Schematic overview of the phonological evolution from Primitive Quendian to Third Age Quenya


Most of the material on this page is taken from information organised and provided by Helge Fauskanger on his Ardalambion site. He studied and collected this material from J.R.R. Tolkien's manuscripts published by his son Christopher Tolkien and the staff of Vinyar Tengwar and the Elvish Linguistic Fellowship. My gratitude to all of them!


The use of these tables assumes familiarity with Quenya. I highly recommend studying first Helge Fauskanger's Quenya course. It also assumes having read (or being in the process of reading) his study on the Evolution of Primitive Elvish to Quenya, from which most of the data here has been extracted, yet without his excellent explanations, which are fundamental to understanding the matter fully. Another article I recommend is Helge Fauskanger's 'Primitive Elvish: where it all began'.


I have not marked unattested forms or non-Tolkienian reconstructions. Neither do I give references. Please consult above-mentioned 'Evolution...' for these very important details.

Short Chronology


One-hundred and forty-four Elves awoke at the mere of Cuivinen, somewhere to the East of the Blue Mountains (Q. Lnoronti, S. Ered Luin) and the Misty Mountains (S. Hithaeglir). In this first period of their existance, which lasted for some 520 solar years, they invented and consolidated their own language.

This period ends with the Separation, when the Three Tribes (the Vanyar lead by Ingw, the Noldor lead by Finw and the Lindar (or Teleri) lead by the brothers Elw and Olw) leave Cuivinen to follow the Vala Orom and travel westwards towards Aman, the abode of the Valar accross the Western Sea. They are know as the Eldar. Of the group of Elves that stayed behind, the Avari, we hear no more.


The Great March to Aman accross land and finally accross the Sea lasted some 270 solar years. Although the language of the Eldar was one, diversifications in dialects already existed. Notably, Lindarin (Telerin) showed peculiarities in pronounciation.

During this journey west, some elves changed their mind and decided to stay in Middle-Earth. The first group separated east of the Misty Mountains (Hithaeglir), and established themselves in Greenwood the Great (later called Mirkwood) and in the vale of the Anduin. These are the Silvan or Woodland Elves. Legolas, in the Lord of the Rings, belongs to this group of elves.

After crossing the Blue Mountains, at the river Gelion, the Nandor of the Lindar (Teleri) Tribe broke away and went south to settle in Ossiriand: the are called the Laiquendi (Green Elves).

Later more Lindar (Teleri) decided to abandon the march. Among them, Elw and his followers stayed and populated Beleriand, especially the woodland of Doriath. Elw adopted the name of Thingol. From him many noble elves and men descend: (the elf) Elrond of Rivendel and (the man) Aragorn are two illustrious ones. These elves are known as Sindar (Grey-elves), and from their Lindarin (Telerin) dialect arose Sindarin.

Another group of Lindar (Teleri) stayed near the shores of the Western Sea: the Falathrim (Elves of the Falas). Crdan was their lord.

Finally the Vanyar, Noldor and, sometime later, part of the Lindar (Teleri) reached Aman: they are known as the Calaquendi (Elves of the Light). This event foreshadows the end of the Common Eldarin period.


The languages of the Three Tribes in Aman slowly drifted apart. The Vanyar and the Noldor in Valinor evolved very similar dialects, Quendya and Quenya, while Telerin, the speech of the Lindar (now properly called Teleri) who arrived later in Tol Eressa, became practically a separate language. The focus on this page is on Quenya, the Noldorin speech.

This period can be subdivided into two sub-periods.


During this period Quenya was still not recorded. This period ends with the invention of the alphabet by Rmil.


This period extends from the invention of the Rmilian alphabet to the departure of the Noldor.

3500 years did the Noldor live in Aman, before they rebelled against the Valar and, following Fanor, left Valinor and returned to Middle-Earth. This happens in the last years of the Age of the Trees, just before the First Age.

In Middle-Earth, the Noldor adopted Sindarin as the spoken language, but did not forget their native Quenya. This period covers the First Age with the continued wars against Melkor and his final defeat; the Second Age, with Sauron and the forging of the Rings, the rise and fall of Nmenor, the Hiding of Valinor and the destruction of many geographical features on Middle-Earth (all lands to the west of the Blue Mountains were covered by the Sea, and the earth was reshaped round); the foundation of Gondor and the Northern Kingdom and the defeat of Sauron; the Third Age covers the second rise of Sauron, and it ends after the War of the Ring and the final defeat of Sauron, when all Elves returned to Valinor never to come back.

Primitive Quendian


The consonants in Primitive Quendian

In the following table, dark bold idicates Primitive Quendian original consonants; grey bold marks the early nasalised consonants in Primitive Quendian; the others consonants shown in the table will be developped at later stages. Alternative graphemes are shown in brackets.

The table shows the primary articulation, classified in (1) manner of articulation and voicing (columns) with aspiration (one column) being indicated separately, and (2) place of articulation (rows) naming the passive component (except in labio-dental, where 'labio', i.e. the lower lip, is active and 'dental', i.e. the upper row of teeth, is passive).

Secondary articulation (labialisation or palatalisation, occuring simultaneously with the primary articulation) is not shown in the table. The voiced semi-vowels w and j can change the quality of a preceding consonant, labialising (w) or palatalising (j) it. Except for kw, secondary articulation seems not to have happened yet in Primitive Quendian. Labialised kw is conventionally written at later stages as qu (in his early Qenya attempts, Tolkien wrote it as q).

