In Sindarin, Noldorin and (partially) Gnomish, the voiceless spirants [x] and [ɸ] (or [f]) developed into vowels before [θ] and (sometimes) [s], forming diphthongs in most cases. Some of the resulting diphthongs had distinct developments from those of primitive diphthongs, indicating this was a comparatively late change, probably after the Old Sindarin/Old Noldorin period, as suggested by David Salo (GS/§4.82, §4.90, §4.91). In the case of the vocalization of voiceless velar spirant [x], it seems these phonetic developments were parallel to vocalizations of the voiced velar spirant [ɣ] at all conceptual stages, which can be used for additional clues on how these sounds evolved.

Most of the Sindarin examples show only the vocalization of [x]. In particular, we see consistent examples of [ax] > [ai], [ex] > [ei] and [ux] > [ui], with probable intermediate forms added for clarity:

• ✶maʒtā > maχtā > S. maetha- “to handle, wield” (VT47/6).
• √PAKAT [> paχthā] > S. paeth “speech” (PE17/126).
• ✶ek-tā [> eχthā] > S. eitha- “to prick” (WJ/365).
• √NEK [> neχthano] > S. neithan “one deprived” (PE17/167).
• √RUKU [> gruχtha] > S. gruitha- “to terrify” (WJ/415).
• ✶nuktā- [> nuχtha] > S. nuitha- “to stunt” (WJ/413).

The sound change [ax] > [ai] is partially obscured because later still [ai] became [ae]. The vocalization of [ex] > [ei] is also sometimes obscured by the fact that later [ei] became [ai] in final syllables:

• ✶nek-tē > neith > S. naith “angle” (PE17/55).
• ✶tek-tā > teith > S. taith “sign, symbol, mark” (PE17/43).

This development in particular is distinct from that of the primitive diphthong [ei], which instead became [ī] in Old Sindarin/Old Noldorin, a strong sign that the vocalization of spirants was a comparatively late sound change.

There is one Sindarin example that might show [ix] > [ī] or [i]:

• √RIK¹ > S. r(a)itha- “try”, with apparent past tense forms rithant (3rd-sg) and rithanen (1st-sg) (PE17/167).

The stem form raitha- is hard to explain, and may be the result of some kind of abnormal Sindarin a-affection (see that entry for details). However, development of the past tense form seems more regular, and it may have derived from: *riktant- > *riχthanth- > *rīthant > rithant, with the long [ī] shortening as it often did in polysyllables. The ambiguous development of the stem form raitha- makes this example hard to interpret, however.

There are no Sindarin examples involving the vocalization of [ox], but there are several examples of the vocalization of [oɣ] that show that [oɣ] > [oi] > [oe]:

• ✶logna [> loʒna > loin] > S. loen “soaking wet, swamped” (VT42/10).
• ✶ugrā > ogra [> oʒra > oir] > S. oer “nasty” (PE22/160).

See the entry on how [ɣ] vocalized before [l], [r], [m], [n] for a more detailed discussion of these examples. After other vowels the Sindarin vocalizations of [ɣ] and [x] generally match, so it is reasonable to assume [ox] > [oi] as well. Assuming this is true, it appears that in Sindarin [x] vocalized to [i] in all cases, so that the phonetic developments were:

• aχ > ai > ae.
• eχ > ei (> ai in final syllables).
• iχ > ī.
• oχ > oi > oe (unattested).
• uχ > ui.

We only have one clear example of the vocalizations of [ɸ] in Sindarin, but its phonetic development seems distinct from the vocalizations of [x]:

• √LEP [> leptā > leɸtha] > S. leutha- “to pick (up/out)” vs. Q. lepta- (VT47/10).

Here it seems [eɸ] > [eu], as opposed to [ex] > [ei]. If this is the general pattern, it may be that the vocalization of [ɸ] and [x] depended on the quality of the spirant in Sindarin: [ɸ] > [u] and [x] > [i]. This was not always the case in early versions of the language, however (see below).

There are no clear examples of vocalizations of [x] before [s] in Sindarin, most likely because the primary source of [xs] in early conceptions was primitive [ks], but it seems that [ks] > [xx] > [x] in Sindarin, for example: S. ach “neck” < aks[ē] < √AKAS (PE17/92). There are examples of such vocalizations before [s] in earlier conceptions of the language, however (see below).

