Compounds in the Noldorin of the Etymologies


Abbreviations used:

LR: The Lost Road and other writings
WJ: The War of the Jewels
VT: Vinyar Tengwar
PE: Parma Eldalamberon

Q: Quenya
N: Noldorin
ON: Old Noldorin
CE: Common Eldarin

Introduction

The Etymologies as published in [1,2,3] contain more than 400 words which are transparently compounds of two or more elements, a good deal of them names. These elements occur in various degrees of tightness (to all of which we will in the following refer to as compounds), we observe forms which are transparently Noldorin phrases, e.g. mîr na Nauglin '*jewel of the dwarves' (LR:375,VT45:37); we find similar expressions joined with hyphens indicating a greater degree of tightness, cf. Taur-nan-Erig 'Eregion' (LR:356) or Toll-ereb '*lonely isle' (LR:394), then compounding into unitary words were nevertheless the individual words can be recognized as valid Noldorin words, cf. Eredengrin '*iron mountains' (LR:379); then compounds in which the phonological changes have significantly altered the shape of the compound (as opposed to the uncompounded words), cf. Maglaður 'Black-sword' (LR:371) (instead of magol ðûr which we'd expect as an adjective phrase) and finally words and names which cannot be interpreted as compounds of Noldorin elements at all but which have to be seen as compound words based on Common Eldarin roots, cf. Delduthling '*fear-night-spider' (LR:354,355,386,396, VT45:11) from DYEL, DO and SLIG with a CE ending. Hence, an analysis of these compounds can provide interesting insight into topics ranging from Noldorin syntax to the developments in phonology from Common Eldarin to Noldorin.

One should note at this point that the linguistic scenario of the Etymologies is often not consistent with what we know from later sources. For example, in WJ:370 Doriath is described as a development from a genitive Dôr Iâth 'Land of the Fence'. While it is still a genitive phrase in the Etymologies, here the literal translation is given as 'Land of the Cave' (LR:358,376,VT45:14,16) - which is evidently Dor-i-'ath with gath 'cavern' (ibid)).

It would likewise be wrong to treat the Etymologies as one consistent source - even within the Etymologies we have ample evidence for changes in Tolkien's ideas, and even if some discrepancies are resolved by older forms being crossed out, that doesn't hold for all forms. We find e.g. mindon derived from minitaun in LR:373 but from minitunda in LR:395, Haldir is derived from SKAL and glossed 'hidden hero' in LR:354,386 but from SKEL and glossed 'fur-hunter' in VT46:14. Thus, there is no a priori reason to assume that the linguistic scenario must be exactly the same throughout the Etymologies. Consequently, early replaced forms will be included in the analysis below, especially when they indicate an interesting external development of the language.

With regard to the method of analysis: The aim is to stay as closely as possible to Tolkien's material, i.e. if several variants of a form are found, they are usually all listed. Arguments based on historical (Old Noldorin or Common Eldarin) forms are usually presented using examples where Tolkien provides a derivation, except if that is not possible a suggested derivation is presented instead if it is reasonably certain. No attempt is made to provide historical forms for all compounds. Likewise, the translation given by Tolkien is provided first, wherever there is reason to believe that this is not a literal translation, it is followed by an attempt of a more literal translation to make the nature of the compound more apparent. Since compounds are frequently more complex than a union of two Noldorin words, the elements of a compound are not given as independent Noldorin words; it makes no sense to analyze e.g. the name Huor as hûn + gorf; instead, the individual elements in the hsitoric stages of the language have to be identified, joined and evolved together. Old forms in the external history of the language (i.e. crossed out and rejected entries) are usually presented alongside (if they were replaced by another entry with a brief note), especially if they indicate an interesting development in external history. The placement of the compounds into distinct groups is certainly to some degree arbitrary, it serves the sole purpose to illustrate some structures. A compound could well belong to more than one group, nevertheless each form appears only once in the lists.

Mutations and assimilations

It is rather difficult to define a difference between mutations and the various assimilations of consonants, but we may go with Tolkien's implicit definition used in the mutation tables in the Early Noldorin Grammar [4] where mutation is the process of altering the first consonant of the second element as a result of a prefixed element ending in a vowel or a prefixed element ending in a relic consonant (for example the preposition o 'from' (WJ:366) ends in a relic d.

