Clipping McDonald's

« previous post | next post »

Commenters on a recent post ("Australian hypocoristics") discussed the vowel quality of the first syllable of McDonald's in detail and at length. The issues involved are interesting enough to deserve a post of their own.

One central question is whether there is more than one qualitatively distinct type of schwa (reduced unstressed vowel) in English, either within or across varieties.

The null hypothesis here, in my opinion, is that there's a single category, representing a phonologically unspecified neutral vowel, which takes on various qualities depending on phonetic environment, speaking rate, and various paralinguistic factors. This null hypothesis might be false, but I've never seen any clear evidence to refute it.

Many people, including some famous phoneticians, have advanced a contrary view, namely that (at least in some varieties of English) there are two distinct schwas, one higher and fronter, the other lower and backer. A supposed near-minimal-pair, for some of the schwa-splitters of my acquaintance, is Alice vs. Dallas. I've never seen any evidence from natural speech that supports this view — performance of (near-) minimal pairs is always suspect in such cases, since it's subject to facultative disambiguation.

Another position, not inconsistent with this first one, is that there are characteristic differences in the pronunciation of such vowels among dialects or varieties of English, with some varieties having higher-fronter schwas, and others having lower-backer ones. This certainly might be true, but again, I haven't seen any real evidence.

Why, besides a general scientific show-me attitude, am I skeptical of these opinions? One reason is that I've learned not to trust my own intuitions in similar cases.

For example, I have the very strong intuition that I pronounce ladder and latter differently. And when I produce these words as a decontextualized minimal pair, there are small but significant average differences between the two words, in the mean values of various phonetic measurements such as the relative duration and amplitude of the medial voiced tap — although the distributions for the two categories are highly overlapped. But in normal usage, these differences evaporate, or are overwhelmed by other sources of variation, or both. And when I try to identify randomly-selected examples of my own pronunciations of latter or ladder, even if the exemplars are taken from decontextualized minimal-pair performances, my performance is indistinguishable from random guessing, .

I reckon that my intuition in this case comes not only from the fact that the words are spelled differently, but also from the fact that latter is a word that I've mostly read rather than heard, while ladder is a word that I've mostly heard rather than read. In any case, I've learned from experiences of this kind that my own intuitions about how I pronounce things are not always reliable.

What about the argument from variable nickname formation, advanced by more than one commenter? Thus

I was surprised to learn recently that Americans sound the first syllable of 'McDonald's' as 'Mick', whereas we Brits pronounce it with a schwa or as 'Mac'. Hence the American hypocorism 'Mickey D's', and the British one 'Maccy D's'.

Other commenters (correctly) denied that any such general trans-Atlantic difference exists, but some still felt that finer-grained dialect differences in the pronunciation of initial unstressed M(a)c might underlie the nickname differentiation.

But entirely independent of how speakers may pronounce schwas, there's quite a bit of variation in how they behave in promoting particular schwas to the status of full vowels. An interesting example emerged from some fascinating (but unfortunately unpublished) work by Marian Macchi, done long ago at Bell Labs. (What follows is based on decades-old memories, and should be evaluated as such…)

Marian had observed that American English speakers generally exhibit anticipatory coarticulation between schwas and the vowels of following syllables: thus in adept the initial vowel is somewhat [ɛ]-like; in adapt the initial vowel is somewhat [æ]-like; and in adopt the initial vowel somewhat [ɒ]-like. She hypothesized that in more carefully-articulated speech, this coarticulatory effect would be reduced or eliminated. So she set up an experiment based on a dictation paradigm, where local high-school students were asked to read lists of words to an experimenter, who occasionally pretended not to have heard them clearly, and interrupted with something like "Excuse me?" or "Could you repeat that?"

What she found was that in the (definitely hyperarticulated) repetitions, some of the subjects indeed produced un-coarticulated (or at least less coarticulated) schwas, [əˈdɛpt] [əˈdæpt] [əˈdɒpt] — but others went in the opposite direction, producing clearly-articulated forms with complete assimilation of the schwa as a full copy of the following vowel, [ɛˈdɛpt] [æˈdæpt] [ɒˈdɒpt] . . .

These were all teen-aged speakers from the same high school in a suburban area of northern New Jersey, apparently all speakers of the same variety of English. Marian's experiment suggests that some speakers apparently are prone to promote predictable context-conditioned phonetic variation to phonemic or even lexical status, while others aren't.

But there's another source of potential variation, which is specifically relevant in clipping McDonalds: speakers are certainly aware that M(a)c is a morpheme, and that there are quite a few names where it ends up with main word stress, e.g.


In those cases, as the spelling "Mac" suggests it should be, the vowel is [æ].

So even someone who perceives the unstressed vowel in McDonald's as closer to [ɪ] than to [æ] might still choose [æ] as the stressed version, based on how the M(a)c morpheme is pronounced when stressed. [And, as a commenter observed, there's the "Big Mac" to provide morpho-phonemic guidance!] Alternatively, someone who is aware of the morpho-phonology might still decide to go with (their perception of) the phonetics.

(There's some relevant additional discussion in "Symbols and signals in g-dropping", 3/23/2011.)


  1. Levantine said,

    July 29, 2013 @ 8:45 am

    I'm the commenter whose post was quoted above, and while I agree that the situation in the States is rather more complex than I first suggested, I do think that the total lack of the 'mick' pronunciation in the UK represents a real transatlantic difference. The 'mick' (or 'mick'-like) pronunciation may not be universal in the US, but it's evidently widespread, whereas it doesn't seem to exist at all in the UK. The corresponding difference between the American and British nicknames surely cannot be coincidental.

  2. AJD said,

    July 29, 2013 @ 9:10 am

    According to Trudgill et al. (2000)'s study of recordings of early New Zealand English, 68% of the speakers in their corpus used what you're terming "two distinct schwas"; that's a study of natural speech.

  3. Jon Lennox said,

    July 29, 2013 @ 9:14 am

    Does the "Rosa's Roses" example refute your null hypothesis about schwas, or is there something about the term "phonologically unspecified neutral vowel" that excludes it?

    [(myl) There's obviously a morphological difference there that might have some phonetic consequences, but again, I think my null hypothesis would be that it's a case exactly like my intuitions about my pronunciations of latter and ladder, so that the morphological difference between Rosa's and roses gives rise to a phonetic intuition (perhaps based on different degrees of perceived contact between the schwa and the final [z]) that does not correspond to a reliable difference in actual pronunciation, except in facultative contexts.

    I could well be wrong about this, but I'd like to see some evidence that goes beyond intuitions about (or measurements of) minimal pairs or other facultative exemplars.]

  4. Levantine said,

    July 29, 2013 @ 9:31 am

    In my variety of English (SE British), the last syllable of 'roses' is pronounced like the word 'is', and not with any kind of schwa. Actually, I also pronounce the last syllable of 'Alice' with a clear I-sound.

  5. AG said,

    July 29, 2013 @ 9:35 am

    As I hear it New Zealanders say something more like "Muckdonald's", with the muck part being pretty fast. Not sure if that fits into the whole schwa discussion at all.

