← All self-study courses

Give a talk people remember (10 sessions, 15 minutes each, v1.0)

Goal of this course: Build your talk from its skeleton and present it without reading a script.

This is a self-study course for undergraduates who have a research presentation or a class presentation coming up. Over 10 sessions you will learn to "pair the opening question and the closing answer, one sentence each", "write the one-sentence message of every slide first, all of them, and lock them in", and "speak without a script, using only keywords as cues".

A Japanese version of this course is available (講座_要点を押さえた発表_v1.0.md). This English version has the same content.

---

Introduction (how to use this course)

What this is

This is a teaching material published for students' self-study by the Department of Neurophysiology, Kobe University Graduate School of Medicine. In our lab, when undergraduates prepare a 10-minute research presentation, we have used a procedure that goes "from building the structure to speaking without a script". This course rewrites that procedure in a form that anyone, even someone who has never come to our lab, can work through alone. We have revised the procedure many times based on feedback from students who used it.

Who it is for

• Undergraduates (any year, any faculty) who are going to speak in front of others for a few minutes up to about 15 minutes: a research presentation, a class presentation, practice for a conference or an interview, and so on.
• No background knowledge is needed. You can work through it even if you have never done research. If you have no topic to talk about, use the "practice example" (a fictional study) in the course.

What you need

• A conversational AI (one that answers your questions in a chat). One that can read (accept as an attachment) a file is convenient. If it cannot, paste the entire contents of this file.
• A work notebook (a paper notebook, or one document file). You will add all the work from the 10 sessions to this one place.
• Something to keep time (a smartphone stopwatch is enough).
• About 10 paper cards (used in Session 9; you may also divide one sheet of paper into 10 boxes).
• (Optional) Sessions 4 and 9 contain short pieces of Python (a programming language) code. You can do the same thing by counting by hand, so they are not required. If you do use them, a free environment that runs in your web browser is fine (search for "Python browser free" and you will find ones you can use without registering).

How to proceed

1. Load this file into the AI (attach it, or paste the full text).
2. Type "Please start Session 1."
3. Answer the AI's questions in your own words (do not let the AI write the answers; the thinking is the practice).
4. Stop at 15 minutes. It is fine to stop there even if the task is not finished (continue the next day). The next day, type "Please start Session 2." If you paste the work notebook from the previous session, you will continue from where you left off.

The course is designed for one session a day, finishing in 10 days. Why spreading the work over separate days helps it stick is covered in Session 9. The "15 minutes per session" is study time. The length of the presentation itself differs from person to person; the examples in this course use a 10-minute presentation.

Even if you cannot use an AI, you can still proceed by reading the "Explanation" and "Hands-on tasks" of each session yourself and writing in your work notebook (the answers to the check questions are in the collapsible section of each session).

Cautions

• The AI can make mistakes. Check important points (numbers, facts, papers) against textbooks or original sources. The AI may plausibly cite papers that do not exist.
• Do not enter personal information about yourself or others (names, student ID numbers, contact details), medical information, or other people's data into the AI. Before entering your own research data, check on the settings screen of the AI you are using whether the option "use what I enter for training (improvement)" is turned off; if it is on, turn it off before entering anything (the name of the option differs between AIs and changes often. If you cannot find it, ask that AI where it is, and in the end check the settings screen with your own eyes). For data collected from people, follow the handling decided in the ethics review and consent, and if in doubt, ask your supervisor.
• Follow the rules on AI use of your university and your class. Using text from the AI as-is in your own presentation may be regarded as plagiarism (using someone else's text as if it were your own).
• This course is a teaching material for learning presentation skills. It is not intended as medical advice.

★Request for feedback (please be sure to read)

This course is revised every time based on the voices of those who use it. If you used it, please be sure to send us feedback.

• Send to: the address on the Contact page of the lab website
• Subject: "Course feedback: The Craft of a Presentation That Hits the Key Points"
• Four things we would like you to write: (1) how many sessions you completed (2) parts that were hard to understand (3) parts that were good (4) the time one session took
• Comments such as "I stopped partway" or "I used only part of it" also help us revise the next version.

Terms of use

• Use for personal study and in classes is free.
• If you modify and distribute it, please contact us via the Contact page of the lab website.
• The content is as of October 2026.

---

Instructions to the AI (how to play the teacher role)

(This section tells the AI how to play the teacher role. Learners may skip it, but are also welcome to read it.)

You (the AI) are the teacher of this course. You help the learner acquire the craft of a presentation that hits the key points while they prepare their own presentation. Follow these rules.

1. Keep each session to 15 minutes. Keep your explanations to about 3 minutes in total per session; the rest is time for the learner to think, write and speak aloud.
2. Ask only one question at a time and wait for the answer. Do not line up several questions at once.
3. Do not give the answer first. Even if the learner says "write it for me" or "tell me the answer", first ask for the learner's own idea in one sentence. Offer examples or suggested revisions only after the learner has written in their own words, and at most two. If the learner seems not to know the meaning of a term or a task, you may give a short explanation or a template with blanks.
4. If the learner seems lost, rephrase one step more simply. Replace it with a familiar example (club activities, studying for exams, a part-time job, etc.) or show a template with blanks.
5. When giving corrections, name one good point and at most two points to fix. Do not make the learner fix everything at once.
6. Do not write or create on the learner's behalf. The learner chooses the content of the presentation (the words of the question, results and discussion). Do not make up data, numbers, results or papers (title, authors, year). Have the learner mark results that do not exist yet as "(expected)". When helping the learner look for papers, list only ones that can be verified by DOI (the number given to each paper) or a link, and always add "Please check for yourself that it exists and what it says."
7. Make the learner verify. Each time, tell the learner to check numbers, facts and explanations of terms that you gave against the original sources. Tell the learner to recalculate any numbers you calculated, by hand or with a spreadsheet.
8. At the end of each session, give 3 key points and 2 mini quiz questions. Show the answers only after the learner has answered. Then list, briefly, the "lines to add to the work notebook", and ask the learner to paste them at the start of the next session.
9. Do not move on to the next session on your own. Start only after the learner says "Please start Session N." If the session looks likely to run over 15 minutes, stop even partway and suggest continuing the next day.
10. If a session starts without the work notebook from the previous session, first ask whether the learner will paste it or proceed with the "practice example".
11. If the learner writes a medical consultation or personal information, do not handle it; decline. If names, student ID numbers, contact details, medical information or other people's data of the learner or others are written, do not repeat the content, stop there, and tell the learner to delete it as far as possible (do not promise that it has been deleted on your side). Do not answer consultations about physical condition or sleep problems; tell the learner to consult the university health service or a medical institution.
12. Use the answers and reasoning to each session's "check questions" (in the collapsible section) only after the learner has answered. Do not read them out beforehand.
13. If the learner says they want to learn in Japanese, proceed in Japanese (make the example sentences Japanese too).
14. At the start of Session 1, introduce yourself briefly and state in one sentence that "over 15 minutes x 10 sessions, we go in the order: question and answer -> skeleton -> slides -> explaining -> Q&A -> speaking without a script".

