TOEIC Link Articles a / an / the: The Three-Question Decision Tree That Decides Every Article Item on Part 5 and Part 6

Articles are the most-overlooked grammar category on TOEIC Link Reading. Candidates assume they are easy because the choices are short, so they skim the item, pick the article that "sounds right," and lose points on roughly a quarter of them. The actual error pattern in our analyzed sample shows article items account for four to six points per form in Part 5 and Part 6 combined, with the bulk of the mistakes concentrated in three predictable contexts: first-mention versus subsequent-mention, abstract noun versus concrete instance, and noun-phrase modifiers that change the article required.

This article is the three-question decision tree that resolves every article item in under ten seconds. It is organized around the three decisions ETS actually tests — definite versus indefinite, indefinite versus zero article, and singular versus plural noun head — because those are the only dimensions that matter on the test. Memorizing rules for "rivers take the" and "countries do not take the" is not the way to pass these items. The way to pass these items is to ask the three questions below in sequence and stop at the first one that resolves the blank.

Why articles deserve a separate study session

Three structural reasons keep article items at four to six points per form despite the small surface area.

Reason 1 — article errors compound across the test. A learner who confuses indefinite and definite articles in Part 5 will make the same mistake on a Part 6 cloze item, and again on a Part 7 reference question that hinges on whether a noun phrase refers back to something already mentioned. Each error is a separate point, but the underlying confusion is one. Fixing the confusion fixes all three.

Reason 2 — articles signal CEFR band sharply. A B1 candidate gets the right article roughly 60% of the time. A B2 candidate gets it right 80% of the time. A C1 candidate gets it right above 95% of the time. ETS knows this and writes article items to discriminate between adjacent bands, especially around the 800-and-up score range where the article-item density rises.

Reason 3 — articles look easy and therefore get rushed. Most candidates spend three seconds on an article item and twelve seconds on a verb-tense item. The error rates are roughly equal on a points-per-second basis, which means article items are actually a poor use of time relative to the points they pay out. Slowing down to ten seconds per article item is the single highest-yield timing change a candidate can make in Part 5.

The three-question decision tree

For every article blank in Part 5 or Part 6, ask the questions in order. Stop at the first question that resolves the blank.

Question 1 — Is the noun phrase referring to something already established?

If yes, the article is the. If no, move to Question 2.

"Established" means one of three things. The noun was mentioned earlier in the sentence or paragraph. The noun is unique by context — the sun, the CEO of our company, the user's password. Or the noun is modified by a restrictive clause that pins down a specific instance — "the report we discussed yesterday."

The most common Part 6 trap on this question is the second-mention test. A sentence introduces "a new system" and the next sentence asks for the article in front of "system." The answer is "the," because the second mention refers back to the first. Candidates who read the second sentence in isolation pick "a" or no article and lose the point.

This question alone resolves roughly 40% of article items on a typical form. For more on how Part 6 tests cohesion across sentences, see our TOEIC Link reading strategies by question type guide, which covers the broader reading framework these items sit inside.

Question 2 — Is the noun head countable singular, countable plural, or uncountable?

If countable singular, the article is a or an. If countable plural, the article is zero unless Question 1 returned "established." If uncountable, the article is zero unless Question 1 returned "established."

This is the question that catches the largest number of B1 candidates. Standard nouns that English treats as uncountable but most other languages treat as countable include: information, equipment, advice, furniture, software, hardware, research, evidence, feedback. A Part 5 item asking for the article in front of "feedback" almost always has zero article as the correct answer, with "a feedback" as the trap distractor.

The same trap reappears in Part 6 with words like "data," "training," and "machinery." If you see one of these words in a blank-article position and the sentence does not refer to a specific previously-mentioned instance, the article is zero.

The countable-versus-uncountable judgment requires word-by-word memory. For a working list of business-context uncountable nouns that TOEIC Link tests repeatedly, see the business email vocabulary cluster which marks countability on every entry.

Question 3 — Does the noun head start with a vowel sound?

If yes, the indefinite article is an. If no, the indefinite article is a.

