The Goldogrin Past Tense

by Patrick H. Wynne


The Gnomish Lexicon (GL) records over 230 past tense forms, many of which Tolkien labeled pret., i.e., preterite, a term that is synonymous with past tense (from Latin praeteritus, past participle of praeterire to go or pass by; praeter beyond, by + ire to go). The term past tense appears only once in GL, in the entry for gî(laith) once, in past, which notes that the particle is used loosely as a mere ‘aid’ to a past tense. (For more on this particle, see the Appendix at the end of this article.)

Goldogrin, like its successors Noldorin and Sindarin, has both strong and weak preterites.1 In brief, strong preterites in Goldogrin are characterized by either 1) lengthening of the stem vowel (which generally also results in a change in vowel quality or diphthongization), e.g., bab- drink, quaff, pret. bôbi (with the usual Goldogrin development of ā > ō); or 2) nasal infixion of the verbal stem, e.g., tag- fix, make firm, construct, pret. tanci (*ta‑n‑k- < root TAKA fix, fasten in QL). Both vowel lengthening (the most common type of preterite formation in Goldogrin) and nasal infixion are also accompanied by addition of the suffix ‑i. Weak preterites in Goldogrin are formed simply by the addition of a suffix (‑i, ‑thi, or ‑ni) to the stem, which remains otherwise unchanged,2 e.g., celu- trickle, pret. celwi-; ol- appear, seem, pret. olthi; rûtha dwell, remain, pret. rûthani.

It is not uncommon for a Goldogrin verb to have two preterite forms. Sometimes both are strong, e.g., nag chew, gnaw, pret. nôgi or nanci; and sometimes both are weak, e.g., palta- beat, pret. palti or paltathi. Sometimes one is strong and the other is weak, e.g., sana- can, know how to, pret. sôni (strong) or santhi (weak). A few verbs even have three preterite forms, e.g., sô- wash clean, bathe has one weak preterite (sôthi), one strong (sûvi), and a mixed form combining both strong and weak characteristics (sûthi). It is possible that the order in which Tolkien lists multiple preterite forms is significant, i.e., the first form might be the more common. This is suggested by the verb tur- (turthi, tauri) can, have power to — a separate entry for pret. tauri states that turthi is more common.

The lists of verbs and their preterites in this article have been standardized using the formula verb (preterite) translation, followed by notes, if any, placed in square brackets (these usually pertain to emendations judged to be of morphological interest, and have also been standardized so that the final form is given first); e.g., the entry that appears in GL as caltha- wax. grow. flourish. côli. [cal- >> caltha-.] is given here as “caltha- (côli) wax, grow, flourish. [caltha- << cal-.]”

STRONG I — Root vowel lengthening

The lengthening of the root vowel of a verb-stem as a means of forming the preterite probably dates back to the Eldarin period,3 since many Goldogrin preterites of this type have clear cognates in Qenya. In the following examples, the Qenya forms are taken from QL:

  • Gn. lav- lick, pret. lôvi — Q lava- lick, pret. lāve-.
  • Gn. mel- love, pret. mîli — Q mel- to love, pret. mēle.
  • Gn. lor- sleep deep; dream, pret. lûri — Q lor- slumber, pret. lōre.
  • Gn. gwil-(1) sail, float, fly, pret. gwaili — Q ’wili- sail, fly, float, pret. wīle-.
  • Gn. tur- can, have power to, pret. tauri — Q turu- can, is able, pret. tūre-.

Note that the root vowels a, e, o, i, u in the examples cited above have undergone a change in quality as well as quantity in the preterite, or else have become diphthongs, these changes being due to the historical development of primitive ā, ē, ō, ī, ū > ô, î, û, ai, au in stressed syllables4 in Goldogrin. Also note that a suffix ‑i is added to these preterites, equivalent to Q ‑e. The etymology of this preterite suffix appears in GL s.v. gôtha- (gui) possess, have, hold. According to this entry, gôtha- derives from ı̯otta, and the pret. gui from ı̯ōı̯ē. The latter pair of forms shows that Goldogrin pret. ‑i derives from ‑ı̯ē. This closely resembles the etymology given in the chart of Qenya verb endings associated with the Qenya Lexicon, in which the Qenya impersonal singular pa.t. ending ‑e is from ı̯ĕ (with short final ĕ), and the neuter singular pa.t. ending ‑ie is from ı̯ē < ‑ı̯ĕ + ə (PE14:31).

The change of stressed ū > au in Goldogrin was not universal, however, and beside the Strong I verbs in which the root vowel ‑u- becomes ‑au- in the preterite (e.g., cug- ponder, pret. caugi) there is another group of Strong I verbs in which the root vowel ‑u- becomes ‑ū- in the preterite (e.g., cum- lie (down), pret. cûmi). All the verb-stems in the latter group end in a labial consonant, either the bilabials b, m (tub-, cum-, gum-, hum-, mum-, num-) or the labiodentals f, v (nuf-, ruf-, luv-).5 It is apparent from this that original ū resisted diphthongization in Goldogrin when it was followed by a labial consonant.6 Two of these verbs have alternate (and probably analogical) preterite forms in ‑au-: gum- to lade, pret. gûmi or gaumi; and ruf tear, pret. raufi or rûfi.

STRONG I-A — Root vowel lengthening in basic verbs

The majority of Strong I preterites in Goldogrin belong to basic verbs (Strong I-A), i.e., verbs that are directly formed from a root without the addition of a derivative suffix, such as nag chew, gnaw < NAKA bite, cwel- fade, wither < QELE- perish, die, decay, fail, nor- run, roll < NORO run, ride, spin, gwidh- weave < GWIĐI- *weave, and tul- bring; come to < TULU- fetch, bear, bring; move, come.7

Goldogrin and Qenya also possessed a number of sonant verbs, i.e., verbs from roots with the vowels , , (e.g., KḶKḶ gather, SṆTṆ esteem, and MṚĐṚ grind in QL). These were all evidently Strong I-A verbs in origin, though this is not readily apparent in the descendant languages, primarily due to different developments of short , , versus long , ṇ̄, . For example, cwintha- (cwanthi) to fill seems to form its preterite by qualitative ablaut (substitution of one stem vowel for another), but it actually derives from vocalic lengthening of the vowel in the root QṆTṆ: present stem *qṇt- > cwinth-, Strong I-A preterite stem *qṇ̄t- > cwanth-. Given the complexity of the phonological and analogical processes involved in the sonant verbs, they are presented in a separate section below, following the discussion of the weak verb classes.

Root-vowel A

  • bab- (bôbi) drink, quaff.
  • bag- (bōgi) sell, trade.
  • bas- (bôsi) bake.
  • brath- (brôthi) cook (tr.)
  • caf- (côfi) to taste.
  • cwas- (cwôsi) wag, shake, flap (tr.)
  • daf- (dôfi) strip, flay, peel skin.
  • dala- (dôli) sing or ring.
  • drab- (drôbi) tr. to labour, work; impersonal, to irk, annoy.
  • fab- (fobi) handle, touch, feel.
  • faf- (fofi) puff, blow, pant.
  • far- (fori) separate, sever, divide (intr.); depart, leave (c. gen.)
  • fas- (fôsi) wash.
  • fau (fui(2)) impersonal it smells, gives off a smell.
  • gada- (gôdi, archaic †ganthi) join, connect, unite. [gada- << gad-. pret. gôdi archaic ganthi << pret. ganti.]
  • gal- (gôli) shine (golden, as the Sun).
  • gama- (gômi) call, shout to. [gama- << gam-.]
  • gav- (gôvi) produce, yield, bear fruit.
  • gratha- (grôthi, grathi) rub, scratch, fray.
  • gwar- (g(w)ôri) watch, all senses; guard.
  • hadha- (hodhi, †hanni) cling, cleave to [hadha- << hadh-.]
  • hag- (hôgi) sit, sit down.
  • hala- (hôli) drag, draw, pull … slip on. [hala- << hal-.]
  • haw-, hau (pret. hui < hôwi; or hauthi) to lie.
  • lag- (lôgi) gather up, pick up, get.
  • lang- (longi) to blare, clang, ring.
  • las- (lôsi) look at, glance at.
  • lav- (lôvi) lick.
  • mal- (môli) chew.
  • mav- (môvi) I like.
  • nab- (nôbi) take, lay hold of.
  • naf- (nôfi) suspect, have inkling of.
  • nag (nôgi and nanci) chew, gnaw.
  • nam- (nômi) withdraw, draw out, draw back, take back; also intr. withdraw, retire.
  • rag- (rôgi) break asunder, burst.
  • ras- (rôsi) scorch, toast.
  • sad- (sôdi) reck, care, value, esteem; show respect for; consider.
  • sam- (sômi) arrange, put together; adjust, settle, reconcile.
  • sana- (sôni or santhi) can, know how to; have knowledge, craft or skill.
  • tab- (tôbi) cut to shape, fashion.
  • thar- (thôri) to saw (up).
  • thas- (thôsi) shave.

There are five preterites in this group that are given in GL with short ‑o- rather than ‑ô-: fab- (fobi) handle, faf- (fofi) puff, far- (fori) separate, hadha- (hodhi) cling, and lang- (longi) to blare. Of these five, only longi *blared appears to have a short ‑o- for a discernible phonological reason: ng here probably represents /ŋg/ developed from original /ŋ/,8 so that *lāŋ- > *lōŋ- > long-, with ō shortened to o with the closing of the syllable (â, ê, î, ô, û do not generally occur in closed syllables in Goldogrin with very few exceptions, and a long vowel before ng does not occur at all). There appears to be no phonological reason for the short ‑o- in fobi, fofi, fori, and hodhi; it cannot be due to the initial consonant in the verb-stem (compare fas- (fôsi) wash, hala- (hôli) drag) or to the medial consonant (compare bab- (bôbi) drink, caf- (côfi) to taste, thar- (thôri) to saw (up), and noun tôdhi paddock). It seems probable, then, that in these four instances Tolkien simply neglected to mark the vowels as long.