The grapheme h can represent different phonemes and can be confusing. The ach-laut is ch, the ich-laut is hj. The phonemes f and v seem to have been originally bi-labials, later changing to labio-dental; the was dental, not inter-dental as in english 'thing'. The nasals mb, nd and g are shown twice in the table, but this does not mean there were two phonemes for each of them: they are simply nasalised voiced plosives. The grapheme is not the spanish ee (palatalised n), it is conventionally used by Tolkien to represent the ng sound as in sing, while g represents the ng sound in finger. The phonemes l and r are traditionally called liquids. The plosives with aspiration ph, th, kh are called aspirates.

manner of articulation plosives
nasals trills
laterals semi-vowels
voicing - - + - + - + - + - + - +
aspiration +
place of articulation
bi-labial p ph b, mb f v hm m, mb hw w
dental t th d, nd (dh) nd
hn n hr r hl l
sibilant s z
palatal hj (hy) (h) j (y)
velar k (c) kh g, g (ng) ch (h) , g (ng)
glottal h 3 (h)

Initial clusters in Primitive Quendian

mb, nd, g were nasalised plosives (so on their own they don't properly belong in this table). kw became labio-velar probably before the Separation (end of this 500-year period). hj is only mentioned once, was it a cluster or the ich-laut?

sk skj skw
sj sw
nd ndj
g gj gw
gj gw
kj kw

The vowels in Primitive Quendian

There are five short vowels and their corresponding long counterparts. From these are formed diphthongs in i and in u. The diphthong ou changed into au very early. The diphthong ei was lost in Quenya. ae and ao are treated as diphthongs at some time but are not found in Quenya.

short vowels   a e i o u
long vowels  
diphthongs in I   ai ei - oi ui
diphthongs in U au eu iu ou -
A diphthongs ae ao

Evolution of Primitive Quendian

See Helge Fauskanger's article 'Primitive Elvish: where it all began' for a fuller explanation.

The X represents a consonant, the o a vowel.

basic root pattern XoX Xo XoXoX
extended root pattern   XoXo and oXoX oXo, XoX and XoXo

The root is called sundo, the basic root vowel is the sundma, the vowel used to extend the root is called mataina or mataima. The vocalic extensions have the same quality as the sundma, but not necessarily the same quantity: for example, if a is the sundma, the vocalic extension may have a or .

The processes that occurred in Primitive Quendian can be summarised:

  • syncopation - the second of two identical short vowels, if unstressed, is dropped before a derivation.
    note 1 many words escaped syncopation
    note 2 syncopation of the first vowel occurs only in DARAK darak > d'rak > Q. rca "wolf"
    note 3 resulting final root j > i, w > u: TUJ, TUJU tujule > tujle > Q. tuil; LAW, LAWA laware > lawre > Q. laur;
    note 4 this is the first occurrence of syncopation, in a later period a second syncopation occurs.
  • elision of final vowels - some final vowels disapeared
  • plural - changes - the plural marker, -, was shortened following a long vowel (or maybe, any vowel)
  • diphthong from long vowel + i - short vowel + i was already a diphthong

The following processes may have occured at the end of this period, or at the begining of the Common Eldarin period:

  • homo-organic stop + nasal have metathesis
    pm > mp ?
    bm > mb e.g. labm > lamb > Q. lamba "(physical) tongue"
    tn > nt ?
    dn > nd e.g. adn > and "gate"
    k > k ?
    g > g ?
  • bn > mb sometimes, both forms coexist e.g. stabn and stamb Q. samb "room"
  • kw labialisation
    unitary use of kw, attested because it is not treated as a cluster
  • aspirate reduction before s
    phs > ps
    e.g. none
    ths > ts
    e.g. KHOTH khoths > khots "assembly" Q. ?
    khs > ks
    e.g. LOKH lokhs > loks Q. lox "hair"
  • aspirate reduction before t ?
    e.g.only example: RIKH or RIK rikt or rikht > rikt Q. rihta- "to jerk"
  • Telerin kw > p
    only in Lindarin (Telerin) dialect, this labialised velar unvoiced plosive becomes bi-labial unvoiced plosive

Common Eldarin


These changes were early ones, before the Nandor separated:

  • dentals d, th, t > s before t
    dt > st
    e.g. LED, LED led-t > lest Q. lesta- "to leave";
    tht > st
    e.g. KHOTH khoth-t > khost Q. hosta- "to collect";
    tt > st only if perceived to belong to different morphemes
    e.g. PUT put-t > pust Q. pusta- "to stop"
    but putt- (with medial fortification TT) > putt Q. putta "(punctuation) stop".
  • nasals m, > nasal dental-alveolar n before dentals d, t
    md > nd
    e.g. PLIM pilimd- > pilnd- Q. pilin (pilind-) "arrow"
    mt > nt
    e.g. KEM? kemtan > kentan Q. centano "potter"
    t > nt
    e.g. *WIN wit > wint Q. winta-, *vinta- "to scatter"
  • -w > -w (final)
    e.g. nidw > nidw Q. nirwa "cushion"
  • final -m > -m after dental
    -tm > -tm
    e.g. jatm > jatm Q. yanw "bridge"
    -dm > -dm
    e.g. no examples
    note 1 these will change to -tme, -dme, then to -nwe in Valinorean Q.
  • w > gw (medial)
    e.g. liwi > ligwi Q. lingw "fish"
  • first element of compound shortens final long vowel
    e.g. tuil + lind > tuilelind Q. tuililindo "spring-singer, swallow"
  • initial nasalised plosives may develop forms with nasal syllabic consonant
    mb- > mb-
    e.g. mbarta > mbarta Q. umbar (umbart-) "fate"
    nd- > nd-
    e.g. ndn > ndn Q. andn "sunset"
    g > g-
    e.g. golod > golod Q. Ingoldo "The oldo", but also Q. oldo
    e.g. gol > gol Q. ingol "deep lore, magic", but also Q. ol "lore, knowledge"
    gj- > dj-
    e.g. gj > dj Q. indyo "grandchild, descendant"
    gw- > gw-
    e.g. gwal > gwal Q. ungwal "torture"
    note 1 these will all develop an initial vowel in Valinorean Quenya: mb- > umb-, nd- > and-, g- > ig-, gw- > ugw-.
  • haplology
    e.g.kukw > kw, kuw Q. cu, cua "dove"; tuilelind > tuilind Q. tuilindo "springsinger, swallow"

At some time, a infixion before e and o, analogical to primitive a infixion before i and u, creates new dipthongs ae and ao:

  • ae e.g. MEL mael Q. mla "loving, affectionate" (diphthong ae later > long > )
  • oe e.g. KOL kaul Q. clo "burden" (diphthong oe later > long > )

Other changes, some after the separation of the Nandor.

