Is Luke the Author of a Gospel and the Acts of the Apostles?
cs.CL · cs.AI · cs.DL
Quick verdict
Delta and impostor-verification stylometry consistently support one author for Luke and Acts; identifying the historical Luke himself overreaches the evidence.
Summary
The paper re-examines whether Luke's Gospel and Acts share one author using two stylometric paradigms on Koine Greek texts (Nestle-Aland via Bibelwissenschaft.de; KJV only for illustration). Section 4 revisits word-frequency arguments (e.g., τε occurs 9 times in Luke vs 151 in Acts) with two-tailed hypergeometric intervals, and a Dickens control (Oliver Twist vs Edwin Drood) shows comparable deviations arise within a known single-author oeuvre. Section 5 applies Burrows' Delta over 11 feature sizes (50-800): Acts' nearest neighbor is always Luke's Gospel, but Luke's Gospel is nearest to Matthew unless Matthew and Mark are removed, after which Luke and Acts cluster. Section 6 applies the Koppel-Seidman impostor verification to 5,000-token chunks (three per book) against the Pauline letters, John, and Revelation; across six similarity/distance measures, ten feature sizes, and word, character n-gram, and truncation representations, mean second-order similarities (0.76-1.0) exceed the 0.8 threshold. Section 7 reports Z-scored markers (past-tense εἶπεν vs present λέγει; genitive articles). Limitations: a nine-book closed candidate set of mutually dependent texts, ad hoc thresholds (d=0.8; softmin c=20) without calibration or false-accept rates, only three chunks per book, and no code release. The evidence supports common authorship of the two books rather than identification of the historical Luke.
Review complete
Assessment at a glance
6
risk flags identified
● high 0
● medium 3
● low 3
5 strengths · 5 weaknesses · 5 author questions
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Burrows' Delta applied with eleven feature sizes indicates Luke's Gospel and Acts were written by a common author, most clearly when Matthew and Mark are excluded.
- The Koppel-Seidman verification model, across eight stylistic representations, six similarity/distance measures, and ten feature sizes, always returns a same-author verdict for Luke-Acts chunks.
- Single-word frequency deviations between the Gospel and Acts cannot establish different authorship, as shown by a control on two novels known to be by Dickens.
- Reliable stylometric decisions on this corpus require at least 100 stylistic features.
- Distinctive markers of the common author include the past-tense εἶπεν (vs the present λέγει preferred by other writers) and frequent genitive forms of the definite article.
Strengths
Methodological triangulation
Burrows' Delta with 11 feature sizes (Table 5) is complemented by the Koppel-Seidman verification across 6 similarity/distance measures, 10 feature sizes, and three representation families (words, character n-grams 3/4/5, trunc-n 3-6; Tables 7, 8, A.3), all converging on the same answer.
Negative control against classic frequency arguments
The Dickens comparison (Table 4) shows large single-word deviations ('the': 9,419 vs 4,247) between books known to share one author, directly undercutting the two-author arguments of Clark (1933) and Argyle (1974) built on words like τε.
Domain-aware experimental design
Transparent reporting of weak cells
Interpretable stylistic findings
Weaknesses
Overclaim relative to what the method can establish
Closed, interdependent candidate set
Headline Delta result is conditional
Ad hoc decision parameters without calibration
No code release and possible topic leakage
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
- All 5 strengths & 5 weaknesses, analyzed in detail
- 6 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
Did this study prove that Luke wrote both the Gospel of Luke and Acts?
No. It provides consistent quantitative evidence that the two books share a single author, using Burrows' Delta and an impostor-based verification model on the Greek text. Because no text indisputably written by Luke exists, stylometry can support common authorship but cannot confirm that this author was the historical Luke.
What methods were used in the paper?
Two main paradigms on the Koine Greek text (Nestle-Aland edition): Burrows' Delta with eleven most-frequent-word feature sizes (50 to 800), and the Koppel-Seidman (2018) verification algorithm comparing 5,000-token chunks of Luke and Acts against impostor texts (four Pauline letters, John's Gospel, Revelation) using six similarity/distance measures, word, character n-gram, and truncated-word representations, with 1,000 random feature-subset iterations.
Why did some earlier scholars think Luke and Acts had different authors?
Mainly because certain words differ sharply in frequency between the books, e.g., the conjunction τε appears 9 times in the Gospel but 151 times in Acts. This paper shows, via a control comparing two novels both known to be by Dickens, that such single-word deviations also occur within one author's oeuvre, so they are not reliable evidence of different authors.
Is the data or code for this study available?
The corpus is publicly available: the Nestle-Aland Greek New Testament and the King James Version, downloaded from Bibelwissenschaft.de. The paper describes methods and parameter settings in detail, but it does not mention releasing the analysis code.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Figure 1 (score histogram) is only described in the extracted text; its shape could not be independently inspected.
- Tables were taken from a text extraction; Greek diacritics and some table formatting may be imperfect, and quoted values follow the extraction as given.
- No supplementary material, code, or data files were available; the review relies solely on the embedded paper text.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.