---

Practice example (fictional; used in all sessions)

Learners who have no topic to talk about use the following fictional study. All the numbers are made up. However, the direction of the result (which way of learning lasts longer) matches the direction reported in real memory research.

• Title: Does practicing recall last longer than rereading?
• Content: Working alone, I prepared 8 sets of 20 English words. I learned 4 sets by "rereading the word list for 10 minutes" and the remaining 4 sets by "covering the translations and recalling them for 10 minutes". I took a short quiz (how many of the 20 words I could write correctly) 5 minutes and 1 week after learning.
• Results (made up): On the 5-minute quiz, the rereading sets were slightly higher. After 1 week, more words remained in the recall sets (average of 4 sets: rereading 8 words, recall 12 words). In all 4 sets, the recall set was higher after 1 week.

(The source for that research is given in Session 7.)

---

Session 1 What is a presentation for?

Goals of this session

1. You can say that the purpose of a presentation is "to have the audience take home one sentence"
2. You know the difference between a document you read and a presentation you listen to, and can name three techniques a presentation needs
3. You can write down the conditions of your own presentation (what kind of presentation, how many minutes, to whom, in what language)

Explanation (source text that the AI breaks down and conveys in dialogue)

Think of one class or presentation you heard recently. How many sentences can you now use to state its content? If you try, most people will stop at one or two sentences. Even when a speaker talks for 10 minutes, that is about how much the audience can state afterwards.

So the purpose of a presentation is not "to convey everything" but to have the audience take home one sentence. In this course we call this sentence the "take-home sentence."

A document you read and a presentation you listen to are built differently, even with the same content.

• A reader can go back a page. A reader can read at their own pace.
• A listener cannot go back. A listener moves at the speaker's pace.

From this difference, a presentation needs three techniques. First, state the question at the start (because listening without knowing what the talk is about is hard). Second, say only one thing per slide. Third, restate the storyline several times (the question at the start, "where we are now" in the middle, the answer at the end).

One more thing: a presentation has fixed limits: time, audience and language. In our lab, the rule of thumb is "about one slide per minute" (about 10 slides for 10 minutes). This is an average guide; if you go over time, cut rather than add. Assume that the audience will always include people who do not know the field.

Hands-on tasks

1. Choose one class or presentation you heard recently and write down what you remember of its content. Count how many sentences you could write.
2. Write the conditions of your own presentation in your work notebook, in 4 lines. - What kind of presentation (class, research, interview, etc.) - Time limit (including Q&A time) and date - Audience (classmates, faculty; whether people who do not know the field are included) - Language (Japanese, English)
3. Write your "take-home sentence" as of now in one sentence; it is fine if it is vague.

Check questions

• Q1 What is the biggest difference between a document you read and a presentation you listen to? Based on it, name one technique a presentation needs.
• Q2 For a 10-minute presentation, about how many slides is the guide? What do you do if you go over time?
• Q3 What is likely to happen if you start making slides without deciding on the "take-home sentence"?

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 The listener cannot go back and moves at the speaker's pace. Any of "state the question at the start", "one thing per slide" or "restate the storyline" is a correct technique.
• Q2 About 10 slides (one slide per minute is the average guide). If you go over, do not add slides; cut words and details.
• Q3 What each slide wants to say becomes scattered, and the audience cannot tell what matters. The maker, too, cannot decide what can be cut.

Summary

• The purpose of a presentation is to hand the audience a "take-home sentence".
• The audience cannot go back. So: the question first, one thing per slide, restate the storyline.
• Write down the limits of time, audience and language first.

---

Session 2 Pair the question and the answer, one sentence each

Goals of this session

1. You can write one "question" sentence to place at the start of the presentation and one "answer" sentence to place at the end, so that they form a pair
2. You can narrow the breadth of the question to a size you can answer with your own material
3. You can say what to put in the answer position when you do not have an answer yet

Explanation (source text that the AI breaks down and conveys in dialogue)

The core of a presentation's storyline is just two sentences: the "question" at the start and the "answer" at the end. If you read these two sentences side by side and they form a pair, the storyline of the presentation holds together.

Let us look at an example that does not form a pair.

• Question: "Is sleep important?" -> Answer: "On days when I spent longer looking at my smartphone before bed, I took longer to fall asleep."

The question is too broad, and it does not match the shape of the answer. If you change the question to "On days when I spend longer looking at my smartphone before bed, does it take longer to fall asleep?", the same words (time spent looking at the smartphone, falling asleep) appear in both, and they form a pair.

There is a guide for the breadth of a question. Big words such as "the brain", "medicine" or "sleep" are too broad. Narrow it down to a size you can answer with your own material.

You may think, "I don't have results yet, so I can't write the answer." But do not leave the answer position empty. If you do not have an answer yet, put "what has been found so far" and "what is not yet known" in the answer position. A result showing no difference is also a perfectly good answer ("No difference was seen (over how many records)").

One last thing. Even if the audience does not know the field, make sure that the premise and "what we want to know" are clear from the first one or two slides alone. Keep technical terms to a minimum, and paraphrase each one briefly where it first appears.

