Loading ...
Sorry, an error occurred while loading the content.

320RE: [Lambengolmor] Re: Tyelpetema and phonetics vs. phonology in Quenya

Expand Messages
  • Pavel Iosad
    Feb 18, 2003

      Ales Bican has raised several most interesting points in response to my
      proposals regarding the various domains of phonetics and phonology in
      Quenya with regard to the status of the tyelpetéma, for which I am most
      grateful. Now, on to the specifics.

      >**First of all, I think there is no published writing of Tolkien's
      >where he would state Quenya does not permit, to simplify it, long
      >vowels before consonant clusters. Or course, I may be wrong, so
      >correct me.

      Indeed you are correct. Nevertheless the lack of vowel lengthening in
      the perfect of _lelya-_, which exhibits nasal infixion, or in the
      present tense of verbs like _lanta_- shows that this shortening in
      closed syllables is regularly forced by the phonotactics. It would
      nicely be described by a generative-style phonology, where the
      morphological module would give a form like _lántar_, and the 'phonology
      proper' module would then give the actual output _lantar_).

      [snip examples of bimoraic segments before clusters]

      > Diphthongs, at least for purposes of the stress, are
      > treated like long vowels and are bimoraic.

      True. See below, however.

      > [_aista-_ 'to dread' (V:358) -- PHW]

      Also _aistana_ 'blessed' in AM, and _aista-_ in _Alcar i Ataren_.

      This may have to do with etymology, since *_aistana_, the editors inform
      us, can be connected with an old base like GAYA(S)- (VT43:38), which
      gives a formation like _*(g)aiastanâ_. Now the syncope of the second _a_
      is a bit hard to explain, it being in a long syllable. Or are we dealing
      with a special status of the _st_ group (it would also be possible to
      write the long vowel of _Hrísto_ off to the suggestion)? I am however at
      a loss at the moment as to how to explain it.

      >It must also be realized that what is written as Cy does not have to
      >stand for a palatal(ized) sound. Surely _my_ in _lamya-_ "to sound"
      >(Etym s.v. LAM) cannot stand for a palatal _m_, as no palatal _m_
      >exists (as far as I know it is impossible).

      Unless it were a coarticulated palato-labial nasal plosive, which is
      nowhere phonemic, though imaginable. (Well, there are no phonemic
      palato-whatever coarticulated stops, though it is not inconceivable that
      a _kp_ shift to a palato-labial before, say, front vowels. Just
      guessing, you might figure out)

      >It might be a palatalized _m_ (which is not unusual).

      We the Slavic-speaking ought to know ;-)

      > The question is whether the _y_ in _lamya_
      > stands for [j] or whether _my_ is a digraph for [m'] (palatalized _m_,
      > I will use the apostophe for palatalization).

      As I suggested, it might as well stand for a [t_jj], or, in your
      notation, [t'j], with the [j] approximant rather than the more of a
      fricative than it is in, say, Russian.

      >And the same can be applied to other Cy combinations:

      This is true, and in fact, can be used as an argument for my
      interpretation. If we consider the _my_ palatalized (which we apparently
      agree upon), it does mean that palatalization is a phonemically relevant
      feature. Offhand, I think that at least in Europe, languages which have
      a palatal series but do not consider palatalization phonemically
      relevant prevail over those which do both. Off the top of my head, the
      only instance of the second-type language is Macedonian. On the other
      hand, it has, as far as I remember, all but eliminated the old Slavic
      palatalization opposition. I am eager to be corrected though.

      > it is hard to say whether _ry_ in
      >_máryat_ stands for [R] (palatal _r_, I will use capitals for palatals)
      > or [r'] or [rj] (resp. [r'j]).

      In the official IPA table (cf., for instance,
      http://www2.arts.gla.ac.uk/IPA/fullchart.html; this one is quite in
      accord with the latest version of the IPA handbook I have access to),
      palatal taps/flaps or trills are not shaded (which means the
      articulation is considered possible), but do not have a symbol assigned,
      which means that no described languages has a phonemic sound of those
      types. Thus Quenya would be entirely untypological here, were there a
      palatal [r] (I am a bit uncomfortable with [R], as it stands for the
      uvular fricative in X-SAMPA)

      >I wonder how is this dealt with in the tengwar[...]

