In notes on the Common Eldarin Article (CEA) from 1969, Tolkien described vocalizations of ꜧ [xʲ] < sy and ƕ [xʷ] < sw. For the origins of these sounds, see the entry on how medial [sj], [sw] became [xʲ], [xʷ].

Tolkien described the later developments of sy > ꜧ as follows:

Initially & medially χ (< hy < sy) became a back spirant with no trace of the lost y; but at the period of the {vocalic infections [sic], when the final vowel i (of various origins) had not yet disappeared, it was still a front spirant which affected preceding vowels in the same ways: sc. a > ei; e > i; o > œi > ei; u > y in ways similar to the... >>} vocalic affections (when final vowels derived from CE long vowels and diphthongs had not yet disappeared) it was still a front spirant and a glide ı̯ developed before it that formed diphthongs similar to those developed before CE kt > S ı̯þ; these were also subject to affections if -i had followed the ꜧ. Note that the {older >>} later affection by which i, u > e, o when a vanishing -a (from CE ā) stood in the old final syllable affected i/u > e/o. Thus CE asy- > aech; osy- and usy(a) > oech; esy- and isy(a) > eich; isy- > {īch >>} ich; usy- > uich. The affections were aech > eich; oech > uich; eich > {ī̆ch >>} ich; uich > ȳch (PE23/139).

In this document Tolkien seems to (briefly?) switch the order of a-affection and i-affection (revising older >> later), but I think this was probably a transient idea since there is plentiful evidence elsewhere that a-affection was first (e.g. Feir, gail, gwein, main). None of Tolkien’s examples in CEA depend on this new ordering, so I’m ignoring it in my analysis. Furthermore, the vocalization of ꜧ to ı̯χ must have occurred after i-affection, otherwise the resulting i-diphthongs would have blocked the i-affection changes. As such, the order of the sounds changes seems to be:

1. Original sy > ꜧ.
2. a-affection or i-affection.
3. Vocalization of ꜧ to ı̯χ.
4. Further diphthong changes like oi > oe.

Tolkien’s statement that the change of ꜧ to ı̯χ occurred “when final vowels derived from CE long vowels and diphthongs had not yet disappeared” implies that this change would occur medially, was well as finally, though this is not absolutely certain since some of the changes described (such as i-affection of e to i) only occur in final syllables. Incorporating the sounds changes described elsewhere in this lexicon, the possible developments are:

• No affections: (1) isyǝ, esyǝ, asyǝ, osyǝ, usyǝ > (2) iꜧ(ǝ), eꜧ(ǝ), aꜧ(ǝ), oꜧ(ǝ), uꜧ(ǝ) > (3) iχ, eı̯χ, aı̯χ, oı̯χ, uı̯χ > (4) ich, aich, aech, oech, uich
• a-affections: (1) isya, esya, asya, osya, usya > (2) eꜧ(a), eꜧ(a), aꜧ(a), oꜧ(a), oꜧ(a) > (3) eı̯χ, eı̯χ, aı̯χ, oı̯χ, oı̯χ > (4) aich, aich, aech, oech, oech
• i-affections: (1) isyi, esyi, asyi, osyi, usyi > (2) iꜧ(i), iꜧ(i), eꜧ(i), yꜧ(i), yꜧ(i) > (3) iχ, iχ, eı̯χ, yı̯χ, yı̯χ > (4) ich, ich, aich, uich, uich

The expected plural patterns are therefore ich, aich → ich; aech → aich; oech, uich → uich. These results differ from Tolkien’s description on two points.

First, he represents the results of esy-, isya, asyi as eich rather than aich. Elsewhere normally ei > ai in final syllables, but there are some exceptions in monosyllables (e.g. Feir, gwein versus gail, main) which could represent some conceptual vacillations on Tolkien’s part.