Another variation occurs in verbs of this group having final ‑u or ‑w in the stem: haw- or hau to lie, pret. hui; and fau it smells, pret. fui. Tolkien notes that pret. hui is derived from earlier hôwi, the latter exhibiting the form typical of this group, and it probably follows that fui is similarly derived from earlier *fôwi (compare the root FAWA smell in QL). The diphthong oi becomes ui in Goldogrin, as shown by such pairs as Q moile, Gn. muil tarn; and Q moina, Gn. muin(1) safe, secure (these Qenya cognates are given in GL). This suggests that earlier hôwi, *fôwi were shortened to *hoi, *foi, which then became hui, fui.

There are three deleted entries in GL for verbs in this group: cav- (côvi) to bend, cav- (côvi) to bend, make stoop, and tath- (tanthi, tôth[i]) count.

Root-vowel E

  • cwel- (cwîli) fade, wither, etc.
  • gwel- (gwîli) boil, bubble (intr.)
  • gwer- (gwîri) wind, turn, bend (tr.)
  • gwes- (gwîsi) greet, welcome.
  • hel- (hîli) freeze (intr.)
  • mel- (mîli) love.
  • thê- (thai) see.
  • ther- (thîri) to tie.

Despite its atypical form, thai pret. of thê- see most likely belongs with this group. The primitive form of the underlying root is given in the Qenya Phonology as √þeχe (PE12:21), which appears as SEHE in QL with verbal derivative sehta to see, pret. sie. QL also provides the etymology of the preterite form: sie = (sēχı̯ə). This shows that it was in origin a Strong I-A preterite, though this was obscured by various phonological changes in the later form. Etymological forms recorded in GL show that original became ai in Goldogrin (probably first becoming ei) before certain consonants, notably t and s, for example aith thorn < eχtă-, and lais green-sward, glade < leχ-sa. Pret. thai suggests that also became ai before the semivowel ı̯ — perhaps Strong I-A *þēχ-ı̯ē with lengthened stem-vowel and suffix ‑ı̯ē (compare the pret. gui < ı̯ōı̯ē discussed above) > *thei-i (with stem *thēi- shortened to *thei-; diphthongs with a long initial vowel do not occur in Goldogrin)9 > thai.

This group originally included gwedh-(1) (gwîdhi) wind, turn, bend (intr.), but this verb was subsequently emended to gwed-(1) (gwenthi) with the same gloss, changing its class to Strong II (preterites formed by nasal-infixion).

Root-vowel O

  • dol- (dûli) dig. [dol- << dal-.]
  • gol- (gûli) stink.
  • lor- (lûri) sleep deep; dream (tr.).
  • mog- (mugi) detest, hate.
  • nor- (nûri) run, roll.
  • og- (ûgi) am able, can.
  • ol- (ûli or olthi) appear, seem.
  • sô- (sôthi; also sûvi and sûthi) wash clean, bathe (tr. & intr.)
  • sog- (sûgi) drink.
  • thol- (thûli) roll.

The short u in mugi, pret. of mog- detest, hate, is probably a slip (compare og- am able, pret. ûgi). A deleted entry for odra or od(2) (pret. onti) am able, can originally also included an alternate pret. ûd (reading uncertain), which was struck out before the entire entry was rejected.

Of the three preterites of sô- wash clean, bathe, only sûvi belongs to the Strong I-A class; the ‑v- derives from ‑ƀ- in the original root, which is given as soƀ- in the entry for sôn(1) pure, clean (compare SOVO and SOW̯O wash in QL), i.e., *sōƀ-ı̯ē > sûvi. For the other preterites of sô-, see the discussion of pâ-, rô-, and sô- below in the section on Weak II-A verbs.

Root-vowel I

  • cwir- (cwairi) to stir; stir round, make spin.
  • cwiv- (cwaivi) am awake.
  • gil- (gaili) gleam, shine pale and silver, as of the moon.
  • gima- (gaimi) hear. [gima- << gim-.]
  • gwidh- (gwaidhi, †gwinni) weave (tr.)
  • gwil-(1) (gwaili) sail, float, fly [gwaili << gwîli.]
  • gwir- (gwairi) look, look at.
  • îr- (airi or irthi) am willing; will, intend to, mean to.
  • lin- (laini and linthi) sound (intr.)
  • lir- (lairi) sing.
  • lith- (laithi) go, depart, be over, finish, end, die. [An alternate pret. linthi was deleted.]
  • nig (naigi) steal, creep, do or go by stealth.
  • rig- (raigi) twist, contort.
  • rin- (raini) (intr.) revolve, return, come back — do again.
  • thin- or thinta- (thaini) skim.
  • tif- (taifi) whistle.
  • tir- (tairi or tirthi) look for, look out for, watch for, await, expect.
  • uir-, older uvir (uthairi or gwirthi) not to wish, not mean to, mean not to. [The entry îr- am willing compares uir (uthairi or uvairi).]

In two instances, Strong I-A verbs with root-vowel i originally had preterites in î instead of ai: the preterite of gwil-(1) sail, float, fly was first written as gwîli, emended to gwaili; and an earlier pencil entry for tif (tîfi) whistle was replaced in the later ink layer by tif- (taifi) whistle.

The verb uir-, older uvir to not wish, not mean to is transparently the negative prefix û- + îr- am willing; will, intend to, mean to. The preterite uvairi points to earlier *uvīr- (compare Q negative u- or ūv-, prefix mainly used before vowels, = un- in QL); for gwirthi, see the discussion of present-tense gwirtha- below under the Weak I verbs. Pret. uthairi is a difficult form to explain. It might, like uvairi, simply consist of a form of the negative prefix û- + pret. airi. GL lists ug-, um-, un-, ub- as strengthened forms of negative û-, to which we can add uv- from uvir, uvairi. If there was in fact a negative prefix *uth-, its only other occurrence is in uthairiol unintentional; reluctant, clearly cognate with pret. uthairi. The preterite of the negative verb û- not to be, not to do was originally given in GL as ûthi (i.e., Weak II û- + ‑thi), and reinterpretation of this as a Weak I preterite (ûth- + ‑i) might have given rise to *uth- as a variant negative prefix. However, ûthi and its plural ûthin were struck out. We might also posit a verb *thair- or *thir- be willing, mean to as the source of uthairiol and uthairi, but no such verb is recorded in GL.

Root-vowel U

U > AU

  • cug- (caugi) ponder; resolve, think; consider, etc.
  • cur- (cauri) turn, curdle, congeal (intr.)
  • fug- (faugi) smoulder.
  • fur- (fauri) to conceal, to lie.
  • gudh- (gaudhi) am hot (intr.)
  • gum- (gûmi or gaumi) to lade.
  • lur- (lauri) frown, scowl.
  • mug- (maugi) keep silent, say nothing (about).
  • mul- (mauli) low, bellow.
  • nur- (nauri) growl, grumble.
  • nus- (nausi) take notice, perceive.
  • ruf (raufi or rûfi) tear.
  • tul- (tauli or tulthi) bring; come to.
  • tur- (turthi, tauri) can, have power to.
  • tûs- (tausi-) tease wool, comb out.
  • us- also usta- (ausi) leave, depart.

An earlier, deleted version of the entry for cur- turn, curdle, congeal gives the preterite as caur. One other deleted verb in GL belongs in this group: gul- (gauli) ooze, trickle.

U > Û

  • cum- (cûmi) lie (down).
  • gum- (gûmi or gaumi) to lade.
  • hum- (hûmi) sleep, drowze.
  • luv- (lûvi) to hang, lower, of clouds.
  • mum- (mûmi) low, bellow.
  • nuf- (nûfi) get smell of, perceive scent of, smell (tr.)
  • num- (nûmi) sink, decline, slope down, descend.
  • ruf (raufi or rûfi) tear.
  • tû- (tûvi) receive; take; get; become. [An earlier pencil version gives (tûvi or analogical tauvi) get, acquire, receive, take; plur. tuin or tuvin.]
  • tub- (tûbi) cover.

The verb tû- apparently derives from a root *tuƀ- (given in QL as TUVU, whence tuvu- receive, accept, take), with the ƀ becoming (semi)vocalic in the present stem (*tuƀ- > *tuu̯- or *tuu- > tû-) while being retained as v in intervocalic position in the pret. tûvi (for intervocalic ƀ > v in Goldogrin, cf. forms such as Ivon < ı̯əƀánna, in GL s.v. gav-). The rejected verb sû- (sûvi, sûthi) wash clean < √soƀ (sû- was first written as sûtha-) closely resembles tû- (tûvi) in form and etymology; this was replaced by sô- (sôthi; also sûvi and sûthi) wash clean, probably because Tolkien realized that √soƀ would yield *sou̯- > sô- rather than sû-.10 It is unclear why luv- (lûvi) to hang did not develop a present tense **lû- analogous to tû-; perhaps dialectal variation was involved, or the v from the preterite may have been reintroduced to the present stem to avoid confusion with lu occasion, time.

STRONG I-B — Root vowel lengthening in derived verbs

Derived verbs consist of a root (usually verbal) + a derivative suffix (‑tha, ‑ta, ‑na, etc.). Sometimes the suffix changes the meaning of the root, e.g., making it causative; compare basic sog- drink with derived soctha- give to drink, quench, both from SOKO (PE12:85). But often a derivative suffix does not change the meaning of the root but merely acts as a sort of extension; e.g., antha-(1) give < ANA- give, send towards (PE12:31).

