Exotic-sounding sounds

« previous post | next post »

A quick follow-up on this part of Bill Poser's post on the pronunciation of Beijing (and building on Ran Ari-Gur's comment, as I discovered while composing this post):

The article mistakenly asserts that the sound [ʒ] does not occur in English. It is indeed found in English, not only in measure but in such words as azure, pleasure, leisure, and treasure. What is true is that all of the words in which it occurs are loans from French, so the sound apparently has an exotic flavor even though it has existed in English for centuries.

Some readers may be a little puzzled by this. Many if not most English speakers, I think it's fair to say, don't know that the words in question are borrowings from French, and in any event (as Bill points out), these have been English words for a very, very long time. So how is it that [ʒ] retains this 'exotic flavor' to English speakers? I don't have the definitive answer to this question, but I do know one thing that undoubtedly plays a part in that answer.

First: a given sound doesn't simply 'exist' in a language. (I'm not implying that Bill was misleading our readers; he was just simplifying for the sake of the point of his post.) Specifically, different sounds have different distributions in a language, and some sounds have wider distributions than others. Still simplifying somewhat, consider some of the distributional facts concerning the sounds [l], [p], and their combination in English.

word-initial word-medial word-final
[p] pack hopper keep
[l] lack holler keel
[pl] play apply
[lp] helper kelp

Note the blank cells in this table: the sequence [pl] does not occur word-finally in any words of English, and the sequence [lp] does not occur word-initially. (Complicating things just a little: [lp] is not terribly common word-medially either, other than in suffixed verbs like help+er and in some proper names like Alpo.) So what this table shows is that the distributions of [p] and [l], at least in relation to each other, are not the same in English.

Back to [ʒ] in English: the distribution of this sound is incredibly limited. A great deal of the English words with this sound are just like the ones Bill cites in his post: two syllables, stress on the first syllable, and medial [ʒ] followed by -ure (pronounced as a rhotic vowel [ɚ]). And, with the exception of azure and (for some speakers) leisure, the vowel preceding the [ʒ] is also the same in these words: [ɛ].

As Ran Ari-Gur points out, another bunch of English words with [ʒ] are borrowings from Latin like vision, version, fusion, aspersion: two or three syllables, stress on the second-to-last syllable (which is the first syllable in the two-syllable words), and medial [ʒ] followed by -ion.

Note that more recent borrowings from French, such as genre, rouge, and beige, sound somewhat more exotic. This is likely because they do not conform to the distribution of [ʒ] in the more plentiful, older, and more frequent words that Bill cited. And frequency matters: American English speakers pronounce garage with a final [ʒ] — British English speakers pronounce it with a final [ʤ], to rhyme with marriage — and this puts less frequent words like barrage and mirage that rhyme with garage somewhere in the middle on the exotic-to-normal continuum. (It may also help to make other words with final [ʒ], like rouge and beige, less exotic to speakers like Ran Ari-Gur.)

And now, back to Beijing pronounced as [bejʒɪŋ]. This word has two syllables and the [ʒ] is medial, but the stress is on the second syllable and the [ʒ] is not followed by -ure or -ion. This all appears to support Bill's hypothesis, that this pronunciation of the name of China's capital thrives in part on the exotic nature of [ʒ], especially in contexts where it does not otherwise (or sufficiently frequently) occur in English.


  1. Bill Poser said,

    August 16, 2008 @ 7:13 pm

    Further evidence of the strangeness of [ʒ] in contexts other than the onset of an unstressed syllable for English speakers is the fact that they not infrequently have an affricate in words like genre, in which [ʒ] would otherwise be at the beginning of a stressed syllable, and words like rouge and beige, in which [ʒ] would otherwise be syllable-final. genre is a particularly nice example, I think, because even speakers who use an affricate nonetheless have the vowel [a] in the first syllable, as in French, indicating that they are emulating, imperfectly, the French pronunciation, not merely using an English spelling pronunciation, in which case the vowel would be [ɛ].

