ResearchOps.ai

Agent manifest configuration

Qualitative transcript concept rules

Sixty-two rules for turning interview quotations into verb-forward, participant-centered concept records that remain tied to exact transcript evidence.

This manifest governs one stage of qualitative analysis: selecting language that expresses what a participant was thinking, feeling, or guided by, separating distinct concepts, and drafting one verb-forward summary for each source-linked record. It configures how an agent handles boundaries, wording, cognition type, ambiguity, and failure conditions so a named researcher can inspect the quotation and proposed summary together. It is not intended as a shortcut around qualitative research or researcher judgment.

What it does

Turn transcript quotations into verb-forward, participant-centered, evidence-linked concept records for researcher review.

What it does not govern

  • Cross-participant grouping
  • Themes or research findings
  • Product requirements
  • Recommendations

George Jensen created and maintains this configuration through qualitative research practice and study. Indi Young’s work especially informed its formulation, but it does not represent her method or imply her endorsement. To learn Thinking Styles from its source, explore training with Indi Young.

Download agent manifest (.md) | How this manifest is used →
Total rules
62
Groups
5

01

Quote to Summary Criteria

What makes a candidate admissible and a summary faithful.

14 rules
  1. QS-001Keep language that reveals inner thinking, an emotional reaction, or a guiding principle—not mere facts or scene-setting.
  2. QS-002Give each distinct concept its own record.
  3. QS-003Assign every phrase to only one record, splitting at phrase boundaries when concepts differ.
  4. QS-004Label every record as inner thinking, emotional reaction, or guiding principle.
  5. QS-005Start the summary with a verb that carries the cognition.
  6. QS-006Follow the verb with the key point the participant is thinking or feeling about.
  7. QS-007Extend the quotation only when more context is necessary to support the key point.
  8. QS-008Retest the verb and cognition type rather than forcing the first interpretation.
  9. QS-009Write from the participant’s viewpoint rather than explaining them from outside.
  10. QS-010Prefer the participant’s own memorable words over researcher terminology.
  11. QS-011Borrow a missing word only if the participant would plausibly use it, preferably elsewhere in the interview.
  12. QS-012Write in first-person present tense with the implied subject omitted.
  13. QS-013Reorder and prune freely while preserving the participant’s language and meaning.
  14. QS-014Produce a verb-plus-key-point concept that can later stand in for the quotation.

02

Words to Try

Rules for widening the field, never a controlled vocabulary.

5 rules
  1. WP-001
    Search for a verb that names the participant’s thought process.

    Thought-process verbs: accept, acknowledge, adopt, agree, anticipate, assume, avoid, choose, decide, deliberate, ensure, figure out, forgive, justify, make sure, plan, ponder, promise, realize, recognize, struggle, suspect, think, weigh, wonder, and others.

  2. WP-002
    Express emotional reactions with “feel + emotion” or a verb that already carries the emotion.

    Emotion-carrying verbs: adore, appreciate, crack up, cry out, distrust, enjoy, envy, fear, hope, light up, marvel, panic, reel, tear up, worry, and others.

  3. WP-003
    Brainstorm several emotions and move from a general label to the most specific one supported.

    Specificity example: Surprised — startled, confused, amazed, excited, shocked, dismayed, disillusioned, perplexed.

  4. WP-004
    Attach supporting details as phrases rather than adding a second sentence.

    Connectors: about, after, although, as if, at, because, before, besides, beyond, by, despite, due to, even though, except, from, if, including, instead of, like, past, since, so that, that, through, when, whereas, while, who.

    Use cautiously: “and” and “or” can invite a second sentence.

  5. WP-005
    Treat vague, passive, exterior, and session-mode verbs as warning signs rather than banned words.

    Vague — communicate, consider, deal with, do, expect, manage, organize, plan on, use.

    Passive — be, discover, experience, find, get, have, hear, know, let, need, observe, read.