The derivative suffix used in the present tense of Strong I-B verbs does not appear in the preterite form, which instead exhibits the bare verb-stem conjugated in the same manner as the Strong I-A verbs. For example, cartha- make, finish consists of a stem car- (= KARA- do, make in QL) + derivative suffix ‑tha, while the preterite côri lacks the suffix ‑tha, being formed directly from the verb-stem car- with vocalic lengthening and addition of the suffix ‑i. The same vocalic changes that occur in the preterites of Strong I-A verbs (ā, ē, ō, ī, ū > ô, î, û, ai, au) also occur in Strong I-B verbs.

  • antha-(1) (ōni) give.
  • cacha-, also cancha (cachui or côgi) laugh.
  • caltha- (côli) wax, grow, flourish. [caltha- << cal-.]
  • cartha- (côri) make, finish. [cartha- << car-.]
  • gôtha- or gai (gui) possess, have, hold.
  • gwista (gwais) be ignorant of, not to know.
  • hosta- (hûsi) gather, collect.
  • intha-(2) (aini) go; fare, proceed.
  • ista- (aisi) know, am aware, perceive, feel.
  • rautha- (rôvi) hunt, chase, pursue.
  • saptha- (†sôbi or sapthani) to dig, to bite into.
  • teltha- (irregular pret. tîli) cover in (close with a roof, lid, canopy, etc.).
  • tertha-, tercha- (irregular pret. tîri) devour, destroy.
  • thin- or thinta- (thaini) skim.
  • urna- (auri) blaze, burn (intr.). [urna- << urtha-, with no pret. given.]
  • us- also usta- (ausi) leave, depart.

Numerous examples of Qenya Strong I-B preterites also occur in the Qenya Lexicon, though only two have direct equivalents in the Gnomish Lexicon:

  • Gn. antha-(1) give, pret. ōni — Q anta- gives, pret. āne.
  • Gn. rautha- hunt, chase, pursue, pret. rôvi — Q rauta- chase, hunt, pursue, pret. rāve.

The preterite gwais of gwista be ignorant of, not to know lacks the suffix ‑i. It is difficult to determine whether this is intentional or a slip; two other examples of Strong I preterites without final ‑i occur in GL: tath- count, pret. tôth (given as tôth[i] in the published text, on the assumption that this form was a slip), and cur- curdle, pret. caur — but both of these entries were deleted (a later version of the entry for cur- gives the preterite as cauri).

Two preterites in this group are called irregular: tîli, pret. of teltha- cover in, and tîri, pret. of tertha-, tercha- devour, destroy. Since these two preterites appear to be typical Strong I-B verbs, it is probable that irregular was meant to describe the Strong I-B class as a whole, in contrast to the far more common weak derived verbs, which outnumber the strong derived verbs by about three to one.

STRONG II — Nasal infixion

Strong II preterites are formed by nasal infixion, i.e., a homorganic nasal (m, n, ŋ)11 was added to the verb-stem before the last consonant. This means of preterite formation was also common in Qenya, and many cognates of Gnomish Strong II preterites appear in the Qenya Lexicon, for example:

  • HEPE bind (infixed *he‑m‑p-) — Gn. heb- (hemfi) bind; Q hepin (hempe) I bind.
  • MATA eat (infixed *ma‑n‑t-) — Gn. mad- (manti) eat; Q mat- (mante-) eat.
  • TAKA fix (infixed *ta‑ŋ‑k-) — Gn. tag- (tanci) fix, make firm; Q taka- (tanke) fix, fasten.
  • HAŘA remain (infixed *ha‑n‑ð-) — Gn. hadha- (†hanni) cling, cleave to; Q harin (hande) remains.

It may also be noted that the same ending ‑i (= Q ‑e) that appears in the Strong I class is added to the Strong II preterites.

STRONG II-A — Nasal infixion in basic verbs

Most of the Strong II preterites in Goldogrin occur in basic verbs (Strong II-A) with stems ending in the voiced stops b, d, g from original p, t, k — the Qenya Phonology notes that the combinations mp, nt, nk were ones particularly favoured by Qenya (PE12:25), and clearly the same held true for Goldogrin. There are also a few Strong II-A verbs with stems ending in the spirants dh, th. Many Strong I-A verbs also have stems ending in b, d, g, dh, th, and it is not clear in most cases why some basic verbs with these final consonants formed their preterites by vowel lengthening while others employed nasal infixion. A few verbs of this form have both Strong I and Strong II preterites: gada- (gôdi, archaic †ganthi) join, gwidh- (gwaidhi, †gwinni) weave, hadha- (hodhi, †hanni) cling, and nag (nôgi and nanci) chew. The Strong II pret. †ganthi of gada- join is called archaic, and the poetic Strong II prets. †gwinni and †hanni are probably also older forms, suggesting that Strong II verbs tended to gravitate toward the dominant Strong I class.

Notably absent from the Strong II-A preterites are stems ending in r, l, s, v, w, m, n, ng, all of which are found as final consonants in the Strong I-A preterites; apparently stems ending in these consonants were not receptive to nasal infixion — clearly a factor in the dominance of the Strong I verbs, since this meant that a larger number of verb-stems were suitable for that class. Basic stems ending in f did allow for nasal infixion, but while there are seven Strong I-A verbs of this type, e.g., caf- (côfi) to taste, there is only one occurrence of a stem ending in f in the Strong II class, the Strong II-B derived verb tefla- (temfi) to scorn < TEFE (whence Q teve- (tembe, tēve) hate, dislike); see the discussion of Strong II-B verbs below.

Tolkien appears to have hesitated over the medial development of original mp, nt, nk in Goldogrin, and several Strong II-A preterites ending in ‑mpi, ‑nti as first written were later emended to ‑mphi, ‑nthi; e.g., gab- (gampi) bark >> gab- (gamphi),12 and hada (hanti) throw at >> hada (hanthi). None of the Strong II-A preterites ending in ‑nci were emended to ‑nchi, although an example of this change does occur in the sonant verbs (see below), in which the analogical pret. flinci of flig- hew was emended to flinchi. Tolkien did not carry out these changes systematically or completely; the impulse to emend ‑mpi, ‑nti, ‑nci >> ‑mphi, ‑nthi, ‑nchi seems to have died out after the L-entries in GL.13

Stems with final B:

  • cab- (camfi, camphi) jump, leap. [camfi, camphi << campi.]
  • gab- (gamphi) bark, bay; of dogs. [gamphi << gampi.]
  • heb- (hemfi) bind, bind round. [hemfi << hempi.]
  • leb- (lempi) pick, pluck, take or feel or touch with the fingers; cull.
  • lib- (limpi) to drip.
  • lob (lompi) run, gallop (of animals).

Stems with final D:

  • bada- (banthi-) build.
  • cwed- (cwenthi-) say, tell. [cwenthi- << cwenti.]
  • dod- (donthi) fall down, drop. [donthi << donti.]
  • gada- (gôdi, archaic †ganthi) join, connect, unite. [gada- << gad-. pret. gôdi archaic ganthi << pret. ganti.]
  • gwed-(1) (gwenthi) wind, turn, bend (intr.) [<< gwedh-(1) (gwîdhi), with same gloss.]
  • hada (hanthi) throw at; c. dat. aim at. [hanthi << hanti.]
  • loda (lonthi) swallow, gulp down. [loda (lonthi) << lod- (lonti).]
  • lud (lunti) flow, stream, float.
  • mad- (manti) eat.

GL also gives one deleted entry belonging to this group: odra or od(2) (onti) am able, can.

The entry for gada- join, connect, unite includes a note stating that in pret. ganthi, pret. of yat- and yṇt- are united. This refers to the fact that ganthi was the preterite form of two separate verbs, gada- (from root yat-) and intha-(1) or gintha- to join to, add, increase (from root yṇt-). For intha-(1), gintha- see below under Sonant Verbs.

Stems with final G:

  • fag- (fanci) cut.
  • log- (lonci) curl, bend (intr.); wind (twine round anything), twine.
  • nag (nôgi and nanci) chew, gnaw.
  • tag- (tanci) fix, make firm, construct.
  • thig- (thinci) pick out, select, choose.
  • *thug (thunci) sink.14
  • tug- (tunci) hit, reach mark etc.

Stems with final DH:

  • gwadh- (gwanni) dwell. [gwanni << gwandi.]
  • gwidh- (gwaidhi, †gwinni) weave (tr.)
  • hadha- (hodhi,hanni) cling, cleave to [hadha- << hadh-.]

Stems with final TH:

  • cwas- and cwath- (pret. to both is either cwanthi or cwasti) shake, nod, wag (tr. & intr.)
  • gretha- (grenthi) save, rescue, preserve.

The entry for lith- (laithi) go, depart originally included an alternate pret. linthi that was subsequently deleted. Two other entries belonging to this group were deleted in their entirety: tath- (tanthi, tôth[i]) count, and reth- (renti-) save, rescue, preserve. Note that the latter entry is unusual in that the final spirant ‑th of the verb-stem reth- is despirantized to ‑t in pret. renti- (this reading is clear in the manuscript).

STRONG II-B — Nasal infixion in derived verbs

The small Strong II-B class consists of derived verbs that form their preterite by nasal infixion. Just as in the Strong I-B class, the derivative suffix used in the present tense of Strong II-B verbs does not appear in the preterite, which is formed instead from the bare verb-stem, conjugated in the same manner as the Strong II-A verbs: tefla- to scorn, bare stem tef- > pret. temfi. There are only two verbs in GL that belong to this class:

  • sibra-, †sib- (simpi) weep, whimper.
  • tefla- (irregular pret. temfi) to scorn.

