Isn’t it nice when you get those interesting short-form videos sent to your mobile phone? I love clicking on content that interests me, but time and again, the subtitles leave me puzzled.

How can it be that the machine confuses the words “convey” and “combine”? They don’t even sound alike!
Okay, to be fair, the speaker had an accent. But although English is not my mother tongue – and I certainly don’t get to hear English spoken with accents every day – even I can hear the difference between those words.

Call or write to me if you are concerned about getting the subtitles right on your videos. It’s a shame when errors distract from content. Looking forward to hearing from you. 

One of the joys of working with fictional texts is that they often contain delightful dialogue, and making those turns in a conversation sound good is arguably one of the most critical aspects in (fictional) writing.  

When it comes to transferring your creations to another language using a machine translation engine, it pays to check for critical mistakes. My “blunder radar” has recently detected some unusual practices related to how grammatical constructions are rendered in German. The way some English TV and film dialogue is translated impacts what I will call the “continuity” of the conversation – essentially, the verisimilitude of the speakers’ interaction.

You may have come across it if you have been watching foreign-language shows and films in your language, when you’ve probably registered it subconsciously: a conversation where the speakers don’t seem to be conversing as the language would require.

You might hear something like this interaction:

A: Wie geht es Ihrer Tochter?
B: Auf ewig in Ihrer Schuld.

A is asking “How is your daughter?” and B replies “Forever indebted to you”. Perfectly grammatical. And a clever play on the words. But in the English version only. The German version misses not just the (grammatical) point but grammar as such. 

If the English version flows naturally, it can be disappointing to see the effect lost in translation:

Das geht Sie doch einen verdammten Dreck an! — Leider doch, Monsieur, wenn die Angelegenheit geklärt werden soll.
Und wo wollen Sie jetzt hinfahren? — Etwas Lippenstift und einen dicken Schutzhelm und dann ab aufs Schlachtfeld.

While examples like these won’t hinder comprehension, there are some that catch one’s eye (or ear, for that matter) a little more forcefully. Therefore, if you have doubts about the quality of your machine translation, send me a message and we’ll discuss how to save your masterpiece. It doesn’t cost the earth!

Ever wondered what the fuss is all about? Machines save us time and money, don’t they? Or do they?

Machine translation has come a long way. It's like watching a toddler grow into a teenager—impressive progress, but still prone to spectacular, sometimes hilarious failures. These AI systems can now translate the literal meaning of words with remarkable accuracy, giving the dangerously misleading impression that human translators might soon be obsolete.

However, here's the uncomfortable truth: what appears to be a perfectly translated document is often just a hastily concocted solution that can harm your business.
​
Take these care instructions. Ignoring the lack of punctuation for a minute, let’s look at the semantics of the underlined phrase in this real-life example. The literal meaning in English of this phrase would be “…clean with a moist bread” (as in the bread that you can eat!).

Not the result you want.
 
Why This Matters for Your Business

Poor translations can be worse than no translation at all. They can:

• Damage your company's credibility
• Create misunderstandings with potential clients
• Lead to costly mistakes in contracts or technical documents
• Make your marketing materials ineffective or even offensive

 
Machine translation is a useful starting point, but it's just that—a starting point. The real magic happens in the editing process, where human expertise transforms mechanical text into meaningful communication.

So, the next time you're tempted to skip the human review stage, remember: machines don't understand jokes, cultural references, or context. They've never put all other things on hold to find the right word in a given register or dialect. They've never felt the satisfaction of crafting the perfect sentence.

But I have. And I'm here to ensure your translations don't just translate words but translate success.
​
Ready to ensure your translations actually make sense? Contact me today for a free consultation!

Once upon a time, there was an ambitious author named Klaus who wanted to publish his masterpiece "The Infinite Story of the Finite Mind" as an e-book. Unfortunately, Klaus knew as much about formatting as a penguin knows about desert racing. What follows is a collection of his most spectacular mistakes – everything that should never happen when formatting an e-book!