  2. Eric Bakovic said,

    August 16, 2008 @ 7:47 pm

    Thanks for pointing this out, Bill. I think the rest of genre ([ʒanɹə] also sounds somewhat exotic (though it's not much like the original French). There are words that end in [ɹə] (Sarah, mora) but not when preceded by a consonant like [n]…

  3. Ryan said,

    August 16, 2008 @ 8:00 pm

    Being a bit nitpicky (or perhaps wrong), doesn't [pl] occur word-finally with a syllabic [l], as in apple – [aepl] ? Even so, the sequence does seem fairly uncommon.

  4. Sridhar Ramesh said,

    August 16, 2008 @ 8:17 pm

    Perhaps the best [ʒ]-word, in terms of bringing up "exotic" images, is "Asian"

    Perhaps the worst [ʒ]-word, in terms of bringing up "excotic" images, is "Caucasian".

    My uneducated feeling is that the Beijing pronunciation with [ʒ] instead of [ʤ] isn't just because of an exotic feeling attached to [ʒ], but also because of experience breeding a feeling that "If it isn't English, then doesn't make a [ʤ] sound; try something else" (as exemplified by French, German, Spanish, Portuguese, Polish, Czech, etc.)

  5. Sridhar Ramesh said,

    August 16, 2008 @ 8:18 pm

    I guess angle brackets get misparsed and swallowed into HTML. The last quoted part above should read "If it isn't English, then <j> doesn't make a [ʤ] sound; try something else"

  6. Amy Stoller said,

    August 16, 2008 @ 8:32 pm

    "American English speakers pronounce garage with a final [ʒ] — British English speakers pronounce it with a final [ʤ], to rhyme with marriage"

    Isn't that oversimplified? Many AmE speakers use the affricate in words such as garage, massage, and barrage. And many BrE speakers use the affricate in the word garage if they don't use the "Broad A" (in BrE, the "Long A"); hence "garridge" to rhyme with carriage or marriage.

    Where AmE speakers differ more consistently from BrE speakers in pronouncing many such French loanwords is in assigning syllable stress. AmE nearly always stresses the ultimate syllable, BrE the penultimate.

    By the way, a couple of other candidates for incorrect "fricativization" are Raj and Taj Mahal.

  7. Michael said,

    August 16, 2008 @ 8:34 pm

    I'd like to point out that in rural Indiana, at least, we pronounce garage with a final [ʤ]. The final [ʒ] sounded pretentious until relatively recently. (Maybe still does; I don't live there any more.)

  8. mae said,

    August 16, 2008 @ 8:59 pm

    Is that exotic sound also the exotic sound of apostrophe-s in "How's your mother?" ?

  9. Ran Ari-Gur said,

    August 16, 2008 @ 9:16 pm

    Re: genre, rouge, garage, mirage, barrage: Oh, yes, those. Somehow I couldn't think of many examples before.

    Also: wow, I've never seen my full name so many times in one blog post. Please, feel free to call me "Ran".

  10. Tom said,

    August 16, 2008 @ 10:05 pm

    I for one am really annoyed by the focus on a "correct" pronunciation that I first heard on NPR and now have read in news stories repeated elsewhere. Can you imagine someone claiming that saying "Spain" would be offensive to spaniards because that's not how they say it? (Or saying "Seville" or "Madrid" if you want a closer parallel). Surveying those around me, I can't find anyone who's heard the so-called correct pronunciation spoken by an American (any American with exposure to Chinese is not really a valid example since the question should be the pronunciation in American English). Wouldn't that make the so-called correct pronunciation an annoying hypercorrection? Or has that existed as an alternate in regions of the language I haven't been exposed to (or have I just not noticed?)?

