GAZEBO.

I had always taken gazebo to be, in the OED’s words, “a humorous formation on GAZE v., imitating Lat. futures like videbo ‘I shall see’ (cf. LAVABO),” but the OED goes on to say “but the early quots. suggest that it may possibly be a corruption of some oriental word,” and William Sayers, in “Eastern Prospects: Belvederes, Kiosks, Gazebos,” Neophilologus 87 (2003): 299-305, tries to pin down that “oriental” origin. As the abstract says:

An etymology for gazebo is sought in Hispano-Arabic and a likely candidate meaning ‘mirador, viewing platform’ is found in the work of the medieval Cordoban poet Ibn Guzman. The eighteenth-century British occupation of Tangiers may have provided an avenue for the importation of this lexical isolate, although the architecture of the octagonal garden pavilion now designated gazebo would have had multiple paths to Britain.

You can only read the first page at that link, but Dr. Techie at Wordorigins.org quotes a good chunk of the remainder of the article, which discusses two possible sources, North African and Hispanic Arabic qasbah ‘citadel’ and Ibn Quzman’s qushaybah. The discussion is interesting from both philological and cultural points of view.

Comments

  1. Gazebos fit well in the Hispano-Arabic esthetic of fountains, parks and gardens.

  2. Gazebos fit well in the Hispano-Arabic esthetic of fountains, parks and gardens.

  3. woody mellor says:

    But it just wouldn’t have packed the same punch if the Clash had sung “Rock the Gazebo.”

  4. Charles Perry says:

    I don’t know. It seems a stretch from qsiba (“little reed, little flute, little Casbah”) to gazebo (“turret, lantern or roof on a house”), if for no other reason because the first two consonants of qsiba are unvoiced. Although qaf often becomes gaf in North Africa, the first vowel in CaCiCa words always drops and the now-adjacent voiceless consonant should prevent devoicing. (Qutban, the Moroccan word for shish kebab, is a plural of classical qadib “wand,” which became qdib and then qtib.) I would expect the English to be ixeba or kisseba.
    The German etymology looks more plausible, except for the part about its being wildly unlikely.

  5. Charles,
    I’m sorry, but where does qsiba come from?
    the first two consonants of qsiba are unvoiced
    We’d have to know more about the transmission of the word, but as I’m sure you know, the Spanish letter [z] represents the sounds [s] or [θ]. The word could have been borrowed in its printed form.
    the first vowel in CaCiCa words always drops
    Not true of Hispanic Arabic and some Maghribi dialects (at least according to Handbuch der Arabischen Dialekte).

  6. You really need to read the Knights of the Dinner Table comic about the gazebo sometimes.

  7. q and s unvoiced? Where?
    I would point to Qassem (male name) pronounced more or less “gahsum” and qasr “castle” pronounced more or less kahsur or gahsur. The pronunciation of qaf is as I understand something of a Jordan/Saudi (?) shibboleth with the g or k pronunciation, also westerners can’t really pronounce it.
    I’ve never heard “Casbah”, although I saw it in a novel set in the West Bank (Leon Uris I think–shouldn’t he know?). I say “rock the suq”!

  8. mollymooly says:

    Have to quote footnote #5:

    American English also has the derogatory term gazabo (var. gazebo) meaning ‘fellow, guy, awkward figure’ (Dictionary of American Regional English, first attestation 1889) and Spanish gazapo ‘sly fellow’ seems a likely source. The matter will not be furthered pursued in the present context. Turkish kaz boku ‘worthless shit’ suggests itself but from Turkey to Spain and the American South-west is a long migration.

    My theory is that gazebo derives from Aurangzeb, whose name means “builder of tiny bandstands”.

  9. “Gesehbau” and “Gassebau” seem very un-German. “Gesehen” as an infinitive would be really archaic (Old German, and with other vowels), and “Gassebau” is unlikely because it misses an epenthetic -n-. These are pure speculation by a non German speaker.

  10. Turkish kaz boku ‘worthless shit’ suggests itself but from Turkey to Spain and the American South-west is a long migration.
    That’s very unlikely -i mean, a Turkish origin. “Kaz boku” literally means “Goose shit” in Turkish and it is not a slang term either (it doesn’t mean “worthless shit” -it doesn’t mean anything).

