A familiar flashcard can create a false sense of security. The answer appears almost before the prompt has finished loading, so the learner marks it easy and moves on. Two days later, the same word is gone. For a B1 office worker reviewing English on the Neihu MRT, the useful moment sits between instant recognition and complete loss: the word has faded enough to require retrieval, yet remains recoverable. Vocabulary review should make room for that effort.
Easy recall can be too easy
Seeing waive six times in twelve minutes produces fluency with the card. It says little about whether the word will return during Friday's supplier call. The distinction matters because recognition supplies clues that conversation withholds. A bilingual card may show 免除, the first letter, the deck position and the colour of the template. Spoken production begins with an intended meaning and no visible answer. This is why a vocabulary card needs a way back to English. Retrieval direction and retrieval timing solve different problems. A Chinese-to-English prompt practises production. A delay makes that production effortful enough to diagnose memory. Combine them and the card begins to resemble the demand that matters: finding waive while explaining a contract.
Jeffrey Karpicke and Henry Roediger tested foreign-language word learning under schedules that changed what happened after an item had been recalled correctly. Their 2008 Science paper found that repeated study after a successful recall added little to delayed performance, while repeated retrieval produced a large benefit. Learners also predicted their later performance poorly. Smooth study felt more durable than it proved to be.
One laboratory result cannot set a universal calendar. It does establish a useful boundary: a correct answer once is weak evidence of a stable word.
Spacing has costs as well as benefits
Longer gaps make retrieval harder. They also increase the chance of failure and can lengthen a session because forgotten items need relearning. Tatsuya Nakata, Yuichi Suzuki and Xuehong He examined both sides in a registered study of 170 Japanese-speaking university students. Participants learned 20 English-Japanese word pairs under four combinations of short and long spacing during initial learning and later relearning. The 2023 Language Learning paper reported better long-term retention and more words learned per trial when longer spacing was used in both phases, despite the extra time those intervals required.
The target was a small set of paired words in controlled sessions. A learner studying phrases, pronunciation and use across sixteen weeks faces a messier task. Spacing still deserves a place, though the study does not license a promise that one app setting will work for every word. Abstract words such as tentative may need another example. A transparent word such as disagreement may survive a longer gap.
Relative timing remains unsettled too. In a study of 128 Japanese learners, Nakata compared equal intervals with intervals that expanded after each encounter. The 2015 Studies in Second Language Acquisition article found an advantage for expanding spacing on receptive vocabulary scores across several absolute schedules. Other experiments cited in later work have produced mixed comparisons between equal and expanding patterns. The sturdy conclusion concerns spaced retrieval itself. The perfect staircase of intervals is still up for argument.
App settings can hide this uncertainty behind labels such as hard and easy. A learner should inspect what those buttons actually do, because an interval that jumps from eleven days to five months after one lucky answer gives weak diagnostic information, while an interval that never grows keeps spending time on settled words. The choice depends on the deck, the deadline and the cost of forgetting the term.
Wait a little.
Use evidence. If a learner has recalled the same item accurately from a Chinese prompt, supplied it inside a fresh sentence, and pronounced it clearly after several days away, the next gap can grow with far more justification than a single tap on an easy button provides. For a certification exam next Tuesday, shorter gaps may be sensible because the use date is close. For workplace vocabulary expected to survive the next quarter, the review needs to cross weekends, changing tasks and the ordinary interference created by other English words.
Let success lengthen the interval
A practical deck can use performance as evidence. When a word returns accurately and promptly, send it farther away. When recall fails, show the answer, use the word in one meaningful sentence, then bring it back sooner. The interval follows the memory instead of a motivational streak.
Consider a seven-word set for an account manager: allocate, backlog, comply, discrepancy, reconcile, tentative, and waive. A sample schedule might revisit a successful item after two days and then six days. A failed item could return later in the same session before rejoining the shorter interval. Those numbers are an example, not a law. The useful test is whether the next appearance still requires a small search.
Keep the response strict. For reconcile, accept a spoken phrase such as “reconcile the invoice with the purchase order” only when the learner can produce it before revealing the back. Typing every answer can make a twelve-minute commute awkward. Speaking quietly into a phone works. If pronunciation matters, the card can play a model after the attempt rather than before it.
I would rather see seven words recalled across six days than 43 cards skimmed twice before Zhongxiao Fuxing.
Review inside a sentence
Word pairs isolate form and meaning, which is useful during early learning. They leave grammar and collocation thin. After the first accurate recalls, change one prompt. Replace “免除” with “The bank agreed to ___ the fee.” Later use “The clause ___ liability in one narrow case.” The second item also checks the inflected form waives. This shift adds difficulty, so avoid changing timing, direction and context all at once. A missed contextual card may reflect the sentence rather than the word. Keep one plain meaning prompt beside one sentence prompt until both are stable. For an unfamiliar technical term, a picture or short definition may carry cleaner meaning than a Chinese equivalent.
The stopping rule should be visible. End a review when each due item has received one honest retrieval attempt and each failure has been corrected. Extra laps through already fluent cards mainly improve today's feeling. Tomorrow morning, open the deck and start with yesterday's misses: discrepancy first, answer hidden.