Main accent in the Lost Tales period

In SQ (PE12:4ff.) and EQG (PE14:67) the three stages described above correspond to three clearly differentiated periods of the history of the Elves and their languages. The first stage was that of Primitive Eldarin or Common Eldarin, which at that time did not mean — as later in the Silmarillion — the language common to the Elves that started the March to the West, but the primitive speech of all the Elves, i.e. what Tolkien would later call the Quendi.3 That stage lasted almost till the sundering of the Noldoli, so it encompassed not only the oldest form of the Elvish language, but also some derived languages of the sundered Kindreds of the Elves in Valinor.

In those old languages the accent was free, i.e. there were no phonological characteristics of words that might be used to predict which syllables had to be accented. Accent position was mainly determined by etymology and grammatical categories (the past tense is cited as an example in EQG). And unlike in later stages, the accent was often borne by the final syllable. SQ provides some examples of Eldarin with the accent in different places, such as the oxytone kṇdōrā́ a bright dawn or paroxytone păχī́ănd an oration, among others (PE12:5, 8),4 and many more examples may be found in the Noldorin Word-lists and Dictionary written at the time of EQG (PE13:133‒65).

The second stage concerns only the languages of the Elves that remained in Valinor after the Flight of the Noldoli (there is no detailed account of the changes in the accentuation of Gnomish). That was called the Cor-elda or Cor-eldarin system in SQ, referring to the languages of the Eldar that remained in the city of Cor.5 In EQG Kor-eldarin refers generally to the languages of the Three Kindreds that dwelt in Kor since their coming in Valinor, including the Noldoli, so there the system of this second stage is called Valinorian (encompassing both Qenya and Telerin; see PE14:61).

In the Cor-elda or Valinorian system the main accent became attached to the root of the word or its basic syllable (or the one that was felt to be radical). That was normally the initial syllable, except in words with prefixes that were clearly identified as such. Thus, the Eldarin examples given above became kṇ́dṑra and pắ(h)īan(d).

There is an interesting case in the pair taχ‑sṓđā́ and táχsŏđa, two forms of an old compound related to the Q stems TAHA high and SORO sit, literally meaning seated high (PE12:21, 85, 87).6 Their accentuation suggests that the latter might be a Cor-eldarin development of the former, but it might instead be just a primitive variant, since both forms show the sounds χ and đ that had already developed into h and r (through z) in Cor-eldarin (PE12:16, 18, 24). In any event, the words with both accentuations seem to have coexisted (the paroxytone due to the recognition of taχ‑ as a prefix), giving two different forms of the adjective lofty in Qenya: tahōra < taχ‑sṓđā́, with χs > χχ > h (PE12:20), alongside tāra < táχsŏđa with the same changes, but additional vanishing of h after the stressed vowel (PE12:18), and regular development of the hiatus ā̆ŏ > ā (PE12:14), i.e. *táhŏza or *táhŏra > *táŏra > tā́ra.

Finally, the third stage was that in which Qenya, at some unspecified time in its later history, adopted its characteristic trisyllabic law. This forward displacement of the accent, which normally could not be further back than the antepenultimate syllable, is explained in SQ as an interchange between the main accent (formerly on the first syllable) and the secondary accent — see about the patterns of secondary accents below. Thus Cor-eld. kṇ́dṑra became Q kandṓra, whereas pắīan kept the initial accent because it developed a diphthong and became disyllabic in Q páiyan or páian (PE12:8). Other examples given in SQ of accent changes from Cor-eldarin to Qenya are kṛ́pàlima > kàrpálima tricky, clever, wily, lṇ́tànwa > lantánwa (seemingly a form of the verb lant‑ drop, fall), and ū́kā̀rĕlĕ > ukā́rele (perhaps derived from *ū̆kar‑ undo).

Main accent in the Silmarillion period

In later accounts, the primitive stage of free accent was associated with the Q(u)enderin period (in English called Primitive Quendian, Common Quendian or just Quendian), which meant the primitive common language of the Quendi (PE18:55 n.111). Strictly speaking, that would refer only to their language before the sundering of the Eldar and Avari, but the accentual pattern was retained in Common Eldarin and its derivative languages, with some small divergences between them.