  11. Gazebo may “derive from German” in some way or another, but all this “Gesehbau”, “Gassebau” and “gesehen” (= approximately “see”) is codswallop. Gazebo can not possibly derive from this, because this is not any language at all.
    qbdp is quite right. For “Approximately” “see”, read “exactly” “seen”. What is this “Gesehbau”?? I’ve never encountered the like in 40 years of speaking, writing and reading German, from Luther down to Luhmann, via Nietzsche, Fontane, the Manns, Frisch, comic books, you name it. Even Arno Schmidt would have spurned it. The non-existent “Gassebau” should be “Gassenbau”, as qbdp says – but it would mean “construction of narrow (inner-city) streets (alleyways)”, like “Strassenbau” means “street construction”. It doesn’t mean “street-structure”, whatever that may be. Here, the author seems to be losing his grip on any kind of sense.
    There’s nothing here for historical linguistics, because this is not the German language in any form, at any time or any place.
    By the way, the first paragraph of the extended excerpt from Sayer’s article contains much stately evidence-weighing (not involving German), with learned citations. If you compare it with the (short!) OED entry on gazebo, you may get the impression, as I do, that much of Sayer’s learning is lifted straight out of that short entry. Same goes for the contributor of “Gesehbau”.

  12. Be more open-minded, Grumbly. Who are we to say what is or isn’t German? Let a hundred flowers bloom!

  13. Be more open-minded, Grumbly. Who are we to say what is or isn’t German? Let a hundred flowers bloom!

  14. Charles Perry says:

    bulbul and nijma: Qsiba is the North African equivalent of the Classical diminutive of qasaba, qusaiba (cited in the article from Ibn Quzman).
    Qaf often becomes gaf in Maghribi Arabic. A glance at Richard Harrell’s “A Dictionary of Moroccan Arabic: Arabic-English” shows 4 pages of roots beginning with gaf and 13 with qaf, some words appearing in both sections. Qsiba appears there as a diminutive of qesba/gesba, but gsiba does not.
    As for transmission, it’s hard to imagine the word would have come from written Arabic — or medieval Andalusian Arabic either. Or Spanish (lack of attestation), for that matter. Everybody seems to be speculating about 18th century Maghribi.
    If it came from a dialect that preserves an audible vowel in the first syllable, I still doubt the s would become a z.

  15. Oy, the etymology seems every bit as farraginous as that of gazpacho, discussed last year. Is some concealed connexion too far-fetched? After all: gazebo, gazpacho; and both seem to have Hispano-Arabic roots. Hard to see what is in common semantically, of course.

  16. In fact gazpacho was discussed in 2007. On reflection, both have to do with a building, since γαζοφυλάκιον (implicated in etymology of gazpacho) can be “treasure house”. In the New Testament, so more likely to be known around the place than many Greek words; and just the sort of thing to have followed a tortuous trajectory through Arabic.

  17. Qaf often becomes gaf in Maghribi Arabic.
    What?
    In the Jordanian/Palestinian/Bedouin dialect(s?) I’m used to, there is a qaf ق with a pronunciation that wiggles around a lot and a kaf ﻚ (final form); (initial form)ﻛ that seems to have a more stable pronunciation–but nothing called a gaf.
    Jordanians can’t really understand Moroccan Arabic. But why is the discussion about Arabic in Morocco and not Spain? Where did the conquista Arabs come from? Surely the Morrocan form developed from contact with the Berbers….
    If it came from a dialect that preserves an audible vowel in the first syllable I don’t understand this at all. The Arabs attach definite articles to the front of practically everything. Even the Iraqi refugees can give long discourses about sun letters and moon letters that define which definite article to use.
    The Castillian z is pronounced more like th.

  18. J.E.: So it’s just creative writing, is that what you’re saying? All the judiciousness in the contributions about German really is very misleading, then. You ask, who are we to say what is or isn’t German? Does “we” mean everybody? If so, why does anyone go to the trouble of citing anyone else on a point of German etymology?
    When Nijma writes “Jordanians can’t really understand Moroccan Arabic”, then anyone could have said that with the same weight? I can’t judge what Nijma says, yet occasionally I make a mental note – but definitely not for later use in speculating about Arabic.
    “It’s only German”, appears to be the motto – but that cuts two ways. From English-speaking, literate Westerners, at least those concerning themselves with linguistics, I expect a command of one or more of French, Spanish, Italian, German, Polish, Danish and so on – and consequently a strong awareness of one’s limitations in respect of the others. When people can be so culpably frivolous about a homebase language – and then off-hand about it, when challenged – what can we expect to come out when they start holding forth about Quechua? I invite the reader to imagíne some non-English speaker butting in to tell us innarestin facs about Engrish etymology.
    These be flowers of evil. I say, let a thousand nonsenses about German be nipped in the bud.

