How to start a literature search: what to read first
By Mubarak MutesasiraUpdated 9 min read
Short answer:
Start a literature search by defining a provisional question, finding a relevant review, and using it to identify the vocabulary, foundational papers and active disagreements. Then read recent studies and follow references in both directions. Keep a search log from the start so you can explain how you found the evidence and notice what the initial search may have missed.
For a new topic, the immediate goal is orientation: learning enough to search intelligently and judge relevance. A systematic review has a different standard of completeness and documentation. This guide explains the first steps and how the Research reading list can help, within the wider route to publishing a research paper.
Step 1: Turn the topic into a provisional question
A topic such as remote learning is too broad to guide useful reading. Add the group or setting, the phenomenon or intervention, and what you want to understand. You do not need a final research question yet; you need enough direction to recognize a relevant paper.
For example, “remote learning” might become “How do university students describe barriers to participating in online seminars?” That wording points towards participation, experience and higher education rather than every study of educational technology.
Write down the purpose of the search as well. Are you choosing a dissertation question, planning a method, writing an introduction or deciding whether a practical intervention has evidence? The same topic will produce different reading priorities for each task.
Keep a note of assumptions in the initial question. Perhaps you assume a barrier is technical when recent work treats it as organizational. Early reading should be allowed to change the question. Do not force every source into the framing you happened to start with.
Step 2: Build a small vocabulary map
List the core concepts and likely alternatives. Include abbreviations, older terms, spelling variants and phrases used in neighbouring fields. Separate concepts from synonyms so you can combine them clearly later.
| Concept | Possible search language in an illustrative search |
|---|---|
| Setting | university, higher education, undergraduate |
| Activity | online seminar, virtual classroom, synchronous learning |
| Issue | participation, engagement, barriers, access |
| Perspective | student experience, perceptions, interviews |
These terms are examples to test, not a complete search strategy. Read titles and abstracts to see which terms authors actually use. A term with an everyday meaning may have a narrower technical meaning in the field.
Avoid narrowing too soon with every detail of your intended study. A query requiring the exact population, location, method and outcome can miss useful evidence expressed differently. Begin broad enough to discover terminology, then refine based on what you learn.
Step 3: Read a useful review first
A relevant review can show the main approaches, debates and sources in a field. Check its scope and search dates before treating it as a current map. A recent publication can still summarize a search that ended considerably earlier.
Read the review's purpose and methods. Is it systematic, scoping, narrative or another form? What populations, designs, languages or dates did it include? Those boundaries explain what the review can and cannot tell you.
Use it to build questions rather than simply borrowing its conclusions. Which studies carry the main claims? Where do authors disagree? Are there methods or definitions you do not yet understand? Add those questions to your reading notes.
If the topic has no useful review, start with a well-matched empirical paper and inspect its references, then compare the framing with another paper. For a structured introduction to organizing the work, see literature review tools.
Step 4: Find foundational papers without worshipping citation counts
Foundational papers introduce a theory, method, dataset or result that later work repeatedly uses or debates. A highly cited paper can be a useful lead, but citation count is not a quality score and does not establish that the original conclusion remains accepted.
Ask why the paper is cited. It may provide a measurement instrument, a definition, a result later challenged, or simply a widely used reference point. Read the relevant section of the original before citing it for a particular claim.
Older papers have had more time to accumulate citations. Newer studies, smaller fields and research in less widely indexed languages can be important without large counts. Balance influence with relevance to your question and with later evidence.
When a foundational method is central to your project, read enough detail to understand its assumptions and limits. A summary in a review may be sufficient for orientation but not for choosing an analysis or defending an operational definition.
Step 5: Add recent work that updates the map
Search beyond the review's last search date and look for studies that test, refine or challenge the main claims. Recent work can show changes in measurement, context or available evidence. It can also reveal that a supposedly open question has already been addressed.
Do not assume the newest paper is the strongest. Examine its design, evidence and relationship to earlier work. A small new study may add a useful perspective without overturning a larger evidence base.
Read conflicting findings deliberately. Check whether they differ in population, definition, method, setting or analytical choices. A disagreement can reflect genuinely different questions rather than one paper being right and the other wrong.
Use publication dates as one clue, and inspect version information where relevant. A preprint, accepted manuscript and final article may represent versions of the same work. Avoid counting them as independent studies or citing an outdated version without reason.
What Research reading list does
Type a topic into Research reading list. It finds review articles, the most-cited foundational papers and recent work from the scholarly index, then orders them into “read first,” “then” and “to go deeper,” with a reason for each. The papers are real records you can verify.