    Exterior — approve, believe, explain, feel like, feel that, hate, judge, like, love, prefer, want.

    Session mode — compare, complain, critique, remember.

03

How the Work Goes

Practical checks used while shaping and reviewing the records.

16 rules
  1. TC-001Use the participant’s own verb immediately when they have already supplied a good one.
  2. TC-002Brainstorm two to five verbs or emotions, generating more when the interpretation is uncertain.
  3. TC-003Let the best-fitting verb help determine the cognition type.
  4. TC-004Prune the quotation before typing it so the concept becomes visible.
  5. TC-005Restore pruned language when it turns out to contain the memorable wording.
  6. TC-006Before keeping an implied concept, test your confidence, why it was mentioned, and whether it belongs to the remembered moment.
  7. TC-007Require stronger evidence for the concept itself than for supporting details.
  8. TC-008Separate emotions by what they concern and when or where they occurred.
  9. TC-009Join repeated expressions to the existing concept record instead of duplicating it.
  10. TC-010Include supporting details only when they will help someone remember the concept and participant later.
  11. TC-011Repair an unclear key point when writing the details exposes a dangling reference.
  12. TC-012Read the summary aloud and revise wherever the wording makes you stumble.
  13. TC-013Preserve personal context until the later grouping stage.
  14. TC-014Keep one final summary per record.
  15. TC-015Enrich thin summaries using vivid language found elsewhere in the transcript.
  16. TC-016Move memorable material from the details into the key point when that is where the cognition really sits.

04

Judgment Calls

Choices the method does not reduce to a single automatic answer.

17 rules
  1. JC-001Accept that several faithful summaries may be valid, with clarity and recall deciding between them.
  2. JC-002Hold cognition type loosely because the same concept may support more than one type.
  3. JC-003Choose a verb that matches the shape of the thinking, not merely its subject matter.
  4. JC-004Cite where the participant used any word borrowed from elsewhere in the interview.
  5. JC-005Preserve uncomfortable participant language rather than substituting the researcher’s preferred term.
  6. JC-006Add a qualifier only when it helps distinguish this concept from others.
  7. JC-007Reduce genuine double verbs to one main verb, unless dropping one loses real meaning.
  8. JC-008Keep “try to” only when the participant is genuinely struggling, working at, or aiming toward something.
  9. JC-009Replace a non-emotional adjective with the emotion behind it when the evidence supports one.
  10. JC-010Look behind descriptions of another person’s action to find the participant’s actual emotion.
  11. JC-011Let self-description emotions stand when a more precise recasting becomes forced.
  12. JC-012Treat bodily responses as supporting details rather than emotions themselves.
  13. JC-013Split the record when candidate emotions scatter across different concepts.
  14. JC-014Use “wish” and “hope” freely unless they are merely disguised preferences.
  15. JC-015Favor clear, faithful wording over mechanical conformity to the rules.
  16. JC-016Resist granting motives or considerations the participant never expressed.
  17. JC-017Prefer over-splitting to over-merging because separate concepts can be regrouped later.

05

Fail Conditions

Conditions that send a record back or remove it instead of polishing it.

10 rules
  1. FC-001Drop language spoken to the interviewer now rather than from inside the remembered moment.
  2. FC-002Drop approval, preference, description, and fact unless they conceal genuine interior cognition.
  3. FC-003Return any summary whose verb has no clear key point.
  4. FC-004Rewrite gerunds as direct verbs.
  5. FC-005Split compound sentences containing two subject-and-verb concepts.
  6. FC-006Rewrite third-person summaries into implied first-person, present-tense language.
  7. FC-007Enrich generic summaries from the transcript rather than inventing specificity.
  8. FC-008Reject researcher analysis and rewrite directly from the participant’s quotation.
  9. FC-009Loosen summaries that preserve the participant’s original word order too rigidly.
  10. FC-010Record insufficient evidence and drop the candidate rather than manufacturing a concept.