Conceptual Development: Tolkien seems to have introduced the vocalization of spirants early on, but the nature of this sound change evolved quite a bit over different iterations of the language.

Gnomish Developments (1910s): Roman Rausch discussed the analogous Gnomish sound changes in his Historical Phonology of Goldogrin (HPG/§2.6, 2.7). Analysis of Gnomish phonetic developments is complicated by the fact that the voiceless spirants [x] and [ɸ] were part of the phonetic inventory of the earliest version of Primitive Elvish from the 1910s. Furthermore, the combinations pth and cth were quite common in Gnomish, and did not vocalize universally as they did in Sindarin/Noldorin:

• G. captha- “to startle” (GL/24) vs. ᴱQ. kapta- < ᴱ√KAPA (QL/45).
• G. celeptha “silvern” (GL/25) vs. ᴱQ. telepta “of silver” < ᴱ√TELEPE (QL/91).
• G. leptha “finger” (GL/53) vs. ᴱQ. let (lept-) < ᴱ√LEPE (QL/53).
• G. tectha- “to write” (GL/69) vs. ᴱQ. tekta- < ᴱ√TEKE (QL/90).
• G. mectha- “to aim at” (GL/57) vs. ᴱQ. mekte- < ᴱ√MEKE (QL/60).
• G. ectha “sword” (GL/31) vs. ᴱQ. ekte < ᴱ√EKE (GL/18).

There are, however, examples where it appears that voiceless spirants vocalized. These vocalizations seem to have occurred primarily (but not exclusively) in final syllables:

• ᴱ✶kahta > G. caith “cause, reason, motive” (GL/24) vs. ᴱQ. ká “deed, act, fact” < ᴱ√KAHA (QL/43).
• ᴱ✶ektī [> eith] > G. aith(i) “sword” vs. ᴱQ. ekte (GL/18); compared to G. ectha above (GL/31).
• ᴱ✶ektḷ [> eithl] > G. aithl “spring” (GL/18) vs. ᴱQ. ektele “fountain” (QL/35).
• ᴱ✶leχ-sa [> leis] > G. lais “green sward, glade” vs. ᴱQ. lé “sward, glade” (GL/53).
• ᴱ✶foχo [> foχtha- > foitha-] > G. fuitha- “to hide” (GL/36).
• ᴱ✶foχo [> foχsa > fois] > G. fuis “hoard” (GL/36).
• [*saptha >] G. sauth “hole, tunnel” (GL/67) vs. ᴱQ. sat (sapt-) < ᴱ√SAPA (QL/82).
• [*alepthion >] G. alewthion [aleuthion] “having fingers” (PE13/109) vs. leptha “finger” above.

Roman Rausch suggested that the likely rule was that pth and cth were preserved medially, but [x] and [ɸ] were vocalized when part of a final consonant cluster. There are counterexamples to this, but they can mostly be explained by later compounds or analogical leveling, for example: ᴱ✶eχt·taþ· > G. aithos “thorn-bush” (GL/18) might have been revised by analogy with G. aith “thorn”. The last example G. alewthion [aleuthion] with medial vocalization appeared as a somewhat later word in the Gnomish Lexicon Slips rather than in the Gnomish Lexicon itself, and may represent a shift towards the Early Noldorin pattern of universal vocalizations (see below).

Based on the examples above, it seems that [x] vocalized to [i] and [ɸ] to [u] in the Gnomish of the 1910s. This is partly obscured by the fact that in Gnomish later [ei] became [ai] much like in Sindarin, except that in Gnomish this was a universal change rather than being limited to final syllables (the diphthong ei does not appear in the Gnomish Lexicon proper, though it does begin to appear in the aforementioned Gnomish Lexicon Slips). Another obscuring change was the fact that [oi] became [ui] in Gnomish (not [oe] as it did in Sindarin).

