What would you evaluate?
You may know that a translation sounds wrong and still struggle to explain why it deserves a penalty. That explanation matters in linguistic quality assurance: another reviewer needs to understand your judgment, the translator needs actionable feedback, and the person using the translation needs the right problem fixed.
Two employer descriptions illustrate the difference. BURG's LQA Specialist role describes an independent quality function accountable for quality outcomes rather than production volume. It includes MQM-based evaluation and documented severity reasoning. Alboum's Spanish QA Specialist role mixes reviewing and editing translations with assessing translators, evaluating tests and maintaining linguistic resources. Neither description makes bilingual fluency alone the qualification.
Read the duties before deciding that a title fits:
- Translation production: you deliver target-language text that meets the brief.
- Machine-translation post-editing: you revise machine output to the required quality level. Evaluating that output is a separate responsibility unless the assignment includes both.
- Linguistic QA: you identify and document quality problems against agreed references, classify them and explain their consequences. Some roles also require you to edit the text.
- In-product localization testing: you check localized content in its interface or game context, reporting issues such as clipped text, incorrect language or broken presentation with enough detail to reproduce them. This can overlap with linguistic review, but it is not just a source-to-target text assessment.
- Localization project management: you own delivery coordination, resources, schedules and handoffs. Coordinating a review does not establish that you performed the linguistic assessment.
These are different responsibilities, not a fixed career ladder. Search for linguistic quality assurance, LQA specialist, translation reviewer, and quality assurance specialist plus your language. An unqualified “QA” search can lead to general software testing or other work with no translation-review component.
Check experience, country and hours first
A sample cannot resolve a hard experience requirement or a schedule you cannot sustain. Use the descriptions below as examples of what to check, not as a promise that either employer is still accepting applications.
| Check | BURG: LQA Specialist | Alboum: Spanish QA Specialist |
|---|---|---|
| Languages | English at C2/native level; active working proficiency in Spanish or another language. Spanish is preferred. | Native Spanish and excellent English. |
| Experience and domain | At least three years of hands-on LQA/localization QA experience, MT post-editing evaluation experience and a specified regulated-content background. A relevant degree or equivalent professional experience is accepted. | Four or more years in translation, editing, proofreading or QA. |
| Tools and methods | memoQ proficiency; MQM and severity-weighted scoring proficiency. | Knowledge of Trados and TermBase. Plunet knowledge is preferred. |
| Quality responsibilities | Evaluation, documented severity reasoning, partner-performance analysis, training and linguistic-asset management. | Reviewing/editing, assessing translators, glossaries and style guides, workflow issues, Spanish translation-memory alignment and evaluating new linguists' tests. |
| Advertised location and basis | Full-time remote, “anywhere in the world.” Confirm the engagement basis for your country. | Full-time remote employee position, “anywhere in the world,” subject to required availability. |
| Required availability | US Central hours, stated as “8:30 to 5:00 CST.” | “10 – 2 EST each business day.” This is an availability window within a full-time role, not a four-hour workday. |
| Compensation still to resolve | No numeric salary range disclosed; application requests a salary range. | No numeric salary range disclosed; application requests gross salary requirements in local currency and states monthly payment in local currency. |
Alboum expressly includes translation, editing and proofreading among its accepted experience areas. BURG asks for hands-on LQA/localization QA experience. Do not treat those requirements as interchangeable, or assume that the degree-equivalence option waives BURG's other requirements. Check its full description for the specific regulated-content background; a general practice exercise is not evidence of that expertise.
For a reader outside the United States, “anywhere” is a starting point for a question, not confirmed country eligibility. Ask the employer through a verified route: “Can you engage someone resident in my country, on what employee or contractor basis, and with what total weekly hours?” Then establish the required overlap, any coverage outside it, and the governing clock. Ask whether “CST” or “EST” means a fixed offset or the local US time zone as it changes seasonally. Only then convert the schedule to your local hours.
Leave missing pay information unknown. A US job-board estimate does not establish the salary, currency or benefits available to you.
Build a small sample you can defend
If your experience fits, a practice evaluation can make your reasoning visible. It is not employment history, a qualification or independent validation of your ability. It does not satisfy years-of-experience requirements, and not every employer asks for a portfolio.
Use self-created material or material you have permission to use. Choose a language pair and locale you genuinely know, with non-sensitive content you can assess competently. Include the source, target, specifications and evaluation boundaries so that someone else can follow your decisions.
