Wednesday 15 June 2011

glomerulus

What’s a glomerulus? I imagine most of us would have to confess that this word is not part of our vocabulary, unless we have a particular interest in disorders of the kidney.

According to the Concise Oxford Dictionary, a glomerulus is ‘a cluster of nerve endings, spores, or small blood vessels, esp. of capillaries at the end of a kidney tubule’. Wikipedia explains it more succinctly as ‘a capillary tuft that performs the first step in filtering blood to form urine’.

More relevantly for today’s blog, how do we pronounce this word?

It’s not in LPD or CPD. But COD tells us that it is ɡlɒˈmɛr(j)ʊləs. ODP, from the same stable, agrees, while using its special weak-vowel symbol ᵿ (U+1D7F, LATIN SMALL LETTER UPSILON WITH STROKE), which your browser may or may not be able to display.

Why the bracketed (j)? Why is the palatal element optional or variable?

There’s something awkward about the sequence rj. We long ago dropped the yod that presumably once existed in words such as rude (rɪud → rjuːd → ruːd), brew, true and so on, where the vowel is strong. The awkwardness arises in words where the vowel is weak. The yod tends to be preserved in weak syllables. The two best-known words that exemplify it are virulent and garrulous.

In the case of virulent, both LPD and EPD prefer the yodless ˈvɪrʊlənt for BrE, while also listing variants with -jʊ-, -jə- or -ə- between the liquids. For garrulous they both prioritize ˈɡærələs, but aɡain also admit variants with a yod.

Introspecting, I feel that in slow careful deliberate speech I want to include a j in these two words, and hence by extension in glomerulus. But in faster or less studied speech it’s too much of an effort to articulate the rj sequence, so I drop the yod.

Strangely, this does not apply if the -rj- derives from the compression of ri-, as in disyllabic glorious, merrier, variant. (See this blog for 16-17 January 2007.) In cases like these there is no tendency whatever to omit the palatal element.

21 comments:

  1. This is to do with syllabification. I'd break glomerulus into syllables as ɡlə.mɛr.jʊ.ləs, which puts the r and the j in separate syllables.

    But then I've got a rhotic accent so that's fine for me. For non-rhotics, including the yod would surely force the r to be dropped, wouldn't it? That would actually give ɡlə'məːjələs, not *ɡlə'mɛrjələs.

    So the weak vowel on its own is not enough to permit yod-retention; you also need a preceding syllable to accept the r. That's why initial rju- (or rjə-) with a weak vowel - admittedly very rare - also forces yod-dropping: rutacea rə'teɪʃə, Rutupiae rə'tjuːpiiː, Rupelian rə'piːliən, rucervus rə'səːvəs.

    ReplyDelete
  2. Nice idea, Pete. Unfortunately it doesn't square with the facts, since non-rhotics do NOT say ɡləˈmɜːjʊləs, ˈvɜːjʊlənt, ˈɡɑːjələs.

    ReplyDelete
  3. So you have a preconsonantal r in these words? Or does the j change to a weak i?

    ReplyDelete
  4. We have -rj-, or however you choose to represent r plus nonsyllabic i.
    Compare the r plus syllabic l that you can get in nonrhotic barrel.
    The rule deleting r before a consonant evidently ignores the consonants derived through these processes.

    ReplyDelete
  5. Another difference between /j/ and compressed /i/ (I think) is that the latter can't coalesce with a preceding coronal: I can't imagine anybody saying /kə'neɪdʒən/ for Canadian

    ReplyDelete
  6. Yes Pete, we do have a preconsonantal r in these words. If you call j a consonant, which seems reasonable here because indeed it does not change to a weak i.

    It's annoyed me since childhood, when I was already hearing this in such words, and before I even had any idea what phonology was, I felt it was a phonological impossibility. I had nothing as glamorous as even the S.O.D. then, but I looked them up in various lesser Oxford dictionaries, believing even then that Oxford was the still point of the turning world, and they all seemed to agree on respelling them with (-ro᷅o᷄-) for /rʊ/. (Unfortunately the oldest I now seem to have is a 1963 edition of the COD, but it confirms my recollection of that respelling.) I have been collecting examples ever since, and in ever greater increments they have appeared in all sorts of dictionaries with /rjʊ/ or equivalent, often to the exclusion of /rʊ/. What's going on? It seems to me that AmE has always gone in for rjʊ/rjə a lot more, and I have always thought the same as you, Pete, namely that rhotics can do the necessary syllable division to make it phonotactically possible, though there's still something funny about it, or why isn't it ɡlə.mɝ.jʊ.ləs etc, which would make as much sense for rhotics as ɡlə'məːjələs would for us? But I hardly think this is another case of AmE influence. I think it's more likely to be spelling pronunciation in both AmE and BrE. It's a known thing for spelling pronunciations to fly in the face of phonotactics.

