JLPT vocabulary or grammar first? Follow your mistakes

Use a sentence to separate the jobs

In 友達と行きます。, recognising 友達 and 行く is vocabulary work. Understanding と as “with” in this context, and the form 行きます, involves grammar. These skills interact but are not the same error.

When you know the words but misidentify a condition or relationship, adding more word cards may not address the problem. When unfamiliar words dominate, strengthen vocabulary recognition as well.

A fifteen-minute loop

First six minutes: review some old words and add a few new ones. Try recalling reading and meaning before looking at the answer.

Next six minutes: take one pattern, check its connection and example, then compare a confusing alternative. Identify a specific difference rather than memorising the same English gloss twice.

Final three minutes: redo a missed item and note the kind of error. Return to repeated difficulties next time. This is an example, not a guaranteed preparation formula.

Compare complete contexts

Two expressions translated as “after” may require different forms before them or describe different relationships between events. Matching translations do not make every sentence interchangeable.

Listening to an example exposes pronunciation and flow. It does not replace full listening-comprehension tasks; preparing for the complete test also needs longer reading and listening practice.

Give each tool a clear job

Aptenyo Vocabulary handles level-organised cards, readings, meanings and examples. A dictionary may be better for investigating an unfamiliar word, while an integrated exam tool may span more skills.

Aptenyo Grammar is a separate app submitted for Apple review; it was not publicly downloadable at the time checked. A vocabulary purchase is not a grammar entitlement or a complete listening-and-reading course.

Treat historical frequency as reference

Before trusting an appearance count, ask what was counted, whether each source can be traced and whether the organiser published the statistics. Aptenyo’s relevant counts are internally compiled references, not official statistics.

A high past count cannot establish what will appear next. Balance study priorities with your own understanding. Counts without independently verified provenance should not be sold as predictions.