research-document

Explanatory-progression coding rubric v1.0

Explanatory-progression coding rubric v1.0

Frozen: 2026-08-02 after one pilot revision.
Mission: MS-COMM-2026-0004

Coding protocol

  1. Segment the bounded text where the dominant rhetorical function changes.
  2. Assign one primary move and any secondary move from the codebook.
  3. Record the explicit textual cue, dependency on prior material, warrant source, assumed prerequisite, omission, and plausible alternative reading.
  4. Separate O observations from I interpretations and H untested outcome hypotheses.
  5. Code once in source order, wait until the other works are coded, then recode a matched opening/middle/closing sample without consulting the first pass.
  6. Resolve disagreements by narrowing claims; do not manufacture agreement.

Codebook

Code Move Observable test Common confusion
OR Orient States situation, problem, purpose, or target Background that has no local directional function
DF Define Fixes meaning, role, scope, or boundary Mere synonym or example
PR Prerequisite Supplies or invokes a dependency needed later Unmarked assumed knowledge
DC Decompose Splits a task, system, or claim into parts A list with no dependency function
SQ Sequence Makes an order or dependency explicit Chronology that does not guide inference/action
CL Claim Advances the target proposition Summary of an already established proposition
WR Warrant Connects support to claim: proof, precedent, evidence, or norm Support listed without an inferential bridge
EX Example/analogy Instantiates or maps an abstract relation Evidence intended as direct warrant
OB Objection/limit Raises countercase, uncertainty, boundary, or exception Generic qualification
RS Resolution Answers an objection or reconciles tension Repetition of the initial claim
SY Synthesize Compresses prior moves into an integrated result Local transition
NX Navigate Cross-reference, heading, numbering, or preview supports retrieval Decorative formatting
AC Action/closure Specifies what follows, what is constructed, or what must be done Descriptive consequence only

Required contextual fields

  • intended audience and plausible resistance;
  • audience starting state and target state;
  • medium, length, and stakes;
  • authority and proof obligation;
  • unit and dependency boundaries;
  • explicit versus implicit transitions;
  • treatment of uncertainty and alternatives;
  • accessibility and translation dependence;
  • local strength, systemic cost, and transfer condition.

Pilot revision and reliability check

The pilot initially combined WR and EX. Darwin's analogies, Brown's legal precedent, Euclid's deductions, and RFC definitions showed that an example can make a relation imaginable without warranting it under the domain's proof rules. Version 1.0 therefore separates them and adds NX for specifications.

A repeated blind-to-first-pass recode covered each work's opening, one middle unit, and closing (15 units). Primary-code agreement was 12/15 (80%). Three disagreements were resolved as dual-coded moves: Darwin analogy (EX+WR), Descartes rules (DC+SQ), and RFC terminology (DF+NX). This is an internal stability check, not independent inter-rater reliability. A second analyst and predeclared adjudication procedure remain required before using the rubric for comparative measurement.

What the rubric cannot establish

Move counts do not measure quality, comprehension, truth, or ethicality. The rubric can expose architecture and generate outcome hypotheses; reader studies are needed to determine effects.