The pret. temfi is called irregular, probably for the same reason that two Strong I-B preterites were also labeled as irregular: it is far more common for derived verbs to have a weak preterite. A rejected earlier pencil entry gives this verb as Strong II-A: tef- (temfi-) to scorn. GL also has a rejected entry odra or od(2) (onti) am able, can that belongs to the Strong II-B verbs (odra, with bare stem od- > pret. onti).

All attested Strong II-B verbs end in the derivative suffixes ‑la (tefla- scorn) or ‑ra (sibra- weep, rejected odra can); in fact, the only preterites given in GL for derived verbs in ‑la, ‑ra are the three Strong II-B forms just cited. However, there are a substantial number of derived verbs in ‑la and ‑ra given in GL without any indication of the preterite form, e.g., fugla- smoke (a pipe) (cp. fug- (faugi) smoulder), rumla- make a noise (cp. rum noise); and fadra- to sate, weary; feed up (cp. fad enough), cwivra- awaken (intr.) (cp. cwiv- (cwaivi) am awake). While many of these verbs ending in ‑la, ‑ra would accommodate Strong II-B preterites — e.g., fugla- smoke, pret. *func(h)i; fadra- sate, pret. *fant(h)i — it is clear that others such as rumla- make a noise and cwivra- awaken would have to form their preterite in some other fashion, since their stems (rum-, cwiv-) end in consonants that are not receptive to nasal infixion.

WEAK I — Suffixion of ‑i

The Weak I verbs form their preterite by addition of the suffix ‑i. This class consists almost entirely of derivative verbs, with the exception of basic cwas-, cwath- shake, nod, wag (which has an apparently suppletive preterite cwasti) and na-(1) is (with irregular pret. thi << ni). Weak I verbs ending in ‑tha, the most common type, omit the final ‑a and replace it with pret. ‑i (bactha- walk, pret. bacthi), as do those ending in ‑ta (nosta- am born, pret. nosti). Verbs ending in the extension ‑u comprise a third and relatively common subgroup of the Weak I class; in these verbs, the ending ‑u is retained in the preterite, with ‑ui > ‑wi (felu seem, pret. felwi).

Derived verbs in ‑tha:

  • bactha- (bacthi) walk.
  • ciloba- or ciloptha- (pret. always cilopthi) twitter, chirp, chirrup.
  • cuitha- (cuithi) am alive, live.
  • gratha- (grôthi, grathi) rub, scratch, fray.
  • gultha- (gulthi) endure.
  • gwirtha- (gwirthi, properly to uir) not to wish, not mean to, mean not to.
  • haitha- (haithi) hie, go, fare, walk.
  • lentha- (lenthi) come towards speaker, approach, draw near.
  • lintha- (linthi and linthani) sound (tr.); strike or ring bell, etc.; play an instrument.
  • mactha- (macthi) slay, kill.
  • murtha- (murthi) dream.
  • mútha- (mûthi) depart, go, leave.
  • nactha- (nacthi) bite.
  • nictha (nicthi) it is raining, hailing, is snowing.
  • rôtha· (rôthi) embrace.
  • sactha- (sacthi) fight.
  • tortha- (torthi) scorch.

Several Weak I forms in GL were rejected. The entry baidha- to clothe (with no preterite given) replaced an earlier version deleted in ink: baitha- to clothe, pret. baithi. The verb bara- or barna- dwell in, till (tr.) land was first written as bartha-, and the entry originally included a preterite barthi that was subsequently struck out, perhaps at the same time that bartha- was emended to bara- or barna-. The verb dortha- to settle (tr. and intr.) originally included tr. pret. dorthi (Weak I) and intr. dorthani (Weak III), both of which were struck out and not replaced. The entry for sô- wash clean, bathe (tr. & intr.) was first written as sôtha- (sôthi) bathe, tr. & intr.

The verb gwirtha- not to wish is unusual in that it is an analogical present tense, said to be a new formation modeled after pret. gwirthi, which is properly the preterite of uir- not to wish. In the preterite form, the u- in uir- (neg. û- + îr- am willing) became consonantal u̯-, which regularly developed into gw- in initial position; thus *u̯irthi > gwirthi (compare u̯iri > gwir- look and u̯ar > gwar- watch, PE11:46).15 So while gwirthi was in origin a Weak II preterite formed by suffixion of ‑thi to uir-, it came to be perceived as the Weak I preterite of analogical gwirtha-.

Derived verbs in ‑ta:

  • nosta- (nosti, ‑athi) am born.
  • palta- (palti, ‑tathi) beat.

Derived verbs in ‑u:

  • celu-, celwa- (celwi-) trickle.
  • felu (felwi) seem.
  • lenu, pl. lenwin (lenwi(r)) come towards speaker, approach, draw near.
  • nelu (nelwi) ring (tr. & intr.).
  • pelu- (pelwi) fence, enclose; usually in form gobelu-, gobelwi.
  • telu- (telwi) close, end, finish.

An entry lenu- (lenwi) to stretch, extend in length, trail was deleted, probably because Tolkien decided to assign this verb the meaning come, approach instead (as above). The verb belu-, belwa- (beluthi) unroll, unfurl (intransitive); belly (of sails), swell was first written as belu- (belwi-) unroll, unfurl.

The entry for lenu come, approach cites plural forms for both the present and preterite tenses: present pl. lenwin, with lenwi(r) = pret. sg. lenwi, pret. pl. lenwir. Present plurals marked by the suffix ‑in are well attested in GL, e.g., na is, pl. nain; get, acquire, pl. tuin or tuvin; and û not to be, not to do, pl. uin(1) (this list is not exhaustive). The form lenwir is our only example of a preterite plural in ‑r, although the Gnomish Grammar refers to the existence in G. of an ‑r plural sign in verbs (PE11:10). The only other recorded example of a preterite plural in Goldogrin appears in the entry for the negative verb û- not to be, not to do, which originally included pret. ûthi, ûthin. Here ûthin has the plural ending ‑n instead of ‑r (the reading ‑n in the manuscript is clear). However, both ûthi and ûthin were struck out and not replaced.


  • cacha-, also cancha (cachui or côgi) laugh.
  • cwas- and cwath- (pret. to both is either cwanthi or cwasti) shake, nod, wag (tr. & intr.)
  • na-(1) (thi) is. [thi << ni.]

The verbs cacha- or cancha laugh, cwas- and cwath- shake, nod, wag, and na-(1) is fall outside of the three main Weak I sub-categories, and all three have unusual preterite forms.

cacha-, cancha laugh has two preterites, cachui or côgi — in the manuscript, the ending ‑ui in the first form is clear; it cannot be read as **cachni or **cachai. This verb derives from the root KAKA laugh given in QL (Q kakin I laugh), perhaps by gemination of the second consonant: *kakka- > Gn. cacha-; compare Q hakka, Gn. hacha the hams, buttocks. The ungeminated stem is seen in the pret. côgi (< *kāk-).16 The stem of the alternate present tense cancha was formed instead by nasal infixion: *ka‑n‑k- > cancha (compare Q kankale laughter in QL). The final diphthong ‑ui in pret. cachui is puzzling. It might result from the final ‑a of cacha- becoming ‑o with the addition of pret. ‑i: *cacha-i > *cacho-i > cachui (oi regularly becomes ui in Goldogrin). A possible parallel occurs in the pair alfa swan, alfuil swan, the latter apparently from alfa + fem. ‑il (compare ainil, fem. form of Ain a God). However, given that alfa is from earlier alchwa (Q alqe), the ‑uil in alfuil may simply result from *alchw-il > *alchuil > alfuil, so this is not definitive corroboration of *a-i > ui in Goldogrin. The forms glóna- to dawn and glonaith it dawns (with apparent pronominal ‑ith; see Goldogrin Pronouns by Patrick H. Wynne) also serve as a counterargument to this development. On the other hand, there is a tendency in Goldogrin for a (of whatever origin) to become o before certain consonants in unstressed final syllables. The Gnomish Grammar states that unaccented ‑an universally becomes ‑on (PE11:13), as in coma disease (in which the ‑a is from ‑ē; cp. the Qenya cognate qāme), gen. comon, with the same change seen in dative comor. The change ‑as > ‑os is also attested, as in leptha finger pl. lepthos, and odinthas 17 with variant odinthos (ibid.). Perhaps the consonantal ı̯ in the original preterite suffix ‑ı̯ē (see the discussion of gôtha- (gui) possess above) had a similar effect in cachui to that of ‑n, ‑r, ‑s in comon, comor, lepthos.17

The verbs cwas- and cwath- shake, nod, wag (tr. & intr.) are clearly cognate with the root QASA in QL, whence Q qasa- (qasse) shake, flap, nod, rustle (intr.) The Goldogrin forms suggest that there were originally two variant roots *qas- and *qath-, which fell together as QASA in Qenya (in which intervocalic th became s) but remained distinct in Goldogrin (in which th was retained). cwanthi is a Strong II-A preterite, formed by nasal infixion of cwath-. cwasti appears to have originated as the Weak I preterite of a derived verb *cwasta- (compare nosta- am born, pret. nosti), perhaps a causative form make something shake or nod — and although the present tense of this derived verb did not survive, its preterite was absorbed into the paradigm of cwas-, cwath-, perhaps accounting for why these Goldogrin verbs are transitive and intransitive, while Q qasa- shake is only intransitive. On a separate page in GL there is another entry for cwas- wag, shake, flap (tr.), with Strong I-A preterite cwôsi. It is possible that this represents a different conception of this verb from that in the entry for cwas- and cwath-, or it could indicate that cwas- (cwôsi) came to be regarded as exclusively transitive in contrast to intransitive cwath-, with cwanthi and cwasti used as the preterites of either form.