  19. gṣəyba would be a perfectly acceptable Maghrebi form – the diminutive is quite productive – but Spanish Arabic normally realised qaf as /q/, not /g/, as attested by borrowings into Spanish such as alcantara

  20. Qaf often becomes gaf in Maghribi Arabic.
    What?
    Seems pretty clear to me. Is it clearer for you if I say /q/ becomes /g/? And al-Andalus was ruled by Arabs and Berbers from Morocco for a long time.

  21. Chill, Grumbl[eb]y. JE was being facetious.

  22. Harrumph!

  23. I’ve been had! Well, serves me right for clomping about with such a chip on my shoulder. I did think that JE was cool, and couldn’t figure why he would write that. It sounded suspiciously like the “exuberant whirling” of another contributor, with whom I would not dare to cross swords on anything substantive.

  24. I’d watch out for Emerson, too. They’re a vicious family. His aunt got hanged for blogging, something like that.

  25. Lewdness.

  26. Lewdness.

  27. And mine for pirouetting on the Sabbath.

  28. They never caught mine–not weighty enough.

  29. In my family it was always barratry.

  30. AJP Crown says:

    That’s it, lewd blogging. It’s a family thing, no pirouetting, thank god. Dante places barrators in the Eighth Circle, fifth pit of Hell. See you there.

  31. Excessive fees for grain storage, actually, but nobody ever dared say anything. They have to outlive us if they want the last say–at the eulogy.

  32. Does usury have a circle?

  33. Who derailed this? Back to the question.
    Is it clearer for you if I say /q/ becomes /g/?
    Yes, if pronunciation is meant. For me the letter qaf ق is already /g/ and not /k/ although a (Jordanian) native speaker would say /q/. (I can’t pronounce it properly, but if I substitute /g/ I can be understood by a native speaker.) The Egyptians use jeem ج to represent /g/. The pronunciation of the hill immediately above Amman’s city center becomes “Gebel Hussein” (Egyptian) instead of “Jebel Hussein” (Jordanian).

  34. Well, the Egyptians use jeem ج to represent /g/ because they never changed the sound to /j/ like most other Arabic speakers. It’s not a sound change like qaf > /g/, as found in various widely scattered dialects.

  35. I think that some people here are inventing fake hanged aunts. Google “wicked house” and you will find my dear Aunt Elizabeth.
    The House of Emerson: wicked since 1686.

  36. I think that some people here are inventing fake hanged aunts. Google “wicked house” and you will find my dear Aunt Elizabeth.
    The House of Emerson: wicked since 1686.

  37. Does usury have a circle?

    I just checked. Usurers are in the same boat with blasphemers and buggers (seventh circle, ring 3). Them was harsh times, folks! Goddamn rent boys overcharging the clients.

  38. Can I play too? I boast a brother who is in the pokey for “frivolous litigation” – what the smart set is calling barratry. Any relation, Hat?
    (Actually for tax evasion, aggravated by self-righteousness. It runs in the family.)

  39. Well, the Egyptians use jeem ج to represent /g/ because they never changed the sound to /j/ like most other Arabic speakers.

    Hmm? I thought when Arabic arrived in Egypt, جیم was already [dʒ]. I don’t think I’ve read it anywhere explicitly, but I had tentatively assumed it came from Coptic, in the way that much of what is distinct about North German phonology when speaking Standard German comes from Low German.

  40. Not to be pedantic, but, hard though it is to believe, there is reason to suspect that lower Egyptian /g/ is a change back towards the original Semitic value, not a retention. The fact that it assimilates to the article, like coronals, and words like wišš “face” g seems to have first taken place in pre-Islamic times among the Bedouins of northern Arabia, but for the most part only spread out from there several centuries after the Islamic era through their various migrations. In North Africa it is generally taken to have been introduced by the Banu Hilal. Since the dialects of northern Morocco are pretty generally /q/-using (see http://books.google.co.uk/books?id=Jg71xrTV-W0C&printsec=frontcover&dq=morocco+heath&lr=&ei=lGduSZPTMpyqMquPkIcE, p. 141), and there wasn’t any major later Bedouin invasion of Spain as there was of North Africa, one wouldn’t expect the /g/ realisation in Spain – but inter-dialectal borrowing usually messes neat statements like that up.