  6. Faldone said,

    July 29, 2013 @ 10:40 am

    One thing that has to be remembered is that people are generally not very good at self-analyzing their pronunciations of words. One specific example came about when Skaneateles, New York had its fifteen minutes during the Clintons visit there in 1999. When asked how the name of the town was pronounced, the mayor clearly said, "/'skeɪ ni "æt lɨs/. But when he pronounced the name later in normal speech it was just as clearly "/'skɪ ni "æt lɨs/.

  7. NW said,

    July 29, 2013 @ 11:06 am

    Of course there are accents like RP where the unstressed vowel is a KIT vowel, but that's not by any means a schwa or neutral vowel. Presumably we're interested in the existence of accents with a three-way distinction between Rosa's, roses/rose's, and rose is (where we want the strong form of 'is', so put it at the end of a clause: I'm going if Rose is).

  8. Ellen K. said,

    July 29, 2013 @ 11:10 am

    Seems to me there's a difference between comma and commie that's not just in our heads. So, there can be a vowel contrast in a non-stressed syllable. So what are the conditions that allow for contrast in an unstressed syllable, versus the typical lack of contrast? Is it a matter of degree of stress? Is it only at the end of a word? (Or perhaps end of a morpheme.)

    I'm also thinking, people who might contrast the two words in the phrase "Rosa's Roses", if they had reason to say it, wouldn't necessarily pronounce them differently when saying the words individually.

    [(myl) Certainly the final vowel of e.g. city is not a schwa in any variety of English that I'm familiar with. And it could be argued that the final vowel of e.g. fellow is another unstressed vowel that's different again. There are also clearly contextually-determined varieties of schwa, as e.g. in the adept/adapt/adopt experiment , or those that are colored by [r] or ]l], or the clear effects of adjacent coronal consonants. I'm just not convinced that there are lexically-distinct varieties of schwa.]

  9. Peter S. said,

    July 29, 2013 @ 11:12 am

    One experiment that might clear this up is to tell Americans a story about /ɛrɪn/ or /ɛrən/, and then later ask them questions about the story, and see which pronoun they use to describe him/her. I know I've been confused about the sex of people named Erin/Aaron on occasion, and I think (although as you say, it's really easy to fool oneself) I was taking my cue from the weak vowel, since they're otherwise pronounced identically in most of the U.S.

  10. Levantine said,

    July 29, 2013 @ 11:14 am

    Ellen, my pronunciations of 'Rosa's' and 'roses' would always be distinct from each other, and I believe this is true of many (most?) Brits. I would never pronounce the second syllable of 'roses' with anything other than a distinct I.

  11. Mark F. said,

    July 29, 2013 @ 11:32 am

    Ellen – I don't think the second vowel sound in 'commie' is any kind of schwa at all.

  12. Rodger C said,

    July 29, 2013 @ 12:07 pm

    I write private notes in phonemic tengwar and have never encountered the need for more than one schwa symbol.

    @AG: Aren't New Zealanders also known for pronouncing "it is" with two schwas?

  13. Ellen K. said,

    July 29, 2013 @ 1:09 pm

    @Mark F. I didn't say or suggest that the second vowel sound in commie is a schwa. I said it's a vowel in an unstressed syllable. We are, after all, talking about vowels in unstressed syllables, and contrast or lack there of. To say it's not a schwa and therefore irrelevant is begging the question, it seems to me. The issue is not whether there's contrast between syllables that have what we call a schwa. The question is whether or not there is vowel contrast (or variation) in non-stressed syllables, or in certain kinds of non-stressed syllables, regardless of how we label them. The contrast that some people apparently have between Rosa's and Roses is relevant even if we say that one of the two words does not have a schwa.

    The question could perhaps be stated: When is there a distinctly audible quality difference in vowels in unstressed syllables, versus differences that exist in people's heads but don't exist in speech. Or so it seems to me.

  14. Jonathan Mayhew said,

    July 29, 2013 @ 1:23 pm

    I thought we were discussing vowels that could be schwas or not, and whether there are two schwa sounds or one.

    I would never say "comucks" for comix but I might say "comuck relief."

    [(myl) Assuming that this is true, an alternative explanation would point to coarticulation with the final [s].]

  15. Pflaumbaum said,

    July 29, 2013 @ 1:35 pm

    @ Levantine –

    No-one doubts that you hear a clear difference in your own and others' speech. I do too – though like (I think) most linguists I'd classify it as a close central /ɨ/, similar but not identical to the /ɪ/ of KIT.

    But the issue, as MYL is arguing, is whether that impression actually corresponds to any real systematic difference.

  16. Levantine said,

    July 29, 2013 @ 1:46 pm

    Pflaumbaum, I can assure you that 'roses' or 'Alice' with a schwa is simply not possible in my accent. I'm sure others from the south east of England would agree. Tony Blair is the only person with an otherwise southern British accent whom I've ever heard using a schwa-like sound in such contexts, and that may have something to do with his having spent some of his youth in Australia, where I believe 'roses' and 'Alice' are likely to be pronounced with schwas.

  17. SK said,

    July 29, 2013 @ 2:01 pm

    Following up on what's been said by Levantine, Pflaumbaum and others: I'm a BrE speaker, with a more-or-less RP accent, and I certainly share the impression that I pronounce 'Rosa's' and 'roses' differently – but I do take MYL's point that in principle I may be imagining a systematic difference which isn't really there. However, it might be relevant that of all my RP-ish friends, there is one in particular that I always notice *doesn't* make this distinction – and this feature of his accent stood out for me very quickly when I first met him. Presumably the fact that I notice this specifically with him suggests that other people really are making the distinction – otherwise, what would I find so striking about him not making it?

  18. Pflaumbaum said,

    July 29, 2013 @ 2:11 pm

    @ Levantine –

    Well, I'm from South-East England too, and my impressions strongly agree with yours.

    It's still begging the question though. MYL had strong intuitions about latter/ladder. I was speaking to a fellow-Londoner the other day who denied passionately that she pronounced the /t/ of top differently from that of stop and pot.

    I wonder if mis-spellings provide evidence here. UK Google searches give 4.45 million hits for benifit, and only 322,000 for benafit. It's hard to measure though, as presumably most people who heard schwa wouldn't make a typo at all, since 'e' so commonly represents schwa (though planitary only gets 24,000, compared to 86,000 for planatary). Likewise, the common rediculous isn't much use, since re- is so much more common and productive a prefix than ri-.

    Interestingly, I'm not completely certain in all my judgments. I'm not sure which of the two (assuming there’s a real difference) I pronounce in quarantine, for instance. Quarintine beats quarentine in hits, but I guess most of the schwa-hearers were again absorbed into the group who spelt it correctly.

  19. Levantine said,

    July 29, 2013 @ 2:17 pm

    SK, that's sort of the point I was trying to make with my Tony Blair example. His use of schwas always sounded very peculiar to me, precisely because it was unconventional. I've just asked some friends of mine to say 'Rosa's roses' out loud (without telling them why), and they're definitely pronouncing the two words very differently.

  20. Levantine said,

    July 29, 2013 @ 2:22 pm

    Pflaumbaum, while I understand that people's self-perception can be off, I am as certain of the fact that I don't pronounce 'roses' or 'Alice' with a schwa as I am that I don't say 'castle' or 'grass' with a short A.

  21. Eric P Smith said,

    July 29, 2013 @ 2:27 pm

    I am not a linguist but a keen amateur.

