Chess German

[Roger de Coverly]

German chess articles are not too difficult to read if you can fall back on O level knowledge and a few specialist terms. Google translate can throw up some funnies.

Reading https://de.chessbase.com/post/dietmar-kolbus-1966-2024 there's a paragraph about a King and Pawn ending which runs

Original

Und zwar rettet Schwarz der Zug 1. ... Kd2. Wenn jetzt 2. a4, dann 2.... f5, 3. Kd4 und jetzt nicht 3....f4 und der Bauer fällt, sondern 3.... Ke2 und nach 4. Ke5 (4. a5 wird mit f4 usw. beantwortet mit Zugumstellung oder gleichzeitiger Bauernumwandlung) 4....Ke3 und nach 5. Kxf5 spielt Schwarz Kd4 und ist plötzlich im Quadrat, um den weißen Bauern abzuholen. Versucht Weiß gleich 2. Kd4 Ke7 3. Ke5, um das zu verhindern, spielt Schwarz 3.... f5+ und nach 4. Kxf5 Kd6 5. a4 Kc4 und ist wieder im Quadrat - remis!

Google's attempt
And this is where Black of the train 1. ... Kd2. If now 2. a4, then 2.... f5, 3. Kd4 and now not 3....f4 and the farmer falls, but 3.... Ke2 and after 4. Ke5 (4th a5 is answered with f4 etc. with migration change or simultaneous farmer conversion) 4....Ke3 and after 5. Kxf5 plays Black Kd4 and is suddenly squared to pick up the white farmer. Try White equals 2. Kd4 Ke7 3. Ke5 to prevent this, plays Black 3.... f5+ and after 4. Kxf5 Kd6 5. a4 Kc4 and is squared again - draw!

That Zug has a double meaning of train and move, I was familiar with. I wasn't aware of Bauer = farmer though as well as Bauer = pawn. Remis = draw isn't always known to translation engines, but Google seems to have it right.

Characteristic German compound words. Bauernumwandlung = Pawn conversion/promotion
The idioms can be different. Freibauer means passed pawn rather than the literal free pawn, as in loss of material without compensation.

[Ian Thompson]

Original

Und zwar rettet Schwarz der Zug 1. ... Kd2. Wenn jetzt 2. a4, dann 2.... f5, 3. Kd4 und jetzt nicht 3....f4 und der Bauer fällt, sondern 3.... Ke2 und nach 4. Ke5 (4. a5 wird mit f4 usw. beantwortet mit Zugumstellung oder gleichzeitiger Bauernumwandlung) 4....Ke3 und nach 5. Kxf5 spielt Schwarz Kd4 und ist plötzlich im Quadrat, um den weißen Bauern abzuholen. Versucht Weiß gleich 2. Kd4 Ke7 3. Ke5, um das zu verhindern, spielt Schwarz 3.... f5+ und nach 4. Kxf5 Kd6 5. a4 Kc4 und ist wieder im Quadrat - remis!

Google's attempt
And this is where Black of the train 1. ... Kd2. If now 2. a4, then 2.... f5, 3. Kd4 and now not 3....f4 and the farmer falls, but 3.... Ke2 and after 4. Ke5 (4th a5 is answered with f4 etc. with migration change or simultaneous farmer conversion) 4....Ke3 and after 5. Kxf5 plays Black Kd4 and is suddenly squared to pick up the white farmer. Try White equals 2. Kd4 Ke7 3. Ke5 to prevent this, plays Black 3.... f5+ and after 4. Kxf5 Kd6 5. a4 Kc4 and is squared again - draw!

How odd. When I put that paragraph in Google translate I got

Black is saved by the move 1. ... Kd2. If now 2. a4, then 2.... f5, 3. Kd4 and now not 3....f4 and the pawn falls, but 3.... Ke2 and after 4. Ke5 (4. a5 is answered with f4 etc. with a move rearrangement or simultaneous pawn promotion) 4....Ke3 and after 5. Kxf5 Black plays Kd4 and is suddenly in the square to pick up the white pawn. If White immediately tries 2. Kd4 Ke7 3. Ke5, to prevent this, Black plays 3.... f5+ and after 4. Kxf5 Kd6 5. a4 Kc4 and is back in the square - draw!

which is a pretty good translation. Did you perhaps only put the paragraph into Google translate, whereas I put the whole page in? Maybe the translation engine realised from the whole page that the subject matter was chess, but didn't deduce that from just the paragraph.

I wasn't aware of Bauer = farmer though as well as Bauer = pawn.

I've got/had a German postcard somewhere which makes a joke based on the double meaning of Bauer in German.

[Roger de Coverly]

Did you perhaps only put the paragraph into Google translate, whereas I put the whole page in? Maybe the translation engine realised from the whole page that the subject matter was chess, but didn't deduce that from just the paragraph.

Seems a reasonable hypothesis!

[Bob Kane]

using deepl to translate you get a much better result .

And Black is saved by the move 1. ... Kd2. If now 2. a4, then 2.... f5, 3. Kd4 and now not 3....f4 and the pawn falls, but 3.... Ke2 and after 4. Ke5 (4. a5 is answered with f4 etc. with a change of move or simultaneous pawn conversion) 4....Ke3 and after 5. Kxf5 Black plays Kd4 and is suddenly in the square to pick up the white pawn. If White immediately tries 2. Kd4 Ke7 3. Ke5 to prevent this, Black plays 3.... f5+ and after 4. Kxf5 Kd6 5. a4 Kc4 and is squared again - a draw!

[Joey Stewart]

German chess might as well be the same as chess in any other language but English - literally impossible to understand

[AustinElliott]

Neither specifically chess nor German related, but one of the things that sometimes throws the translation algorithms is abbreviations, notably if the abbreviation is actually also a word in the relevant language. E.g. when I get my French online weather forecast translated into English, the days Wednesday and Thursday mysteriously appear in English as 'Sea' and 'Game' (because the French Mercredi and Jeudi are abbreviated on the page to 'Mer' and 'Jeu', which are both French words with the above meanings).

Reading what people have written here, I guess some of the AI-infused algorithms are contextually intelligent enough to avoid this sort of stuff. The advance in the machine translations over the internet years has been impressive, to my non-expert mind - when machine auto-translators for German first appeared a decade plus ago, the output was fairly unreadable (much worse than the machine translations for French), but now it is really pretty good.

[Christopher Kreuzer]

Agreed (DeepL, mentioned by Bob above, is a good example). I do hope that those involved in those advances are documenting the progress and will publish the full story at some point, or even that it is being tracked indpendently by various linguistics researchers (or whatever the subspeciality is for translation).