The entry in GL for na-(1) is describes this verb as quite irregular, referring to the fact that the preterite thi and participle ol· derive from different roots than the present stem,18 a phenomenon called suppletion. The preterite of na-(1) as first written in GL was ni, a Weak I form from the same root as the present. This was replaced by thi, which also appears to be a Weak I preterite, the present tense of which would be *tha- is, although this form does not occur as an independent verb in the Goldogrin material. However, the causative/inchoative verbal suffix ‑tha (applied to verbal, adjectival, and nominal stems) is probably cognate; examples include captha- startle < cab- jump, leap, fairtha- set free < fair free, and fuitha to fall (of night); to grow dark < fui night. Compare the IE root *bheu-, *bhū- come into being, become, source of the verb to be in many of the descendant languages, including English be, Welsh bod, Latin fuī I have been, etc.19 This same IE root was the source of the Latin imperfect tense marker ‑ba- (amō I love, amābam I was loving), in light of which it seems unlikely to be coincidental that the Weak II and Weak III preterite suffixes ‑thi and ‑ni (see below) are identical in form to preterite thi and its earlier form ni. There is, however, no evidence suggesting that the Goldogrin preterite was a continuous or progressive past tense.

WEAK II — Suffixion of ‑thi

The Weak II class consists of verbs that form their preterite by addition of the suffix ‑thi; for the probable etymology of this suffix, see the discussion of na-(1) is above. This is the only Weak class in Goldogrin to include a substantial number of basic verbs; the Weak I class contains only one basic verb (na-(1) is), and the Weak III class consists entirely of derived verbs.

WEAK II-A — Suffixion of ‑thi in basic verbs

In most Weak II-A verbs, the suffix ‑thi is added directly to the present stem, whether this ends in a consonant or a vowel, e.g., gai to possess, pret. gaithi; ol- appear, seem, pret. olthi. In two instances, the final ‑a in the present stem is dropped before ‑thi is added: ciloba- chirp, pret. cilopthi; and sana- can, pret. santhi.

  • ciloba- or ciloptha- (pret. always cilopthi) twitter, chirp, chirrup.20
  • gai to possess (gaithi he had).
  • haw-, hau (hui < hôwi; or hauthi) to lie.
  • îr- (airi or irthi) am willing; will, intend to, mean to.
  • lin- (laini & linthi) sound (intr.).
  • ol- (ûli or olthi) appear, seem.
  • pâ- (pôthi or pâthi) do, put.
  • rô- (rûthi or rôthi) remain, stay, stand — (dat.) endure.
  • sana- (sôni or santhi) can, know how to; have knowledge, craft or skill.
  • sô- (sôthi; also sûvi and sûthi) wash clean, bathe (tr. & intr.)
  • tir- (tairi or tirthi) look for, look out for, watch for, await, expect.
  • tul- (tauli or tulthi) bring; come to.
  • tur- (turthi, tauri) can, have power to.
  • ug- (unthi) negative verb.
  • uir- (uthairi or gwirthi) not to wish, not mean to, mean not to. [For gwirthi as the preterite of uir-, see the discussion of the Weak I verb gwirtha- above.]

Rejected Weak II-A forms in GL include en(1) (enthi) am called, an entry that was emended to enn name; and gwara- (gwarathi) wander, roam, travel (far), which was emended to gwada- with the same gloss, but with no preterite provided. Also, the entry in GL for the negative verb û- not to be, not to do originally included Weak II-A preterites ûthi and ûthin (singular and plural, respectively), but these were struck out. However, the Weak II-A verbs include another negative verb ug-, with pret. unthi. The entry for û- negative prefix with any part of speech (a separate entry from the verb û- not to be) states that this prefix also often occurred in the strengthened forms ug-, um-, un-, ub-. The strengthened forms ug- and un- are apparently used verbally in ug- (unthi), with ug- used for the present stem and un- for the preterite.

The verbs pâ- do, rô- remain, and sô- wash clean,21 in addition to their Weak II preterites (pâthi, rôthi, sôthi), also have alternate mixed-class preterite forms in which the Weak II suffix ‑thi is added to a Strong I-A form of the verb-stem (pâ- > pô-, rô- > rû-, sô- > sû-): thus pâ-, pret. pôthi; rô-, pret. rûthi; sô-, pret. sûthi. GL also includes a deleted entry fâ- (fothi) have, hold, possess that has a mixed-class form as its only preterite. sô- has a third preterite, the purely Strong I-A form sûvi (the etymology of which is discussed above under the Strong I-A verbs).

WEAK II-B — Suffixion of ‑thi in derived verbs

The Weak II-B verbs end in the derivative suffixes ‑ta, ‑tha and ‑u, ‑wa:

  • belu-, belwa- (beluthi) unroll, unfurl (intransitive); belly (of sails), swell.
  • enu-, enwa- (enuthi) am called, am named.22
  • ilta- (iltathi) to stick in, prod, prick. [The preterite form only appears in the sentence on iltathi nin pieg *he stuck me with a pin.]23
  • lantha- (lantathi) fall onto, settle on, alight. [lantha- << lanta-.]
  • nosta- (nosti, ‑athi) am born.
  • palta- (palti, ‑tathi) beat.

Rejected Weak II-B forms in GL include the entry canca- (cancathi) to laugh; the preterite was emended to canghathi, and then the entire entry was struck out, being replaced later in the same group of entries (listed under the headword cag a joke, jest) by cacha-, also cancha (cachui or côgi) laugh. The entry for elma- marvel at, admire originally included pret. elmathi, but this preterite was struck out and not replaced.24 The verb lantha- fall onto was lanta- as first written, and its pret. lantathi has a light pencil stroke through it, perhaps indicating that it was rejected (probably at the same time that lanta- was emended in pencil to lantha-).

WEAK III — Suffixion of ‑ni

The Weak III class consists entirely of derived verbs ending in ‑tha that form their preterite by addition of ‑ni; for the probable etymology of this suffix, see the discussion of na-(1) is above.

  • lintha- (linthi and linthani) sound (tr.); strike or ring bell, etc.; play an instrument.
  • rûtha (rûthani) dwell, remain.
  • santha- (‑thani) show, declare, point out, make known or clear.
  • saptha- (†sôbi or sapthani) to dig, to bite into.

Two Weak III preterites in GL were rejected. The entry for dortha- to settle, as already noted above in the discussion of the Weak I verbs, originally included two preterites, tr. dorthi (Weak I) and intr. dorthani (Weak III), both of which were struck out. The entry failtha- go pale also originally included a preterite [fail]thani that was struck out and not replaced.

SONANT VERBS — from roots with vocalic Ḷ, Ṇ, Ṛ

The vocalic system of Primitive Eldarin in the Lost Tales period included , , (PE12:3), which are also referred to collectively in the Qenya Phonology as sonants (PE12:10). Like the other vowels (a, e, o, i, u), the sonants , , could be either long or short, and they appear as the vowel in a large number of roots — QL includes many examples such as FḶKḶ cleave, hew, LṆQṆ catch, find, get, and VṚT͡YṚ serve.

In Qenya and Goldogrin , , underwent a process the Qenya Phonology calls sonant-resolution (PE12:10). In Qenya, this meant that the sonants became diphthongs consisting of u, a, or i + l, r, or n. The resulting diphthongs varied according to two factors: 1) the length of the original sonant, and 2) the articulation position (q, k, k̑, t, p) of the consonant following the original sonant. For example, short became ul before q and p (e.g., sulpa soup < SḶPḶ) and il before k, k̑, and t (e.g., silkesse hay, harvest < SḶKḶ(1)), while long became al in all instances (e.g., salpa bowl < SḶPḶ, salka scythe < SḶKḶ(1)). The development of was the most varied, yielding , , , an, um, or am depending on the two factors noted above. A full discussion of this process may be found in the Qenya Phonology (PE12:10–11).

In Goldogrin, the most common sonant resolutions were as follows:25

Short > li glith dale, hillslope < nḷđ.
Long > al nal dale, vale < *nḹđ.
[QL has NḶDḶ (nḷřḷ), whence Q nal (nald-) dale, dell.]
Short > in in(d)ra long < ṇdr.
Long ṇ̄ > an annai far away < ṇ̄dai.
[QL has NĐN stretch, whence Q ande, andea long.]
Short > ri tricthon (fibrous, fine) root < tṛk.
Long > ar tarc root (especially edible roots) < *tṝk.
[QL has TṚKṚ, whence Q tarka root.]

This process of sonant-resolution gave rise in both Qenya and Goldogrin to a number of ablaut verbs. These were in origin Strong I-A verbs (i.e., basic verbs with a lengthened vowel in the preterite) derived from roots with sonant vowels; and the different resolutions of short sonants and long sonants often resulted in apparent qualitative ablaut in these verbs in the descendant languages, i.e., the preterite appears to have been formed by substitution of a different stem vowel than that in the present stem.26 In Goldogrin, the present tense of these sonant verbs almost always has stem-vowel i (from resolution of the short sonants , , > li (or il), in, ri), with stem-vowel a in the preterite (from resolution of the long sonants , ṇ̄, > al, an, ar). For example:

  • — Gn. ilt- (galti) to yoke, join < YḶTḶ. Compare Q tulpu- (talpe) uphold, support < TḶPḶ.
  • — Gn. tinta- (tanti) dance < TṆTṆ. Compare Q tunq- (tanqe-) hear < TṆQṆ.
  • — Gn. briga (*barchi) I fear < VṚKṚ (with analogical pret. braigi- later substituted for etymological *barchi). Compare Q tirty- (tartye-) divide, part < TṚT͡YṚ.