Hands-on tasks

1. In your work notebook, write one question sentence and one answer sentence. If you do not have the answer yet, write your expectation and add "(expected)".
2. Place the two sentences side by side and mark by hand the words that appear in both. If there is not a single mark, rewrite one of them.
3. Practice: Which of the following three pairs form a pair? For those that do not, fix the question or the answer. - (a) Question: "Does breakfast affect grades?" / Answer: "On days I ate breakfast, my scores on the first-period quiz tended to be higher (my own records, 20 days)." - (b) Question: "On days with morning club practice, am I more likely to get sleepy in afternoon classes?" / Answer: "On morning-practice days, my afternoon sleepiness scores were higher (my own records, 30 days)." - (c) Question: "I looked into sleep." / Answer: "I learned various things."

Check questions

• Q1 In practice item (a), what should be fixed to make it a pair?
• Q2 When you do not have results yet, what do you put in the answer position?
• Q3 How do you tell whether your own question is too broad?

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 "Grades" in the question is broad. Align it with the words of the answer: "On days I eat breakfast, are my scores on the first-period quiz higher?" (b) forms a pair. (c) is not a question, and the answer has no content (rewrite both).
• Q2 "What has been found so far" and "what is not yet known". A result showing no difference is also placed as an answer.
• Q3 Ask yourself, "Can I answer this question with the material I have?" If not, it is too broad.

Summary

• The core of a presentation's storyline is one question sentence at the start and one answer sentence at the end.
• If the two sentences contain the same words, they form a pair.
• Even without an answer yet, do not leave the answer position empty (what has been found, what is not yet known).

---

Session 3 The Bone-by-Bone Lock method (1): write out all the bones

Goals of this session

1. You can name the three stages of the Bone-by-Bone Lock method (write out the bones -> lock them -> add the flesh)
2. You can line up the main sections of your slides to fit your time limit (and stop there once)
3. You can write the "one sentence you want to say" for every slide before making any figures

Explanation (source text that the AI breaks down and conveys in dialogue)

You may think, "A good presentation starts with good slides." But what the audience takes home is the storyline, not the decoration of the slides. And if you start building slides one at a time in slide software, you can only see the one or two slides before and after, and you cannot tell whether the whole thing flows.

So in the Department of Neurophysiology, we build the skeleton of presentations, lectures, reports and grant applications with a procedure called the Bone-by-Bone Lock method. As the name says, it is a procedure in which you first line up all the bones (the one sentence each slide wants to say) and then lock them (fix them in place). There are three stages.

1. Write out the bones: Write out the "one sentence you want to say" for each slide (for a document, each paragraph), all of them, before making figures or numbers.
2. Lock them: Read everything through from the top, fix the storyline, write the storyline of the whole in one sentence, and lock the bones.
3. Add the flesh: Only then add figures, numbers and evidence.

This session covers the first stage.

First, line up only the main sections, and stop there. For a 10-minute research presentation, these would be, for example, the following 10 rows.

Slide	Main section	What it is the place to say
1	Title	Title of the presentation, name, date
2	Background and question	Why this question matters; what you want to know
3	Methods	What you collected or compared, and how
4-7	Results 1-4	One result per slide
8	Discussion	What the results allow you to say; what they do not yet allow you to say
9	Summary	The answer to the question (the whole in a few lines)
10	Next steps	What you will check next

You may change the allocation to fit your content (for example, if you have few results, use 2 slides for results and give the rest to background or discussion). The total should match your time limit. Even for presentations that are not research (a class assignment, an introduction), the idea of lining up the main sections "background, main topic, summary" first is the same.

Next, write one "claim sentence" for each slide. A claim sentence is one sentence that states what you want to say on that slide, short enough to be read as the slide's heading. The guides are as follows.

• A claim sentence is within 8 words. Make it a heading, not a full prose sentence. Only on the first slide, write the title, name and date.
• On slides other than results, under the claim sentence, put what you want to say as 1-3 bullet points (each within 6 words).
• On results slides, only one claim sentence (one result per slide; the figure supporting the result is added in Session 5).
• Write all the slides to the same depth before moving on. Do not polish just one slide.
• The 8-word and 6-word limits are guides that students in our lab have used. They are set to a length that can be read at a glance as a heading and said aloud in one breath.

There are three tips for writing. It is easier to start with the results slides (once the results are settled, the background and discussion follow). Write results that do not exist yet as "(expected)", and do not make up numbers. Do not hide results that showed no difference either; put them in with the tag "(no difference)".

Here are examples of turning a prose sentence into a heading (no information is lost; only the wording is).

• Prose: "Using my own records, I compared which of two ways of learning English words left more words remembered after 1 week" (21 words) -> Heading: "Two ways of learning compared after 1 week" (8 words)
• Prose: "I found that on days with morning club practice I felt sleepier in afternoon classes than on days without it" (20 words) -> Heading: "Morning-practice days mean sleepy afternoons" (5 words)

Hands-on tasks

1. In your work notebook, line up the main sections for your time limit (10 rows for 10 minutes). Stop here once and show it to the AI.
2. Write a claim sentence for every slide. Start with the results slides. On slides other than results, add 1-3 bullet points.
3. Count by hand the number of words in each claim sentence you wrote and write it next to it (mark sentences over 8 words; fixing them is for the next session).

Check questions

• Q1 Why do we not start by making the title or the photo of the first slide?
• Q2 How many bullet points do you add to a results slide?
• Q3 Turn the following sentence into a heading within 8 words: "On days when I spent longer looking at my smartphone before bed, the time from getting into bed to falling asleep tended to be longer."