  11. Josh G. said,

    August 16, 2008 @ 11:42 pm

    The substitution of [ʒ] for [ʤ] is not just a result of hyper-foreign pronunciation. I've also heard it by native English speakers in words like "exogenous." There is some sense that [ʒ] is a "more sophisticated" free variant of [ʤ], perhaps due to its occurence in polysyllabic and learned English vocabulary, not just in words borrowed from French. The corresponding voiceless sounds [ʧ] and [ʃ] don't quite have this status (though we do find something similar in the pronunciation of "Chavez" and some other Spanish-derived words), likely due to the fact that [ʧ] and [ʃ] bear considerable contrastive load in everyday English vocabulary, while [ʒ] and [ʤ] do not.

  12. Garrett Wollman said,

    August 17, 2008 @ 12:40 am

    Sorry, I don't get this whole "exotic flavor" business, not for [Z] anyway. [y], maybe. I'd be inclined to agree with Sridhar Ramesh, that there's some intuition that [dZ] is a fairly unusual sound in non-English languages (probably primed by experience with other European languages). This seems like it ought to be testable, although I haven't a clue about phonology experiment design.

  13. wren ng thornton said,

    August 17, 2008 @ 2:47 am

    My off the cuff analysis would be that the [ʒ] pronunciation is so prevalent because of the preceding [j]. Given the palatal+(dental+palatal) sequence in conjunction with the fact that yogh is not that rare in English, assimilation is easy. Which also fits the Azerbaijian example. We do have [jʤ] as well (age, stage, cage, page,… many of which can be followed by [ɪŋ]) so assimilation can't be the whole story, but it does add its weight to the forces of exoticism.

  14. Rubrick said,

    August 17, 2008 @ 4:26 am

    I simply could not let this discussion pass without providing a link to this classic Simpsons moment in which Moe makes fun of Homer for his use of the fancy French word "garage".

  15. Justin L said,

    August 17, 2008 @ 4:46 am

    I saw the stress pattern as somewhat unusual for English, but as for the affricate, I just thought [dʒ] becomes [ʒ] because [jdʒ] is a very awkward cluster of consonants, particularly with the syllable break in between the [j] and the [dʒ]. The other words I can think of with a similar pattern–"aging", "paging", and the like–all are stressed on the first syllable, and almost seem to have an extra [d] before the syllable break (like "aging" as [ejd.dʒɪŋ]) in the speech of me and my roommates.

    Of course, if we spelled the capital of China "Baging", it would probably be mistaken as a typo for "bagging".

  16. Nancy Irving said,

    August 17, 2008 @ 7:20 am

    Oddly, of the words originally given–azure, pleasure, leisure, and treasure–as supposedly importing the "exotic" [ʒ] sound from the French, none in fact has the [ʒ] sound in the French words from which they are derived.

    Azur, plaisir, loisir, tresor–all are spoken with the regular English [z] sound.

    So wherever the [ʒ] came from here at least, it wasn't from France.

  17. Ian Tindale said,

    August 17, 2008 @ 8:15 am

    You've just alerted me to a something: I read the word "azure" and like so many other occasions in my past 47 and a half years (after the initial boot-up), mentally pronounced it exactly like it's written. In other words, like the words "a lure" or "a sure" or "a pure".

    Now that I think about it, I don't think I've ever heard another person say the word to me, to gain any reference from. I'm sure there's plenty of words in my – oh, what's that word where you have lots of words in your head; ah yes, vocabulary – that I might have only ever seen written and never heard pronounced. This is probably one of them.

    Alternatively, if I have in fact been hearing people say this regularly and with frequency all the time non-stop and ceaselessly, it's probable that I simply never made the connection between the word that sounds like "a pure" and whatever mish-mash was coming out of their mouths.

  18. John Cowan said,

    August 17, 2008 @ 10:12 am

    Nancy: Saying that [ʒ] is found in words of French origin is not the same as saying it came directly from France. [y] in French loanwords such as "azur" became [ju] in English, splitting the front rounded vowel into separate tokens of frontness ([j]) and roundedness ([u]), as in the name of the letter U. Then [zj] became [ʒ] by palatalization, just as in the -sion words.

