XXX PE23/136

Sindarin soft mutation is pervasive in the language and serves numerous functions: marking the direct object of a verb, the modified form of a noun following an article, the modified form of an adjective following a noun, etc. This entry primarily discusses the mutational process itself. The conditions under which mutations occur is addressed in other entries.

The soft mutations of Sindarin are very similar to those of Welsh. Historically, Sindarin soft mutation (also called lenition) arose from the phonetic development of consonants after vowels. This is most noticeable in modern Sindarin with the mutations of consonants after the definite article i “the”: i beth “the word (peth), i dhraug “the wolf (draug). It involves a “softening” of the sound from voiceless stops to voiced stops, voiced stops to voiced spirants, voiceless spirants (s, h) to “softer spirants” (h, ch).

In many cases, the vowel that conditioned the soft mutation was lost, and the mutational process was generalized to cases where no soft mutation would have occurred from a purely phonological perspective. This is the case, for example, with adjectives following nouns, since most nouns originally ended with vowels that were ultimately lost. After these losses, the soft mutation generalized to those rare cases where the noun did not originally end in a vowel. The soft mutations of Welsh have a similar historical origin.

Soft mutations modify the method of articulation of the consonant (voiceless stops become voiced) but not the place of articulation (labials remain labials, dentals remain dentals, etc.). Thus, it is easiest to discuss soft mutation in terms of method of articulation.

Voiceless Stops (p, t, c): The soft mutations of voiceless stops p, t, c are the voiced stops b, d, g. This is a result of the sound change whereby voiceless stops voiced after vowels. This mutation is extremely well attested, and applies even to consonant clusters beginning with voiceless stops, as in Drann the soft mutation of Trann “Shire” (SD/129). Welsh has the same mutation, and Tolkien established the soft mutation of voiceless stops all the way back in Gnomish of the 1910s (GG/7). Some examples from Sindarin:

• peth “word” → beth (LotR/953).
• tíriel “having watched” → díriel (LotR/729; PE17/25).
• calan “green” → galan (RC/349; UT/281).

The second example is of particular interest because the environment where it appears (palan-díriel) would produce -thíriel (nasal mutation) as a purely phonetic phenomenon. Thus it is an example of a mutation that is exclusively grammatical, as discussed by Tolkien in a 1958 letter to Rhona Beare (footnote, Let/278-9).

Voiced Stops (b, d, g): The soft mutations of voiced stops b, d, g are voiced spirants such as v, dh, except that g vanishes completely: i ’aladh “the tree (galadh)”. For clarity Tolkien sometimes marked the lost g with an apostrophe ’, but this is not required and has no affect on pronunciation. This mutation is a result of the sound change whereby voiced stops became spirants after vowels: v, dh, ʒ (IPA [v, ð, ɣ]); later on the voiced velar spirant ʒ (IPA [ɣ]) vanished. This mutation is extremely well attested, and applies even to consonant clusters beginning with voiced stops, as in ’laur the soft mutation of glaur “gold(en)” (UT/253). Welsh has the same mutation except [v] is spelled “f” and [ð] is spelled “dd”; Tolkien established the soft mutation of voiced stops all the way back in Gnomish of the 1910s (GG/7). Some examples from Sindarin:

• beleg “great” → veleg (RC/536).
• dínen “silent” → dhínen (WJ/333).
• gaear “sea” → aear (LotR/238; PE17/27).

The initial combination gw- is a somewhat special case. Like other initial clusters beginning with g, this sound was lost under soft mutation leaving only w-, as in i-wath “the shadow (gwath)” (PE17/41). The historical origin of this mutation is somewhat different, however. In Sindarin’s phonetic history, an initial [w] became [gw], but only at the beginning of the word, not after a vowel. Thus the clitic i- “the” didn’t cause the g in gw- to vanish, but rather prevented its development in the first place.

This is mainly a historical curiosity, however. In modern Sindarin, the soft mutation of initial gw-clusters behaves like other g-clusters: the g vanishes.

Nasals (m): The dental nasal n does not mutate, but the labial nasal m becomes v, just as it does in Welsh (except in Welsh the mutated form is spelled “f”). The net result is the same as the mutated form of b, which introduces some ambiguities into the language: “the queen (bereth)” and “the festival (mereth)” are both i vereth. The ultimate change of m to v was later than that of b to v, with m passing first through a nasalized spirant Tolkien sometimes represented as ṽ but more often as mh. This nasalized ṽ may even have survived in some dialects like North Sindarin and/or Doriathrin prior to the fall of Beleriand (PE17/131, 133).

