INQUIRING LINE

What are the three orthogonal axes that structure the argument scheme periodic table?

This explores Wagemans's 'Periodic Table of Arguments' — the three coordinate axes he uses to map every possible argument scheme onto a single closed space, and what that buys you over a hand-built list.


This explores Wagemans's Periodic Table of Arguments and the three coordinates that define it. The corpus is precise on the axes: an argument's place is fixed by (1) its **subject–predicate structure** — whether the conclusion adjusts the subject or the predicate of a claim, (2) whether it is **first-order or second-order** reasoning — an argument about the world versus an argument about another argument, and (3) the **pairing of proposition types** it links together Can three axes organize all possible argument schemes?. Because the axes are orthogonal, every combination is a valid cell, and the cells exhaust the space — there's no scheme that falls outside the grid.

The deeper move is what this replaces. The dominant framework before Wagemans was Walton's catalog of 60-plus schemes, assembled by family resemblance — useful but open-ended, with no principle telling you when the list is complete or where a new scheme would sit. Three generative axes turn that contingent list into a finite, predictive structure, exactly mirroring chemistry's jump from cataloging substances to the periodic table that predicted undiscovered elements Can argument schemes be organized by formal principles instead of lists?. The payoff isn't tidiness — it's that empty cells become hypotheses about argument types nobody has studied yet.

The most surprising axis to sit with is the second one. Wagemans argues that the first-order/second-order split does double duty: it lines up with the *classical* distinction between internal and external topoi, and with the *modern* line between reasonable and fallacious arguments. That implies fallacies aren't just dialectical bad manners — they may have a specifiable formal-linguistic signature, a structural location on the grid rather than a judgment call Do first-order and second-order arguments unify classical and modern divisions?.

Worth knowing as a coda: a clean coordinate system on paper doesn't mean machines can read the coordinates. Classifying which scheme an actual argument instantiates demands recognizing inferential patterns spread across a whole passage, not local cues — and models that top 0.80 F1 on tagging argument components stall at 0.55–0.65 on scheme classification Why does argument scheme classification stumble where other NLP tasks succeed?, with LLMs only crossing that bar given few-shot examples and explicit scheme descriptions Can large language models classify argument schemes reliably?. The periodic table tells you the space of answers is closed; placing a real argument inside it stays genuinely hard, partly because the same text often admits more than one valid reconstruction Why do different people reconstruct the same argument differently?.


Sources 6 notes

Can three axes organize all possible argument schemes?

Wagemans's Periodic Table maps all argument schemes onto coordinates across three axes: subject-predicate structure, first-order versus second-order reasoning, and proposition-type pairings. This combinatorial approach replaces Walton's open-ended list with a closed, systematic space enabling computational analysis and discovery of unstudied scheme types.

Can argument schemes be organized by formal principles instead of lists?

Wagemans shows that three orthogonal axes generate a closed, finite classification space for all argument types, replacing the family-resemblance logic behind Walton's 60+ schemes. This mirrors the chemical periodic table's shift from contingent lists to predictive structure.

Do first-order and second-order arguments unify classical and modern divisions?

Wagemans proposes that the first-order vs second-order argument distinction reflects both the classical internal-external topoi divide and the modern reasonable-fallacious distinction. This suggests fallacy theory operates through specifiable formal-linguistic structure rather than purely dialectical criteria.

Why does argument scheme classification stumble where other NLP tasks succeed?

Scheme classification requires recognizing inferential patterns across distributed text spans, not local surface features. Models plateau at F1 0.55–0.65 while the same systems exceed 0.80 on component tagging and stance, suggesting the integrative reasoning demand is fundamentally different.

Can large language models classify argument schemes reliably?

Zero-shot prompting fails uniformly across models. Few-shot with scheme descriptions helps, but only larger models exceed F1 0.55, with Claude reaching 0.65. Smaller models plateau around 0.53, suggesting a representational capacity threshold.

Why do different people reconstruct the same argument differently?

Multiple valid argument reconstructions exist for the same text with no ground truth. This is not annotation error but an inherent feature of the task—different formalization schemas are each internally valid.

Research prompt for your LLMexpand ↓

Copy into ChatGPT or Claude to take this line of inquiry further — it asks the model to find newer work and re-test which earlier constraints still hold.

You are a researcher re-evaluating Wagemans's Periodic Table of Arguments—a three-axis classification scheme for argument types—against the latest LLM and argumentation research (2024–2026). The question remains open: do these three orthogonal axes (subject–predicate structure, first-order vs. second-order reasoning, proposition-type pairings) remain the most durable coordinate system for understanding argument structure, or has capability scaling or new empirical work revealed a different decomposition?

What a curated library found — and when (dated claims, not current truth):
Findings span 2021–2026; treat them as timestamped constraints, not present reality.
• Wagemans's three axes (subject vs. predicate, order level, proposition pairing) form a closed, generative space that unifies classical topoi and modern fallacy distinction (~2021–2024).
• LLMs classify argument components reliably (0.80+ F1) but scheme classification stalls at 0.55–0.65 F1; only few-shot prompting + explicit scheme descriptions push performance higher (~2024).
• Argument reconstruction is fundamentally underdetermined—the same text admits multiple valid readings—making ground-truth labeling intrinsically ambiguous (~2024).
• Recent work (2024–2025) shows LLMs exhibit prompt sensitivity, content effects on reasoning, and improved multimodal reasoning about social situations; chain-of-thought variants (including argumentative querying) boost structured outputs (~2025–2026).

Anchor papers (verify; mind their dates):
• arXiv:2404.00750 (2024-03): Can Language Models Recognize Convincing Arguments?
• arXiv:2410.12405 (2024-10): ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs
• arXiv:2412.15177 (2024-12): Critical-Questions-of-Thought: Steering LLM reasoning with Argumentative Querying
• arXiv:2601.06002 (2026-01): The Molecular Structure of Thought: Mapping the Topology of Long Chain-of-Thought Reasoning

Your task:
(1) RE-TEST THE AXES. For each of the three orthogonal dimensions, judge whether newer model capabilities (scaling, instruction-tuning, multimodal reasoning) or training methods (e.g., argumentative chain-of-thought, critical-question scaffolding) have relaxed or overturned the scheme-classification bottleneck (0.55–0.65 F1). Separate the durable claim (the axes themselves may still map the space) from the perishable limitation (models' ability to reliably *locate* arguments within that space). What, if anything, has lifted classification performance toward human agreement?
(2) SURFACE CONTRADICTING OR SUPERSEDING WORK. The library hints at argumentative querying (2024-12) and molecular-structure mapping of chain-of-thought (2026-01) as recent alternatives to flat scheme classification. Do these frameworks propose a different decomposition of argument structure, or do they layer atop Wagemans's axes? Cite papers that explicitly challenge the three-axis model or propose a richer topology.
(3) PROPOSE 2 RESEARCH QUESTIONS THAT ASSUME THE REGIME HAS MOVED:
   a) Can argumentative chain-of-thought (critical questions, counterclaims) recover the underdetermined latent structure of a passage—i.e., does steering LLM reasoning through argument-aware prompts resolve the multiple-valid-reading problem and converge scheme classification toward 0.80+ F1?
   b) Does the subject–predicate axis remain orthogonal to first-order/second-order reasoning and proposition-type pairing once you condition on multimodal social reasoning (e.g., arXiv:2507.20409), or do pragmatic/social dimensions introduce couplings that break orthogonality?

Cite arXiv IDs; flag anything you cannot ground in a real paper.

Next inquiring lines