Post 4APJ5aeRww8

Fiona Jallings Jan 22, 2015 (23:01)

+Jenna Carpenter  and I have been chatting about Sindarin pronunciation, and the bit about the pronunciation of the letter I came up, specifically its mention in The Road Goes Ever On. In there, it's mentioned that I should be /ɪ/ when short. We hunted through recordings of Tolkien reciting Sindarin and Quenya Poetry, and found no consistency. It seems that he pronounced short Is like the I in "machine" most of the time, but it became like the I in "bit" before nasals and Ls - a typical Englishization of the pronunciation. This is the only place I've seen it mentioned to pronounce short Is in such a way. Any thoughts?

Olga García Jan 23, 2015 (16:54)

Here's some info on Quenya, phonetic and otherwise:
http://en.m.wikipedia.org/wiki/Quenya

Olga García Jan 23, 2015 (16:57)

And here's the info for Sindarin:
http://en.m.wikipedia.org/wiki/Sindarin

Roman Rausch Jan 23, 2015 (18:57)

Tolkien had a noticeable English accent when reciting Elvish, I'm afraid, so I wouldn't use the recordings as a guideline.
In RGEO it reads:
The short vowels may be rendered as in E. sick, bed, hot, foot (for ŭ), though ŏ is intended to be rounder than in modern E.
This is ambiguous, since English only has short [ɪ], no short [i] (there is the same problem with [u] vs. [ʊ] and [a] vs. [ɑ], btw). I would say that the intended sound is [i] and English 'sick' [sɪk] is just given as an approximation.
In the LotR appendix the sound is compared to English machine [...] irrespective of quantity. That would be [i:] minus the length = [i].

Fiona Jallings Jan 23, 2015 (20:38)

Ah, so he was probably referring to the actual length of the vowel, not its lax/tenseness, or that is the consensus here.

Jenna Carpenter Jan 23, 2015 (22:43)

Regarding 'no short [i] (in English)' that's not entirely correct. In NRP the HAPPY vowel has undergone phoneme neutralisation and is now much closer to the KIT vowel than the FLEECE. (Mark I say NRP only as it remains ɪ in Scots and Northern English dialects and [e] in RP).

Fiona Jallings Jan 24, 2015 (18:42)

In my dialect, (Northwestern, lower-48 USA English) in mono-syllable words, we're seeing some changes to VC at the ends of them. If the consonant is a D or T, then it's sometimes dropped, (more often it's only halfway articulated, and you can't hear the it) and the vowel is lengthened or left short depending on if it's voiced (D) or not (T). So, "bet" and "bed" become /bɛ/ and /bɛ:/.

As interesting as this case of compensatory lengthening is, It's not the dialect that Tolkien was using for reference.

Jonathon Omahen Feb 05, 2015 (21:57)

Tolkien is rather famous for essentially biffing it on his own pronunciation with regards to Sindarin and Quenya. Admittedly, he has been pretty upfront about not being a "native" speaker, his accent dominates his pronunciation of the languages. Working from his written descriptions, there is room to argue for some dialectical variation.

While it would be consistent with proto-Elvish, Qenya, and Quenya to argue for vowel length being phonemic, you could easily make a case for Sindarin (and related) to not being marked by length, but by quality. Moving a front-high vowel to a mid- or low- vowel position would be arguably compensatory for a loss in length distinction. This would be analogous to what we see in much of the Celtic world (re: Welsh, Gaelic, Breton, etc.). Accenting and the short/long distinction can be wrapped up in the allophonic realisation of the vowel, rather than in the measured length (of the morae).

Or maybe I'm just way off here?

Fiona Jallings Feb 06, 2015 (04:11)

Hmmm... there's three different vowel lengths, not just long/short. Maybe in a dialect of Sindarin such a thing could have happened, but that's not what it seems Tolkien was describing in the back of the LotR.

Jonathon Omahen Feb 06, 2015 (18:16)

+Fiona Jallings , forgive me but where are you seeing three distinct vowel lengths? Though not beyond possibility, I haven't seen any examples of phonemic vowel length beyond long and short (hence IPA doesn't provide an easy way to notate, for instance 'middle', vowel length).