Changes in consonant clusters:

  • aspirate reduction after another consonant
    Xph > Xp
    e.g. PHIR ?lphirin- > ?lpirin- Q. *ilpirin "inmortal" reformed to ilfirin by analogy to firin
    Xth > Xt
    e.g. none
    Xkh > Xk
    e.g. none
  • tkh > kk
    tkh (? > tk) > kk
    e.g. KHAYA *et-khair > Q. eccaira "remote, far"
    note 1 this may also imply tk > kk
    note 2 once metathesis tk > kt, et-kel "out-flow" > ektel Q. ehtel "spring, issue of water"
  • tk > kk or tk > kt ? - see previous entry
  • s voiced to z before voiced plosive - new voiced sibilant spirant z
    sd > zd
    e.g. esd > ezd "rest"
    sb > zb
    e.g. none
    sg > zg
    e.g. MASAG masg > mazg "pliant"
    note 1 these clusters (and gd, bd) will get unvoiced later after the shortening of final long vowels in polysyllabics, before the appearence of Rmil's alphabet: zd > st, zb > sp (not attested), zg > sk (and dg > kt, bd > pt [> ps])
  • d unvoiced before s
    ds > ts
    e.g. sjads > sjats "cleft, gash" Q. hyats
    note 1 what about plosives b, g: do they get unvoiced too before s?

Unvoicing and loss of glottal 3:

  • 3 unvoiced to ch before unvoiced consonant - new unvoiced velar spirant ch (ach-laut)
    3U > chU (U = unvoiced consonant)
    e.g. MA3 ma3t > macht Q. mahta "to handle"
    note 1 this unvoicing implies a shift in the articulation, from glottal to velar, the glottal unvoiced being h like in english 'his'
  • all other medial 3 lost
    e.g. ma3iti > maiti "handy" (Q mait)
    note 1 unvoicing of 3 to ch before unvoiced consonants had already occured, so this rule applies to the remaining medial glottal
  • loss of final glottal 3 - all cases
    but -u3 > -uw
    e.g. cu3 > cuw Q. c "bow"
    note 1 the loss of final 3 seems to have occured much later, not in this period!
  • initial 3- will change to h- sometime later, and may even be lost.

Loss of nasal :

  • loss of medial and final (except where it nasalises g, gw, k, kw or changed w > gw) - all cases
    e.g. te > te > Q. ta "indicates"
    e.g. pee > p Q. p "lip"
    note 1 initial - persisted. It was changed to n- in Exilic times

Effects of loss of 3 and :

  • effects on vowels brough in contact by loss of 3 or
    vowels remain in hyatus
    or form normal diphthongs
    short similar vowels > long vowel
    e.g. ee >
    e.g. pee > pe'e > p Q. p "lip"
    short + long similar vowel dissimilate
    e.g. e > i
    e.g. TE te > te' > ti Q. ti "path"
  • effect of a vowel and a consonant brought in contact by loss of 3 or
    short vowel + 3X
    short vowel + X
    > long vowel + X (X = consonant)
    e.g. do3m > dm Q. lm "twiligh, night"

Other final-vocalic changes:

  • loss of final short vowels -a, -e, -o
    loss of -a
    e.g. swanda > swand Q. hwan (hwand-) "sponge"
    loss of -e
    e.g. i-ndise > indis "bride"
    loss of -o
    e.g. ndro > ndr Q. nr "land"
    note 1 resulting final -l (from -la) becomes syllabic
    e.g. hekla > hek-l
    note 2 resulting monosyllabic word compensates by lengthening its vowel
    e.g. kwene pl. kwen > kwn pl. kwen Q. qun pl. queni "person"
    note 3 resulting final -w > -u
    e.g. angwa > angw > angu "snake"
    note 4 resulting final -j > -i
    e.g. none
    note 5 plurals were analogically reformed
    e.g. swanda pl. swandai > swand pl. swandi
  • short final -i > -e
    e.g. dmi > dme Q. lm "twilight"
    short final -u > -o
    e.g. angu > ango Q. ango (angu- pl. angwi) "snake"
    note 1 in forms of the word where these vowels where not final (e.g. with case endings) these vowels were preserved
    note 2 later stages of Common Eldarin

Pre-Record Quenya


Very early stage:

  • mb-, nd-, g- are still valid (initial nasalised voiced plosives)
    e.g. Mbelekro "Melkor"
    note 1 they will drop the voiced plosive in Valinorean Quenya: mb- > m-, nd- > n-, g- > -.
  • long final vowel in compounds (except plural -) are shortened
    e.g. Eru-khn > Erukhna Q. Eruhin (Eruhn-) "Child of Eru"
    note 1 this leads, in later stages, to the total loss of the final vowel, see below.
    note 2 the first element of a compound had already shortened its final vowel.
    note 3 many compounds were formed after this rule became inoperational
    e.g. Elen-tr Q. Elentri "Star-queen" was not reduced to **Elentri > **Elentar, because it was compounded later.
  • some cases of metathesis -
    dl > ld not infrequent: Edl > Eld


