Week 4 — Speech Workshop / Rehearsal Studio · "Research, Evaluate, and Cite"

Course: Public Speaking — Fundamentals of Oral Communication (COMM 1) · Silver Oak University (fictional sample) · Prof. Marchetti
Objective: Objective 3 — find, evaluate, and correctly cite credible supporting material; catch fabricated citations · SLO A (compose — write verified oral citations) · SLO B (critical analysis — evaluate source credibility; identify fabricated sources)
Worth 50 points · Speech Workshops group = 15% of the grade · Workshop 4
Format this week: a research and citation drill — no recording required this week; the work is finding real sources, evaluating them, writing verified oral citations, and catching an AI's fabricated ones.

This is the course's citation-integrity Workshop. Week 4 is the week the fabricated-citation risk moves front and center. The signature AI-critique this week is not about hollow praise — it is about AI-generated source fabrication: you will ask a chatbot for sources, then verify every one at the actual source, and flag the invented ones. This is the most important AI skill in this course.

---

Part 1 — The Big Picture

Public speaking runs on evidence. An audience extends trust to a speaker who can say: "Here is where I got this — a real, named, dated, qualified source." Speakers who cite fabricated or unverifiable evidence don't just lose credibility; they mislead. And in a world where AI chatbots confidently generate plausible-sounding citations that don't exist, the responsibility for verification sits entirely with you, the speaker.

This Workshop is built around a simple but powerful drill: find real sources, evaluate them, cite them — and then catch the AI inventing ones.

The guiding question: "Can I prove that my evidence is real — and can I spot when an AI is making it up?"

---

Part 2 — The Drill: Find, Evaluate, and Cite Two Real Sources

Choose one topic from the list below — or use a topic your instructor assigns. The topic should be researchable in a class period and useful as evidence in a speech.

Suggested topics (choose one, or use your own with instructor approval):
- The effect of sleep deprivation on academic performance
- Campus food insecurity among college students
- Physical activity and mental health in young adults
- E-waste and electronics recycling rates
- The health benefits of regular walking

Do this:

Step 1 — Find two real sources.
Use at least two different types of locations:
- A library database (ask your campus librarian for access to JSTOR, EBSCO, PubMed, or your campus catalog)
- A credible web source (a .gov agency, a university research page, a recognized public health or research organization)

For each source, record this information in the table below:

Field	Source 1	Source 2
Author(s) or organization
Title of article/report
Publication or organization name
URL (or database + article title)
Date of publication
Date you accessed it

Step 2 — Evaluate each source using the CRAAP criteria.
Fill in the table. Write at least one sentence per criterion for each source.

Criterion	Source 1	Source 2
Currency (is the date recent enough for this topic?)
Relevance (does it address your specific claim?)
Authority (who wrote/published it, and what are their credentials?)
Accuracy (is it supported by evidence? verifiable?)
Purpose (why was it written — to inform, advocate, sell?)
Your verdict (credible to cite? any caveats?)

Step 3 — Write a complete oral citation for each source.
Use the three-part format: "According to [source/author identity] — [qualification/why credible] — [date] …"

Write a full sentence for each:
- Oral citation, Source 1: __
- Oral citation, Source 2: __

Verification record: for each source, confirm you can access it:
- Source 1 URL/database location (accessed [date]): __
- Source 2 URL/database location (accessed [date]): __

This is the load-bearing step: you must have actually located these sources. A citation you cannot point to at a real URL or database is not a verified citation.

---

Part 3 — Self-Assessment of Your Source Work

Rate each on a 1–5 scale and add a one-line honest note:

What to check	Score (1–5)	Honest note
Both sources located at verifiable URLs or databases
CRAAP evaluation complete — all five criteria for each source
Oral citations complete — all three parts for each
Verification records noted — URL/database and date accessed

Pick your ONE thing: what is the single most useful improvement you could make to your source work for a real speech? Write it in one sentence: "Next time I will ______."

---

Part 4 — Analysis Questions

Answer in a sentence or two each:
1. Which of the five CRAAP criteria did you find hardest to assess? Why?
2. Did you have to discard any source during your research? What made it not credible enough?
3. When you wrote the oral citation, was there a part you had trouble completing (source identity, qualification, or date)? How did you resolve it?
4. What is the difference between citing a source in a speech and citing it in a written paper? Why does it matter?

---

Part 5 — Rehearsal-Coach Moment (BYOAI)

Bring in your approved chatbot (Gemini, Claude, or ChatGPT) as a rehearsal coach.

1. Share your two oral citations and ask: "I'm building a speech on [your topic]. Here are my two oral citations — give me specific, actionable feedback on whether they are complete and correctly formatted."
2. Check whether the coach correctly identifies all three parts (source identity, qualification, date) and gives useful, specific feedback.
3. Apply its best concrete suggestion and revise any incomplete citation.

---

Part 6 — AI-Critique Moment: Catch the Fabricated Citations (REQUIRED — the signature drill of this week)

This is the most important step in this Workshop. Complete it carefully.