Sonant verbs in Goldogrin fall into two broad groups: 1) those that retained their ablaut preterite, either as their sole past tense or with concomitant analogical preterites modeled after other verbal classes; and 2) those that lost their ablaut preterites, replacing them with analogical forms. The first group includes the following verbs:

  • cinga- (cangi-) to go wrong, get tangled. [Apparently from a root *kṇg-, whence also cang tangled, confused, mixed up, awry, wrong, which GL compares with cing spider. Possibly related is the root KANGA- weave in QL (whence Q kangale a web, kangaris spider.]
  • cintha- (canthi) to light, set alight. [QL has KṆŘṆ shine, whence Q kanda- (kandane or kande) blaze.]
  • cwintha- (cwanthi) to fill. [<< cwinta- (cwintathi or cwanti) to fill. QL has QṆTṆ or QATA, whence Q qanta- (qante, qantane) fill, complete.]
  • gartha- (garthi) respect, admire, have affection or respect for. [GL gives the root as gṛþ- = Q karda, with gṛþ- << kṛþ-; karda does not appear in QL, nor does any root analogous to kṛþ- or gṛþ-.]
  • grith- (graithi, analogous for garthi) attend to, tend, care, care for, mind. [The forms were changed in ink << gridh-, graidhi & gardhi.]
  • gwinta- (gwanthi) see [gwanthi << gwanti. Apparently from a root *gwṇt-, whence also gwint face, gwintha- to face.]
  • ilt- (galti; also ilti or iltathi) to yoke, join. [QL has YḶTḶ, whence Q yalta yoke.]
  • intha-(1) or gintha- (ganthi) to join to (intr.); add, increase; add to. [<< int- (ganti). QL has DẎṆTṆ, whence Q ʼyanta- (yante-) enlarge, increase, add to.]
  • tinta- (irregular pret. tanti) dance. [tanti << tinti. QL has TṆTṆ, whence tanta- (tante-) bounce, bound, rebound and tanta- (tantane) dance (tr.), dandle, wave.]

The verb gartha- (garthi) respect, admire is said to be from a root gṛþ-, and is unusual among the Goldogrin sonant verbs in having a present tense form with stem-vowel a rather than i. This is probably due to an original long vowel in the present stem: gartha- < *gṝþ-. Another possible example is canga- to tangle (with a Weak II preterite cangathi struck out), which appears to derive from the same root *kṇg- as cinga- to go wrong, get tangled, only with a long vowel in the present: canga- < *kṇ̄g-. As noted in footnote 21, some present tense forms in Goldogrin have a long stem vowel, and in some instances this long vowel appears to be original, e.g., lada- to smooth out has a variant lôda which must derive from *lāt-. grith- attend to, tend, care, like gartha- respect, admire, apparently derives from gṛþ- (these two verbs are grouped together in GL) but has instead ri < short . grith- has the analogical pret. graithi, modeled after Strong I-A preterites such as gwir- (gwairi) look. The analogical form may have arisen to avoid potential confusion resulting from the fact that garthi (the historical preterite) was identical to that of gartha- respect, admire; the lack of an asterisk beside garthi in the entry grith- implies that it continued as a living form beside graithi (compare this with the asterisked forms in group 2 below). ilt- to yoke, join has three preterites, one historical (galti < *yḹt-) and two analogical (Weak I ilti and Weak II iltathi).

The entry for tinta- dance refers to tanti as an irregular pret., and a separate entry for the preterite form describes tanti as the irregular pret. of weak tinta-. Evidently, in later Goldogrin tinta- was perceived as a derived verb < *tin- + ‑ta, one which ought to have a weak preterite in ‑i or ‑thi, such as nosta- (nosti, ‑athi) am born and palta- (palti, ‑tathi) beat. Hence the preterite tanti was regarded as irregular. It seems likely that this sort of reinterpretation of sonant verbs ending in ‑ta or ‑tha as derived verbs might also have taken place with other verbs from this first group (cintha-, gwinta-, intha-, etc.).

The second group, sonant verbs that replaced their historical ablaut preterites with new analogical forms, includes the following:

  • briga (braigi-, new form, analogical for *barchi) I fear, am afraid of. [briga << brig-; for *barchi << *barci. QL has VṚKṚ, whence Q varkin I dread, I fear.]
  • clib- (claibi, analogical for talpi*) uphold, support; reflexive with allative, to lean. [GL gives the root as tḷp-; QL has TḶPḶ (Gn. clib-), whence Q tulpu- (talpe) uphold, support.]
  • crib- (analogical pret. crimpi) gather, pluck. [QL has KARPAR pluck.]
  • drib- (analogical pret. draibi) rot, decay, wear out. [GL gives the root as nṛq-; QL has NṚQṚ to wither, fade, shrivel, whence Q narqa- wither (intr.).]
  • drith- (new analogical pret. drinthi- for *nardi-) it smells, smacks of, savours of. [GL gives the cognate noun drith smell, savour, flavour from root nṛþ-. QL has NṚŘṚ smell sweet, whence Q narda- smell sweetly.]
  • flig- (flaigi or flinchi, both analogical for *falci) hew. [flinchi << flinci. QL has FḶKḶ cleave, hew.]
  • thlib- (thlaibi, for salpi*) sup, lap up, suck. [QL has SḶPḶ, whence Q sulp- (salpe) lick, sup, lick up, sup up.]
  • thlid- (thlinti, for salti*) to sort out; sift, sieve; discriminate. [QL has SḶTḶ, whence silt- (salte-) sort out, sift, winnow.]

Here the original ablaut preterite forms were lost, apparently at a relatively early period in the development of Goldogrin, since when these are cited they are marked with an asterisk (*barchi, *nardi-, *falci, etc.), indicating a hypothetical earlier form reconstructed from later recorded forms. (Tolkien’s usual practice in GL was to place an asterisk in front of a hypothetical form, but in the case of talpi*, salpi*, and salti* the asterisk follows; if this was meant to be significant, the distinction being made is obscure.)

A comparison of this second group of sonant verbs with the first suggests a reason why the historical ablaut preterites were sometimes abandoned. In the second group, the abandoned ablaut preterites are usually quite different in appearance from their present tenses: briga (*barchi); clib- (talpi*); drith- (*nardi-); flig- (*falci); thlib- (salpi*); thlid- (salti*).27 This disparity is due to the fact that these verbs are all from sonant roots in , — short , in the present stem resolved to li, ri, resulting in initial consonant combinations that frequently underwent further changes; for example, *tḷ- > *tli- > cli- (in clib-), *nṛ- > *nri- > dri- (in drith-), *sḷ- > *sli- > thli- (in thlib-). In the preterite, however, , resolved to al, ar so that no initial consonant combination occurred: *tḹ- > tal- (in talpi*), *nṝ- > nar- (in *nardi-), *sḹ- > sal- (in salpi*). This is in marked contrast to the first group, most of which are derived from sonant roots in , which always resolved to a vowel + n ( > in, ṇ̄ > an), meaning that the present tenses in this group remained close in form to their ablaut preterites, usually differing only in variation of the stem-vowel: cinga- (cangi-); cintha- (canthi); cwintha- (cwanthi); gwinta- (gwanthi); intha- or gintha- (ganthi); tinta- (tanti).

The formal disparity between the ablaut preterites and present tenses in the second group must have acted as a strong impetus to replace the ablaut forms with analogical preterites more closely resembling the present tenses. It is noteworthy in this regard that even in the first group, the verb whose ablaut preterite least resembles its present — ilt- (galti) to yoke, join — developed two analogical preterites more closely resembling the present tense: ilti, iltathi.

There are a few other verbs listed in GL that are probably derived from sonant roots, but for which no ablaut preterite form (retained or lost) is mentioned. For example, rig- *to snarl, sneer (with cognate noun rig a snarl, a sneer) is compared with arc fierce, harsh, ill-tempered < r̄k (i.e., ṝk) and brigla- change, vary is from vṛt (QL has vṛt-, whence Q varta change). Other examples might include ilta- (iltathi) to stick in (QL has ILTIL thrust in, whence Q iltin I thrust home), irtha- peck (QL has IRTYIR- peck, prick, whence Q irtin I peck, prick), and sinta- to sparkle (QL has SṆT͡YṆ twinkle, whence Q sinty- (santye) sparkle).

The past-tense auxiliary particle

The GL entry gî(laith) once, in past includes the following note: often, in form , used loosely as a mere ‘aid’ to a past tense. Unfortunately, there are no actual examples of used as a past-tense auxiliary in the Goldogrin corpus, though grouped with gî(laith) are two other related forms, gîr yesterday and gîrin bygone; old, belonging to former days. However, in two charts of early Noldorin verb conjugations from the Leeds period, a prefix gi- occurs in two aorist forms: (gi) mennin 1 sg. aorist of mad- eat (PE13:129) and gindengin 1 sg. aorist of dagion I kill (PE13:130). Written below gindengin is an alternate form di·nengin, in which the prefix di- appears to correspond to gi-. It seems likely that this early Noldorin aorist prefix gi-, di- is the conceptual successor of the Goldogrin preterite auxiliary .

QL has the root DYĒ (also dyeye and dyene) behind, back (before of time), whence ʼyenye(n) yesterday, die yesterday, diéra yesterday’s; bygone, over, passed, and other derivatives. The simplest form of this root, DYĒ, is probably the source of Gn. . Original initial *dy- became g- in Goldogrin, as in Gn. gulin patient, enduring, long suffering (lit. burdened) < √dyulu- (PE11:38), and gantha more < yṇt (PE11:37), a root that appears in QL as DẎṆTṆ. Thus *dyē > Gn. (with the usual development of *ē > Gn. ī). Early Noldorin gi- probably has the same etymology, with the alternate form di- also derived from the same root DYĒ, though via a form in which the Y became vocalic: *dyē > *die (compare Q die yesterday) > di-. GL has several forms in di- that are probably derived from DYĒ in a similar manner, including dîn once (upon a time), long ago, formerly and dim former, past, over.