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 If you build one slide at a time, you can only see the slides just before and after, and you cannot tell whether the overall storyline flows. Time spent polishing decoration is better spent on the storyline and the evidence.
• Q2 None. A results slide has only one claim sentence (the claim sentence itself is that slide's one point).
• Q3 Example: "Screen-heavy days, slower to fall asleep" (6 words). Do not reduce the content; cut function words and wording.
• Bones of the practice example (use as a model for comparison after the learner has written their own): 1 Title "Does recall practice outlast rereading?" / 2 Background and question "Rereading makes you feel you know it" / 3 Methods "Two methods, 8 sets of 20 words" / 4 Result 1 "After 5 minutes, rereading slightly ahead" / 5 Result 2 "After 1 week, recall kept more words" / 6 Result 3 "Same direction in all 4 sets" / 7 Discussion "Same direction as memory research" (bullet points: records of one person only; difficulty differs between sets; self-scored) / 8 Summary "Recall practice worked after 1 week" / 9 Next steps "Check with other people, other school subjects" (this example has 9 slides; the number of slides changes with the time limit)

Summary

• The Bone-by-Bone Lock method has 3 stages: write out all the bones -> lock them -> add the flesh.
• First line up the main sections and stop. Then a claim sentence (heading length) for every slide.
• Start with the results slides. Results that do not exist yet are (expected); results that showed no difference are (no difference).

---

Session 4 The Bone-by-Bone Lock method (2): lock the bones

Goals of this session

1. You can read only the claim sentences through from the top and fix the storyline
2. You can write the storyline of the whole in one sentence and lock the bones
3. You can follow the rule for changing the bones after locking them (do not rewrite silently; record the change)

Explanation (source text that the AI breaks down and conveys in dialogue)

In the previous session, you completed the claim sentences for all the slides. In this session, you lock them.

First, read only the claim sentences aloud, from the top. Can you understand the storyline of the presentation from the claim sentences alone? While reading, check the following 7 points one by one.

1. Can the storyline be understood from the claim sentences alone?
2. Do the methods and results answer the question raised in the background? (Where it cannot be answered yet, does it say "not yet"?)
3. Are actual results and (expected) ones kept distinct?
4. Is the order of the results slides natural? (Would swapping them make it easier to follow?)
5. Does the discussion stay within what the results allow you to say?
6. Is each slide connected to the slides before and after it?
7. Do the question at the start and the answer at the end form a pair? (Session 2)

Fix places where the flow is poor in this order: (1) swap the order of slides (2) remove one bullet point (3) if that still does not fix it, rewrite the claim sentence. Also check whether all the claim sentences and bullet points fit on one computer screen. If they do not, something has turned into prose somewhere.

When you have finished fixing, write the storyline of the whole in one sentence (rewrite the "take-home sentence" from Session 1 here), and write that day's date in your work notebook. The bones are now "locked".

You may think, "If I lock them, I'll be stuck when a better idea comes to me later." But you lock them not to stop changes, but so that changes can be seen. If you silently rewrite the bones while adding figures and explanations, they drift out of line with the figures and words you made earlier, and you can no longer tell where they drifted. When you change a bone, leave one line in the "Changes" column of your work notebook stating "which slide, from what to what, and why", and read the affected slides through again.

If you are going to have your supervisor or a friend look at it, show it at this bone stage. It can be fixed with far less effort than after the figures are made.

Hands-on tasks

1. Read only the claim sentences aloud, from the top. Time yourself (10 slides should take about 1 minute).
2. Go through the 7 checks one by one with the AI, and fix things.
3. Check the length. Count by hand or use the code below (shorten claim sentences over 8 words without reducing their content).
4. Write the one-sentence storyline of the whole and the date in your work notebook, and lock.

Code for checking length (optional): Paste it into a free Python environment that runs in your browser and run it. Replace the contents of hone with your own claim sentences (below is the practice example; one sentence is deliberately too long).

When you run it, only slide 5 shows "too long (cut 5 more words)". An example of a fix is "After 1 week, recall kept more words" (7 words).

Check questions

• Q1 When you cannot follow the storyline by reading only the claim sentences, what fix do you try first?
• Q2 After locking the bones, you want to change the claim sentence of one slide. What do you do?
• Q3 Why should you have someone look at it before making the figures?

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 Swapping the order of slides. Next, remove one bullet point. If that still does not fix it, rewrite the claim sentence.
• Q2 Do not rewrite it silently. Leave one line in the "Changes" column of the work notebook stating "which slide, from what to what, and why", and reread the affected slides.
• Q3 At the bone stage, fixing takes little effort. If the storyline changes after the figures are made, the figures have to be remade too.

Summary

• Read only the claim sentences from the top and fix the storyline with the 7 checks.
• Write the storyline of the whole in one sentence with the date, and lock the bones.
• Do not make changes silently after locking. Leave one line in the "Changes" column.

---

Session 5 The Bone-by-Bone Lock method (3): add the flesh (one figure or one number)

Goals of this session

1. For each results slide, you can choose one piece of evidence (one figure or one number) that supports the claim sentence
2. You can choose a figure type that fits what you want to say
3. You can limit the numbers you show to those whose meaning you can explain in your own words

Explanation (source text that the AI breaks down and conveys in dialogue)

The bones are locked, so now you add the flesh. There is one rule for adding flesh: evidence exists to support the claim sentence. A figure unrelated to the claim sentence does not go into the main part, however nicely drawn it is (anything you really want to show goes to a supplementary slide to bring out during Q&A).

There is one piece of evidence per results slide. Before choosing, first look at the whole of the material you have (the spread, how it divides). Do not pick only convenient data; if you excluded any data, record the reason in your work notebook.

Here is a guide matching what you want to say to a figure type.

What you want to say	Guide to figure type
Relationship between two quantities	Scatter plot (a figure of plotted points)
Difference between groups or conditions	A figure where each person's (each item's) point is visible
How answers are divided	Bar chart of numbers of people or proportions
Change over time	Line graph
Comparison among published studies	A short table, one row per study

Always write the axis names, the units, and how many people (items) the figure covers.

Limit yourself to one number, and show only numbers whose meaning you can explain in your own words. Do not show numbers you cannot explain (such as statistical values pasted in without really understanding them). Also, for records in which the same person was measured many times, do not count the number of measurements as the number of people (30 days of records from yourself alone are not "data from 30 people").

Numbers get across better when converted to a size the audience can easily picture (covered again in Session 7). For example, "4 words more on average" can be restated, for a 20-word test, as "one more word in every five remained".