Strictly referring to the Appendix (LOTR AE), we see that:
   * Vowels can differ in length and quality
   * Sindarin vowels don't vary in quality between long and short (LOTR AE p4: "In Sindarin long e, a, o had the same quality as the short vowels, being derived in comparatively recent times")
   * References to long vowels in Sindarin would then logically mean that vowel length in Sindarin is phonemic, with no corresponding change in vowel quality (quite different from what I wrote above, incidentally)
    *That vowels represented by English orthography  "...were of normal kind, though doubtless many local varieties escape detection" (LOTR AE p4), seems to indicate that dialectical variation can come into play (as above), wherein conceivably some Sindarin speakers could exhibit quality differentiation (or allophonic alternation) using a more "Quenya"-like "open" and "close" vowel differentiation
   * Stressed monosyllables and vowels had an especial weight given to them, though this is not unlike English in many ways, and of course Welsh. This stress would be indicated in the usual ways: tone, length, volume; it would be necessarily indicated by a stress marker in IPA. I would not call this "especially long" as does Fauskanger, but it would presumably seem more "weighty" as it is stressed.

All in all, the reading of only the Appendix would lead to a rather simple vowel system, with phonemic vowel length, and Sindarin exhibiting no change in quality of vowel between long and short, presumably only length.

I might be missing something? I focus most of my attention purely on Quenya, and the necessary proto-E->Q->Qu development, so Sindarin is more a side issue for me. All errors and oversights are mine, and I would beg forgiveness on their account.

Fiona Jallings Feb 06, 2015 (22:15)

+Jonathon Omahen
 In the latin orthography of Sindarin, there's both accute and circumflex over vowels. for example, "dûn" versus "dúnadan", and before you say it's because "dûn" is a monosyllable, there's also "annûn". So, there must be a difference between Û and Ú. The difference can't be stress, because the U of both "dûn" and "dúnadan" is stressed, and the U of "annûn" isn't. So, I think that the difference is length.

Though not beyond possibility, I haven't seen any examples of phonemic vowel length beyond long and short (hence IPA doesn't provide an easy way to notate, for instance 'middle', vowel length).
There is an IPA annotation for middle-vowel length. It's (ˑ), the top half of (ː). It must be rare, because so few people seem to have heard of it, but it's common enough to be built into the IPA keyboard I use, without even having to use a difficult key combination to get to it, not even a shift key.

Jonathon Omahen Feb 13, 2015 (04:54)

+Fiona Jallings :  I will humbly apologise for my lack of consulting the IPA charts more closely.

There is an IPA annotation for middle-vowel length. It's (ˑ), the top half of (ː).

Indeed, I take this correction gladly! I was also interested to see some languages that use phonemic middle-vowel length. It certainly seemed viable to me as a real distinction, and I appreciate the correction.

In the latin orthography of Sindarin, there's both accute and circumflex over vowels.

I concur entirely that the orthographic conventions established by Tolkien show an apparent distinction of three vowel lengths (a, á and â). However, reading the extant sources carefully indicates to me that his intent was for a heavy stress on long monosyllables, resulting in an extra-long pronunciation. This would indicate, to me, a suprasegmental feature (just as accent generally is [though not always]) that is non-phonemic:

"In Sindarin, long e, a,  o, had the same quality as the short vowels..."

"Long vowels are usually marked with the ‘acute accent’, as in some varieties of Fëanorian script. In Sindarin long vowels in stressed monosyllables are marked with the circumflex, since they tended in such cases to be specially prolonged; so in dûn compared with Dúnadan. The use of the circumflex in other languages such as Adûnaic or Dwarvish has no special significance, and is used merely  to mark these out as alien tongues (as with the use of k)"
(LOTR:AE)

Again, here, plain reading would indicate that Tolkien is suggesting, at the very least, that extra-long vowels are the application of the stress on a long-vowel monosyllable adding to the length of the vowel when produced. That is, he seems to be saying that phonemically, we have:

     /'duːn/ and /'duːnadan/

But that the realised phones would be closer to:

     /'duːːn/ and /'duːnadan/
                      or
     /'duːn/ and /'duˑnadan/

I could be off on this, but we see some consistency reflected in the description of Tengwar as well:

"Long vowels were usually represented by placing the tehta on the ‘long carrier’, of which a common form was like an undotted j. But for the same purpose the tehtar could be doubled. This was, however, only frequently done with the curls, and sometimes with the ‘accent’. Two dots was more  often used as a sign for following y"
(LOTR:AE)

That short (or regular) and long vowels are distinguished is important, for these are phonemic and not merely allophonic features of essentially all Eldarin languages. However, there is no case or exception made for 'extra-long' vowels to be indicated with any markings. This would certainly be in line with the extra length being a predictable feature of the accent.