  19. Ran Ari-Gur said,

    August 17, 2008 @ 10:34 am

    @John Cowan: But if you're acknowledging that [zj] became [ʒ] by palatalization in English, then the rest of the argument kind of falls apart. "It feels exotic because it's the result of a common English process that has occasionally applied to French loanwords"?

  20. Jean-Sébastien Girard said,

    August 17, 2008 @ 11:18 am

    Ryan, I'm fairly sure than in words like "apple", triple, and so on, the /l/ is analysed as syllabic consonant /l̩/ (not to mention fairly strongly velarized, but that would probably not count), and hence technically (as an alternative notation is /əl/) not the same sound sequence as that in "play".

  21. Aaron Davies said,

    August 17, 2008 @ 12:23 pm

    @Ran: Well, yeah, what else? Just because the French words go through some sort of shift on their way into English doesn't mean they don't still have some sense of foreignness about them. It's in the vowels if nothing else–"azure" should be *[æʒure], not "[aːʒure]".

  22. Aaron Davies said,

    August 17, 2008 @ 12:31 pm

    Which reminds me, this whole debate reminds me of another example of "hyper-correction of foreign words" (one which I think has been noted on LL before, though not recently)–American speakers tend to assume that /æ/ is an exclusively American sound, and automatically render "a" as "aː" when dealing with "foreign" words. For instance (and since someone already brought up Starbucks), I doubt anyone aware that "grande" has two syllables would ever render it [grænd iː] or [grænd eɪ]–they "automatically" know better. (This is reasonably correct, of course, as an Italian borrowing, but I can't think of a decent hypercorrection off the top of my head. Still, it illustrates the point.)

  23. Ben said,

    August 17, 2008 @ 1:32 pm

    Several commenters, especially Sridhar Ramesh, have touched on what I think accounts for the "foreignness" of [ʒ] in American English, namely that it's not so much that [ʒ] feels foreign as that [dʒ] feels like a mispronunciation in a borrowed word. It's coarse to say corsage with a [dʒ] (although I realize I have an interesting doublet in my idiolect: homage and homage, the latter with [ʒ] and ultimate stress, and they play distinct roles).

    I'm glad the various American pronunciations of "garage" came up; I've encountered four, the product of two variables: (1) [ʒ] versus [dʒ], and (2) the first syllable realized as either an unstressed syllable or a consonant cluster. I suppose there's probably a geographic element to these pronunciations, but my gut tells me there's more sociology behind them than geography. Data would, of course, be nice.

    Aaron Davies: I think the best examples of hypercorrecting /ae/ to /a/ are in words borrowed from Arabic, like "Islam." And yes, I know the Arabic vowel isn't exactly /ae/, but it's much closer to /ae/ than to /a/.

  24. Ran Ari-Gur said,

    August 17, 2008 @ 3:13 pm

    @Aaron Davies: Interesting! In my dialect "azure" is said nothing like either of your trisyllabic pronunciations, but instead is said disyllabically, as ['æʒɚ] — just like "after", but with [ʒ] instead of [ft]. Regardless, you seem to be intentionally choosing the most exotic of the words; out of measure/seizure/pleasure/treasure/etc., and vision/television/lesion/recursion/etc., how many sound exotic to you? There are simply too many of these words, and they're too regular, with too many of them having come from too many sources and been in the language for too long, for [ʒ] to automatically sound exotic to any speaker of Modern English. What's relevant, as Dr. Bakovic points out, is that [ʒ] has a limited distribution in what you might call "normal" words, and so words that use it outside of that distribution (can) sound exotic.

  25. KCinDC said,

    August 17, 2008 @ 3:24 pm

    Ben, isn't the fancier word, with [ʒ], spelled "hommage"?

