I have no idea what this site is about other than being a showcase for texts in apparently invented languages; this recent post, for example, begins:
Aża cīfir zuoj Thorenur khioj thurenur, řa duiker. Siřover taekar ofo, ici se laima şiekhar ifi koiden, siekar, beifir ata zirenur roidir, vu go ifi sibe. Eçe aba diamur ikhi mēli khiari cui ozho, thai mebi, koimir, nofo uzhu, şiala da çuizir reizir, roidir cerenur poi merenur. Seiş lui me khugo eve do feamur soilir. Vēs unu çīmir, ei thoamur imu siekar çafa galu deidir feamur, ethe i vueni çei petenur o apha ucu raifir soelo theamur, lāmur pharenur vai, ri bī vāş boiben thaga morenur ořo eve goekher? Vobe uru zuifir ata çişur laima khoel.
Şarenur udu, zhutenur feamur bamu. Patenur ule reilir ata ucu zuelo bamo goekher? Bui uřu laidir feamur te feuş te, tařover obo uamur ukhu, çui o işi vueni çala ene, şurenur zai mola coekhar lai ifi keala çala khioj ikhi, foimir, efe ukhu mēli noelo otho şarenur noej sibe. Kufa idi çuizir iři imi, řoi vu ono. Uzhu ene zeamur roidir zueş. Da bui eşe sitenur eve rei, khiari ephe theamur ça leilir tuifir işi gaşur zoima thuku, theamur voşur, çofe thuipen khe. Nudo zo goekher? Bai be zhai, çoekar nofo nuima patenur vu feamur kheidir ubu řili rogu ukhu zuelo tāmur eze citenur sibe. Ese vāş neşur. Zirenur khāva şetenur. Şēkhar ethe khiari phī, roidir gifa ucu gei tāmur asa. Zuelo zeizir daimo guşu itenur ru atha feuş khugo akha oco leidir. Ava, ifi aga kufa ephe kāmur řīlir, iekar keka. Ulu fetenur sago da thurenur theidir, zhī boiben, oro khī iamur, vēs, khogo go, ifi rauş siřover ede ephe çuizir pheamur sişur soilir. Coizir phe řuamur, zuoj.
Zuoj indeed, not to mention Iä. Make what you will of it; there are more links, with various speculations, at the MeFi post where I found it. But stay safe out there!
Could serve as a replacement for lorem ipsum, maybe. Lorem ipsum is supposed to help you concentrate on typesetting matters rather than the text, but it does the exact opposite for me.
It looks like rather the kind of site where I’d be reluctant to click on any links …
Though it may be that I am paranoid, and the links would only lead to some Hell dimension.
I clicked on several of them, and I̴̧̛̛̛̥̼̙̣̩̲̦͌̑͊̾͒̐̓̅̅̎͗̈́͑͝’̸̛̛̠͔̙̫̻̎̐̃͊̎͒̽̔̇̂͋m̶̛͔͙̮̟̭̀́̊̒̑̃̄̿̕̕͘ͅ ̸̦̠̲̤̞̙̻̊͋́̂͛͊͊̍̈́͘͝͠ͅf̷̻̟̠̯̝̯̪̪̙̫͇̟̞̟͙̔͒̇̿̍͐͑͛͂̓͐̑͘͝ï̷̢̨̛͈̫̯͈͕̮̦̓̃̍̊͒̃̀͊̇͑͑͝͝ņ̵͇̗̣̹͉̳͉̠̩̬͒͌ͅě̵̡̨͍̯̤͖̜̗̠͕́̅̃͑̚͠!̷̛͕͕̬̙̙́͐̾͛́͌͝
DeepSeek says the language is Ithkuil, invented by one John Quijada (details on Wikipedia), and means:
I can’t vouch for DeepSeek’s paragraphing. Burárum!
I just put the text into DeepSeek again, and got a totally different hallucination… So I guess it’s not even Ithkuil…?
Burárum!
We discussed Ithkuil back in 2012. Burárum!
