03 · What You Need to Know
Source Provenance Can Disappear Long Before You Write the Citation
Researchers often think of citation as something that happens during writing. By then, however, the more difficult problem may already have occurred.
You read a paper. You highlight a passage. You paraphrase it into your notes. Later, you copy the paraphrase into a thematic note. You combine it with evidence from two other papers. Months afterward, you use the resulting idea in a draft.
At each step, information moves farther from the document in which you originally encountered it.
A reliable literature-review workflow therefore preserves provenance throughout the movement from source to synthesis.
Think of Provenance as a Chain
For an ordinary source-derived idea, the chain might look like this:
Source The identifiable article, book, chapter, report, or other work.
Location The page, section, paragraph, table, figure, or other place where the relevant material appears.
Source-derived note Your quotation, paraphrase, summary, or record of the finding.
Analytical connection The theme, question, claim, or argument to which you connect the note.
Synthesis The interpretation you eventually develop across one or more sources.
You should be able to travel backward through this chain. If you reach an idea-level note but cannot identify the source from which one of its claims came, part of the chain has broken.
Record a Source Identifier With Every Source-Derived Note
Do not create orphan notes such as:
Students may rely on AI feedback even when it contains errors.
Instead, preserve enough information to identify the source. This might be an author-year citation, citation key, reference-manager link, DOI, database identifier, or another stable connection to your bibliographic record.
You do not necessarily need to repeat the complete reference beside every sentence. What matters is that the identifier unambiguously points to the correct source.
If you maintain a consistent record for every important paper, source identifiers can connect individual notes back to that authoritative record rather than duplicating full bibliographic information throughout your system.
Record the Location Before You Close the Source
Knowing which paper contains an idea is helpful. Knowing where in a 30-page paper the relevant evidence appears is much better.
For paginated sources, record page numbers for direct quotations and for important paraphrases, definitions, findings, or claims you expect to revisit. For sources without stable page numbers, use another meaningful locator when available, such as a section heading, paragraph, table, figure, appendix, or timestamp appropriate to the source.
This serves two purposes. First, it makes later verification faster. Second, it helps you recover the original context rather than relying on a compressed note.
Source identified
I know which paper this idea came from.
Source traceable
I can quickly return to the specific part of the paper needed to verify the idea and its context.
Mark Direct Quotations Immediately
A copied sentence is particularly dangerous when its status is not recorded.
Imagine pasting a useful passage into your notes without quotation marks. Months later, you edit the surrounding notes and no longer remember whether the wording is yours. The sentence may eventually enter your manuscript as though you composed it.
Whenever you copy exact wording, mark it as a quotation at the moment of capture and preserve the appropriate locator. If you later paraphrase it, keep enough provenance to verify the paraphrase against the source.
APA Style distinguishes direct quotation from paraphrase and requires direct quotations to be attributed to the original author with an appropriate locator. The practical lesson for note-taking is straightforward: preserve that information before the source disappears from your immediate working context.
A Paraphrase Still Needs a Source Connection
Putting an idea into your own words does not change where the idea came from.
A paraphrase may be more vulnerable to source loss than a quotation because it no longer visually resembles the source. Without a citation marker, it can begin to look like your own thought.
This is why the distinction between what the authors said and what you inferred yourself should be recorded during note-taking rather than reconstructed during drafting.
A simple entry might contain:
PARAPHRASE — Author, year, p. X: [your paraphrase]
The particular notation is optional. The provenance is not.
Do Not Let Thematic Reorganization Strip Away Citations
Source information is relatively easy to preserve while notes remain organized by paper. Everything inside a paper note has an obvious parent source.
The risk increases when you reorganize material by idea.
Suppose you copy notes from six papers into a document called Student Trust in AI Feedback. If you move only the statements and leave their citations behind, the thematic note may become analytically useful but evidentially fragile.
Watch Out
Whenever source-derived material moves from a paper-level record into a theme, claim, outline, matrix, or draft, move its source connection with it. Never assume that proximity to another citation will be enough to reconstruct provenance later.
A workflow that combines paper-level traceability with idea-level synthesis can reduce this problem substantially.
One Idea Note May Need Several Sources
An idea may not have one source because it emerged from your comparison of several sources.
