On the diphthongs ei, ai in Noldorin and Sindarin

Roman Rausch

Jan. 5th 2008

tm, tn > tw (nt) > nt, tt
What a muddle.




There are some phonological details in Tolkien’s languages where we see an apparent hesitation between two possibilities. Sometimes both of them are allowed, sometimes only one is given and conceptions change to and fro in the course of time.
One such detail concerns the diphthongs ei and ai in Noldorin and its successor Sindarin. This article deals with the oscillation between the two throughout three main stages – Early Noldorin (c.1923), the Noldorin of The Etymologies (c.1937 and later) and finally Sindarin (roughly LotR and everything after). Similar variations are ae/oe (or ai/oi) [1], œ/e or -dl-/-gl-.
This matter should be of especial interest for regularization attempts of Noldorin vocabulary into Sindarin.

1  Mechanisms of creation

At first, one has to determine what to look for. We will only consider cases where both ei and ai may be yielded. For example, the change a > e by i-affection (mostly pluralization) is irrelevant here, but an original non-syllabic ı̯ in the adjacent syllable may cause both ei and ai. See [3] for a detailed discussion of Noldorin/Sindarin vowels.

1.1  original diphtongs

The direct change ei > ai can be observed in Early Noldorin. Note that the reverse change ai > ei never occurs (neither in Early Noldorin nor later). Original ai remains as such in Early Noldorin, changes > oe in the Noldorin of The Etymologies and > ae in Sindarin. Therefore, it will not be included.

1.2  vocalization

Consonants like k, g, χ, gh, p are vocalized to i mostly before original t, s. In Early Noldorin n may be vocalized as well. Vocalization is of interest if the preceding vowel is e < e, i, so that ei is created. If the preceding vowel is a, the change is aC > ai in Early Noldorin and aC > ae in the Noldorin of The Etymologies, as well as Sindarin.

1.3  vowel mutation

The diphthongs ei, ai may be created from every other vowel by vowel affection (= vowel mutation) and i-infixion in the plural; or often due to the suffix -yā. In the case of u, i it requires the lowering by a-affection yielding o, e which merge with original o, e [3]; hence:

1.4  contraction ehi > ei

The combination -esi changes > -ehi by lenition, which in its turn becomes -ei (Noldorin of The Etymologies).

2  Early Noldorin

Early Noldorin basically includes two wordlists written around 1923, as presented in PE13. Only words with primitive forms that are directly given by Tolkien or deducible from Qenya cognates will be presented below (they form the majority).

2.1  vocalization

2.2  mutation of a

2.3  mutation of e, e < i

2.4  original ei

2.5  Discussion

The pattern in Early Noldorin seems to be very regular and can be easily extrapolated. It appears that at first, ei is created by any of the mechanisms. It becomes ai in the ultimate syllable and remains ei otherwise. So for example aith ’thorn, *spear’ < ekte shows ei > ai in the ultimate syllable (which is here the only one), but the plural *ektin- becomes eithin because ei is now non-ultimate. The only case where ei is allowed in the ultimate syllable or in monosyllables, is when it appears before r by pluralization of a, although the variant development a > e is also possible. Hence e(i)rch, he(i)rdh, ge(i)rth. This is essentially the same system which is found in Welsh.
The only exception is dain, dein, where dein might be an archaic form. Compare also the following Old Noldorin examples all showing the pluralization a > ei (and subsequent ei > ai/e in later Noldorin):

Note that ei/ai is never created by mutation of o, the pattern is rather o > y, i (e.g. odog ’much’, pl. ydig, edyg (PE13:151)).

3  Noldorin of The Etymologies

The next major stage in the external development of Noldorin is the compilation of The Etymologies in 1937 and the following years. It will be seen, however, that the patterns here are not as regular. In fact, The Etymologies are probably not a fully consistent source, having been compiled over a significant span of time. Therefore, percentages of the respective occurrences will be given in brackets.
References will be made by the roots of the respective entries.

3.1  vocalization (60% ei, 30% ei|ai , 10% ai)

3.2  mutation of a (67% ei, 14% ei|ai, 19% ai)

3.3  mutation of e, e < i (92% ei, 8% ei|ai, 0% ai)

3.4  mutation of o, o < u (100% ei, 0% ei|ai, 0% ai)

3.5  contraction ehi > ei (100% ei, 0% ei|ai, 0% ai)

3.6  Discussion

At first one has to note that the combination ei is now more favoured by Tolkien and also appears in the ultimate syllable. In fact, the plural affection a > ei is a very dominant pattern and similarly e, o > ei is met in almost all cases only with the exception of deleted breig, braig.
There is however a variation between ei and ai in about one third of the words where vocalization took place. But actually breitho, eithel, leithio, leithian, teitho all have ei in the non-ultimate syllable. Assuming that ei can never change to ai in such a position, we are left with one sample of ei only, one sample of ai only (aith, although apparently changed >> eith) and three samples with ei|ai.
It may be that Tolkien reintroduced the idea of ei > ai (if not obligatory, perhaps at least an optional development). In fact, Tolkien’s way of presenting entries is to give an earlier form, followed by a later form just separated by a comma – and the ei/ai variants are presented in just the same way.
It should be also noted that Tolkien apparently considered the idea of assigning the variation ei|ai to different dialects. We meet two words mentioned as ’Feanorian’, i.e. spoken in the house of Feanor:

If Feanorian Maiðros equals common Noldorin Meiðros, then the same distinction is probably already true for meið and maið, so that this dialect would effectively turn ei > ai. In fact, the change -ðw > -f is later part of the Northern dialect of Sindarin, see VT41:8, but there is no similar mention concerning ei|ai.

