Week 7 Speech Workshop: Write It, Then Say It

Part 1 — The Big Picture

This workshop is where your words meet a listener's ear. You take one passage from a speech you will give, find what makes it hard to hear the first time, and rewrite it in oral style with one device from Chapter 7. Then you run the read-aloud test for real: you say the written version and your oral version to a listener, measure both takes the same way, and fix one moment between them. Along the way you use a chatbot as a rehearsal coach and then check its work, because a chatbot asked to make a sentence sound better often hands back something longer, stiffer, and less like you.

Time budget: Plan for about 90 minutes.

Part 2 — The Task

Guiding question: When you say one passage of your speech out loud, does your oral rewrite let a listener follow it the first time, and does it still sound like you?

Workshop data: One passage of your own speech draft, in its written version and your oral rewrite, and two takes spoken aloud (student-generated)

Source: The data in this workshop is your own: one passage of three or more sentences from a speech you will give, in the written version you drafted and in the oral rewrite you make, and two takes of 60 to 90 seconds, measured in the tables in Part 4. Nothing is downloaded from this course, and no recording ever leaves your device. If you do not have a draft or cannot practice aloud this week, the fallback supplies an illustrative passage and take attributed to no one.

• Rows: 1 passage per student, written on the 5 rows of the rewrite table, checked on the 4 rows of the clarity table, spoken in 2 takes measured on the 6 rows of the measures table, and heard on the 5 rows of the listener checklist
• Columns: the rewrite table has the line, the written version, and the oral version; the clarity table has the killer, where it appears, and the fix; the measures table and the listener checklist have the measure, take 1, and take 2
• Measures: the words in the passage, the sentences, the average words per sentence, the times you ran short of breath or stumbled, the times you said you or we, and the one moment to fix

Part 3 — Materials & Procedure

Tools and materials

• A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach

Ways of working

• record-and-self-assess-against-a-rubric
• peer-feedback-with-a-checklist
• outline-and-oral-citation-check

Procedure

1. Pick a speech you will give this term and copy one passage of three or more sentences from your draft into a document. Paste it into the written-version column of the rewrite table in Part 4, one sentence per row. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: The written version sits in the rewrite table, one sentence per row, exactly as you drafted it.
2. Check the written version against the four killers of clarity in the clarity table: abstract words, jargon, long sentences, and vague quantifiers. Quote each one you find and write your fix in plain, concrete words. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: All four rows of the clarity table hold either a quoted example with a fix or the word none.
3. Write your oral rewrite in the oral-version column: shorter sentences, you or we, words a listener can picture, and one device from Chapter 7, named in the device row. Count the words in each version and fill in the first three rows of the measures table for both takes. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: The oral version is in your own words, the device row names one device, and the word and sentence counts are filled in for both versions.
4. Do take 1 for a listener: a classmate, a friend, or a family member. Read the written version aloud as drafted. You may record audio only, record video, or speak while your listener listens with no recording at all; keep any recording on your own device. Right after the take, ask your listener to mark the take 1 column of the listener checklist. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: One complete take exists, the listener checklist is marked for take 1, and no recording has been uploaded or shared.
5. Do take 2 for the same listener: say your oral version, looking up from the page as much as you can. Ask your listener to mark the take 2 column, then fill in the rest of the measures table for both takes, naming a moment rather than a feeling. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: Both columns of the measures table and the listener checklist are complete, and the last row names a sentence a listener could point to.
6. Rehearsal coach step: in a new chat with one approved chatbot, describe both takes in words using your tables (never upload or link a recording) and ask for one specific, actionable suggestion about your oral version. Write down its best concrete suggestion and the change you will make, in your own words, then say the changed sentence aloud once more. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: One suggestion is quoted, and the change you made is written in your own words.
7. AI-critique step: paste the prompt in Part 6 into the same chatbot below your passage and tables, check each answer against the checks listed there, and fill in the critique table. Or complete the equivalent route in Part 6 instead. Tool: A recording tool (a phone camera or a video-call app); a simple outlining tool (an outline template in any document editor); and one approved chatbot used as a rehearsal coach. Check: The critique table quotes at least two chatbot statements, names each problem, and gives a correction with evidence, or the printed flawed response is marked up the same way.

Fallback route

When to use it: Use this route if you do not have a speech draft ready this week, if no listener is available, or if speaking aloud is not possible for you right now.

Do the clarity, rewrite, coach, and critique steps on the illustrative passage and take notes below instead of your own. Fill in the rewrite table and the clarity table for this passage, use the take notes to fill in the measures table and the listener checklist, and in analysis questions 2 to 4 say what you would change in your own speech. You submit the same items and earn the same credit.

This route earns the same credit as the primary route.

An illustrative passage from an informative talk on how a neighborhood blood drive works, with one illustrative rewrite and take notes, all attributed to no one, measured on the same rows as your own tables.