A second (and more difficult to explain) variation is ȳch as the i-affected (plural) form of uich < usy- versus uich as the i-affected (plural) form of oech < osy-. In both cases the i-affected forms should have developed to yꜧ(i) > yı̯χ > uich, and it is hard to explain the difference in plurals with uich → ȳch. My best guess is the ȳch-plurals of uich are some kind of analogical leveling, but with what isn’t clear. See the discussion of the diphthongs yı̯, yu̯ became ui for further details.

As for sw > ƕ, Tolkien said that it vocalized to u̯χ:

In the case of final chw, since the χ was not only a back spirant but also highly rounded, chw becoming final > u̯χ (wch), producing with the preceding vowels the diphthongs awch, owch (> awch), ewch, iwch (> ȳch), {uwch >>} uch (> {ūch >>} ŭch), with affections ewch, ewch, iwch (> ȳch), {ȳch >>} uich (PE23/139 and note #17).

Following the normal phonetic rules of this lexicon, the possible developments in final syllables are:

• No affections: (1) iswǝ, eswǝ, aswǝ, oswǝ, uswǝ > (2) iƕ(ǝ), eƕ(ǝ), aƕ(ǝ), oƕ(ǝ), uƕ(ǝ) > (3) iu̯χ, eu̯χ, au̯χ, ou̯χ, uu̯χ > (4) ȳch, euch, auch, auch, uch
• a-affections: (1) iswa, eswa, aswa, oswa, uswa > (2) eƕ(a), eƕ(a), aƕ(a), oƕ(a), oƕ(a) > (3) eu̯χ, eu̯χ, au̯χ, ou̯χ, ou̯χ > (4) euch, euch, auch, auch, auch
• i-affections: (1) iswi, eswi, aswi, oswi, uswi > (2) iƕ(i), iƕ(i), eƕ(i), yƕ(i), yƕ(i) > (3) iu̯χ, iu̯χ, eu̯χ, yu̯χ, yu̯χ > (4) ȳch, ȳch, euch, uich, uich

Since these changes happen only finally, the process of ƕ > u̯χ aligns with u-intrusion and was therefore probably later than ꜧ to ı̯χ. Here the expected plural patterns align more closely with the ones given by Tolkien: ȳch, euch → ȳch; auch → euch; uch → uich. The simplest explanation of the last plural pattern is that both yı̯ and yu̯ > ui; it seems Tolkien also considered but rejected yu̯ > ȳ.

Conceptual Development: In Definitive Linguistic Notes (DLN) from 1959, Tolkien described a slightly different development:

N.B. final -w (left after loss of vowels) in Sindarin was dropped after labials (-mw > mm anyway): after other consonants [it] became ŭ or was intruded like y but without alt[ering] of the preceding vowel. So matwā [>] madw̯ > maud or madu. teswā “[?chip]” > teχwā > teχw̯ > tewch. lisyā “sweet” > liχı̯ā > leχı̯ > leich > laich, pl. lîch (PE17/148).

Here Tolkien seems to indicate sy, sw became χ followed by a full semi-vowel: [xj], [xw] rather than [xʲ], [xʷ]. These still produced diphthongs, but via the more ordinary process of i-intrusion and the more obscure process of u-intrusion, and were thus limited to final positions in both cases. Prior to 2025, this lexicon assumed this was the normal process both finally and medially. This seems to have been Tolkien’s thinking in 1959 where the change was explicitly compared to i-intrusion (PE17/148). This further supported by draft descriptions of the diphthongal developments of ꜧ in CEA, which was likewise (almost) identical to i-intrusion:

... vocalic infections [sic], when the final vowel i (of various origins) had not yet disappeared, it [ꜧ] was still a front spirant which affected preceding vowels in the same ways: sc. a > ei; e > i; o > œi > ei; u > y in ways similar to the... (PE23/139 note #14).

It seems the new paradigm for the vocalization of ꜧ to ı̯χ instead resembling kt > ı̯þ was introduced in CEA itself, apparently in the middle of writing this paragraph.

Neo-Sindarin: For purposes of Neo-Sindarin, I would stick with the system described in CEA, largely because that gives us a good description of the plural patterns for words originating from vocalized spirants.