Consequently, we might think of assimilations as processes altering the ending of the first element along with the start of the second one (according to this tentative definition, quite a number of developments shown in Sindarin mutation tables for stop or nasal mutation e.g. in [5] would actually be classified as assimilations, for example ed s- > es s- does not alter the second consonant but assimilates the first one instead). We can show the difference using the development of the cluster -n-t- in compounds. There are three possibilities:

Mutation and assimilation can (and probably often do) coexist - we can explain derivations like andatekta > andeith 'long-mark' (LR:391) as syncope and mutation to an intermediate *and-dektha followed by the assimilation ndd > nd. In other cases we can be less sure. For example, Edegil 'great bear, *seven-stars' (LR:379) could be explained from odog-gil as lenition *odog-'il or as the result of an assimilation gg > g

It seems that it is a good assumption that mutations act first, assimilation occurs later - often we see assimilations in variant Noldorin forms whereas there is seldom a mutated and an unmutated variant attested.

Mutations

It is beyond the scope of this essay to establish the complete pattern of mutations, for the most part it agrees with what is commonly assumed for Sindarin. However, there are a few cases where more attention to detail is rather fruitful, among them the behaviour of nasalized stops and the mutation pattern seen after stops and liquids.

The pattern of mutations is difficult to predict. In loose compounds, probably from recent contractions of Noldorin phrases, lenition or sometimes no mutation is observed, all other types occur in tighter compounds though. In tight compounds however the outcome is not predictable. Indeed, we even find different patterns as variants, e.g. Certhan next to C(e)irdan 'shipbuilder' (LR:365,390). The most likely story external explanation is that the system of sound changes is not based on hard rules but rather on possible developments, allowing Tolkien to experiment and settle for the form that would most appeal to him. The story-internal explanation for lenition as a variant to other mutation types is readily found in the final vowels of the first element which occured frequently in Old Noldorin, presumably leading to lenition of the second element before being lost. Examples include lhasbelin < lassekwelênê 'autumn' (LR:366,367,VT45:24), Gothmog < Gothombauk '*dread-tyrant' (LR:359,372), andeith < andatektha 'long-mark' (LR:391), Fingon < Phindekâno '*skill-valour' (LR:362,381), mindon < minitaun 'tower, *isolated hill' (LR:373) and Glingol < lingikâle '*hang-light' (VT45:19,27).

On the other hand, we observe other mutations if the primitive elements are not separated by a vowel, cf. ifant < infant < inpanta < yen-panta 'year-full' (LR:358,400,VT46:23), echui(w) < et-kuiwê 'awakening' (LR:366), Mormagli < morn-magli 'Blackbear' (VT45:28) and nelthil < neltildi 'triangle, *three-point' (LR:376,393). Thus, we may speculate that the difference between Certhan and Cirdan reflects an underlying difference in derivation from *cir-táno and *cirja-táno.

To put it mildly, the behavior of nasalized stops seems difficult to understand. If they are placed between vowels by intensification of the CE root by prefixion of the sundóma [6], we can study the possible outcome of MB in LR:372 where we find ambar, amar 'Earth' from MBAR. Thus, intervocalic -mb- is allowed to stand or contracted to -m-. This does not seem to be the case if the cluster is placed between vowels by compounding, as we have annabon < andambundâ 'elephant' (LR:372), not **annamon or **annambon; see also Angband < *angâ-mbandô 'Hell, Iron prison' (LR:371, VT45:6,33). But unfortunately we also find Gothmog < Gothombauk '*dread-tyrant' (LR:359,372), indicating that also -m- is a possible outcome. The ultimate problem for a predictable classification is then posed by Findobar, Findabar < phind-ambar '*skill-home' (LR:372,381) where inside the element ambar the development -mb- > -b- occurs (thus, all three possible outcomes are attested for ambar).

The situation looks similar for other nasalized stops, we see for example -nd- > -nn- in Balannor < Balandor < bala-ndore 'Valinor' (LR:350) or -nd- > -n- in Curunir 'a man of craft, wizard' (VT45:24) but e.g. ceredir 'doer, maker' (LR:354) - the variation here may be due to a partial blending of DER and NDER however. Likewise, we see ng > g in Talagant, Talagand < tyala˝gando 'harper, *play-play' (LR:377,395) but ng > ng in Fingolfin < *Phind(e)-ngolfine '*skilled in magic skill' (LR:381) - the latter may however rather be interpreted as a contraction -ndng- > -ng.