1. The Font Apocalypse
Klaus loved fonts. ALL fonts. On a single page, you could find Comic Sans, Papyrus, and of course his favorite: Wingdings for particularly dramatic passages. Readers later reported that their e-readers had begun to weep in despair.

2. The Image Drama
Klaus firmly believed that more is always better. That's why he added at least three high-resolution photos to each page – each at least 20 MB in size. His 120-page book quickly reached the size of a medium-sized planet. When opening the e-book, devices heated up so much that you could easily fry eggs on them.

3. The Table Tragedy
To structure his complex thoughts, Klaus created nested tables with 27 columns. On a smartphone, the result looked like a three-year-old's failed attempt to recreate a QR code with crayons.

4. The Spacing Nightmare
Spacing between paragraphs? Klaus knew a trick: just hit the enter key until it looks good! On his computer, that was exactly 17 blank lines. On e-readers, this resulted in sometimes three pages of emptiness gaping between two sentences – perfect for readers who wanted to meditate in between.

5. The Formatting Fantasy
Klaus had heard that text can be made bold and italic. So he made EVERY OTHER WORD bold AND italic AND underlined AND in capital letters. Eyewitnesses reported spontaneous migraine attacks while reading.

6. The Table of Contents Inferno
Tables of contents are important, Klaus thought. So he created one – but completely manually, without hyperlinks. His helpful instruction at the beginning of the book: "To get to Chapter 7, please turn the page 342 times."

7. The Metadata Mess
When Klaus uploaded his work, he left the metadata fields empty or filled them with creative entries. Under "Author" he put "The one you all know," and for the book title he chose "⭐️? BEST BOOK OF ALL TIME - BUY NOW!!! ?￢ﾭﾐ️". In bookstores, his work finally appeared under "Miscellaneous Cookbooks."

Klaus's e-book eventually became so infamous that it was included in the National Museum's collection of digital curiosities – as a warning for future generations. So if you one day format your own e-book, think of Klaus and do exactly the opposite!

Who was your musical idol in your youth?
Did he or she sing in your native tongue, in the language you grew up speaking?

In my January post, I mentioned a musical genius from Austria who charmed the masses 200 years ago. A singer-songwriter who shared similar success in Austria but much more recently is Rainhard Fendrich, who turned 70 this week. An idol of my youth, he charmed us with his apt, often very funny, and always thought-provoking lyrics and catchy melodies. His hits include: Strada Del Sole, Tango Korrupti, Macho Macho, Midlife Crisis, Blond, Weus’d a Herz hast wia a Bergwerk, Haben Sie Wien schon bei Nacht gesehen?, Es lebe der Sport…

Fendrich sings in his Viennese-Austrian dialect, which is relatively easy to understand for learners of German, and the songs resonate so much with the audience that you’ll find that most Austrians are triggered by phrases such as ‘midlife crisis’ or ‘I am from Austria’ (practically our unofficial anthem) to hum the tunes. If you’d like to discover Fendrich’s oeuvre and aren’t sure whether you understand the songs, feel free to contact me for help.

Rainhard Fendrich forever!
HAPPY BIRTHDAY! ALLES GUTE ZUM GEBURTSTAG! 

Jubilation may not be on your mind right now, but there is a good reason to start the year on a musical note. ​How about waltzing into the new year? 

This year marks 200 years since the birth of an Austrian musical genius who delighted the crowds with his danceable orchestral compositions at a time when censorship, revolutionary unrest, and armed conflicts shaped people’s lives. 

​
Johann Strauss II (1825–1899), often referred to as the “Waltz King”, made his debut in 1846, at the age of 21, with his own compositions and orchestra in the dance Café Dommayer in Vienna’s Hietzing district – and the craze, a “dancing mania of near epidemic proportions”[1], hit the city and the countryside.​
​

And there is one piece that resonates with Austrians like no other, you could say, and which you may have heard on an Austrian Airlines flight: the Blue Danube Waltz, which you can listen to by clicking on the button below. It uplifts and stirs the soul, which is no bad thing at the start of something new.  Happy New Year!

