Week 4 · Lecture outline

Week 4 — Lecture Outline · Research & Supporting Materials

Public Speaking · COMM 1 Fall 2026 · Prof. Marchetti Fictional sample

Course: Public Speaking — Fundamentals of Oral Communication (COMM 1) · Silver Oak University (fictional sample) · Prof. Marchetti
Objective covered: Objective 3 — Find, evaluate, and correctly cite credible supporting material; distinguish expert from lay testimony, credible from non-credible sources, and oral citation from plagiarism or fabrication.
SLOs touched: A (compose & deliver — the oral citation drill) · B (critical listening & analysis — evaluating source credibility)
Meeting pattern: 2 sessions × 75 min = 150 min. Segment minutes below total ~150; scale to your own pattern.

Week at a Glance


The week's big question	"Where does good evidence come from — and how do I prove it's real when I cite it out loud?"
By the end of the week, students can…	(1) identify and define the main types of supporting material (examples, statistics, testimony — expert vs. lay); (2) evaluate a source for credibility using the CRAAP criteria; (3) deliver a correct oral citation (source + qualification + date); (4) catch a fabricated citation by verifying AI-supplied sources at the primary source.
Key vocabulary	supporting material, brief example, extended example, hypothetical example, statistics, testimony, expert testimony, peer/lay testimony, source credibility, CRAAP (Currency, Relevance, Authority, Accuracy, Purpose), oral citation, written citation, plagiarism, fabrication, paraphrase, primary source, secondary source
Materials	slides (Deck 4), the week's readings + library guide link, one approved chatbot (Gemini / Claude / ChatGPT) for the AI-critique moment and tutorial, the Speech Workshop source-verification drill
Timing note	8 segments, ~150 min total. Session 1 = Segments 1–4 (~75). Session 2 = Segments 5–8 (~75).

Segment 1 — Hook & the Promise (10 min) · Session 1 opens

Hook. Open with a scenario, not a definition: "Imagine you're about to give a speech on sleep deprivation. You ask an AI chatbot for a statistic. It tells you: 'According to a 2023 study published in the Journal of Sleep Medicine, 73% of college students get fewer than six hours of sleep per night.' You write it down. You put it in your speech. You say it in front of 30 people. Then your instructor looks it up — and the study doesn't exist. The journal doesn't have that article. That statistic was invented. What happens next?"

Take a few responses. Then the turn: "That is a real scenario — it happens every term. This week is about making sure it never happens to you. And the fix isn't complicated: verify before you cite. But to do that, you need to know what counts as a credible source, how to find one, and — most importantly — how to say it out loud in a speech so the audience can judge it too."

The promise: "By Friday you'll be able to name the types of supporting material, evaluate any source with five questions, write a complete oral citation, and catch a chatbot's invented evidence. Those four skills are the ethical engine of every speech you'll give."

Segment 2 — Types of Supporting Material (22 min)

Set it up: "Evidence comes in three main flavors. Knowing which type you're using — and which fits your claim — is how you choose strong support instead of weak support."

Type 1 — Examples
- Brief example: a single, quickly-stated instance that illustrates a point. ("A campus that installed automated lighting cut energy costs significantly.") Vivid, fast, easy to follow — good for opening or illustrating a familiar point.
- Extended example: a developed story or case study that takes time to tell. Good for making an abstract idea feel real and memorable. The detail is the point.
- Hypothetical example: a made-up scenario you explicitly label as invented. ("Imagine you're a first-generation college student with a 20-hour work week…") Useful for putting the audience in a situation, but must be clearly flagged — presenting a hypothetical as if it happened is fabrication.

Type 2 — Statistics
- Numbers that summarize large amounts of data. Powerful — but only when:
- They come from a credible source (a government agency, peer-reviewed research, a reputable research organization — not an anonymous website or an AI chatbot).
- They are used accurately — not cherry-picked, not rounded up to sound stronger, not cited without the date (a statistic from 2001 may be meaningless in 2026).
- They are made meaningful — "one in five adults" lands better than "21.4 percent."
- Misconception to address: "more statistics = more persuasive." No — one well-chosen, well-sourced number beats a wall of unverified figures.

Type 3 — Testimony
- Expert testimony: a statement from someone with recognized training, credentials, or professional experience in the relevant field. A cardiologist on heart disease; a structural engineer on bridge safety. The oral citation must establish why they are an expert.
- Peer/lay testimony: a statement from someone without specialized credentials — an eyewitness, a person with lived experience, an ordinary person affected by the issue. Valid and powerful, but it is NOT expert testimony. Presenting lay testimony as expert testimony is misleading.

One fully worked comparison:

Topic: campus food security.
- Brief example: "A student at this university went three days without a full meal during finals week." (quick, vivid)
- Statistic: "For example, a speaker might cite a survey from a university basic-needs office showing that a substantial percentage of students have experienced food insecurity — citing the office name and year of the report." (illustrative placeholder — if using a real figure, verify at the source)
- Expert testimony: "A registered dietitian at a campus health center, commenting on their clinical observations, provides expert commentary." (qualified voice)
- Peer testimony: "A student who experienced food insecurity describes their experience." (authentic, human — but not expert)

Memory hook: "Examples make it vivid. Statistics make it real. Testimony makes it human. Know which you're using."