1. See Carl F. Hostetter’s article The Past-Tense Verb in the Noldorin of the Etymologies: A Formal Classification. Tolkien uses the term weak once in GL, referring to the form tanti as the irregular pret. of weak tinta-, the latter glossed as dance. There are no verbs explicitly labeled strong in GL, though the contrastive existence of a class of strong Goldogrin verbs logically follows from the reference to weak tinta-.

2. Although minor changes do occur to the verb-stem in a few weak preterite forms, these changes serve no grammatical function but are instead purely phonological, resulting from a different phonological environment in the preterite. Thus felu- seem, pret. felwi; ciloba- chirp, pret. cilopthi; uir- not to wish, pret. gwirthi; etc.

3. In the Qenyaqetsa the terms Eldarin and primitive Eldarin refer to the original speech once common to all Elves, the language that would in Tolkien’s later writings be referred to as Primitive Quendian: All who were awakened in Palisor at Koivienéni are therefore called Eldar and that kindred the Eldalie; for which reason doth Rûmil say that there was once a single Eldarin spoken by all these Eldar in common (PE12:1).

4. These same changes are also found in nouns and adjectives, as in the following examples (the Qenya cognates are all cited in GL):

  • Gn. ôm mind, heart, thought — Q āma.
  • Gn. Cîmir, name of Yavanna — Q Kēmi Mother Earth.
  • Gn. sûr long, trailing — Q sōra.
  • Gn. cail lip — Q kīla.
  • Gn. baul body, trunk — Q pūle.

It is interesting to note that with the exception of ā > ō, these are the same changes that occurred in English during the Great Vowel Shift, which took place between the 12th and 18th centuries and affected the point of articulation of long stressed vowels. Very simply put, Middle English ā, ē, ō were raised to ē, ī, ū respectively, and Middle English ī and ū, which could not be raised as they were already at the highest points of articulation for front and back vowels, became the diphthongs ai and au. The shift of Goldogrin ā > ō is reminiscent instead of a change that took place in the development of Old English into Middle English — by the early 13th Century, OE ā had become ME ǭ in all dialects south of the Humber river, e.g., OE stān became ME stoon /stɔ:n/, Modern English stone. Other examples include Modern English bone (OE bān), boat (OE bāt), and toe (OE ). (See Joseph Wright’s An Elementary Middle English Grammar, §51.)

In unstressed position, different developments apparently occurred to original ā, ē, ō, ī, ū in Goldogrin. For example, according to the Gnomish Grammar (GG), original final ‑ā, ‑ē, ‑ō in nouns all fell together as ‑a in Goldogrin — hence Q elda, Gn. egla fairy (< ‑ā); Q qāme, Gn. coma disease (< ‑ē); and Q noldo, Gn. golda gnome (< ‑ō). GG also states that phonologically, final ‑u, ‑i in nouns such as culu gold and brindi queen only refer to ‑ū, ‑ī. But whether these statements in GG are borne out consistently in the nouns recorded in GL remains to be seen; a full and systematic study of Goldogrin phonology has yet to be done.

5. No explicit statement exists that Goldogrin f and v were labiodental rather than bilabial. In the Qenyaqetsa, however, it is noted that Eldarin f was a bilabial spirant (PE12:15), which in Qenya became labiodental f initially, while medially between vowels it became f labiodental > v, presumably labiodental v (PE12:20).

6. This phonological rule is also seen in various nouns and adjectives in GL, for example (Qenya forms are cited from QL):

  • Gn. hûm sleep, slumber — Q fūme (e) sleep.
  • Gn. lûm time — Q lūme (i) time.
  • Gn. nûmin the west — Q nūme west.
  • Gn. ûmi large — Q ūmea large.

The only possible exception to this rule to be found in GL is the noun auba a shout, which is compared to the verb uptha- shout. These forms suggest an original root *UP-, and auba might derive from *ūp-. On the other hand, GL notes that the prefix was used in forming [a] number of adjs. and occasionally nouns — unaccented and probably of various origin, and it could be that auba represents instead *a‑up-.

The rule that ū does not become au in Goldogrin when followed by a labial consonant presents another intriguing parallel with English phonology. As noted in footnote 4, during the Great Vowel Shift Middle English ū became the diphthong au. However, there was an exception to this: ME ū did not diphthongize when followed by a labial consonant (especially p and b), e.g., ME droupen > Mod. Eng. droop, ME toumbe > Mod. Eng. tomb, ME roum > Mod. Eng. room (all with /u/). Remarkably, this is precisely the same exception seen in Goldogrin.

This phonological exception shared by both Goldogrin and Middle English, along with Goldogrin’s parallels with the Great Vowel Shift, begins to suggest an answer to a question that has long puzzled Tolkien scholars: why isn’t Goldogrin more similar to Welsh? Twenty years ago, Christopher Tolkien noted in his introduction to the Appendix to names in The Book of Lost Tales, Part I that his father’s early languages have several resemblances to early English that are obviously not fortuitous, citing Gn. hôr old (OE hār hoary, grey, old) and rûm secret (whisper) (OE rūn mystery, secrecy). Moreover, plural formation in Goldogrin, via the suffixes ‑in or ‑th (sometimes dissimilated to ‑s or ‑f), more closely resembles plural formation in Middle English (eye eye, pl. eyen; fowel bird, pl. foweles) than it does the far more complex Welsh system of plural formation, in which vowel changes play a major role (brân crow, pl. brain; maneg glove, pl. menig; etc.), in addition to a wide variety of suffixes (arf weapon, pl. arfau; cath cat, pl. cathod; pechadur sinner, pl. pechaduriaid; etc.). It appears that Tolkien may have intended Goldogrin to be a sort of Anglo-Welsh hybrid, having close phonological, morphological, and grammatical similarities to English as well as similarities to Welsh.

7. These etymologies are not given as such in GL and represent my own suppositions. The roots here are cited from QL, which typically gives roots in the form CVCV, in which the second vowel is usually a repetition of the first and not essential to the meaning of the root. Roots cited in GL often omit this final repeated vowel and have the form CVC; for example, the S-entries in GL include the roots sab- (= SAPA dig, excavate in QL), sog- (= SOKO *drink in QL), and soƀ- (= SOVO, SOW̯O wash in QL).

8. The development of original intervocalic ŋ > ng (i.e., /ŋg/) evidently occurred in both Goldogrin and contemporary Qenya; e.g., QL has EŊE whence Q enga mead, with Goldogrin cognate enga plain, vale in GL. Also, the table in the Qenya Phonology listing the development of consonants in initial, medial, and final position (PE12:28) gives ‑ng- as the medial development of ŋ. The development of intervocalic or medial ŋ changed in Tolkien’s later writings, however. In the Etymologies, ŋ (represented by the letter ñ) appears to have usually vanished medially, as in Q téra, N tîr straight, right < *teñrā (s.v. entry TEȜ-); and N line, row < *tēñe (s.v. entry TEÑ-). From Quendi and Eldar, Appendix D (c. 1959–60) refers to intervocalic ñ, later lost in Quenya (VT39:6–7), with examples such as Q lip < *peñe, dual peu the two lips, the mouth-opening < peñū (VT39:11).

9. The only occurrence of a diphthong with a long initial vowel in GL is gâi to possess, and this form, significantly, was emended to gai.

10. In all its other occurrences in GL, earlier *ou̯ regularly yields either ô (χou̯-ră > hôr old, aged, ancient, lou̯me > lôm gloom, shade) or o (ou̯χē > ocha fleece, sou̯þ- > soth bath).

11. Homorganic nasal — a nasal pronounced using the same speech organs employed in pronouncing the following consonant, i.e., a nasal of the same articulatory position as the consonant it precedes. m is the homorganic nasal of p (labial); likewise n is the homorganic nasal of t (dental), and ŋ of k (velar).

12. In spelling Goldogrin words, Tolkien occasionally employs ph to represent the sound /f/ when derived from /p/, which is also one of the uses of ph employed in later Sindarin orthography (see Appendix E to The Lord of the Rings, entry PH); in this regard, note that Tolkien gives both camfi and camphi as alternate spellings of the preterite of cab- jump, leap. However, in contrast to Sindarin, in which f represents /v/ in word-final position, e.g., in Nindalf, Fladrif (Appendix E, entry F), f in Goldogrin spelling evidently always represents /f/ and never /v/. In those instances where a final ‑f in Goldogrin corresponds to ‑v- medially — e.g., gôf fruit, pl. gôvin — this is due to the devoicing of voiced spirants in word-final position. Compare the devoicing of final /ð/ > /þ/ in forms such as Gn. tath hedge, fence, pl. tadhin, probably from root TAĐA in QL, whence Q tanda hedge, fence, enclosure, tar fence, etc. This phenomenon provides another phonological parallel with Middle English, in which final ‑z, ‑v, and ‑ð were devoiced to s, f, and þ, as in the Northern dialect gif to give, luf to love beside Midlands dialect ȝiven, lȯven (examples from Fernand Mossé’s Handbook of Middle English, §45).

13. Nor were these changes restricted to preterite verbs — limp, limpelis the drink of the fairies was hesitantly changed to limfa, limfelis; Gilthalunta Silvership (name of the moon) >> Gilthaluntha (though the separate entry lunta a ship was left unchanged); and lencos distance >> lenchos, to cite only a few examples. It is possible that Tolkien intended the introduction of forms with medial mph, nth, nch beside those with mp, nt, nc to represent dialectal variations in Goldogrin; note, for example, that tent toe has two plurals, one of each type: tenthin and tentin.

