Lesson 10: Defense Slides Blueprint

Slide 1 0:00 – 0:30 (30 sec)

Title Slide

On the slide:

Full title: "Application of Kit-Build Concept Mapping for Japanese Reading Comprehension"
Your name, NIM, study program
State Polytechnic of Malang
Date

What to say "Good morning. My name is Dicha Zelianivan Arkana. Today I will present my thesis on applying Kit-Build concept mapping for Japanese reading comprehension among beginner learners."

Tip: Don't read the title. Just introduce yourself and move on.

Slide 2 0:30 – 1:30 (1 min)

The Problem

On the slide:

Japanese uses 3 scripts: Hiragana, Katakana, Kanji
Indonesian learners face structural differences (SOV, conjugations, particles)
709,479 Japanese learners in Indonesia (Japan Foundation, 2018) — motivation is high, but difficulty is real
Memorisation alone doesn't work — need structured understanding

What to say "Japanese has a unique three-script system — Hiragana, Katakana, and Kanji. For Indonesian learners whose native language uses Latin script, this creates a substantial barrier. Indonesia has over 700,000 Japanese learners, so motivation is high. But motivation alone doesn't remove the structural difficulty. Memorisation without understanding doesn't stick. So the question is: how can we help learners organise meaning, not just remember words?"

Slide 3 1:30 – 2:30 (1 min)

The Gap & Kit-Build Solution

On the slide:

Concept mapping helps — but open-ended maps are hard to assess automatically
Kit-Build (Hirashima, 2015): close-ended reconstruction with automatic diagnosis
Used in EFL reading (Alkhateeb 2016, Andoko 2020) — but NOT yet in Japanese
Research gap: no Kit-Build platform for Japanese reading comprehension

What to say "Concept mapping can help learners externalise their understanding. But open-ended maps are hard to grade automatically — every student creates different maps. Kit-Build solves this. The teacher prepares a goal map, decomposes it into a kit, and learners reconstruct it using the same components. The system can then compare maps automatically at the proposition level. Kit-Build has been used for English reading, but not yet for Japanese. That's the gap this study addresses."

Slide 4 2:30 – 3:30 (1 min)

Yomilink — What You Built

On the slide:

Screenshot of Yomilink interface (Kit-Build reconstruction view)
Feature list: goal map prep → kit generation → learner reconstruction → automatic diagnosis
Plus: consent, pre/post-test, TAM questionnaire, analytics
Open-source, MIT license, modern web stack

What to say "I developed Yomilink — a modern web-based platform that implements the full Kit-Build workflow. It handles goal map preparation, kit generation, learner reconstruction, and automatic proposition-level diagnosis. Beyond the core Kit-Build features, Yomilink also integrates the entire research pipeline — consent collection, pre-test and post-test delivery, TAM questionnaire, and researcher analytics. It's open-source under the MIT license."

Tip: If you can do a 15-second live demo, do it. If not, a clear screenshot is enough.

Slide 5 3:30 – 5:00 (1.5 min)

Methodology

On the slide:

Design: one-group pretest-posttest pilot
N = 35 beginner Japanese learners, 2 Marketing Management classes
3 sessions: Day 1 intro (offline), Day 2 calibration (offline), Day 3 main (online)
Reading material: "わたしのうち" (Minna no Nihongo Ch. 10), 7 sentences, 13 Kanji with furigana
Goal map: 7 propositions (spatial + functional relationships)
Pre-test → Kit-Build activity → Post-test → TAM → Feedback (all in one online session)
10 multiple-choice questions, same at pre and post, passage visible, dictionaries allowed

What to say "The study used a one-group pretest-posttest pilot design. Participants were 35 beginner Japanese learners from two Marketing Management classes at State Polytechnic of Malang. We conducted three sessions — an introduction, a calibration session where we refined the test instrument, and the main online session. The reading material was a passage from Minna no Nihongo Chapter 10 describing a neighbourhood. The goal map contained 7 propositions. During the main session, participants completed a pre-test, the Kit-Build reconstruction activity, a post-test, the TAM questionnaire, and open feedback — all through Yomilink in the browser."

Slide 6 5:00 – 6:00 (1 min)

Results: Score Direction

On the slide — big numbers:

	Mean	Median	SD
Pre-test	7.06	8	2.41
Post-test	7.77	8	2.34
Delta	+0.71	1	1.99

Trend: 20 improved (64.5%), 6 same (19.4%), 5 declined (16.1%)

Paired t-test: t = 2.219, p = 0.017, Cohen's d = 0.399

Wilcoxon signed-rank: W = 52.000, p = 0.014

What to say "The mean score increased from 7.06 in the pre-test to 7.77 in the post-test — a gain of 0.71 points out of 10. About 64% of participants improved. The paired t-test confirmed this increase as statistically significant, with p = 0.017 and a small-to-medium effect size. The Wilcoxon signed-rank test, which doesn't assume normality, confirmed this result at p = 0.014."