Segment 3 — Finding Sources & Evaluating Credibility (22 min)

Set it up: "Anyone can find a webpage. The skill is finding a credible one — and knowing the difference matters enormously when your name is attached to the claim."

Where to find credible sources:
- Library databases (JSTOR, EBSCO, your campus library catalog) — peer-reviewed journals, reliable reference works.
- Government and institutional sites (.gov, .edu, research institutes, professional associations) — primary data, authoritative reports.
- Quality news outlets (national/regional outlets with editorial standards) — current events and recent data.
- Expert interviews — a primary source you create by speaking to a credentialed expert.
- What to avoid*: anonymous blogs, unverified social media posts, AI chatbot output cited as a source (the AI is not a source — it may be pointing at a source, which you then have to verify).

The CRAAP criteria — present factually as one standard framework:

Introduce it plainly: "One widely-used checklist for evaluating sources is the CRAAP test, developed by librarians at California State University, Chico. The five criteria apply to any source — web page, article, or database."

Criterion	The question
Currency	When was it published or updated? Is that recent enough for this topic?
Relevance	Does it actually address your claim, for your audience?
Authority	Who wrote or published it? What are their credentials? Is this a peer-reviewed journal or a personal blog?
Accuracy	Is it supported by evidence? Are claims sourced? Can you verify the information elsewhere?
Purpose	Why was this written — to inform, sell, advocate? Does it acknowledge complexity or only present one side?

Model side-by-side comparison (describe two sources on the same topic):

Topic: the effect of sleep deprivation on academic performance.
- Source A: A peer-reviewed article from a sleep science journal, authored by university researchers, dated within the last five years, with full methodology and references. → CRAAP: Currency ✓, Relevance ✓, Authority ✓ (credentialed researchers), Accuracy ✓ (peer-reviewed), Purpose ✓ (academic).
- Source B: A blog post from an unidentified author on a wellness website, no date, no citations, advocating the site's supplement products. → CRAAP: Currency ?, Relevance marginal, Authority ✗ (no credentials), Accuracy ✗ (no sources), Purpose ✗ (commercial, biased).
"The CRAAP criteria don't make the judgment for you — they make the gap between the two sources impossible to ignore."

Quick interaction (~5 min): put two brief source descriptions on a slide; students apply two or three CRAAP criteria and call out the strengths or problems. Fast classification round-trip.

Segment 4 — Oral Citation: Saying the Source Out Loud (15 min) · Session 1 closes (~75)

The distinction that surprises students: "Most of you learned to write a bibliography — a Works Cited page — in a previous class. A bibliography is a written citation at the end of the document. In public speaking, you do something different: you cite the source out loud, in the middle of the speech, so the audience can hear it and evaluate the evidence while they're listening. That's an oral citation. It is not a bibliography; it is not optional."

Why it matters: an audience can't flip to your bibliography. If you don't say the source, they have no way to evaluate your evidence, no way to check it later, and no reason to trust it over a random claim. Oral citation is what makes evidence evidence rather than just assertion.

The oral citation format (teach it as a sentence):

"According to [who/what the source is] — [why they are credible] — [the date] …"

Model oral citations (format only — these show the pattern; always use your own real, verified sources):
- "According to the U.S. Bureau of Labor Statistics — the federal agency that tracks employment and earnings — in their 2024 annual report …"
- "According to Dr. Jane Smith — a registered dietitian and researcher at a university nutrition department — writing in 2023 …"
- "According to a 2025 survey conducted by a campus basic-needs office on student food security …"

What makes an oral citation complete: source identity + qualification/why it's credible + date. All three.

What makes an oral citation incomplete (address each):
- Just naming the source with no date: "According to CNN…" — when?
- Date without credential: "According to a 2024 study…" — by whom?
- Qualification without source: "An expert says…" — which expert, and where?

The ethics: oral citation and plagiarism
- Using a source's idea or words without saying the source = incremental plagiarism, even if you paraphrased.
- Presenting invented evidence — including AI-invented evidence — as if it were a real citation = fabrication. This is among the most serious academic integrity violations in a speech.
- Paraphrase (putting the source's idea into your own words while crediting the source) is preferred over long direct quotation in speech — but you still cite the source.

Memory hook: "Source + why you trust it + when. All three, every time."

Segment 5 — Why AI Fabricates Citations (and What to Do About It) (15 min) · Session 2 opens

Hook back in: "Last session we talked about what a credible source looks like and how to cite it. Now the question that matters for this course and beyond: why can't you just ask an AI chatbot for sources?"