Many of the characteristics of the primitive stage described in the Lost Tales period were kept, particularly the frequency of oxytone (final-accented) words, and the grammatical significance of the accent, although in QP this was a feature of Lindarin, less present in Noldorin and Telerin (PE19:56).7 Some examples are:

• The accentual patterns of the past tense, which had already been mentioned in EQG. The accented suffix ‑nḗ of the weak pa.t., as in Eld. ortānḗ raised, is the most commented upon (PE18:56, 106; PE19:56; PE22:96, 104, 135); TQ2 compares it with the strong pa.t. ending in ‑ē, usually accented on the previous syllable, as in mántē ate (PE18:106).
• The accented ‑ā́ used for the present tense, such as in talātā́ is slipping down, istā́ is aware or mātā́ is eating (PE22:134)
• The stem of causative verbs ending in accented ‑tā́ or ‑yā́, such as ortā́ raise, in contrast with other derivative verbs with similar but unaccented endings such as órtā̆ rise, ascend (PE18:106, PE22:114, 135, 159).
• Oxytone adjectives ending in ‑ā́, like kiryā́ swift (PE18:106; PE19:56). More specific long adjectival suffixes are mentioned in TQ1 and in a text about the Eldarin noun structure contemporary to TQ2, such as ‑rā, ‑dā, ‑nā, ‑yā ‑wā (PE18:56, PE21:82).

A detail added in TQ is a set of restrictions for the placement of the primitive accent, which in SQ were only commented on as a result of changes in the Cor-eldarin system, namely that the main accent could never be on a final syllable ending in a short vowel, and it was seldom on short syllables adjacent to long ones (PE18:55, 105); thus, forms such as tā̆rā̆kắ or tārắkā̆ were disallowed. However, the second of these rules was a softer restriction, which allowed some exceptions; thus initial short syllables, which were often the main stem syllables, could be accented even if followed by a long one, as in tắrākā̆. And according to TQ1 (but not in TQ2), before a final long syllable that was recognized as a suffixal or inflexional element, a preceding short syllable could also bear the accent, as in tărắk-ā. The lack of accent on final short vowels is also mentioned in QP, although there it is said to be a specifically Lindarin modification (PE19:56).

There was a later period in which the tongues derived from Eldarin (Lindarin/Quenya, Noldorin/Sindarin and Telerin) changed their accent patterns, but the particular details of those changes differ between texts. The most prevalent idea is that — as in the earlier conception — at least in Ancient Quenya the main accent was at some time retracted to the first stem syllable, which was normally the initial syllable (cf. PE22:105, R:68).

According to QP, this happened to all Eldarin tongues, and it actually happened first in Telerin and later in Noldorin, whereas Lindarin maintained the older accent-pattern with lesser modifications for a much longer time (PE19:60 n.152). On the other hand, in a later text, written c. 1968, the opposite is implied: commenting on the words for the flat of the hand — Q palta, T plata, S plad, all derived from CE palátā — Tolkien noted that Telerin and Sindarin had lost the Eldarin unstressed vowel of the first syllable, whereas in Quenya the position of the Common Eldarin stress was not important, since at an early period Quenya had shifted the stress back to the first syllable (VT47:8‒9).

The accent shift that resulted in the system dicated by the penultimate law happened in Quenya before the classical period, so the position of the main accent in Parmaquesta was already as described in The Lord of the Rings, although it was a gradual change, and the later Tarquesta exhibited some differences in the acoustic characteristics of the accent (PE19:61), which will be described in the next sections. How and when the other tongues converged to the same system is not described in the available texts.

Secondary accent in the Lost Tales period

Among the earlier texts there is no single full account of the placement of secondary accents through the history of the Eldarin languages. For the first, Primitive Eldarin, period there is no information at all. Regarding the second stage it has already been noted, in the discussion about pitch, that according to EQG secondary accents were often placed, with raised pitch, in the position of the older main accent. This also seems to be the case of some Cor-eldarin words given earlier in SQ. However, there is a noticeable change between SQ and EQG with respect to this.