    Most authorities acknowledge at least 3 unstressed vowels in English, namely /ɪ/ as in 'rabbit', /ʊ/ as in 'to each', and schwa. I don't think anyone is claiming that the vowels in the unstressed syllables of 'rabbit' and 'to each' are schwa. The question is how we categorise unstressed vowels other than those two.

    My dialect is Scottish Standard English (SSE). I have long believed that unstressed vowels in SSE reduce less far than unstressed vowels in most other dialects. In SSE the initial vowels in ‘theistic’, ‘intense’, ‘except’, ‘absurd’, ‘obtuse’ and ‘O'Rourke’ are all unstressed and all contrast with each other, and in my view they are partly reduced forms of the vowels in FLEECE, KIT, DRESS, TRAP, LOT and GOAT respectively. I believe that I can back up that claim with the following minimal pairs in my speech and in that of many other older SSE speakers: ‘elicit’ /iˈlɪsɪt/ ‘illicit’ /ɪˈlɪsɪt/; ‘insure’ /ɪnˈʃur/ ‘ensure’ /ɛnˈʃur/; ‘except’ /ɛkˈsɛpt/ ‘accept’ /akˈsɛpt/; ‘ablation’ /abˈleʃən/ ‘oblation’ /ɒbˈleʃən/; ‘Orion’ /ɒˈraɪən/ ‘O’Ryan’ /oˈraɪən/. Which of the unstressed SSE vowels in 'except', 'absurd', 'obtuse' and 'O'Rourke' we call schwas depends, it seems to me, on the definition of schwa: but in my view they are all clearly different.

  22. J.W. Brewer said,

    July 29, 2013 @ 2:35 pm

    Is there room in the intersection of phonology and onomastics for secondary stress (especially in this perhaps unusual context of proper names with identifiable component morphemes) having some sort of weird impact on vowel articulation? When I pronounce McDonald, the primary stress is on the second syllable, but the initial syllable doesn't seem as unstressed/reduced/half-swallowed as the final syllable. A parallel might be the surname Van Buren – primary stress in the middle, but enough secondary stress on the "Van" that I don't think the vowel is reduced all the way down to a schwa although I'm not sure if it's exactly the same vowel as you get when you say "Van" as a standalone word. Are "partially-reduced" vowels a thing? If the stress pattern was somehow cuing (some) people to end up with neither a fully-reduced schwa nor the fully unreduced vowel you get in first-syllable-stress surnames like Macintyre, maybe you would get somewhat unpredictable results?

  23. Levantine said,

    July 29, 2013 @ 2:36 pm

    Eric, regarding the first batch of words you give, I would use a schwa only for 'absurd' and possibly 'obtuse'. My accent is more or less RP, and I would make most of the same distinctions that you would.

  24. Pflaumbaum said,

    July 29, 2013 @ 2:40 pm

    @ Levantine – I think Blair's shwas are residual Geordie (at least… those of us who accept that there's a /ə/-/ɨ/ split in most English English accents would agree it is neutralised in Geordie as well as Aussie).

    SK – interesting point. I suppose, to take the position of MYL's absurdly unqualified advocate, I'd say that it's important to note whether you were aware of the supposed distinction before you noticed your friend's accent. If so, you may have put whatever it is that's 'odd' about his speech into that category incorrectly.

    The point being that there is no 'Southern English schwa', as such, just a continuum of productions varying within and between speakers. For example, on page 146 of this paper are plotted the vowels of a group of female Romanian speakers. The value of schwa (recorded there, annoyingly, as /ʌ/ for disambiguation purposes) varies quite widely. Some speakers' schwas overlap with not only others' /ɨ/, but also /e/, /a/ and even in some cases /o/.

    Which isn't to say /ə/ and /ɨ/ aren't separate phonemes in Romanian. I'm just saying that you may have heard something odd about his schwas – maybe they're more open than most people's? – and interpreted according to your pre-existing classification as a neutralisation of the close/mid contrast in weak vowels.

    Well, that's my best effort…

  25. J.W. Brewer said,

    July 29, 2013 @ 2:40 pm

    Note that (at least according to wikipedia), John Wells' initial lexical sets proposal had three for unstressed vowels, but only in final position – the keywords being happY, lettER, and commA. Since unstressed vowels commonly occur in non-final syllables, there is clearly More Work To Be Done, unless it turns out to be the case that they can all be squeezed for all varieties of English into the commA set.

  26. J.W. Brewer said,

    July 29, 2013 @ 2:51 pm

    I think for prosodic reasons the nickname "Mickey D.'s" requires a two-syllable first name (um, a trochee specifically, after I use wikipedia to remind myself of what the opposite of an iamb might be). Since "Macca" is not part of the inventory of common AmEng nicknames and "Mac" (which is) is prosodically unsuitable, "Mickey" turns out to be the best fit available under the circumstances, without signifying much about how "McDonald's" is pronounced in AmEng – since as noted somewhere above AmEng speakers are perfectly happy to call someone surnamed McDonald "Mac" without the nickname matching the pronunciation of the same morpheme in the context of the surname. Note the asymmetry in the other direction that the AmEng nickname inventory has "Mickey" but lacks clipped/monosyllabic "Mick," which, despite being borne by numerous prominent non-American rock musicians, remains as foreign-sounding to the AmEng ear as "Nigel" or "Trevor."

  27. Levantine said,

    July 29, 2013 @ 2:59 pm

    J.W. Brewer, how would you then explain the British nickname, which is 'Maccy D's'? Also, several American commenters responding to the earlier post agree the they pronounce 'Mc' as 'mick' or something close to it, which is not something you'd encounter in British English.

  28. J.W. Brewer said,

    July 29, 2013 @ 3:06 pm

    OK, earliest instance of "Mickey D's" for "McDonald's" I could find in google books is from a 1979 book called "Dating Habits of Young Black Americans" and the majority of pre-1985 hits seem to be from magazines serving a primarily black readership (Ebony, Jet, Black Enterprise – although Railfan & Railroad may be an outlier there). This suggests it may have started as an AAVEism. Refine your phonological theories accordingly! (Elsewhere in American black popular culture, there was a 1975 song titled Mickey D's by an R&B/soul/proto-disco group called The People's Choice, associated with Gamble & Huff and the then-in-vogue Philly Sound – you can find it on youtube but it's an instrumental so it's difficult to discern if the title actually alludes to McDonald's.)

  29. Pflaumbaum said,

    July 29, 2013 @ 3:12 pm

    Getting people to say Rosa's roses or elicit – illicit is really not much good here. MYL already discounted that as subject to 'facultative disambiguation'. He's talking about natural, connected speech here. And though if I had to call it, I'd bet my Language Log subscription on their being an /ə – ɨ/ contrast in some accents, I'd bet my house on most of Eric P Smith's contrasts evaporating in ordinary speech.

  30. J.W. Brewer said,

    July 29, 2013 @ 3:18 pm

    Hey, I may well pronounce semi-reduced "Mc" kinda/sorta like "mick" my own self (although now that I'm self-conscious about it my self-reporting is not going to be reliable). I'm just saying that I don't infer much from that. There are more or less by definition no trochaic nicknames with a schwa in the right place, and an established/pre-existing male nickname that fits the prosody is necessary for the poetic conceit to work. Or that's my thesis. Maccy is just as absent from the AmEng nickname inventory as Macca or Mackers would be. Or Jezza or Gazza or Shazza. Yet another instance of how non-Americans talk funny. (Or, put a different way, how even as informal-seeming a subset of the lexicon as nicknames/hypocorisms turns out to be rulebound in ways that vary from variety to variety.)