    I have been assuming that the transcriptions in the OED's 1989 2nd ed online version March 2011 only differed from the previous online version in being alphabetized versions of the ridiculous old graphics, but in case there was any evidence of a trend in the direction of spelling pronunciation I checked out the ones brought up so far, and /gləʊˈmɛr(j)ʊləs/ /ˈvɪrjʊlənt/ and /ˈgærələs/ are the same. I may get round to checking some more, as 'querulous' reveals that they are trying to rewrite history:

    Online version March 2011: /ˈkwɛr(j)ᵿləs/ , U.S. /ˈkwɛr(j)ələs/
    Previous online version linked to from current version: ˈkwɛrjʊləs

    But I saved vast numbers of entries from the old version to mht files before it was taken down, and it was in fact kwɛr(j)ᵿləs/, U.S. / kwɛr(j)ələs/ in that too.

    What does it all mean?

    ReplyDelete
  7. That's r᷅o᷄o for /rʊ/, to the best of my ability.

    ReplyDelete
  8. Army, we will just have to eschew not being able to imagine anybody saying /kə'neɪdʒən/. And for 'Acadian' we have the venerable spelling 'Cajun'.

    ReplyDelete
  9. Pete,
    I'm not quite sure JW was talking about the words you were talking about, which I assumed to be the ones he had referred to as not being ɡləˈmɜːjʊləs, ˈvɜːjʊlənt, ˈɡɑːjələs. On closer inspection of his 12:00 post it seems more likely to me he was talking about disyllabic glorious, merrier, variant, as it is in them that "the rule deleting r before a consonant evidently ignores the consonants derived through these processes". If indeed yod tends to be preserved in weak syllables even after r, then its preservation as a consonant in 'glomerulus' et al is a non-process, and if it is not preserved and then reintroduced, that seems to me to be a different process from any he's talking about. Anyway I obviously was talking about 'glomerulus' et al, and I hope I succeeded in answering your question to some extent, tho I could go on for ever if given half a chance.

    ReplyDelete
  10. John,
    «COD tells us that it is ɡlɒˈmɛr(j)ʊləs. ODP, from the same stable, agrees, while using its special weak-vowel symbol ᵿ»

    You don't object to that transcription, but the stress mark indicates syllabification with an unchecked ɒ. We are seeing things like that a lot more, it seems to me, and they too look suspiciously like spelling pronunciations which fly in the face of phonotactics, though not to the extent of ˈvɪrjələnt etc. More like 'tattoo', which you blogged on the December before last, but I guess less puzzling for you, as you have to syllabify 'tattoo' as tæ.ˈtuː which is anomalous whichever way you look at it, whereas I suppose you can deal with 'glomerulus' as ɡlɒm.ˈɛr(j)ʊləs – I'll look the other way while you're syllabifying the rest!

    In OED 'glomerulus' is under 'glomerule': «(ˈglɒməruːl) Also in mod.L. form glomerulus (gləʊˈmɛr(j)ʊləs)» This is more like what you would expect once the o loses the stress, with the open syllable gləʊ, which of course moves on to glə once the door of the Oxford stable is left unbolted.

    ReplyDelete
  11. Oh I see! Well I was asking if non-rhotics have a preconsonantal r in glomerulus, virulent and garrulous.

    I know that glorious and the others have a mandatory r for all speakers, because the following i in these words is definitely a vowel. But it seems to be that rj in the glomerulus-type words constitutes a preconsonantal r, which I understood to be illegal in non-rhotic speech.

    ReplyDelete
  12. Pete,
    Thanks for confirming that I got all that right. And that we agree that the reason 'glorious' and the others have a mandatory r for all speakers is that the following i in these words CAN definitely be a vowel, however phonetically indistinguishable from [j] it may become, whereas the non-rhotics' r in glomerulus, virulent and garrulous precedes a [j] that can definitely NOT be anything syllabic enough to be called a vowel.

    This compulsory [j] in such words makes the r a preconsonantal r, which you understood correctly to be illegal in non-rhotic speech. I hope you sympathize with my annoyance at that illegality.

    John says it's strange that yod-dropping does not apply if the -rj- derives from the compression of ri-, as in disyllabic glorious, merrier, variant, and that in cases like these there is no tendency whatever to omit the palatal element, but it doesn't seem strange to me in the light of our observations.