There is a comment inserted in SQ stating that the secondary accent appears to have been strong in Cor-eldarin, and able to stand on a syllable adjacent to that of the main stress (PE12:5). EQG, on the other hand, restricts this possibility, noting that when the old accent fell on a syllable immediately following the new, no trace of the older accent was left (PE14:67). Thus many of the Cor-eldarin examples from SQ, such as kṇ́dṑra, kṛ́pàlima, lṇ́tànwa or ū́kā̀rĕlĕ, would not be valid in EQG. This implicitly means that for all those cases, the penultimate law of later Qenya was no longer explicable as a shift between the main and secondary accent, as it was in SQ.

SQ also provides some details about the distribution of secondary accents in the later Qenya stage (PE12:5, 7, 26‒27). The general rule was that secondary accents were placed on alternate syllables in either direction from the main stressed syllable. This is reminiscent of the usual pattern of Finnish — although apart from that, the accent system of Qenya and other Elvish languages is not like that of Finnish, in spite of its influence on other phonaesthetic aspects.9 That rule, however, could be broken for various reasons:

• The relative weight of adjacent syllables — following the hierarchy of weights explained in the exceptions to the penultimate law — could displace the accent one position, leaving two adjacent unaccented syllables (but not more than two). This could happen by heavy or long syllables attracting the accent, or by the weakest syllables pushing it away.
• The first syllable of a word could bear a slight secondary inital accent, even if it was adjacent to the main accent. Note that this pattern (ò‑ó initially) does not clash with the ó‑ò forbidden in EQG. This might be due to the special significance of inital syllables that is noted in later texts.
• Conversely, final short syllables tended to be unaccented, even if they were in a position that favored a secondary accent — normally in a proparoxytone word (ó-o-o), where there is one syllable between the main accent and the final. This weakening of the accent was more marked in vowel-ending words, such as noldova belonging to gnomes vs. noldomar gnome-land, although the details on this point diverge between fragments of the same text. Thus, according to PE12:5 a final vowel never bore a secondary stress, while a slight beat was observed in such words if the vowel was followed by a consonant.10 On the other hand, PE12:27 tells that in words ending with a short final vowel the secondary stress was comparatively weak, implying that they could bear some stress, although less marked than if the word ended in a consonant (or a long vowel, which was rarer, perhaps impossible in later Qenya).

Secondary accent in the Silmarillion period

Most of what can be said about the secondary accent according to Tolkien’s later conception can be found in the first version of QP (PE19:57‒60). The description found in that text has many points in common (even in the wording) with that of the Qenya secondary accent in SQ, excluding the inserted commentary about the possibility of adjacent stress in an earlier stage, which had been precluded afterwards in EQG. Actually, the patterns of secondary accent adopted in QP from SQ were not exclusive of the later stages of the languages; they existed since their primitive or archaic stages, and were very similar in all branches.

The points shared between the Qenya of SQ and the archaic accent system described in QP are:

• Secondary accents were generally placed on alternate syllables in either direction from the main stress (ó‑o‑ò, ò‑o‑ó), although this pattern could be broken by the tendency of long syllables to attract the secondary accent and short ones to be unaccented. Thus, in sequences such as ó‑ŏ‑ŏ‑ō (or ō‑ŏ‑ŏ‑ó) the secondary accent would fall on the last (or first) long syllable, instead of the short one that alternates with the main accent, as seen in the example of kárpalimā̀.11
• The secondary stress could not fall on a syllable adjacent to the main one, except if it was the initial syllable of the word — adding that this happened only in the rare cases of compounds with a stressed monosyllabic first element that were still felt as loose compounds, such as *tàlgárta high boot < tál foot + gárta hide, leather.12
• Short final syllables never received a secondary accent (as in mándălĕ, where the last syllable alternated with the main accent but remained unnacented), clarifying that final syllables ending in a consonant counted as long, as in the case of nínkwĕrĭtàr — yet another example of the ó‑ŏ‑ŏ‑ṑ pattern commented on above.13