14. This entry does not appear as such in GL, but is reconstructed here based on the following evidence. In the Gnomish Grammar, the preterite nûmi sank (also in GL) in the phrase nûmi Galmir i·saroth Galmir sank into the sea (PE11:12) was first written as thunci, presumably also preterite sank. Given such attested Strong II-A verbs as tug- (tunci) hit, reach mark and tag- (tanci) fix, make firm, it seems probable that thunci is the preterite of a verb *thug-, and GL does in fact list a form thug, though with no gloss given and with no final hyphen (a final hyphen is lacking in many verb-stems in GL, such as nag chew in the list above).

15. The same development of negative u- > gw- appears in gwivrin barren, earlier uivrin (which still survives, ui being replaced by analogy) < neg. û- + ivrin fertile.

16. Alternatively, the ending ‑cha in cacha- might originate from a derivative suffix *‑ka: *kak-ka- > cacha-. An apparent derivative ending ‑ca, ‑cha occurs in a few Goldogrin verbs, for example glisca- to like sweet things (cp. glis sweet) and halcha- drag on ground (roughly) (cp. hala- drag).

17. It should be noted, however, that original *ai generally remained unchanged in Goldogrin, e.g., Ain a God (Q ainu a pagan god < AY̯A honour, revere in QL). A few examples of ai > e occur, for example Elwing lake foam, with the older spelling Ailwing preserving the initial element ail a lake, pool unchanged. Original ai also remained unchanged in unaccented final syllables, e.g., annai far away, for which GL gives the etymological form ṇ̄dai.

Another possibility is that ‑ui in pret. cachui is an analogical borrowing from the Weak I verbs in final ‑u, although the latter consistantly have preterites in ‑wi, not ‑ui. Perhaps it was felt that cachui was simply more euphonious than **cachwi (note that the combination chw is allowed in Goldogrin, e.g., celchwed glassy and sachwen summer, midsummer). However, all other attested Weak I verbs in final ‑u are derived from roots ending in either ‑l (belu-, celu-, felu, nelu, pelu-, telu-) or ‑n (lenu, enu-), which makes cacha- an unlikely candidate for analogy with these forms.

18. The participle ol· is perhaps derived from the root Ō be, exist given in QL with the sole derivative ō- am.

19. It is also interesting in this regard that the Noldorin Word-lists (from the Leeds period, during which Tolkien reworked Goldogrin into an early form of Noldorin) include a verb tha- to make, cause to be, pa.t. aist (from *s’ta-, asti-), of which Tolkien writes, From this arises the causative suffix, ‑tha (PE13:153). The etymological forms of tha- and aist cited in this entry bear a striking resemblance to other Indo-European roots and their derivatives associated with the sense to be: with *s’ta- compare IE *stā- stand, whence Spanish estar to be; and with asti- compare IE *es-, whence Latin est is, Modern English is, and (most notably) Sanskrit as-, 3 sg. asti.

20. ciloba- twitter, chirp, chirrup does not have the C(C)VC structure characteristic of almost all basic verbs, such as lav- lick and cwir- to stir. The associated nouns cílobi, cilobinc robin are cognate with Q kilapi, ‑ink robin listed in QL along with kilapitya- to chirp. No root is provided in QL, probably because kilapi is simply an onomatopoeic approximation of the warbling song of the European Robin (Erithacus rubecula). The verbs Gn. ciloba-, ciloptha- and Q kilapitya- were apparently derived from the nouns, in the sense *sing like a robin. The form ciloptha-, with derivative ending ‑tha, shows that ciloba- was regarded as basic for morphological purposes. The preterite cilopthi apparently represents a convergence of the Weak I preterite of ciloptha (cilopth-i) and the Weak II preterite of ciloba- (cilop-thi).

21. Goldogrin present-tense forms with a long vowel are not uncommon, occurring in a wide variety of verb-types, both basic (û- not to be, îr- am willing, tûs- tease wool, lûda- to bloom, etc.) and derived (lûtha- pass (of time), fôtha it snows, etc.). These long vowels are various in origin, a full account of which is not possible here, though a few examples can be cited. Some derive from original long vowels, as lada- to smooth out with variant lôda < *lāt-. Some are later lengthenings of original short vowels, as gôtha- possess < ı̯otta (an etymological form given in this entry in GL). Others apparently result from the loss of an original intervocalic consonant, which either caused compensatory lengthening of the preceding vowel, or perhaps two short vowels of the same quality were brought into contact when the intervening consonant disappeared, as thê- see, probably from *theʼe- < √þeχe (this root cited in the Qenya Phonology, PE12:21); also compare the etymology of mark, line, which according to GL is from teȝ́e, while of the same meaning is from .

22. The entry for this verb in GL and the entry immediately below it appear in the published text as:

enu-, enwa- am called, am named. [Pencil.]
enos, enuith title. (enuthi.) [<< “enos title.”]

Although this reproduces the actual arrangement of forms in the manuscript, enuthi was probably not intended as an etymological form of enos, enuith title (as the published text of GL suggests); rather, enuthi is most likely the preterite of enu-, enwa-, having been written on the line below this verb due to a lack of space following the gloss am called, am named.

23. ilta- might be a sonant verb < *ḷt- rather than a derived verb < *il- + ‑ta; see the section on Sonant Verbs.

24. elma- marvel at, admire appears to be derived from the noun elm a wonder; a singular, marvellous, or unique thing; something strange; these forms and cognates such as eleg other, else and elfel different, strange, like something else point to a root *EL- other, different (not given in QL). Compare olma- I dream, probably from noun olm a dream < OLO(3), a root compared in QL to LORO doze, slumber.

25. Although a full discussion of sonant-resolution in Goldogrin is not practical here, a few additional developments may be noted. Short sometimes became il or ul instead of li, as in the prefix il- (ul-) denoting the opposite, the reversal, i.e. more than the mere negation, confused with ul- < √ul … because, being from , it gave ul- before m, b, w, etc. — compare the GL entry for ulm jar, flagon, which originally included the etymological form ḷm (later stuck out), a further example of > ul- before m as stated in the entry for il- (ul-). Short after a labiovelar appears to have become ur instead of ri; thus curdhu sin, wickedness, evil < qṛđ (QL has QṚŘṚ, whence qarda bad), and gurth *death < ngwṛþ- (s.v. Gurtholfin; QL has GWṚĐR die, whence Q urdu death). The sonants and were sometimes retained unchanged in word-final position, e.g., aithl a spring < ektḷ, ethl(in) ivy < etḷ-, and idhr (indr, inr) long, a long time (av.) < ṇdr. But compare the development of final > li in bothli oven < mbāsḷ, as well as final > ‑il, ‑l in Danigwethil, Danigwethl the great mountain of Valinor (with ‑thḷ < ‑tḷ).

26. Tolkien himself uses the term apparent ablaut to describe these forms in The Qenya Verb Forms, four loose pages detailing the conjugation of Qenya verbs tucked inside the front cover of the Qenya Lexicon. A note to the past-tense suffix ‑e on the chart labelled Endings states that in addition to the endings this tense is distinguished by (1) ‑n- strengthening, or (2) vowel-length strengthening, or (3) ‘apparent’ ablaut, or (4) suffix between stem and ending (PE14:31). The qualitative ablaut in Qenya (and Goldogrin) is only apparent in that it is actually the result of divergent phonological development in the later languages of short and long forms of the same original vowel; whereas in the Indo-European languages qualitative ablaut was a matter of actual substitution of one vowel for another. Proto-Indo-European also made use of quantitative ablaut, based on alternations in vowel length. The most prominent qualitative ablaut series in IE was e / o / ø, as exemplified in Greek: peíth-ō I persuade (e), pé-poith-a I trust (o), é-pith-on I persuaded (ø) (Modern English has many verb forms resulting from this process, e.g., sing / sang / sung). Greek also provides an example of quantitative ablaut in the declension of the word for father: nom. patḗr (ē), acc. patéra (e), gen. patrós (ø). (Examples are cited from An Introduction to the Indo-European Languages by Philip Baldi.)

27. GL does not give the original ablaut preterites of crib- (< *kṛp-) and drib- (< nṛq-), but these can be reconstructed as *carpi and *narcwi, respectively.

Baldi, Philip. An Introduction to the Indo-European Languages. Carbondale and Edwardsville: Southern Illinois University Press, 1983.

Hostetter, Carl F. The Past-Tense Verb in the Noldorin of the Etymologies: A Formal Classification. Published in Tengwestië, Dec. 4, 2003.

Mossé, Fernand. A Handbook of Middle English, Fifth Printing, Corrected and Augmented. Translated by James A. Walker. Baltimore: The Johns Hopkins Press, 1968.

Wright, Joseph and Elizabeth Mary Wright. An Elementary Middle English Grammar, Second Edition. London: Geoffrey Cumberlege, Oxford University Press, 1928.

Wynne, Patrick H. Goldogrin Pronouns. Published on The Cottage of Lost Play, Dec. 5, 2003.

See also the general Tengwestië Bibliography.

  • 2008-11-08 14:42:18: Formatting changes only: converted text to Gentium/Basic, deprecated all Gentium tags, converted combining diacritics to modifiers where possible

Copyright ©2004 Patrick H. Wynne

Quotations from the works of J.R.R. or Christopher Tolkien are the copyright of their publishers and/or the Tolkien Estate, and are used here with their kind permission. The word TOLKIEN is a registered trademark of The J.R.R. Tolkien Estate Limited. The characters and scripts of Tolkien’s invented languages and works in those languages are the copyright of the Tolkien Estate.


First published on April 1st, 2004

Last modified on Wednesday, February 26th, 2020 6:52 pm PST

This page has been visited 6523 times since April 1st, 2004

Page design by and copyright ©2004 Carl F. Hostetter

Site owned and operated by Carl F. Hostetter