The ordering helps you begin when a search result list gives you no sense of sequence. A review may supply vocabulary, a foundational paper may explain a central method, and recent work may show how the discussion has changed. You still decide whether each item fits your question.
The tool costs 3 credits when signed in. Signed-out visitors can try a few free AI runs a day on tool pages. It does not provide a guaranteed exhaustive search, establish paper quality or replace access to and reading of the original sources.
Use a specific topic rather than a single broad noun. Include the population, problem or method if you know it. If the results reveal a different vocabulary, revise the topic and compare what changes. Keep the useful records and your reasons for selecting them.
Step 6: Read in passes rather than from first page to last every time
First assess relevance: title, abstract, question, design and main finding. If the paper does not address the issue you need, record why and move on. Save detailed reading for sources that shape your argument or method.
For a relevant paper, inspect the methods and results before accepting the discussion's interpretation. Check who or what was studied, how the key concept was measured and what comparisons support the conclusion. Read tables and figures with their captions.
Then ask what role the paper will play in your work. Is it background, a methodological precedent, evidence for a claim, a competing explanation or a limitation to investigate? Writing this role down prevents a folder of PDFs from masquerading as understanding.
For a foundational or contested source, read closely enough to explain its reasoning. Efficiency means matching effort to purpose, not relying on abstracts for claims that require full-text evidence. If you cannot access a needed paper, use your library's legitimate access routes or look for an authorized repository version.
Step 7: Follow references backward and citations forward
Backward searching means inspecting a useful paper's reference list. It helps identify the studies and concepts the authors rely on. Forward searching means finding later papers that cite it, which can reveal applications, critiques and updated evidence.
Use both routes. Backward searching alone can keep you inside an older conversation; forward searching alone may miss foundational work that current authors assume readers know. Keep returning to your question so citation trails do not expand indefinitely.
Watch for clusters that repeatedly cite one another. A single network is not necessarily the whole field. Search alternative terminology and neighbouring disciplines, especially when the topic spans methods or professions.
A scholarly index is a discovery source, with coverage limitations. For a substantial project, choose appropriate disciplinary databases with library support. The deep research tool can support further exploration, but a report still needs source verification and a documented search when the task requires one.
Step 8: Keep a search log and evidence notes
Record the database or index, search date, query, filters and useful results. Include why a query was changed. You will otherwise struggle to reconstruct why a paper appeared in your notes or whether you already tested a synonym.
For each source you read closely, keep a compact evidence record:
- Full reference and a stable identifier or source link.
- The question and study design.
- Population, setting or material studied.
- The finding relevant to your question.
- Limitations affecting how you can use it.
- Page or section for a quotation or specific claim.
- Your interpretation, clearly separate from the authors' words.
Distinguish direct quotations from paraphrases immediately. Store source locations before drafting. The Citation checker can help identify reference problems later, but it cannot recover the meaning of a passage you never recorded accurately.
A worked first-reading sequence
Invented example: a student is exploring barriers to online seminar participation. Their first search is dominated by studies of course grades, which do not directly answer the experience question. A relevant review supplies terms about belonging, interaction and access.
The student reads that review, then a foundational measurement paper it relies on. They discover that engagement is measured differently across studies. They add recent interview research and a contrasting survey, noting that each provides a different kind of evidence.
The resulting reading plan is not “read everything about online learning.” It is a sequence: learn the main concepts, understand a key measure, compare recent findings, then identify what remains unanswered for the intended population. The student records each search and revises the provisional question accordingly.
This example illustrates a workflow rather than recommending specific studies. A generated reading list should point you to verifiable papers for your actual topic, not supply a fictional bibliography that merely looks plausible.
Know when orientation must become a formal search
An introductory reading list can support topic selection or early framing. It is not enough if you intend to claim a comprehensive evidence synthesis. Systematic and scoping reviews need explicit methods appropriate to their purpose, including search sources, eligibility rules, screening and documentation.
Read systematic review search strategy before treating exploratory queries as a review method. For choosing between review types, see scoping review versus systematic review. Involve a librarian early when search quality is central to the study.
Even for an ordinary research-paper introduction, avoid saying that no study exists solely because a starter list did not include one. Test alternative terms, check recent work and state the boundary of your claim. The research gap finder can suggest directions to investigate, not certify absence.
Decide what to read next
At the end of each reading session, write down the main question you can now answer and the one you still cannot. Choose the next source because it addresses that gap in understanding. This gives you a reason to stop collecting unrelated papers.
You have a useful initial map when you can explain the vocabulary, main approaches, important disagreements and the evidence relevant to your purpose. Return to the search when the question changes or new work appears. Start with Research reading list, then build a documented, source-checked search around what you learn.