Early Noldorin Developments (1920s): In the Early Noldorin of the 1920s, the medial combinations pth and cth disappeared, so that the vocalization of [x] and [ɸ] before [θ] became universal. As described by Roman Rausch in his Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.1.3), it seems that vocalization of [x] remained [i] as in Gnomish. Here are some relatively straightforward examples (with a few intermediate changes added for clarity):

• ᴱ✶kṛkta [> kraχta] > ᴱN. craith “circle” (PE13/140).
• ᴱ✶surı̯akta > ᴱN. huiriaith “gale” (PE13/163).
• ᴱ✶ekta > ᴱN. aith “thorn” (PE13/136).
• ᴱ✶g’lamektá or ᴱ✶k’lamektā > ᴱN. glavaith “blaze” (PE13/144, 162).
• ᴱ✶loktu > ᴱN. luith “magic, spell” (PE13/149).
• ᴱ✶siktā [> seχta] > heith > ᴱN. haith “moist” (PE13/147, 163).

These Early Noldorin vocalizations are still obscured by various other sound changes. In the last example, the sound changes are obscured by a-affection, a new feature of Early Noldorin. Other vocalizations are obscured because [ei] > [ai] and [oi] > [ui] as in Gnomish, but these diphthongal changes seem to be less universal in the Early Noldorin period. For example, we see ᴱN. eitheb “thorny” vs. ᴱN. aith “thorn” (PE13/136, 158) and ᴱN. eithlos “fountain” vs. ᴱN. aithl “spring, fount” (PE13/158), which seems to indicate that [ei] > [ai] mostly in final syllables in the Early Noldorin of the 1920s (as was also the case in later Sindarin). Similarly, there are a fair number of examples where [oi] remains unchanged:

• ᴱN. oith “feud” vs. ᴱQ. ohta “war” (PE13/151, 164), likely from primitive *✶oktā; compare later ᴹ✶oktā > N. auth (Ety/KOT).
• ᴱN. hoith “coitus” vs. ᴱQ. pukta or puhta (PE13/147, 163), likely from primitive *✶puktā; here a-affection plays a role.
• ᴱN. oif “terror, phantom” vs. ᴱQ. oswe, both from ᴱ✶okswē (PE13/151, 164).

The last word oif seems to show a vocalization before [f] not seen in any other examples; it’s not clear under what conditions the combinations [xɸ] or [xf] are possible. Since both [ui] and [oi] seem to derive from [ox] in final syllables for Early Noldorin, it’s not clear if there was a consistent phonetic rule here, or if Tolkien was gradually abandoning [oi] > [ui] in Early Noldorin as a step in the direction of the later Sindarin sound change whereby [oi] became [oe].

As was the case in other conceptual stages, there are far fewer examples of vocalizations of [ɸ] in Early Noldorin. As suggested by Roman Rausch (HPITN/§4.1.3), the few examples seem to indicate that [ɸ] > [i]:

• ᴱ✶kapse > ᴱN. cais “leap” (PE13/140).
• ᴱN. maith “ravishment; seizure” vs. ᴱQ. mapta (PE13/163).

If this was the general rule, perhaps both [x] and [ɸ] vocalized to [i] at this conceptual stage. However, there is also the following pair of (unglossed) examples from a table of Early Noldorin diphthongal developments written in the early 1920s (PE15/64):

• ᴱ✶lopse > ᴱN. laus
• ᴱ✶lepse > ᴱN. leus

If we account for the sound change [ou] > [au], these two examples indicate the vocalization of [ɸ] > [u], as was the case in Gnomish. These two examples may represent lingering Gnomish ideas, serving as bridge between the Gnomish pattern of the 1910s (where the vocalizations of [ɸ] were distinct from those of [x]) and the Noldorin pattern of the 1930s (where the vocalizations of [ɸ] and [x] were the same).

Early Noldorin Plural Developments: Early Noldorin words demonstrating [x] vocalization also have plural forms showing i-mutation:

• ᴱN. archaic †îth or †ith plural of aith “thorn”, with non-archaic plurals eithin or eithir (PE13/136, 158).
• ᴱN. crith plural of craith “circle” (PE13/140).
• ᴱN. eithlir plural of aithl “spring, fount” (PE13/158).
• ᴱN. †glevith plural of glavaith “blaze”, with non-archaic plurals gleveithin or glaveithin (PE13/158).
• ᴱN. uith plural of ᴱN. oith “feud” (PE13/151, 164).
• ᴱN. huith plural of ᴱN. hoith “coitus” (PE13/147, 163).
• ᴱN. uif plural of ᴱN. oif “terror, phantom”, with variant plurals oifir or oifar (PE13/151, 164).

