The Missing Tengwar Modes

Writing Chinese with Tolkien's Elvish script

Years ago I found some obscure pages describing Tengwar modes for Polynesian languages, and another for Lojban. I thought: if someone has done Lojban, surely all the major language families are covered.

They are not.

Tengwar—the Elvish writing system from Tolkien's legendarium—has been adapted to write dozens of languages. There's a Polish mode. A Korean mode. Multiple Japanese modes arguing about which is correct. But nobody had published a mode for any Chinese language. Not Mandarin. Not Cantonese. Not any of the others.

The sole precedent for any tonal language was a Vietnamese mode from 2016.

So I decided to fix that.

Why this is actually a reasonable project

Mandarin is surprisingly well-suited to Tengwar. Its phonology is small and regular: 21 initial consonants, a handful of vowels, only two possible coda consonants, and a strict syllable template. The entire phonetic inventory fits neatly onto the Tengwar consonant grid—four columns for place of articulation, rows for manner.

The challenge is tone. Every Mandarin syllable carries one of four contrastive tones. Miss the tone and you've said a different word. The script has to encode them.

The solution: below-tengwa marks that trace pitch contour. A horizontal bar for Tone 1 (high level). A rising stroke for Tone 2. A chevron for Tone 3 (the dipping tone). A falling stroke for Tone 4. The shape shows you the pitch movement—the same principle Pinyin diacritics use, made of the same strokes.

Things you might learn about Chinese phonology

Working on this project means wrestling with how Chinese languages actually work:

Mandarin's stops aren't voiced. The contrast everyone hears as "b vs. p" is actually /p/ vs. /pʰ/—unaspirated vs. aspirated, not voiced vs. voiceless. This matters for mapping to Tengwar grades.

"Seven Chinese languages" undersells it. Cantonese and Min are as different from Mandarin as Portuguese is from Romanian. Min preserves voiced stops that Mandarin lost a thousand years ago. Wu (Shanghainese) has a three-way tonal distinction based on register that requires different Tengwar conventions entirely.

The letter "i" in Pinyin lies to you. After zh, ch, sh, r, z, c, s, the written "i" isn't the vowel /i/ at all—it's a syllabic continuant, the consonant itself stretching through where a vowel would be. Some analyses say these syllables have no vowel. The mode represents this by having no vowel tehta. A bare consonant with a tone mark means syllabic consonant.

Tones are contours, not pitches. Tone 3 in Mandarin isn't "low"—it's a dip that falls and then rises. Cantonese has six tones. Min has seven. The marks need to show shape, not just relative height.

The actual project

The repo contains seven complete mode specifications:

• Mandarin — Pinyin as the reference romanization

• Cantonese — Jyutping romanization, six tones, final stops -p -t -k

• Hakka — Taiwan MOE romanization

• Gan — no standard romanization exists, so the project defines one

• Xiang — the Hunan varieties

• Min — Tai-lo/POJ, seven tones, voiced stops

• Wu — Shanghainese and relatives, five-tone register system

Mandarin, Cantonese, Hakka, Gan, and Xiang form "Group A"—their modes are mutually compatible because none have voiced obstruents. Min and Wu form "Group B" with classical Tengwar voicing semantics, incompatible with Group A but phonetically correct for languages that actually distinguish voicing.

Each spec includes sample texts. The Mandarin mode has all the Lord of the Rings title translations—both the Mainland and Taiwan editions—rendered in Tengwar. 魔戒 (mó jiè, "Magic Ring") becomes {malta}[o]²{quesse}{+pal}{anna}[e]⁴.

I also had to extend a Tengwar font. The Alcarin Tengwar fork adds nine glyphs for the below-tengwa tone marks.

The whole thing

It's on GitHub: https://github.com/MarkAtwood/tengwar-sinitic.

The modes are CC BY-SA. The font extensions are OFL. Use them, extend them, tell me what's wrong.

Nobody needed this to exist. But the gap was there, and filling it meant learning how Chinese phonology actually works, how Tengwar modes are designed, and why all the existing modes make the choices they do. If you read through the specs, you might learn some of that too.