Epenthesis of i (insertion of i into the interior of a word) occurs in both Sindarin and Noldorin where i or y [j] became final and the preceding syllable ended with a single consonant. This lexicon uses the terms i-epenthesis or (more casually) i-intrusion for this phenomenon. Like other forms of i-affection, this sound change plays a significant role in the development of Sindarin and Noldorin plurals, particularly in final syllables. Tolkien discussed i-epenthesis several times in the context of diphthongal developments in his notes on the Noldorin usage of the Feanorian Alphabet from the 1930s:

The old long diphthongs and the [newer] long diphthongs produced (a) by vowels + vocalized consonants, (b) by long vowels + epenthentic ı̯; [emphasis added] (c) by diphthongization of ON ō > ou > au: were usually denoted by two vowels signs. Thus archaic ]. = [au], h. = [ou], h`B or h~B = [oi]; l`B or l~B = [ei]; ]l = [ae̯], hl = [oe]; .`B, .~B = [ui]. The modifier -Ö peg dadol “double dot”, in ON called ambayō “raised y”, was used thus lÖ = ĕı̯ produced by epenthesis of ı̯ to ĕ [emphasis added]: as in băr “home”, pl. bĕı̯r (later by lengthening before sonorous consonants bār, be͡ır) written w]7 wlÖ7 (PE22/36).

Tolkien also mentions i-epenthesis in a chart of diphthongs from the same document (PE22/39-40). The relevant entries are listed here:

• [ĕı̯] e or affected a + epenthetic ı̯. lÖ later became either [e] or becoming long was identified with [e͡i].
• [œ̆ı̯] affected ŏ, ŭ in unstressed syllables + epenthetic ı̯. hÖ| œi later > ei, and so to [e] or [e͡i].
• [ui] ... ON ū + epenthetic ı̯; also ON au, ō + epenthetic ı̯ or + i in contraction ... .`B .~B later G .Ö - usually .Ö (beside .~B).

Tolkien mentioned i-intrusion (or y-intrusion) for Sindarin in notes from the late 1950s or early 1960s:

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 [emphasis added] 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 [emphasis added] (PE17/148).

In this note, Tolkien also seems to be considering w-intrusion as a sound change parallel to i-intrusion, but there is no evidence for it beyond this note. I think it’s likely that w-intrusion was only a transient idea.

Over the years, various analyses of this sound change have been published, the earliest appearing in Jim Allen’s Introduction to Elvish (1978, pp. 112, 115, in an article written by Chris Gilson and Bill Welden), which covered only e > ei > ai given the lack of published evidence in the 1970s. A somewhat more comprehensive analysis appeared in David Salo’s Gateway to Sindarin (GS/§4.99, 2004). I consider the most complete analysis to be the one written by Bertrand Bellet in his Vowel Affection in Sindarin and Noldorin (VASN, 2005), and I mostly follow his analysis in this entry. Bertrand Bellet used the linguistic term epenthesis to describe this change, as did Tolkien, but I often use the less formal term i-intrusion, following the suggestion of Elaran from a private Discord chat on 2018-08-25.

Mechanics of i-intrusion: As discussed by Bertrand Bellet, the likely mechanism for this sound change is as follows. When there was a final i or ı̯ in a Sindarin or Noldorin word (possibly after becoming final as the result of earlier vowel loss), it palatalized the preceding consonant. This palatalized nature transmitted forward into the preceding syllable, breaking the preceding vowel into an i-diphthong. Somewhat later the now-final consonant lost its palatalization. Adding some intermediate stages to the example leχı̯ > leich above (PE17/148), the probable phonetic developments were:

• leχı̯: [lexı̯] > [lexʲ] > [leʲxʲ] > [leixʲ] > [leix]: leich.

This palatalization broke most preceding single vowels (both short and long) into a new i-diphthong. The scenarios listed by Bertrand Bellet (VASN) are:

• -eCi > -eCʲ > -eʲCʲ > -eiCʲ > -eiC.
• -œCi > -œCʲ > -œʲCʲ > -œiCʲ > -œiC.
• -ūCi > -ūCʲ > -ūʲCʲ > -uiCʲ > -uiC.
• -ǭCi > -ǭCʲ > -ōʲCʲ > -oiCʲ > -oiC.