For these plurals to have developed, the vocalization of [x] must have occurred after the various stages of i-affection, such as i-raising and i-fronting. Whether this was true in later conceptual stages is unclear (see below). These plural forms also indicate that iχ > ī and uχ > ui in Early Noldorin, which fits the general pattern of [x] vocalizing to [i].

Noldorin Developments (1930s): In the 1930s, Tolkien partially described the vocalization of [x] in notes on the Noldorin usage of the Feanorian Alphabet (PE22/39-40):

The long diphthongs — ON diphthongs, diphthongization of ON ō, or new diphthongs from short vowel + vowel (in contractions or in contact with vocalized ʒ, χ), or from long vowels + epenthetic ı̯ ...

• [o͡u] ON au, ō or ŏ + ʒ, χ. archaic h. later > [au] q.v. ...
• [a͡e] older ai, as above; or ă + ʒ, χ. ]l later G[ondolic] ]Þ — usually ]Þ (beside ]l)...
• [e͡i] e + ʒ, χ or affected a + ʒ, χ; ... l`B l~B later G[ondolic] lÖ — usually lÖ (beside l~B)...
• [ui] ON ui; affect[ed] o, u + ʒ, χ ... .`B .~B later G[ondolic] .Ö — usually .Ö (beside .~B)...

Based on these notes it seems that [x] vocalized as follows in Noldorin:

• aχ > ai > ae.
• ex > ei.
• oχ > ou > au.
• uχ > ui (but see below for examples of [ux] > [ū]).

Although not listed, presumably iχ > ī. These sound changes are the same as in the Early Noldorin of the 1920s and the Sindarin of the 1950s, except for oχ > ou. The notes on the Feanorian Alphabet did not indicate the conditions under which [x] vocalized, but based on examples in The Etymologies from the 1930s, it seems this sound change occurred before [s] and [θ]. This is consistent with the rules whereby [p], [t], [k] spirantalized before [s] in Noldorin and before other voiceless stops and aspirates: the combinations [xs] and [xθ] would therefore have been fairly common. There are Noldorin examples for many of these phonetic developments in The Etymologies, with probable intermediate developments added for clarity:

• ᴹ✶taksē [> taχsa > taisa] > N. taes “nail” (Ety/TAK).
• ᴹ✶yakta- [> yaχthe > yaithe] > N. iaeth “neck” (Ety/YAK).
• ᴹ✶b’rekta [> breχtha > breitha] > N. breitho “break out suddenly” (Ety/BERÉK).
• ᴹ✶k’rikta- [> kriχtha > krītha] > N. critho “reap” (Ety/KIRIK).
• ᴹ√LOKH [> loχsa > lousa] > N. lhaws [laus] “hair” (Ety/LOKH).
• ᴹ✶oktā [> oχtha > outha] > N. auth “war” (Ety/KOT).
• ᴹ√YUK [> yukte > yuχthe] > N. iuith “use” (Ety/YUK).

In the case of [ax] > [ai] and [ox] > [ou], the vocalizations are obscured by the later sound changes whereby [ai] became [ae] and [ou] became [au]. There are also examples in The Etymologies where [ux] > [ū] rather than [ux] > [ui]:

• ᴹ√SUK [> suktu > suχtho] > N. sûth “draught” (Ety/SUK).
• ᴹ√LUK [> lukte > luχthe] > N. lhûth “spell, charm” (Ety/LUK).
• ᴹ√LUK > N. lhûtha- “to enchant” vs. ᴹQ. luhta- (Ety/LUK).

This competing phonetic development [ux] > [ū] is actually a bit more frequent than [ux] > [ui], which only occurs for the root ᴹ√YUK in The Etymologies. These variations may represent some vacillation on Tolkien’s part. Alternately, perhaps the sound change of [ux] > [ui] was conditional and only occurred in combinations like [jux-] > [jui-].