      So do I.

      >With this are connected what I would call primary and secondary Cy
      >combinations. Primary Cy combinations are those that existed from
      >Primitive Quendian (resp. Common Eldarin). They occur exclusively

      That isn't as obvious. Apparently you mean that certain word-initial
      consonants were palatalized, but then your use of 'combinations' is
      somewhat misleading. Combinations of C+_j_ were clearly present on the
      CE level, since medial combinations of this kind consistently turn out
      as tyelpetéma-consonants in Quenya, but also cause i-affection in

      > What we do not
      > know is whether e.g. _n + y_ in _vanya_ produced [N] or [n'j]/[nj].

      Indeed. That is what the whole problem hinges on, anyway.

      > The same holds for _máryat_ and _hiruvalye_. There are some
      > indications that (some) secondary Cy combinations stand for two
      > I think I mentioned it in earlier posts: (a) in VT42:27 Tolkien
      > mentioned that "_atatya_ remained [unreduced] because the second _a_
      > was not syncopated, being in a long syllable";

      Indeed. This is a very instructive example.

      Now the gist of my explanation is the following: sound-changes can be
      broadly divided into two classes, viz., phonetically driven and
      phonologically driven, and we should clearly distinguish the two fields
      when dealing with this part of the Eldarin language system.

      Syncope, for instance, is driven phonetically. From a phonological (here
      phonotactical) point of view, there is nothing inherently wrong with
      three consecutive syllables sharing a similar nucleus, phonetics
      however, aiming at easing of articulation, is the driving force behind
      the simplification of the 'redundant' elements. Since this is a phonetic
      rather than a phonological phenomenon, we should consider the _ty_ as a
      phonetic unit. According to my suggestion, _ty_ is *phonetically*
      bimoraic, and therefore the second syllable of _atatya_ is indeed

      (As a rather important aside, I have quite forgotten to explain how, in
      my theory, the syllable boundary splits the [t_j] and the [j] if they
      are permissible word-initially. It is possible to suggest that
      bisegmental sequences are forbidden on both the phonetical and the
      phonological levels, so word-initially the _ty_'s and sundry could be
      pronounced without the glide owing to these constraints, but with it
      intervocalically. In post-pausal position, as after the nasals, it would
      of course be also pronounced in a single segment, but this does not
      create a lot of problems with regard to syllable division, since the
      preceding syllable would be closed anyway)

      [snip _óne_/_óle_ stuff]
      > However, there is another fundamental thing to remember: published
      > sources are from different stages of the development of Quenya and
      > they do not have to compatible.


      >Another thing must also be mentioned. Quenya does not like sequences
      >of consonants much and if there is a sequence, it does not consist
      >of more than two members. In other words, we do not see combinations
      >of three and more consonants in Quenya. Nevertheless, we see
      >combinations CCy: _nty_, _ndy_, _rty_ (_lty_ not attested), _sty_,
      >and _hty_.

      Which further suggests that they are phonologically single, as Ales
      notes. One should however note the Ñoldorin shift _Quendya_ > _Quenya_
      (XI:361), which suggests the instability of the aforementioned
      combinations (this instability, coupled with the general abhorrence of
      voiced stops, contributes to the total loss of the [d_j]), and thus
      their somewhat indefinite status.

      > The theory about morphemic boundary from _Introduction to Elvish_ was
      > snipped, Pavel rejects it himself.

      It doesn't stand up to evidence anyway, since _hir+uva+lye_ is stressed

      > _lyenna_ - the only example of
      > initial _ly_ - is somewhat doubtful,

      > Anyway, _lyenna_ is clearly a grammatical word. [...]
      > By this I mean that ['grammatical words']
      > bear[] certain grammatical marks (it is therefore marked) and such
      > forms may behave otherwise than other, unmarked, forms. An example
      > of this may be _ciryant_ which appears in the Plotz Letter. In Letters
      > no. 347 Tolkien stated that Quenya did not tolerate final consonants
      > other than dentals; he repeated the same idea in VT42:7, mentioning
      > _t, n, l, r_ (he forgot _s_). Hence _ciryant_ contradicts this
      > statement, but as it is a grammatical (inflected) word, it is
      > allowed to exist.

