Editing, Proofreading, Quality Assurance
Red-penciling before green-lighting
For your German content that needs polishing, I offer different levels of editing and quality assurance, depending on your text’s purpose and target audience. Typical steps may include: spelling and grammar checks, improving your draft’s readability and style, some shortening or structural optimization, a thorough comparison of source and target versions, fact-checking or a final round of proofreading before your document gets printed or your website goes live.
In some scenarios, using machines to pre-translate your content may be beneficial. But such a workflow requires careful checks and corrections to make sure the translations really serve your needs. Since this is a separate type of service, please see my MTPE page for more details on that.
Overview
- What?
- Various types of corrections
- Independent review of translations
- Validation of layout and formatting
- Checking facts and consistency
- How?
- In digital or printed document formats
- With or without “Track Changes” mode
- Always in line with client-specific style guides
- Use of LQA tools such as Xbench as needed
Editing or proofreading?
Have you ever needed someone to polish your content but felt confused by all the terms swirling around online? What’s the difference between “editing” and “proofreading,” what do “line editing” and “copy editing” mean? And what is “revising” all about? Don’t worry, I’ve got you covered!
Let’s start with revising because that’s the first step after writing your initial draft. You want to make sure you have a solid structure, your ideas get conveyed clearly, and your sequence of arguments is convincing. When revising your text, you focus on its content, not on its form. (Having said that, people may sometimes use “revising” in the more general sense of “going over a text again to improve it.” If in doubt, ask for clarification!)
Now we move on to editing and its subcategories. Essentially, proper editing transforms your text into a piece of writing your audience will enjoy. You may need help with different aspects. If you’re not sure where you’re going with your draft (especially when writing fiction), consider getting a developmental edit. If you’re at war with language or often struggle to find the right words, you need a line edit that will make your writing clearer and more engaging. If your concern is grammar, spelling, punctuation, adherence to a given style guide (if applicable), and the correctness of all facts, a copy edit should do.
Last but not least, there’s proofreading. The term goes back to the traditional printing process when a so-called galley proof had to be checked for formatting or spelling errors (introduced by the typesetter) before letting the printing press do its job. In a modern context, proofreading usually means checking the text for any objective language and formatting errors (such as typos, incorrect punctuation, inconsistent style of headings or captions, and so on). A proofreader typically does not check your tone of voice, your facts or your train of thought.
Naturally, there’s some overlap between these different tasks, so make sure you know what you need and what your chosen collaborator is offering.
Language is subjective, but we can objectively assess some aspects of it. Assume you have two texts of the same length, let’s say 1,500 words. One of them has 14 typos, the other only 2 – clearly, there’s a winner and a loser here. However, some errors may be more severe. For example, if a translation is free of grammar and spelling errors but gets numerous technical terms wrong, it’s still useless!
This is where evaluation frameworks like the Multidimensional Quality Metrics (MQM) come into play. They define error categories and weights so the quality of different texts can be assessed and compared. If you use such a QA system, I’ll be happy to fill in the respective scorecards.
Fact-checking in action
Nowadays, anybody can trigger the automatic creation and dissemination of any type of content with the click of a button – welcome to the age of AI slop! While obvious nonsense such as the Google-approved glue pizza is easy to spot, other types of false claims and misinformation require more scrutiny. This is why my fact-checking services help you identify problematic and misleading statements in your texts that could harm your reputation.
A while ago, I was part of a QA team that had been hired to review and fact-check a large collection of career guide articles for a well-known job website. The content targeted the end client’s German readers, so color me surprised when I came across a section saying (in German): “If you want to become an orthodontist, you need to complete a 5-year bachelor’s degree in dentistry.”
The thing is, in Germany you have to pass a state exam in order to get your license to practice dentistry, and a simple bachelor’s degree won’t suffice. Plus, the default length of a bachelor’s program would be 3 to 4 years, not 5 years.
Then there was this excerpt from a different article: “Expect a law degree to take 9 semesters. For admission you need a university entrance qualification and a specific minimum grade point average (numerus clausus).” Here, the devil is in the details. Law degree programs in Germany now take 10 semesters by default. Until recently it was 9, and you still find many outdated pages online – apparently, the copywriter had only done cursory research. Furthermore, the wording implied that the GPA and numerus clausus are synonyms. But they are not. A numerus clausus is a method for limiting the number of students admitted to certain study programs. The GPA is just one of various possible criteria.
This may seem like nitpicking but remember: The client wants to be seen as a trustworthy source for people seeking career advice. If their website can’t even get the basics right, how likely would you be to pay for some of their premium services?
Payment and Terms
For editing, QA, and similar services, my standard rate is €50/hour. The necessary effort for turning your texts into publication-ready content will depend on the scope of your project and the specific steps to be performed. To give you at least some pointers: When proofreading a text that was written with care, a feasible throughput is about 2,000 words per hour. If style and readability need to be optimized as well, about 1,500 words per hour are more realistic.
Payments can be made via SEPA credit transfer (for clients in Germany/Europe) or via PayPal (for non-European clients). As usual, my standard terms are 14 days, and my Terms & Conditions apply.
A short editing sample (in German)
Given my native language, I usually provide editing and proofreading services for texts in German. This sample is for people with other native languages who wouldn’t get much value out of a mere before-after comparison of an edited German text. If you do speak German yourself, switch to the German version of this page and check out the detailed annotated sample illustrating my style.
