Mixed mutation was a system first proposed by David Salo to explain some irregularities in the use of the genitive preposition en. In his book, Gateway to Sindarin, he described the effects of mixed mutation as that of a soft mutation followed by a nasal mutation (GS/79-80). Although the genitive preposition en is the most notable example of mixed mutation, it is not the only one.

Tolkien considered several etymologies for en, such as ancient ena or alternately ani > eni > en (PE17/97). For purposes of this discussion, all that matters is that its original form (a) ended in a vowel that was later lost and (b) included a nasal; for a more detailed exploration of the origin of en see the entry on the genitive. In this entry we will represent the lost vowel of en with the schwa ǝ: enǝ.

This means that, prior to the point when short vowels vanished before morpheme boundaries, the mutations of en developed the same way as soft mutations. However, after the vowel was lost, the nasal came into direct contact with the consonant producing further nasal effects. One of the other prepositions with (possible) mixed mutations provides a clue on these specialized effects:

S an, dative chiefly with pronouns or persons. < ana, hence vocalic mutation, but takes form m before m, b (PE17/147).

The development of this preposition an(a) would be similar to that of en(ǝ), and hints at special forms before nasal m and voiced stop b. As we will see below, mixed mutation differs from soft mutation primarily in its lack of mutation for voiced stops b, d, g and nasal m.

Voiceless Stops (p, t, c): For voiceless stops, the consonant after the vowel follows the ordinary process of soft mutation, because the nasals came in contact with the stop only after the sound change whereby the stop would be spirantalized. However, the nasal comes in contact before the sound changes whereby nasals vanished at word boundaries before a single consonant, so that the nasal then vanished. The result would be enǝ·t > enǝ·d > en·d > e·d. There happens to be an example of this development:

• Narn e·Dinúviel “Tale of the Nightingale (Tinúviel)” (MR/373).

Compare this to the nasal mutation development: in·t > in·th > i·th, *i thinúvil “the nightingales”.

Voiced Stops (b, d, g): For voiced stops, the consonant after the vowel follows the ordinary process of soft mutation up until the point where the vowel is lost between the consonants. The result would be enǝ·d > enǝ·ð > en·ð. At this point we would need to determine how the nasal and the voiced spirant interact when they come in contact. Fortunately we have a couple of clear examples to guide us:

• Narn e·Dant Gondolin “*Tale of the Fall (dant) of Gondolin” (MR/373).
• Methed-en-Glad “End of the Wood (glad)” (UT/153).

The first example indicates that the dental spirant ð was restopped after the nasal, and then the nasal was lost just as it was with nasal mutations: enǝ·d > enǝ·ð > en·ð > en·d > e·d. The second example indicates that the velar spirant ʒ (IPA [ɣ]) was likewise restopped, as in enǝ·gl > enǝ·ʒl > en·ʒl > en·gl, but in this case the nasal was not lost because it appeared before a cluster, as was also the case with nasal mutation. Thus it seems that the voiced spirants restopped after nasals, with the net result that voiced stops effectively do not change under mixed mutation.

Ancient Nasalized Stops ([m]b, [n]d, [n]g): With ancient nasalized stops, the likely development is: enǝ·mb > en·mb > em·mb > e·mb. As with nasal mutation, the double-nasal prevents the loss of the nasal before the voiced stop, but then the long nasal shortens. The attested examples mostly fit this pattern:

• Narn e·mbar Hador “*Tale of the House (mâr, [mb-]) of Hador” (MR/373).
• Taur e-Ndaedelos “Forest of the Great Fear (daedelos, [nd-])” (LotR/1134; PE17/151).
• Bar-en-Danwedh “House of Ransom (danwedh, [nd-])” (S/203).
• paeth e Ngolodhrim “*speech of the Noldor (golodhrim, [ng-])” (PE17/126).

The third example may represent a variant in the split between the preposition and the mutated noun, en-d rather than the expected e-nd as seen in the second example. The last example appears after a variant e golodhrim, perhaps reflecting some uncertainty on Tolkien’s part; it is also a class plural form which may also have a different effect; see Plural Genitives below.

Sibilants and Spirants (s, h, f, th): We do have an example of a mixed mutation for breath h in a discussion of origin of the word heryn “lady”:

• Roch e Cheruin “Steed of the Lady (†heruin)” (PE17/97).