  41. I need to remember to stop using angle brackets in comments here. That first paragraph was meant to continue:
    wišš “from” wajh; see http://books.google.co.uk/books?id=wv2eBP9lPskC&pg=PA160&dq=egyptian+Arabic+g+palatal&lr=&ei=Y2BuSYSOIZi8M-_Cxd0D for some philological arguments.

  42. It’s not a sound change like qaf > /g/, as found in various widely scattered dialects.
    Perhaps the medieval Castillians reached the same solution for pronouncing this difficult sound that I did. I have absolutely no problem pronouncing any sound in Spanish. My pronunciation is close to native, or so my students tell me.
    The /q/ > /g/ shift seems like a common cross-cultural sound change–but not one that goes the other direction.

  43. Not to be pedantic, but, hard though it is to believe, there is reason to suspect that lower Egyptian /g/ is a change back towards the original Semitic value, not a retention.
    Huh, I learn something every day, especially here. And for heaven’s sake, don’t apologize for pedantry; we eat that up here at the Hattery!
    Oh, and for angle brackets use & gt ; (greater than) and & lt ; (less than), closing up the spaces: >, <.

  44. Oh, that’s beautiful, Lameen. For anyone else who’s in my unhappy position of having to use a US proxy to see the books’ contents (odd, normally if these things don’t work from an Irish IP address they won’t work from a UK IP address either), here’s a relevant bit from the first book:

    “By the twelfth century the shift to [ǧ] had been completed in Egypt and by the seventeenth century the ج was pronounced affricative [ǧ]. In the seventeenth century, however, a reversed sound shift had started to take place in urban Egypt.
    Little data are available from the sixteenth and the seventeenth centuries, but data derived from both Muslim and Judeo-Egyptian Arabic reveal that in the latter century the present-day velar stop pronunciation [g] was not yet the norm. In fact, the evidence indicates that in the beginning of the seventeenth century the sound shift from [ǧ] to [g] had only begun.
    Yūsuf al-Maģribī (ms. 1606) cites a special pronunciation ragl in some rural areas for the regular Egyptian rāǧil ‘man’ (p.156). He also mentions the word lagan ‘basin’ (p. 253) writing it with ڭ (ك with three dots), but claims it comes from Turkish (see also Blanc 1981:192-3). It is then obvious that the normal pronunciation was the affricate [ǧ].”

    I love the internet.

  45. I assume others have noticed that a Google Books search turns up another William Halfpenny work before the one cited by the OED, from 1749, referring to a “gazebo room,” and without the Chinese association.
    Before that, in Wetenhall Wilkes’ An essay on the pleasures and advantages of female literature (1741), are these lines:

    Unto the painful ſummit of this height
    A gay Gazebo does our Steps invite.
    From this, when favour’d with a Cloudleſs Day,
    We fourteen Counties all around ſurvey.

    It’s in ECCO, but not GB. I know the OED plans an automated search for older quotations in a broad corpus, so it may well turn up something even better when it does happen.
    And that Lemon already identified gaze as the source in 1783.
    On the more fanciful side, how about Ça, c’est beau!?

  46. Not to be pedantic, but,
    No, it’s not pedantic, but it’s highly technical, and therefore more accurate if less accessible information. I appreciate highly technical language, and if there’s something I don’t understand, I’m sure there are plenty of people who can point in the right direction, if I only ask. Sometimes I have to read something with one finger on the paste button over the google toolbar or even save it for later if I don’t have time to work it out, but I love it, especially if it’s about Arabic. I do appreciate the nod to those of us whose backgrounds are more on the practical, hands-on side of language, but sometimes those esoteric links are the best ones.

  47. Well, the nontechnical takeaway is that I was completely wrong about the history: rather than preserving old Arabic /g/, Egyptian Arabic changed it to /dž/ like everybody else, then changed it back to /g/ a few centuries ago. That’s the kind of perverse historical detail I love.

