Sindarin Stress — A Muta cum Liquida Rule?

In Appendix E Tolkien described Sindarin (and Quenya) stress as follows:

In words of two syllables it falls in practically all cases on the first syllable. In longer words it falls on the last syllable but one, where that contains a long vowel, a diphthong, or a vowel followed by two (or more) consonants. Where the last syllable but one contains (as often) a short vowel followed by only one (or no) consonant, the stress falls on the syllable before it, the third from the end. (LotR:1116)

This is also the most basic rule for Latin stress. Usually this is described in the following terms:

If a syllable contains a long vowel (L Rōmānus, Q Elentāri) or ends in a consonant (L sacer·dōs, S Pelar·gir) it is called heavy, otherwise it is called light. Tolkien says for the second case that the syllable’s vowel is “followed by two (or more) consonants” since the maximal onset principle would dictate that a single consonant would count towards the following syllable (meaning that e. g. in S Dene·thor, the second syllable would be light. This example also shows that what counts is the spoken word: while ⟨ph, th, ch⟩ are written with two letters, they each represent a single consonant, /f, θ, x/ in IPA). With those terms one can define the stress placement as follows:

• In words with one or two syllables, the stress always falls on the first one (L púer, S rhū́n)
• In words with at least three syllables, the stress falls
  • on the second‐to‐last syllable (the penultimate) if it is heavy (L sacérdōs, S Pelárgir),
  • on the third‐to‐last syllable (the antipenultimate) otherwise (L fḗmina, S Dénethor).

As the given examples illustrate, I use a variant orthography for the purpose of discussing stress here: Long vowels ī, ȳ, ē… are universally written with macrons and the main stress in indicated by the acute é, á, ū́…. Where necessary, syllable brakes are marked by a middle dot ·, and secondary stress is marked by a grave ò, ì ỳ,…

Proposal

This is not however the complete rule for Latin stress placement. Among other exceptions, there is a rule traditionally referred to by the name of muta cum liquida:

there are also so‐called ambivalent syllables when either a short vowel is followed by two consonants the first of which is some muta [= the stops p, t, k, b, d, g] or the semivocalis f and the second a liquid [= l, r], or … (Aelius Donatus, Ars grammatica, Liber II, p. 369)

In the terms above this rule means that the combinations are not split across syllable boundaries, but form the onset of the following syllable together, meaning that e. g. ínte·grō, téne·brae are stressed on the antipenultimate. The reason why this is possible is, that words and therefore syllables can start with those combinations, for the combinations gr‑, br‐ e. g. grex, brevis.

This is not an absolute rule in Latin, consider for example the following verse:

et pri|mo simi|lis volu|cri, mox |vera vo|lucris (Ovid, Metamorphoses, 13:607)

Here volucri(s) is once divided according to the muta cum liquida exception and once as voluc·ris with a heavy second syllable. However the scanning according to muta cum liquida seems to be the rule rather than the exception, at least in Classical time.

In the poetry from the Lord of the Rings there are (at least) two places where a Sindarin name appears to be stressed not according to the simple rule from Appendix E:

Séek for |the swórd |thát was |bróken
In Ím|ladrìs |it dwélls (LotR:246)

The metric feet in this poem are quite irregular, but every line contains three raises, which in the second one can be only achieved if Ímladrìs is stressed on the first and third syllable, not on the second as the rule from Appendix E would predict.

Of mígh|ty kíngs |in Nár|gothrònd
And Gón|dolìn, |who nów |beyónd (LotR:316)

Here every line is made up of four iambic feet, making it necessary to stress Nárgothònd like Imladris above.

Based on those examples, some have proposed that, similarly to Latin, Sindarin might have a muta cum liquida exception, making the stress in Nárgo·thrònd and Ímla·drìs regular. On first glance this sounds like a fairly reasonable explanation. Like Latin, Sindarin permits complex onsets, and thr‑, gr‐ are among them. First to get an overview of the scope of such a rule, let’s list all the possible complex onsets.