    ReplyDelete
  13. Three possibly interconnected datapoints: I'm sure my biology teacher pronounced "glomerulus" /ɡlɒmərˈʌləs/, unless it was /ɡlɒməˈrʌləs/. I've also heard Irish people say /kə'neɪdʒən/, perhaps jocularly. I can't remember ever hearing /j/ in "garrulous" or "virulent" in Ireland.

    ReplyDelete
  14. Just a minute, everyone!

    My understanding, which you are all welcome to correct, is that the following concepts exist at the level of the phoneme, not the phone:

    * The syllable
    * The distinction between syllabic and non-syllabic sounds
    * The distinction between semivowels and their vocalic congeners (e.g. i vs. j)

    In which case much of the preceding discussion seems misguided. The distinction between, say, /glɔ.rjəs/ and /glɔ.ri.əs/ makes sense at the phonemic level, but not at the phonetic level.

    ReplyDelete
  15. "The distinction between, say, /glɔ.rjəs/ and /glɔ.ri.əs/ makes sense at the phonemic level, but not at the phonetic level."

    Does that mean that it's not possible to hear the difference between those two? Feel free to ignore this question if it's too elementary.

    ReplyDelete
  16. Army1987: "I can't imagine anybody saying /kə'neɪdʒən/ for Canadian"

    As the token Canadian here, I wouldn't discount /kə'neɪdʒən/ and so Mollymooly above rightly questions that claim. Consider Injun from Indian and Cajun from Acadian. Definitely possible.

    My instinct wants to shy away from the /ɹj/-sequence in glomerulus altogether and use a straight /ɹ/. So I'd tend to say /gläˈmɛrələs/. (Just to reclarify, my Canadian English /ɑ/ appears to be more central than back, yet distinct from /æ/ in man.)

    ReplyDelete
  17. mallamb said...

    In OED 'glomerulus' is under 'glomerule': «(ˈglɒməruːl) Also in mod.L. form glomerulus (gləʊˈmɛr(j)ʊləs)» This is more like what you would expect once the o loses the stress, with the open syllable gləʊ, which of course moves on to glə once the door of the Oxford stable is left unbolted.

    It's perhaps instructive to go back to the original, i.e. NED of 1900 (this volume edited by Bradley). And also to note the additional word glomerular, which one might expect to follow the same pattern as glomerulus.

    Glomerular (glǫme·rŭlăɹ)
    Glomerule (glǫ·mĕrul)
    Glomerulus (glome·rⁱŭlɒ̆s)

    So, only the one which is clearly Latin in form has the optional yod after the r.
    The variation in the vowel of the first syllable is puzzling. Bradley must have felt that the vowel in the word glomerular was directly influenced by the pronunciation of glomerule, in a way that the Latin word (even in English mouths) was not.

    This also highlights a problem with the conversion of the old NED/OED phonetic transcriptions to IPA. The glo- of glomerulus has become gləʊ- in the modern dictionary. Yet the original symbol indicated neither a diphthong nor a long vowel, nor yet a schwa. Presumably, at the time, the standard long o of English, transcribed ō in NED, was in fact pronounced and not ǝʊ. In unaccented positions, it could be reduced to a short monophthong having the quality of the first element of the diphthong: IPA o, in other words. Modern transcriptions do not seem to admit this possibility, so we are left with choosing between ə and əʊ. The OED has gone for the latter, but it leads to some rather affected-looking pronunciations, as here. There is a similar subtlety in the use of u rather than ū in the final syllable of glomerule.

    ReplyDelete
  18. Steve, I quite agree. (In fact, I wanted to have a look around if anybody had noticed the loss of short o and the connection to the change of the first part in the diphthong, otherwise write it up myself.)

    ReplyDelete
  19. Steve:
    Presumably, at the time, the standard long o of English, transcribed ōᵘ in NED, was in fact pronounced oʊ and not ǝʊ. In unaccented positions, it could be reduced to a short monophthong having the quality of the first element of the diphthong: IPA o, in other words. Modern transcriptions do not seem to admit this possibility, so we are left with choosing between ə and əʊ.

    Wikipedia has the symbol ɵ for this (which I hereby christen the "O"bey lexical set).

    ReplyDelete
  20. Here's how the NED treated the other words:

    Garrulance (gæ·rŭlăns)
    Garrulity (gărū·lĭti)
    Garrulous (gæ·rŭləs)

    Virulent (vi·riŭlĕnt) [etc.]

    ReplyDelete
  21. @vp: thanks.

    That's

    U+0275 LATIN SMALL LETTER BARRED O

    for any who need to use it.

    ReplyDelete

Note: only a member of this blog may post a comment.