  31. Brett said,

    July 29, 2013 @ 3:24 pm

    @J. W. Brewer: Your post reminded me that McDonald's actually tried to promote the nickname "Mac D" in its advertising for a while. (Remember the burger that was served with the hot side hot and the cool side cool?)

    My American ear also doesn't find anything foreign-sounding about "Trevor," at least as a first name. It's not as common in American as Britain, but it's not rare (like Nigel). I had three elementary school classmates named Trevor (although not all at the same time); the one I remained close friends with through high school came from a family that was almost stereotypically middle American.

  32. Levantine said,

    July 29, 2013 @ 3:30 pm

    But Pflaumbaum, if I try to say 'roses' or 'Alice' with a schwa, that's precisely what it feels like — trying! It sounds entirely artificial to me, as if I were faking an Aussie accent. By contrast, I can pronounce 'lettuce' either with a schwa or as 'lettis', and both sound fine (the latter is posher to my ears). I don't see why it's so unreasonable to maintain that certain accents invariably differentiate the vowel sounds of such pairs as Rosa's/roses and Dallas/Alice.

  33. J.W. Brewer said,

    July 29, 2013 @ 3:33 pm

    The Howard Johnson's chain of hotels/restaurants is now perhaps largely defunct, but the nickname "Hojo/HoJo/Hojo's" etc. seems to have been extant at least as far back as 1965 (and maybe that could be antedated with another few minutes playing with google books). The standard AmEng pronunciation of "Hojo" (rhymes with "mojo", hence the blues-afficionado injoke of the 1972 NRBQ song title "Howard Johnson's Got His Ho-Jo Working") does not bear any relationship to the standard AmEng pronunciations of "Howard" and "Johnson."

  34. Vardibidian said,

    July 29, 2013 @ 3:34 pm

    Curious about an experiment with the mc/mac/mick–if I asked my children to elongate the syllables, I'm pretty sure they would choose 'mick' rather than 'mack'. That's how they whine the word, anyway. Does this represent them thinking of the syllable as 'mick', even though they actually pronounce it with a schwa normally? I can't imagine an American accent whining the word with a 'mack' sound, but I can sorta imagine an English accent doing so.


  35. Brett said,

    July 29, 2013 @ 3:34 pm

    @J. W. Brewer: Also, regarding the impossibility of "Gazza" in American English: There is a business by that name near where I live. I don't know what goods or services they provide. I only remember the place's existence because its name is so odd—an utterly impossible nickname in my dialect.

  36. Levantine said,

    July 29, 2013 @ 3:40 pm

    J.W. Brewer, clearly the existence of 'Mickey D's' is not dependent on all or most Americans' pronouncing 'Mc' as 'mick', but I don't think it's unfair to suggest that the apparent frequency of this pronunciation helped the nickname to gain ground. In the UK, by contrast, the American nickname was all but untenable and got changed to 'Maccy D's'.

  37. Pflaumbaum said,

    July 29, 2013 @ 3:47 pm

    @ J.W.Brewer –

    Rza and Gza might disagree with you there.

    @ Levantine –

    I don't see why it's so unreasonable to maintain that certain accents invariably differentiate the vowel sounds of such pairs as Rosa's/roses and Dallas/Alice.

    There's nothing unreasonable about the theory – it's widely accepted among phonologists! Again, I 'maintain' it too. But if we're just sitting here saying lettuce to ourselves, that's all it is – maintaining it. It's not science. Subjective impressions are just not a good way of finding out whether such things are true.

  38. Eric P Smith said,

    July 29, 2013 @ 3:52 pm

    I'd bet my house on most of Eric P Smith's contrasts evaporating in ordinary speech.

    I quite take the points about 'facultative disambiguation', and I do know that one can claim to perceive contrasts in one's own speech that have no objective basis. But I would reiterate my claim that the contrasts I mention do exist in the ordinary speech of many older SSE speakers.
    As a reductio ad absurdum, if all of the contrasts ‘elicit’ /iˈlɪsɪt/ ‘illicit’ /ɪˈlɪsɪt/ ‘insure’ /ɪnˈʃur/ ‘ensure’ /ɛnˈʃur/ ‘except’ /ɛkˈsɛpt/ ‘accept’ /akˈsɛpt/ ‘ablation’ /abˈleʃən/ ‘oblation’ /ɒbˈleʃən/ ‘Orion’ /ɒˈraɪən/ ‘O’Ryan’ /oˈraɪən/ were illusory, then we would have the unstressed vowels in ‘elicit’ and ‘O’Ryan’ matching!

  39. Levantine said,

    July 29, 2013 @ 3:55 pm

    Pflaumbaum, then what's the point in any of us contributing to these posts? I don't have a language lab in my house to monitor my speech. I can still be certain that there are particular sounds I would never produce, and a schwa in such words as 'roses' and 'Alice' is among these sounds.

  40. dw said,

    July 29, 2013 @ 5:09 pm

    I believe that someone once bet his salary that he could distinguish between Coke and Pepsi.

    I hereby volunteer for a similar study. I would be willing to bet a lot (although perhaps not my salary: I have a mortgage, wife and kids), that I can distinguish, in playback of my own connected speech, between pairs such as "affect" and "effect".

  41. J.W. Brewer said,

    July 29, 2013 @ 5:31 pm

    Dear Pflauzza, if I may be so familiar: I will confess to being only vaguely familiar with the works of the Wu-Tang Clan and affiliated individuals, but I do have a sense that they have some reputation for innovative/idiosyncratic language use. But it would be interesting if it turned out to be the case that RZA and GZA (leaving aside the orthography for the moment) were nicknames formed according to some morphological-derivation process that unbeknownst to me is reasonably regular and productive in some variety of AmEng.

    I was going to mention in the Australian-nickname thread that nicknames in -o seem rare/idiosyncratic/not-productively-formed in AmEng, but one can nonetheless find occasional examples, as witness the famous "Book 'em, Danno. Murder One" tagline from Hawaii Five-O. (I have also I think encountered at least one Stevo and one Jonno in the U.S., neither of whom appeared to be of Australian origin.)

  42. dainichi said,

    July 29, 2013 @ 6:44 pm

    If it is in fact the case that there is only one reduced unstressed vowel in some/many/all English varieties, one reason it's so hard to accept might be that some of its allophones are phonetically closer to unreduced vowels than to each other. ("The vowel in roses is close to the KIT vowel, and the vowel in lettuce is close to the CUT vowel. They cannot possibly be allophones of each other")


    "if I try to say 'roses' or 'Alice' with a schwa, that's precisely what it feels like — trying!".

    That just proves that what you perceive as a schwa (maybe the reduced vowel in 'lettuce') is not the allophone that you use in 'roses' or 'Alice', but not that the vowel that you actually use in these words is contrastive from it. Maybe it is contrastive, but not reduced, i.e. maybe it's the KIT vowel. But then it's off topic, since we're discussing reduced unstressed vowels. I.e. your dialect might reduce fewer vowels but still only have one reduced vowel phoneme.