  • occasional loss of final short vowel in tri- (or more) syllabics
    e.g. Mbelekro > Mbelekr "Melkor"
  • regular loss of final short(ened) vowel in compounds
    e.g. (Eru-khn >) Erukhna > Erukhn Q. Eruhin (Eruhn-) "Child of Eru"
  • effect of loss of final short vowels:
    -w > -o i.e. resulting final -w > -u > -o
    -j > -e i.e. resulting final -j > -i > -e
    e.g. talrunja > talrunj > talruni Q. tallun "sole of foot"
    note 1 this is essentially the same rule that was operative in Common Eldarin.
    note 2 after this stage, this rule becomes inoperative

The softening of Quenya has begun

Loss of initial s-clusters, with introduction of new phonemes: hl (unvoiced l), hr (unvoiced r), hw (unvoiced w) and hj (hy) (ich-laut):

  • initial s + unvoiced plosives p, t, k > aspirates ph, th, kh (there are no cases attested of initial s + voiced plosives). These will later become spirants:
    sp- > ph-
    e.g. spang > phang Q. fanga "beard"
    st- > th-
    e.g. stank > thank Q anca > sanca "split"
    sk- > kh-
    e.g. skarw > kharw Q. charw > harw "wound";
    skj- > khj-
    e.g. skjapat- > khjapat Q. hyapat "shore"
    note 1 this is not directly attested by Tolkien for Quenya, only for Sindarin
    note 2 Quenya and Sindarin have the same parallel development, Amanya Telerin does not: it retains all initial s-clusters
  • initial s + liquids l, r unvoices them > hl, hr - later, in 3rd Age, these are pronounced voiced. In Etym (1930's) the s was simply lost without any intermediate step.
    sl- > hl-
    e.g. slk > hlk Q. hlc "dragon" exilic Q. lc
    note not all hl- were voiced to l-: hlapu- "blow", hlar- "hear", hloni "sounds" remain unvoiced if attested forms are correct.
    sr- > hr-
    e.g. srw > hrw Q. hrv "flesh"
    e.g. sraw > hraw > Q. hroa "body"
    note mi-srawanwe Q. mirroanwe "Incarnate" shows that in medial position, -sr- > -rr-
  • initial s + nasals m, n drop the s
    sm- > m-
    e.g. smalin > malin Q. malina "yellow"
    sn- > n-
    e.g. snard > nard Q. narda "knot"
  • initial sj > hj (ich-laut)
    sj- > hj-
    e.g. sjalm > hjalm Q. hyalma "shell, conch"
  • initial sw > hw
    sw- > hw-
    e.g. swanda > swand > hwand Q. hwan (hwand-) "sponge"

Transformation of other initial clusters and phonemes:

  • initial unvoiced velar k + j > palatalised dental tj
    kj- > tj
    e.g. kjelep > tjelep Q. tyelp "silver"
  • initial gj, dj, gj reduced to semi-vowel j
    gj- > j-
    e.g. gjern > jern Q. yerna "old, worn"
    gj- > j-
    e.g. gjalm > jalm Q. yalm "clamour"
    but gj > dj Q. indyo "grandchild, descendant"
    dj- > j-
    e.g. djelt > jelt Q. yelta "to loathe"
  • initial aspirate ph + u > hwu
    phu- > hwu-
    e.g. phuin > hwuin Q. huin "gloom, deep shadow"
    note 1 hwu- becomes hu- in Valinorean Quenya.
    note 2 (VT41:8 n.1) there was a tendecy to slacken f to hw, both being bilabials. This was later checked in Noldorin when f became labio-dental.
  • initial khj remains
    note 1 later khj > chj when spirant ch was introduced in Valinorean Quenya

The stress retraction period, followed by syncopation:

  • the stress moved to the first syllable
    but verbal augment and reduplication never carry the stress, even in this period.
    note 1 this process antecedes syncopation
  • syncopation (2nd occurrence) - where a word of three or more syllables commences in two concomitant syllables that contain identical short vowels, the second of these vowels is normally syncopated (cf. Letters:426), unless there is a consonant cluster before or after the second vowel:
    e.g. galad > gald Q. alda "tree", but aklar Q. alcar "glory" and kjelepn Q. t(y)elemna "of silver"
    note 1 kw had already become labialised by this time, for it is not considered a cluster: salakwe > salkwe Q. salqu would not have occurred. (this is reported in PQ period)
    note 2 words that shortened their second vowel after this rule ceased to operate, do not syncopate: terw > terew Q. terva "fine, accute"

Effect of syncopation on j, w, and vowels

  • j before consonant > i and can form diphthongs
    jX > iX
    e.g. gajak > gajk > gaik Q. aica "fell, terrible, dire"
  • w before consonant > u and can form diphthongs
    wX > uX
    e.g. tawarin > tawrin > taurin Q. taurina "of wood"

Changes undergone by phoneme d:

  • initial d- > l- (regular)
    e.g. dme > lme Q. lm "twilight, night"
  • initial d- > n (seldom)
    e.g. dr > nr Q. nr "man" (analogical to n)
    e.g. doron > dorn > norn Q. norno "oak" (by n assimilation)
    but only exception: d'rka > rka Q. rca "wolf"
  • medial intervocalic d may > l by assimilation with another l
    e.g. g-lada- > glala- Q. lala- "to laugh"
  • d > l may cause dissimilation of further l in the word
    e.g. DAL dl > ll > ld Q. lra "flat" by dissimilation!
  • medial dj > lj
    e.g. DEL (> LED) ledj Q. lelya- "to go"
    note 1 DEL del- > lel- (regular) > led- (dissimilation)
  • post-vocalic d >
    e.g. this would have happened: DEL edel > edl > *el > *ello
    except that metathesis had occured: DEL edel > edl > eld Q. Eldo arch. variant of Elda "Elf"
    (see below)
    note 1 this will later become z, then r
Last chance for original glottal 3:
  • initial glottal 3 > h - this h is the 'english' unvoiced glottal h in 'his'
    note 1 Tokien changes his mind constanly about this initial 3, sometimes assuming it has always been h. See Etym 3AR.
    note 2 this initial 3 > h may be lost:
    e.g. 3r > (hr >) hr Q. r "inner counsel (heart)" (VT41:11-15)