But QP also included new details that introduced some asymmetries in the archaic accentuation patterns:

• In sequences of three short syllables after the main accent (noted as rare), the second syllable received a secondary accent, keeping the usual alternating pattern ó‑ŏ‑ŏ̀‑ŏ (e.g. in nínkwerìta). But before the main accent the secondary was pushed back to the third syllable, resulting in the pattern ŏ̀‑ŏ‑ŏ-ó (e.g. kằlŭmălánda or kằlŭmălṓ); and a long syllable adjacent to the main stress could move it even further back, as in èrĕmăloitḗ.14
• As a general rule, the secondary accents that preceded the main were the strongest. However, preceding accents that fell on short syllables — as in the situations just described — were weaker than accents borne by long syllables separated by one or two syllables in either direction.

According to this account, the principles that governed the position of secondary accents were hardly affected by the Quenya accent shift. The only exception that is noted is that a secondary accent (often with low tone) could stand on an initial syllable imediately before a long main stress (PE19:60). Since this was already allowed in the archaic period for loose compounds, that comment might mean that in Quenya this pattern was generalized to other types of words. This is corroborated by the example of cṑmállo, which though unglossed seems to be the ablative declension of the noun *cōma (PE19:59).

Accent and vowel length in the Lost Tales period

According to SQ, the development of the Qenya phonology favored short vowels, since in all the historical transitions between accent systems there were circumstances in which long vowels were shortened. In the transition from the primitive to the Cor-eldarin stage this shortening happened sometimes to long unstressed vowels that followed an accented long vowel in initial position (words starting with ṓ‑ō > ṓ‑ŏ), and also to long vowels before any consonant group, as well as after certain groups if the vowel was final; later in Qenya all unaccented syllables between two accents were shortened as well (PE12:4‒5, 7).

Eventually the vowels of all unaccented syllables in Qenya became short (PE12:6). This, combined with the accentuation rules commented on above, means that adjacent syllables with long vowels would be rare in Qenya; if ever, they would only be allowed in an initial sequence of accented syllables of the type ṑ‑ṓ. In the Qenya Lexicon, the only (dubious) example that can be found is lāmáre flock, perhaps representing *lā̀mā́re (PE12:50).15

Also the vowels of closed syllables — those followed by consonant groups or long consonants — were always short in Qenya, although such syllables were considered long. This can be seen in the contrast between etymologically related words with long stem vowels that were shortened before consonant groups like lūme time vs. lukta period (PE12:56), nōre native land, nation, family, country vs. nosta birth, birthday (PE12:66), and many more.

The quantitative laws of Qenya purportedly dictated other circumstances under which vowels could be lengthened (PE12:8‒9), but they were not described explicitly. According to the text, vowel length in Qenya was no longer etymological and morphological purely (…) but is mostly to be determined by that vowel’s place in the word and its neighbourhood (PE12:6). However, in many words of the Qenya Lexicon vowel lengthening seems to happen as a grammatical feature. Thus, we see consistently lengthened vowels before the adjectival suffix ‑va in mirūva like wine, winy, palāva wandering, turūva wooden, urūva like fire (PE12:61, 71, 96, 98), or in the past tense forms seen in kāle shone, līre sang, mēle loved, and in the tables of verb forms attached to the Qenya Lexicon (PE12:44, 54, 60, PE14:28ff).

EQG shows another example of vowel lengthening that can also be said to be a grammatical feature, although it has a phonological ground: the plural declension of vowel-ending nouns, which is formed by ‑li or other derived suffixes, comes with lengthening of the last stem vowel in the case of proparoxytone nouns like tantare dance, pl. nom. tantarēli, gen. tantarēlion, etc. (PE14:44, 73). This is explained as a restoration of historic length, although it also helps to preserve the prosody of the basic Qenya form. The addition of the declensional suffix moves the syllable with the original main accent (the antepenult) to a position where it can no longer bear it due to the penultimate law; but by virtue of lengthening the vowel before the suffix, the main accent is moved forward and leaves the originally accented syllable in a position that favors a secondary accent, thus: sg. tántare, pl. tàntarḗli.