Slide 7 6:00 – 7:00 (1 min)

Results: Kit-Build Performance

On the slide:

Kit-Build score formula: S = (M / T_Goal) × 100%
Goal map: 7 propositions
Mean Kit-Build score: ~73% (≈ 5 of 7 propositions reconstructed)
Range: 0% to 100% — wide variation
Spearman ρ = 0.198, p = 0.284 → NOT significant

What to say "For the Kit-Build reconstruction, the mean score was about 73%, meaning learners reconstructed roughly 5 of the 7 propositions on average. There was wide variation — some learners reconstructed the full map, others struggled significantly. Interestingly, the Spearman correlation between Kit-Build score and score change was not significant. This may reflect construct contamination — the Kit-Build task requires Kanji decoding while the multiple-choice test provides recognition cues. So the two measures capture different aspects of comprehension."

Slide 8 7:00 – 8:00 (1 min)

Results: The Ceiling Effect

On the slide:

Pearson r = −0.549, p = 0.001 (pre-test vs. gain — strong negative)
Median split: median pre-test = 8

Group	n	Mean Pre	Mean Post	Mean Δ
Low (pre ≤ 8)	19	5.68	7.21	+1.53
High (pre > 8)	12	9.42	9.08	−0.33

Welch t = 2.701, p = 0.013. Hedges' g = 0.988 (large effect).

What to say "An important finding: the ceiling effect. Participants who scored low on the pre-test gained an average of 1.53 points. Those who scored high actually declined slightly — minus 0.33. The difference is large, with Hedges' g near 1.0. This means the test was too easy for advanced beginners. The measurable range of improvement was compressed. Future studies need harder or more variable material."

Slide 9 8:00 – 8:45 (45 sec)

Results: TAM & Feedback

On the slide:

TAM (5-point Likert): Perceived Usefulness > 3.5, Perceived Ease of Use > 3.5 → positive acceptance
Top feedback: "Helps learning" (24), "More vocabulary needed" (14), "Initial confusion" (12)
Engagement dropped: 87% → 61% (fatigue from single-session design)

What to say "Participants reported positive initial acceptance — both perceived usefulness and ease of use were above the 3.5 threshold. The most common positive feedback was that the platform helped them learn. The most common suggestion was to add more vocabulary practice. Engagement did drop from pre-test to post-test, which we attribute to fatigue from the single-session design."

Slide 10 8:45 – 9:30 (45 sec)

Contributions

On the slide — 3 bullet points:

Platform: Yomilink — modern open-source Kit-Build platform (MIT license)
Evidence: First pilot-tested procedure for Kit-Build in Japanese reading comprehension
Insight: Orthographic complexity (Kanji) may moderate Kit-Build's effect — concept mapping frameworks from alphabetic languages need adaptation

What to say "This study makes three contributions. First, Yomilink — a modern open-source platform under the MIT license. Second, the first pilot-tested procedure for applying Kit-Build to Japanese reading comprehension. Third, an important insight: concept mapping frameworks developed for alphabetic languages like English may need adaptation for languages with complex orthography like Japanese. The script-decoding barrier may attenuate Kit-Build's effect on comprehension."

Slide 11 9:30 – 10:30 (1 min)

Limitations

On the slide — grouped:

Category	Limitation
Design	No control group → cannot claim causation
Instrument	Same questions in pre-test and post-test → possible practice effect
Conditions	Dictionary + passage visible throughout → measured assisted comprehension
Carryover	Day 2 familiarisation → inflated pre-test scores → ceiling effect
Sample	N = 35 from two classes in one program → not generalisable

What to say "Key limitations: no control group, so we cannot claim causation. Same questions used twice, which may introduce practice effect. Dictionaries and the passage were always available, so we measured assisted comprehension. The familiarisation session may have inflated pre-test scores. And the sample is limited to two classes in one program."

Critical: Name your limitations before the examiners do. It shows mastery and honesty.

Slide 12 10:30 – 11:30 (1 min)

Future Work

On the slide:

Controlled experiment with comparison group (summarisation or selective underlining)
Parallel or equivalent test forms for pre-test and post-test
Delayed post-test to measure retention
Larger, more diverse sample across proficiency levels
Stronger scaffolding: pre-taught vocabulary, inline glosses, simplified material
Split online session across multiple days to manage cognitive load

What to say "For future research, I recommend a controlled experiment with a comparison group, parallel test forms, a delayed post-test for retention, and a larger sample. I also recommend stronger scaffolding — such as pre-taught vocabulary lists or inline glosses — to better isolate the structural-understanding component from the script-decoding barrier."