Have you ever started a sentence only to veer off in an unexpected direction? You might have inadvertently employed anacoluthon, a rhetorical device where a sentence abruptly changes its grammatical construction mid-stream.

This linguistic phenomenon, derived from Greek meaning "lacking sequence," occurs when a speaker or writer begins with one syntactical structure but switches to another before completing the first. While it may seem like a mistake, skilled orators and authors often use anacoluthon intentionally for emphasis, to mimic natural speech patterns, or to create a jarring effect.

Often, the jarring effect comes about because we expect to be led somewhere else, such as in this German example from a Duden dictionary[1]:

Ÿ   "Sie ist sicherlich schon gegangen, weil ihr Mantel hängt nicht mehr an der Garderobe."

Here, the expectation is for the subordinate-clause word order:

Ÿ   Sie ist sicherlich schon gegangen, weil ihr Mantel nicht mehr an der Garderobe hängt.

According to Claude, an AI assistant created by Anthropic and prompted by Alexandra, Shakespeare was fond of this device. In "Macbeth," Shakespeare writes:

Ÿ   "The heavens, as troubled with man's act, / Threatens his bloody stage."

Notice how "heavens" (plural) doesn't match with "threatens" (singular)?

In everyday speech, anacoluthon is common in moments of excitement or distraction: "Oh, the party tonight—did you bring the snacks?" Here, the speaker abandons the initial thought about the party to ask an urgent question.

Notice how the punctuation helps us decipher what is going on here in processing terms? Audio transcribers in their work aim to translate this linguistic reorientation into written language to help readers during text reception.

Whether deliberate or accidental, anacoluthon adds a touch of spontaneity and realism to language, reminding us that even in grammar, sometimes the journey is more interesting than the destination.
  
What do you think?

[1] © Duden ‒ Das Wörterbuch der sprachlichen Zweifelsfälle, 9. Auflage, Berlin 2021

Retrieval augmented generation is an approach that aims to improve the performance of large language models (LLMs) and other AI models by incorporating external knowledge sources during the generation process. The idea behind this technique is to augment the model's knowledge base with relevant information from external sources, such as databases, documents, or the internet, to enhance the quality and accuracy of the generated output.
 
In traditional language models, the generation process is solely based on the model's internal knowledge, which is learned from the training data. However, this knowledge can be limited and may not cover all the information required for a specific task or domain. Retrieval augmented generation addresses this limitation by retrieving and incorporating relevant external knowledge during the generation process.
 
The retrieval augmented generation process typically involves the following steps:
 
1. Query Formulation: The model formulates a query based on the input or the context of the generation task.
 
2. Information Retrieval: The query is used to retrieve relevant information from external knowledge sources, such as a document database, the internet, or a knowledge base.
 
3. Knowledge Integration: The retrieved information is integrated with the model's internal knowledge, either by concatenating it with the input or by using attention mechanisms to attend to the relevant information during the generation process.
 
4. Generation: The model generates the output based on both its internal knowledge and the retrieved external knowledge.
 
This approach has several advantages over traditional language models. By incorporating external knowledge, retrieval augmented generation can produce more accurate, informative, and up-to-date outputs, especially in domains where the training data may be limited or outdated. It can also improve the model's ability to handle out-of-distribution or open-ended tasks, as it can leverage external knowledge sources to fill in knowledge gaps.
 
However, retrieval augmented generation also comes with challenges, such as designing effective retrieval mechanisms, integrating diverse knowledge sources, and ensuring the coherence and consistency of the generated output. Additionally, the quality of the generated output depends heavily on the relevance and reliability of the retrieved information, which may require careful curation and filtering of the external knowledge sources.
 
Overall, retrieval augmented generation is an important area of research in the field of AI and natural language processing, as it has the potential to significantly enhance the capabilities of language models and other AI systems by leveraging external knowledge sources.


Written by Claude, an AI assistant created by Anthropic, prompted by Alexandra.

In our digital age, machine translation has become a ubiquitous tool, allowing us to bridge linguistic barriers with the click of a button. However, as impressive as these technologies are, they are not infallible. Just like a rough diamond requires skilled craftsmanship to unleash its full brilliance, machine-translated text often needs the discerning eye and deft hand of a human linguist to truly shine.
 