Spirantisation process continues. Appearence of the new spirants: (unvoiced) f, , ch, (voiced) v, (dh), 3 (re-introduced). Plosives change to spirants in certain positions:

  • initial and post-vocalic voiced plosives b, d, g > voiced spirants v, , 3
    initial b > v
    e.g. bes > *ves Q. vezu, veru "married pair"
    post-vocalic b > v
    e.g. laib > laiv Q. laiv "ointment"
    lb > lv
    e.g. Q. elva "starlike"
    post-vocalic d >
    e.g. ta+da (allative) > tad > ta Q. taz, tar "thither"
    e.g. mi+da (allative) > mid > mi Q. miz, mir "to the inside, into"
    note 1 later > z > (Exilic Q.) r
    initial g > 3
    e.g. galad > gald > 3ald Q. alda "tree"
    post-vocalic g > 3
    e.g. weg > we3 Q. vi "manhood"
    lg > l3
    e.g. ulgund > ul3und Q. ulundo "monster"
    note 2 all 3 are lost in post-Rmilian period
  • unvoiced aspirates ph, th, kh > unvoiced spirants f, , ch - also from early sp-, st-, sk-
    ph > f
    e.g. phini > phine > fine Q. fin "a single hair"
    th >
    e.g. thindi > thinde > inde Q. ind, sind "grey"
    kh > ch
    e.g. khithm > chim Q. chiw, hisw "fog"
  • initial palatalised and labialised clusters - note hj- (ich-laut) was already < sj-, and hw- (unvoiced w) was already < sw, phu
    khj- > chj- > hj-
    e.g. khjeles > chjeles > hjeles Q. hyell "glass"
    skj- > khj > chj- > hj-
    e.g. skjapat- > khjapat > chjapat > hjapat Q. hyapat "shore"
    khw- > chw- > hw-
    e.g. skwarm > khwarm > chwarm > hwarm Q. hwarma "crossbar"

The shortening of final vowels in polysyllabic words happened near the end of this period.

  • the change -i > -e and -u > -o (final short vowels) ceased to operate
  • the introduction of spirants has already occurred
  • long final vowels are shortened in polysyllabics - also analogically in inflected forms, except accusative
    note 1 the sequence of events indicates that the plosive > spirant change has already occurred, but Etym shows cases of preservation of plosives and shortening. This may be an inconsistency.
    note 2 the long accusative presents a problem. The old accusative marker -h (-3) is supposed to have disappeared earlier, like all final -3.

Devoicing of medial clusters:

  • zb > sp e.g. no examples
  • zd > st e.g. esd > ezd > ezde > est Q. Est (a Vali) "rest"
  • zg > sk e.g. masg > mazg > mazga > maska Q. maxa "soft"
  • bd > pt e.g. libd > libda > lipta Q.lipsa "soap"
  • gd > kt e.g. khagd > chagda > chakta Q. chahta > hahta "mound"

Other changes to clusters:

  • pt > ps
    e.g. (lepet >) lepte > Q leps "finger"
    note 1 probably dropped by Tolkien later, some forms retain pt
  • hwu > hu
    e.g. (phuin > hwuin >) hwuine > huine Q. huin "gloom, deep shadow"
  • simplification initial mb, nd, g to full nasals, (except when they had developped syllabic nasals, for which new vowels appear):
    mb- > m- e.g. mband > mbando > mando Q. mando "custody"
    nd- > n- e.g. ndeur > ndeuro > neuro Q. neuro "successor"
    g- > - e.g. golod > gold > goldo > oldo Q. oldo "oldo"
    sometimes medially in compounds, if the word is also used as simplex
    e.g. bal-ndre > Valinor (not Valindor)

Syllabic consonants expanded:

  • mb- > umb- e.g. MBARAT mbarta [final vowel uncertain] > mbart > umbart Q. umbar (umbart-) "fate, doom"; another theory is the - is a prefix, cf. Sindarin name marth "Ill Fate"
  • mb- > amb- e.g. MBARAT mbarta [final vowel uncertain] > mbart > ambart Q. ambar (ambart-) "doom"
  • mb- > amb- e.g. MBAR mbar > mbar > ambar Q. ambar "world"
  • nd- > and- e.g. ndn > ndne > andne Q. andn "sunset"
  • nd- > ind- e.g. NDIS ndis > ndise Q. Indis "bride, a name of Nessa"
  • g- > ig- e.g. GOLOD golod > golod > gold > goldo > igoldo Q. Ingoldo "The oldo"
  • dj- > idj- e.g. GY, GYON gj > dj Q. indyo "grandchild"
  • gw- > ugw- e.g. gwal > wgwale > uwgwale Q. ungwal "torture"
  • l- > il- e.g. laphirin > lphirin Q. ilfirin "immortal"
  • -l > -il
    e.g. makla > makl Q. macil "sword"
    e.g. ? sirla > sirl (stopping rl > ll assimilation) Q. siril "rivulet"
  • others:
    e.g. kwentr > kwentr Q. quentaro "narrator"
    e.g. tamr > tambr Q. tambaro "knocker, woodpecker"
    e.g. istro > istr Q. istar "wizard"