The following is a fictional English-to-Spanish practice exercise for a low-stakes photo-organizing app. It illustrates annotation structure, not medical or legal review, a certification method or a hiring-test answer.
- Audience and use: adults organizing personal photos; help text explains available actions, not parser input.
- Locale: Spanish for Spain, es-ES.
- Versions: self-created source v1 and target v1; glossary v1 and style guide v1.
- Glossary: the app's “album” object must be called “álbum.”
- Style rules: direct, informal instructions; both “elige” and “selecciona” are permitted for “select”; decimal values use a comma.
- Evaluated size: all four segments below, totaling 23 English source words when counted by whitespace. No other text or app behavior was evaluated.
These records include the complete short source and target segments. In a longer sample, retain the segment identifier and quote the exact affected spans without losing the surrounding context needed to judge them.
S01: A glossary violation
- Source
- “Create a new album.”
- Target
- “Crea una nueva colección.”
- Specification
- Glossary v1 requires “álbum” for this app object.
- Proposed correction
- “Crea un nuevo álbum.”
- MQM category
- Terminology: the target conflicts with the approved terminology resource.
- Proposed severity and user impact
- Minor in this exercise. The instruction remains understandable, but the object name is inconsistent with the app's agreed terminology. There is no evidence here that the mismatch prevents the action.
- Uncertainty and next check
- Confirm that glossary v1 is current and applies to this object. If the app deliberately distinguishes collections from albums, investigate the source or glossary rather than forcing the replacement.
The reason for changing “colección” is the applicable glossary, not a claim that the word is inherently bad Spanish. If the reference is wrong or does not apply, the judgment changes.
S02: A meaning-changing error
- Source
- “You can rename the album after saving it.”
- Target
- “No puedes cambiar el nombre del álbum después de guardarlo.”
- Specification
- Source v1 says renaming after saving is possible.
- Proposed correction
- “Puedes cambiar el nombre del álbum después de guardarlo.”
- MQM category
- Accuracy, specifically mistranslation: the added negation reverses the source meaning.
- Proposed severity and user impact
- Major in this exercise. The help text falsely tells the reader that an available action is impossible, seriously undermining its intended use.
- Uncertainty and next check
- The source-to-target contradiction is clear. Actual feature behavior has not been tested. If the product no longer supports renaming, raise a source-version problem instead of silently treating the source as current.
An English back-translation, “You cannot rename the album after saving it,” can help explain the contradiction. It does not establish that an applicant can assess Spanish fluency or the rest of a Spanish translation.
S03: An acceptable alternative
- Source
- “Select an album to continue.”
- Target
- “Elige un álbum para continuar.”
- Specification
- Style guide v1 permits both “elige” and “selecciona.”
- Proposed correction
- None. Retain the target.
- MQM category
- No error. If the workflow requires a record of the reviewer's suggested stylistic change, annotate it as a neutral preference under Style, not as a penalized defect.
- Severity and user impact
- Neutral, zero penalty in this exercise. The instruction preserves the meaning and complies with the supplied style rule.
- Uncertainty and next check
- A later, applicable requirement to use one consistent UI verb could change the decision. The reviewer's personal preference alone cannot.
Leaving this sentence alone is part of the evaluation. Finding more changes is not the same as finding more errors.
S04: A locale-formatting error
- Source
- “The sample file is 2.5 MB.”
- Target
- “El archivo de muestra ocupa 2.5 MB.”
- Specification
- The es-ES style rule requires a decimal comma.
- Proposed correction
- “El archivo de muestra ocupa 2,5 MB.”
- MQM category
- Locale conventions: number formatting does not follow the specified locale rule.
- Proposed severity and user impact
- Minor in this exercise. The value remains interpretable in this low-stakes help text, but its presentation violates the specification.
- Uncertainty and next check
- This judgment assumes display text only. If the string were input to a parser or used in a consequential calculation, investigate that context before assigning severity.
For your own sample, choose content within your competence rather than copying this language pair. Keep an unresolved question visible; do not manufacture certainty to make the record look finished.
Explain severity before the score
Category asks what is wrong. Severity asks what consequence it has here. The MQM-Core typology distinguishes terminology, accuracy, linguistic conventions, style, locale conventions, audience appropriateness and design/markup. Those distinctions prevent a reviewer from treating every disliked expression as the same kind of problem.
The MQM Council's scoring documentation starts with project specifications and end-user needs. Minor errors have limited impact; major errors seriously affect intended usability, understandability or reliability; critical errors make content unfit for purpose or pose serious harm. The project selects relevant categories, severity rules, sampling and thresholds. Not every employer uses MQM, and MQM's sample scorecards are not mandatory rubrics.