The question of what can actually start a word in Sindarin is obscured somewhat by the presence of consonant mutation. In their unmutated form, for example, no Sindarin word begins with chw‐, but under mutation this can occur: gwath ‘shadow’ → a chwath ‘and a shadow’ (PE17:41). In this case one could argue that a chwath actually represents a single unit /ax.waθ/ where chw‐ doesn’t actually starts a word. Similar arguments can be made for most types of mutation, but in case of grammatical lenition the wordhood of the mutated element is clearer:

• bast → penim vast ‘we have no bread’ (PE17:144)
• daer → Athrad Dhaer ‘great ford’ (WJ:338)

No attested Sindarin word shows the cluster mv or ddh, so these must both be two separate words.

With this sorted out, here is a list of all the attested complex onsets in Sindarin, including those produced by lenition (len.):

• pl/pr, tr, cr/cl (plad, trann, claur) – those are irrelevant for this rule, since p, t, c can never appear medially (where they became b, d, g).
• br/bl, dr, gl/gr (brass/Celebrant, draug/nedrui, glad/aglar)
• fl, thr, thl (flad/N Alflon, Thranduil/athra‐, lhaw → len. thlaw/Gwathló) – Thranduil seems to be a special reduction starting in thr‐, the base word for spring is tharan. However a lenition of rh‐, e. g. rhûn, → len. thrûn would be an option as well (for the lenition and nasal mutation of rh‑, lh‐ see PE17:147)
• vr, dhr, (brand → len. vrand/silivren, draug → len. dhraug/Caradhras)

When applying the muta cum liuida rule Latin usually pays attention to morpheme boundaries, counting gr‐ in inte·gro as a complex onset, but -br‐ in the compound ab·rumpo as divided between the syllables. Since Nargothrond and Imladris are both compounds (of narag‐ost‐rond and imlad‐riss/rist respectively), we will assume that this does not apply to the Sindarin version of the rule, otherwise we would loose the reason for why we proposed the rule in the first place.

Counter‐evidence

Sindarin Poetry

However I believe there is considerable evidence against this view. Firstly, the examples above are from English rather than Sindarin poetry. In Tolkien’s Sindarin poetry one can find two relevant examples:

A Él|berèth |Gilthó|nièl
silív|ren pén|na mī́|rièl (LotR:238)

This poem appears to be written in a iambic tetrameter, which would dictate that silívren is stressed on the penultimate as Appendix E explains, not as síli·vrèn as the muta cum liquida rule predicts.

ánnon e|dhéllen, |édro hi |ámmen
fénnas no|góthrim, |lásto beth |lámmen (LotR:307)

Here each half‐line seems to contain a dactylic followed by a trochaic foot, dictating a stress nogóthrim, where muta cum liquida would predict nógo·thrìm.

This shows that at least all Sindarin poetry can be explained by the rules in Appendix E and indeed in the two cases, where muta cum liquida could apply, it is absent. The first example (and the rest of A Elbereth Gilthoniel) also show the rules of secondary stress in Sindarin, namely that counting from the main stress in both directions syllables alternate between being unstressed or having secondary stress.

English Poetry is Unreliable

Since the rule was derived from observations about Sindarin words in English poetry, it might prove useful to look for more examples there as well. A very useful source for this are the Lays of Beleriand, since they contain huge amounts of poetry mostly in two forms:

• alliterative verse: Here words from each half‐line must alliterate, that is the first consonant of their stressed syllables must be the same (Here marked with underlining).
• octosyllabic rhyming couplets: Here lines are composed of iambic or trochaic tetrameters and two consecutive lines usually rhyme. For two words to rhyme, usually the first syllable of the rhyme must be stressed.

Thangorodrim – The rule is not universal

The muta cum liquida rule – even in English poetry – is not universal, which can be best seen with the name Thangorodrim:

Then Thalion was thrust / to Thàngoródrim, (8:92)
of Thàngoródrim’s / thunderous mountain. (33:714)
of thunder‐riven / Thàngoródrim. (37:867)

In the alliterative poems it constantly participates in th-alliteration, implying the stress Thàngoródrim, but never in g-alliteration as the muta cum liqida rule would predict with Thangóro·drìm.

from Thàn|goród|rim’s stó|ny crówn. (212:1653)
on Thàn|goród|rim’s thúnde|róus |tówers. (227:2051)
that wéighs |on Thàn|goród|rim’s fóot (282:3528)

In the rhyming couplets Thangorodrim also always occurs as Thàngoródrim.