  48. LH:…rather than preserving old Arabic /g/, Egyptian Arabic changed it to /dž/ like everybody else, then changed it back to /g/ a few centuries ago. That’s the kind of perverse historical detail I love.
    Yes, I took away that much, but there is more detail that I can’t seem to stop chewing on. In the reference cited about Egyptian jeem ج sound shift between the 8th and 11th centuries, it says,

    Whereas in the beginning of the period, velar stop pronunciation may have continued to be pre-Islamic [g], toward the end of the period, the shift to the affricate [g with carrot ˇ on top symbol] may have been completed (as indicated by Ibn Sina, for example). In sum, the shift was in progress in this period from a velar stop [g] to a fronted [g'] to an affricate [g with carrot ˇ on top symbol]. In the seventeenth century, however, a reversed sound shift had started to take place in urban Egypt.

    It goes on to say the shift had only started at the beginning of the 17th century the and cites several examples.
    The “[g]” sound here is obviously the “Latin” (?) rendering of the IPA /g/, and the [g with carrot ˇ on top] is /d3/. (Is that legible as a “j” as in judge? Is there no IPA character set? Oh, I see Hat writes/ dž/.) Okay so it shifted from what we would call a “g” to “j” and back to “hard g” with a transitional “fronted [g']” the first time and maybe the second time too. This transitional form of jeem “fronted [g']” I’m having a hard time visualizing, but it seems to have something to do with the way Texans pronounce g like d.

  49. David Marjanović says:

    Wow. That’s the only such back-shift known outside of some Samoyedic branch or other (where /si/ became /ki/).

  50. David Marjanović says:

    And of course I can confirm that the speculations about German are kaz boku.

  51. This transitional form of jeem “fronted [g']” I’m having a hard time visualizing, but it seems to have something to do with the way Texans pronounce g like d.

    No wonder you’re having a hard time visualizing it. I’m having a hard time just recreating what this would sound like. Mebbe ain’t no such thing, I ‘spec. The “IPA” in the citation sure don’t help, anyways. Notice the weasel-word “almost”, sneakily counterbalanced by “important”:

    This fronted “g” also results in the peculiarity of Texas speech which changes the “g” almost into a “d” sound. This occurs only when “g” is followed by “l”, as in “dlA:is” (glass) and “dlimps” (glimpse). This variant is important because it appears to be used by many Texans.

    Course Texas be a wide, welcoming place. While I was bein raised in El Paso, and later was learnin in Austin, nobody dlimpsed into their dlasses. But I’m not a phonologist, so I’m not trained to hear what’s not audible from the get-go.

  52. I take it back. After fruitlessly experimenting for an hour, off and on, with g, d (the tongue position more than the sound), and “L” in the isolated word “glimpse”, I decided to bring speech ductus into the picture (don’t know if one says this in English, what I mean is the way the speech of individual speakers flows).
    To my surprise, I found I could say “hadda [d]limpse th’other day” with no effort. The “[]” is intended to indicate that the d-sound is almost not there. I think I’ve heard people talking like that, but didn’t know that it was “used by many Texans”. The important thing is, there’s no g-sound in there at all. And merely a whisp of a d. But that’s just me talking.
    At least in the context of this Texan pronunciation, I find it misleading to talk about a “fronted [g']” (but then I’m not a phonologist, as I never tire of pointing out). I would say it’s not so much “pronouncing g as d”, as not employing a g-sound at all. Phonetic orthography is oriented primarily towards one dialect (or related group of dialects), as a rule (is that right?). When an illiterate person says “hadda [d]limpse”, it’s surely misleading to say “he’s pronouncing a g as a d”, cozzen of how he don’t know a g from a hole in the ground. He’s just speaking natural-like. If you try to “pronounce a g-sound as a d-sound”, you break your mouth.
    I got fooled by the d in the IPA “dlimps”, and by the locution “pronounce g like d” (but then …).

  53. For reference, Khan 1990:5 from that reference is here.
    I rather suspect that [g̟] looks awful in pretty much every browser. Has someone made a font for IPA with decent hints?

  54. I have to take that back even further, dammit! I can find a kind of g, sometimes. It’s like a minuscule gulp. The g is there in addition to the d, and follows it by a millisecond. Something like “[d][[g]]limps” (but then, I am no phonologist etc.)
    I’m not accustomed to scrutinizing the inside of my speakers. My eyes have gotten all twisted around from the inside.