What you're doing: you are going to ask an AI chatbot for sources or statistics on your topic, then verify each one at the actual source and record what you found — real, partial match, or fabricated.

The steps:

1. Open your approved chatbot.
2. Ask it: "Give me three statistics or source citations about [your topic]. Include the author, publication name, and year for each."
3. For each citation the AI provides, attempt to verify it:
   - Go to the named organization's or journal's website.
   - Search for the specific article, report, or statistic by title or author name.
   - Try a library database (Google Scholar is acceptable for this verification step).
   - Record what you found.

Fill in this verification table:

AI-supplied citation	What I searched	What I found	Real / Partial / Fabricated
Citation 1:
Citation 2:
Citation 3:

4. Write a 3–4 sentence paragraph reporting:
   - How many of the AI's citations were real and verifiable?
   - How many were fabricated or unverifiable?
   - What would have happened if you had used the fabricated ones in a speech without checking?
   - What rule will you follow from now on when an AI supplies a source?

Why this drill exists: AI language models are trained to produce plausible text, not verified facts. They can generate citations with specific author names, journal titles, volumes, and years that simply do not exist. Every term, speakers trust these citations without checking, then stand in front of an audience citing evidence that was invented. This drill is built to make sure that never happens to you.

---

Part 7 — What to Submit

Submit a single document (or text entry) with:
1. Your completed source-information table (Part 2, Step 1)
2. Your completed CRAAP evaluation tables (Part 2, Step 2)
3. Your two complete oral citations with verification records (Part 2, Step 3)
4. Your self-assessment scaffold (Part 3) including your "one thing to fix"
5. Your Part 4 analysis answers
6. Your AI-critique verification table and paragraph (Part 6)

Due Sunday, Sep 27, 11:59 p.m. (50 points).

---

Instructor answer key & model responses — REMOVE BEFORE PUBLISHING TO STUDENTS

Students research different topics and find different sources, so there is no single correct source table. The key grades completeness of the CRAAP evaluation, correctness of the oral citation format, and honesty and specificity of the AI-critique.

Model oral citation (illustrative format — verify any real source used in class at the primary source):
- "According to the U.S. Centers for Disease Control and Prevention — the federal agency responsible for public health research and data — in their [year] report on [topic] …"
- "According to Dr. [Name] — a [credentials] at [institution] — writing in [year] in [publication] …"

Note to instructor: if you demonstrate a live oral citation in class this week, verify the actual source at the primary URL before class. Never use an AI-generated statistic or citation in your demonstration. The model above shows the FORMAT only — fill in only sources you have personally verified.

Expected AI-critique results:
- Most chatbots, when asked for three citations on a topic, will produce some mix of: real sources (verifiable at the named location), partially-matching sources (the author and journal exist, but not that specific article), and fully fabricated citations (the article, author, or both do not exist at the named source). Some chatbots are better than others; none are reliable.
- A student who verifies all three and finds at least one fabricated citation has passed the drill. A student who finds all three real receives full credit with a note: some topics yield better AI results; try the drill again with a different topic or model if all three were real.
- Full credit for Part 6 requires the student to name a specific citation they checked, state whether it was real or fabricated, and articulate a clear rule for future behavior (e.g., "I will always verify at the source before citing").

Common ways students lose Workshop points:
- CRAAP evaluation with only two or three criteria — requires all five per source.
- Oral citation with two parts (usually missing the date or the qualification).
- AI-critique paragraph that says "the AI was right" without showing the verification steps.
- Verification record with a vague "found it online" rather than an actual URL.

---

Grading rubric — 50 points

Criterion	Full	Partial	None
Sources located and CRAAP evaluation (Parts 2–3) — both sources found at verifiable locations; CRAAP evaluation complete for all five criteria for each source (15)	15	8–12	0–6
Oral citation completeness (Part 2, Step 3) — both citations include all three parts (source identity, qualification, date) in complete sentences; verification records noted (15)	15	8–12	0–6
Self-assessment and analysis (Parts 3–4) — honest, specific scaffold responses; all four analysis questions answered with real engagement (10)	10	5–8	0–4
AI-critique drill (Part 6) — verification table completed with three AI-supplied citations checked at primary sources; paragraph names specific results and states a clear rule (10)	10	5–8	0–4

Quality gate (self-checked): the four rubric criteria sum to exactly 50 (15 + 15 + 10 + 10 = 50). ✓

Citation-integrity gate (self-checked): this Workshop asserts no specific statistics or real source citations as facts. The oral citation examples in the model answer above show format only, and the instructor note explicitly directs any real demonstration source to be verified live before class. The AI-critique drill is built to expose fabricated citations, not to present them as real. The CRAAP criteria are referenced factually as a standard framework (developed at CSU, Chico — no proprietary claim). ZERO fabricated facts, statistics, or citations in this Workshop. CITATION-INTEGRITY GATE PASS.