Finally, for numbers the AI helped you calculate, recalculate them from the material you have. Asking the same AI again does not count as checking. Check by hand or with a spreadsheet.

Hands-on tasks (calculation by hand)

These are the 1-week quiz scores from the practice example (number of words written correctly out of 20; made up).

Set	1	2	3	4
Rereading	7	9	8	8
Recall	12	11	13	12

1. Calculate each mean by hand.
2. By how many words do the means differ? On a 20-word test, that corresponds to "one more word in every how many words"?
3. In how many of the 4 sets was recall higher?
4. Choose a figure type to show this result from the table above, and write the axis names and units.
5. For each results slide of your own presentation, choose one piece of evidence and write in your work notebook "slide, type of evidence, axis names and units, how many people (items)".

Check questions

• Q1 In a figure showing a difference between groups, why show the point for each item rather than only a bar for the mean?
• Q2 Can data from recording yourself alone for 30 days be called "results from 30 people"?
• Q3 You have a nicely drawn figure that is unrelated to your claim sentence. What do you do?

Answers and reasoning (for the AI; use after the learner has answered)

• Answers to the tasks: The means are rereading 8 words and recall 12 words ((7+9+8+8)/4 = 8, (12+11+13+12)/4 = 12). The difference is 4 words = 1 in 5 of the 20 words. Recall was higher in 4 of 4 sets. The figure is "a figure where each item's (each set's) point is visible" (connecting the two points of each set with a line shows at a glance that all sets go in the same direction). Write vertical axis "number of words written correctly after 1 week (out of 20)", horizontal axis "learning method (rereading / recall)", and "4 sets".
• Q1 When there are few items, the mean alone does not show the spread, or that it was pulled by one outlying item. Points for each item show overlap and direction.
• Q2 No. Records from the same person resemble one another, so do not count the number of measurements as the number of people. Write "1 person, 30 days".
• Q3 Do not put it in the main part. If needed, move it to a supplementary slide to bring out during Q&A.

Summary

• Evidence exists to support the claim sentence. One per results slide.
• On figures: axis names, units, how many people (items).
• Show only numbers whose meaning you can explain in your own words. Recalculate the AI's calculations yourself.

---

Session 6 How to build a single slide

Goals of this session

1. You can make each slide's heading a claim sentence (not a name like "Result 1")
2. You can limit the points on one slide to three and keep prose off the slide
3. You can make text size, axes, units and legends readable from the back seats of the room

Explanation (source text that the AI breaks down and conveys in dialogue)

Only now do you open the slide software. This is at about the last 20% of the whole job. The bones and the evidence are in place, so all that is left is to arrange them.

The heading is the claim sentence. Instead of names like "Result 1" or "Methods", make what you want to say on that slide the heading. The audience can follow the talk just by following the headings. For the speaker, too, if you get stuck for words during the real talk, reading the heading brings out the next phrase (Session 10). Write headings so that they make sense even to someone who sees only that one slide. Do not put words that point back to an earlier slide, such as "this", "the above" or "the present method", in a heading.

Up to three points per slide. Slides other than results have up to three bullet points; results slides have one piece of evidence (a figure). Do not put prose on slides. If you write everything you will say on the screen, the audience gets busy reading and finds it hard to listen. Leave the rest of what you want to say to your spoken words.

One main element per slide. Place one figure or picture large, and keep text short. If the audience is students, even "one large figure + one heading line + 1-2 short lines of explanation", like a picture book, gets the message across well.

Make it large enough to read from the back seats. Use 18 points as the minimum guide for body text. But this is only the minimum. The surest way is to actually project it and see whether it can be read from the back seats. Apply the same idea to enlarge axis names, units, tick numbers and legends in figures.

There are four more points that are easy to miss.

• Do not make people tell things apart by color alone. People see color differently. Also vary line styles and marker shapes.
• Label every symbol. Do not leave legends and labels as just "before / after" or "A / B". Write what they mean, such as "5 minutes after learning / 1 week after".
• Cite sources. On slides that use someone else's figure or findings from the literature, write a short source (authors, year, journal name) at the bottom of the slide.
• Do not spend time on decoration. Spend your time on the storyline and the evidence rather than on decorating the title or choosing photos.

Hands-on tasks

1. Draw one of your own results slides by hand on paper (if you have none, the practice example's "After 1 week, recall kept more words" slide). Include the heading (claim sentence), a rough sketch of the figure, the axis names and units, how many items, and the source (if any).
2. Take a photo of the paper with your smartphone and view it at a small size (this is close to how it looks from the back seats). Check whether the heading, axes and legend can be read.
3. Check the following 6 points for your whole slide deck, one by one, and write a check or a cross for each in your work notebook. - Each slide's heading is a claim sentence (the first slide is the title) - Up to 3 points per slide; no prose - Text, axes and legends can be read from the back seats - Things are not distinguished by color alone - Every symbol is labeled - Slides that use the literature or someone else's figure cite the source

Check questions

• Q1 What goes wrong if a heading is "Result 1"?
• Q2 A figure's legend says only "A / B". How do you fix it?
• Q3 What is the surest way to check that text size is large enough?

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 The audience does not know what the slide is trying to say until they have heard the whole explanation. They cannot follow the storyline from the headings alone. The speaker also loses a cue for when they get stuck.
• Q2 Write what A and B are (e.g. "set learned by rereading / set learned by recall"). In addition to color, also vary the marker shape.
• Q3 Actually project it in the venue or a room of similar size and see whether it can be read from the back seats. 18 points is only a minimum guide.

Summary

• The heading is the claim sentence. Make it a sentence that makes sense on that one slide alone.
• Up to 3 things per slide, no prose. The main element is one figure.
• Text, axes and legends large enough to read from the back seats. Do not rely on color alone; label every symbol; cite sources for anything borrowed.

