GPT-assisted writing shows higher lexical diversity, richness, and noun/non-stopword ratios, while independently authored writing shows higher adverb ratio and sentence-length variability¶

Claim · All claims
Evidence · 1 study · causal r1 · q2 · i3 large
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i3 Across 90 participants, paired t-tests showed GPT-assisted documents had higher TTR (r = .43), Hapax Ratio (r = .46), Word Entropy (r = .42), NSR (r = .50), Noun Ratio (r = .47), Verb Ratio, and Adjective Ratio, while independently authored documents had higher Adverb Ratio (r = -.34) and Sentence-Length Variability (r = -.13). → Rajesh Kumar 2026

Evidence¶

Rajesh Kumar 2026¶

Rajesh Kumar, Nabeel Siddiqui, Alexander Fuchsberger. (2026). Detecting GPT-Assisted Writing Using Interpretable Stylometric Features. https://arxiv.org/abs/2609.26687 doi:10.48550/arxiv.2609.26687

q2 · i3 · causal · r1

Paired t-tests and point-biserial correlations on document-level median feature values across 90 participants (Table 2). NSR showed the strongest positive association (r = 0.50) and Adverb Ratio the strongest negative (r = -0.34), described as "the strongest negative association".

"GPT-assisted documents have higher TTR, Hapax Ratio, Word Entropy, NSR, Noun Ratio, Verb Ratio, and Adjective Ratio, while independently authored documents have higher Adverb Ratio and Sentence-Length Variability."

Discussion¶

Related Claims¶

• SHAP analysis identifies Hapax Ratio as the most influential stylometric feature in Random Forest predictions, followed by NSR, Noun Ratio, Adverb Ratio, and TTR — related
• Random Forest using nine interpretable stylometric features detects GPT-assisted student writing with held-out ROC-AUC of 0.870 and F1 of 0.842 — related
• AI-assisted journal entries show higher lexical diversity and more emotion-word usage than unassisted entries — possibly the same claim (merge candidate)
• Humanization works by systematically suppressing detector-salient surface cues — long-token ratio, AWL density, and lexical diversity — rather than fragmenting syntax — related
• Type-token ratio showed a non-monotonic relationship with predicted difficulty, with both very low and very high lexical diversity increasing predicted text difficulty — related
• Elevated detector scores track long-token and Academic Word List density, with numeric and non-alphabetic token ratios correlating negatively — related