I will not be foolish enough to suggest the argumentum ad silencio, because a lack of an extra-long vowel carrier or mark is not evidence against phonemic distinction of three vowel lengths. There are many very essential phonemic features of languages (examples including the 'okina of Hawai'ian, vowel length in Hawai'ian prior to the modern period, the pitch-accent of Tokyo-based standard Japanese, or the vowels in Hebrew) that are not usually notated in the orthography. However, given the description by Tolkien above, and browsing through some corpus examples, the explanation of the Tengwar for vowel notation comports with my current understanding.

In addition, I would argue that as a well-informed philologist, Tolkien was well-enough read in historical and comparative linguistics (especially owing to his Old English and Welsh knowledge) to know that the most natural formation of unique phonemes in a child language (such as compensatory vowel lengthening) would derive from a parent feature that has been changed or modified from other phonological processes in the newly forming language. Again, my specialisation is really in Quenya, but I would like to see a good case for a phonemic 'extra-long' vowel made from derivation from PE -> OS -> S. If a plausible case could be made, repeatable and testable, that could be very convincing.

I have tried to find others that look at the vowel lengths as more than long and short and have found some discussion on Elfling and Lambengolmor from 6-10 years ago, but no real consensus on it. If you have other papers or research, I'd be extremely happy to hear of it!

So how then do we deal with perceived outliers?

The crux of your argument is that annûn violates two premises:

    - Stress should fall on the first syllable
    - This is polysyllabic

Based on regular rules of stress and accent, the final long vowel should simply be long, not 'extra-long'. Knowing that language will be inconsistent, and that exceptions may exist (for historical or aesthetic reasons), this one example cannot be enough to call the entire framework into question. If there are examples in the corpus of multisyllabic 'extra-long' and unaccented vowels occur, in any significant number, that is certainly worthy of extra attention. In addition, significant (or at least consistent) demonstration of minimal pairs involving the 'extra-long' vowel would just about shut me up on the matter (other than to pitch in on the understanding, of course).

There are some plausible ways to account for the strangeness of annûn. For instance, it may be that Tolkien is indicating that the stress should be on the final syllable, contra to standard accent patterns. This could be for historical reasons (derivation), aesthetic reasons (it pleased the Sindar ear more) and/or disambiguation to distinguish annûn from another word or words that would otherwise be homophonic.

These are just ideas I'm throwing out, since I am really unfamiliar with the Sindarin corpus compared to the Quenya one. If you have anything you can point me to to better understand where you're coming from, that would be amazing. I am putting my primary stress on Tolkien's description of the language first, then on his extant corpus, and lastly on his tertiary writings during various phases of development. Ideally, we would work solely from the corpus, but due to its scant nature, and that Tolkien most certainly made errors (which I like to think of as happy, natural-looking inconsistencies we see in any language), I tend to exegete his description, then his examples in light of such.

Thank you so much for giving me some thoughtful responses! It's really exciting to think about, and I had to do some additional research in a new area. It's refreshing. :)

Jonathon Omahen Feb 13, 2015 (04:56)

...I realise how off-topic I have taken this! Much apologies, I should have posted as a separate post!

Fiona Jallings Feb 13, 2015 (06:00)

I think that it did start out as an allophone of vowels in stressed monosyllables, but may have extended beyond that to become a straight up phoneme. Tolkien also uses the acute accent when writing some stressed monosyllables ( comes to mind), so I think that at some point it became a separate phoneme, but this 3-length vowel system seems to be only in I, Y, U, and O. I can't think of any Sindarin words with É or Á in them. (which neatly fits into the categories -high and -round.)

Annûn and amrûn may have stress on the final syllable - true. It could have been that it was to show that they came from compounds: an+dûn, am+rhûn. If so, was the stress becoming phonemic in Sindarin, rather than its presumed allophonic state? What about other words with long vowels at their ends? Tolkien was distressingly vague on the topic of pronunciation. I wish he had just described the pronunciation linguistically, so we didn't have to push through this fog.

Jonathon Omahen Feb 19, 2015 (01:11)

+Fiona Jallings: Forgive me, but I don't really see this from the extant evidence. Have you written out this theory more fully anywhere? It is quite interesting at least, and I'd be down to learn more.

Fiona Jallings Feb 27, 2015 (04:19)

+Jonathon Omahen
 This is the most i've written out this idea. I'm just pointing out some more ways that the evidence could be interpreted; the point being that the evidence isn't conclusive.