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

Yet more on voiced stops

Expand Messages
  • Pavel Iosad
    Hello All, ... Which leads us to the problem of a proper phonological analysis of Quenya. To wit: are the nasalised voiced stops /mb nd ñg/ single phonemes,
    Message 1 of 15 , Jun 7, 2002
    • 0 Attachment
      Hello All,

      Answering my question on Quenya intervocalic -d-'s, Ivan wrote:

      > The prime directive is: No /d/ in Quenya
      > except in the combinations /ld nd rd/ (LR:1155).

      Which leads us to the problem of a proper phonological analysis of
      Quenya. To wit: are the 'nasalised' voiced stops /mb nd ñg/ single
      phonemes, or biphonemic sequences?

      I hold it is the latter.

      But evidence is scanty, and I would like to discuss this question with a
      more knowledgeable company. So what would the assembly's opinion be,
      taking into account the following pieces of evidence:

      1) Consistent use of the word _combinations_ in Appendix E when
      referring to Grades 2 and 4 seems to argue that the sequence is
      biphonemic.

      2) These groups were clearly considered "long" for purposes of stress in
      early (and later) Sindarin, as per LR:1089 (it appears I am using a
      different edition than Ivan above). This would imply that in the
      structure of a word like _*periandath_ the _a_ in the penult qualifies
      as a vowel followed by two consonants, ergo /nd/ is biphonemic. This
      meets two objections - even early Sindarin is not Quenya (though I'd
      assume the phonological workings of the two would be exceptionally
      similar), and the second of a more fundamental nature. If we assume /nd/
      is monophonemic, the stress would still fall on the penult in this case,
      since /nd/ (be it mono- or biphonemic) is an impermissible onset, since
      no word in Quenya begins with it. On yet another hand, this latter
      argument could be taken as evidence for the biphonemic status of the
      group in question, as the restriction could then be explained in terms
      of the restriction on initial clusters in Quenya.

      3) There is little reason to distinguish /nd/ from /ld/ and /rd/. The
      latter are clearly biphonemic. It would then seem that a voiced stop is
      in a strong position when clustered with an alveolar sonorant. It is
      unclear whether /b/ shifted to /v/ after /l/ as a matter of some later
      dialect, or of a regular phonological process (since it appears that the
      Elves themselves did use _lb_ (LR:1095)). To clarify: /g/ shifted to a
      voiced /h/ regularly in Ukrainian and southern Russian dialects,
      however, there was no process of a regular voiced stop > homorganic
      voiced fricative shift. The only argument to see nasalisation as
      phonemically relevant attribute is its typological justification.
      Otherwise, we could as well argue that /ld/ is a phoneme while /nd/ and
      /rd/ are biphonemic. This doesn't seem likely at all

      4) However, there are clear cases of metathesis (e.g. in the past tense
      of basic verbs). A biphonemic sequence yielding a single phoneme is not
      at all impossible (cf. the conduct of Slavic *tj and *dj). Why would
      *_tek-ne_ yield *_tencë_? An answer might be positing not a metathesis
      (i.e. not the development of two sounds), but a nasalisation of the last
      consonant of a CVC- root as a phonological process a bit like the Irish
      attenuation and broadening (caolú and leathnú). Such an interpretation
      seems to be an argument for the monophonemic status.

      Overall, I still think the biphonemic interpretation is the better one,
      not the least because it is the less complex one. I am sure there is
      more to it than the outline above.

      Any comments?

      Pavel
      --
      Pavel Iosad pavel_iosad@...

      'I am a philologist, and thus a misunderstood man'
      --JRR Tolkien, _The Notion Club Papers_
    • Candon McLean
      ... They can only be biphonemic. The question is are they coarticulated or not? For example, Ladefoged ignores phonemes like [tS] and [d3] in the IPA, in
      Message 2 of 15 , Jun 7, 2002
      • 0 Attachment
        --- Pavel Iosad <pavel_iosad@...> wrote:

        > Which leads us to the problem of a proper phonological analysis of
        > Quenya. To wit: are the 'nasalised' voiced stops /mb nd �g/ single
        > phonemes, or biphonemic sequences?