For the fictional exercise above, suppose all error-type weights are 1 and the selected severity multipliers are neutral 0, minor 1, major 5 and critical 25. One major error at 5 plus two minor errors at 1 gives 7 penalty points. The neutral preference adds 0. These are illustrative counts from invented text, not a real applicant-test result or a recommended hiring threshold.
Seven raw points do not establish an “accuracy percentage” or a pass/fail result. The evaluated text size, counting method, normalization and agreed rubric matter. This exercise covers only four segments and 23 source words; it cannot support a judgment about an entire translation or someone's general performance. Do not assume that every critical flag automatically fails every scoring model.
MQM's values and scores guidance also allows neutral annotations for preferences, incorrect or missing terminology resources, and problems beyond the translator's control. Escalate a defective reference or source problem to its owner. A useful flag is not automatically a justified translator penalty.
Make the application truthful
Describe what you have actually reviewed: language direction and locale, content domain, source and target material, review scope, tools used directly, references applied and feedback delivered. Separate hands-on evaluation from coordinating reviewers, and practical tool use from knowing the terminology.
If you have edited translations but have not used a severity-weighted rubric professionally, say so. You can explain how your editing experience relates to the duties and label any MQM exercise as independent practice. Do not turn that practice into a claim of professional MQM experience.
Use one real, permitted example or clearly labeled practice example to answer five questions:
- What exactly was wrong, or why was the wording acceptable?
- Which source, glossary entry or style rule settled the decision?
- What would happen to the intended user, and why did that severity fit?
- What remained uncertain, and who could resolve it?
- If feedback or adjudication changed the annotation, what evidence changed your mind?
Only describe a revision or outcome that actually happened. You do not need an invented improvement percentage to explain a well-supported correction.
Before undertaking a substantial employer test, clarify expected time, scope, payment if any, permitted tools and AI assistance, confidentiality, intended use of the output and evaluation criteria. Follow the actual assessment instructions. Permission to make an independent practice sample is not permission to obtain assistance on a hiring test.
Independently confirm the vacancy and recruiter through the employer's official route before submitting sensitive information or installing software. BURG's observed application form requested a CV, DISC results, internet-speed results and a requested salary range; that does not remove the need to verify the route and understand what a submission requires. Requests to pay to unlock work, an unclear employer or an unspecified production-sized test warrant stopping and verifying. These are general safeguards, not allegations about either example employer.
Before accepting and starting work
Resolve the rate or salary, currency, gross basis, pay period, total hours, employee or contractor terms, tool access and applicable benefits. Alboum's description states monthly payment in local currency and requests gross salary requirements in that currency. That is one employer's wording, not an international payroll rule or a numeric offer.
For project, per-word or per-task review, also ask what volume is committed and what counts as payable work. Are preparation, calibration, corrections, meetings and disputed annotations paid? A task rate without those boundaries cannot be compared directly with a salaried specialist role.
Before your first review, agree the following with the person responsible for quality:
- The exact source and target versions, intended audience, use and locale.
- The current glossary, style guide and other applicable references.
- The selected error categories and examples of each severity level.
- The sampling scope, evaluated text size, repeated-error policy and scoring thresholds.
- The feedback format and the person authorized to adjudicate disagreements.
A useful starting practice is to calibrate a small initial sample with the lead reviewer before assessing the rest. Record disputed judgments and their resolution, correct annotations when evidence changes, and flag recurring source, glossary or workflow problems rather than repeatedly penalizing their symptoms. These are recommended working practices derived from the framework and role duties, not promises that every employer follows them.
Keep confidential client material out of public portfolios and unapproved services, including AI tools. Evaluating a sample does not certify the entire translation or give you authority over release or compliance decisions.
Find a suitable review role
Continue with Remoote's Translator search, which includes translation and review work. It is a mixed title-based search, not an LQA-only, worldwide or MQM filter. Some specialist openings use other titles, so keep the additional search terms above in your search routine.
Open a relevant role's full duties and country terms. Identify the most consequential gap: missing domain experience, uncertain country eligibility, incompatible hours or insufficient evidence of your review work. If the requirements fit, apply with a truthful example of your judgment. If one missing fact decides the fit, ask that specific question first.
The search offers Set alerts for daily emails with matching new openings. Treat them as leads, not confirmed matches: each still needs checks for qualifications, country, hours and current availability.