This is another well-attested mutation for Sindarin, but unlike the soft mutation of voiceless and voiced stops, m-mutation did not appear in the mutation tables of either Gnomish of the 1910s (GG/7) or the Early Noldorin of the 1920s (PE13/120-121). It was established in the mutation charts of Noldorin of the 1930s, however (PE19/18-19).

Sibilants and Spirants (s, h): The voiceless fricatives f, th do not undergo soft mutation, but the sibilant s and the breath h do mutate, to h and ch respectively. These mutations are both well attested for Sindarin:

• sain “these” → hain (LotR/305; PE17/42).
• haered “remote(ness)” → chaered (LotR/238; PE17/25).

These mutations do not occur in Welsh, and are features unique to Sindarin. The mutation of s arises from the (Old) Sindarin sound change whereby intervocalic [s] became [h]. It seems this change occurred differently at morpheme boundaries than it did medially, since medial s > h ultimately vanished, whereas an initial h from soft mutation survived. On the other hand, h-mutation is the indirect result of the sound change whereby initial [x-] became [h-]. As with gw-mutations, the presence of a preceding vowel inhibited this phonetic development, so the soft mutation of h is the result of the prevention of an initial sound change rather than the replication of a medial sound change.

The h-mutation appeared in the mutational tables of Gnomish from the 1910s (GG/7), but s-mutation did not. Somewhat strangely, h-mutation disappeared from the Early Noldorin tables of the 1920s (PE13/120-121), but both h- and s-mutation appeared in the Noldorin tables of the 1930s (PE19/18-19). The h-mutation was deleted from the 1930s Noldorin chart (PE19/18, note #7), but there is solid evidence of it elsewhere, such as N. Aran Chithlum “King of Hithlum” (Ety/TĀ, PE22/33); loose genitives like this caused soft mutation in the Noldorin of the 1930s and 40s.

Voiceless Liquids (lh, rh): Voiceless initial liquids are voiced to r, l by soft mutation in Welsh, but it is not clear exactly what the mutations are for Sindarin. These sound shifts are not mentioned in the Gnomish soft mutations of the 1910s (GG/7), but this is not especially strange since initial voiceless lh, rh were not part of the phonetic inventory of Gnomish. Mutations for lh, rh likewise did not appear in the Early Noldorin mutation table from the 1920s (PE13/120-121), but these sounds did begin to appear in contemporaneous word lists, such as ᴱN. lhith “dust” (PE13/149) and rhód “course” (PE13/152), with deleted gloss “free from”. Given compounds like ᴱN. yrlhith “dustless” (PE13/156) and ᴱN. darhod “not free” (PE13/141), they may have been unchanged by mutation.

Voiceless initial liquids were well established by the Noldorin of the 1930s, however, with an origin and system of soft mutations essentially identical to that of Welsh. In particular, all ancient initial liquids l, r became unvoiced to lh, rh (PE19/19), and their soft mutations were the result of the inhibition of this phonetic development after vowels, much like gw- and h-mutations as described above. The same holds true for Welsh, except voiceless l is spelled “ll”. The clearest Noldorin example is:

• N. rhass “precipice” → i-rass (Ety/KHARÁS), an analogical soft mutation for older †i-chrass [< *i-khrassē].

While writing drafts of The Lord of the Rings in the 1940s, Tolkien gradually decided that ancient l, r were not unvoiced to lh, rh, and many Noldorin words beginning with voiceless liquids were switched back to regular liquids, as in: N. rham “wall” (Ety/RAMBĀ) >> rammas “great wall” (WR/288; RC/512) and N. lham(b) “tongue” (Ety/LAB) >> Lamben “Tongue” (TI/280) and later S. lam (PE17/46; WJ/394). This removed the basic historical foundation for voiceless liquid mutations in Sindarin.

However, unlike Gnomish, voiceless initial liquids still appeared in Sindarin as the result of other sound changes, in particular:

LH represents this sound when voiceless (usually derived from initial sl-) ... RH represents a voiceless r (usually derived from older initial sr-) (Appendix E, LotR/1114).

