How to Track Terminology Changes Across Decades of Research

A term that names your topic today may have carried a different meaning, or no meaning at all, thirty years ago. Tracking terminology across decades means building a search and documentation method that captures those shifts instead of assuming a single keyword spans the whole period. The practical answer is to treat terminology as a variable: define eligibility first, run free-text and subject-heading searches side by side, log how each database translated your terms, and keep a dated trail of every change you make.

This guide lays out a complete sequence with checkpoints, a worked example, a reusable log, and clear rules for when to stop or escalate. It is written for researchers revising a manuscript, closing a project, handing off a lab, or starting a new review where the vocabulary itself is part of the evidence.

The Cochrane Handbook's guidance on searching for and selecting studies starts from eligibility criteria, not from a word list. It recommends planning searches from those criteria, using multiple sources, combining free-text terms with subject headings, documenting the process, deduplicating records, screening in stages, linking multiple reports of one study, and rerunning searches before publication. Each of those steps has a terminology dimension: subject headings change over time, free-text usage drifts, and the same study may appear under several labels.

That is why a single saved query is not enough. You need a record of what each term meant in each source at the time you searched, so a reader or coauthor can reconstruct your reasoning.

Before touching a database, write down where you actually are. Most terminology-tracking problems trace back to one of a few starting conditions.

• You have a modern keyword and nothing else. You know the current label but not its predecessors.
• You have an old term and need to reach the present. You are extending a historical search forward.
• You have a mixed pile of records collected over years with inconsistent labels and no log.
• You are revising a manuscript and a reviewer questioned whether your search covered earlier vocabulary.
• You are handing off a project and the next person needs to reproduce your search without you.

Name your condition in one sentence at the top of your working document. That sentence becomes the scope statement your later decisions must satisfy.

Start where Cochrane starts: eligibility criteria. Write the population, intervention or exposure, outcome, and study-design boundaries in plain language. Then, for each concept, list the labels you expect to encounter.

A useful checkpoint here is a concept-to-label map: one row per concept, with columns for the current label, suspected historical labels, and the source where you will test each. You do not need to confirm every historical label yet; you need to know which ones you will test.

If you are starting from scratch and need candidate papers to seed that map, a scholarly search interface is the natural first move. WisPaper search lets you find academic papers through a scholarly search interface, which is enough to begin collecting the labels authors actually use.

Cochrane's recommendation to combine free-text terms with subject headings is the core mechanic of terminology tracking. Free-text searching catches the words authors wrote; subject headings catch the controlled vocabulary a database applied. When a field's vocabulary shifts, the two diverge, and that divergence is your signal.

PubMed's User Guide is helpful here because it documents field tags, phrase searching, filters, Advanced Search, saved searches, Similar Articles, citation lookup, and links to full text when available. Critically, PubMed's search details reveal how terms were translated and combined. That translation view is your audit window: it shows what the database did with your phrase, which is exactly the record you need when a term's meaning is in question.

Checkpoint: for each concept, save both the free-text string and the subject-heading string, and capture the translation details for each. If the database translated your term into something you did not intend, log it.

Documentation is where most terminology projects quietly fail. Cochrane explicitly recommends documenting the search process, and that recommendation scales directly to multi-decade work.

Build a query log with one row per search event. At minimum, record the source, the date, the exact string, the subject headings used, how the source translated the terms, and the number of records returned. The translation column is the one people skip and later regret.

Google Scholar's Search Help documents searching by author and title, date sorting and date limits, Cited by and Related articles, citation export, library access links, email alerts, and ways to locate alternative versions of a work. Those features support a different job than a controlled-vocabulary database: they help you move forward and backward through a citation network and find alternate versions of a work that may carry different labels. Use them as a complement, and log them in the same table so your trail stays unified.

Cochrane recommends deduplicating records and linking multiple reports of one study. Terminology drift makes both steps harder, because the same study can appear under an early label in one record and a later label in another.

When you deduplicate, keep the variant labels attached to the merged record rather than discarding them. A merged record that preserves its label history is far more useful downstream than a clean record with no provenance.

Cochrane's staged screening model fits terminology work well. Screen titles and abstracts first, then full texts, and at each stage record why a record was included or excluded, including any judgment about whether an older label refers to the same concept.

This is where a shared workspace earns its place. If you are organizing the records you keep, My Library lets you save or upload papers and organize them in the WisPaper library, so the label history you are building stays attached to the papers themselves rather than scattered across local folders.

Checkpoint: every included record should carry at least one label variant and a note on which concept it maps to. If a record has no label note, it is not ready.

Once papers are gathered, terminology questions become comparative: did authors in the 1980s use this term the same way as authors in the 2010s? Answering that by rereading every paper is slow.

If your papers are in My Library or added from search results, Library QA lets you ask questions based on those papers. That supports a targeted check: pose a question about how a specific term is used and compare the answers against the passages you have already logged. Treat the answers as pointers back into the source text, and confirm each one against the paper before it enters your log.

A terminology log is only as good as the references attached to it. When your log feeds a bibliography, screening the BibTeX against real academic databases catches references that do not correspond to real works. TrueCite supports BibTeX citation screening for exactly this step, flagging hallucinated references before they reach a manuscript or a handoff document.

Checkpoint: every citation in the log resolves to a real work, and every work in the log has a label note. Both conditions must hold before you call the trail auditable.

Imagine a lab that studied a phenomenon under an older name for two decades; the field later adopted a new label. A departing researcher must hand the search to a successor.

1. Scope statement: "Reproduce the lab's full literature base on [concept], covering both the historical and current labels."
2. Concept-to-label map: current label, two suspected historical labels, and the sources to test each.
3. Parallel searches: free-text and subject-heading strings for each label, with translation details captured.
4. Query log: one row per search event, dated, with translation notes.
5. Deduplication: merged records retain all label variants.
6. Staged screening: inclusion notes state which label each record used and why it maps to the concept.
7. Corpus questions: targeted checks on how each label was used across decades, confirmed against source passages.
8. Citation screening: the BibTeX for the handoff document is checked before transfer.
9. Rerun: searches are rerun before the handoff document is finalized, per Cochrane's recommendation to rerun before publication.

The successor can now reconstruct every decision from the log alone.

The trail is ready when a reader who has never seen your project can follow the scope statement through the query log to the included records and out to the screened citations, without asking you a question. Run that test literally: hand the log to a colleague and see where they stall.

Stop and escalate when any of the following occurs:

• A term's meaning is genuinely contested in the sources. Do not resolve it silently; record the disagreement and flag it for a coauthor or reviewer decision.
• A database translation changes your intended meaning. Log it and revise the string rather than accepting the translation.
• A citation does not resolve. Remove it from the log until it is confirmed.
• A label cannot be mapped to a concept. Keep it in a separate unresolved list rather than forcing a mapping.
• The scope statement no longer matches the searches. Revise the statement first, then the searches.

For longer efforts that span many sources and need a structured literature review, Scholar Agent fits WisPaper's Scholar Agent area and its literature review, inspiration, or reproduction use cases. If the work is organized as a defined effort with its own boundaries, Research Projects is the WisPaper Research Projects area where you can create a new project to hold it.