Skip to content
LinguaCommons
← Writing Systems

Tengwar — Tolkien's Fëanorian script, and the modes he never wrote

The Tengwar is a writing system devised by J. R. R. Tolkien, probably in the late 1920s or early 1930s. Its first published appearance is the Lonely Mountain jar inscription in *The Hobbit* (1937); its full description is Appendix E of *The Lord of the Rings* (1955).1 The single most useful thing to understand before you start is this: **Tengwar is not a language.** It is a script, and a deliberately flexible one — the same letters carry different sound values depending on which *mode* you are writing in.5 Quenya, Sindarin, Black Speech, Old English and modern English have all been written in it, by Tolkien himself.3

Two histories: the invented one and the real one

Inside the fiction, the Tengwar were invented by the Elf Fëanor in Valinor and used first for Quenya and Telerin; a *tengwa* is a letter (plural *tengwar*), replacing the older term *sarat*, from Rúmil's earlier alphabet, the Sarati. When the script reached Beleriand the Sindar called the letters *tîw* (singular *têw*), against their own native *cirth*.1

Outside the fiction, the Tengwar has ancestors of its own. The Sarati (described in *Parma Eldalamberon* 13) already anticipates vowel-marking by diacritics and some sound/shape correspondences; the Valmaric script (*Parma Eldalamberon* 14), which Tolkien used c. 1922–25, is closer still, sharing letter shapes, the inherent vowel [a], and an arrangement very like the classical Quenya table.1 Jim Allan compared the whole design to Francis Lodwick's 1686 Universal Alphabet, on the grounds of both letter shapes and the shape/sound correspondence principle.1

How the script is built

Tengwar is *featural* to an unusual degree: the shape of a letter tells you the place and manner of articulation of its sound. Most letters are made from a vertical stem (*telco*) plus one or two rounded bows (*lúva*).

  • **Series** (*témar*) — the four columns — correspond to places of articulation, distinguished by which side the bow sits on and whether it is closed. In the classical Quenya mode: I *tincotéma* dental, II *parmatéma* labial, III *calmatéma* velar, IV *quessetéma* labiovelar. A fifth palatal series, the *tyelpetéma*, has no letters of its own; it is written with a following-y diacritic (usually two underposed dots).15
  • **Grades** (*tyeller*) — the six rows — correspond to manner. Doubling the bow voices the consonant; raising the stem above the line makes it a fricative; shortening the stem to bow height makes it a nasal (grade 5) or, at grade 6, an approximant.1
  • Some tengwar have an **extended** form, with the stem projecting both above and below the line. These are rare, and mode-specific: a Quenya mode uses extended *parma* for /pt/ and extended *calma* for /kt/, and the English general use employs extended *ando* as an abbreviation for *the*.15
  • **Vowels** are handled in one of two ways, and this is the deepest split in the system. In *ómatehtar* modes the vowels are diacritics (*tehtar*) sitting above a consonant — which makes those modes abugidas rather than alphabets. In *full* modes the vowels get their own tengwar; the well-known one is the Mode of Beleriand, the script on the Doors of Durin.1
  • The vowel tehtar resemble Latin diacritics: circumflex /a/, acute /e/, dot /i/, left curl /o/, right curl /u/. Long vowels other than /a/ may be written by doubling. A vowel with no consonant to sit on rides a **carrier** — short *telco*, shaped like a dotless i, or long *ára*, like a dotless j.12
  • Whether the vowel tehta goes above the **preceding** consonant (Quenya) or the **following** one (Sindarin, English, and the One Ring inscription) is a property of the mode, not of the script.1

The character inventory

Tengwar has no Unicode encoding (see below), so no chart here can show you the glyphs — this table gives the 24 primary tengwar by their Quenya "full names" from Appendix E, arranged as Tolkien's own table arranges them, with the classical Quenya values. Read across for place of articulation, down for manner.2

Grade (tyellë)I tincotéma (dental)II parmatéma (labial)III calmatéma (velar)IV quessetéma (labiovelar)
1 — descending stem, single bowtinco "metal" /t/parma "book" /p/calma "lamp" /k/quesse "feather" /kw/
2 — descending stem, doubled bowando "gate" /nd/umbar "fate" /mb/anga "iron" /ŋg/ungwe "spider's web" /ŋgw/
3 — raised stem, single bowsúle < thúle "spirit" /θ/formen "north" /f/aha "rage" (< harma "treasure") /x/hwesta "breeze" /xw/
4 — raised stem, doubled bowanto "mouth" /nt/ampa "hook" /mp/anca "jaws" /ŋk/unque "a hollow" /ŋkw/
5 — short stem, doubled bownúmen "west" /n/malta "gold" /m/noldo < ñoldo "deep-elf" /ŋ/nwalme < ñwalme "torment" /ŋw/
6 — short stem, single bowóre "heart, inner mind" (weak medial r)vala "angelic power" /v/anna "gift" (absent consonant / y)vilya < wilya "sky" (v < w)

Twelve further tengwar fall outside the four-by-six table. Their values vary more than the primaries do.2

