How to help your readers' intuition, or lack thereof, when talking about probabilities

Bayes' famous theorem is widely regarded as the most important theorem in statistics. But that doesn't mean that it is easy to understand or interpret; not by any means.

Today, trying to come to grips with it myself, I stumbled on a vital piece of information for any technical writer or into-English technical translator who wants to make any text that mentions probabilities more accessible to their readers.

Fortunately, you can use this tip whether or not you've managed to understand the theorem, so feel free to jump to the shortcut.

Optional: Want to know more about Bayes' theorem?

Here are three links:

• An Intuitive (and Short) Explanation of Bayes’ Theorem
• High-speed intro to Bayes's rule
• Bayes theorem, the geometry of changing beliefs
• The Bayesian Trap by Veritasium.

Shortcut

From An Intuitive (and Short) Explanation of Bayes’ Theorem:

People prefer natural numbers. Say “100 in 10,000″, or something similar, rather than “1%”. This helps people work through the numbers with fewer errors, especially with multiple percentages (“Of those 100, 80 will test positive” rather than “80% of the 1% will test positive”).

Who are you talking to? An expert or a member of an elite

We all know that effective translation, like effective writing, typically hinges on a good understanding of the target readership and a clear idea of what the text and its translation are intended to achieve. But that's once you have the job. What about the phase where you're still pitching your services to the client, especially if the person you are pitching two comes from higher up the management chain than the person or team who will eventually manage the work?

I recently stumbled upon this excellent piece by Rob Henderson entitled Experts and Elites Play Fundamentally Different Games.

Henderson's piece explains all sorts of situations that I observed over the years and many in which I was personally involved.

I now see more clearly why I had so much trouble and for so long attempting to convince clients that it wasn't "straight translations" they needed but client- or target reader-centred adaptations. And when I started achieving a measure of success, it was probably partly because by then I was talking to "expert" technical journalists rather than high-flying engineers who saw themselves as members or aspiring members of the elite.

Goodness; and to think that we had to wait for 2025 to find someone who could explain these things to us in relatively simple terms.

AI tools for background searches

Need to gather and organise background documents for a translation project or, indeed, anything at all?

If so, here are some tips and tricks inspired by an extended conversation with my colleague Roger Depledge in Toulouse, France.

1. Which AI tool?
   As of this writing, Perplexity appears to be one of the best of those that are readily accessible, easy to use, and free of charge.
   For everyday enquiries and fast response, and also to minimize the use of cloud resources, stay with the default Search option.
   For more thorough research, choose the Research option. Be aware, however, that this takes longer, consumes more cloud resources and may be limited to just a few searches per free-access user and per day.
   
2. How to formulate your query
   To minimize the use of cloud resources, consult:
   -  The Top 10 Ways to Use AI in 2025
   -  or google: 2025 Top ten tips using AI.
   Or, go to Perplexity and ask it something like this: What are your top ten tips for formulating queries?
   
3. Top trick #1
   When you are happy with your query and Perplexity's response, follow up with the supplementary question:
   Could you now generate a cynical reply to my query?
   
4. Top trick #2
   Now ask this question:
   Is there an elephant in the room in this context and, if so, what is it?
   

You will, I think, be quite surprised how much you will learn.

ChatGPT, a drafting aid for translation by emulation

On 17 October 2011, I published the first of two posts summarising my general approach to the type of translation/adaptation services I was then offering to my clients. The first was entitled Translation by emulation, take #1. The second, Translation by emulation, take #2, was posted on 12 November 2011.

In ChatGPT is a marvel of multilingualism, Johnson of The Economist writes "It is, in the jargon, a large language model. That means it is very good at predicting what kinds of words tend to follow which others ... (my bold)".

I conclude, though I have yet to conduct any tests, that one way to use ChatGPT before drafting a translation/adaptation by emulation, or using any similar approach, would be to ask ChatGPT a question, in the target language, on the subject addressed by the source text. It should often, I anticipate, come up with a text rich in key expressions and terminology already in use by others who have already published material on the said subject. 

