agentleFS
Jeng-linLiJeng-linLi/humanities-research-skills/skills/emi-transcript/SKILL.md

emi-transcript

Skill0 starsChanged 49 days agoMITView on GitHub

Classroom transcript analysis for EMI and multilingual settings — episode segmentation, language-choice and function annotation, with frequency tables and anonymised output. Use for EMI 課堂語料, classroom discourse, translanguaging analysis, translingual practice, or when the user needs to annotate and quantify classroom talk.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If consent status is unknown, stop and ask.

    line 20
  2. Distinctive:

    Keep timestamps. Analysis without timing cannot answer questions about sequence or amount of talk.

    line 28
  3. Very specific:

    Default dimensions, each with allowed values.

    line 33
  4. Very specific:

    Do not reconstruct inaudible text.

    line 68
  5. Very specific:

    Every interpretive claim about why a pattern occurs is strength: I under the evidence contract and must list the segments it rests on.

    line 71

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

EMI classroom transcript analysis

Annotation engine for classroom talk in English-medium and multilingual settings. Design goal: the annotation scheme is explicit and replaceable, so the user can swap in their own framework instead of inheriting a black box.

Step 0 — Ethics gate

Before any analysis:

  • participants pseudonymised (T01 teacher, S01 student, P01 participant)
  • institution, course and date generalised
  • confirm consent covers the analysis being requested

If consent status is unknown, stop and ask. Do not proceed on assumption.

Step 1 — Segment

Split the transcript into episodes (activity boundaries) and turns.

segments.tsv: segment_id, episode_id, speaker, timestamp_start, timestamp_end, text

Keep timestamps. Analysis without timing cannot answer questions about sequence or amount of talk.

Step 2 — Declare the scheme

Default dimensions, each with allowed values. Override them if the user has a framework — record the override in scheme.md so the coding is reproducible.

dimensiondefault values
language choiceL2-only · L1-only · mixed-intra-turn · mixed-inter-turn
discourse functionexplaining · questioning · scaffolding · managing · evaluating · other
pedagogic moveinitiation · response · feedback · other
participantteacher · student · group · unknown

Definition of mixed must be stated explicitly (word-level? clause-level? turn-level?) because the count depends entirely on that unit.

Step 3 — Annotate

One row per segment, every dimension filled or marked n/a. Never leave a cell blank — blank means "coder skipped it" and corrupts the counts.

Step 4 — Quantify

Report only what the data supports:

  • distribution per dimension (counts + % of segments)
  • cross-tabulation of two dimensions, e.g. language choice × discourse function
  • episode-level comparison, if episodes are comparable
  • talk time by participant, if timestamps are reliable

Step 5 — Interpret

Link patterns back to episodes: what was the class doing when the pattern appears? Numbers alone are not findings.

Hard constraints

  • Timestamps and speaker labels are never modified or inferred.
  • Do not reconstruct inaudible text. Mark [inaudible] and exclude from counts; report the excluded proportion.
  • ASR output is treated as provisional: flag likely errors, do not silently correct them.
  • Quantification is descriptive. Do not run inferential statistics on a single classroom without saying so.
  • Every interpretive claim about why a pattern occurs is strength: I under the evidence contract and must list the segments it rests on.

Files

scheme.md
segments.tsv
annotations.tsv
frequency.md
sources.yml

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.