Enter the art of machine translation post-editing (MTPE) and translation review/editing. These processes involve a linguistic expert meticulously refining and polishing a machine-generated translation, ensuring that it reads naturally, conveys the intended meaning accurately, and adheres to the target language's nuances and conventions.
 
Imagine a masterpiece painted by a skilled artist, but with a few smudges and imperfections marring its beauty. That's where the post-editor steps in, acting as a linguistic conservator, carefully retouching and restoring the work to its intended grandeur. They scrutinize every word, every phrase, every nuance, ensuring that the final product is a flawless linguistic masterpiece.
 
But MTPE and translation review/editing are more than just correcting errors; they are about capturing the essence of the source text and rendering it in a way that resonates with the target audience. It's a delicate balance of preserving the original meaning while adapting it to the cultural and linguistic norms of the new language.
 
Just as a skilled jeweler can transform a rough gemstone into a dazzling work of art, a professional linguist can take a machine-generated translation and meticulously refine it into a polished, culturally-attuned masterpiece. It's a process that requires not only linguistic expertise but also a deep understanding of the source and target cultures, ensuring that the final product is not just technically accurate but also culturally relevant and impactful.
 
So, whether you're a business seeking to expand your global reach, an author hoping to share your work with a wider audience, or simply someone who values the power of clear communication, consider enlisting the services of a skilled linguistic services provider. With their expertise in MTPE and translation review/editing, they can transform your machine-generated translations into linguistic gems, ensuring that your message shines brightly across languages and cultures.
 
 
"Written by Claude, an AI assistant created by Anthropic"

‘You are the woman with the single high cheekbone.’ Having said this, he wrapped his arms around Maddy, his forty-year-old wife of forty years, who was a bit of a bother. It made him feel protected rather than protective. Still smarting, he said to her, ‘That was clever.’

Maddy, who was no shrinking bush, and standing ten centimeters (actually, six inches) bigger than her husband, no slip-up of a girl either, defended herself with the words, ‘Well, I didn’t want to trip up the passing traffic. I’d slept on a spoon all night, you see.’

John, her husband, who wasn’t six people (or six-packed) and let his shoulders hang like a boxer when he wanted to square them, then stood back when he wanted to stand up.

Maddy smiled and sat on her face, having shut her mouth with her hand. This was when Maddy realised that she should not have taken such a big sip of wood. She had eaten a cupped icecake [sic] earlier and now felt like throwing up. So she threw herself in a bush …

Okay, let's stop here with the story I started in my last post. I think you get my point about machine translation. Now let's hear from you! Did you have similar experiences? What is your take on these (new) AI developments?

AI is careless, short-sighted, sexist and prudish. And it gives bad (dietary) advice!
 
It’s Christmastime again, the season of fun and telling (tall) tales. This year, let me tell you a story:

Once upon a time...
...there was this guy who ate restaurants and wore eye glasses with tree branches and had the body of a dog. He was an expiring writer who also translated from nudity to his mother tongue whenever he found the time. A shy person by nature, he liked to keep his arms in his hands when spoken to or rattle his hands, especially if his interlocutor was a beautiful young woman; he would throw his eyes – which were situated on his elbows – at her and feel the urge to start eating. He instinctively knew that she would not agree to drink tea and toast because women tended not to appreciate that he was stupid and nosy, rather than stupidly keen to tease, which would have been fun, perhaps. They also felt that his reserves needed sorting (as much as his reserve).

On most days, therefore, when he was busy, our young man liked to sit in a chair in the garden and watch the sky wobble, or he would ride around the fields in his pram (in lieu of a pushbike). One day at about thirty (3:30 p.m.), as he was sitting in his garden, a vehicle stepped out from behind a tree. Dust-Moths (and dust motes, too) flew, distress rolled in waves towards the man, and he ran his tongue over his lip, which lay across the table from him. But then he took courage and the man whose surface he was … uhm … surfaced and he did not tense, he looked forward to something terrible happening.

