A Japanese vocabulary app can show 駅, offer four meanings, and make you feel certain that the word is secure. Ten minutes later, someone asks where the station is and えき will not arrive. Nothing mysterious happened. The first task asked you to recognize an answer that was already present; the conversation asked you to retrieve a sound, choose a useful form and place it in a sentence without seeing the solution. Those are related abilities, but they are not the same proof of knowing a word.
A useful review system therefore needs more than one card direction and one confidence score. This guide follows a single word through four checks: recognize its meaning, connect written form with reading and sound, retrieve the Japanese from an intention, and use it in a changed sentence. KanaWay is an upcoming app, and the concrete workflow described here reflects how its vocabulary path is designed to separate those connections. You can also run the entire diagnosis with paper cards, recordings and your own example sentences; the method does not depend on the app.
Why one correct tap does not prove that you know a word
Recognition is valuable. When you meet 駅 in a notice and understand that it refers to a station, reading continues instead of stopping. The problem begins when that success is treated as evidence for every other direction. A multiple-choice answer may be reached through the first kanji, the shape of the options or simple elimination. It does not require you to supply えき, identify the same word in connected speech, or request directions with 駅はどこですか. The score is accurate for the task it measured and misleading only when we give it a broader meaning.
Think of word knowledge as a small network rather than a single box. The written form should connect to a core meaning, the reading, the sound and a few natural sentence patterns. Meaning or situation should also lead back to the Japanese form when you want to speak or write. One route can be quick while another is missing. The aim of practice is not to make every route equally difficult; it is to reveal which route fails before an easy route hides it.
A Japanese vocabulary app should test four kinds of knowledge
The first check is receptive meaning: see 駅 in a short sentence and identify the relevant idea without translating every part. The second is form and sound: read えき from 駅 and recognize えき when the written form is hidden. The third is active recall: start from a picture, intention or compact meaning cue and produce 駅 or えき before any options appear. The fourth is contextual use: choose or build a new sentence in which the word fits the speaker, grammar and situation. Each check asks for different evidence.
These four checks are not a ladder that every word must climb in one sitting. A beginner may need strong receptive access to many words and active access to a smaller set. A travel word such as 駅 deserves speaking practice sooner than a rare descriptive noun. The useful question is what you need the word to do. Mark the target connection, test it directly and let the result decide the next review. That produces a more honest study plan than counting every correct recognition as complete mastery.
Check one: recognize meaning inside a real scene
Begin with a sentence or picture that supplies enough context but does not reveal the answer. For 傘, a rainy street scene plus 忘れました may be more useful than an isolated gloss. Ask what role the target word plays and which part of the scene makes the meaning likely. Then confirm the core meaning and one limitation. 傘 names an umbrella; it does not by itself tell you whether someone is opening, buying or forgetting it. Context should clarify the word, not allow a learner to ignore its form.
To prevent guessing, change the scene after the first success. Put the same word in a shop request, a weather warning and a lost-property exchange. If the meaning survives while the surrounding vocabulary changes, recognition is becoming portable. The Japan Foundation's Marugoto teacher material recommends organizing vocabulary by topic and paying close attention to the context where language appears. Topics make encounters coherent, but varied scenes stop one photograph from becoming the only cue you know.

Check two: reach the reading and recognize the sound
A written word can carry meaning while its reading remains slow. Show 駅 without furigana and give yourself two seconds to say えき. Then hide the writing, play a clear recording inside a short sentence and identify the word from sound alone. These are two directions: form to sound and sound to meaning. Keep them separate in the result. If you read the kanji but miss the audio, more visual cards will not repair the weak listening connection.
Use one focused contrast when sound is the problem. Compare びょういん and びよういん, or ここ and こうこう, only after you understand both items and have reliable audio. Do not grade an accent against one speaker's voice. The check is whether the important mora, long vowel and word boundary remain available enough for comprehension and a clear response. After feedback, repeat once correctly, then delay the next attempt. Immediate imitation can confirm pronunciation, but delayed access shows whether the connection remains.