The WP article on Ithkuil has the following Romanised example:
Even allowing for the possibility of very different transcription schemes, it seems difficult to interpret this as the same language as in Hat’s extract.
many CVVCVC words. Usually CVVCVr.
and
CVVmur
but
CVCVnur
The blog attributes the entry to an ‘Ayden’, a plausible Turkish name. It includes Turkish characters ç and ş, but oddly also r-hacek which isn’t Turkish. And the letter distribution is way off.
And yeah, I’d be wary of clicking any links, as David Eddyshaw says.
I would be astonished if Deepseek (or any LLM) actually can translate Ithkuil.
(I just discovered once again that GT, admittedly far from the poster child for machine translation nowadays, can’t even reliably translate from Hausa, which must be well to the easier end among African languages for APEs and their cousins. Lots of stolen text for them to phagolyse.)
The Ithkuil snippet from WP suffices to show that the language is Oti-Volta:
p’amḿ: Kusaal pamm “a lot, greatly”
tua: Kusaal tua “pound in a mortar”
tʊ꜔꜖a: Kusaal tʋlla “be hot” (with the negative enclitic)
kit: Kusaal kɛt “allow” (imperfective aspect)
(The translation given on the WP page seems to be very free indeed, and it is clearly not to be relied upon in detail.)
The autoglottonym Ithkuil probably reflects a phrase cognate with Mooré
d kuili (= Kusaal ti kul) “(that) we go home”, a cross-linguistically common and semantically transparent pattern for self-designations.
The language of Hat’s excerpt is clearly not Oti-Volta, and quite possibly not even Gur at all. However. the frequent noun plurals and finite verb forms in -Vr do suggest an affinity with the northern branch of Scandi-Congo.
If it is invented, what is the point of it? Apart from having fun and intellectual stimulation.
Strangely, my Google translate says it’s Breton, and while failing to translate it in full, she managed to give me a few bits in Russian, but not in English or French.
This passage:
Çala pēl bejov ini phī řoilir pheno mo fouş, rauş motenur ugu phelu fei semo, çajev şarenur ŗoilir eke cu oro żoamur vutenur, citenur deidir khirenur norenur lebe cema dīlir zoimir.
is in Russian:
Если вы хотите, чтобы я был в восторге от того, что вы делаете, я могу сделать это, когда вы хотите, чтобы это произошло, и вы можете сделать это, когда вы хотите, чтобы вы сделали это, и вы должны сделать это, чтобы получить больше информации.
Which in English goes like this:
If you want me to be excited about what you do, I can do it when you want it to happen, and you can do it when you want it to happen, and you should do it to get more information.
If I put the whole thing into Google translate it tells me it’s Breton and translates the title as ‘Find the ciphers’, but if I put only the title in it tells me it’s Slovenian and means ‘What is the number of days?’
I liked the first version, because my first guess when I looked at it was that it was a substitution cipher, but it’s not Breton, or at least doesn’t use Breton diacritics.
Putting in a sentence at a time is kind of fun – apparently ‘Ava, ifi aga kufa ephe kāmur řīlir, iekar keka’ means ‘If you are a person who is not a person, you are a cake’!
Definitely not Breton.
It would be moderately interesting to guess just why GT misidentified it as Breton, I suppose.
Probably a combination of matching a few random letter sequences with Breton, combined with having too little pirated Breton data overall to correct the initial jumping to conclusions. I suspect that it would never misidentify it as something like French or German.
But asking “why?” of an “AI”‘s behaviour is about as sensible as taking up haruspicy.
I was just thinking of taking up haruspicy! Got a liver I can borrow?
I would look for prominent text features like:
1. Goekher only and always appears as the last word in a question.
2. There are many V1CV2 words. In virtually all of these words V1 = V2; to me the number and placing of these words is inconsistent with them functioning as a grammatical category like “pronoun”, “preposition”, etc.
Six occurrences of zh. But it also shares that, and th and ç, with Albanian…
The analysis in the Strange Minds post linked on Metafilter seems to suggest that it’s at least partially an English relex– note how well the text lines up with idioms like “as well as.”