      Indeed. It can be suggested that the instability referred to above
      contributed towards the gradual elimination of the strict phonotactical
      system. The presence of sequences usually realised bisegmentally in
      initial position and the suppression of the bisegmentality can be a
      driving force behind this weakening of structure. The weakening leads to
      possible overriding of phonotactics by the constraints of grammar (resp.
      the _ciryant_ case) or phonetics (which is what we are discussing).

      After all, it must be noted that the mere fact of the phonologically
      irregular stress in _hiruvalye_ points that the form is marked,
      otherwise its formation would be blocked by a rule higher on the
      hierarchy (i. e. more marked)

      >**In other words, an initial _d_ is not found (though _Aldudénie_ might
      >be an example of this under certain assumptions), which does not
      >necessarily mean it is forbidden.

      But cf. L:1094-5, where the phrase definitely says that the /b g gw/
      were only met in conjunction with the nasals, and /d/, from the turn of
      the phrase, is only met after /n l r/. (There's also Christopher's
      apparent slip in UT, where he gives the name of a Númenórean city as
      _Almaida_, it should apparently be #_Almalda_)

      > One would be inclined to say that
      > a medial _ky_ is not permitted if _Erukyerme_ (UT) and _Ekyanáro_
      > (VT41:14) were not attested.

      By the way, if you asked me, I'd give these as probable candidates for
      palatal rather than palatalized stops; the articulatory mecanism is the
      same as behind the palatal status of _hy_, for which see my original
      post. If we suggest that _ty_ is palatal, then these are just
      alternative orthographies, if it is palatalized, Quenya falls into the
      Macedonian category. A third possibility is that _ky_ is an alternative
      orthography for palatalized _ty_, but this is unlikely.

      >> In the above discussion I have carefully avoided using the words
      >> 'palatal' or 'palatalized' with reference to _tyelpetéma_ consonants,
      >> since we should first determine whether they are the former or the
      >> latter. I suggest they are palatalized.

      >**I think members of the tyelpetéma are palatal consonants.
      >Tolkien himself said that Quenya had a palatal series (tyelpetéma,
      >LotR, Ap. E).

      The distinction between 'palatal' and 'palatalized' can be blurry,
      especially for one who isn't deep into synchronic phonology. Tolkien
      surely knew about it, but it wasn't his primary area of interest (which
      is probably why we seldom get any coherent synchronic picture of
      Tolkien's invented languages - it doesn't possess an independent value
      for Tolkien, and is justified by writing poetry, cf. the passage in 'The
      Secret Vice'). If _ty_ were palatal, it is difficult to justify why
      Tolkien described it as 'similar to English _t_ in _tune_' (L:1088)',
      since the sound is not palatal (but rather, in British speech,
      pronounced with a glide!). Also he might have included the Spanish ñ as
      a description of NY in the Appendix (though this is of no value as an

      [Snip another example of probable confusion between 'palatal' and

      >> One obvious reason is that we would then have problems with
      >> _r_. A *palatal* _r_ (as opposed to palatalized) is simply
      >> in the world's languages, and even if it is, the nearest _I_ can come
      >> a palatal trill or flap is a retroflex approximant [...]

      > **That is an importand fact, but actually it may mean nothing. The
      > that a palatal _r_ seems to be nonexistent does not mean that it does
      > not exist in Quenya. Although I am not familiar with any language
      > possessing a palatal _r_, I read it occurred in a language called
      > Malayalam (I read this in _Trends in Phonological Theory_ by Eli
      > Fischer-Jorgensen p. 65).

      See above, on the IPA table.

      > It is also supposed to have existed in old
      > Czech (as a reflex of _rj_), but this does not mean anything.

      That is rather off-topic, but that is somewhat strained to me. As I
      understand, the transition from palatalized [r] to the Czech r-haczek
      (coarticulated [r] and [Z] as in 'pleasure') only implies the addition
      of the fricative character and removal of the flap, and does not
      necessarily imply a back-and-forth place of articulation shift.


      > _hy_ is somewhat a problem, since it apparently _is_ palatal.

