Felo

Hunnic Language Translator Online

Convert English text into a constructed Yeniseian-inspired 'Hunnic' lexicon and back, using phrase pairs, a word list, intraword character substitutions and a regex rule. Unlike a bare textarea tool, it reports which tokens matched, which were left unchanged, and applies the passes in a documented order. Try it online.

What is Hunnic Language Translator Online?

Hunnic Language Translator Online is a conversational text converter for a constructed, Yeniseian-inspired 'Hunnic' lexicon. You type English (or Hunnic) text, and the Agent substitutes known phrases and words, applies a small set of intraword character substitutions, and runs one regex rule — then hands back copyable plain text. It is built for the same job as the classic two-textarea translator: quick, playful two-way substitution. The important difference is transparency. Instead of silently emitting placeholder values, the Agent tells you which tokens matched a phrase, which matched a single word, which were transformed by character rules, and which were left untouched.

A word of honesty up front: Hunnic is very poorly attested. The lexicon here is a constructed conlang with Yeniseian-style core vocabulary and Turkic, Mongol, Uralic and Indo-European loanword flavor. This is a creative tool, not a scholarly reconstruction. The conversion data is also deliberately small: five phrase pairs and a handful of word pairs whose source targets were real rather than 'undefined' placeholders.

Key Features

Bidirectional substitution. Send English and get Hunnic, or send Hunnic and get English. Phrase-level matches are checked before single words, so multi-word expressions like 'good luck' resolve as a unit.
Documented pass order. The Agent applies phrases, then words, then intraword character substitutions, then the regex rule — and states that order in its reply so results are reproducible.
Coverage reporting. Every response lists unmatched tokens. You immediately see which words have no entry rather than receiving a silent placeholder.
Glossary mode. Ask for the glossary and get the known phrase pairs and the known word pairs as a clean table you can copy.
Random sentence generation. Ask for a random sentence and the Agent builds one only from the embedded entries that have real targets.

How to Use

Type your text and say which direction you want, for example: English to Hunnic: my wife and good luck at Attila's Court. If you omit the direction, the Agent infers it from the script and vocabulary and states its assumption.
Read the converted block first. It appears in a copyable code block, followed by a short match report naming phrase matches, word matches, character substitutions and unmatched tokens.
Refine in a follow-up: ask for the glossary, ask why a word was left unchanged, or ask for a random sentence. You can also paste Hunnic text back for the reverse direction.

Example Use Cases

Worldbuilding flavor text. A tabletop GM needs a short line of constructed Hunnic dialogue for a nomadic-court scene and wants to know which words are real lexicon entries versus left in English.
Reverse lookup for a phrase. You found tanno in a note and want to know it maps to 'good luck'; paste it and ask for the English direction with an explanation.
Checking coverage before writing. You have a paragraph of English and want to know, before committing, how many of its words actually have Hunnic entries — the coverage report answers that in one pass.

Tips for Best Results

Keep phrases intact. 'good luck' converts as a phrase; splitting it into separate words may change the result.
Ask for the match report if you only see the converted block — it tells you exactly which rules fired.
Expect many unmatched words. The embedded lexicon is small, so most English words will be preserved unchanged and listed as unmatched; that is the intended, honest behavior.
Treat output as creative conlang, not as evidence about the historical Hunnic language.

FAQ

What input format does it accept? Plain text, typed directly into the chat. There is no file upload. Very long passages are handled in sections, and the Agent tells you when it splits.

How accurate is the result? Accurate to the documented rule set: the embedded phrase pairs, the embedded word pairs, three intraword character substitutions and one regex rule. It is not a linguistic reconstruction, and unmatched words are preserved rather than guessed.

How is this different from a bare textarea translator? This Agent applies a stated pass order and reports unmatched tokens, so you can see the limits of the lexicon instead of discovering them by accident. It never emits placeholder values for missing entries.

Can it translate a whole language? No. It substitutes a fixed lexicon and a few character rules. It does not generate grammar, morphology or syntax.