        They can only be biphonemic. The question is are they coarticulated
        or not? For example, Ladefoged ignores "phonemes" like [tS] and [d3]
        in the IPA, in fact they aren't in the IPA, because his phonetic work
        has shown that they are two sounds that are coarticulated (see
        Ladefoged _A Course in Phonetics_ 1975. 4th edition). Clusters like
        [mb], found in some African languages etc., are also not in the IPA
        because they are coarticulated.

        I believe by phoneme you mean coarticulate and by biphonemic you mean
        two independently articulated segments.

        A crucial test (which may be beyond us) is to see if the cluster
        splits into a coda and an onset:

        e.g. /lambe/ > 1. [lam.be] or 2. [la.mbe].

        If (1) then the cluster has two independently articualted segments;
        if (2) then the cluster is coarticulated and thus a "phoneme."

        An example from English is 'judging:' /d3Ud3 + Ing/ > [d3U.d3Ing],
        where /ng/ = the sound ingma, i.e. the velar nasal. In the English
        example it's clear that /d3/ is coarticulated as the sound does not
        split across syllable boundaries.

        I don't remember if Tolkien has given us discriptions of the
        syllabification of these clusters. If he hasn't then we need to
        listen again to his recordings (but these maybe inaccurate as he
        wasn't a native speaker of Quenya, alas).

        Candon


        __________________________________________________
        Do You Yahoo!?
        Yahoo! - Official partner of 2002 FIFA World Cup
        http://fifaworldcup.yahoo.com
      • pavel_iosad
        Hello, ... Evidence? :-) I do agree, but still... :-) ... As a matter of fact, I was referring to the phoneme as a unit of structural analysis (which And
        Message 3 of 15 , Jun 7, 2002
        • 0 Attachment
          Hello,
          Candon wrote:
          > > Which leads us to the problem of a proper phonological analysis
          > > of Quenya. To wit: are the 'nasalised' voiced stops /mb nd ñg/
          > > single phonemes, or biphonemic sequences?
          >
          > They can only be biphonemic.

          Evidence? :-) I do agree, but still... :-)

          > I believe by phoneme you mean coarticulate and by biphonemic you
          > mean two independently articulated segments.

          As a matter of fact, I was referring to the phoneme as a unit of
          structural analysis (which And rejects, just don't beat me now :-))
          and not as a segment in the speech.

          > A crucial test (which may be beyond us) is to see if the cluster
          > splits into a coda and an onset:
          >
          > e.g. /lambe/ > 1. [lam.be] or 2. [la.mbe].
          >
          > If (1) then the cluster has two independently articualted segments;
          > if (2) then the cluster is coarticulated and thus a "phoneme."

          I believe coarticulation in Quenya at least isn't the primary test,
          as demonstrated by the fact that the Quenya _qu_, which was
          pronounced as a cluster (in the Third Age at least), was still
          permitted word-initially, demonstrating it was not a cluster
          phonologically.

          Besides, the syllabification test allows both interpretations, as
          the whole of the /nd/ group obviously is included in the prceding
          syllable, as per the Maximum Onset Principle. Quenya words do not
          start in /d/, ergo /d/ is an impermissible onset. Ditto with /nd/,

          Pavel
          --
          Pavel Iosad pavel_iosad@...

          'I am a philologist, and thus a misunderstood man'
          --JRR Tolkien, _The Notion Club Papers_
        • Candon McLean
          Hi, ... The evidence is in acoustic phonetics which shows overlapping wave-forms for sounds like [tS], i.e. [t] peaks and before its wave has ended [S] begins.
          Message 4 of 15 , Jun 8, 2002
          • 0 Attachment
            Hi,

            --- pavel_iosad <pavel_iosad@...> wrote:

            >>> Which leads us to the problem of a proper phonological analysis
            >>> of Quenya. To wit: are the 'nasalised' voiced stops /mb nd �g/
            >>> single phonemes, or biphonemic sequences?

            Candon wrote:
            >> They can only be biphonemic.

            Pavel wrote:
            > Evidence? :-) I do agree, but still... :-)

            The evidence is in acoustic phonetics which shows overlapping
            wave-forms for sounds like [tS], i.e. [t] peaks and before its wave
            has ended [S] begins.

            I believe this work was started in the 50's and 60's at Edinburgh,
            which perhaps means Tolkien was aware of it.

            Candon wrote:
            >> I believe by phoneme you mean coarticulated and by biphonemic you
            >> mean two independently articulated segments.