Changes in medial consonant clusters:

  • lr > ll e.g. talrunja > talrune Q. tallun "sole of foot"
  • nl > ll e.g. nenl > nenle Q. nell "brook"
  • ls > ll e.g. khjeles > chjeles > hjeles > hjels > hyelse Q hyell "glass"
  • rs > ss e.g. neres > Nersa Q. Nessa "Nessa" (name of a Vali)
  • sm > mm e.g. asmal > asmale Q. ammal "yellow bird"
    but there are exceptions
  • sr > rr e.g. mi-srawanwe > mirrawanwe Q. mirroanw "incarnate"
  • r > e.g. neth-r > ner > nea Q. nessa "young"
  • l > ll e.g. *elo > *ello
  • r > rr e.g. no examples, but confirmed by Tolkien
  • ln > ld e.g. chalna > chalda Q. halda "hidden"
    ln > ll occasionally this may be due to infixion in past tense, or metathesis: cf nl.
  • nm > nw e.g. tinme Q. tinw "sparkle, glint; star"
  • m > w e.g. chime Q. chiw, hisw "fog"
  • sk remains e.g. Q. rusca "red-brown"
    sk > ks metathesis Etym e.g. miska Q. mixa "wet"
  • rg > rj e.g. targa Q. tarya "stiff"
  • lg > lj deprecated
  • kt > cht Q. spelling 'ht'
    e.g. ektele Q. ehtel "spring"
    note 1 caused blending of kt and cht roots
    note 2 echt > ehjt, icht > ihjt sometime later... the Q. spelling is 'ht' anyway
  • pl > lp ? (VT42:31 n. 58)
  • bl should have become vl (now > ?), or supposing it had methasesis (not mentioned) it would be lv ?
  • tl > lt ? (VT42:31 n. 58)
  • dl had become l (now > ll)
  • kl > lk metathesis e.g. aklare Q. alcar, alcar "glory" cf. (VT42:31 n. 58)
  • gl had become 3l (I suppose this > l)
  • pr > rp ? (VT42:31 n.58)
  • br should have become vr (now > ?), or supposing it had methasesis (not mentioned) it would be rb ?
  • tr > rt metathesis e.g. netre Q. nert "nine"
  • dr had become r (now [> zr] > rr) e.g. njadr Q. nyarro "rat"
  • kr > rk metathesis e.g. lakra Q. larca "swift"
  • gr had become 3r (I suppose this > r)
  • pm > mp (VT42:24) (metathesis)
  • bm had metathesised into mb in PQ/CE
  • tm > nw e.g. maratme Q. maranw "destiny"
  • dm had become m (now > ?; if left so, m > zm > rm)
  • km > ngw e.g. nakma Q. nangwa "jaw"
  • gm should have become 3m but
    gm > ngw e.g. sagma Q. sangwa "poison"
  • pn > mn e.g. Utupno Q. Utumno "Melkor's first stronghold"
    pn > mp (VT42:24) e.g. lepene Q. lempe "five" (metathesis)
  • bn had metathesised to mb in PQ/CE sometimes e.g. stabn > stamb Q. samb; lebn Q. lemba "left behind"
    or bn should have become vn but
    bn > mn e.g. abno Q. amno, samno "carpenter"
  • tn > nt (VT42:24) (metathesis)
  • dn had metathesised into nd in PQ/CE
  • kn > k (VT42:24) e.g. stakn Q. anca "cleft, split" (metathesis)
  • gn should have become 3n or ...?
    gn > g ? (metathesis?)

The s in voiced surroundings:

  • vowel + s + voiced consonant or unstressed vowel: s > z (including sj but excluding sw)
    e.g. ausa Q. auza, saura "foul"
    but -sanwe Q. sanw "union of 2 thoughts, telepathy"
    but asely "with you", as prep. "with" + -e-ly "you" (VT43:29)
    but tanya-salpe "Bowl of Fire". (LT1:266) salpe would be related to salpa- "lick up, sup, sip" [SLAP]
    but tarma-sundar "Roots of the Pillar". sundar would be related to sundo "base, root, root-word" [SUD]
    note 1 VT44:20 note 4: (1960s) this change did not occur when s was followed by a stressed vowel
    note 2 in Exilic Q. z > r
  • final post-vocalic s
    final post-vocalic s > z
    e.g. (Etym) LOS olor "dream"
    final post-vocalic + s remains s
    e.g. (later) LOS olos (olor-) "dream"
    note 1 Tolkien changed his mind in favour of the restricting rule
    note 2 in Exilic Q. z > r

Changes in consonants:

  • w >
    e.g. w-mentie Q. menti "meeting (of two)"
  • > z
    e.g. niwa Q. nizwa, nirwa "cushion"
    note 1 in Exilic Q. z > r
  • non initial ti > si - Tolkien changed his mind later
    e.g. mait pl maisi, later changed to maiti

Changes in vowels:

  • -ina > -in (adjectival)
    e.g. firina > firin "dead"
    note 1 some do, most don't, some forms coexist
  • diphthong ae > long > (ae disappears)
    e.g. maela Q. mla "loving"
  • diphthong ao > long > (ao disappears)
    e.g. caolo Q. clo "burden"
    note 1 Book Q. genitive for words in -a is - < -ao
  • ei >
    e.g. neite Q. nt "moist"
  • ii > i
    e.g. awwii > awwiie > awwie (old prefect "has passed away" replaced by avni)
Final consonants and clusters:
  • -nt remaind in dative dual
  • -cht > -t (cht is spelled ht)
    e.g. jacht (yaht) Q. yat (yaht-) "neck"
  • -ks > -s
    e.g. peltaks Q. peltas (peltax-) "pivot"
  • -Xt > -X
    e.g. Umbart Q. Umbar (Umbart-) "fate"
    e.g. Mandost Q. Mandos (Mandost-) "Mandos"
  • -Xd > -X
    e.g. Q. hwan (hwand-) "sponge"
    e.g. Q. Taniquetil (Taniquetild-)
  • -g > -n
    e.g. Q. Laurelin (Laureling-)
  • -double > -single
    e.g. Q. tol (toll-) "island"
  • -m > -n
    e.g. Q. talan (talam-)
  • -k > -t
    e.g. Q. filit (filic-)
  • final vowel + w drops the w
    -aw > -a ?
    -ew > -e ?
    -iw > -i ?
    -ow > -o ?
    -uw > -
    e.g. KU3 > kuw Q. c "bow"
    note 1 the loss of final -3 was reported in Common Eldarin, but may have happened now