The excerpt below is from a Pegasystems e-book about IT strategies in government institutions. I was hired by an agency to review another translator’s work because the end client kept complaining about the German translations being too stiff and literal. As it turned out, the “translator” had run the whole text through DeepL and made only superficial edits. Presumably, this was an attempt to cut corners and costs – but it backfired because the poorly edited DeepL translation required so many changes that the agency had to pay me for essentially re-doing the translation from scratch!
English source text:
Welcome to the twelfth edition of Governments Building for Change, a collection of government client case studies showcasing the power of workflow automation and AI powered decisioning in overcoming challenges and unlocking new opportunities. As we navigate the complex and ever-evolving landscape of public administration, the transformation that leaders have highlighted here show the potential of technology to drive positive change and deliver efficient, citizen-centric services.
Challenges and issues: English is a fairly “lightweight” language. You can stuff a sentence with nouns, participles, and subordinate clauses without your readers instantly falling asleep. German? Not so much. Our words tend to be longer, our nouns feel heavier, and nested sentences run on and on. Therefore, when translating from English into German you often have to re-order things or break up sentences to preserve the smoothness of the source. Machines rarely do this. They might get the grammar and vocabulary right, but the output will feel clunky and boring.
This is a key problem with the original translator’s version. The underlined part below is a seemingly endless string of nouns (the capitalized words) and tedious to read. The whole paragraph is also too literal and passive – it makes you want to delete the e-book before you’ve even reached page 2!
Bad translation:
Willkommen zur zwölften Ausgabe von Governments Building for Change, einer Reihe verschiedener Case Studies von Regierungsbehörden, die die Möglichkeiten der Workflow-Automatisierung und der KI-gestützten Entscheidungsfindung bei der Bewältigung von Herausforderungen und der Erschließung neuer Chancen aufzeigen. Angesichts der komplexen und sich ständig weiterentwickelnden Verwaltungslandschaft verdeutlichen die hier vorgestellten Veränderungen das Potenzial der Technologie, einen positiven Wandel voranzutreiben und effiziente, bürgernahe Dienstleistungen zu schaffen.
In my edit, I broke up the first sentence and replaced the clunky nouns (“Bewältigung” and “Erschließung”) with active verbs (“bewältigen” and “erschließen”). I also replaced an opaque anglicism (“Case Studies” becomes “Fallstudien”) and some words that instantly identified this text as a translation (“Willkommen zur zwölften Ausgabe…” and “Verwaltungslandschaft”). The result is easier to read, doesn’t sound like a translation, and made the client happy as well.
Edited translation:
Die vorliegende zwölfte Ausgabe von „Governments Building for Change“ enthält auch diesmal wieder eine Reihe von Fallstudien zu innovativen Strategien in Behörden und Regierungsstellen. Die Beispiele zeigen, wie sich mithilfe von Workflow-Automatisierung und KI-gestützter Entscheidungsfindung ganz unterschiedliche Probleme bewältigen und neue Chancen erschließen lassen. Denn das Verwaltungswesen wird immer komplexer und entwickelt sich ständig weiter. Umso wichtiger ist es, Technologien gezielt einzusetzen, um einen positiven Wandel voranzutreiben und effiziente, bürgernahe Dienstleistungen bereitzustellen.
Frequent Questions
Will there be a sample edit (for longer documents)?
Yes, a sample edit is common for long-form content (such as novels or text books). It serves as a test run and helps both you and me decide whether we’re a good match. Depending on factors such as your desired length of the sample and my schedule, I might either do it for free or I will charge for it but later reduce my fee for the full project by the corresponding amount.
Can “AI” do fact-checking and quality assurance?
Let’s take this one step by step: “AI” is an umbrella term without a clear definition. It comprises a wide range of methods and techniques that work in very different ways. Consider something as simple as spellcheckers. Originally, those were rule-based systems without any sort of “artificial intelligence.” But modern versions often integrate machine learning, which means the line between “AI” and “not AI” gets blurred. Technically, you could say, “yes, AI can help with quality assurance tasks” (here: with spellchecking).
However, there’s also generative AI that calculates its outputs based on probabilities that are derived from massive amounts of training data. These data sets are incomplete and full of biases. A simple example: If you ask a GenAI system about the capital of France, it’ll tell you it’s Paris. This fact is present in the training data with such a high frequency that another response is unlikely to have a higher probability of being correct. This creates the illusion that the system understood your question. And that’s dangerous, because when you trust this system and then ask more specific questions that do not have such a clear answer in the training data, it will generate the most probable response without verifying whether it’s true or false.
In short: Do not let the current hype get to you! Instead of asking whether “AI” can help you, determine what kind of assistance you actually need. And then look for specific tools – or competent humans – that meet those needs.
Is a professional editor/reviewer really necessary when the budget is tight?
Well, editing and proofreading are neither rocket science nor witchcraft. You can learn these skills, and there are many helpful blogs and books teaching you grammar, punctuation, and how to write in an engaging way. Practice makes perfect, and if you’re willing to invest enough time, it’ll save you money in the long run.
Keep in mind, however, that editing your own texts can make it difficult to “kill your darlings” and to come up with potentially better alternatives. You’ll be too invested into your own material. Hence, it’s a good idea to let someone else look at your drafts. Decide on a case-by-case basis how important your text is and how much you might hurt your product or your reputation with content that’s a snoozefest or riddled with errors.