            Pavel wrote:
            > As a matter of fact, I was referring to the phoneme as a unit of
            > structural analysis (which And rejects, just don't beat me now :-))
            > and not as a segment in the speech.

            The acoustic analysis is clear that this isn't so, but for
            convenience and a shorthand lable, perhaps we could call
            coarticulated segments a phoneme (as long as we remember that they
            are not).

            Candon wrote:
            >> A crucial test (which may be beyond us) is to see if the cluster
            >> splits into a coda and an onset:
            >> e.g. /lambe/ > 1. [lam.be] or 2. [la.mbe].
            >> If (1) then the cluster has two independently articualted segments;
            >> if (2) then the cluster is coarticulated and thus a "phoneme."

            Pavel wrote:
            > I believe coarticulation in Quenya at least isn't the primary test,
            > as demonstrated by the fact that the Quenya _qu_, which was
            > pronounced as a cluster (in the Third Age at least), was still
            > permitted word-initially, demonstrating it was not a cluster
            > phonologically.

            I'm not sure I follow this. In English we have word initial /kw/ and
            it is biphonemic. Quenya <qu> seems to be of the same sort.

            Pavel wrote:
            > Besides, the syllabification test allows both interpretations, as
            > the whole of the /nd/ group obviously is included in the prceding
            > syllable, as per the Maximum Onset Principle. Quenya words do not
            > start in /d/, ergo /d/ is an impermissible onset. Ditto with /nd/,

            I'm not sure it's obvious. It's true that Quenya only allows
            palatalized or labialized consonant clusters word initially, but what
            happens word internally isn't clear.

            So, your claim is because /d/ is never an onset /nd/ will not split
            word internally. I don't believe this _has_ to be true. But let's
            say it is. What about the cluster /nt/? /t/ is allowed in onset
            position. So the syllabification of a word like _tintalle_ > 1.
            [tin.tal.le], or 2. [tint.al.le] will help us decide if clusters are
            coarticulated or not.

            Perhaps we can make an argument based on Tolkien's asthetic tastes to
            help us decide. It's clear that Tolkien was interested in creating a
            euphonic language. Which is the more euphonic syllabification of
            Quenya _sinda_? 1. [sin.da] or 2. [sind.a] It seems clear to me
            that (1) is more euphonic, and it is easier to articulate (the same
            can be said of _tintalle_(1)above. Ease of articulation also seems
            to have been important to Tolkien (cf. /n/ + /s/ > [ss] (e.g.
            _Elessar_).

            Notice also that when segments assimilate (for ease of
            articulation) they don't disappear. This would indicate that both
            segments [ss] in _Elessar_ are pronounced (as compared with
            _*elesar_.

            If both segments are indeed pronounced, this in turn seems to
            indicate that the cluster is _not_ coarticulated as the best way to
            make [ss] salient (i.e. perceivable) is to split the cluster [s.s]


            If we don't have Tolkien's ideas on syllabification (and I haven't
            had time to look into it), then his desire for euphony and ease of
            articulation perhaps can shed some light on whether quenya clusters
            are coarticulated, i.e. "phonemic," or not.

            Candon

            __________________________________________________
            Do You Yahoo!?
            Yahoo! - Official partner of 2002 FIFA World Cup
            http://fifaworldcup.yahoo.com
          • pavel_iosad
            Hello, I m still sceptical about acoustic evidence, nevertheless. Candon wrote: [...] ... But I was trying to find out exactly whether they are or are not! :-)
            Message 5 of 15 , Jun 8, 2002
            • 0 Attachment
              Hello,

              I'm still sceptical about acoustic evidence, nevertheless.

              Candon wrote:
              [...]
              > The acoustic analysis is clear that this isn't so, but for
              > convenience and a shorthand lable, perhaps we could call
              > coarticulated segments a phoneme (as long as we remember that they
              > are not).

              But I was trying to find out exactly whether they are or are not! :-)


              > > I believe coarticulation in Quenya at least isn't the primary
              > > test, as demonstrated by the fact that the Quenya _qu_, which
              > > was pronounced as a cluster (in the Third Age at least), was
              > > still permitted word-initially, demonstrating it was not a
              > > cluster phonologically.
              >
              > I'm not sure I follow this. In English we have word initial /kw/
              > and it is biphonemic. Quenya <qu> seems to be of the same sort.