The first three scenarios match the Tolkien’s list from his notes on the Noldorin use of the Feanorian Alphabet from the 1930s given above (PE22/39-40). The fourth scenario is not mentioned separately in Tolkien’s notes, because in the Noldorin period, [ǭi] became [ui], so that the end result was the same as the third scenario. See the next section below for more discussion of this fourth scenario in the context of Noldorin and Sindarin.

As noted by Bertrand Bellet (VASN), there are two vowels that seem to resist i-intrusion: i and y (a more rounded i-sound). The vowel i cannot move any further forward; as for y see the discussion under Plurals of Monosyllables with ô below. The remaining vowels listed above (short e, œ; long ǭ, ū) are actually a complete list of all vowels that could appear before final -i, because:

• primitive short a had become e due to i-fronting.
• primitive short o had become œ due to i-fronting.
• primitive short u had become y due to i-fronting.
• primitive long ā had developed into ǭ.
• primitive long ē had developed into ī.
• primitive long ō had developed into ū.

As was the case for the vowel i, any existing i-diphthong such as ai or ui also resisted i-intrusion.

I-intrusion and u-diphthongs: Whether i-intrusion could affect u-diphthongs is a bit more ambiguous. At this stage of Sindarin’s (and Noldorin’s) phonetic evolution, the only possible surviving primitive u-diphthong was au: primitive [eu] and [iu] developed into [ȳ] in Sindarin ([ū] or [io] in Noldorin) and primitive [ou] developed into [ū] (or [au] in Noldorin). In Noldorin, any au-monosyllables mutate to ui in the plural, such as:

• N. guin plural of N. gwaun “goose” (Ety/WA-N).
• N. thuin plural of N. thaun “pine-tree” (Ety/THŌN).
• N. nuig plural of N. naug “dwarf” (EtyAC/NAUK).

These plurals fit the Noldorin i-epenthesis rules of au/ō [ǭ] described by Tolkien in his notes on the Feanorian Alphabet from the 1930s:

[ui] ... also [from] ON au, ō + epenthetic ı̯ or + i in contraction (PE22/40).

The primitive diphthong [au] merged with [ǭ] in Old Noldorin, so that the likely Noldorin phonetic development is:

• *naukī > *nǭki > *nǭik > *noig > nuig; where [ǭi] > [oi] > [ui], the same development as the primitive diphthong [oi].

In Sindarin, the plural of S. naug “dwarf; stunted” (< ✶naukā) is noeg (WJ/413; UT/100). There is a similar (but deleted) plural thoen of thaun “pine-tree” (PE17/81); apparently this was the proper Sindarin phonetic development, but Tolkien couldn’t use it because he had already published a (Noldorized?) word thôn in Orod-na-Thôn “Pine-mountain” (LotR/469; RC/384). Their developments might have been similar to that of Noldorin except for the development of the resulting diphthong [oi], as suggested by Betrand Bellet (VASN):

• *thānī > *thǭni > *thǭin > *thoin > thoen.
• *naukī > *nǭki > *nǭik > *noig > noeg.

The problem here is that, unlike Noldorin, there is no evidence that primitive [au] > [ǭ] prior to i-intrusion as a universal rule (though it probably did so at a later point since [au] became [o] in many polysyllables). In fact, there is evidence that ǭ > au before final vowel losses and hence before i-intrusion: arāme < arǭmæ < araum(a) < araum̌, arauv < araw (WJ/400). Thus the plural form of both these words was probably thauni and nauki immediately prior to i-intrusion. A more likely development would be:

• -auCi < -auCʲ < -auʲCʲ < -oiCʲ < -oiC

I think it’s likelier the diphthong au monophthongized to o (or perhaps ǫ, IPA [o̞]) as part of the vowel breaking, ultimately resulting in the diphthong oi. Hence:

• *thānī > *thauni > *thǫin > *thoin > thoen.
• *naukī > *nauki > *nǫik > *noig > noeg.