In WJ:367 we learn that with regard to the preposition ed: [It] retains its consonant in the form ed before vowels, but loses it before consonants, though es, ef, eth are often found before s, f, th.. It is probably worth noting that in the only compound using this element before s- in the Etymologies, this is not what happens: In LR:356 we find ethir < *et-sir 'mouth of a river'. Since ts > tth > th(þ) is a regular development, cf. litse > litthe > lith 'ash' (LR:369) this really ought to be read as eþir rather than **et-hir (as wrongly suggested previously by myself and others). Thus, the 'often found' variant **essir is not realized in this case.

Contrary to the suggestion made in Helge Fauskanger's Sindarin mutation tables, the conjectured effects of liquid mutation -ld- > -ldh- and -lb- > -lv- are not usually seen in the Etymologies. In fact, in most of the cases no mutation is seen in spite of the fact that lenition could be an option quite independent of the last consonant of the first element. We find no mutation b- > v e.g. in Elbereth 'Varda' (LR:355), Gilbrennil 'Varda, *star-lady' (LR:358), Thar-bad '*crossway' (LR:392) (the root for 'path' in the Etymologies is BAT rather than PAT - the latter is associated with 'open') and Dolbaran '*brown-knob' (LR:351); only in two cases is mutation observed: herven 'husband' (LR:352,364) and herves 'wife' (LR:352). Likewise, d- > dh- is never observed in compounds, but we have Boldog 'Torment-Slayer' (LR:375,377), Delduthling'*fear-night-spider' (LR:354,355,386,396, VT45:11) and meldir, meldis 'friend' (LR:372) for no change of the cluster following -l. It should be remarked that it is actually reasonable to look for these changes in compounds (not only prepositions and short prefixes) as other manifestations of the liquid mutation occur rather frequently, cf. -lt- > -lth- in Gilthoniel, Gilthonieth 'Varda, star-maker' (LR:358,390) (derived from TAN in the Etymologies) or Barthan 'Aule, Earth-builder' (LR:358,372,390). We may conclude that -ld- and -lb- are usually allowed to stand in Noldorin compounds.

Assimilations

The most common assimilations are:

Some special developments

The formation of compounds in Noldorin is significantly more complex than merging two Noldorin elements and leniting the second word. As a rule of thumb, compound formation is more complex if the compound is derived from the older layers of the language. In addition to mutations and assimilations, various other phenomena connected with the evolution history of the compound may arise.

Some of the most important of these developments are:

Genitives in normal order

From non-compounded genitives and loosely compounded names we can read off the usual order of the genitive phrases in Noldorin: the word in genitive follows the one it describes. We may note at this point that there occur no phrases or loose compounds in the Etymologies where this order is inverted (inverted order only occurs in tight compounds), so it seems justified to call the order with the genitive noun second 'normal' and the opposite 'archaic'. This agrees well with the linguistic scenario laid out much later for Sindarin in WJ:370.

Genitives must have been marked by case endings in Old Noldorin, this is apparent from the existence of an ON ending -o marking the 'old partitive' (VT45:17) and the derivation thoron < thoronen from the ON genitive singular (evidently here -n (LR:392)). Apparently at this time this was likewise the ending in Quenya, cf. Q: sorne > sornen (ibid). From the fact that Dor-na-Thuin 'Land of Pines' is the Noldorin translation of Ilkorin Dor-thonion (ibid) one might infer that -ion is a gen. pl. valid only for Ilkorin - but in fact Eredwethion '*mountains of shadows' is clearly marked as Noldorin (and even contrasted with a different Ilkorin form), so it seems that here #gwethion is the genitive plural of gwath. The same ending seems to be relevant in Eregion '*of Hollies' (LR:356).