The vowels of accented open syllables (followed by a single consonant or another vowel in hiatus) could be short (unimoric) or long. In the latter case, they could be either half-long (bimoric) or full long (trimoric), depending on the length or weight of surrounding syllables. In general, vowels were longer in syllables under the main accent than under secondary accent, and also in trochaic position (followed by a single short syllable) compared with dactylic (followed by two short syllables), such that: (a) vowels in dactylic position could only be long (bimoric) if they carried the main accent, and (b) vowels in trochaic position were bimoric if they had a secondary accent, and trimoric if it had the main stress (PE12:6). These rules can be illustrated with some words of the Qenya Lexicon — with diacritic marks altered from the original to represent more accurately the purported accentuation and vowel length:

• Trimoric vowels in anû́va doughty or fanốre a day-dream (PE12:31, 37), where the main accent is on open syllables in trochaic position.16
• Bimoric vowels in the related words anū́vie manhood, doughtiness or fanṓriva absent-minded, where the main accent has been moved to a dactylic position.
• A bimoric vowel in the secondary accent of mṑrilánta nightfall, which is in trochaic position (PE12:62). The main accent falls on a closed syllable, which therefore has a unimoric vowel.
• Unimoric vowels in the inflected forms of Teleriness‑ the valley (or land) of the Elves (PE12:91), e.g. accusative *Tèlerinéssen,17 where the secondary accent is in dactylic position, and the main accent is on a closed syllable.

This set of rules also specifies that vowels under main stress in spondaic position are bimoric. Spondaic patterns (two consecutive long syllables), as discussed above, could only happen initially, with the main stress on the second; examples of this could be lìndṓrea singing at dawn (esp. of birds) (PE12:54), or the case of *lā̀mā́re commented on above. There is no indication of how long would be the first, secondarily accented vowel in such spondees, if they ever existed in long open syllables as in the latter example.

Accent and vowel length in the Silmarillion period

In later accounts of the Eldarin phonology, the variation between short and long vowels was part of the base-structure or sundokarme explained in TQ1 and TQ2. The most primitive Quendian bases were formed by syllables with short vowels, but ómataile (vowel-lengthening) was one of the processes used to make derivative bases (PE18:45, 95). Vowel-lengthening also continued to be a grammatical feature of Quenya, as in the conjugation of verbs — not only the past, but also of the present, as in tul‑ come, tūla am coming (cf. PE22:95‒96), and in the declension of nouns. But as in the previous account, during the development of the Eldarin languages vowels tended to be shortened, essentially in the same circumstances related to the accent, although adjusted to the new historical framework.

In TQ1 the shortening of vowels in unstressed syllables adjacent to long stressed syllables, or before consonant groups, was originally described as a process that happened in Common Quenderin, although this was emended to date it to the Eldarin period (PE18:56‒57). In QP this was further restricted to a feature of Quenya that happened before the accent shift. In later Quenya, after the accent shift, vowels tended to keep their length, although they were still shortened in some circumnstances, and in Parmaquesta the archaic length was still preserved, at least in writing.

One of those later shortenings happened to the long final vowels, which had survived in Old Quenya (cf. PE21:69), but now were always in an unaccented position except in monosyllables. As a result, for instance, the mark of the accusative declension was lost in Tarquesta (VT6:14, PE19:77). QP also comments on the shortening of long vowels in the antepenult syllable just before the main stress (i.e. in paroxytone words), with the exception of trisyllables where the long vowel occurred in the initial stem-syllable (PE19:59). It is uncertain, however, what kind of Old Quenya words would have had a long vowel in such a position, since non-initial syllables next to the main stress could not be accented, and unaccented syllables had already been shortened before that time.

