"Yuchtman and Seashore in 1967" is not a citation

Here is a sentence from a real literature review, lightly anonymised:

The construct was first operationalised by Yuchtman and Seashore in 1967, and refined considerably in the decades that followed.

Two surnames, a year, an attribution of intellectual work. Everything a citation has — except that it may not be one, and the difference matters both to a reader and to anything checking your document.

When is a mention a citation?

The rule is narrower than people assume, and it is not about whether you named someone.

A CITATION, narrative form
  Yuchtman and Seashore (1967) operationalised…
  → the year is in parentheses, marking a pointer
  → an entry must exist

A CITATION, parenthetical form
  …was first operationalised (Yuchtman &
  Seashore, 1967)…
  → same claim, different placement

PROSE
  …was first operationalised by Yuchtman and
  Seashore in 1967…
  → names people and a date
  → points at nothing
  → strictly, no entry is required

The distinguishing mark is the pointer, not the content. A citation is a formal undertaking that a corresponding entry exists in your reference list. Prose naming a researcher and a year makes a historical statement and undertakes nothing.

A citation is not "a mention of who did the work". It is a promise that the reader can find the work, and the punctuation is where the promise is made.

Five shapes that look like citations and are not

These are the recurring ones — worth recognising because they will appear in any report that reads your document carefully.

1  ATTRIBUTED PROSE
   "…as Okafor demonstrated in 2019…"
   → a name and a year, no pointer

2  AN INSTITUTIONAL ACTOR
   "…the Ministry announced in 2019 that…"
   → an organisation and a year, describing an
     event rather than citing a document

3  A NAMED PERIOD
   "…following the 2008 financial crisis…"
   → a year attached to an event

4  A QUANTITY
   "…in the larger cohort (n = 412)…"
   → parentheses and a number, and nothing to
     do with references

5  A CROSS-REFERENCE
   "…as shown in Table 3 (Chapter 4)…"
   → a pointer, but into your own document

Numbers three and four are the reason a checker cannot simply trust parentheses, and number two is the one that produces the most argument — because "the Ministry announced in 2019" often should carry a citation, and frequently does not.

Why a careful check surfaces these anyway

A reference checker that only recognised perfectly formatted citations would be easy to build and would miss the defects worth finding. Real documents contain citations with a missing bracket, a year outside the parentheses, an ampersand where a comma should be, and every other consequence of four years of editing.

So extraction is deliberately generous: it captures anything citation-shaped and sorts afterwards. The alternative — recognising only clean forms — silently drops exactly the malformed citations that most need attention, and a report that omits them looks better while being less useful.

OVER-CAPTURE
  prose is caught and shown as a candidate
  → you glance at it and dismiss it
  → cost: a few seconds

UNDER-CAPTURE
  a malformed citation is not recognised as one
  → it never appears in any report
  → it has no matching entry and nobody knows
  → cost: a defect that survives to submission

The first is visible and cheap. The second is
invisible and permanent.

Which is why candidates that turn out to be prose are not a failure of the check. They are the price of catching the citation someone typed with a bracket missing — and the mirror of the problem numeric styles have, where the risk runs entirely the other way.

What to do with a near-citation

Three questions, in order, and most items are settled by the first.

Is it prose? If it names people or a body and makes no pointer, dismiss it. "The 2008 crisis" is not a citation and never was.

Is it a citation with a formatting error? A year that fell outside its parentheses, a missing closing bracket, an et al. without a year. These look like prose to a machine and are citations to you — fix the formatting, and the item becomes a normal citation with an entry to match.

Or is it an attribution that should have been a citation? This is the interesting third case, and the one worth slowing down on.

The third case: prose that owes a citation

Return to the opening sentence. "Operationalised by Yuchtman and Seashore in 1967" is prose, and it is also a specific claim about a specific piece of work — one a reader may reasonably want to check.

Whether it needs a citation is a judgement rather than a rule, and it turns on what the sentence is doing:

BACKGROUND, uncontroversial
  "Following Darwin in 1859, the field…"
  → nobody needs a pointer to locate Darwin
  → prose is fine

A CLAIM YOU ARE RELYING ON
  "Yuchtman and Seashore in 1967 established
   the two-factor structure this thesis adopts"
  → your argument depends on it
  → a reader must be able to reach it
  → this owes a citation

The test we would suggest: if a reader might disagree with the sentence, or want to see the work for themselves, it needs a pointer. If it is orientation — a date, a name, a piece of common ground — prose is enough.

And there is a specific reason to care beyond convention. An attribution with no reference is a claim about someone's work that cannot be checked. Most of the time the work exists and the omission is stylistic. Occasionally it does not — and a narrative attribution is precisely where a source that was never verified tends to survive, because prose is not audited the way a reference list is. Presenting work that does not exist as the foundation of an argument is academic misconduct whether it appears in a bibliography or in a sentence, and the sentence is the harder place to notice it.

The narrative form, and why it hides things

One structural point that explains a surprising share of what a check finds. Narrative citations — Okafor (2021) argued… — are legitimate everywhere and harder to audit than parenthetical ones, for a reason that has nothing to do with correctness.

PARENTHETICAL
  …the effect persisted (Okafor, 2021).
  → sits at the end of a clause
  → visually separate from the argument
  → easy to scan for, easy to count

NARRATIVE
  Okafor (2021) found that the effect persisted.
  → embedded in the sentence's grammar
  → reads as prose, because it IS prose
  → your eye processes it as a subject, not
    as a reference

This is why narrative citations are over-represented among the ones people miss when tallying by hand. You are reading a sentence about what Okafor found; you are not reading a citation. The information is identical and the attention it receives is not.

It also produces a specific defect worth checking for: a narrative citation whose year drifted. Okafor (2021) found… where the entry says 2019 is easy to write and almost impossible to notice, because the sentence reads perfectly and the number is doing grammatical work rather than sitting in a bracket asking to be verified.

If you tally by hand, the practical advice is to search for the pattern rather than read for it. Searching your text for (20 catches parenthetical years; searching for each surname in your reference list catches the narrative ones, which is the same backward pass that finds orphans and can be done once for both.

A note on institutional actors

Category two deserves its own paragraph, because it is where the boundary is genuinely blurred rather than merely technical.

"The Ministry announced in 2019" describes an event. "The Ministry's 2019 guidance requires…" describes a document — and a document can be cited, should be cited, and frequently is not, because it does not feel like literature. Government reports, standards, regulator guidance and institutional policies are all citable works with authors, dates and often identifiers — and the ones most likely to move or vanish.

If your thesis leans on official documents, it is worth searching your text for the organisations you name and checking whether each mention is an event or a document. The documents belong in your reference list; the events do not.