The particle na can be used to mark the genitive. A variant nan is attested in Taur-nan-Erig 'Eregion, Forest of Region, *forest of hollies' (LR:356) - if this variant is supposed to indicate plural, it has to be optional as we also have Dor-na-Thuin 'Land of Pines' (LR:392). Neither gloss suggests that either na or nan would imply any degree of definiteness. From the fact that thoron is described as genitive singular of thôr (LR:365,392) we can infer that in Old Noldorin na must have been a preposition followed by the genitive. This may also explain the plural pattern seen in mîr na Nauglin (LR:375,VT45:37). However, if Eregion is indeed a valid Noldorin genitive, then Taur-nan-Erig (LR:356) would argue against this idea. The formation of the genitive using na(n)is attested in the following examples:

Alternatively, genitives using the definite article i in singular or in in plural are seen. Usually the use of the definite article seems to imply some measure of definiteness, the only exception being HauðiNdengin 'hill of slain' (LR:363,375). Here we may speculate that the article is used as a means to indicate that the participle dangen is here used as a noun 'slain ones' rather than an adjective. The attested examples are:

The most commonly found pattern is the genitive implied by word order and lenition - the noun in genitive is found second and usually lenited. From Cilthoron '*pass of the eagle' (LR:365,392) (thoron < thoronen is gen. sg. we can infer that this goes back to the Old Noldorin case inflection which has been lost in most Noldorin words. We find this genitive pattern in various degrees of tightness, but alternative forms like Rathloriel, Rath Loriel '*golden bed, *path of gold' (LR:368,383,VT46:11) suggest that they share a common evolution. Evidently this formation can express definite genitives (cf. taur egledhrim 'King of the Exiles' (LR:389,VT46:16)) as well as indefinite ones (cf. Cilgalad 'Pass of Light' (LR:365)) Attested examples include:

Note that for some of the examples presence of lenition cannot be established. The lenited form of rh is r- in the linguistic scenario of Noldorin as apparent from e.g. rhass > i-rass (LR:363). A relatively large number of examples however does not show lenition at all. As the replacement Din-Garadras > Din-Caradras (LR:354,VT45:9) indicates, this may reflect the change from genitive lenition in the Noldorin scenario to the unlenited genitives seen in later Sindarin. A further number of instances might be explained by the observation that d > ð, dh is not carried out consistently and hardly ever in compounds, cf. among others the variations Geleið/Geleid (LR:373,377,VT46:3).

Genitives in inverted order

The Etymologies contain a large number of compounds which can be interpreted as genitives, albeit with the noun in genitive in front position. In some cases the nature of the compound as genitive is apparent from the translation, cf. camland 'palm of hand' (LR:367), for most words this is however not the case and the classification as a genitive is essentially an educated conjecture, see e.g. daedelu 'canopy' (LR:391) where we might perhaps literally translate '*hood of shadow'.

It is probably a reasonable assumption that this group of compounds represents older genitives. We don't have a single example of a non-compounded phrase or variant; likewise there is no example of a hyphenated compound. Instead, for a relatively large number of forms the derivation from primitive forms is given, indicating explicitly that we don't deal with recent compounds.

The pattern of consonant changes is hardly predictable - for the same CE affix tanô we find both lenition C(e)irdan 'shipbuilder' (LR:365,390) and liquid mutation Certhan (earlier variant of Cirdan) (ibid) or Barthan 'Aule, Earth-builder' (LR:358,372,390). Likewise, for the combination nasal + stop, both nasal mutation ifant < infant < inpanta < yen-panta 'year-full' (LR:358,400,VT46:23) and lenition Thorondor 'King of Eagles' (LR:392) (from taur) is seen.

In one example, the noun in genitive is accompanied by an adjective. We find Gondolin 'heart of hidden rock' (LR:355) (apparently from gond+dol+ind).

We may sort according to forms which are consistent with a simple development, i.e. either no change or lenition and those who show a more complex development. Into the first group can be counted:

The second group consists of the following forms:

Adjectives in trailing position

There are a number of compounds which can be readily interpreted as nouns followed by an adjective. In more than 2/3 of the instances where we are able to tell the difference the trailing adjective is lenited in the compound, we may assume that this is the favoured pattern. There is attested instance which would require complex phonology (though there are a few assimilations, cf. gg >g in Tarag(g)aer 'Ruddihorn' *deadful horn' (LR:391) or dhth > th in Arthoren 'Fenced Kingdom' (LR:360,VT45:17)), thus we may assume again that these compounds are from a recent date. However, in a few cases the preservation of an ON final vowel suggests an older formation, cf. Maglaður, Maglaðhonn 'Black-sword' (LR:371). We observe number agreement between noun and adjective if the noun is plural, cf. Eredengrin '*iron mountains' (LR:379) with angren 'of iron'. Thus, we have indication that the adjective in Noldorin follows the noun, agrees in number and is often lenited just like in Sindarin. The complete list of forms is:

