I regret not posting it before Canvas 2025, but hopefully it'll be useful for people playing it in 2026. All letters are 5 pixels tall, and most 3 pixels wide (some 4, a few 5). I've also included a few Cyrillic letters and the digits.

I tried to make it even smaller, but it gets really funky.

Interesting video on the stone that allowed researchers to decipher Ancient Egyptian. Check comments for a few notes.

Additional links with press coverage: ArcheologyMag, Oxford.

For context:

The Huns were nomadic people from Central Eurasia; known for displacing a bunch of Iranian (e.g. Alans) and and Germanic (e.g. Goths, Suebians etc.) speakers, that ultimately invaded the Roman Empire. They reached the Volga around 370 CE, and one of their leaders (Attila) is specially famous. Often believed to be a Turkic people, but if the study is correct they're from a completely different language family instead.

The Xiōng-Nú are mentioned by Chinese sources as one of the "Five Barbarians" (i.e. non-Han people). They would've lived in Central Eurasia between 300 BCE and 100 CE or so, and eventually became Han tributaries.

The Paleo-Siberian language in question would be an older form of Arin, a Yeniseian language. Yup, that same family believed by some to have relatives in the Americas.

For further info, if anyone is interested, Stephen Bax claimed a decade ago to partially decode the manuscript; here's a video with his reasoning, as well as the paper he released. Sadly Bax passed away in 2017 (may he rest in peace), so the work was left incomplete.

The main idea behind this language is to become evolutionary food for other languages of my conworld. As such I'll probably never flesh it out completely, only the necessary to make its descendants feel a bit more natural.

Constructive criticism is welcome.

Context and basic info

The conworld I'm building has three classical languages, spoken 2~3 millenniums before the conworld present: Old Sirtki, Classical Tarune, and Mäkşna. And scholars in the conworld present are reconstructing their common ancestor, that they call "Proto-Sitama".

What I'm sharing here, however is none of their fancy reconstructions. It's the phonology of the language as it was spoken 7 millenniums before the conworld present. Its native name was /kʲær.mi.'zɑst/, or roughly "what we speak"; the language itself had no written version but it'll be romanised here as ⟨Cjermizást⟩.

Its native speakers were a semi-nomadic people, who lived mostly of livestock herding. They'd stay in a region with their herds, collect local fruits and vegetables, and then migrate for more suitable pasture as their animals required.

It was quite a departure from the lifestyle of their star travelling ancestors, who were born in a highly industrialised society in another planet.

Grammar tidbits

Grammar-wise, Cjermizást was heavily agglutinative, with an absolutive-ergative alignment and Suffixaufnahme. So typically you'd see few long polymorphemic words per sentence. Those morphemes don't always "stack" nicely together, so you often see phonemes being elided, mutated, or added to the word.

Consonants

Cjermizást features a contrast between "soft" and "hard" consonants. "Soft" consonants are palatalised, palatal, or post-alveolar; "hard" consonants cannot have any of those features. Both sets are phonemic, and all those consonants can surface outside clusters.

Palatalised consonants spawn a really short [j], that can be distinguished from true /j/ by length.

Although /j/ and /w/ are phonetically approximants, the language's phonology handles them as fricatives, being paired with /ɣ/ and /vʲ/ respectively.

/r rʲ/ surface as trills or taps, in free variation. The trills are more typical in simple onsets, while the taps in complex onsets and coda.

The contrast between /m n/ is neutralised when preceding another consonant in the same word, since both can surface as [m n ŋ]; ditto for /mʲ nʲ/ surfacing as [mʲ ɱʲ ɲ].

Coda /g/ can also surface as [ŋ], but only in word final position; as such, it doesn't merge with the above.

Liquids clustered with voiceless fricatives and/or stops have voiceless allophones.

Vowels

Proto-Sitama's vowel system is a simple square: /æ i ɒ u/. They have a wide range of allophones, with three situations being noteworthy:

• /ɒ u/ are typically fronted to [Œ ʉ] after a soft consonant
• /æ i/ are backed to [ɐ ɪ] after a hard velar
• unstressed vowels are slightly centralised

Accent

Accent surfaces as stress, and it's dictated by the following rules:

1. Some suffixes have an intrinsic stress. If the word has 1+ of those, then assign the primary stress to the last one. Else, assign it to the last syllable of the root.
2. If the primary stress fell on the 5th/7th/9th/etc.-to-last syllable, move it to the 3rd-to-last
3. If the primary stress fell on the 4th/6th/8th/etc.-to-last syllable, move it to the 2nd-to-last.
4. Every two syllables, counting from the one with the primary stress, add a secondary stress.

Phonotactics

Max syllable is CCVCC, with the following restrictions:

• complex onset: [stop] + [liquid]; e.g. /pl/ is a valid onset, */pw/ isn't
• complex coda: [liquid or nasal] + [stop or fricative]; e.g. /nz/ is a valid coda, */dz/ isn't

If morphology would create a syllable violating such structure, an epenthetic /i/ dissolves the cluster.