  43. Levantine said,

    July 29, 2013 @ 7:03 pm

    dainichi, the vowel I use in 'roses' and 'Alice' is indeed the I of 'kit', or something very, very close to it (I'm not sure why this makes it off-topic in your view). It is unstressed though not reduced, and clearly distinguishable from a schwa. I believe this distinction is universal in the speech of people from the London area, whether cockney or RP.

    Regarding 'lettuce', the traditional (and posh) pronunciation is 'lettis', with a clear I. The pronunciation with a schwa, which is what I normally say, seems to be more recent, and probably came about because of the spelling.

  44. Edward said,

    July 29, 2013 @ 7:32 pm

    A paper I wrote with Stephanie Johnson a few years ago has data on the Rosa's-roses distinction in American English (it is a distinction for most speakers): Flemming & Johnson (2007) 'Rosa's roses: reduced vowels in American English. Journal of the International Phonetic Association 37:1.
    The situation in Southern British English is somewhat different (see the paper for a brief discussion), and still awaits a phonetic study as far as I know.

    [(myl) The procedure in the cited paper was essentially the same one that I used in the unpublished study of my own pronunciations of latter and ladder: a list of sentences of the form "Now say 'roses' again" or "Now say 'Rosa's' again" was read in a laboratory setting. Although the minimal pairs were randomized into different halves of the list, which is a less facultative setting than if they were adjacent, the focus of the study will have been plain to all participants.

    The study also reports a perception test, interpreted to mean that "The minimal pairs are generally easy to distinguish". But the format of the test was to present a minimal pair from a given speaker, and to ask the subject to identify the order of presentation. This is much easier in general than a standard identification task is. and yet the performance was just 88% correct (where 50% is chance performance). This certainly shows that there is some perceptually relevant information in the signal, but it leaves me suspecting that exemplars from normal usage would not be reliably identifiable.

    What other explanation might there be for such results, besides the hypothesis that these speakers have two underlyingly different reduced vowels? Well, we know that "schwa" (in the sense of an unstressed and fully reduced vowel) is raised and fronted in the context of coronal consonants. This coarticulation is normal and inevitable for all vowels, but in the case of "schwa", it's especially strong because the vowel is relatively short (and therefore the coarticulated portion is larger in proportion), and also perhaps because the vowel has little identity of its own.

    So in "Rosa", the first portion of the final vowel will be coarticulated with the medial [z], but the later portion will revert to the neutral centralized position normal for final schwa in English. Since a final vowel will be fairly long (especially in isolation or in a citation context), the centralized portion will be a strong one. No such model exists for words like "roses", of course.

    As a result, when a speaker wants to make the difference between Rosa's and roses clear to a listener, it's natural for her to emphasize the centralization that would characterize Rosa in isolation, in contrast to the full coarticulation that would apply in roses — perhaps lengthening the relevant portion of the vowels as well. That's a somewhat more elaborate explanation of what I mean by "facultative disambiguation".

    I'm not certain that's what's happening here, but I still haven't seen convincing evidence to the contrary.]

  45. Levantine said,

    July 29, 2013 @ 8:07 pm

    Thank you for the link, Edward. I just read the part about Southern British English, and I agree that pairs like Lenin/Lennon and rabbit/abbot have differently pronounced final vowel sounds. I would pronounce the first of each pair with an I and the second with a schwa.

  46. Matt said,

    July 29, 2013 @ 8:34 pm

    But it would be interesting if it turned out to be the case that RZA and GZA (leaving aside the orthography for the moment) were nicknames formed according to some morphological-derivation process that unbeknownst to me is reasonably regular and productive in some variety of AmEng.

    The standard explanation is that they are acronyms. The most commonly given expansions are "Rakeem Zig-Zag-Zig Allah" and "God Zig-Zag-Zig Allah" respectively — IIRC RZA himself offers the former in The Tao of Wu — but you also see "Ruler Zig-Zag-Zig Allah", "Genius Zig-Zag-Zig Allah", etc.

    Another explanation with a long pedigree (I remember reading it back when people got information about music in magazines!) is that they are onomatopoeic, the sounds of records being scratched on the sound of a word. "RZA" is said to be a record scratch on "Rakeem" (sometimes "Razor"), "GZA" a record scratch on "Genius". ("God" is never proposed as a source word here, presumably because its initial consonant would not match the pronunciation of "GZA.") They were using the names "Rakeem" and "Genius" before they began using "RZA" and "GZA", incidentally, so the chronology checks out.

    Note also that their given names are Robert Diggs and Gary Grice, respectively.

  47. J.W. Brewer said,

    July 29, 2013 @ 9:22 pm

    Apparently GZA is not pronounced anything like "Gazza," but oddly enough the standard Br/Aust rule will apparently get you "Gazza" from "Gary." Although I don't think you can standardly get anything in -zza from "Robert," whether or not homophonous with RZA. There is BrEng "rozzer" for "cop," the etymology of which seems to be unknown although there are various speculative hypotheses Out There on the internet. And of course one of the BrEng synonyms for "rozzer" is "bobby" . . .

  48. Matt said,

    July 29, 2013 @ 9:34 pm

    And of course one of RZA's other aliases is "Bobby Digital". The Cockney law-enforcement roots of the Wu-Tang Clan are revealed at last.

  49. Peter Taylor said,

    July 30, 2013 @ 2:07 am

    @J.W. Brewer, Gazza can come from Gary (I've known one or two) or from the surname Gascoigne (the more salient alternative, due to the fame of a certain footballer).

  50. Akito said,

    July 30, 2013 @ 2:38 am

    Re "unstressed but not reduced" vowels. As a learner of American English, I have always pronounced the complementizer that with unstressed [æ], the way I heard it spoken. When I went to England for the first time in the 1970s, I was surprised to hear [ə] for this word among many people–unstressed and reduced. Current pronunciation dictionaries prefer [ə] (unless emphatic) but don't point to any BrE/AmE differences. What do native speakers think?

  51. dw said,

    July 30, 2013 @ 3:42 am


    I would generally expect schwa in unstressed "that" in both AmE and BrE.

    However, I do have an impression that there is a certain pensive style of speaking, perhaps typical of natively AmE-speaking (but not BrE-speaking) academics, where complementizer "that" is more likely to be stressed:

    "I … would … say … th[æ]t … " (deep thought follows).

  52. Levantine said,

    July 30, 2013 @ 4:02 am

    Akito, I'm not sure how relevant this is to your question, but I've noticed that Americans tend not to reduce the final syllable of 'saucepan', which I would pronounce as a schwa. On the other hand, my boyfriend, who is American, says 'Thailand' with the same final syllable as 'England', whereas I don't reduce it (but this may just be the way he and I speak). Then there's the British pronunciations of 'mobile', 'versatile', etc. versus the reduced American ones.

    I'm not sure about transatlantic differences in the reduction of 'that', but I think my usual pronunciation is indeed of the reduced kind.