The answer may be implicit in the examples given in the text: torṓmànna > Parmaquesta torō̆mánna, where the antepenult was shortened in Tarquesta torománna, vs. the trisyllabic kṓmallṑ where the length of the antepenult was retained in Q cōmállō̆. Although not stated explicitly, these words seem clearly to be the allative and ablative cases of Q torṓma < torṓmā and cṓma < kṓma, respectively.18 And the choice of such examples might indicate that the situation whose resultion was being described, i.e. the contact between long vowels and penult main stress in Old Quenya, typically happened in the derivation of new words or grammatical forms that were particular to Quenya. To support this, it can be noted that various texts of this period stated that the declension of nouns was not fully developed in Eldarin, and the adverbial cases, especially the allative ‑nna, were a particular development of Quenya (PE21:63, 66, 69, 72, 79). Notice also that the forms torṓmànna and kṓmallṑ that are given in the text do not even coincide with the situation that is being discussed (long vowel before the main stressed penult), which would rather be *torōmánna and *kōmállo, so perhaps they were not actual archaic forms, but analytical reconstructions of the raw forms — before accentual and vocalic adjustment — resulting from adding the adverbial suffix to the noun stems.

Regarding the quantity of long vowels, it is known from the Appendices to The Lord of the Rings that Sindarin had overlong vowels, especially in stressed monosyllables or words related to them, where such vowels are usually marked with a circumflex accent, e.g. dûn west, rhûn east, or Annûn sunset Amrûn sunrise (LR:1093). And although they were not included in the published Appendices, other linguistic texts related to their making and later writings contain examples of Quenya monosyllabic stems that, when standing as full words, are likewise marked with a circumflex, so they might also be considered to have such specially prolonged vowels: e.g. nôr / nŏr‑ land (PE17:107), lîs / līr‑; honey (PE17:154), or yên / pl. yēni long year of the Elves (NM:348).19

If related to the like cases in Sindarin, those Quenya examples could be used to interpret the concept of stressed monosyllables mentioned in the Appendix of LR, as monosyllabic words derived from Eldarin stems that received the main stress in derivatives, compounds, and phrases that contained them. This idea is reinforced by a commentary in the section of QP that deals with the development of vowels in stressed stem-syllables, which describes the origin of trimoric vowels (of triple length) in Eldarin languages.

This text explains the general development of stressed stem-syllables in Eldarin words of any length, not only monosyllables. It notes that trimoric vowels were not original, but they were formed in Common Eldarin by lengthening of older bimoric vowels, e.g. by compensatory lengthening after loss of ʒ in māʒ > mâ hand, or in ā + a from āʒă.20 Still during the prehistoric period, in medial position those vowels were reduced to normal length (i.e. bimoric), but their full trimoric length was kept in final syllables not followed by a consonant (PE19:53, 105‒106).

During the development of the Eldarin languages, as explained above, original long vowels were regularly shortened when they fell on unaccented syllables. Therefore, it may be assummed that the preservation of the trimoric quality of those final vowels was bound to their coincidence with the syllabic accent, which could happen since Common Eldarin words were often oxytone. On the other hand, in Quenya (as in Sindarin, although that is not dealt with in QP) the main stress was generally removed from the last syllable in polysyllabic words. Accordingly, it is said that in Parmaquesta these longer than normal vowels could only occur in monosyllables (PE19:53).21 And if the tendency to unstress and shorten final long vowels also affected monsyllabic words, they would be shortened to normal length, such that the existence of stressed monosyllables with overlong vowels should be considered to be a particular feature of Parmaquesta alone.

On the other hand, in poetry the older quality of those vowels could be partially restored, again under the influence of the accentual patterns emphasized in verse. Indeed, discussing the prosody of Namárië, Tolkien noted that in Quenya final vowels, though normally short and unstressed,

had nearly all formerly been long vowels (…) so that in the very frequent cases of words ending in two short syllables, as ūnōtime, tellumar, lumbule, hīsië, etc., they received a light stress that could be used metrically. (…) In exclamatory words such as namārië, the length of the final vowel was often retained, and could in a farewell cry be much extended. (R:69)