  26. marie-lucie said,

    August 17, 2008 @ 3:34 pm

    gar- vs gr- :

    Years ago I lived in a rural area where the way to deal with garbage was to burn it some distance from the house, so we did so too, in a cleared patch partially surrounded with grass. One day the grass caught fire and started to burn. Unable to douse the flames with buckets of water, I called the fire department, which immediately sent two fire-trucks. The leading fireman said "Where is the garage?" I must have looked astonished, so he continued "You said you had a garage fire! – No, I said a grass fire". The firemen might not have been so prompt if they had not heard my (French-accented) grass as their g(a)rage.

  27. Ollock said,

    August 17, 2008 @ 5:21 pm

    [ʒ] isn't just restricted to Beijing in English. In my experience quite a few people — TV announcers and otherwise use it for pinyin and . Sometimes I wonder about how this affects Chinese speakers listening to English language news reports, since these sort of mispronunciations — along with general confusion over what the pinyin vowels represent ([lu] for [liow]/ [aw] for [ow]) and the various romanizations — might cause even more confusion overtop of the more tolerable inability of English speakers to produce Mandarin tones (not that any of these variants are really inherently intolerable — given the understanding that some of the distinctions do not exist in English and most English speakers are unfamiliar with Chinese).

  28. Bryn LaFollette said,

    August 17, 2008 @ 9:50 pm

    @wren – I believe Azerbaijan is the usual spelling, which doesn't have an 'i' following the 'z'. But even without that being the case, I don't think palatal assimilation is really what's going on here.

    @Ran, Sridhar, Garrett, etc – It seems like there is a pretty good argument that can be made that the highly restricted distribution of [ʒ] in English combined with the sense of [ʤ] being a non-foreign sound, leads to [ʤ] > [ʒ] as a hypercorrection pattern (and maybe not just for "foreign" words, but also for "prestige" words). The [a]/[æ] and [ej]/[i] alternations seem like good examples of the same process, I think. Although, I think the flip-side to Aaron Davies' argument might be the tendency of BrE speakers to go over-board in the direction of [æ].

  29. Bryn LaFollette said,

    August 17, 2008 @ 10:01 pm

    Also, just a point on the "correctness" issue of which pronunciation is more "right". I don't think that that's really a part of this issue, even if that's how the news stories may have been framing it. I would argue that once a word gets adopted by a language, the only "right" pronunciation is going to be whatever your locale speech community is going to accept without blinking.

    The interesting point in this discussion, I feel, is really that given we know what the "original" pronunciation of the native Beijinger for the name of the capitol of China is, how did the common American English pronunciation with [ʒ] as the onset of the second syllable arise. I think any issue of "correctness" or "authenticity" is only going to be a point of divisiveness and distract from the real issue here.

  30. Eric Bakovic said,

    August 17, 2008 @ 11:13 pm

    Just now catching up on (some of) these comments:

    @ Amy Stoller: yes, I oversimplified. My bad.

    @ wren ng thornton: [jʤ] as in paging is not a palatal-dental-palatal sequence. The digraphic transcription of affricates in the IPA is a bit misleading; the initial stop portion is actually articulated in the same palatal (technically, postalveolar) position that the final fricative portion is. (Try pronouncing paid and page, paying close attention to where your tongue is at the beginning of the [d] and the [ʤ], and I think you'll see what I mean.) So place assimilation cannot be part of the story here.

    @ Tom: I agree, and I wrote about this on phonoloblog a couple of times. Sadly, the blog is down at the moment, but if you Google "foreign pronunciation phonoloblog", you can see the cached posts.

    @ Nancy Irving: it's true that Modern French does not have [ʒ] in the relevant words — but English didn't borrow those words recently enough for Modern French to be the relevant form of the language to look at for answers. I generally agree with John Cowan's response (pace Ran's reply), but it'd be nice to know the history of the pronunciation of these words in French. I hope to follow up on this at some point soon.

  31. marie-lucie said,

    August 18, 2008 @ 12:33 am

    @Eric Bakovic: it'd be nice to know the history of the pronunciation of these words in French

    Intervocalic -s- was [z] in Old French and still is in Modern French. Stressed [o] as in -sion became [u] (written ou) in Anglo-Norman (eg nacioun 'nation') and later followed the patterns for OE [u].