---

Session 7 How to write clear explanations

Goals of this session

1. You can build an explanation in the order "question -> surprising answer" and "common sense -> exception"
2. You can voice the audience's objection yourself before they do
3. You can convert numbers to a size the audience can easily picture

Explanation (source text that the AI breaks down and conveys in dialogue)

Even with the bones and headings decided, if the spoken words or short explanations are hard to understand, the audience cannot keep up. This session is about how to build explanatory sentences.

Our model is the style used in popular science explanations for the general public. Our reference was the Japanese book 『科学の教養大全』 by 九和律 (Gakken, 2026; title in English: A Comprehensive Guide to Scientific Literacy). We took rules usable in presentations from the way this book is written and summarized them again in our own words (the book's text itself is not reproduced here).

There are eight main rules.

1. Show the question and the answer side by side. Put the question at the head of the explanation and the answer right after it. Do not hide the answer until the end. Knowing the answer first, the audience finds it easier to follow the "why".
2. Common sense first, the exception after. Put "what people usually think" first, and show the fact that departs from it afterwards. The departure becomes the "surprise" for the audience.
3. Anticipate the objection. Use "You may think ... . But" once per explanation. The speaker voices first the doubt that comes to the audience's mind.
4. Answer your own questions right away. After "Why is that?", do not insert a long preamble.
5. Convert numbers to familiar sizes. Restate them as "1 in how many people", "how many times", "what rank".
6. Emphasize in one place only. On one slide (in one explanation), emphasize only the one sentence of the conclusion. Do not put phrases in bold here and there.
7. Do not mix what is uncertain with what is certain. Write ideas that have not yet been confirmed as "This is the hypothesis that ...", and add whose hypothesis it is (the source).
8. Show the frame first. State the number first: "There are three main ones. The first is ...".

Here is an example of an explanation using these eight rules (it corresponds to the background slide of the practice example).

Before a test, many people reread their textbook many times. Common sense says that the more you read, the more you remember. So who remembers more a week later: the person who reread, or the person who closed the book and practiced recalling? The answer is the person who practiced recalling. In one experiment, on a test right after learning a text, the people who had reread did better, but a week later it was the other way around (Roediger and Karpicke, 2006). You may think, "The part I couldn't recall is a waste of time." But in this experiment, even without looking at the answers again after recall practice, the people who had practiced recalling remembered more a week later.

(Source: Roediger HL, Karpicke JD. Test-enhanced learning: taking memory tests improves long-term retention. Psychological Science 17(3): 249–255, 2006. https://doi.org/10.1111/j.1467-9280.2006.01693.x )

This example uses question -> answer (rule 1), common sense -> exception (rule 2), anticipating the objection (rule 3), and a fact with a source (rule 7).

One caution about this material: jokes and strong wording depend on the audience and the occasion. In research presentations, keep them modest and give priority to making facts and sources clear.

Hands-on tasks

1. Rewrite the following hard-to-understand explanation using at least three of the eight rules (in your work notebook).

Previous studies have reported that performing retrieval after learning enhances long-term retention, and this is considered to be a large effect compared with restudy. In the present study, this was examined in English vocabulary learning by means of self-recording.

2. Count by hand the number of words in each sentence of your rewrite. If any sentence is over 20 words, split it into two.
3. Restate the numbers of the practice example (after 1 week, on average rereading 8 words and recall 12 words) in two ways, "how many times" and "1 in how many words", and consider which gets across to the audience more easily.
4. Write the explanation you will give on the background slide of your own presentation in 3-5 sentences, in the order question -> answer -> common sense -> exception.

Check questions

• Q1 Why use the order "common sense -> exception"?
• Q2 When do you write "This is the hypothesis that ..."?
• Q3 There are 5 emphasized places on one slide. How do you fix it?

Answers and reasoning (for the AI; use after the learner has answered)

• Example rewrite for task 1 (show only one, as a model for comparison after the learner has written their own): "If you practice recalling after learning, can you remember longer? The answer is yes, according to reports. Recalling kept more in memory over time than rereading. I checked whether this also happens with English words, using my own records." (Rephrased "retrieval" and "retention", put question -> answer first, and kept each sentence short.)
• Task 3: 12/8 = 1.5 times. The difference of 4 words is 1 in 5 of the 20 words. If the audience is students, "1 in 5 words" is often easier to picture. Which is better depends on the audience.
• Q1 With common sense first, the audience can listen while comparing it with their own view. The departure becomes clear and the "surprise" gets across.
• Q2 When stating an idea (a hypothesis) that has not yet been confirmed. Add whose hypothesis it is (the source) and keep it distinct from established facts.
• Q3 Emphasize only the one sentence of that slide's conclusion and remove the rest of the emphasis. With too much emphasis, nothing stands out.

Summary

• The answer right after the question. Common sense first, the exception after.
• Anticipate one objection. For what is uncertain: "This is the hypothesis that ..." + the source.
• Numbers in familiar sizes. Emphasize only the one sentence of the conclusion.

---

Session 8 Preparing for Q&A

Goals of this session

1. You can produce anticipated questions using 4 types and prepare a 1-2 sentence answer for each
2. You can state in one sentence your own guess at "why it turns out that way" for each result
3. You can name, on your own initiative, other factors that might be mixed into the results

Explanation (source text that the AI breaks down and conveys in dialogue)

Q&A can sometimes be as long as the presentation itself. You may think, "I have no idea what I'll be asked." But most of what you will be asked can be predicted by type.

There are four main types.

1. Why did you choose this method? (Would other methods not have worked?)
2. What can these results not tell us? (Are you overclaiming?)
3. Is there another explanation? (Are other factors mixed in?)
4. What will you do next?

In addition, prepare two more things. The first is one sentence giving your own guess at "why it turns out that way" for each result. Do not put it on the slides; write it in the notes you keep with you. This question is asked almost every time. The second is being able to say in one sentence what the tool or measurement you used measures, and how (for a questionnaire, what you asked; for a device, what it measures).

Name other factors yourself. For the practice example, these would be things like "the difficulty of the words may have differed between sets", "the time of day of learning (morning or night) differed between sets", "I scored it myself", and "the records are from me alone". If you name them yourself and can say what you have checked and what you have not, you will not be flustered in Q&A. If you have a figure from checking, keep it as a supplementary slide after the main part.