Indeed, Tolkien used these sound changes to devise new etymologies for Noldorin forms he wanted to retain in Sindarin, such as lhaw “ears”, rhûn “east” and Rhovanion “Wilderland”. Assuming most initial lh, rh in Sindarin were the result of ancient sl, sr, we can deduce the soft mutations of voiceless liquids from the medial developments of these sounds. In Noldorin of the 1930s medial [s] became [θ] before [l], [r], and there is evidence of this for Sindarin as well:

• ✶kasraya > cathrae “tressure” (VT42/12).

That would suggest that perhaps the soft-mutations of lh, rh are thl, thr, as the most probable result of their phonetic history. However, there is a note from 1957 where Tolkien sheds some light on the nasal mutation of voiceless stops:

na (< nā) “to, towards” of space/time. with vocalic mutation. before vowel n’ ... S na, before vowels nan (nasal mutation), means “with” in sense of possessing, provided with, especially of characteristic feature ... na “to” and na “with” are therefore distinct before vowels and b, d, g, p, t, c, m, s but same before h, f, þ, r (rh), l (lh). Late forms as nan-h as for vowels, archaic nath-r, nath-l for nan-rh, nan-lh. nan|sr > nassr > nathr.

In this quote Tolkien is comparing the prepositions na “to” and na(n) “with”, the first causing soft mutation and the second causing nasal mutation. He goes on to say that soft and nasal mutation produces the same results for a bunch of consonants, including voiceless lh, rh. He then says “archaic nath-r, nath-l for nan-rh, nan-lh”. This seems to imply that the nasal mutations thl, thr for voiceless lh, rh are archaic, and in modern Sindarin these forms are unmutated lh, rh. This in turn implies voiceless liquids likewise do not undergo soft mutation in modern Sindarin, and the mutations thl, thr are archaic.

This is consistent with our only example: the adjective rhúnen “east” shows no mutation in Talath Rhúnen “East Vale” (S/124), and normally adjectives undergo soft mutation in that position. Absent more information, I think it’s safest to assume voiceless liquids do not change under soft mutation except to thl/thr in archaic/poetic speech.

Ancient Nasalized Stops ([m]b, [n]d, [n]g): In Primitive Elvish, some ancient words began with nasalized stops, such as ✶mbar(ă) “home”, ✶ndorē “land”, ✶ñgolodō “Gnome”. In Sindarin’s phonetic history, these initial nasals vanished before stops, so that the Sindarin form of such words began with simple stops: bâr, dôr, golodh. However, in the conditions leading to soft mutation, these clusters would instead undergo their historic intervocalic developments, in which [mb], [nd] became [mm], [nn] and (normally) [ŋg] survived. It seems these soft mutations were not simply identical to the intervocalic forms, however. Tolkien represented these soft mutations in various ways throughout his life.

In the Gnomish Grammar of the 1910s, the soft mutations fully restored the ancient nasalized stops: i·mbar, i·ndor, i·Ngolda from G. Golda “Gnome” (GG/7); these voiced nasal-stop clusters appeared medially in Gnomish as well. However, in the Early Noldorin Grammar of the 1920s, the intervocalic forms further reduced to simple nasals, as in i·mâr, i·nôr, i·ngoloth from ᴱN. Goloth “Gnome” (PE13/120). In phonetic tables of the 1930s, Tolkien indicated the soft mutation of ancient [m]b was long mm, so that (hypothetical) mutated form would *i·mmâr, but the mutation of ancient [n]d was nd- (PE19/20). In one of the few clear examples of Sindarin soft mutations for an ancient nasalized stop, we see mb as in: i mbas “the bread” in the Sindarin translation of the Lord’s Prayer from the 1950s (VT44/27). But in notes from the late 1960s Tolkien said:

In Sindarin in absolute initial position the nasal was lost: mb´ to b´, but if a proclitic word, such as the article i “the”, preceded, mb remained and developed, as medially, > mm > m (PE17/104).