Valinorean Quenya

  • ay, ai > e before unaccented vowel ay > ai > e
    e.g. daj > laj > layo, laio Q. lo "shade"
    e.g. gajr > 3ajr > ayr, ayr Q. r "sea"
    e.g. phaj > faya, faia Q. fa "spirit"
    note 1 Fayanra and Fanro coexist in Feanor's time
  • y > ai when previous rule ceased to operate
    e.g. Mja > Maia
  • aw > o before unaccented vowel aw > au > o
    e.g. sraw > hrawa > hraua > hroa "body"
    but TAWAR tawar Q. tavar "wood"
End of the retraction period. New stress rules:
  1. stress moves to penultima syllable if long, otherwise to ante-penultima
  2. three syllable words usually shorten the vowel in the second syllable to keep the stress on the first:
    e.g. arta Q. arata "exalted"
    e.g. gl > gl > gle Q. ingol "philosophy"
    but ndn > ndn > ndne Q. andn "sunset, west"
  3. long vowel following stressed syllable is shortened
  4. long vowel preceding stressed syllable is usually shortened
    e.g. w-nn > -nna Q onna "twin-born"
    e.g. Q. man "vowel" pl. omandi
  5. four or more syllable words ending in 3 short syllables lengthen the penultima to attract the stress:
    a noun of more than three syllables cannot end in three short syllables
    e.g. Q. Eldali "Elf-folk" possesive Eldaliva
    e.g. Q. huin "shadow" possessive huiniva diphthong ui is treated as hiatus in this rule
    e.g. Q. Cuivinen = cuivi "awakening" + nn, -nen "water"
    but Q. Altariel from T. Altriell (should be altariel = alat > alta "shine by reflection" + riel "wreath, garland") S. Galadriel
    but Q. Anrion "Sun-son" anar, anr + -ion "son"
    but Q. ancalima "brightest" (an-calima)
    note 1 some compounds are exempted
  6. long vowels are shortened before a consonant cluster
    long vowels are not allowed before a consonant cluster
    e.g. Mnwe > Manw
    note 1 cluster with -j- (-rj- etc) may be exempted: mryat "her 2 hands"
  • ij + vowel > i + vowel
    e.g. oiya Q. oia "ever"
  • uw, w + vowel > u + vowel - the resulting u- is short
    e.g. kwa > kua Q. cua "dove"
  • long , + vowel > short u, i + vowel
  • post-vocalic w > v except -ow- and -aiw-
    e.g. hrwe Q. hrv "flesh"
    but Q. n "conception" pl nwi
    but Q. aiw "bird"
    but jagw > jagwe > ya3we Q. yw "ravine" (not yv, cf. note 1)
    note 1 this rule was temporary, and ceased to operate before the loss of glottal 3
    note 2 initial w > v is Exilic Q. properly
    note 3 possessive pl. -aiva is analogical to sg. -ava (-a nouns)
The fate of glottal spirant 3. After the first disappearence, it was re-introduced in Pre-Record Quenya: 3 < g. It was recorded by Rmil. Then it disappeared again, after the post-vocalic w > v changes became inoperational.
  • initial 3 + consonant may develop anticipatory vowel - the vowel has the same quality as the one following the cluster
    e.g. grauc- ?grauc/graucu > 3rauc- > 3arauc- Q. arauco, rauco "terrible creature"
    note 1 later in this period, all 3 are lost
  • loss of 3
    final e.g. Denwego > Denweg > Lenwe3 Q. Lenw
    medial intervocalic e.g. weg > we3 > we3o Q. wo, vo "man"
    medial -l3- e.g. ulgund > ul3undo > ulundo "monster"
    medial post-vocalic 3 + consonant has compensation e.g. magr > magra > ma3ra Q. mra "good"
    initial e.g. gass > 3ass > 3assa Q. assa "opening"
  • effect on vowels brought in contact
    • long vowel contradictory statements:
      Etym long vowel is shortened
      e.g. tgu > tgo t3o Q. tuo "muscle" (not to)
      e.g. rg > rge > r3e Q. ri "crown" (not r)
      PM:347 long vowel remains
      e.g. rg > rga > r3a > ra "wreath" (not ria)
    • identical vowels are dissimilated
      oo > uo e.g. Kh-gor > Ch3ore > Cho'ore (shortened o) Q. (dissimilation) Chuor, Huor (masc. name) "heart-vigour"
      ee > ie e.g. weg > we3 > we3e > we'e Q. wi, vi "manhood"
      but ugu- > u3u- Q. - "negative prefix"
  • bi-labial spirant f becomes labio dental - first in Telerin, then in Quenya (but not in Vanyarin Quendya, who are accused of confusing f and hw)
    e.g. try this: pretend you are going to make the sound for B, but instead of saying B, just blow through your tight lips... this is the old bi-labial F, the way it was pronounced before the change. It changed to 'our' F, which is labio-dental.
  • secondary syncope in long compounds
    e.g. Artaher "noble lord" = arata-her
    e.g. Ambarto (mother-name of Pityafinw) "upwards exalted" = amba-arato
  • ndy > ny - only in the speech of the Noldor
    e.g. Quendya > Quenya
  • the change > s begins at the end of this period in Valinor - only in Noldorin Q.
  • the change z > r begins at the end of this period in Valinor - only in Noldorin Q.