    I agree with John Cowan: when borrowed into Middle English, French [y] merged with [u], which later developed an initial glide through the English vowel shift. Then sequences s or z + j became palatal fricatives, hence the current pronunciation of the English words in -sure and -sion (and azure is close enough to the -sure words to have been caught up in the pattern).

    I also agree with you that because of the odd distributional pattern of the voiced palatal fricative and its restriction to words which are not part of the general vocabulary (except for the -sure and -sion words), a foreign word which "out to sound exotic" will tend to be pronounced with the palatal fricative rather than the affricate (as in Abidjan pronounced as Abijan, as I commented earlier). Words which become very common (like garage) acquire the English stress pattern on the first syllable and convert the consonant of the suffix to an affricate, continuing a much earlier process (as in carriage, courage, etc).

  32. Tristan McLeay said,

    August 18, 2008 @ 2:52 am

    The original post has it right. [ʒ] is only normal in English when it can be looked on as /zj/. [j] can only follow a consonant if the next vowels are /ə/ and /u:/ (and maybe /ʊ/ or /ʊə/ or /ɔ:/ depending on what your dialect does with words like " pure ”. Any other context is valid only in foreign words, so /ʒɔnrə/ has a foreign feel to it that /dʒɔnrə/ lacks. In that regard, [ʒ] isn't really even a proper phoneme of English; it just kind of was granted that honor on account of the comparable changes that happened to its friends /t s d/ and the fact that it fills in a gap at /tʃ ʃ dʒ/. Personally I'm inclined to convert all of these unusual cases of /ʒ/ into /dʒ/. So if we look at it in that environment, " Beijing " with a /ʒ/ must be a perceived-foreign thing.

  33. Aaron Davies said,

    August 18, 2008 @ 9:29 am

    @Ran, my IPA shortcomings may be showing–I think the [re] endings were a massive thinko on my part. "Azure" is definitely two syllables for me, something like "ah, sure", except with all the stress on the first syllable and a "zh", not an "s" (alternately, like the first two syllables of "aujourd'hui", except with an "ah", not an "oh"). (I'm falling back on basic "sounds like" notation. Assume General American (except for the French obviously) if there's any question.)

    re Azerbaijan, i take it the "correct" pronunciation is [æ zɚ baɪ ʤɑn], as indicated at Wikipedia? this then would be an example of both points, as i commonly hear it as [aː zɚ baɪ jaːn]. (amusingly, the wikipedia .ogg sound sample quite clearly says [æ zɚ baɪ ʤaːn], despite clearly notating a [j]. perhaps one was changed without the other….) incidentally, i wonder if there's some vowel harmony principles going on in this example? i find myself right now wanting to fix both stressed vowels as either [æ] or [aː]; using one of each seems much more difficult to remember.

  34. Ken Brown said,

    August 18, 2008 @ 3:50 pm

    The exotification of the names of foreign capital cities was very noticeable when Afghanistan started coming into our broadcast news.

    At first the news readers, like most British people, put the stress on the second syllable and used a long rather un-English vowel for it.

    Later on, when they got more used to it, they said the word as if it was an English town whose name was spelled "Corble".

  35. mollymooly said,

    August 18, 2008 @ 4:39 pm

    Maybe we should just go back to using "Peking". Most other European languages still use their traditional Pekin-type spellings. The burden of being the World Language is that other countries feel they have the right to insist on a particular toponymic form being used in English, not just in their own languages; they don't seem to care what the Germans, Dutch or Italians write.

    In the days of the USSR, Azerbaijan, Ajaria, Tajikistan, etc. all had -dzh- rather than -j- in the Romanization of their Cyrillic spellings. Ironic that an obviously non-English spelling convention should be less prone to English mispronunciations.

RSS feed for comments on this post