The effect of i-intrusion on any later u-diphthongs that might have arisen in Sindarin from the vocalization of [ɸ] is unclear, but perhaps they fell in with [ȳ] developments (for [eu] or [iu]) or [au] developments (for [ou]).

Effects of i-intrusion on plurals: The results of i-intrusion manifested in plural mutations for certain final syllables in Sindarin and Noldorin. As noted by Betrand Bellet (VASN), the relevant plural patterns are:

• au in final syllables mutates to oe (Sindarin) or ui (Noldorin).
• a in final syllables mutates to ai (Sindarin) or ei (Noldorin).
• ū in final syllables mutates to ui (Sindarin and Noldorin).
• o in final syllables mutates to ei (Noldorin only).

The complexities of the plural mutation of au are discussed in the previous section. The plural mutations of a to ai (S.) or ei (N.) in final syllables are more straightforward and much better attested. Some examples include:

• S. craban, pl. crebain (PE17/25).
• S. aran “king”, pl. erain (PE17/40, 111).
• S. bâr “house, home”, pl. bair (PE17/97, 164).
• N. habad “shoe”, pl. hebeid (Ety/SKYAP).
• N. pân “plank”, pl. pein (Ety/PAN).
• N. bár “home”, pl. beir (PE22/36).

As representative of this phonetic development, the plural for S./N. bar “home” would likely be derived from:

• *barī > *beri > beir > bair.

The last change is Sindarin-only, where [ei] became [ai] in final syllables, a sound change that did not (usually) occur in Noldorin. This Sindarin-only sound change is the likeliest reason for the ai/ei variations in Sindarin/Noldorin plurals. There are a few Noldorin plurals that show ai, such as erain plural of N. aran “king” (Ety/ƷAR) and (m)belain plural of Balan “God” (Ety/BAL, KIRIK), but these can be explained as early manifestations of the Sindarin phonetic development. By the time Tolkien started writing The Lord of the Rings drafts in the 1940s, most of these plurals show ai, such as N. bair “houses” (WR/380) instead of earlier beir from the 1930s (PE22/36).

The plurals where û mutates to ui are also straightforward phonetically. The main challenge is finding clear examples, but there are at least a couple:

• S. duir plural of S. dûr “dark” in Emyn Duir “Dark Mountains” (UT/280).
• N. luin plural of N. lhûn “blue” in Eredluin “Blue Mountains” (Ety/LUG²).

The plurals where o mutates to ei seem to be a Noldorin-only phenomenon. In Sindarin, polysyllabic words with an o in the final syllable consistently mutate to y in plurals. This is the result of i-raising and i-fronting on the final syllable, for example:

• S. Gelydh plural of S. Golodh “Noldo” (S/238; WJ/364, 379).
  • *ŋgolodī > *goludhi (raising) > *gœlydhi (fronting) > gelydh.

As noted in the entry on i-raising, in Noldorin the raising of o > u > y is inhibited under some conditions. As such, some Noldorin plurals underwent a different phonetic development that involved i-intrusion:

• N. Goeloeidh [gœlœið] > Geleidh plural of N. Golodh “Noldo” (Ety/ÑGOLOD).
  • *ŋgolodī (no raising) > *gœlœdhi (fronting) > gœlœidh (intrusion) > geleidh.

Without i-raising, the o in the final syllable developed into [œ], which became the diphthong [œi] after i-intrusion. Later on, the vowel [œ] became [e] and the diphthong [œi] became [ei], producing the same result as i-intrusion combined with the vowel e. This Noldorin phonetic development was mentioned by Tolkien in his notes on the Feanorian Alphabet from the 1930s:

[œ̆ı̯] [from] affected ŏ, ŭ in unstressed syllables + epenthetic ı̯. hÖ| œi later > ei, and so to [e] or [e͡i] (PE22/39).

Although the diphthong [œi] did not arise in the development of Sindarin plurals, in theory it could occur in other circumstances, such as in primitive words with the suffixes -yā or -yē and the base vowel o or u. Presumably the phonetic development would be: -uCya > -oCya > -oCia > -œCi > -œiC > -eiC > -aiC. [@@@ revisit, what if œ > e after ei > ai. Also possibly fuir “north”.] There are no clear examples of this, however, and in fact the few examples we have seem to contradict this phonetic development:

• S. ruin “fiery red” vs. Q. runya (PM/366), presumably derived from *runyā.
• S. eryn “wood, forest” < ✶oronyē (PE17/119).