Adjectives in front position

This is the second most numerous group of compounds found in the Etymologies - only compounds with the noun in genitive in front position occur more often. This makes it likely that in Old Noldorin (like in Quenya) the adjective could frequently come before the noun and that many of these phrases developed into compounds. In two cases we even see the adjective preceding the noun: Sarn Athrad, Sarnathrad '*stony ford' (LR:383,385) and Sern Gebir, Sarn-gebir 'lone stones, *stony ridges' (LR:363,VT45:20) ( from ceber 'knob, top; ridge'). In both cases, more closely compounded variant forms are attested, but it seems that in some cases the Noldorin adjective may precede the noun even in modern expressions. Note that #cebir is lenited in this case, lenition marks a close grammatical relationship, not the adjective as such.

Many are consistent with lenition or no mutation:

A smaller group requires a more complex explanation, for example nasal mutation or some assimilations:

Prefixes

Compounds with short, prefixed elements are rather common. In many cases the prefixes have the meaning of prepositions, or is explicitly listed as both a prefix element and a preposition in LR:379 with the meaning 'above, over'. However, it seems more likely that at least some of these compounds have to be interpreted as old formations rather than recent contractions for a number of reasons:

First, we know (albeit from later sources) that prepositions do not agree in number, we find e.g. o galadhremmin ennorath 'from the tree-woven lands of Middle-earth' (LOTR, RGEO72) where the preposition shows no sign of being pluralized, cf. the singular Celebrimbor o Eregion 'Celebrimbor of Hollin' (LOTR). Nevertheless, i-affection (in plural or for other reasons) is observed with some prefixed words, cf. Alchoron pl. Elcheryn 'Ilkorin, *outside Kôr' (LR:349,367,VT45:5,25), gwador pl. gwedeir 'brother' (LR:394) or esgeri 'cut around, amputate' (LR:379) (from os-) - though we also see counterexamples, cf. †adlegi '*to free' (VT45:27) or ortheli 'roof, screen above' (LR:391) (no un-affected prefixes in plural are attested).

Then, prefixes can appear inside compounds and are then eroded by mutation. For example, aronoded 'withour reckoning, countless, innumerable, endless' (LR:349,378,399) is composed of ar + gonoded where the second element contains in turn the prefix go- which is reduced to 'o- by mutation. A similar development is seen in Tavrobel '*woodpecker-village' (LR:380,390) which in turn contains gobel 'walled house, village' (LR:380).

Finally, for some forms the underlying phonological development is quite complex. gwa-star 'hummock' (LR:388,399) is from thâr 'grass' (ibid) - but presumably the compound was formed before the development st- > th- word-initially, so this cluster is preserved. For eithel 'spring, *out-run' (LR:363) a very interesting development is given in which the first consonant cluster switches, etkele > ektele > eithel.

Thus, clearly a fair share of compounds must be explained by more than a contraction of Noldorin prepositional phrases - for others we don't quite know. We find compounds consistent with no mutation or lenition:

Some forms require liquid or stop mutation for their interpretation:

Finally, a few more require more complex development not expressable in terms of the standard set of consonant mutations:

Numbers

A few compounds involving numbers are found. In all instances, the number constitutes the first element. The absence of compounds involving numbers in the second element is rather telling: While a fair share of adjectives were found in second position, in essence indicating that in modern Noldorin the adjective usually followes the noun, and this is retained in recent compounds, numbers may in fact not be grammatically adjectives (as some writers have assumed in the past). The hypothesis that numbers regularly precede the noun agrees in fact well with the only direct evidence we have for the use of numbers in a Noldorin sentence: In [7] we find Lheben teil brann i annon ar neledh neledhi gar godrebh '*five feet high [is] the door and three may walk abreast' where the number precedes the noun in lheben teil '*five feet'. All in all the list of compounds involving numbers is rather short:

Descriptive compounds

In a compound of this type, both elements are descriptions of the object described by the compound. in [8] David Salo develops a more sophisticated classification, but for the purpose of this essay it is enough to note that these compounds, unlike all classes treated previously, can in fact not be assumed to have originated easily as contractions of independent phrases, they must have been compounds early on. For example, basgorn 'loaf' (LR:372) is 'round' and 'bread' - both elements describe the object, whereas angweð 'chain' (LR:397) is a bond of iron, the first element describes the second one. Thus, while angweð may be explained as originating from case-inflected anga in front of weda, no such explanation is possible for basgorn. As a consequence, we would expect that compounds of this type had to originate rather early, and indeed Tolkien frequently provides etymological forms.

Descriptive compounds involving verbs

In some cases, verbs seem to be part of the compound. This i smost clearly evident from megli < mad-lî 'bear' (LR:369) which seems to be nothing but the short description 'eats honey'. In other cases, a translation would require a participle (although only the verb sten is part of the compound), cf. Glingol < lingikâle '*hanging light' (VT45:19,27). In all cases the verb stem is the first element. However, since only roots are found and not inflected verbs, it would be premature to use this to draw any conclusions on word order in Noldorin. The (rather short) list is:

Suffixes

Some elements in compounds resemble suffixes rather than independent words. Possibly the most prominent case is -iand, denoting a region or a land, for which no independent form is known. Presumably it roughly corresponds to the English '-ia' as e.g. in 'Utopia' which can't exist as an independent element either. If so, Beleriand might literally be translated as 'Balaria' (Tolkien translates Rohan as 'Hippia' in Letters:178). Other than that, -dir might be grouped in here, although compounds involving this element have been listed in a different group if e.g. the translation explicitly suggests a genitive. It is unclear if this group should be treated as compounds at all.

Unclear formations

Several compounds cannot be analyzed with reasonable certainty within the context of the Etymologies, at times because one of the elements cannot be identified, at times because they may rather be Doriathrin or Quenya. They are given here with partial translations and remarks for the sake of completeness.

Conclusions

As this investigation has hopefully shown, the formation of compounds as seen in the Etymologies is significantly more complex than choosing two elements and joining them with the use of a mutation table. Indeed, the very difference between mutations, assimilations and regular phonological evolution becomes blurry at times when discussing older compounds.

In general, compound formation in later Sindarin seems to follow similar trends, although there are at times pronounced differences. Perhaps most prominent among them is the non-lenition in Sindarin genitive phrases. We have, for example, Aran Moria 'Lord of Moria' (LOTR) or ennyn Durin 'Doors of Durin' (ibid). This has some implication for genitive compounds in Sindarin, for example the days of the week appear in unlenited form (as opposed to their Noldorin counterparts described above), cf. Orgilion, Orgaladhad, Ormenel or Orbelain (LOTR). A comprehensive investigation of the similarities and differences between compounds in Noldorin and Sindarin is however beyond the scope of this essay.

Acknowledgements

A warm 'Thank you!' goes to Erna Aichholzer for proofreading the article, pointing out omitted and wrongly typed references and sparking some interesting discussions.

References

[1] The 'Etymologies' in 'The Lost Roand and Other Writings', edited by Christopher Tolkien
[2] Addenda and Corrigenda to the 'Etymologies' part I , Vinyar Tengwar 45, edited by Patrick H. Wynne and Carl F. Hostetter
[3] Addenda and Corrigenda to the 'Etymologies' part II , Vinyar Tengwar 46, edited by Patrick H. Wynne and Carl F. Hostetter
[4] 'Early Noldorin Grammar' in 'Early Noldorin Fragments' , Parma Eldalamberon 13, edited by Christopher Gilson, Bill Welden, Carl F. Hostetter and Patrick Wynne
[5] Sindarin - the Noble Tongue by Helge K. Fauskanger
[6] Intensifying prefixes in the Etymologies by Thorsten Renk
[7] 'Tolkien - Artist and Illustrator' by Wayne G. Hammond and Christina Scull
[8] 'A Gateway to Sindarin' by David Salo

Thorsten Renk
thorsten@sindarin.de

Back to the index