              Quenya doesn't allow initial clusters at all. Thus, _qu_, which is
              permissible initially, is _not_ a cluster

              > I'm not sure it's obvious. It's true that Quenya only allows
              > palatalized or labialized consonant clusters word initially, but
              > what happens word internally isn't clear.

              I'd say that the palatalized and labilaized sounds are precisely
              monophonemic.

              > So, your claim is because /d/ is never an onset /nd/ will not split
              > word internally. I don't believe this _has_ to be true. But let's
              > say it is. What about the cluster /nt/? /t/ is allowed in onset
              > position.

              Good point, but it is obvious that the unvoiced stops have much
              fewer phonotactical restrcitions imposed on them than the voiced
              ones.

              On the ohter hand, this example amply demonstrates that /mp nt ng/
              are biphonemic sequences. This would mean that plosives
              (phonemically) present a rather strange system /p/ ~ /b/ ~ /mb/.
              Such a system is highly untypological. The only structurally
              analogous situation I can think is the traditional PIE
              reconstruction (substitute aspiration ofr nasalisation). But that
              may precisely have been the inspiration! It would be "very
              Tolkien" :-)

              [...]
              > If we don't have Tolkien's ideas on syllabification (and I haven't
              > had time to look into it), then his desire for euphony and ease of
              > articulation perhaps can shed some light on whether quenya clusters
              > are coarticulated, i.e. "phonemic," or not.

              Still, I do not see any direct correlation between coarticulation
              and monopohnemic status.

              But perhaps the better-learned ones here will clarify it for me...:-)

              Pavel
              --
              Pavel Iosad pavel_iosad@...

              'I am a philologist, and thus a misunderstood man'
              --JRR Tolkien, _The Notion Club Papers_
            • Ivan A Derzhanski
              ... [...] ... [...] ... Acoustic phonetics is not about phonemes at all. Whatever sequence of wave forms the language treats as a phoneme is by virtue of that
              Message 6 of 15 , Jun 8, 2002
              • 0 Attachment
                Candon McLean wrote:
                > --- pavel_iosad <pavel_iosad@...> wrote:
                > >>> are the 'nasalised' voiced stops /mb nd сg/
                > >>> single phonemes, or biphonemic sequences?
                [...]
                > >> They can only be biphonemic.
                [...]
                > The evidence is in acoustic phonetics which shows overlapping
                > wave-forms for sounds like [tS], i.e. [t] peaks and before
                > its wave has ended [S] begins.

                Acoustic phonetics is not about phonemes at all. Whatever sequence
                of wave forms the language treats as a phoneme is by virtue of that
                fact a phoneme.

                > > I believe coarticulation in Quenya at least isn't the primary test,
                > > as demonstrated by the fact that the Quenya _qu_, which was
                > > pronounced as a cluster [...], was still permitted word-initially,
                > > demonstrating it was not a cluster phonologically.
                >
                > I'm not sure I follow this. In English we have word initial /kw/
                > and it is biphonemic. Quenya <qu> seems to be of the same sort.

                English allows word-initial (and generally syllable-initial)
                clusters. Quenya doesn't. So the evidence of English isn't
                automatically relevant to Quenya.

                > Pavel wrote:
                > > Besides, the syllabification test allows both interpretations, as
                > > the whole of the /nd/ group obviously is included in the prceding
                > > syllable, as per the Maximum Onset Principle. Quenya words do not
                > > start in /d/, ergo /d/ is an impermissible onset. Ditto with /nd/,
                [...]
                > So, your claim is because /d/ is never an onset /nd/ will not split
                > word internally. I don't believe this _has_ to be true.

                As a matter of fact, it does not. Think of Finnish medial /ht/.
                It has to split as /h/+/t/, because a cluster can be neither an
                onset nor a coda, but we have to live with the fact that /h/ can
                be a coda of a non-final syllable (though not a final one).

                --Ivan
              • fr3dr1k_s
                ... Few sounds in speech are *not* coarticulated in that sense. For example, by anticipatory coarticulation, /k/ has lip rounding in the word coo . The
                Message 7 of 15 , Jun 8, 2002
                • 0 Attachment
                  Candon McLean wrote:

                  >>> They can only be biphonemic.
                  >> Evidence? :-) I do agree, but still... :-)
                  >The evidence is in acoustic phonetics which shows
                  >overlapping wave-forms for sounds like [tS], i.e. [t] peaks and
                  >before its wave has ended [S] begins.