Whether these are aberrations or represent some special Sindarin-only development is unclear.

Noldorin plurals with a that mutates to e: There are a few Noldorin plurals in The Etymologies that show e in final syllables instead of the expected ei, such as:

• N. eder plural of N. adar “father” (Ety/ATA).
• N. deren plural of N. doron “oak” (Ety/DÓRON).
• N. Neweg plural of N. Nawag “dwarf” (Ety/NAUK).

The last example appears beside variants neweig and (deleted) newaig (EtyAC/NAUK). Many of these examples can be explained by the Noldorin sound change whereby [ei] sometimes became [e] in unstressed final syllables; this sound change is also seen in words that are not plurals, such as ᴹ✶talrunya > tellœin > tellein > N. tellen “sole of foot” (Ety/RUN, TAL). This reduction of ei to e is mentioned by Tolkien in his notes on the Feanorian Alphabet from the 1930s:

[ĕı̯] e or affected a + epenthetic ı̯. lÖ later became either [e] [emphasis added] or becoming long was identified with [e͡i] (PE22/39).

There are few monosyllabic plurals with a mutating to e that require a different explanation; see the next section.

Consonant clusters resist i-intrusion: As noted by Bertrand Bellet (VASN), in Noldorin there is clear evidence that consonant clusters prevent i-intrusion. We see, for example:

• N. lelf plural of N. lalf “elm-tree” (Ety/ÁLAM).
• N. perf plural of N. parf “book” (Ety/PAR).
• N. rhenc plural of N. rhanc “arm” (Ety/RAK).
• N. Enfeng “Longbeards” plural of N. An(d)fang (Ety/ÁNAD, SPÁNAG).

There are similar examples in Sindarin:

• S. Enfeng “Longbeards” plural of S. Anfang (WJ/10) as above.
• S. nern plural of S. narn “tale” (MR/373).
• S. úgerth “*trespasses” plural of S. *úgarth (VT44/28).
• S. -end “lands” plural of S. -and “land” (Let/383; UT/318).

However, there are other Sindarin examples (and at least one Noldorin example) where i-intrusion seems to have occurred despite consonant clusters:

• S. morchaint “dark shapes, shadows” plural of S. *morchant (VT42/9).
• S. eilph plural of S. alph “swan” (UT/265).
• S. eigl plural of S. ogl “evil”, beside variant plurals eigil and egl (PE17/149).
• N. eirch or erch plural of N. orch “Orc” (LR/406), elsewhere appearing as more a regular plural N. yrch (Ety/ÓROK).

There are further Sindarin examples where i-intrusion occurred before an old consonant cluster [-ss] that was later reduced:

• S. lais plural of S. las(s) “leaf” (PE17/62), likely derived from *lassī.
• S. nimrais “pale horns” plural of S. nimras(s) (PE17/89), likely derived from *-rassī.

In the example above, it is possible that the cluster reduction of [-ss] > [-s] occurred before i-intrusion, however, and thus the i-intrusion was in fact regular. There is also the following example of an e-plural before a reduced consonant cluster:

• S. cem plural of S. cam “hand” (VT50/22), likely derived from *kambī.

Absent further evidence, it is hard to determine the reason for these Sindarin plural variations. Phonetically, it makes sense that the palatalization from final i had difficulty penetrating more than one consonant, and the plural variations showing e before consonant clusters would therefore be the “expected” pattern. The variations showing ai (or ei) could represent (a) normalization to the plural patterns of single consonants, (b) indication that some consonant clusters did not inhibit i-intrusion, or (c) conceptual vacillations on Tolkien’s part.

Plurals of monosyllables with ô: Based on the phonetic rules discussed above, the expected plural form of monosyllables with overlong ô would be overlong ŷ, for example:

• N. pŷd plural of N. pôd “animal’s foot” (Ety/POTŌ).
• N. dȳr (earlier dy̆r) plural of N. dôr “land” (PE22/38).