4  Sindarin

The Sindarin stage involves material written during the completion of The Lord of the Rings and after its publication. This is a time span of more than 15 years and so inconsistencies due to different conceptions are far more likely to occur, but I do not believe that a further splitting does any good because there are no such handy sources as the Noldorin wordlists anymore. We are rather dealing with lots of separate notes and several essays, each contributing a couple of samples.

4.1  vocalization

4.2  mutation of a

4.3  mutation of e, e < i (25% ei, 67% ei|ai, 8% ai)

4.4  mutation of o, o < u

4.5  Discussion

The distribution of ei/ai in Sindarin seems to be similar to the Early Noldorin system – ai becomes more dominant again. In particular, the dominant pluralization pattern is now a > ai (attested in a vast amount of examples). An exception is met with eilph < alph, but considering also nern pl. of narn ’tale’ (MR:373) sern, sairn pl. of sarn ’stone’ (TI:366,283, WR:98,132) we may assume that a > e(i) is a common plural formation before r and l; and ei is retained in such a case as in Early Noldorin.

However, it is quite apparent that words with mutated e are mostly given in two variants – one with ei and another with ai – and only a few examples show either ei or ai without a mentioned alternative.
As an explanation one could once again assume that there is a later change ei > ai and the ei-forms are perhaps still allowed, albeit probably being archaic. This goes well with the mention of an intermediate form hein that further shifted to hain. So far the Early Noldorin/Welsh pattern seems to work in Sindarin.
On the other hand two parallel ei/ai-forms are never created by mutation of a. So if we have e.g. *atarī ’fathers’ and gilyā ’silver spark’ in Common Eldarin; then *edeir and geil at some stage in early Sindarin, why is it that the final results are edair and geil, gail, but *edeir does not appear? Perhaps there was another conception with a direct change a > ai and e > ei; and the latter combination could then optionally merge with ai. This, of course, does not go well with hein > hain anymore; but could explain why teleir is mentioned in the same passage with belair, gevair – the former comes from e, the latter from a. The same situation is found in ’Quendi & Eldar’ (a well-considered essay, not just a collection of notes), where forms like levain < lavan, #tail < *talya are found beside Feir < Firya.

A further complication is added by the derivatives of √RIK- *’try’ (PE17:167). At first Tolkien wrote reitha *’strive’ with the apparent past tenses rithantem *’we strived’, rithantin *’I strived’. These forms can be well understood phonologically: CE *riktā > *reχþa (a-mutation and spirantization) > reitha (vocalization of χ). The past tense stem would be *riktant- > *riχþant- > rithant-, now without a-mutation because it only operates with final -ā (PE17:152). But Tolkien then wrote raitha which would be a rare example of ei > ai also in a non-ultimate syllable (another one is deleted bainia- (PE17:149)). Still, an e written over the word seems to indicate that the alternative reitha is also possible.
Since these notes were written hastily and the readings are not entirely certain (according to the editors), raitha is likely to be an experimental form. There is a number of examples with retained ei in a non-ultimate syllable (again according to the Early Noldorin/Welsh pattern) from essays and texts that are not hastily written notes: Ereinion, teithant, Neithan, gleina-, seidia-.
Another exceptional case, namely ei < a appearing in the ultimate syllable and not changing to ai also occurs once, in the name Edhelvein, but it stands against a large amount of examples with a > ai.

In the case of o the dominant mutation pattern is now o > y (and rarely o > ui, o > e) rather than o > œi, ei as in The Etymologies. So the only relevant example here is ogl, pl. eigl. Even within the conception of an overall change ei > ai this should remain unaltered as it is probably still derived via intermediate œi. This means we would have for example *œigl, eir at one point in time changing into eigl, air.

For the ’update’ of Noldorin material into Sindarin I would therefore propose the following recipe:


Helge Fauskanger. Ae or oe? http://www.uib.no/people/hnohf/oe.htm
Helge Fauskanger. The question of nd or n(n) http://www.uib.no/people/hnohf/ndnn.htm
Bertrand Bellet. Vowel affection in Noldorin and Sindarin http://www.tolkiendil.com/langues/english/i-lam_arth/vowel_affection_sindarin_noldorin

update: Feb 2nd 2008 — discussion of raitha/reitha added
update: Apr 9th 2008 —
eilph, Edhelvein added and discussed
update: Sep 11th 2008 —
bainia- added
update: Jun 5th 2009 — a couple of examples, mostly from PE17:140-141 added


This document was translated from LATEX by HEVEA.