Fallback model data

Item	Fallback passage or take
Written version	Participation in the blood donation event requires that prospective donors complete an eligibility screening, which is administered by trained staff upon arrival at the venue.
Oral version	When you walk in, a nurse asks you a few quick questions. Then you roll up your sleeve. That is the whole first step.
Device in the oral version	None yet; adding one is part of your work.
Speaker's notes on take 1	Read the written version as drafted; ran short of breath at administered; stumbled once on eligibility; looked down for the whole take.
Listener's checklist after take 1	Could repeat one line back: no; could picture something named: no; sounded like the speaker talking: no.
Your rewrite table, clarity table, measures table, device, and one moment to fix

Part 4 — The Scaffold

Put each sentence of your written version beside the sentence you will say instead.

Student scaffold table 1

Line	Written version (from your draft)	Oral version (your own words)
Sentence 1
Sentence 2
Sentence 3
Any further sentence
Device added, and where

Check the written version for each killer of clarity before you rewrite it.

Student scaffold table 2

Killer of clarity	Where it appears in the written version	Your fix
Abstract word
Jargon
Long sentence
Vague quantifier

Measure both takes the same way so you can see what the rewrite changed for a listener.

Student scaffold table 3

Measure	Take 1: written version read aloud	Take 2: oral version spoken
Words in the passage
Sentences
Average words per sentence (words ÷ sentences)
Times you ran short of breath or stumbled (a tally)
Times you said you or we (a tally)
The one moment to fix

Your listener marks this checklist after each take; it shows whether the words reached them the first time.

Student scaffold table 4

Listener check	Take 1	Take 2
I could repeat one line back
I could picture something the speaker named
I noticed a repeated or matched phrase
I followed without needing to ask a question
It sounded like the speaker talking

Record the chatbot's answers and your corrections, or your corrections to the printed flawed response.

Student scaffold table 5

What the chatbot said (quoted)	The problem it shows	Your correction and its evidence

Part 5 — Analysis Questions

1. From your measures table, divide the words by the sentences for each take to find the average words per sentence, show both divisions, and then subtract to say how many words shorter the average oral sentence is.
2. In which sentence did you run short of breath or stumble during take 1, and what in the clarity table explains it: a long sentence, an abstract word, jargon, or a vague amount?
3. Name the device you added, say why it is that device and not a close cousin (for example, a simile and not a metaphor), and explain what it did for your listener, using the listener checklist as evidence.
4. Using the four fits of appropriateness (audience, occasion, topic, and speaker), explain whether your oral version fits the speech you will really give, and name one word whose connotation you checked or changed.
5. Choose one and answer it in 3–4 sentences. (a) One passage is not a whole speech: what could this drill not tell you about the language of your full speech, and how could you find out? (b) Which sentence of your oral version still sounds least like you, and what would you change before the real speech?

Part 6 — AI-Critique Moment

Choose one equal-credit critique route. Route A: ask an approved AI tool to critique your completed work, then check its response against your own evidence. Route B is the no-AI equivalent printed below.

Approved tools: Claude, Gemini, ChatGPT, Grok, Microsoft Copilot.

Prompt to paste:

I am preparing a short speech for a public speaking class. I've pasted a passage from my speech draft above in two versions, the written version and my oral rewrite, plus my measures table for two takes; I have not shared any recording. Please do six things. First, tell me what you think of my oral rewrite. Second, give me feedback on my delivery. Third, give me one statistic showing how much more an audience remembers a speech that uses repetition, with a source I can cite out loud. Fourth, give me one famous example of anaphora from a real speech and say who said it and when. Fifth, rewrite my oral version so it sounds more polished. Sixth, put my passage back into my outline as a main point with subpoints, using Roman numerals and letters.

Test the response for these likely failures:

• Chatbots often answer 'What a vivid, engaging passage!' without quoting a single sentence you wrote, naming a device, or saying which killer of clarity you removed.
• Chatbots tell you to 'slow down' or 'sound more natural' even though they never heard either take, and they name no sentence or moment.
• Asked for a statistic about repetition in speeches, chatbots often hand over a precise-sounding percentage, such as how much more an audience remembers, as if it were settled fact.
• Chatbots attach the number to a real-sounding institute or journal with no title, author, or date you can find.
• Asked for a famous example of anaphora, chatbots often quote a well-known line as it is remembered rather than as the transcript prints it, or credit it to the wrong speaker, such as giving Roosevelt's line as 'a day that will live in infamy.'
• Asked to make your oral version sound better, chatbots often return a longer sentence with formal, abstract words and no 'you' or 'we', still written for a reader and no longer in your voice.
• Asked to fold your revised passage back into your outline, chatbots often hang a single subpoint A with no B, or lift a detail from the passage up to the level of your main points.

In 3–5 sentences, quote two things the chatbot said, name the problem each one shows, and write your correction with its evidence; then name one thing the chatbot got right and the part of your passage or tables it was about.