This question only fires after Question 2 returns "countable singular" and Question 1 returns "not established." The trap on this question is that ETS does not test vowel letters — it tests vowel sounds. "An hour," "an honest mistake," "a university," "a one-time fee" are the four canonical traps. Candidates who memorize "vowel letter ⇒ an" lose all four of these items.

The traps reappear with abbreviations that read as letter names. "An MBA candidate" because M reads as "em." "A URL" because U reads as "you." "An FAQ" because F reads as "ef." If the item has an abbreviation in the noun head position, read the abbreviation out in your head before picking the article.

The three high-frequency contexts

The decision tree above resolves every item, but three contexts cause disproportionate errors and deserve a dedicated rehearsal pass.

Context 1 — First mention versus subsequent mention in Part 6

Part 6 always tests this because the format is built for it. A four-sentence paragraph introduces an entity in sentence one with "a" and refers to it again in sentence two with a blank article. The blank is "the." Candidates who skim sentence one and read sentence two in isolation lose this item every time.

The discipline is to read the full paragraph first, identify the entities, and only then return to the blanks. Five seconds of paragraph reading saves twenty seconds of guessing per item.

Context 2 — Abstract noun versus concrete instance

The abstract noun takes zero article. The concrete instance of that abstract noun takes "a" or "the" depending on Question 1. "Innovation drives growth" — abstract, zero article. "An innovation we shipped last quarter" — concrete instance, indefinite. "The innovation the marketing team led" — concrete instance with restrictive modifier, definite.

This trap appears most often with words that occupy both abstract and concrete slots: education, training, experience, service, work, design, language. ETS writes Part 5 items that hinge specifically on whether the candidate can tell the abstract reading from the concrete reading.

The signal that distinguishes them is the modifier. If the noun is bare or modified by an adjective, the reading is abstract and the article is zero. If the noun is modified by a number, a possessive, or a restrictive clause, the reading is concrete and the article is "a" or "the."

Context 3 — Proper nouns and named entities

Most proper nouns take zero article. Exceptions are tested heavily.

Rivers, oceans, seas, mountain ranges, and plural country names take the. The Nile, the Pacific, the Mediterranean, the Alps, the Netherlands, the Philippines, the United States. Singular country names, cities, individual mountains, and lakes take zero article. France, Tokyo, Mount Fuji, Lake Geneva.

Organizations follow the noun pattern: if the organization's name reads as a common noun phrase, it takes "the." The World Bank, the United Nations, the Federal Reserve. If the organization's name reads as a proper noun without a common-noun pattern, it takes zero article. Microsoft, Toyota, Anthropic.

Document titles, publication names, and product names follow stricter conventions. Newspapers take the by convention — the New York Times, the Wall Street Journal — except when the title itself starts with a proper noun pattern that suppresses the article. Most software products and consumer-product names take zero article: "Microsoft Excel," "Google Search," "Slack."

Timing target and self-check

Article items should run at eight to ten seconds each on Part 5 and at twelve to fifteen seconds each on Part 6 because the cross-sentence read costs more time. Across thirty Part 5 items, a typical form contains three to four article items, which means article items should consume roughly forty seconds of your Part 5 budget. Anything above seventy seconds suggests the three-question tree is not yet automatic.

The self-check after a practice set is to count your article-item errors against your verb-tense and preposition errors. If article errors are higher than either of those, the fix is the three-question tree above, not more general grammar review. If article errors are lower, your remaining grammar points are concentrated in the heavier categories covered in our TOEIC Link verb tenses and TOEIC Link prepositions guides.

How to drill articles

Twenty article items per session, three sessions per week, for two weeks. After each session, classify your errors by which question failed — Question 1 (established versus not), Question 2 (countable judgment), or Question 3 (vowel sound). The classification will concentrate in one or two of the three questions, and that is where the next session focuses.

After two weeks, retest yourself on a full Part 5 and Part 6 timed set. Article items should be at or above 90% accuracy. If they are, you are CEFR B2 on this category and can stop drilling. If they are below 90%, the failing question on the classification log is the bottleneck, and that question's drill set should continue for another week.

For the broader Part 5 timing and elimination framework these items sit inside, return to the per-section playbook in our TOEIC Link reading strategies by question type guide. For the underlying section structure, the TOEIC Link reading module overview is the parent reference.