ORT English cheat sheet

Main idea = what the whole passage is about; it must cover every paragraph, not just the first one.

Detail questions are answered by locating a paraphrase in the text, not by matching identical wording.

Inference answers must be logically forced by the text; if two options are possible, the weaker is usually wrong.

Vocabulary-in-context: substitute each option into the sentence and keep the one that preserves the surrounding logic.

Tone is signalled by evaluative adjectives and adverbs (e.g. "merely", "remarkably"), not by topic alone.

Reference words (it, this, they) resolve to the nearest preceding noun phrase that agrees in number and makes sense.

Order-of-events questions require timestamps or sequence markers (then, after, once, by the time) mapped to a timeline.

True/False/Not Given: "Not Given" means the text neither confirms nor contradicts; absence of evidence is not contradiction.

Skimming for structure: first and last sentences of paragraphs usually carry the topic sentence and conclusion.

Signpost phrases ("firstly", "moving on", "to sum up") mark the transition between sub-answers in a lecture.

Correction cues ("actually", "sorry", "rather") mean the speaker is replacing an earlier number or name; record the later value.

Numbers, dates and prices are the highest-risk items: write them as digits immediately, then verify later.

Distractors are often true statements about the wrong subject; match the answer to the exact question word (who/what/when).

Spelling of proper nouns is usually spelled out or repeated; capitalisation and double letters are checked.

In maps and diagrams, follow the speaker's directional language (opposite, adjacent, beyond, at the far end).

Note completion answers must fit the grammar of the gap: singular/plural and word class are constrained by the sentence.

Signal words for contrast (but, however, although) introduce the point examiners most often test.

Predicting answer type from the question before audio starts reduces the load during the single play.

Present perfect (have/has + past participle) links past action to present relevance; use it with for/since, not with a finished past time.

Past simple is required with finished time markers (yesterday, in 2019, last week); present perfect is wrong there.

Conditionals: zero = general truth; first = real future; second = unreal present (were, not was, in formal usage); third = unreal past.

Reported speech shifts tense back one step and changes time/place words (now → then, here → there, tomorrow → the next day).

Passive = be + past participle; only transitive verbs passivise, and the agent is dropped when unknown or unimportant.

Articles: a/an for first mention and countable singulars; the for shared or previously mentioned referents; zero article for general plurals.

Collocation beats synonymy: make a decision, do research, take responsibility, pay attention, heavy rain, strong wind.

Countable vs uncountable governs much/many, few/little, and a number: information, advice, furniture and news are uncountable.

Relative clauses: defining clauses take no commas and allow that; non-defining clauses take commas and never take that.

Task response requires every bullet of the prompt to be addressed; an unaddressed bullet caps the score regardless of language quality.

Paragraph structure: one controlling idea per paragraph, a supporting example or reason, and a linking sentence.

Cohesion is scored on linkers used correctly (however, therefore, in contrast); misused linkers damage the score more than omitting them.

Formal register excludes contractions, slang, rhetorical questions and the second person "you" in academic tasks.

Word count is a floor, not a target: under-length answers lose marks for development, not for grammar.

Sentence variety is shown by mixing simple, compound and complex sentences; a run of short simple sentences reads as immature.

Data-description tasks report trends with verbs (rose, fell, plateaued) plus adverbs (sharply, steadily, slightly), not bare numbers.

Editing checklist: subject–verb agreement, tense consistency, article before singular countable nouns, and spelling of homophones.

Fluency is judged on continuity and pace, not on accent; self-correction is acceptable if it does not break flow.

Part-style long turns reward a clear structure: direct answer, reason, example, then a closing comment.

Extending answers uses the AREA pattern (Answer, Reason, Example, Alternative) to avoid one-line responses.

Paraphrasing the question back shows lexical resource; repeating the examiner's exact words does not.

Fillers that buy time naturally ("that's an interesting question", "let me think") are preferable to long silences or L1 words.

Pronunciation is scored on intelligibility: word stress and sentence stress matter more than individual phonemes.

Speculative language ("it might be", "I suppose", "probably") is required for abstract or hypothetical prompts.

Discourse markers ("on the whole", "having said that", "to be honest") signal range and are explicitly rewarded.

Interactive tasks require responding to the examiner's follow-up, not restarting a prepared monologue.