Tolkien discussed this phonetic development in his notes on the Feanorian Alphabet from the 1930s:

[ȳ]** only occurred after archaic period from lengthening of stressed y before voiced final consonants as dȳr < dy̆r “lands” (PE22/38).

Overlong ô could only appear in Sindarin and Noldorin from the lengthening of a short ŏ in monosyllables. Based on the rules above, the expected phonetic development of these forms would be the raising of o to u, the fronting of u to y and finally the lengthening of y as was common in stressed monosyllables:

• *(n)dori > *duri (raising) > *dyri (fronting) > dyr > dȳr (monosyllabic lengthening).

There would be no i-intrusion because (at least in Noldorin of the 1930s) the front vowel y seems to have resisted i-intrusion. However, there are several Sindarin and Noldorin ô-monosyllables with ui in their plurals instead:

• N. duil “hills” plural of N. dôl (SM/225), from The Lord of the Rings drafts of the 1940s.
• S. thuin plural of S. thôn “pine” (PE17/81).
• S. thuil plural of S. thôl “helmet” (PE17/188).

The last example is of particular interest, because it seems to be a reformation of an older and irregular plural S. thely, but this “reformation” only makes sense if ui was the normal plural pattern for ô-monosyllables. The likeliest explanation I can think of is that y did not resist i-intrusion and instead broke into the diphthong yi > ui. Presumably in this scenario the round front y dissimilated to the round back vowel u in order to remain distinct from the i in the newly formed diphthong ui. This would be a fifth scenario for i-intrusion in addition to the four discussed above:

• -yCi > -yCʲ > -yʲCʲ > -yiCʲ > -uiC.

This sound change could have happened in both monosyllables and polysyllables, but with polysyllables it seems the unstressed yi then simplified to y instead of becoming ui, as evidenced by plurals like emyn and gelydh. There are a couple more examples, however, that seem to show ui diphthongs at the end of polysyllables resulting from i-intrusion that later develop into y:

• N. emuin > emyn plural of N. amon “hill” (Ety/AM²).
• ✶herūnī > heruin > heryn “lady” (PE17/97).

Thus it may be that yi > ui was a general rule, and then ui > y in the final syllable of polysyllables. Note that there are plenty of Sindarin words with ui in final syllables, but they are mostly obvious compounds, which may have helped preserve the ui: Baranduin (baran + duin), Draugluin (draug + luin), Orodruin (orod + ruin), Thranduil (tharan + tuil). Conversely, where a monosyllable ended in a consonant cluster no i-intrusion occurred, as in S. dyrn plural of S. dorn “tough” (PE17/181) and S. gyrth plural of S. gorth “dead (person)” (Let/417).

All the attested examples of ŷ-plurals from ô-monosyllables are from Noldorin. It seems likely that the ui-plurals given above represent the normal plural pattern for all ô-monosyllables in Sindarin, and this was a conceptual shift in the behavior of i-intrusion between Noldorin and Sindarin.

Internal i-intrusion: There are a number of examples where i-intrusion occurs at morpheme boundaries rather than at the end of words. This only produces abnormal results in the (comparatively rare) case of second elements of compound that began with a “y”-sound, which is spelled i- at the beginning of Sindarin and Noldorin words. Bertrand Bellet pointed out a number of examples (VASN):

• S. einior “elder” (PM/358) = an- + iaur.
• S. eirien “daisy, *day-maiden” (SD/126) = ar- + -ien.
• N. eilian(w) “rainbow” < ON. elyadme (Ety/ƷEL, YAT).

In the last example, the “y”-sound did not vanish, but instead developed into the vowel [i] as was usually the case medially. This interior i-intrusion at morpheme boundaries seems to be limited to ancient compounds. More recent compounds (as seen from the fact that there is no evidence of i-affection in the initial element of the compound) are not modified:

• S. Arossiach “Ford(s) of Aros” (PM/358) = Aros + iach.
• S. Esgaliant “Bridge of Esgalduin” (WJ/333) = esgal + iant.
• S. Echoriath “Encircling Mountains” (S/138) = echor + iath.
• S. Inglorion “*Son of Inglor” (LotR/80) = Inglor + ion(n).