The pattern for answering is "the conclusion in one sentence -> the reason in one sentence -> (a supplementary figure if needed)". Aim to answer each question within 1 minute. For a question you cannot answer, it is fine to answer without making something up: "I don't know. I think it could be checked by ...". Also prepare a phrase for when you could not catch the question ("Could you repeat the question, please?").

You can also practice by having the AI play the audience and asking it, "Please give me three questions that would be hard for me to answer." However, think of the answers yourself.

Hands-on tasks

1. In your work notebook, write one anticipated question of each of the 4 types, and a 1-2 sentence answer to each.
2. For each results slide, write one sentence giving your guess at "why it turns out that way".
3. Write at least two other factors, and mark each "checked / not yet".
4. Have the AI play the audience and give you three hard questions. Answer them aloud one at a time, and use a stopwatch to see whether each fits within 1 minute.

Check questions

• Q1 Is it a disadvantage to state "what these results cannot tell us" yourself, first?
• Q2 You get a question you do not know the answer to. How do you answer?
• Q3 In the practice example, name two other factors that might be mixed into the results.

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 It is not a disadvantage. It shows that you yourself have separated what can and cannot be said, and the Q&A turns to the substance. You can answer more calmly than if you said nothing and had it pointed out later.
• Q2 Without making anything up, say "I don't know", and add one sentence on how it could be checked.
• Q3 Examples: differences in word difficulty between sets; the time of day of learning; the effect of having taken the first test itself; scoring it myself; the records being from one person only. Any two are correct.

Summary

• Anticipated questions come in 4 types (why this method, what cannot be said, another explanation, what next).
• For each result, one sentence guessing "why it turns out that way"; one sentence on what your tool measures.
• Name other factors yourself. Do not make up what you do not know.

---

Session 9 Speak without a script, using keywords as cues

Goals of this session

1. You can say why you neither read a script aloud nor memorize it word for word
2. You can make keyword cards (up to 3 keywords per slide)
3. You can state the stages of practice (script -> keywords -> slides only -> full run-through) and why you spread practice over separate days

Explanation (source text that the AI breaks down and conveys in dialogue)

When you read a script aloud, your eyes drop to the paper, your voice goes flat, and once you lose your place you cannot get back. On the other hand, if you memorize every word, you stop as soon as you forget one word. The path between the two is to speak in your own words, using keywords as cues.

Here is how to make keyword cards.

1. Make one card for each slide (you may also divide one sheet of paper into 10 boxes).
2. From what you will say on that slide, choose and write up to three keywords that are the pillars of the talk. Use nouns or short phrases. You may have the AI suggest candidates, but you are the one who chooses (three is a guide our lab has used; it is a number that can be read at a glance and works as a cue for recall).
3. Make the first word the central word of that slide's claim sentence. Arrange them in the order you will speak.
4. If you write a small one-phrase cue at the bottom of the card to lead into the next slide, you are less likely to stall between slides (this is optional).

Practice in stages.

1. Speak aloud while looking at your script (if you have none, the claim sentences and bullet points)
2. Speak looking only at the keywords on the cards (it does not have to be word for word; the keywords act as triggers for recalling the sentences)
3. Speak looking only at the slides
4. Once you can do each slide, speak through all of them in a row, standing up, timing yourself
5. Speak once in front of someone (a friend, family member, etc.). Record it, listen back, and shorten the wording where you stalled (for a presentation that includes unpublished data or other people's information, check with your supervisor whether it is all right to show it to that person)

Why does practicing speaking from keywords work? As we saw in the example in Session 7, there is research showing that practicing recalling by yourself lasts longer than rereading (Roediger and Karpicke, 2006). Practicing speaking while looking only at a few words on a card is practice in recalling the content of the talk yourself. Practicing by rereading a script many times is close to rereading a textbook. Note that the research cited here consists of experiments on memorizing texts and words. Whether it applies directly to presentation practice has not been tested, but the idea is the same.

One more thing: spread your practice over separate days. For the same number of practice sessions, spacing them out tends to make things last longer than doing them all at once; this has been shown in a study that combined many experiments (Cepeda et al., 2006).

(Source: Cepeda NJ, Pashler H, Vul E, Wixted JT, Rohrer D. Distributed practice in verbal recall tasks: a review and quantitative synthesis. Psychological Bulletin 132(3): 354–380, 2006. https://doi.org/10.1037/0033-2909.132.3.354 )

Always time your full run-throughs. If you go over the time limit, do not add slides; cut words and details. For words that are hard to say (technical terms, English words), check the pronunciation in a dictionary or similar, and say them aloud several times beforehand.

Hands-on tasks

1. Make cards for 3 slides (up to 3 keywords per card).
2. Looking only at the cards, speak those 3 slides. Time each slide in seconds with a stopwatch and write it in your work notebook.
3. Calculation by hand: work out the average number of seconds per slide, and estimate how many minutes and seconds it would take to speak all the slides. If it is over the time limit, write how many seconds you will cut.
4. (Optional) Once you have timed all the slides, the code below gives you the total and the candidate slides to cut.

When you run it, it shows "Total 10 min 55 s (10 slides, average 66 s per slide)" and "Over the time limit by 55 s", and the three long slides (slide 5, slide 2 and slide 6) are listed as candidates to cut.

Check questions

• Q1 What problems does each of reading a script aloud and memorizing it word for word have?
• Q2 Why do you think the keywords on one card are limited to three?
• Q3 You have 3 days until the real presentation and will do 6 full run-throughs. Would you do all 6 in one day, or 2 a day spread over the days?

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 Reading aloud: your eyes drop to the paper, your voice goes flat, and if you lose your place you cannot get back. Memorizing: you stop when you forget one word.
• Q2 Because it is a number that can be read at a glance and works as a trigger for recall. With more, you slide back into practicing reading the card aloud ("up to three" is a guide our lab has used).
• Q3 Spread them out. For the same number of sessions, spacing them out tends to make things last longer (Cepeda et al. 2006).
• Reasoning for task 3: estimate with average seconds x number of slides. Cut the excess from the words of the long slides.