      >I cannot say whether there is a difference between
      >_thy_ and _sy_ (whether the distinction strident/not-strident might
      >exist even among palatals).

      I don't think so, as the strident/non-strident distinction is very
      difficult to handle by the dorsum. The IPA also confirms that the
      distinction, even if existent, is apparently nowhere phonemic.

      > (By the way, if combinations Cy stand for palatals, then there
      > must have existed /S/ (i.e. a palatal _s_), realized as [Z] between
      > vowels -- this was the source of _ry_ [R].)

      Let me note that the presence of intervocalic voicing of _s_ (with or
      without rhotacism) does not necessarily imply voicing of other
      fricatives. True, the Germanic languages voice both the _s_ and the /f þ
      x/ series, but Latin had rhotacism but not voicing of other fricatives.
      This is not a very good example, since Latin does not possess
      intervocalic fricatives word-internally (the initial ones stem from the
      IE voiced aspirates, which were realized word-medially in Italic as
      voiced unaspirated stops), and thus the morpheme boundary (as in
      _de+fici+o_ or _de+fend+o_) can contribute to the lack of voicing. I am
      sure other examples can be found though.

      >> My suggestion is the following. The stress in _hiruvalye_ is to be
      >> explained by the fact that Quenya stress, not being phonological, is
      >> determined by purely phonetic environs, unlike the (phonemically
      >> relevant) length.
      >> I would suggest that _ly_, being palatalized,

      >**I think it is a palatal, being distinct to a palatalized _l_, which
      >occured between _e, i_ and a consonant, cf. App. E s.v. L: "[_l_] was,
      >however, to some degree 'palatalized' between _e, i_ and a consonant,
      >or finally after _e, i_".

      The part about 'to some degree' worries me a lot. I don't know much
      about degrees of palatalization in European languages (since the Russian
      non-palatalized [l] is heavily velarized, which interferes a lot), but
      it seems possible that the palatalization distinction could possess
      three grades.

      >> Now with regard to length, there is a different situation. Stress,
      >> phonologically irrelevant in Quenya, can be determined by the purely
      >> phonetic environs. Length, being phonologically important, should be
      >> judged on the phonological level. Now if _ry_ is a single consonant
      >> apparently is *phonologically*),

      >**I do not think it is obvious.

      I never said it was obvious. This situation is difficult to resolve
      (just as in natural languages), since not one of the two tests
      applicable to Quenya (whether the consonant makes a syllable closed;
      whether it can be word/syllable-initial) is not applicable to the sound
      in question - the first test is what we are discussing, the second one
      is inapplicable in case of word-initial [ry] because there's just no
      possible etymology for an initial [ry] (unless a RY-root is found), in
      case of internal syllable-initial because the first element of a cluster
      is either a nasal or a spirant fricative, both of which tend to
      assimilate with [r] and blur the distinctions.

      >> then the syllable division in _máryat_
      >> is _má+ryat_. There is the question of unattested syllable- (=word-)
      >> initial _ry_, but the tendency for open syllables must be much
      >> than the tendency for maximum onsets.
      >**Morphological criterion might also have played its role, because
      >there is a morpheme boundary between _má_ and _rya(t)_. The same
      >morpheme boundary is, however, between _ma_ and _nna(r)_ in
      >_mannar_ (FS, LR:72). Here the geminated (or long) _n_ may also
      >be phonologically a single consonant, though phonetically a succession
      >of two identical consonants. For that matter there would be the same
      >syllable division.

      The forcing of the vowel shortening here definitely shows that the
      syllable division is _man+nar_. Had Quenya not possessed single
      intervocalic nasals, the situation would be just like the Archi system
      described in my original post, but it just doesn't.

      >> Thus, my suggestion is the following: a CV[Cy]V sequence is
      >> (for purposes of stress) divided into syllables as CV[C+y]V and
      >> phonemically (relevant to length) as CV+[Cy]V.
      >**While this may be possible, I think there are many uncertainties.

      Indeed there are, that's just why I posted this :-)

      Pavel Iosad pavel_iosad@...

      Is mall a mharcaicheas am fear a bheachdaicheas
      --Scottish proverb
    • Show all 22 messages in this topic