Run a KanaWay cycle without letting recognition lead
In the planned KanaWay vocabulary flow, begin with a word from the lesson or a word you deliberately saved from the dictionary. Read its example and hear the pronunciation once. On the next encounter, answer before revealing support: first read the form, later identify the audio, and on a separate review retrieve Japanese from a compact situation cue. The result stays attached to the connection that was tested. A missed listening answer can return as listening practice without erasing a successful reading result.
Keep the set small enough to finish the hard prompts honestly. If twenty new words produce a wall of guesses, reduce new items and preserve review of older words. Use the hint only after a real attempt, and treat the hint as feedback rather than a correct response. KanaWay is still upcoming, so there is no store link, public release date or claim that this workflow has produced user outcomes. The useful principle already works elsewhere: hide the answer, name the requested direction, attempt, check, then schedule a later attempt of that same direction.
Check three: retrieve Japanese from meaning or intention
Now remove the Japanese. Show a simple intention such as 'ask where the station is,' a photograph of a station entrance, or the beginning 駅は… with the key word still hidden. Say or type the target before viewing it. A short silence is part of the task. If an answer appears only after the first kana, record that as assisted recall rather than full recall. The distinction is useful because it tells you whether the next review should begin with a smaller cue or no cue.
Direction matters. Research indexed by PubMed on term-definition learning found that retrieval practice improved the practiced retrieval direction, while transfer to the opposite direction was limited and depended on repeated practice. A Japanese-to-English card therefore cannot be assumed to build English-to-Japanese production automatically. You do not need to duplicate every card forever. Choose active recall for words you expect to say or write, and receptive checks for words whose main job is supporting reading and listening.

Check four: use the word in a changed sentence
Producing a dictionary form is not yet flexible use. Give the word a job. With 駅, ask for a location, describe what is near it, or explain where you will meet: 駅はどこですか, 駅の前にカフェがあります, and 三時に駅で会いましょう. The surrounding grammar changes, but the word keeps a coherent meaning. With a verb, change person, time or politeness and check the needed form. With an adjective, use it to make a decision rather than merely attach です to a memorized phrase.
Grade the sentence on two levels. First, did the word fit the intended meaning and collocation? Second, was the sentence understandable with the grammar you chose? Do not reject a useful sentence because it differs from the model. Feedback should identify one repair, then ask for a new sentence that applies it. This final check matters because memorized card wording can support recall while still failing to transfer. A changed message is evidence that the word is connected to a language system, not only to its original prompt.

Read a miss as a diagnosis, not a verdict
Use four compact labels: meaning, reading or sound, recall, and use. When a word fails, mark the exact label and keep the evidence. 'I forgot 傘' is too broad. 'I recognized 傘 and read かさ, but could not retrieve it from a rainy-scene cue' gives the next session a clear task. One successful repair does not need to reset the whole word, and one failure does not erase working connections. Review can target the weak route while occasionally checking the others for drift.
Also distinguish a memory failure from a language problem. You may retrieve 乗る quickly but use the wrong particle because the sentence pattern is uncertain. That calls for grammar feedback and a complete example, not ten more meaning cards. You may know the reading of 大切 but confuse its use with 大事 in a particular context; that calls for paired examples. Precise diagnosis keeps spaced repetition from becoming a machine that simply shows the same easy card more often.
Build a seven-day review that moves from seeing to using
On day one, introduce no more than five to eight useful words through a coherent scene, pronunciation and one clear sentence. On day two, test written recognition and reading separately. On day three, begin with audio, then retrieve two high-value words from meaning or pictures. On day four, mix old and new items so position cannot reveal the answer. On day five, create or choose changed sentences. Day six is a light listening or reading encounter; save only the words that genuinely block meaning or deserve active use. On day seven, retest the failed directions without the original hints.
The weekly result is not a single percentage. Count how many words passed each connection after a delay. Perhaps eight are readable, six are audible, four can be recalled and three can be used in a fresh message. That profile is more useful than declaring that you know eight words. Next week, keep the strong receptive set moving through reading and listening while giving the smaller active set more retrieval and sentence work. A Japanese vocabulary app earns its place when it helps choose that next action rather than merely celebrating familiar taps.
Finish a word when it has enough access for its purpose, not when every fact is perfect. A word needed for conversation should receive active recall and a personal sentence. A low-frequency word in a graded reader may need only recognition. Recheck after time has passed and in a different context. If the connection holds, widen the interval. If it breaks, repair that route with feedback and one new example. Progress then means more reliable access to Japanese when the real task asks for it.