  53. Douglas Bagnall said,

    July 30, 2013 @ 5:46 am

    @New Zealand English mentioners:

    The current theory on KIT and COMMA in NZE (from my memory of e.g. Hay, Maclagan, and Gordon, "New Zealand English", Edinburgh University Press 2008) is that the two are indistinguishable. At the same time unstressed vowels are rather less likely to be reduced than in most varieties of English, which keeps the overall number of schwas under some control (I think Hay et. al. speculatively frame this as a move away from stress timing, possibly due to the influence of the mora-timed Māori language).

    Anyway, that spares us from worrying over /ɪ/ vs /ə/ for "mick", but it leaves the /æ/ in "mack" out on its own, unlikely to be reduced to the schwa. I think I would personally accept either variant for "McDonalds", but tend to use "Mick" (or as AG hears it, "Muck").

  54. Akito said,

    July 30, 2013 @ 6:44 am


    Thank you for your input. Pensive style — that explains it. All my teachers have been the pensive kind!


    Thanks for sharing those examples.

  55. Peter S. said,

    July 30, 2013 @ 7:33 am

    I am an American, and I pronounce the complementizer that /ðɪt/, and I cannot imagine my using a schwa there.

  56. Edward said,

    July 30, 2013 @ 9:39 am

    myl: I think we broadly agree. I don't think our results represent facultative contrast – based on debriefing I don't believe the subjects were aware of the minimal pairs (unlike your self study). However, Rosa's-roses is a derived contrast, not permitted within monomorphemes. The fact that it is perceptually relatively subtle may explain that fact (although my impression is that part of the difficulty in identification arises from variation between speakers), and the coarticulatory/reduction effects you allude to are the reason why 'Rosa's' and 'roses' can be realized very similarly. I also agree that Am. English dialects allow only one vowel quality in most fully unstressed contexts, but a clear exception is word-final position where diagnostics like flapping indicate that a 'schwa' vowel, short [i] and [oʊ] can occur unstressed. (Note that diagnosing secondary stressed vs. unstressed syllables is non-trivial, and many of the disagreements in the discussion appear to be disagreements about what counts as 'unstressed').
    Where we seem to differ is that I think the partial preservation of the word-final 'schwa' quality under affixation is a regular grammatical process, not an attempt at 'facultative disambiguation'.
    But on the original topic, the quality of the reduced vowel in the first syllable of 'McDonald's', I share your impression that this is the usual non-final unstressed vowel, which is highly contextually variable, being short and non-contrastive in vowel quality (more on this topic here:

  57. Ellen K. said,

    July 30, 2013 @ 10:10 am


    I tend to think of unstressed "that" in American English as being pronounced like "thet". Perhaps that could be called a partially reduced (or semi-reduced) vowel.

    Similarly, the unstressed form of "than" is homophonous with "then".

    I don't think I'd ever use æ in an unstressed syllable, but I don't think that necessarily means a word which, when stressed, has æ would always be reduced to a full schwa when unstressed. Seems to me that the possibility of something in between is something that someone studying this would have to consider as one possibility.

  58. Ellen K. said,

    July 30, 2013 @ 10:22 am

    To the English people who consider roses to have a completely different vowel sound than Rosa's in the second syllable.

    What about Rosie's? How does that vowel compare to roses? Or Rose's?

    Me (American) I have happy-tensing, which gives Rosie's a very different vowel from roses. As opposed to roses and Rosa's which, while distinct in my head, aren't so clearly distinct in speech.

  59. Levantine said,

    July 30, 2013 @ 11:43 am

    Ellen, 'Rosa's', 'roses', 'and 'Rosie's' are all different in my pronunciation — the first is with a schwa, the second with the sound of the word 'is', and the third with the sound of the word 'ease'. 'Rose's' and 'roses' sound exactly the same (as I expect that they do in all varieties of English). Also, 'Rosa's' rhymes with 'posers' in my accent.

    (By the by, not all British people from England would consider themselves English.)

  60. Levantine said,

    July 30, 2013 @ 12:13 pm

    There's a joke in the Australian show Kath & Kim involving confusion between the phrase 'baby cheeses' (meaning small circles of cheese) and 'Baby Jesus'. Glee did something similar with 'grilled cheeses' and 'grilled Jesus' (the episode in question is entitled 'Baby Cheesus'). These jokes would be much harder to pull off in a Southern British accent, where 'cheeses' and 'Jesus' do not rhyme.

    [(myl) Doesn't "Jesus" also end in a voiceless [s] rather than a voiced [z], in most varieties of English?]

  61. Levantine said,

    July 30, 2013 @ 12:23 pm

    myl, that's true, but at least the vowels match in the accents used in the two shows.

  62. J.W. Brewer said,

    July 30, 2013 @ 1:28 pm

    One can find wordplay of the form, e.g., "What a Friend We Have In Cheeses" in AmEng, although I don't think the final-syllable vowels are the same for all (most?) AmEng speakers, even on top of the /s/ v. /z/ issue. Certain sorts of imperfect rhymes are commonly treated as close enough for wordplay – perhaps someone could do a research project trying to quantify how close is close enough. Come to think of it, Life of Brian (a movie primarily made by Southern British speakers) has a bit where someone comically mishears "blessed are the peacemakers" as "blessed are the cheesemakers," despite the lack of perfect homophony.

  63. Levantine said,

    July 30, 2013 @ 1:56 pm

    A cheeses/Jesus pun is not impossible in Southern British English, just not as tenable or funny. The peacemakers/cheesemakers joke in the Life of Brian is more about the absurdity of the malapropism than it about the likelihood of it. What makes the Kath & Kim example so funny is that the character's pronunciation of 'Baby Jesus' really does sound rather close to 'baby cheeses' (see

  64. dw said,

    July 30, 2013 @ 2:54 pm

    @Ellen K:

    For some speakers, including me, Rosa's-roses-Rosie's are all distinct.

    For others, including traditional RP, there is a possible studded-studied merger, in which both the happY and rabbIt vowels can be qualitatively identified with the KIT vowel. In this case the possible merger would depend on whether the second vowel of Rosie's is taken to be the unstressed happY vowel or or the FLEECE vowel with secondary stress.

  65. Jongseong Park said,

    July 30, 2013 @ 5:56 pm

    I'm somewhat surprised by the scepticism regarding the existence of distinct weak vowels in pairs such as Alice and Dallas. But maybe I shouldn't be.

    Here is my conclusion: In some varieties such as Australian English these are by and large completely merged, and in American English there is potential for a complete merger as well.

    If we write the central weak vowel as /ə/ and the higher and fronter one as /ɨ/, then this means that in Australian English both /ə/ and /ɨ/ are realized consistently as [ə], so that Alice and Dallas always rhyme. In American English, meanwhile, /ɨ/ can almost always be potentially realized as [ə]. Some words with diaphonemic /ɨ/ may even be always realized with [ə] in American English. For most words with /ɨ/ though, there will be essentially free variation with different likelihoods of this surfacing as [ə] or [ɪ] based on several factors. It seems that being in proximity to velar consonants increases the probability of /ɨ/ surfacing as something close to [ɪ] even in American English: Alex, excite, legged. (That might be a factor in the reports of [ɪ] in Mac-, whose vowel is not normally considered to be /ɨ/.)