Recall too that ChatGPT is perfectly capable of taking into account the education level, among other criteria, of your client's (hence your) target readership and the aim (information, gentle persuasion, etc.).

If you have access to ChatGPT and practise translation/adaptation by emulation, or any similar approach, and would like to report back, please comment.

Other relevant quotes from Johnson's piece:
"It is also able to instantly blurt out text in more than 50 languages—the precise number is apparently unknown to the system itself."
“I don’t translate from any other language; I look in my database for the best words and phrases to answer your questions.”
"This should not overshadow the achievement of a model that can effortlessly mimic so many languages, including those with limited training data. Speakers of smaller languages have worried for years about language technologies passing them by."

If by any chance you have yet to see for yourself ChatGPT's drafting skills, here's one place you could start.

A very persuasive mistranslation

This morning I participated in a discussion with other translators on how to turn a persuasive source text into an equally persuasive target document. The text under discussion was an interview given by Jean-Marc Jancovici, a French expert in energy and global warming, published in the September 2022 issue of L'Hémicycle.

A tenuous but potentially promising train of thought came to me a few hours later, triggered by an article (Translating the Bible is a vexed task, as a new book shows) that I had read recently in The Economist. The passage of interest is:

"In the Old Testament, Isaiah said an almah would give birth. Today most scholars agree that it probably just meant “young girl”. The translation of Isaiah’s Hebrew into pre-Christian Greek (in the Jewish text known as the Septuagint) rendered it as parthenos, or “virgin”. For Christians, this bolsters the idea that Jesus’s birth fulfilled a prophecy."

My train of thought will, I'm sure, sound provocative to many, but I see here a compelling example of an extraordinarily persuasive mistranslation that will, I trust, interest my colleagues who asked if I could provide any examples of persuasive writing and translation.

I recall learning from other articles and from a translation conference that included a paper by a member of the team working with Natân André Chouraqui (1917-2007) that one reason the Aramaic almah was rendered in pre-Christian Greek as parthenos (παρθένος), or virgin, was to align the translation with the target readership's strong preconceptions regarding the birth of deities and the like, and in this way make the text more persuasive. See, for instance, Translation for Transformation: André Chouraqui and His Unique Contribution to Interfaith Dialogue and Friendship.

Pause now to think for a moment about the impact of this decision on Christians, world history and goodness knows what else over the past two millennia. If ever there was a persuasive mistranslation, this is it.

See also: Young Mary vs Virgin Mary: A mistranslation of epic proportions.

French-English Glossary of Naval Technology v23 (#NavTechGloss v23)

 Update

This glossary is intended primarily for French-to-English translators specialising in naval defence technical journalism. It attempts to deal with terminological challenges specific to journalism, including access to and equivalents for the many synonyms, multiple designations and other devices that journalists use in both English and French -- though perhaps more in the latter than in the former -- to avoid repetitions.

The glossary began as a simple Word document over 25 years ago and has simply grown and grown. To accommodate its constant growth, I have adopted a custom page format (50 cm x 50 cm). The idea is not to print it but to use it on your computer. Better still, index it using an indexing search engine like dtSearch.

To download click here: NavTechGloss_Dyson_FR-EN_v23

Should the link fail due to a hacker or for any other reason, please let me know by submitting a comment.

Steve's terminology finds

Please find below a link to a personal collection of internet 'finds' that may be of interest to other FR => EN translators. Many of the entries come from: 

• tweets by Anglocom and others
• posts by translators on Pet peeves & bêtes noires des traducteurs FR<>EN and other FaceBook pages
• gleanings from news media that I consult regularly
• miscellaneous internet sources

Feedback and comments welcome.

If you have any queries, please contact me directly or leave a comment.

Link: Termino finds_FR-EN_SteveDyson_Nov2022

Should the link fail, whether due to a hacker or for any other reason, please let me know by submitting a comment.