Plain language first — what is actually happening:
- AI language models are trained to produce plausible-sounding text, not to retrieve verified facts. They can generate citations that sound authoritative — specific author names, publication titles, journal volumes, page numbers — that do not exist.
- This isn't a bug being fixed; it is a fundamental feature of how these systems work. They predict what a citation looks like, not whether it exists.
- This behavior is called "hallucination" in technical language, but that understates it: for citations, the better word is fabrication. The model is inventing a source.

The typical failure mode (describe for students):

"You ask: 'Give me three statistics about college students and mental health.' The AI responds with three bullet points, each ending with a citation: a specific article, author, journal name, volume, and year. You Google the first one. The journal exists. The volume year exists. The article title does not. The author in the context of that article does not. The statistic cannot be verified anywhere. The other two citations have similar problems."

The rule, stated clearly: An AI chatbot is not a source. It may point you toward real sources that you then need to verify yourself. Never cite what an AI tells you without first locating the original source and confirming its accuracy.

The fix: when an AI gives you a citation, take these steps:
1. Go directly to the source it claims — the journal, the organization's website, the database.
2. Search for the article, report, or page by its title or author.
3. If you can find and read the original, you may cite it — citing the original source, not the AI.
4. If you cannot find it after a genuine search, do not use it.

Segment 6 — The Speech Workshop, Model Citations, and Verification Drill (18 min)

Set it up: "This is where we put it together. The Workshop this week is built around a signature drill: you will actually go find real sources, evaluate them, write verified oral citations, and then ask an AI for sources and catch the invented ones. Let's walk through what that looks like."

Model the research workflow (describe it step by step):
1. Choose a topic narrow enough to research in a class period (e.g., campus recycling programs, the effect of walking meetings on productivity, hydration and athletic performance).
2. Find two sources — one from a library database or government/institutional site, one from a credible news outlet or university research page.
3. For each source, run it through the CRAAP criteria quickly — five questions, one minute.
4. Write an oral citation for each: source identity + qualification + date.
5. Record the URL and date you accessed it.

Model the AI-catch workflow (describe it step by step):
1. Open your approved chatbot.
2. Ask: "Give me three statistics or sources about [your topic]."
3. For each citation or statistic the AI provides, attempt to verify it:
- Go to the named source's website or a library database.
- Search for the specific article, report, or data point.
- Record what you found — real/verified, partial match, or not found / fabricated.
4. Flag every invented or unverifiable citation explicitly.

Why this drill is the signature of this week: "The first time you catch an AI confidently handing you a made-up statistic — and you can prove it's made-up because you went and looked — you will never unthinkingly trust a chatbot citation again. That's the point."

Segment 7 — Misconceptions + Instructor FAQ (15 min)

Walk through the key misconceptions explicitly (put each on a slide):

❌ "Expert testimony and peer testimony are the same thing."
✅ Cure: Expert testimony comes from someone with recognized credentials in the relevant field. Peer/lay testimony comes from someone without specialized expertise — an ordinary person, an eyewitness, someone with lived experience. Both are valid, but they do different things. Presenting lay testimony as expert testimony misleads the audience.
❌ "More statistics make a speech more persuasive."
✅ Cure: One well-chosen, clearly sourced statistic beats five vague, unverified numbers. Accuracy and relevance matter more than quantity.
❌ "An oral citation is the same as a written bibliography."
✅ Cure: An oral citation is said aloud during the speech so the audience hears it. A bibliography is a written list at the end of a document. Speeches use oral citations. Both serve citation, but they are different forms for different contexts.
❌ "Paraphrasing means I don't have to cite the source."
✅ Cure: Paraphrasing puts the idea in your own words — but you still owe the audience the source. Paraphrasing without attribution is still incremental plagiarism.
❌ "If an AI chatbot gave me the citation, I can use it."
✅ Cure: The AI is not a source. Verify the original. If you can't find it, don't use it.

Instructor FAQ — Common Stumbles

Student says / does	Quick cure
Uses an AI citation without checking it.	The AI is not a source. Verify at the original. If unverifiable, don't use it.
Presents lay/personal testimony as expert evidence.	Ask: what credentials does this person have in this field? If none, it is peer/lay testimony — which is still valid, but must be labeled correctly.
Cites the source in writing only, not out loud.	In a speech, the audience can't see your bibliography. You say it aloud — source + credential + date.
Rounds a statistic up "to sound stronger."	That is distorting evidence — an ethical violation, even if small. Report the number accurately.
Confuses paraphrase with no citation needed.	Paraphrase means your words, not your idea. The source still gets credited.
Uses "a study shows" without naming the study.	Who wrote it, when, and for what organization? The audience deserves all three.

Scope flag

This outline stays within Objective 3 (research, supporting material types, source credibility, oral citation, and avoiding fabrication/plagiarism). The CRAAP criteria are referenced factually as a standard framework (developed at California State University, Chico); they are not proprietary to this course. Statistical examples are described as illustrative formats — any real statistic used in class would be verified live at the source before class, and students are instructed to do the same. No AI chatbot output is presented as a source anywhere in this course.

~ Prof. Marchetti's edition · Fall 2026 · built with thecoursemaker.com