For example:
My synthesis: Students' willingness to act on AI-generated feedback appears to depend partly on whether they can evaluate its reliability.
If that interpretation emerged from Studies A, B, C, and D, record those sources with the synthesis. Do not arbitrarily attach the idea to whichever paper you happened to read last.
You might record:
SYNTHESIS — based on: Study A [locator]; Study B [locator]; Study C [locator]; Study D [locator].
This preserves an important distinction. The individual evidence belongs to the sources. The cross-source interpretation belongs to you.
Be Careful When a Paper Attributes the Idea to Someone Else
One of the easiest ways to lose the original source occurs when you encounter an idea in a paper that itself cites an earlier work.
Suppose Paper A states that a particular theoretical principle was introduced by Paper B.
You have several possibilities:
- you are interested in how Paper A interprets the principle;
- you need the original formulation in Paper B; or
- you need evidence that Paper A itself produced.
These are not interchangeable.
If your claim concerns what Paper B originally argued, the preferable approach is generally to retrieve and read Paper B when it is reasonably accessible rather than relying on Paper A's representation of it. APA Style guidance on secondary sources similarly advises citing primary sources when possible and using secondary-source citation when the original work is unavailable or inaccessible.
Source you read
The work in which you encountered the claim.
Original source of the claim
The earlier work to which the claim is attributed and which you may need to consult before representing it directly.
Do Not Cite an Original Source as Though You Read It When You Did Not
Following a citation chain is useful. Pretending you followed it is not.
If you encounter an important claim attributed to an earlier work but cannot obtain that original source, follow the requirements of your citation style for secondary citation rather than placing the original work in your reference list as though you consulted it directly.
In APA Style, for example, a secondary-source citation identifies the original work in the text but cites the secondary source that was actually read, while only the secondary source appears in the reference list.
The details vary among citation systems, so verify the requirements of the style you are using.
Distinguish Original Empirical Evidence From Later Retellings
Citation chains do not involve only theoretical ideas.
A review article may state that an intervention improved performance by a certain amount and cite an earlier empirical study. A later paper may repeat the same claim. Eventually, several sources may appear to support the statement even though they all trace back to the same original dataset.
If the claim matters to your argument, follow the citation trail far enough to determine where the empirical evidence actually originated.
This prevents a form of accidental evidence inflation in which multiple papers repeating the same finding are mistaken for multiple independent findings.
Keep Source, Evidence, and Interpretation as Separate Fields When Useful
A structured note can make provenance explicit:
| Field |
Example Function |
| Source |
Identifies the paper you actually read |
| Locator |
Points to the relevant page, section, table, or figure |
| Evidence / Claim |
Records what the source reports or argues |
| Original source cited? |
Flags whether the paper attributes the idea to an earlier work |
| Original source checked? |
Records whether you retrieved and examined that earlier work |
| My interpretation |
Preserves your analytical response separately |
| Theme / Argument |
Connects the evidence to your emerging synthesis |
You do not need all these fields for every note. They are useful when provenance is likely to become complicated.
Use Citation Keys, Links, or Identifiers Rather Than Depending on Filenames
A filename such as paper-final-2.pdf is a weak source identifier.
Stable identifiers can reduce ambiguity. Depending on your workflow, this may include a DOI, reference-manager item, citation key, or another persistent record identifier.
The point is not to choose one particular technology. It is to ensure that your note points to a source record that will still make sense after folders are reorganized or files renamed.
Your Literature Matrix Should Preserve Traceability Too
Compression inside a matrix can make provenance disappear surprisingly quickly.
A cell containing:
AI literacy increases critical evaluation.
is much less useful if you cannot determine where that statement came from.
If a matrix entry records an important finding or claim, retain a source identifier and, when appropriate, a page, table, figure, or other locator. A matrix should help you compare studies without becoming detached from them.
This complements the principle of building a literature matrix that supports analysis rather than merely storage.
Do Not Assume Your Future Self Will Remember
The most dangerous source-tracking system is memory.
When you have just read a paper, you may recognize every sentence in your notes. After another fifty papers, several months, and three rounds of reorganizing your literature review, that familiarity may disappear.
Record provenance for the researcher you will be six months from now, not the researcher who still has the PDF open.