There are a smaller number of Noldorin examples of the vocalization of [ɸ]. These examples are not numerous enough to precisely establish a pattern, but the few clear examples from the 1930s seem to match the vocalization of [x]:

• ᴹ√AP [> apsa > aɸsa > aisa] > N. aes “cooked food, meat” (Ety/AP).
• ᴹ√LAP [> lapsa > laɸsa > laisa] > N. lhaes “babe” (Ety/LAP).
• ᴹ✶tupsē [> tupsa > tuɸsa > toɸsa > tousa] > N. taus “thatch” (Ety/TUP).

The last example taus is exceptionally obscure unless you recall that final [e] became [a] after [s], and thus a-affection likely plays a role in the phonological development of this word. It seems the base vowel [u] was lowered to [o] before the vocalization [ɸ], and the resulting diphthong [ou] developed to [au] as usual.

Noldorin and Sindarin Plural Developments: It is is difficult to determine the exact timing of this sound change in the Noldorin of the 1930s and Sindarin of the 1950s and later. The example of N. taus seems to indicate it occurred after a-affection. A couple of notes from the Feanorian Alphabet document hints it may have occurred after i-affection as well:

• [e͡i] e + ʒ, χ or affected a + ʒ, χ [emphasic added]; ... l`B l~B later G[ondolic] lÖ — usually lÖ (beside l~B).
• [ui] ON ui; affect[ed] o, u + ʒ, χ [emphasic added] ... .`B .~B later G[ondolic] .Ö — usually .Ö (beside .~B).

Tolkien does not explain what “affected a” and “affected o” are in these notes, but since a-affection cannot apply to the vowel a, i-affection seems the only reasonable option. If this was the case, these words might have somewhat unusual plurals, similar to Early Noldorin plurals for this class of nouns. For example:

• *taksī > *teχsi > *teis, hypothetical plural of N. taes “nail”.
• *oktā > *uχthi > *uith, hypothetical plural of N. auth “war”.

Unfortunately, none of the obvious examples of [x], [ɸ] and [ɣ] vocalizations have attested plurals in either the Noldorin of the 1930s or Sindarin of the 1950s and later. It is therefore unclear whether the plurals of [x], [ɸ] and [ɣ] vocalizations would show i-affection or not. Assuming they did exist, these plural mutations would have been distinct from words originating from primitive diphthongs, such as:

• √MAG > *maʒri > *meʒri > *meir > *mair, hypothetical plural of S. maer “good”.
• √DAY > *dairi > *daer, hypothetical plural of S. daer “great, large”.

Tolkien said that ae did not change in Sindarin plurals (PE17/25), so even if these plural mutations existed at one point, they most likely were reformed to fit more normal plural patterns. For purposes of Neo-Sindarin, it’s probably simplest to assume that [x] and [ɸ] vocalizations took place before i-affection and just avoid the whole question.

Summary of Conceptual Developments: It seems that in the Gnomish of the 1910s, [x] and [ɸ] had distinct vocalizations, with [x] > [i] and [ɸ] > [u], but the vocalizations occurred almost exclusively in final syllables. This began to shift in the Early Noldorin of the 1920s, where the vocalizations of these two spirants became universal and both of them developed into [i]. In the Noldorin of the 1930s, [x] and [ɸ] shared the same vocalizations, but the exact sound changes were conditioned by the preceding vowel, so that after the vowel [o] (and sometimes also after [u]) both [x] and [ɸ] became [u] instead. By the Sindarin of the 1950s, it seems once again [x] > [i] consistently; there is only one published Sindarin example of the vocalization [ɸ], but here [ɸ] > [u], so perhaps Tolkien had completely reverted to the Gnomish pattern.

It is notable, though, that the only u-diphthong Tolkien said was part of Sindarin is au. So, if Gnomish-style u-diphthongs reappeared in Sindarin as the result of the vocalization of [ɸ], they likely developed into [ȳ] (in the case of [iu]) or [au] (in the case of [ou]). The later diphthong [eu], however, seems to have survived, based on the example leutha-, though it must have been rare.

GS/§4.82 GS/§4.90 GS/§4.91 WGHC/175:§108iv @@@ possible counter examples on order with a-affection: lhûtha-

HPG/§2.6 @@@ also alewthion

HPG/§2.6 @@@