In total Thangorodrim occurs twelve times in the Lays, always with the stress Thàngoródrim. While the muta cum liquida rule in Latin is somewhat optional, a word that defies it with this regularity would be unlikely.

Fingolfin, Balthronding, Unothradin – The rule is not sufficient

Proposing a muta cum liquida rule is also not sufficient to explain similar aberrations in stress. To show this, let’s consider a couple other names that should have their stress on the penultimate, whether one proposes a muta cum liquida rule or not:

Fíngolfìn’s son; / Féanòr’s children, (96:49)
from defeat fending / Fíngolfìn’s son, (98:125)
But Finweg cried / Fíngolfìn’s son (136:1)
to bend whose bow, / Bálthrondìng named, (117:773)

Here the alliteration implies the stress pattern Fíngolfìn, Bálthrondìng, the same that we observed in Nárgothrònd, Ímladrìs, but since lf‑, nd‐ cannot start a word in Sindarin, the muta cum liquida rule does not cover it.

Ninin Ùnothrádin, / Unnumbered Tears. (6:13)
of Ninin Ùnothrádin, / the Unnumbered Tears? (11:218)

Here one can see what is in a sense the reverse case: According to the rules from Appendix E the stress should be unóthradìn since the last two syllables are light, but in the verse the word alliterate with other vowels, implying the stress ùnothrádin. Again this remains uncovered by the muta cum liquida rule.

Conclusion

Those examples leave two possible conclusions:

1. There is a very large class of words for which the stress rules from Appendix E are wrong, with both syllables predicted to be light attracting stress (unothradin) and syllables predicted to be heavy not attracting stress (Fingolfin, Nargothrond…). A subset of these aberrations might be explained with a muta cum liquida rule but there are also some words that consistently don’t participate in this exception (Thangorodrim), so if it exists, it must be part of a complex system.
2. It is a systematic error to predict the stress of a Sindarin word by looking (just) at its occurrences in English poetry. The stress rules for Sindarin itself are as described in Appendix E, but in English poetry Sindarin words can appear with different stress patterns according to poetic license.

I think (2) is the much better explanation of what is going on in this examples, since it is both easier and adheres more closely to Tolkien’s stated stress rules.

The case of Fingolfin provides more evidence for case (2); while it shows up in the alliterative verse as Fíngolfìn, in the rhyming couplets it also occurs as Fingólfin.

Fingól|fin stóod: |his shíeld |he bóre (284:3539)
Fingól|fin líke |a shóo|ting líght (285:3574)
Fingól|fin’s hígh |and hó|ly tómb, (287:3630)

If one proposes that English does reflect the Sindarin stress patterns, that means Sindarin words would have variable stress patters, another deviation from Appendix E. If on the other hand one assumes the irregular stress to be a result of their occurrence in English poetry, it becomes relatively easy to explain: The Sindarin stress Fingólfin fits well into iambic feet, so it can be left unchanged in the couplets, but in the alliterative verse the stress changes to Fíngolfìn, perhaps because Tolkien felt like F was the more ‘distinctive’ consonant in the name which he wanted to participate in the alliteration patterns.

One interesting point to observe is that in all listed cases where the stress doesn’t line up with the system described in Appendix E, the stress is instead shifted to the first syllable of the word. English – like most Germanic languages – has some preference for word‐initial stress, and while there is some debate on precisely how this rule works and especially for English there are large classes of exceptions to this rule, the result is that most speakers would find it acceptable to stress Elvish words on their first syllable. Thus it seems most likely to me that the effects on stress we see in the examples above are an effect of English rather than Sindarin.

Resources

• On Complex Syllable Onsets in Latin (Brill 2010).

---

A muta cum liquida rule for Sindarin has been proposed in various places, e. g.

• https://menegroth.github.io/stress-in-sindarin.html
• http://archives.conlang.info/gho/bhelgho/tree.html
• https://mildred-of-midgard.dreamwidth.org/158375.html
• https://forums.signumuniversity.org/index.php?threads/old-discussion-from-ep-133-the-metre-of-boromirs-prophecy-poem.4203/

and this is meant as a response to those proposals, but my main reason to publish this is less because I feel this is a controversy in the study of Sindarin that needs settling and more to test out the shiny new incarnation of Aglardh, eglerio Adu an adadabad i·had hen.

Article as PDF (again, mainly for testing purposes)