Education · Language brand
Built on the evidence,
not on a hunch.
Vocabulary apps where a cat does the nagging. One language, one character, and the same evidence-backed schedule under all of them. None of us is a linguist, so instead of inventing a method we went to the meta-analyses and built the 6 things they agree on — spacing, retrieval, dual coding, feedback, level-matched input and interleaving.

The language brand inside Pawblesse Education
5 characters · 5 languages
- 3
- 7
- 14
- 30
- 60
- Characters
- 5One per language, across 5 languages.
- Released
- 2 of 5Across 4 store listings on iOS and Android.
- Methods used
- 6Every one of them shared by every app in the brand.
- Papers cited
- 6Each method links to the study it came from, further down.
- CookieNyan · English
- MochiNyan · Japanese
- DalgoNyan · Korean
- BaoNyan · Chinese
- ChurroNyan · Spanish
NyanNyanVoca · the lineup
One language, one cat.
Each character is named after something you eat where the language is spoken — mochi, cookie, dalgona, bao, churro. 2 of the 5 have shipped and the rest have not, which is the whole reason the badges are there.
CookieNyan
Teaches English
iOS · Out nowAndroid · Out nowMochiNyan
Teaches Japanese
iOS · Out nowAndroid · Out now- No artwork for DalgoNyan yet
DalgoNyan
Teaches Korean
iOS · Planned - No artwork for BaoNyan yet
BaoNyan
Teaches Chinese
iOS · Planned - No artwork for ChurroNyan yet
ChurroNyan
Teaches Spanish
iOS · Planned
NyanNyanVoca · the apps
Both of them, without the scrolling.
Switch between them here rather than scrolling past one to reach the other. Every screenshot is the app as it ships.
The evidence
Six methods, and where each one comes from.
Every row names the finding, the effect the research reported, what the app actually does about it, and the paper you can go and read. Nothing here is our own study — we did not run one.
| Method | Reported effect | What the app does | Source |
|---|---|---|---|
| Spaced repetitionA word you just got right is the word you are about to forget. | Reported effectMedium–large48 experiments, 3,411 learners | What the app doesGet a card right and it comes back in 3 days, then 7, 14, 30, 60. Say you are unsure and the clock resets to the start. Kanji and missed JLPT questions run their own 1–3–7–14–30 ladder and only graduate after three passes on three separate days — answering the same card again the same afternoon does not count. Set an exam date and the intervals are capped, so nothing is scheduled to come back after the test. | SourceKim & Webb (2022), Language LearningAcross 48 experiments with 3,411 learners, spacing beat cramming at a medium-to-large effect size. |
| Retrieval practiceBeing asked beats being shown again. | Reported effectMediumBeat restudying on every final measure | What the app doesCards run in two beats — you see the word, then you try to produce the meaning before the answer appears. After that the same word returns the same day wearing a different question: meaning, reverse, listening, pronunciation, example sentence, cloze. Eight formats, so a word you have merely recognised does not pass as a word you know. | SourceBarenberg (2021), Applied Cognitive PsychologyIn a school study, learners who were quizzed outperformed learners who restudied on every final measure, and the gap widened with each round. |
| Dual codingA picture and a sound give a word two ways back into your head. | Reported effect+89%Transfer-test advantage over words alone | What the app doesEvery word arrives as an illustration, native audio and the sentence it lives in, together rather than one after another. The card art is preloaded for the whole session, so the same image is there in the quiz and not just on the card you first met. | SourceMayer, cognitive theory of multimedia learningAcross 11 controlled experiments, learners given words with pictures outperformed words-only learners on transfer tests by an average of 89%. |
| Corrective feedbackA wrong answer is only useful if it comes back. | Reported effectd = 0.64Larger in foreign-language settings | What the app doesMiss one and you get the reading, the meaning and why the sentence needs that word — not just a red mark. Then it returns three ways: first in the day's closing quiz, again in Mochi Notes as a drill you can run on its own, and again on the spacing ladder, where it starts from the beginning. | SourceLi (2010), Language LearningCorrective feedback averaged d = 0.64 across studies, with larger effects in foreign-language settings than in second-language ones. |
| Comprehensible inputYou learn a new word from a sentence you can already almost read. | Reported effectSmall–medium34 studies, 3,942 learners | What the app doesThe short reading at the end of a session is assembled from the cards you studied that day, which is what keeps the share of words you already know high enough for the new ones to stick. It is short — a few sentences — and it is meant to sit next to the drilling, not replace it. | SourceNakanishi (2015), TESOL QuarterlyA meta-analysis of 34 studies with 3,942 learners found reading at the right level improved comprehension, vocabulary and fluency; 95–98% known words is the range where new vocabulary is picked up. |
| Interleaving, in the right orderMixing helps — but not on day one. | Reported effectDelayed testsWhere the gain shows up — reversed for beginners who start mixed | What the app doesA session moves in blocks: review, new words, phrases, grammar, writing, quiz. Across the day the same word crosses between those blocks. Wrong-answer choices are pulled from the same level and the same situation, so telling them apart stays real work instead of a process of elimination. | SourceHwang (2025), Language LearningInterleaved practice wins on delayed tests, but beginners and lower-performing learners do better starting blocked and switching to interleaving later. |
What it does not do yet
MochiNyan does not grade your speaking. The pronunciation questions ask you to pick the right reading, and the handwriting exercise is tracing, not composing a sentence of your own. Producing language under pressure is the one axis the research is clear about and the app is thin on — which is where CookieNyan starts.
Questions
The things people actually ask.
Anything not answered here goes to the same inbox three people read.
- Does it cost anything?
- MochiNyan Japanese is free to download. Pricing for each app is whatever its store listing says — we do not restate it here, because a number copied into a web page is a number that goes stale without anyone noticing.
- Is there an Android version?
- Yes. Both released apps are on Android as well as iOS. The install links are still being wired into this page, which is why you will see a dashed “link coming” marker where a store button belongs — that marker disappears the moment the link lands.
- What language is the app itself in?
- MochiNyan Japanese runs in English or Korean — the Japanese is the content, not the interface. Each app lists its own interface languages on its store page.
- Do I need to already know some Japanese?
- No. Courses run from Starter to Master across JLPT N5 to N1, and the app begins with hiragana and katakana stroke by stroke if that is where you are.
- Does it grade my speaking?
- It depends which one. MochiNyan does not score you — you hear a phrase and say it back, the pronunciation questions ask you to pick the correct reading, and the writing exercise is tracing rather than composing. CookieNyan is the one built around a scored speaking step, with a pass mark that rises as your level does.
- How much space does it need?
- MochiNyan Japanese is around 196 MB on iOS and needs iOS 17 or later. It is rated 4+. The store listing carries the current figure for each platform.
- When are the other cats coming?
- DalgoNyan, BaoNyan and ChurroNyan are planned and not drawn yet. We are not giving dates, because a date we miss costs more than the wait does.
- What does the app do with my data?
- Each app has its own privacy policy, linked from its section on this page and from inside the app. We keep the policy in one place rather than restating it here, because two copies of a policy always end up disagreeing.