A flight attendant (who was female, of course) wearing a wired bra emerged from the vehicle, hit him with a wall and squirted fresh-from-the-box freshly pressed wine at him. How rude! How could she do this to an ambulance driver (miraculously converted after his ambulance-chasing days), a guy who gave French kisses and wasn’t tongue-tied at all? And he hadn’t even had a second’s warning, only a second warning!

All kinds of thoughts crossed his mind (not his face). Never one to cause ‘quiet’ havoc when ‘quite a lot of’ havoc would do, he picked up a contaminated bag he was supposed to have replaced and squeezed it hard when he meant to give it the tiniest squeeze. Then he said firmly, ‘Nicht. Ever. Do. That. Wieder.’

But the mind does not only wonder, it also makes many wonders, and so the young man asked her, ‘Want a hand? Do you want to help me? It’s eighteen fifteen (8:15 a.m.) and I really need to make a start on chapter “Zweizig sieben” (27) of my German book.’

Leaning on a piece of bread and chewing a shovel the woman replied, ‘You don’t know me.’

‘Yes. I know.’ (= Yes, I do.) 
------------------------------------------------------------------------------------------------------------------------------------------------------------ 
To be continued…
  
Watch this space for more weirdly wonderful LLM tales. And if you think this story is unbearably ‘creative’, slightly over the top, or just plain weird … well, hard luck. Don’t complain to me. A translation engine (can I call it an ‘AI’?) produced the German words and phrases I used (in their English translations) for this story. What a wonderful world …
 

Lives of great men all remind us
We can make our lives sublime,
And, departing, leave behind us
Footprints on the sands of time. 
​
H. W. Longfellow

Have you ever wondered why some computer-generated translations sound a bit strange to native ears? Well, there are several explanations, but my focus here will be on English words used by German speakers.

“Duplicate Content” is such a word, the ominous “shitstorm” is another. Some English words make it into dictionaries and glossaries, while others hover on the verge of widespread acceptance – or being translated, as it may be. And if a clever linguist comes up with a word to replace the “foreign” English word, this new German word – a neologism or perhaps a coinage – may find its way into the permanent vocabulary.

So if you want to grab the attention of your German readers, go for it: use the English term if it is widely used in different settings and social groups. Comprehension is a key indicator of the success of your localisation.

But how do you know what words to use? – The answer is quite simple, really. Ask an expert. 

​Hot and cold, heißgeliebt or heiß geliebt? 

​Opening a box of Easter egg dyes recently, a smile came across my face when I read the instructions, and I simply had to share some more interesting German language facts with you. 

The phrase ‘kalt gefärbt und heiß geliebt’, ​translated by the manufacturer as ‘cold dyed and much loved’, illustrates a semantic peculiarity of German given some extra attention in the last spelling reform. You may have heard of a group of adjectives consisting of two words or one, depending on how you want to write them or the literal or figurative meaning. The ‘compound’ spelling – two adjectives combined into one word – sometimes conveys a sense that is ‘more than its parts’.

Both spellings are usually correct, as the bible of German, the Duden will tell you, but some controversy over the loss of semantic differentiation (and because Germans simply love compounds!) meant that since 2006 the need to split adjectives is no longer the official spelling rule.

The phrase in the packaging of the Easter egg dyes, on the other hand, illustrates a wordplay based on the opposites of cold and hot: the eggs can literally be coloured using a ‘cold’ dye, and choosing the two-word spelling for 'heißgeliebt' here was really a matter of balance and parallel structures. Nice, huh?

Do you have similar instances of semantic or orthographic peculiarities of German that you would like to discuss with me? Send me a message or an email, and let’s talk.
 
In the meantime, I wish all friends and clients of Witinall Language Services a very

                                HAPPY EASTER!
Yours,
Alexandra

As the temperatures rise, new opportunities and project ideas present themselves. And your German customers are waiting to hear from you. So don't make them wait. Get your translations, transcriptions or subtitles out there now! 
Get your free quote from Alexandra today!