As pointed out by Bertrand Bellet, there are also few examples of interior i-intrusion in Noldorin and Sindarin words that are not at morpheme boundaries, most involving the verbal suffix -ia from primitive ✶-ya:

• N. teilio “to play” < ✶tyalı̯ā-, beside more regular telio (Ety/TYAL; EtyAC/TYAL).
  • Also related te(i)lien “sport, play” (Ety/TYAL).
• S. seidia- “to set aside” < ✶satya- (VT42/20).
• S. gleina- “to bound, limit” vs. T. glania- (VT42/8).

These examples are very much in the minority, however. The vast majority of Sindarin and Noldorin -ia verbs show no such interior i-intrusion. These may be transient conceptual shifts or perhaps some dialectical variation.

Conceptual Development: As discussed by Roman Rausch in his Historical Phonologies of Ilkorin, Telerin and Noldorin around 1923 (HPITN/§4.2.1), there are numerous examples of i-diphthongs in the final syllables of plurals in the Early Noldorin of the 1920s, so some form of i-intrusion must have existed at this early conceptual stage. There are quite a few variations in the Early Noldorin plural patterns, however, making it difficult to determine the exact phonetic rules. There are also numerous examples of i-diphthongs in plurals before consonant clusters, so it seems that such clusters did not inhibit i-intrusion in Early Noldorin (at least, not in all cases):

• ᴱN. eglainn plural of ᴱN. aglann “ray of light” (PE13/136, 158).
• ᴱN. failch plural of ᴱN. falch “cleft, ravine” (PE13/143).
• ᴱN. ewaist plural of ᴱN. awest “summer” (PE13/160).
• ᴱN. (n)guilt plural of ᴱN. golt “gnome” (PE13/123, PE15/62).
• ᴱN. guinn plural of ᴱN. gonn “stone” (PE13/123).

It seems the basic process for i-intrusion was the same for the Noldorin of the 1930s and the Sindarin of the 1950s and 1960s, but the phonetic developments are somewhat masked by shifting conceptions for diphthongal developments. To summarize the discussion above, Bertrand Bellet (VASN) identified four scenarios for i-intrusion, and I would add two more Sindarin-only scenarios:

• -eCi > -eCʲ > -eʲCʲ > -eiCʲ > -eiC.
• -ūCi > -ūCʲ > -ūʲCʲ > -uiCʲ > -uiC.
• -ǭCi > -ǭCʲ > -ōʲCʲ > -oiCʲ > -oiC.
• -auCi < -auCʲ < -auʲCʲ < -oiCʲ < -oiC, Sindarin variant of the above.
• -œCi > -œCʲ > -œʲCʲ > -œiCʲ > -œiC; much more common in Noldorin.
• -yCi > -yCʲ > -yʲCʲ > -yiCʲ > -uiC; monosyllables only, probably just in Sindarin.

There is strong evidence for the first two scenarios in both Sindarin and Noldorin, assuming you compensate for obscuring phonetic developments: ei surviving in Noldorin but becoming ai in Sindarin. The third and fourth scenarios are basically the same, but account for the early change of au > ǭ in Noldorin vs. the pre-intrusion change of ǭ > au in Sindarin. The fifth scenario could (in theory) have happened in Sindarin, but all the clear examples are from Noldorin, where they were more numerous due to inhibited i-raising of o to y in some Noldorin polysyllables. Possible examples of the sixth scenario started to appear in the 1940s (duil “hills”, SM/225) and it may have become the normal plural pattern for ô-monosyllables in the Sindarin of the 1950s and later.

As discussed above, there is clear evidence that by the Noldorin of the 1930s, consonant clusters inhibited i-intrusion. This seems to be the most common pattern in Sindarin as well, but in later writing the evidence is more mixed, with a few examples that seem to show i-intrusion despite consonant clusters.

GS/§4.95 GS/§4.99 GS/§4.105 GS/§4.231 possible medial examples: eilian(w), Dainthor @@@