Consonant clusters cannot mix hard and soft consonants. When such a mix would be required by the morphology, the last consonant dictates if the whole cluster should be soft or hard, and other consonants are mutated into their counterparts from the other set. For example, */lpʲ/ and */ʃp/ would be mutated to /ʎpʲ/ and /sp/.

Stops and fricatives clustered together cannot mix voice. Similar to the above, the last consonant of the cluster dictates the voicing of the rest; e.g. */dk/ and */pz/ would be converted into /tk/ and /bz/ respectively.

Gemination is not allowed, and two identical consonants next to each other are simplified into a singleton. Nasal consonants are also forbidden from appearing next to each other, although a cluster like /nt.m/ would be still valid.

Word-internal hiatuses are dissolved with an epenthetic /z/. Between words most speakers use a non-phonemic [ʔ], but some use [z] even in word boundaries.

Romanisation

As mentioned at the start, the people who spoke Cjermizást didn't write their own language. As such the romanisation here is solely a convenience.

• /m n p t b d g s x w z l r/ are romanised as in IPA
• /k ɸ ɣ/ are romanised ⟨c f y⟩
• "soft" consonants are romanised as their "hard" counterparts, plus ⟨j⟩
• ⟨j⟩ is omitted inside clusters; e.g. /pʲʎ/ is romanised as ⟨plj⟩, not as *⟨pjlj⟩
• /æ i ɒ u/ are ⟨e i a u⟩

Use this thread to ask questions or share trivia, if you don't want to create a new thread for that.

[Note: the purpose of this thread is to promote activity, not to concentrate it. So if you'd still rather post a new thread, by all means - go for it!]

Quick summary: a tablet written in Hittite, from a likely vassal to their king, recounts how Attaršiya [Atreus?] of Ahhiyawa [the Achaeans] and his sons attacked Taruiša [Troy]. And at the end there's a fragment in another Anatolian language, Luwian, saying the following:

wa-ar-ku-uš-ša-an ma-a-aš-ša-ni SÌ[R
wrath.ACC god(dess).VOC? si[ng

So roughly "Sing, oh goddess, the wrath..."

This is pretty much how the Illiad starts in Greek:

μῆνιν ἄειδε θεὰ Πηληϊάδεω Ἀχιλῆος
mênĭn áeide theā́ Pēlēïádeō Akhĭlêos
rage.ACC sing.IMP goddess.VOC Peleus.GEN Achilles.GEN
Sing, oh goddess, the rage of Achilles [son] of Peleus

Feel free to use this thread to ask small questions or share random language / linguistics trivia, if you don't feel like creating a new thread just for that.

(Just to be clear: yes, if you want to create a new thread for your question/trivia, you can. I'm only trying to stimulate discussion in the comm.)

This infographic is still incomplete; I'm posting it here in the hope that I can get some feedback about it. It has three goals:

1. To explain what federation is. No technobabble, just a simple analogy with houses and a neighbourhood.
2. To explain why federation is good for users.
3. [TODO] Specific info about the Fediverse, plus some really simple FAQ.

Criticism is welcome as long as constructive.

EDIT: OK, too much text. I'm clipping as much as I can.

This is not some sort of fancy new development, but it's such a classical experiment that it's always worth sharing IMO. Plus it's fun.

When you initially mix both solutions, nothing seems to happen. But once you wait a wee bit, the colour suddenly changes, from transparent to a dark blue.

There are a bunch of variations of this reaction, but they all boil down to the same things:

• iodide - at the start of the reaction, it'll flip back and forth between iodide (I⁻) and triiodide ([I₃]⁻)
• starch - it forms a complex with triiodide, with the dark blue colour you see in the video. But only with triiodide; iodide is left alone. So it's effectively an indicator for the triiodide here.
• some reducing agent - NileRed used vitamin C (aka ascorbic acid; C₆H₈O₆), but it could be something like thiosulphate (S₂O₃²⁻) instead. The job of the reducing agent is to oxidise the triiodide back to iodide.
• some oxidiser - here it's the hydrogen peroxide (H₂O₂) but it could be something like chlorate (ClO₃⁻) instead. Its main job is to oxidise the iodide to triiodide. You need more than enough oxidiser to be able to fully oxidise the reducing agent, plus a leftover.

"Wait a minute, why are there a reducing agent and an oxidiser, doing opposite things? They should cancel each other out!" - well, yes! However this does not happen instantaneously. And eventually the reducing agent will run dry (as long as there's enough oxidiser), the triiodide will pile up, react with the starch and you'll get the blue colour.

Here are simplified versions of the main reactions:

1. 3I⁻ + H₂O₂ → [I₃]⁻ + 2OH⁻
2. [I₃]⁻ + C₆H₈O₆ + 2H₂O → 3I⁻ + C₆H₆O₆ + 2H₃O⁺

(C₆H₆O₆ = dehydroascorbic acid) Eventually #2 stops happening because all vitamin C was consumed, so the triiodide piles up, reacts with the starch, and suddenly blue:

EDIT: @[email protected] shared something that might help to circumvent this shit:

Contained in these parentheses is a zero-width joiner: (​)

Basically, add those to whatever you feel that might be filtered out, then remove the parentheses. The content inside the parentheses is invisible, but it screws with regex rules.