Exilic Quenya


In Feanor's times in Valinor, the following change started to occur among the Noldor: > s in all positions. The change was completed in the Exile. The tengwa l (exilic sl) was still used to represent the former .

evolution Exilic Quenya
original s
remains s or turns to r, or is lost in some clusters
Tulkatho > Tulkath
th >
PRQ spirantisation period:

> s

tasar "willow-tree"
sanca "split"

st- > th-
PRQ loss of initial clusters:
thank > thanka

Already in Valinor, just before the Exile, z > r in all cases (there is one exception) in the speech of the Noldor. The tengwa z ceased to be used for the former z; it was renamed ess and used to represent ss.

evolution Exilic Quenya
original r
and r in clusters:
sr > hr, rs > ss, sr > rr, r > , r > rr, kr > rk, tr > rt, rg > ry
postvocalic d

mi-da > mid
awadelo > Audel
nidw > nidw > nidwa
this d >
PRQ spirantisation period:
> z
PRQ late:
z > r
mir "into"
nirwa "cushion"
postvocalic s followed by voiced
except sw, and s followed by stressed vowel
thaus > aus > ausa
besn > vesn > vesno
pisj > pisya
this s > z
PRQ after spirantisation period:
z > r
saura "foul"
verno "husband"
pirya "syrup"
but z (from any origin) > s,
if same or ajacent syllable contains an r

Dwarfish Khazd > exilic loanword *Cazar > Casar "Dwarf"

The only final consonants allowed after all these changes are these five:

unvoiced voiced
nasal    trill     lateral 
t s n r l

The h story...

The inheritance of the voiced glottal spirant 3 is complex. Initially it was either always the unvoiced glottal spirant h (as in 'his'), or turned to h before the softening of the language and the introduction of the new labial and dental spirants, merging with original h; this initial h may get lost: hr or r. Medially it was unvoiced to velar spirant ch before an unvoiced consonant or was lost altogether (in the late Commond Eldarin period). Finally it may have remaind until after the long final vowels were shortened (to explain the Book Quenya long-vowelled accusative). It was re-introduced in the softening period to replace the g in initial, post-vocalic position, or after l, only to disappear completely after the post-vocalic w > v changes ceased to operate in recorded (Valinorean) Quenya.

The velar aspirate kh is also doomed. Original or reinforced by the initial cluster change sk- > kh-, it was finally promoted to unvoiced spirant ch, merging with ch < 3 (before unvoiced consonant). Initial chj- then > ichlaut hy- (khw somehow became unvoiced hw). In the Exile (or earlier) it seems ch > h in all positions except possibly before t: acht, ocht and ucht (spelled aht, oht, uht) represents the ch ach-laut sound, whereas echt, icht (spelled eht, iht) represent the ich-laut (as does the initial hy- combination).

In the Exilic period, consequently, we are left with one h grapheme representing: (1) the 'unvoiced' palatal semi-vowel hy called ich-laut, spelled hy- initially and h in eht, iht; (2) the unvoiced velar spirant ch (spelled h) called ach-laut in the combinations aht, oht, uht; and (3) the unvoiced glottal spirant h in other positions. The 3 and kh have vanished altogether.

Ze w vants to leave...

Voiced bi-labial semi-vowel w has a lot to do: it can labialise a preceeding consonant, turn to u before a consonant or after u, or even be created when clusters were softened... and much more. In the years of classical Valinorean Quenya, uw/w + vowel > uw, surviving postvocalic w > v, except ow and aiw. Final w has disappeared even earlier. Initial w will also tend to change to v, possibly completing this process in Exile in the third age. The v, a labial (later labio-dental) voiced spirant introduced in the softening period, ends up absorbing b's (initial, post-vocalic and after l) and w's.

Initial > n, and the labialised w > nw (labialised n):

  • - > n-
    e.g. oldo (ngoldo) > noldo (a tengwa)
  • w- > nw-
    e.g. walm (ngwalm) > nwalm (a tengwa)

Polysyllabics see the following changes occur in their final syllable:

  • ending -ai > long > -
    e.g. linta pl. lintai > pl. lint "swift"
  • -eai > -ee > -i (dissimilation)
    e.g. laura pl. laurai > pl. lauri "golden"
  • all final long vowels are shortened - including the accusative
    e.g. ciry > cirya (accusative) "ship"
  • long vowels in final closed syllables may be shortened
    e.g. Anr > Anar "Sun"
  • but three(or more)-syllabics may retain this long vowel when endings are added
    e.g. Eruhn > Eruhin pl. Eruhni "Child of Eru"

Other changes:

  • unvoiced hl, hr are pronounced as voiced l, r
  • a glide y may develop between i and another vowel
    e.g. aia, aiya "hail!"
  • consonant assimilation in groups starting with n:
    nl > ll
    e.g. Nmen "West" + lt "flower" = Nmellt "Flower of the West"
    nm > mm
    e.g. elen "star" + macil "sword" = Elemmacil *"Star-sword"
    nr > rr
    e.g. elen "star" + rna "crowned" = Elerrna "Star-crowned"
    ns > ss
    e.g. elen "stone" + sar "stone" = Elessar "Star-stone" or "Elf-stone"
  • rl > ll sometimes
    e.g. Casarli > Casalli "dwarves" (partitive plural)


Back to Quenya...

Alex Grigny de Castro
March 2002