  55. marie-lucie says:

    (Egyptian reverting to ancestral [g]) That’s the only such back-shift known outside of some Samoyedic branch or other (where /si/ became /ki/).
    Is a historical change /si/ to /ki/ actually attested by well-documented evidence, or is it a case that since several dialects had /si/ and only one /ki/, a reconstructor decided that it was more likely that a single dialect underwent a change rather than the opposite, and reconstructed */si/ for the proto-language rather than the */ki/ expected from the natural direction of palatalization attested in many languages (thus /ki/ to /si/, eg from Latin to French)? Some historical linguists (I will not name names, as I believe those people to be undeserving of the designation) think that the “one vs several” rule can trump well-known historical processes, instead of being just a shot in the dark when there is nothing else to go on, subject to revision if more explicit data come to light.

  56. marie-lucie says:

    ([g]limpse or [d]limpse)
    It makes sense that a [gl] cluster should evolve into a king of [dl]: the articulation of [g] (middle/back of tongue against the soft palate) tends to adapt to the following sounds (eg the part of the tongue involved is relatively farther back or front depending on the following vowel) and the articulation of [l] involves the tip of the tongue behind the upper teeth, which in articulatory phonetic terms is quite a distance to travel between the two sounds [g] and [l] if those are to be kept distinct. Bring the articulation of [g] still closer to that of [l] and the tip of the tongue rather than the middle will be involved, resulting in a [d] (since both [d] and [l] make use of the tip behind the teeth). But in this case, as in that of the t in little or button, the tip of the tongue tends to stay in place for the sequence [tl] or [tn], where the difference between the two sounds is one of manner, not of place of articulation. This is why Grumbly does not actually sense a [d] or a [g] but something vaguer just before the [l]. Excellent observations, Grumbly!

  57. quite a distance to travel

    My goodness, thanks for the explanation, Marie-Lucie! I could not have put it in your words, but I immediately understand what you just described, because I experienced all of it during my self-trials, in a confused way. Initially, I was trying to get from an isolated [gl] to an isolated [dl], and thought “that’s a long way to go”.
    What might be the role of “speech ductus” in all this, “across word boundaries”? In one way, it’s just an idea that allowed me, as a non-phonologist etc., to get away from my own “artificial” obsession with “isolated” sounds. But isn’t the very notion of individual words a scholarly construct, with its own advantages and disadvantages? That’s not to say it’s a “wrong” notion – I’m merely trying to tease out the working assumptions into the light of day, out of curiosity.
    I’m certain that this question is not new. Has it ever been taken out of the realm of pure speculation, put into a scientific form, and investigated? Two things:
    1. It seems that [d][[g]]limpse comes more easily when a vowel precedes, as in “hadda …”, but not a consonant. Also, I have this funny feeling that the d in hadda is aiding and abetting the subsequent [d][[g]]. How far has this action at a distance been observed to go? Could it be that “oral transmission” is not just, not even primarily, a matter of “memory”, but rather of a interacting system of phonetic / melodic force fields? Is memory a melody?
    2. Until the 9th century (was it?), Greek and Latin texts were usually written “in a run-on style”, with no optical clues to separate the “words”, as we need them today (or merely welcome as a convenience?). I wonder if this is just because we want to read faster? Or was there perhaps something additional going on, such as a different appreciation, perception of “speech ductus”? As a slip of a lad, I read a book by Owen Barfield called Saving the Appearances. I remember his arguing that, before the discovery of perspective, people “in fact saw the world differently” (or words to that effect).

  58. David Marjanović says:

    Is a historical change /si/ to /ki/ actually attested by well-documented evidence

    All Uralists seem to agree that that’s what happened. Probably only those few Samoyedic languages, of all Uralic ones, have /k/ in that place, and Samoyedic is a large family. — Unfortunately, though, my source is at least one LinguistList post that I can’t find.
    —————-
    There are languages that have changed /gl/ and /kl/ into /dl/ and /tl/, and examples for the opposite direction are also known, AFAIK.
    That words don’t exist in connected speech is usually taken into account by linguists (though it happens occasionally that one forgets doing it). For example, sources like Wikipedia will readily tell you* that stress in French is not word-final but utterance-final (it comes before audible pauses), and Spanish b/v is [b] (as opposed to [β]) only behind m and at the beginning of utterances, not words. The scary, scary tone sandhi of various southern Chinese “dialects” (where a syllable changes its tone depending on the tone of the preceding syllable) also operates within utterances, and so on.
    * I haven’t actually checked that, except for the last example.

Speak Your Mind

*