CATTI Level 3 flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

CATTI Level 3 pass mark

Each subject is scored out of 100 and 60 is the passing score. Both 综合能力 and 实务 must reach 60; a high score in one subject cannot offset a failure in the other.

笔译实务 time budget

The paper lasts 180 minutes. A workable split is about 80 minutes per direction plus 15-20 minutes for checking numbers, terminology consistency, and untranslated source words.

意合 vs 形合

Chinese favors parataxis (clauses joined by meaning and commas); English favors hypotaxis (explicit connectors, relative clauses, non-finite forms). Translation must convert one into the other.

Omission and addition errors

In CATTI 实务, dropping source content or adding explanation not present in the source counts as an accuracy error. Amplification is allowed only to make implicit grammar explicit, not to add ideas.

Terminology consistency

A recurring technical or institutional term must receive one fixed rendering across the whole passage. Inconsistent renderings are treated as errors even when each rendering is individually acceptable.

亿 and 万 conversion

万 equals 10,000 and 亿 equals 100 million. Thus 5亿 is 500 million, and 3.2万 is 32,000. Convert before writing the English sentence to avoid order-of-magnitude errors.

Proper noun handling

Institution names, treaties, and personal names without an established Chinese form are kept in the source form or transliterated with a gloss, rather than replaced by an invented equivalent.

Register in official texts

Speeches, white papers, and news commentary require formal written style. Colloquialisms, slang, and added personal commentary lower the score because they shift the register of the source.

的-stacking limit

Chinese tolerates few pre-nominal 的 modifiers. More than two chained 的-phrases before a noun usually signals a translated English noun phrase and should be recast into a verb phrase or clause.

English passive to Chinese

English passive clauses often become Chinese topic-comment or active clauses, or take 被/由/受 only where the passive meaning is adversative or agent-focused. Blanket 被 is a common Level 3 error.

笔译综合能力 content

The objective paper tests vocabulary choice, grammar and error correction, sentence completion, and reading comprehension in both languages. It is separate from the translation passages in 实务.

Critical vs minor errors

Critical errors change meaning: wrong numbers, reversed negation, wrong subject, mistranslated key term. Minor errors are stylistic. Graders weigh critical errors far more heavily.

Source-text interference

Copying English word order, article usage, or clause structure into Chinese produces 翻译腔. The fix is to identify the proposition first, then rebuild the sentence with Chinese syntax.

Dates and units conversion

Chinese dates run year-month-day; English runs month-day-year in US style. Units such as 公里, 亩, and 摄氏度 must be converted to the target convention or glossed in parentheses.

Cohesion in Chinese output

Chinese relies on lexical repetition, pronoun omission, and zero anaphora where English uses pronouns and conjunctions. Keeping every English pronoun produces unnatural Chinese.

Proofreading checklist

Before submitting, verify: no source-language residue, numbers and dates converted, one rendering per term, no omitted sentences, and no added commentary. This catches most avoidable deductions.

口译实务 format

Candidates interpret recorded English and Chinese passages consecutively, speaking after each segment of roughly one to two minutes. Note-taking is permitted; the recording is not paused for the candidate.

口译综合能力 tasks

The objective paper tests listening comprehension, gap-filling, and summary writing in the target language, plus vocabulary and grammar items. It must also reach 60 to pass the level.

Note-taking in consecutive

Notes should capture meaning structure: subject-verb-object skeletons, arrows for direction, symbols for cause, contrast, and addition. Verbatim source wording in notes leads to literal, broken delivery.

Sense-group chunking

Source speech is segmented into meaning units of about five to nine words. Each unit is understood and stored as a proposition rather than as a string of words, which supports later reformulation.

Numbers in interpreting

Digits are written in notes and converted on the fly between 万/亿 and thousand/million/billion. The number is read back in the target language before the sentence continues.

Negation and modality

Dropping a negation or weakening must/should/may changes the speaker's commitment. These are critical errors in interpreting and are penalized more than a simplified adjective or example.

Delivery latency

Speaking should begin within about two to three seconds of the speaker's pause. Long silences, restarts, and filled pauses reduce fluency scores even when the content is accurate.

Discourse markers first

Markers such as however, therefore, in addition, and for example define the logical relation between propositions. Capturing the marker first ensures the target sentence carries the right logic.

Paraphrasing unknown terms

When a term is unknown, describe its category or function in the target language and continue. Stopping to search breaks fluency and costs more than an approximate but coherent rendering.

Recency effect in memory

Short-term memory retains the most recent sense group best. Interpreters note earlier groups while holding the latest one verbatim, then deliver in source order without reordering across long spans.

Sight translation in 口译

Some tasks require reading a short written passage aloud in the target language under time pressure. The rule is to read ahead by one clause and avoid word-by-word conversion.

Fidelity hierarchy

Meaning ranks above style: numbers, names, titles, negation, and modality must survive; decorative adjectives and repeated examples may be compressed if time or memory is short.

Self-repair penalty

Restarting a sentence or correcting a number mid-delivery is penalized. A brief, clean correction is acceptable; repeated restarts signal loss of control and reduce the fluency score.

Active listening stance

The interpreter listens for proposition and intent, not surface wording, predicting the speaker's direction from context. This reduces note volume and improves target-language reformulation.

Interpreting register matching

Formal speeches require formal target phrasing; the interpreter should not add colloquial hedges or first-person commentary. Register mismatch is scored under expression quality.

Consecutive vs simultaneous

Level 3 实务 tests consecutive interpreting, delivered after the speaker pauses. Simultaneous interpreting belongs to higher levels and to separate certification streams.