                  Few sounds in speech are *not* coarticulated in that sense. For
                  example, by anticipatory coarticulation, /k/ has lip rounding in the
                  word "coo". The labialized feature of the vowel is anticipated in
                  the realization of the velar stop, [k^w]. That would be an example
                  of coarticulation. But "biphonemic" of course refers to a
                  sequence of two phonemes. These phonemes may or may not
                  be further analysed into sequences of sounds on the phonetic
                  level, but that is irrelevant here. It is important to remember that
                  phonemes, while the smallest units of speech *phonologically*
                  speaking, are not necessarily "atomic" *phonetically* speaking
                  but may be broken down into smaller segments of sound.
                  Affricates are sequences of homorganic sounds on the phonetic
                  level that make up single units on the phonological level: they
                  are phonemes (no scare quotes). In his _Course in Phonetics_
                  earlier referred to, Ladefoged points out that "From the point of
                  view of a phonologist considering the sound patterns of English,
                  the palato-alveolar affricates are plainly single units" (3rd ed.,
                  63). I don't have the 4th ed. though.

                  Sorry if I missed your point and just reiterated the obvious.

                  /Fredrik Ström
                • Candon McLean
                  Ivan and Fredrik both wrote that phonetic analysis isn t relevant to phonemes (or something similar to that effect). I agree. My mistake. Indeed sounds like
                  Message 8 of 15 , Jun 8, 2002
                  • 0 Attachment
                    Ivan and Fredrik both wrote that phonetic analysis isn't relevant to
                    phonemes (or something similar to that effect).

                    I agree. My mistake. Indeed sounds like [tS] are phonemes.

                    The point I was trying to make is that these kind of complex phonemes
                    with coarticulated sounds can't be split, and so if we wanted to test
                    whether Quenya clusters are phonemic or not, we should be able to do
                    so by focusing on the coarticulated properties of these sounds (like
                    affricates, etc).

                    Candon


                    __________________________________________________
                    Do You Yahoo!?
                    Yahoo! - Official partner of 2002 FIFA World Cup
                    http://fifaworldcup.yahoo.com
                  • anthonyappleyard
                    Apart from _Aldudenie_, what known instances of Quenya _d_ between vowels are known? If _Aldudenie_ is the only example, perhaps it is a stray mistake by
                    Message 9 of 15 , Jun 11, 2002
                    • 0 Attachment
                      Apart from _Aldudenie_, what known instances of Quenya _d_ between
                      vowels are known?

                      If _Aldudenie_ is the only example, perhaps it is a stray mistake by
                      Tolkien and if he had lived longer he would have found and corrected
                      it.

                      [It has been suggested many times before that _Aldudenie_, composed
                      by a Vanyarin elf, is a Vanyarin, not Quenya, title. Carl]
                    • Eleder
                      ... The best discussion I can remember about the possibility that *_-dénie_ was the Vanyarin cognate of Noldorin _nainie_, lament , is the #5885 message of
                      Message 10 of 15 , Jun 12, 2002
                      • 0 Attachment
                        > [It has been suggested many times before that _Aldudenie_, composed
                        > by a Vanyarin elf, is a Vanyarin, not Quenya, title. Carl]

                        The best discussion I can remember about the possibility that *_-dénie_
                        was the Vanyarin cognate of Noldorin _nainie_, "lament", is the #5885
                        message of Elfling, by Ales Bican:

                        http://groups.yahoo.com/group/elfling/message/5885

                        As he said, it's hard to believe that it could be a typo by Tolkien,
                        since the word _Aldudénie_ appears in different manuscripts and
                        texts carefully revised by Tolkien.

                        By the way, I introduce myself in this list, as member of the
                        Lambenor Spanish-speaking mailing-list, and the Team of
                        Languages of the Spanish Tolkien Society.

                        ------
                        Eleder

                        "La fantasía se inocula en tu intelecto cual vacuna contra la sórdida
                        subsistencia, cuando el aguijón de John Ronald Reuel Tolkien se
                        inserta en los patológicos hemisferios cerebrales de todo lector que
                        padezca el acierto de acceder a su terapéutica saga."
                      Your message has been successfully submitted and would be delivered to recipients shortly.