Slide 13 11:30 – 12:00 (30 sec)

Closing Statement

On the slide — one sentence, large text:

"Yomilink is feasible for a preliminary Kit-Build activity in Japanese reading comprehension, with positive initial score direction and user acceptance."

What to say "To conclude: this study shows that Yomilink is feasible for conducting a preliminary online Kit-Build activity in Japanese reading comprehension, with positive initial score direction and user acceptance. Future controlled studies are needed to confirm effectiveness. Thank you."

Tip: This is your elevator pitch. Memorise it word for word. If examiners ask "what's the main finding?", this is your answer.

Slide 14 —

Appendix Slides (don't present — have ready)

Keep these hidden but available for Q&A:

Full TAM per-item breakdown
Kit-Build score distribution histogram
Box plot: pre-test vs. post-test
Dumbbell chart: individual score changes
Scatter plot: Kit-Build score vs. delta
Engagement distribution chart
Qualitative feedback theme table
Excluded participants table
Shapiro-Wilk results
Sensitivity analysis (outlier-excluded t-test)

Why appendix slides? Examiners love when you can pull up a chart or table on demand. It shows you know your data deeply. Don't present these — just have them ready.

What	Value	When to use
N	35 (31 valid)	"How many participants?"
Pre-test mean	7.06 / 10	Baseline context
Post-test mean	7.77 / 10	Result headline
Delta	+0.71	Score direction
Paired t	2.219, p = 0.017	Significance claim
Cohen's d	0.399	Effect magnitude
Wilcoxon	W = 52, p = 0.014	Nonparametric confirm
Kit-Build mean	~73%	Reconstruction quality
Propositions	7	Goal map complexity
Spearman ρ	0.198, p = 0.284	No KB↔test link
Pearson r	−0.549, p = 0.001	Ceiling effect evidence
Hedges' g	0.988	Median-split group diff
TAM threshold	> 3.5 / 5	Positive acceptance
Improved %	64.5%	Score trend

Avoid Saying	Say Instead
"Kit-Build improved comprehension"	"Students showed a positive score direction after the activity"
"The study proved that..."	"The results suggest that..."
"Learning gain"	"Within-group score direction"
"The platform is effective"	"The platform was feasible for this pilot activity"
"Unaided comprehension"	"Assisted reading comprehension"
"We found a significant improvement"	"We found a statistically significant positive score direction"
"Novice learners"	"Beginner Japanese learners"

Q1

"Why no control group?"

"This was a preliminary feasibility pilot. The primary goal was to validate the platform workflow, instruments, and initial score direction. A controlled experiment with a comparison group is the recommended next step, and this pilot provides the baseline data to design that study."

Q2

"Can you say Kit-Build improved reading comprehension?"

"Only cautiously. The safe claim is that students showed improved scores after using Yomilink. Because there was no control group, the study cannot fully separate the Kit-Build effect from practice effect, material familiarity, or repeated exposure. The result is a positive score direction for future controlled testing, not causal evidence."

Q3

"Why allow dictionaries?"

"The study measured assisted reading comprehension — how beginners perform when script decoding is partially scaffolded. This is realistic for beginner learning conditions. Dictionary access, the passage with furigana, and the multiple-choice format all reduce the script-decoding barrier. The measured construct is structural understanding under supported conditions, not unaided memorisation."

Q4

"Why did the Kit-Build score not correlate with test improvement?"

"This likely reflects construct contamination from Kanji. In English Kit-Build studies, both the reconstruction task and the comprehension test measure the same thing — understanding of relationships. In Japanese, the Kit-Build score captures both structural understanding and Kanji decoding, while the multiple-choice test captures primarily structural understanding with recognition cues. The two measures are not directly comparable."

Q5

"Is Yomilink better than the original Kit-Build platform?"

"This study does not test that. The contribution is a modern reimplementation with easier web-based deployment, straightforward extensibility, and integrated research features. A direct platform comparison would be needed to claim superiority."

Defense Slides Blueprint

Time Budget

Title Slide

The Problem

The Gap & Kit-Build Solution

Yomilink — What You Built

Methodology

Results: Score Direction

Results: Kit-Build Performance

Results: The Ceiling Effect

Results: TAM & Feedback

Contributions

Limitations

Future Work

Closing Statement

Appendix Slides (don't present — have ready)

Quick Reference: Numbers to Memorise

Phrase Cheat Sheet

Top 5 Examiner Questions & Answers

"Why no control group?"

"Can you say Kit-Build improved reading comprehension?"

"Why allow dictionaries?"

"Why did the Kit-Build score not correlate with test improvement?"

"Is Yomilink better than the original Kit-Build platform?"

Primary Source

Reminder