Forsaking your own: unveiling the delayed recognition of Garfield's work on the "delayed recognition" phenomenon

Delayed recognition (DR) implies that the full scholarly potential of certain scientific papers is recognized belatedly many years after their publication. Such papers are initially barely cited (sleep), and then suddenly, sometime in the future, their citation numbers burst (are awakened). After van Raan (2004a) called them “Sleeping Beauties” the DR phenomenon has drawn considerable attention. However, long before van Raan (2004a) Garfield studied the phenomenon in a series of articles from 1970 up to year 2004. In the present study we ask the pertinent question; Has the phenomenon of DR itself suffered the delayed recognition? In search of an answer we study the citation history of the Garfield (1980a) paper in which Garfield addressed DR directly for the first time. We find that the paper hardly received the attention befitting the Garfield’s stature as an information scientist. Specifically, the paper received a meager of 10 citations up to the publication year of van Raan (2004a) and was then, in 2007, feebly awakened from its deep sleep of twenty-eight years receiving 20 citations in next four years; up to 2010. Being the undisputed giant of information science that even Garfield’s paper on DR can suffer DR is hardly anticipated.

💡 Research Summary

The paper investigates whether Eugene Garfield’s own seminal work on “delayed recognition” (DR) – the 1980 article in which he first coined the term – has itself suffered from delayed recognition. The authors begin by reviewing the sociological roots of the phenomenon, citing early studies by Barber, Merton, and Cole that described how novel ideas often receive little attention initially. Garfield later formalized this as “Sleeping Beauties” (SB), a concept revived and popularized by van Raan (2004). Despite Garfield’s stature, his 1980 paper has been largely ignored, prompting the central research question: has the DR phenomenon itself been delayed?

To answer this, the authors compile citation data from three major bibliometric databases—Scopus, Web of Science (WoS), and Google Scholar (GS)—all searched on September 15 2025 using the exact title of Garfield (1980). The raw counts were 45 (Scopus), 93 (WoS), and 205 (GS). After meticulous manual de‑duplication, correction of mis‑attributed citations, and removal of duplicate records across the three sources, they identified 214 unique external citing papers. In addition, they harvested self‑citations from Garfield’s own publications (20 instances) by consulting the Garfield Library at the University of Pennsylvania, which archives all of his works in the magazine Current Contents (CC) and The Scientist. The final dataset comprised 234 citing items (214 external + 20 self‑citations).

Citation trajectory analysis shows an extreme “sleep” phase: from 1980 to 2002 the paper received at most one citation per year, with many years receiving none. No citations appear between 2003 and 2006. In 2007 a modest resurgence occurs, yielding six unique citations (after de‑duplication). Between 2007 and 2010 the annual citation count rises to about 20 per year, which the authors describe as a “feeble awakening”. They apply Garfield’s own DR criteria (low citation frequency for the first 5–10 years, average ≤ 1 citation per year, and a ten‑fold increase by year 20) and van Raan’s SB definition (sleeping period ≥ 4 years, citation surge ≥ 2‑fold). Garfield (1980) satisfies the sleeping‑period condition but fails to achieve the expected ten‑fold surge; the modest increase is largely attributable to van Raan (2004) – identified as the “prince” that first cited the 1980 paper.

A major methodological contribution is the triangulation of three citation databases with the Garfield Library’s curated records. The authors demonstrate that WoS and Scopus severely under‑represent Garfield’s output because many of his works appeared in non‑peer‑reviewed venues (CC, The Scientist), leading to mis‑attribution (e.g., WoS collapsing 52 Garfield papers into a single editorial citation). Google Scholar offers broader coverage but includes duplicate and erroneous self‑citations, which the authors manually corrected.

The discussion highlights several limitations. First, the inclusion of self‑citations inflates the raw citation count, contrary to contemporary practice that excludes them when assessing delayed recognition. Second, attributing the entire awakening to a single “prince” oversimplifies the diffusion process; other latent citing works may exist but remain undetected due to database gaps. Third, the analysis stops at 2024, omitting any post‑2025 citation dynamics that could alter the interpretation of the awakening phase.

In conclusion, Garfield’s 1980 article, despite being the conceptual cornerstone of DR research, experienced a prolonged period of neglect and only a modest revival decades later. This paradox illustrates how structural biases in bibliometric databases and the prevalence of non‑indexed publication venues can impede the dissemination of influential ideas. The study calls for more inclusive citation tracking mechanisms and a reassessment of how scholarly impact is measured, especially for works published outside traditional peer‑reviewed journals.