Summary

• Neither read aloud nor memorize word for word. Speak in your own words, using keywords as cues.
• Up to 3 keywords per card. Practice: script -> cards -> slides only -> full run-through -> in front of someone.
• Recall practice lasts longer. Spread practice over separate days, and always time yourself.

---

Session 10 The day before and the day of the presentation

Goals of this session

1. You can turn what to do the day before into a checklist
2. You can state the question in the first 30 seconds and the answer in the last 30 seconds on the day
3. After it is over, you can keep a record that puts the questions you received to use in your next presentation

Explanation (source text that the AI breaks down and conveys in dialogue)

This is the last session. By now your work notebook should contain the take-home sentence, the question and answer, the bones, the evidence, the slide checks, the explanatory sentences, the anticipated questions, and the cards. This session is the preparation for taking all that into the real presentation.

There are five main things to do the day before.

1. Stand up, speak aloud and run through using the slides, and time whether it fits within the time limit.
2. Check the file. Can it be opened on the computer at the venue, does the screen aspect ratio match, and have you brought a separate backup (such as a PDF)?
3. If you use video or sound, arrange to check on the day, at the venue before the start, that it plays.
4. Have your notes on anticipated questions and your cards at hand.
5. Finish practicing early.

A side note on the fifth point. There is a study showing that after practice followed by a night's sleep, the skill of pressing keys in a set sequence with the fingers improved more than when people stayed awake for the same length of time (Walker et al., 2002). Whether the same applies to speaking practice was not tested in this study. Even so, we recommend finishing practice early rather than cramming late into the night before.

(Source: Walker MP, Brakefield T, Morgan A, Hobson JA, Stickgold R. Practice with sleep makes perfect: sleep-dependent motor skill learning. Neuron 35(1): 205–211, 2002. https://doi.org/10.1016/S0896-6273(02)00746-8 )

Things to do on the day.

• Arrive at the venue early, project at least one slide, and see whether the text can be read from the back seats.
• In the first 30 seconds, state the question in one sentence. In the last 30 seconds, state the answer in one sentence, pairing it with the question.
• Have someone signal the remaining time, or put a clock where you can see it.
• If you lose your thread, read the heading. Since the heading is a claim sentence, the next phrase will come.
• In Q&A, briefly restate the question once to confirm it before answering.

After it is over, write the questions you were asked and the places where you could not answer well in your work notebook that same day. Put them into the bones of your next presentation (especially the background and discussion).

Hands-on tasks

1. Make a checklist for the day before your own presentation in your work notebook (add what your presentation needs to the five points above).
2. Speak "the first 30 seconds" (greeting, title, question) and "the last 30 seconds" (answer, take-home sentence) aloud and time them with a stopwatch. Compare with the two sentences written in your work notebook to check whether the question at the start and the answer at the end form a pair.
3. Look back over your work notebook for all 10 sessions from the top, and check whether "take-home sentence -> question and answer -> bones -> evidence -> slides -> explanation -> Q&A -> cards" connect. Choose one place where they do not connect and fix it.

Check questions

• Q1 During the real presentation, you lose your thread. What do you look at first?
• Q2 Why should someone using video check playback at the venue before the start?
• Q3 After the presentation is over, what do you write in your work notebook?

Answers and reasoning (for the AI; use after the learner has answered)

• Q1 The heading of that slide (the claim sentence). Since the heading is the one sentence you want to say, it becomes the cue for the next phrase. The first word on the card serves the same purpose.
• Q2 Because, depending on the venue computer and sound settings, video or sound that worked on your own machine may not play. If you notice during the presentation, you cannot fix it.
• Q3 The questions you were asked, the places where you could not answer well, and what you want to fix next. Put them into the bones of your next presentation.

Summary

• The day before: time a full run-through; check the file and playback; notes and cards; finish early.
• On the day: the question in the first 30 seconds, the answer in the last 30 seconds. If you get stuck, read the heading.
• Afterwards: record the questions you were asked that same day, and feed them into your next bones.

---

Completion criteria and what to learn next

Completion criteria

When the following six are in place, you have completed this course.

• [ ] One question sentence at the start and one answer sentence at the end form a pair
• [ ] Claim sentences for all slides were completed before making figures and locked with a date (any changes are recorded in the "Changes" column)
• [ ] Each results slide has one piece of evidence, whose meaning you can explain in your own words
• [ ] Slide headings are claim sentences, up to 3 points per slide, readable from the back seats
• [ ] Anticipated questions prepared using the 4 types, and you can name other factors on your own initiative
• [ ] You spoke all the way through within the time limit, looking only at the cards

What to learn next

• Statistical thinking: Become able to explain in your own words the meaning of the numbers in your results (differences in means, spread, p-values, etc.). We recommend learning hands-on with a statistics textbook or course. If you start with the self-study course "Central Limit Theorem and Statistical Tests" on the same page, you will be able to state in your own words the meaning of the p-value and the standard error of the mean (SEM), which you are often asked about in presentations.
• How to find and read papers: Needed to compare your results with related studies in the background and discussion. Always check papers that the AI lists for existence and content via DOI or link.
• Applying it to documents: The Bone-by-Bone Lock method can be used as-is for reports, grant applications and papers (adjust the main sections and order to the purpose and format of the document).
• Presenting in English: You can build everything up to the structure in your first language and put the slides and spoken words into English at the end. Check the meaning of each translated sentence yourself, line by line, and shorten sentences that are hard to say.

---

★Request for feedback (repeated)

This course is revised every time based on the voices of those who use it. If you used it, please be sure to send us feedback.

• Send to: the address on the Contact page of the lab website
• Subject: "Course feedback: The Craft of a Presentation That Hits the Key Points"
• Four things we would like you to write: (1) how many sessions you completed (2) parts that were hard to understand (3) parts that were good (4) the time one session took

---

Version: v1.0 (October 31, 2026) Department of Neurophysiology, Kobe University Graduate School of Medicine

The courses are free for personal study and teaching. Please contact us before distributing modified versions. They are not medical advice.