TengwaName meansClassical Quenya use
rómen"east"/r/ (full trilled r)
arda"region"/rd/
lambe"tongue"/l/
alda"tree"/ld/
silme"light"/s/
silme nuquerna"silme reversed"/s/ — the reversed form, used where a tehta needs the space above
esse (<< áre < áze)"name" (<< "sunlight")originally /z/; after z merged with r, reapplied to /ss/ and renamed
esse nuquerna"esse reversed"as esse, reversed form
hyarmen"south"/hj/, later /h/
hwesta sindarinwa"grey-elven hwesta"the Sindarin pronunciation of hw
yanta"bridge"consonantal i (/j/)
úre"heat"consonantal u (/w/)

Two more are outside Tolkien's numbered table altogether: *halla* "tall", originally breath-/h/ in the classical mode and in the Mode of Beleriand a hiatus mark for lenited g (there called *gasdil* "stopgap"); and *vaia < waia* "envelope", presumably /v/ from earlier /w/, attested in only one source.2

Modes: the same letters, different values

A mode is a set of conventions for writing one language. Måns Björkman's classification — the standard one — sorts modes not by language but by two structural questions: how are vowels marked, and which series carries the velars?3

ModeVowelsVelars inSample languages
Classical Modeómatehtar, above the preceding tengwaSeries IIIQuenya
General Useómatehtar, above the preceding *or* following tengwaSeries IVEnglish, Sindarin, Quenya, Old English, Adûnaic, Black Speech, Westron, Latin
Quanta Sarmefull (vowel tengwar)Quenya (?) — no certain samples; "used by the Loremasters for special purposes"
Mode of Beleriandfull (vowel tengwar)Series IIISindarin — the Doors of Durin inscription
Later / Westron Convention (Northern variety)fullSeries IVEnglish, Sindarin, Quenya
Qenya AlphabetfullSeries IVEnglish, Old Norse — pre-*Lord of the Rings*, may not be "canonical" in-world

In General Use the four series heads shift from /t p k kʷ/ to /t p tʃ k/, and grade 4 — raised stem with doubled bow — carries voiced fricatives rather than the nasal+stop clusters it carries in Quenya. Grade 6 becomes the approximants. This is why the same glyph is /k/ in one inscription and /tʃ/ in another, and why "what does this tengwa mean?" is not a well-formed question without naming the mode.13

The modes for English — attested, but not a canon

This is where popular accounts go wrong in both directions. It is **not** true that Tolkien never wrote English in Tengwar — most of his surviving Tengwar samples are in English, and the Mellonath Daeron Index of Tengwar Specimina (DTS) catalogues more than a dozen English inscriptions across the General Use, the Later/Westron convention and the early Qenya Alphabet.13 But it is also not true that there is an official English mode. Tolkien said so himself in Appendix E, of the title-page inscription in *The Lord of the Rings*:4

There was of course no 'mode' for the representation of English. One adequate phonetically could be devised from the Fëanorian system. The brief example on the title-page does not attempt to exhibit this. It is rather an example of what a man of Gondor might have produced, hesitating between the values of the letters familiar in his 'mode' and the traditional spelling of English.

So Tolkien wrote English two ways, and both survive in his own hand:4

  • **Orthographic spelling** — each tengwa or tehta stands for a Latin letter or letter-sequence, transliterating English spelling with its silent letters intact. This is what "a man of Gondor" produced. It is not perfectly mechanical: voiced *th* and *s* are distinguished from voiceless, and pronunciation sometimes wins (the *i* tehta for the final vowel of *history*, the *o* tehta for the vowel of *war*). An underposed dot marks silent *e*; a doubled Roman consonant is a line under the tengwa.
  • **Phonemic spelling** — one tengwa per English phoneme. Front closing diphthongs (/eɪ/, /aɪ/) are written as a vowel tehta plus *anna*; back closing diphthongs (/oʊ/) as a vowel tehta plus *vala*. How centring diphthongs would be written is unknown; they evidently did not matter in Tolkien's own English. Short unstressed vowels are sometimes simply dropped.
  • Both use a superimposed bar for a preceding nasal of the same series, a downward hook for a following *s*, and shorthand forms for *the*, *of*, *of the* and *and* (the last written by omitting the vowel sign).4

The modes for Spanish and other languages — community, not canon

Everything beyond Tolkien's own languages plus English is post-Tolkien community work. Modes have been created since 1955 for Spanish, German, Swedish, French, Finnish, Italian, Hungarian and Welsh, and for the constructed languages Esperanto and Lojban.1 These are conventions agreed among transcribers and encoded into tools; they have no textual authority in Tolkien's writings, and competing versions of the same mode exist. Some enthusiasts have added letters not found in Tolkien's writings at all.1

That is a legitimate thing to do — a script belongs to whoever uses it — but it should be labelled. When a tattoo studio, a transcription site or a wedding invitation offers you "Spanish in Elvish", what you are getting is one community's mapping of Spanish phonology onto Fëanorian letters, not something Tolkien wrote or sanctioned. We could not identify a single authoritative published Spanish mode chart with a named author and a stable citation, and we are flagging that rather than presenting one.

