EPSO Lawyer-Linguist common mistakes
EPSO Lawyer-Linguist · Understand the traps and how to avoid them.
Translating the source legal text sentence-by-sentence in the order it appears, so that the target text keeps the source's clause order even where the target language would normally restructure the sentence (e.g. fronting a long subordinate clause before the main verb).
Under time pressure candidates treat the source word order as a safety net: moving clause boundaries feels like a risk of losing content, so they mirror the syntax and only polish vocabulary. Legal texts have unusually long sentences, which makes the pull toward calque strongest exactly where restructuring matters most.
What to do instead
Before drafting, read the whole paragraph and mark the main assertion of each sentence. Draft the target sentence around that assertion first, then attach conditions and provisos in the position the target language expects. Practise by taking five long source sentences per session and rewriting each with a different clause order, then checking that no proposition was dropped or altered.
Rendering EU institutional terms with a plausible dictionary equivalent instead of the established term of art used in the target language's version of EU instruments (for example inventing a translation for 'acquis communautaire', 'subsidiarity' or 'preliminary ruling' rather than using the settled formula).
Candidates rely on general bilingual dictionaries and their own legal vocabulary, and they do not systematically check how the same concept is rendered in the target-language version of the Treaties or in EUR-Lex texts. The invented term reads fluently, so nothing signals the error during self-revision.
What to do instead
Build a personal glossary of institutional terms with the exact target-language wording taken from official EU texts, and revise it before the exam. During practice, mark every term that names an EU institution, procedure, instrument or principle and verify it against a target-language EU source rather than a dictionary. Treat any term you cannot verify as a flag to resolve, not a gap to fill by intuition.
Leaving source-language terminology in the target text because the concept has no obvious equivalent, producing a hybrid text with untranslated terms, unexplained borrowings, or a descriptive paraphrase that quietly narrows the legal scope.
Legal systems do not map one-to-one, so a candidate facing a target-language gap either panics and borrows the source term or over-explains and loses precision. The gap is real, but the handling is improvised rather than chosen deliberately.
What to do instead
Decide consciously between the available strategies: an established target-language equivalent, a functional equivalent with a brief parenthetical gloss where the genre permits it, or a neutral superordinate term that does not overstate. Learn the standard equivalents for recurring system-specific concepts in advance, and practise writing the same gap three ways so the choice becomes a decision rather than a reflex.
Confusing near-neighbour EU law concepts in translation, such as directive versus regulation, opinion versus decision, or the different classes of EU acts and procedures, and therefore choosing target wording that misstates the instrument's legal effect.
Candidates prepare terminology as a vocabulary list without the underlying legal distinctions, so two terms that look interchangeable in the dictionary get used interchangeably. The mistranslation is invisible to a reader who does not know the source, which is why it survives self-checking.
What to do instead
For each high-frequency instrument and procedure, write a one-line definition of its legal effect and who adopts it, and attach the target term to that definition rather than to the source word. Test yourself by translating short authentic passages containing several instrument names and justifying each choice aloud; if you cannot state the effect, you do not yet know the term.
Producing a target text that is grammatically correct but stylistically wrong for the genre: contractions, colloquial connectors, inconsistent register, or a chatty tone in what should be a formal legal instrument, or conversely an archaic register in a document meant to be plain and accessible.
Candidates focus their preparation on vocabulary and accuracy and treat register as decoration. When drafting under time pressure, the register of their everyday language leaks in, and there is no explicit genre model in their head to check against.
What to do instead
Collect short target-language models of the genres that appear in the exam (legislative text, judgment, contract clause, official letter) and note their recurring features: sentence length, connector set, use of passive, person, modality. Before writing, name the genre and its three defining features, then check the finished text against those features in a dedicated register pass.
Submitting a translation that has not been proofread for target-language mechanics, so agreement, case or gender, punctuation conventions, capitalisation of institution names and hyphenation errors remain in the final text.
Time is allocated almost entirely to comprehension and drafting, and revision is whatever is left. Candidates also reread their own text expecting what they meant to write, so surface errors are invisible on a single pass.
What to do instead
Reserve a fixed final block of time, roughly one tenth of the total, for a mechanics-only pass and do not touch content during it. Read the target text without the source in view, sentence by sentence, and check one error class per pass if time allows. In practice, keep an error log of your recurring mechanical slips and review it immediately before the exam.
Spending disproportionate time on one difficult passage or term and then rushing the remaining items, leaving later questions unanswered or answered without reading the full stem.
The computer-based tests are strictly timed and cannot be paused, and a hard item early creates a sunk-cost trap: candidates keep reworking it to protect their accuracy rather than protecting their coverage.
What to do instead
Adopt a hard rule to guess, mark and move on after a set interval per item, then return only if time remains. Rehearse with full-length timed sets so the pacing rule is automatic, and track your split between time spent on answered versus abandoned items to confirm the rule is holding under pressure.
Reading a numerical reasoning item's question stem after starting to interpret the data, then computing a quantity the question did not ask for, or answering with the wrong unit or rounding.
Tables and charts invite immediate calculation, and the arithmetic feels productive, so the stem gets skimmed. Distractor options are built from the intermediate values this produces, so the wrong answer looks confirmed.
What to do instead
Force a fixed sequence: read the question, write down the exact quantity and unit required, then locate only the data needed. Before selecting, restate the question in your own words and check the computed value against the required unit and rounding. Practise by predicting the answer's rough magnitude before calculating, so a distractor result looks wrong on sight.
Treating verbal reasoning as a general knowledge or plausibility test, choosing the option that sounds true about the subject rather than the one strictly supported by the passage, and importing outside assumptions.
The passages are often about real topics, so prior knowledge fills gaps and makes an unsupported option feel correct. Candidates also accept options that are true but do not answer the question asked, because the truth of the statement dominates their judgement.
What to do instead
Answer only from the passage: for each option, locate the exact sentence that supports or contradicts it, and reject any option you cannot anchor to a specific line. Distinguish 'true in the passage' from 'answers the question asked', and practise on topics you know well, where the temptation to import knowledge is strongest.
Drafting the target text without first identifying the source text's structure, defined terms and internal cross-references, so that a defined term is rendered inconsistently later in the text or a cross-reference points to the wrong provision in translation.
Candidates translate sequentially and make each term choice locally, which is defensible sentence by sentence but inconsistent across a long document. Cross-references are especially fragile because they look like ordinary numbers and are not re-checked.
What to do instead
Do a short pre-translation pass: list defined terms with their fixed target renderings, note every cross-reference and where it points, and map the document's section structure. Enforce consistency with a term list used throughout, and in the revision pass verify each cross-reference against the translated structure, not against the source layout.