Tolkien’s vacillations over the proper soft mutations of ancient nasalized stops were probably motivated by a need to keep them distinct from the nasal mutations. This was not an issue in Gnomish of the 1910s, which seems not to include nasal mutations, but nasal mutations were a feature of all later iterations of the language. Analysis of these mutations is further complicated by the fact that Tolkien seems to have used ng to represent both the velar nasal-stop cluster [ŋg] and the simple velar nasal [ŋ] initially. For example, in the 1920s when the labial and dental mutations were simple nasals m, n, the velar mutation was still represented as i·ngoloth. There are notes from the late 1950s or early 1960s indicating that Sindarin likewise possessed an isolated initial velar nasal:

[In Quenya] the change of initial [ŋ] to n ... With respect to pronunciation a curious situation arose. Quenya as it entered Beleriand was, as spoken, a language that did not possess the sounds þ, z, or “free” ñ (sc. not followed by k, g); but in Sindarin þ and ŋ were frequent and their distinction from s, n was essential. After their acquisition of Sindarin the Noldor were thus familiar with þ and ŋ and the more learned often reintroduced the ancient pronunciation of the Feanorian letters that represented these sounds (now pronounced s, n) (PE17/129).

Since any ancient isolated [ŋ] vanished in Sindarin (PE18/104), the most likely candidate for this “free” initial [ŋ] is a mutated form, of which soft mutation is the best candidate since it is the most common consonant mutation. Since [ŋg] survived as a cluster medially, the change [ŋg] > [ŋ] must have been a special development at the beginning of a morpheme boundary. Likewise, the normal post-vocalic development of [mb] would be [mb] > [mm] > [m]. Although there are no attested examples, many Neo-Sindarin writers assume that the soft mutation of ancient [nd] is likewise a simple dental nasal [n], though this would again need to be a special initial development, since medially it shifted to [nn] but did not shorten.

I believe this system of (Neo) Sindarin mutations for ancient nasalized stops was first proposed by David Salo (GS/76), but it is now widely accepted, and is also supported by the late 1960s note mentioned above (PE17/104). In the case of ancient [n]g, many Neo-Sindarin writers (myself included) represent the isolated velar nasal as ñ- rather than using Tolkien’s (ambiguous) representation ng-. Thus these soft mutations would be:

• bâr [mb-] “home” → *mâr.
• dôr [nd-] “land” → *nôr.
• Golodh [ng-] “Gnome” → *ñolodh (vs. more Tolkienic *ngolodh).

Summary: To summarize the above discussion, here are my recommendations for Neo-Sindarin:

• Initial voiceless stops (p, t, c) become voiced stops (b, d, g).
• Voiced stops (b, d, g) become voiced spirants (v, dh, ’).
• Nasal m becomes v.
• Spirants (s, h) become “softer spirants” (h, ch).
• Ancient nasalized stops ([m]b, [n]d, [n]g) become nasals (m, n, ñ).
• Voiceless liquids lh, rh resist soft mutation in modern Sindarin, but became thl, thr in archaic speech.

The last two sets of changes are poorly attested and should be considered somewhat speculative.

Conceptual Development: There are several tables showing soft mutations at various developmental stages for Sindarin’s conceptual precursors, and these can be used to trace Tolkien’s changing thoughts on soft mutation. The earliest appears in the Gnomish Grammar of the 1910s, which followed the above system of soft mutations except (1) s and m do not mutate, (2) ancient nasalized stops are fully restored (mb, nd, ng) and (3) mutations of initial voiceless lh, rh do not appear because they are not part of Gnomish’s phonetic inventory (GG/7-8). A similar table appears in the Early Noldorin Grammar of the 1920s, identical to Gnomish except that ancient nasalized stops mutated to simple nasals m, n, ñ (represented as ng) as described above and for some reason h does not mutate (apparently a transient idea).

The soft mutations for Noldorin of the 1930s can be found in the Comparative Tables of phonetic development, where Tolkien said “Alternatives thus c/g indicate the special initial variations or mutations in Noldorin” (PE19/18-21). In these tables Tolkien introduced the soft mutations of m, s and voiceless lh, rh, though as noted above the phonetic foundation for the mutation of voiceless liquids was different in Noldorin (more Welsh-like) than later Sindarin.

Noldorin also possessed some soft mutations not seen in Sindarin. In particular, these tables indicate ancient initial sw- became f- in Noldorin, with a mutated form chw- (PE19/21), as opposed to Sindarin where it become voiceless hw. Lack of examples makes it hard to tell what the soft mutation of hw would be (if any) in Sindarin, but David Salo suggested it would be chw- (GS/74), similar to the mutation of isolated initial h-. This is partially supported by the variant forms whinn and †chwind for “birch”.