It demonstrates the expected soft mutation of h (as the result of an inhibited sound change for non-initial ch) followed by the nasal loss of n: enǝ·ch > en·ch > e·ch. Presumably something similar would happen for s, but we have no examples: enǝ·s > enǝ·h > en·h > e·h, as in *e had “of the place (sad)”. Both f and th would not mutate, but we would expect to see nasal loss as in enǝ·f > en·f > e·f. Our only apparent example seems to contradict this:

• Taur-en-Faroth “Hills of the Hunters (faroth)” (SA/faroth).

However, the lack of nasal loss in this example is probably because it is a plural genitive (see below).

Nasals (m, n): We have no examples of mixed mutation for m. However, David Salo suggested (GS/79) the still-nasalized spirantal ṽ would be renasalized to m: enǝ·m > enǝ·ṽ > en·ṽ > en·m > e·m, as in *e mereth “of the feast (mereth)”. As for n, it would show no mutation, but we would expect to see nasal loss. However, this is contradicted by our only example:

• Haudh-en-Nirnaeth “Mound of Tears (nirnaeth)” (S/197).

Unlike with faroth above, we don’t have the excuse of this being a plural form, so it isn’t clear what is going on here.

Voiceless Sounds (hw, lh, rh): The phonetic developments here are unclear. On a purely phonological basis we would expect to see en·chw, en·thl, en·thr as the mixed mutations of hw, lh, rh. However, we have what appears to be one possible example of a mixed mutation for a voiceless liquid that contradicts this:

• Narn e·’Rach Morgoth “Tale of the Curse (*rhach) of Morgoth” (MR/373).

Here ’rach seems to be a mutated form, but it can’t be from **grach, because mixed mutation does not mutate voiced stops. One plausible explanation is that the unmutated form is rhach as suggested by David Salo (GS/284). If so, it is an example of the mixed mutation of the voiceless liquid rh. The general consensus among Neo-Sindarin writers is that this pattern extends to voiceless hw, lh as well: *e-’winn “of the birch (hwinn)”, *e-’lûg “of the snake (lhûg)”. We know that the nasal mutations thr, thl of lh, rh are archaic (PE17/147), so perhaps something similar happened with mixed mutation.

Plural Genitives: In a discussion of genitive en (PE17/97), Tolkien said that its “plural form” is enan > en n/, suggesting the plural genitive causes nasal mutation but preserves the nasal in en. Indeed, we do see a number of plural genitives where the preposition en is preserved:

• Bar-en-Nibin-noeg “House of the Petty Dwarves (Niben-nog-plural)” (UT/100).
• Haudh-en-Ndengin “Mound of the Slain (dangen-plural, [nd-])” (S/197).
• Taur-en-Faroth “Hills of the Hunters (fara- + class-plural)” (SA/faroth?).

One counter example is paeth e Ngoloðrim “*speech of the Noldor (golodh-class-plural)” (PE17/126).

These examples are listed here mainly to explain why such plural forms do not show the expected nasal loss because of the presence of the extra n, as in: enǝn·n > enǝ·n > en·n (with loss of the first nasal but not the second). With a nasal mutation it would be enǝn·t > enǝn·th > enǝ·th > en·th or enǝn·d > enǝn·n > enǝ·n > en·n, as in *en·thail “of the feet (tâl-plural)”, *en·nui “of the nights (dû-plural)”. The net result is mutation as with nasal mutation but the preposition en being fully preserved (never losing its n). In pronunciation there is probably a noticeable pause between the en and the mutated plural, marking the lost vowel. See the entry on the genitive for further discussion.

Summary: Mixed mutation behaves very much like soft mutation except (a) voiced stops and nasals are not mutated (or more exactly their original forms were restored) and (b) ancient nasalized stops are fully restored. In terms of its effect on the n in en, it behaves like nasal mutation, causing nasal loss before most consonants but being preserved before clusters. Thus my recommendations for Neo-Sindarin mixed mutations are:

• Initial voiceless stops (p, t, c) become voiced stops (b, d, g).
• Voiced stops (b, d, g) and nasals (m) are unchanged.
• Spirants (s, h) become “softer spirants” (h, ch).
• Ancient nasalized stops ([m]b, [n]d, [n]g) are fully restored to mb, nd, ng.
• Voiceless sounds (hw, lh, rh) are voiced to ’w, ’l, ’r (archaically chw, thl, thr).