The Word List That Did Not Transfer
Areum has ten days before the vocabulary and grammar section. She has spent three months memorising 2,000 words with a flashcard app and scores 82% on isolated word quizzes, but only 55% on cloze items where she must choose the word that fits a sentence. Her grammar score is 60%. Her error log shows repeated problems: affect versus effect; comprise versus consist of; provide someone with something; despite of; subject–verb agreement when a long prepositional phrase sits between the subject and verb; and tense sequencing in reported speech. She studies 30 minutes a day, reviews only cards marked known, and marks a card known after one correct recall. She says she recognises the words in the cloze but the options all feel similar, so she guesses.
1. Explain the gap between 82% on isolated words and 55% on cloze. What does this imply about her flashcard routine?
Read the worked answer
Isolated recall tests form-to-meaning recognition, often with a single plausible answer. Cloze tests selection among near-synonyms inside a grammatical and collocational environment: the item requires knowing the word's syntactic frame, its typical partners, and its register. Her report that the options feel similar is diagnostic. She has stored rough meanings, not usage.
Marking a card known after one correct recall also inflates the sense of mastery, because one retrieval may be tied to the exact card format. The implication is that the flashcard routine is measuring the wrong thing. For each new word she should record one collocation, one grammatical pattern, and one near-synonym it is confused with. She should replace known with a three-day delayed recall plus a cloze sentence. This treats vocabulary as a selection problem, which is what the exam measures, and it will expose the words she only recognises rather than controls.
2. In a grammar item she sees: Despite ___ heavy rain, the event continued. The options are (A) of the (B) the (C) a (D) no word. How should she decide rather than translate?
Read the worked answer
She should test the grammatical frame, not the Korean meaning. Despite is a preposition and takes a noun phrase directly, so despite of is impossible; that eliminates option A immediately. In spite of would be different, so she must read the controller word carefully.
The remaining choices are a definiteness decision. If the preceding sentence has already introduced a specific storm or downpour, then heavy rain is identifiable and the definite article is justified. If the rain is mentioned for the first time as a general condition, no article is also grammatical, which is why this item needs context rather than a rule of thumb.
The procedure is: identify the word class of the controller, eliminate options that violate the frame, then use a secondary criterion such as definiteness, agreement, or parallel structure. Two grammatically possible answers are common in this section; the deciding evidence is usually in the surrounding sentence, not in a translation.
3. Design a 10-day error-log routine that addresses the listed weaknesses in 30 minutes a day.
Read the worked answer
Split the 30 minutes into three blocks. Ten minutes on four high-frequency grammar frames: despite versus in spite of, comprise versus consist of, provide with, and agreement across a long prepositional phrase. Ten minutes on collocation and near-synonym cloze. Ten minutes reviewing yesterday's errors aloud and writing one new sentence for each.
The error log should have four columns: the sentence, the two competing options, the rule or collocation that decides, and a new sentence she writes. She should re-test each entry after one, three, and seven days rather than on the same day. Affect versus effect is a word-class problem, so she should practise sentence transformation between verb and noun rather than memorising definitions. Agreement errors should be drilled with the interrupting phrase bracketed so the subject and verb become visually linked.
By day 10 she should be able to state the rule without looking at the example. The measure is cloze accuracy on unseen sentences, not flashcard streaks.