinction.
Finally, check the journal’s limit and enter the approved wording consistently in the manuscript and submission system where requested. If the form specifies a controlled list, follow that instruction rather than forcing your own terms into an incompatible field. Keep your candidate record as a working aid; submit the final list in the journal’s required form.