    British English shares this variability for large classes of words, though with less of a tendency to realize /ɨ/ as [ə]. However, at least with Received Pronunciation, there are some words that have /ɨ/ in other varieties that do not have the possibility of surfacing as [ə]. Alice, for example, cannot rhyme with Dallas in RP. The Longman Pronunciation Dictionary uses a special symbol next to the pronunciation of Alice with [ə] to mark it as an 'educated non-RP' form. Likewise for Kevin and Martin, which only uses [ɪ] in RP though in other varieties they are most often realized with syllabic nasals, a reflex of /ə/. Same for words like axis and genesis. One (but by no means exclusive) way to analyze this would be to say that words like Alice, Kevin, Martin, axis, and genesis have underlying /ɪ/ in RP but /ɨ/ in other varieties.

    If you're coming from an American English background and you have noticed that /ə/ and /ɨ/ are never reliably distinguished in your own speech, then you might assume that is the case for other varieties of English as well. But a variety like RP maintains this distinction more readily and there are even lexical items where the distinction is obligatory.

    For a typical RP speaker, if I'm not mistaken, Alice always has [ɪ]; surface and roses usually have [ɪ] but rarely may have [ə]; actress may be split equally between [ɪ] and [ə] (if not [e] in strong form); solace may almost always be [ə] although sometimes [ɪ]; and Rosa's and Dallas only ever have [ə]. One could come up with several ways to account for this variation, but to suppose that there is just one underlying unspecified vowel at work would require some explaining.

    It is understandable however that an American English speaker who has [ə] for the majority of these words may not think that more than one vowel category is necessary.

    Not much to do with what I wrote, but a possibly helpful link just for some background: John Wells's phonetic blog: weak ɪ

  66. Eric P Smith said,

    July 30, 2013 @ 6:30 pm

    @myl: Here in the UK, I hear /ʤiːzəz/ more and more.

  67. Michael said,

    July 30, 2013 @ 7:20 pm

    Do the English people here pronounce the unstressed vowels of Erin and Aaron differently? If so, I'm amazed. I pronounced both the stressed and unstressed vowels of those the same.

    @ Akito: I (an American) have noticed that too. The English reduced vowel in that often sounds more centralized to my ear, i.e., more like a "proper schwa." To me it sounds almost like [ʊ] in some cases. It's also my impression that English people use the reduced pronunciation more than I do. Then there's the glottal stop that often occurs at the end of that word, but that's another story. I haven't studied this though, I'm only an amateur linguist.

  68. Bob Wright said,

    July 30, 2013 @ 7:44 pm

    I am US English speaker, born and raised in Detroit in the 50s. My parents were raised in Missouri and Tennessee, although I never noticed a Southern accent. The point: grandma McKnight was called "MOMma MACK-night" (slight pause between syllables), but if saying the last name alone, it was MICKnight. I never was aware of the difference until reading this post.

  69. mimhoff said,

    July 31, 2013 @ 12:33 am

    There may be differences in the name of the restaurant "McDonald's", but is there as much confusion over the name of their signature product, the "Big Mac"?

  70. Sarah said,

    July 31, 2013 @ 1:51 am

    East Midlands English – Rosa's and roses both pronounced the same, with a schwa. Neither Rosie's nor Alice would ever be pronounced with a schwa.

  71. Ellen K. said,

    July 31, 2013 @ 6:44 am

    East Midlands English – Rosa's and roses both pronounced the same, with a schwa. Neither Rosie's nor Alice would ever be pronounced with a schwa.

    Sarah, I wasn't suggesting Rosie's is ever pronounced with a schwa. I think we can all agree it isn't. But in some dialects, words like happy have a lax final vowel, similar to, if not the same sound as, the vowel that some have said they have in the second syllable of roses. My question related to that, with the assumption that Rosie and happy have the same final vowel in dialect with the lax final vowel in happy. Which I now understand may not be a correct assumption.

  72. Levantine said,

    July 31, 2013 @ 7:14 am

    Michael, I pronounce 'Erin' with the I of 'kit' and 'Aaron' with a schwa. Moreover, the opening syllables are entirely differently — the E of 'get' versus the A of 'arrow'.

    And again, it may be safer not to call the speakers in question 'English'. Many Britons who belong to ethnic minorities do not refer to themselves as 'English', even if they were born and brought up in England. To me, 'English' denotes someone of more-or-less Anglo-Saxon ethnicity (a bit like WASP in the States). I am much more comfortable identifying myself as a British person from London.

  73. dw said,

    July 31, 2013 @ 10:17 am

    Do the English people here pronounce the unstressed vowels of Erin and Aaron differently? If so, I'm amazed.

    Yup — both vowels are completely distinct. When I first moved to the US, I was "amazed" that two relatively common (in the US) personal names could be homophonous for so many Americans (in Britain, at least when I lived there, "Aaron" was uncommon and "Erin" virtually unknown).

  74. SK said,

    July 31, 2013 @ 11:26 am

    Like Levantine and dw, I pronounce Erin and Aaron totally differently. But unlike Levantine, my default pronunciation of Aaron uses the vowel of 'aeroplane'/'Mary', not 'arrow'/'marry'. Of course for AmE speakers with the marry-merry-Mary merger, those would come out sounding the same anyway. This makes me wonder: are there any BrE speakers out there who pronounce Aaron with the 'merry' vowel, to complete the set? I suspect not, but you never know.

  75. Levantine said,

    July 31, 2013 @ 11:38 am

    SK, I've definitely heard other Brits pronounce 'Aaron' the way you do. I'm not sure how I ended up adopting the other pronunciation. But if I met someone who introduced himself as Aaron using your pronunciation, I would follow suit and say it the same way when talking to or about him.

  76. Joe said,

    July 31, 2013 @ 12:16 pm

    Taking this back where it started ("I was surprised to learn recently that Americans sound the first syllable of 'McDonald's' as 'Mick', whereas we Brits pronounce it with a schwa or as 'Mac'. Hence the American hypocorism 'Mickey D's', and the British one 'Maccy D's'.")

    Where in the UK does anyone refer to McDonald's as "Maccy D's"?

    I have only ever heard it referred to as "McDonald's", "McD's", and "Mickey D's"

  77. Levantine said,

    July 31, 2013 @ 12:22 pm

    Joe, 'Maccy D's' is fairly common in North London, where I grew up. If you Google it, you'll find that McDonald's even tried to trademark it, as I believe they did 'Mickey D's' in America. I've never heard 'Mickey D's' on this side of the pond.

  78. Joe said,

    July 31, 2013 @ 12:22 pm

    re: "Many Britons who belong to ethnic minorities do not refer to themselves as 'English'"

    I take it you also mean those of us in the "ethnic minorities" living north of Hadrian's Wall, west of Offa's Dyke and across the North Channel / Sruth na Maoile :-) ?

  79. Levantine said,

    July 31, 2013 @ 12:35 pm

    Joe, that's another matter, though I take your point. I was referring to ethnic minorities in England because the commenter was addressing 'English people'. I don't think he was using 'English' to mean 'British', but rather those speakers of English from England. I am one such speaker, though, like many other members of ethnic minorities here in England, I would call myself British instead of English. I've always been curious to know how ethnic minorities in Scotland, Wales, and Northern Ireland refer to themselves.

  80. Joe said,

    July 31, 2013 @ 2:51 pm

    Sorry – off-topic reply to Levantine

    The impression (largely true I hope, though there will always be the bigots who think / act otherwise) that the Scottish Government like to promote is that of an inclusive nation, with the (currently in power) Scottish National Party positioning themselves as the party of all the people who live in Scotland, whatever their place of birth, ethnicity, colour, religion, whatever.