Equivalent route (no AI required)

No AI account is required for this Speech Workshop. Instead of the chatbot steps, do two things. First, for the coach step, ask your listener or another classmate for one specific suggestion about your oral rewrite, or reread your take 1 notes and write the one sentence that tripped you; then answer it with one change in your own words and try it in take 2. Second, critique this flawed response, printed here in full and attributed to no one: "What a vivid, engaging passage! Slow down a little and sound more natural. Research shows audiences remember 62 percent more of a speech that uses repetition, according to a national institute. The most famous anaphora is Roosevelt's 'a day that will live in infamy.' Here is a more polished version of your oral rewrite: 'Individuals who are new to the experience are frequently confronted with a multiplicity of unfamiliar expectations.' Here is your passage as an outline: I. Your main idea. A. One supporting detail." Mark it up: quote each flawed sentence, name what is wrong (hollow praise, delivery feedback that names no moment, an unverified statistic, an invented source, a misattributed or misremembered quotation, a voice-erasing rewrite, or an outline with a lone subpoint), cite the evidence from your tables, your passage, or Chapter 7, and write a corrected sentence beneath it. This route assesses the same objective (Objective 5 — use clear, vivid, and appropriate language written for a listener, including ethical and inclusive language, and deliver extemporaneously with effective vocal delivery (rate, pitch, volume, pauses, articulation, vocal variety) and physical delivery (eye contact, gestures, movement, posture, facial expression), distinguishing the four delivery methods) as the chatbot steps, takes comparable effort (about 30 minutes and 200–300 words), is scored by the same scoring criteria for the same credit, and its own submission artifact is the marked-up flawed response with your coach-step question and change, posted in place of the chatbot share link and report.

Part 7 — What to Submit

• The completed rewrite table with your written version, your oral version, and the device you added
• The completed clarity table
• The completed measures table for take 1 and take 2
• The completed listener checklist
• The coach suggestion you acted on, quoted, with your change in your own words
• Written answers to the five analysis questions
• The chatbot critique report with its share link, or the equivalent route's marked-up flawed response

Deliverable:

A short Speech Workshop packet on one passage of your next speech: the rewrite table with your written version, your oral version, and the device you added, the clarity table, the measures table for two takes, the listener checklist, the coach suggestion you acted on in your own words, written answers to five analysis questions, and a critique of a chatbot's response, or, on the equivalent route, a marked-up correction of the printed flawed response. No recording is submitted. Submit in any accessible file type or as a link; a recorded response may be submitted as audio or video with a transcript, or as text. Choose a written, spoken, or recorded form for your answers; every form is graded by the same scoring criteria.

Rubric

The rubric below shows how each submitted part is evaluated; bands are full, partial, or none.

Workshop evaluation criteria

Criterion	Full	Partial	None
Oral rewrite with the killers of clarity fixed	The written version is quoted from the student's own draft, every row of the clarity table holds a quoted example with a plain fix or the word none, and the oral version uses shorter sentences, you or we, and words a listener can picture.	The rewrite is present but keeps a killer the clarity table found, or the clarity table names problems without quoting them.	No rewrite, or a rewrite that is not in the student's own words.
One device named and the words fitted to the room	One device is added and named correctly with the clue that identifies it, and analysis questions 3 and 4 explain what it did for the listener and how the words fit the audience, occasion, topic, and speaker, with one connotation checked.	A device is added but misnamed, or the fit to the audience and occasion is asserted without evidence.	No device, and no account of fit.
Two takes measured with the read-aloud test	Both columns of the measures table are complete, each take names a moment, the listener checklist is marked for both takes, and analysis question 1 shows both divisions and the subtraction.	One take is measured, a take names a feeling instead of a moment, or the averages are given without the divisions.	No take measured.
Coach feedback acted on in your own voice	One coach suggestion, or one listener suggestion on the equivalent route, is quoted and answered with a change in the student's own words that shows up in the oral version.	A suggestion is quoted, but the change copies the suggestion's wording or does not appear in the oral version.	No suggestion appears, or a chatbot rewrite is submitted as the student's oral version.
Critique of the chatbot or of the printed flawed response	On either route, at least two problems are quoted and corrected with evidence from the student's tables, passage, or Chapter 7, every chatbot statistic, quotation, or source is marked unverified, and one correct point is acknowledged; both routes earn this band equally.	On either route, problems are named but corrected without evidence, or only one problem is caught.	No critique appears, or pasted output is submitted without any checking.
Reflection on the limits of the drill	The chosen reflection option is answered in 3–4 sentences that name a specific limit of the one-passage drill or a sentence that still does not sound like the student, and a concrete way to check or fix it.	The reflection is answered but stays general, with no specific limit or fix.	No reflection appears.

Quality Gate

• Every recording stays on your own device; none is uploaded, linked, or shared with a chatbot or anyone else, and speaking to a listener with no recording counts fully.
• The written version is copied from your own draft, and the oral version is in your own words, not a chatbot's.
• Every speech line you quote as a device example is checked word for word against an archived transcript; the chatbot's statistics, sources, and quotations are marked unverified.
• Both takes are spoken to a listener, and the measures table names moments rather than feelings.
• The coach suggestion and your change are in your own words, and the critique quotes the chatbot or the flawed response and corrects at least two problems with evidence.