Unicode and fonts: the unsolved part

  • **Tengwar is not in Unicode.** Michael Everson proposed an encoding in 1997 (WG2 N1641). The range U+16080–U+160FF in the Supplementary Multilingual Plane was tentatively allocated on the 2023 Unicode roadmap, but has since been removed.1
  • The **ConScript Unicode Registry** — an unofficial allocation of Private Use Area code points — maps Tengwar to **U+E000–U+E07F**, including letters, digits (Tengwar counts in duodecimal as well as decimal, with dedicated marks for the base and for the least-significant digit) and variant forms.1 Fonts implementing it include Everson Mono, Tengwar Telcontar, Constructium, Tengwar Formal Unicode, FreeMonoTengwar and the Kurinto "Aux" faces.1
  • The *de facto* standard in the wider user community is older and worse: Dan Smith's typefaces map Tengwar onto **ISO 8859-1** code points, so the underlying text is Latin gibberish and displays as a string of nonsense characters on any machine without the exact font. Because 191 code points are not enough, an overlapping "Tengwar A" font holds the overflow, and each diacritic needs four code points — one per width of the character it sits on. Tengwar Annatar, Tengwar Parmaitë, Tengwar Élfica and Tengwar Formal all follow this encoding, and they differ in details.1
  • Practical consequence: **Tengwar text is not portable.** You cannot copy it into a search box, sort it, spell-check it, or reliably send it to someone else. Ship images, or ship the transcription source plus the mode name.

Who uses it

  • Tolkien fandom, continuously since the 1950s — calligraphy, inscriptions, jewellery, and a steady stream of tattoos. Every actor playing a member of the Fellowship in the Jackson films except John Rhys-Davies took a Tengwar tattoo of the English word *nine*; the footballers Sergio Agüero and Fernando Torres have their names in Tengwar on their forearms.1
  • Scholarly work in Tolkien linguistics: *Vinyar Tengwar*, *Parma Eldalamberon* and the Omentielva conference proceedings, where the primary manuscripts are edited and the modes argued over.1
  • Online transcribers — Tecendil and the Tengwar Transcriber are the widely used ones — which encode a mode's rules so you can type Latin and get Tengwar out.5
  • The Mellonath Daeron Index (DTS) at Forodrim, the running catalogue of published Tengwar samples, is the reference every serious argument about attestation cites.3

Honest limitations

  • **"Write my name in Elvish" is under-specified.** Which language, and which mode? A name transcribed in the English orthographic mode, the English phonemic mode and a Sindarin mode will look different and mean different things.
  • **Phonemic modes require you to know your own phonology,** and they vary between dialects of English — which makes them harder to read for a wide audience than the orthographic mode.4
  • **Attestation is uneven.** In Björkman's charts, tengwar whose values are marked with an asterisk are *not attested* — reconstructed by inference from the system's logic. Reasonable people fill those gaps differently.
  • **The mode names themselves are provisional.** Björkman notes that every mode name he uses is one Tolkien applied at least once to an orthography matching that description, but that the names may not be correctly applied and future publications may force revision.3
  • **No encoding means no future-proofing.** Anything you write today in a PUA or ISO-8859-1 font is a private convention, not a document.

Learn more

Notes & Bibliography

  1. "Tengwar." Wikipedia — dating (late 1920s/early 1930s; Lonely Mountain jar 1937; Appendix E 1955), Sarati and Valmaric antecedents and the Lodwick comparison, in-fiction history and terminology (tengwa/sarat/tîw/cirth), témar and tyeller structure and the parmatéma worked example, extended stems, ómatehtar vs full modes and tehtar shapes, carriers, classical vs general-mode series values, post-Tolkien modes for Spanish/German/Swedish/French/Finnish/Italian/Hungarian/Welsh/Esperanto/Lojban, Everson's 1997 proposal and the deleted U+16080–U+160FF roadmap slot, CSUR U+E000–U+E07F and digits, the Dan Smith ISO 8859-1 de-facto encoding and its flaws, font list, fandom and tattoos. [source]
  2. Björkman, Måns. "Tengwar Names." Amanye Tenceli — the Quenya "full names" from Appendix E with glosses and classical-mode values for tengwar 1–36, the unnumbered halla and vaia, the tyelpetéma, carriers (telco, ára) and tehta names. [source]
  3. Björkman, Måns. "Tengwar Modes." Amanye Tenceli — the classification by vowel-marking and velar series, the mode table used here with sample languages and DTS references, the caution that mode names are provisional. [source]
  4. Björkman, Måns. "Tengwar — English General Use." Amanye Tenceli — the Appendix E quotation on there being no mode for English, the orthographic/phonemic distinction and its details (silent e, doubled consonants, diphthong writing with anna and vala, dropped unstressed vowels, nasal bar, s-hook, shorthands for the/of/of the/and), and the note on dialect variability. [source]
  5. Gourdol, Arno. "Tecendil: Tengwar Handbook." — Tengwar is a script and not a language; modes as per-language convention sets; the témar/tyeller summary including the tyelpetéma; extended ando as the English abbreviation for "the". [source]