The sphere of the sphere is the sphere of the sphere
« previous post | next post »
In a comment on "Electric Sheep", Tim wrote:
Just want to share a little Google Translate poetry resulting from drumming my fingers on the keyboard while set to Thai:
I want to preserve this gem for posterity (i.e. past the next GT update). So the specific keyboard banging sequence was
่ดฟหกวาวฟไำาพ่้ฟเวฟ่ากวหด่วสฟหกดา่ไ้วำพี้วหกอืวฟหก่้ดสำไอฟ่าส้่หำพะืฟทหอั้ฟส่าาิำืไพ
and here is a screenshot of the result:
Also Google Translate's reading of the output:
And the second repetition, slowed down (for better comprehension, or after a few cybercocktails):
Finally, Google's synthesis of the Thai input:
Roger C. responded:
I'd have believed you if you told me this was one of those passages from the Nag Hammadi codices that are very incompetently rendered into Coptic from Greek.
Viseguy said,
April 22, 2017 @ 9:36 pm
Lovely. Gertrude Stein would approve.
Chas Belov said,
April 22, 2017 @ 10:51 pm
Odd. If I remove the leading apostrophe, the translation is:
The six of them are soooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
[(myl) Indeed they are. But it's not an apostrophe, it's a diacritic, specifically
|่| 0x0E48 "THAI CHARACTER MAI EK"
As Wikipedia explains,
Middle class consonants with a long vowel spell an additional four tones with one of four tone marks over the controlling consonant: mai ek, mai tho, mai tri, and mai chattawa.
In this case there's no preceding "controlling consonant", so the diacritic is displayed over an empty glyph, and looks like an apostrophe:
]
Thorin said,
April 22, 2017 @ 11:54 pm
I tried out the same thing, pressing an entirely random series of keys.
"Drake is a beautiful warrior.
I'm pretty sure I'm right.
I was just about to say that I was justified by the fact that I had to find a doctor."
I guess Drake messed me up pretty good.
Ben Zimmer said,
April 23, 2017 @ 12:19 am
Here's what I got (from typing this):
"It is also known as the distribution of the value of the word distribution. The word is the word of the word. I can not wait to see the results."
[(myl) Again, to preserve the process for posterity, the input sequence was
ๆยรำอืฟรือฟวคภทืฟคฟวืฟวฟ้ฟทฟฟวฟทใือคฟภวทแแิฟำรัฟตฟววผากืดคำวรฟดืฟแาแคยฟงผเัะรำพยฟาวฟดืฟผวอืฟำตค
and the obligatory screen shot is
]
AntC said,
April 23, 2017 @ 12:43 am
SInce the Thai is not merely a repeated two-character sequence, as were most of the Electric sheep prods, does the Thai actually mean anything?
Could you split up the string of characters into word segments more or less? Would any of those segments be "sphere" or "six" or "sky"?
unekdoud said,
April 23, 2017 @ 12:59 am
Sounds like a corrupted version of the geometrical theorem presented in "The Kiss Precise": if four circles in a plane are arranged so that every pair is tangent (hence creating six points of tangency), then the square of the sum of the signed curvatures is twice the sum of their squares.
Of course, there's no chance that Google Translate is improvising on centuries-old mathematical literature… or is there?
mp said,
April 23, 2017 @ 2:50 am
Why only English gets such beautiful poetry? The Polish translation is a buch of ungrammatical nonsense with lots of proper names. And others (I looked at least at Geman, Czech and Spanish) seem to be the same.
[(myl) This is probably related to the fact that the system has many times more training material for English than for other languages.]
Adrian Morgan said,
April 23, 2017 @ 8:36 am
The fear of the spear is the sphere of the seer.
(Which, miraculously, emerges unchanged when translated into Thai and back into English. But say it ten times quickly.)
Rebecca said,
April 23, 2017 @ 9:16 am
@mp –
I'm interested in that question, too. I wondered about similar results in the "tangled web" post, but nobody addressed it. I wonder if Google Translate has some process/architecture running on English target that they don't yet have implemented for other languages.
Tim said,
April 24, 2017 @ 2:43 pm
@mp, @myl
I was surprised to see such a coherent response to my first finger-drumming experiment. Thinking about this from the other direction, translating the same kind of gibberish from Spanish (or French or German, etc.) to English doesn't produce anything reasonable. Perhaps Thai admits more legal word-sequences in random data than Spanish?
[(myl) Two probably-relevant differences:
(1) Thai orthography doesn't usually put spaces between words;
(2) The Thai writing system is an abugida, where consonants have default or intrinsic vowels.
]
Joke Kalisvaart said,
April 24, 2017 @ 3:26 pm
พกกิ่านกกายยใาสลลงใืกสา่ยาเ่กวแรนวฟืหแนรฟืกนอแ้ฆกอื่ว yields "Carrying a bag of cigarettes is a great way to start a drug business." but removing the last two letters from the Thai results in "Carnivorous herbs are available in the form of herbs such as carrots,"
The first sentence actually makes quite some sense, but this is the first time I heard about Carnivorous herbs :)
Rubrick said,
April 24, 2017 @ 5:45 pm
@Chas: The six of them are sooooooooo… what??? I can't believe they're leaving us hanging like that!
bratschegirl said,
April 25, 2017 @ 11:56 am
I bet the carnivorous herbs hang out with the colorless green ideas.