    As a white Briton (born in England, but living in Scotland) I think of my immediate neighbours (in the suburbs of Edinburgh) as Pakistani Scots, Irish and Chinese Scots respectively. I'm not sure how they think of themselves but I suspect the older generations (born in Pakistan / Northern Ireland / Hong Kong) more likely to think of themselves as Pakistani / British / Chinese and their children and grandchildren (who have grown up in Scotland, most of them having been born here) more likely to think of themselves simply as Scottish, but for all I know they maybe all think of themselves as British too.

  81. Joe said,

    July 31, 2013 @ 3:15 pm


    "Mickey D's" -v- "Maccy D's"

    My family (Edinburgh) use "McD's", my brother's family (West Midlands) use "Mickey D's" – maybe we are linguistic outliers – I must try to listen out for what other people say.

  82. Jongseong Park said,

    July 31, 2013 @ 5:19 pm

    According to the Longman Pronunciation Dictionary, in British English, Aaron traditionally is ˈɛər ən (i.e. with the vowel of Mary) and still is usually so for the biblical character, but the personal name may nowadays be either ˈɛər ən or ˈær ən (i.e. with the vowel of marry).

    Erin with a schwa is marked as a non-RP variant, meaning that it always ends in -ɪn in Received Pronunciation.

  83. Levantine said,

    July 31, 2013 @ 5:37 pm

    Joe, thanks for your thoughts. The situation up there is no doubt complicated by the separatist movement. After all, any member of an ethnic minority who lives in Scotland and favours Scottish independence would hardly want to call him- or herself British.

    Regarding the various nicknames of McDonald's, it may well be the case that 'Maccy D's' is commoner here in the South. I discovered a few years ago that 'Hennes' as a way of referring to the clothing store H&M is a London shibboleth.

  84. Eorrfu said,

    August 1, 2013 @ 2:58 am

    If asked to repeat for clarification my Merry unmerges from Marry/Mary. I also think I'm weird that Erin/Aaron are not merged for me. Also, every Rosa I have known was Hispanic so I pronounce it with an /a/ which I picked up from my bilingual mother.

  85. Levantine said,

    August 1, 2013 @ 6:22 am

    My ethnic background is Turkish, and it just occurred to me that schwa (or something very close to it) occurs in Turkish as an actual letter — the undotted I. Vowels in Turkish are paired into A/E, I/İ, O/Ö, and U/Ü; the second of each pair is considered the 'front' version of its 'back' counterpart. The pronunciation of the vowels is easy enough for English speakers to deduce from their spelling, except for the undotted I, which is basically a schwa (the dotted I is the vowel of 'kit'). In no way is this sound conceived of as reduced in Turkish — it is a distinct letter that must be clearly pronounced for the word to make sense.

  86. Pflaumbaum said,

    August 1, 2013 @ 8:32 am

    The undotted 'ı' of Turkish is very similar to the sound traditionally recorded (and the sound that I hear) as the BrE 'roses' vowel – i.e. /ɨ/ (not /ɪ/, which is the English (but not generally the Scottish!) KIT vowel). You can hear it at Wikipedia here. The Turkish one is somewhat further back, and so less similar to schwa than the English version – certainly not 'basically a schwa'. Or to put it another way, if you consider the Turkish one a schwa then you really ought to agree with MYL's point!

    The only language I know that has both /ɨ/ and /ə/ as phonemes (apart from arguably British English) is Romanian. As you can see if you look at the link I put up earlier, their ranges do overlap across speakers, but they're certainly distinct in the modern language, with at least one minimal pair. Some people actually think the Romanian /ɨ/ came from Turkish, though it also exists in Slavic languages whose speakers invaded first. But the period of Turkish domination may have helped it free itself from /ə/ as a separate phoneme…

  87. Levantine said,

    August 1, 2013 @ 8:57 am

    The undotted Turkish I sounds to me like the I of 'cousin', which I pronounce as a schwa. I am a fluent speaker of both Turkish and English, and I'm not the only one who considers undotted I to be very close to schwa (as a quick search on Google Books will reveal). My pronunciation of 'roses' is with a sound equivalent to the Turkish dotted I, which is to say the I of 'kit'. I'm not sure why certain commenters here are so reluctant to believe that I don't rhyme 'roses' with 'Rosa's', particularly when the distinction is well attested in other speakers.

  88. Pflaumbaum said,

    August 1, 2013 @ 4:53 pm

    Well the whole point of MYL's post is that he doubts people like you (and I) have a phonemic contrast between the two, despite our subjective experience. That's why people aren't all just accepting your position.

    As for me, I may have a similar accent to yours, as I'm also 'ethnic' North London (brought up in Haringey, mid-thirties). As I've said, I also clearly 'hear' the difference, though I'm open to the possibility of being wrong.

    Your proficiency in Turkish isn't really relevant. Most native speakers of languages are unaware of the phonemic and allophonic structure of their language. The fact that you hear a Turkish close backish vowel as a mid central vowel makes me suspect you're not immune from this. No offence meant.

  89. Levantine said,

    August 1, 2013 @ 5:48 pm

    Yes, but I don't think MYL is arguing that no accent would ever distinguish between such pairs as Alice/Dallas and Rosa's/roses. After all, that's why we speak of splits and mergers. You said above that /ɨ/ is 'the sound traditionally recorded . . . as the BrE "roses" vowel'. That is not true. It is the sound recorded for certain accents of British English, including that of Birmingham, and not for others such as RP and cockney. I happen to speak an accent that uses an unstressed (though unreduced) I sound in 'roses'. There are many of us out there, and it's very well attested indeed. What's the issue exactly?

    As for the undotted I of Turkish, my pronunciation of English 'cousin' and Turkish 'kazın' (second-person plural imperative of the verb 'dig') are so close that they could be mistaken for each other. I never claimed the undotted I was identical with schwa; I merely noted with interest that Turkish makes an actual letter out of a sound that would be considered indistinct or reduced in English.

  90. Pflaumbaum said,

    August 1, 2013 @ 6:27 pm

    RP is normally transcribed phonetically with /ɨ/. It's true John Wells used /ɪ/, but that was in broad transcription. And I'm familiar with most major London accents – old RP, 'Standard British' (modern RP), Cockney, various shades of working-class non-Cockney, MLE, 'Estuary'… But I've never heard one where the roses vowel wasn't reduced. I'm not sure it's unreduced in any English accent for that matter, except perhaps some African ones that don't seem to reduce much at all.

    'What is the issue exactly' – once again, the only issue is that you're begging the question. 'It's very well attested indeed' – then maybe you can address MYL's post by saying where a phonemic split has been shown experimentally. But I think we've come as far as we can.

  91. Levantine said,

    August 1, 2013 @ 6:50 pm

    I haven't encountered /ɨ/ in transcriptions of RP (I'm not saying they don't exist). The British Library 'Case Studies' page for RP ( uses /ɪ/, and if you scroll down to 'horses', you will the hear the sound that I produce when pluralising 'rose'. To my ears, it is nothing like /ɨ/ as pronounced in the Wikipedia article you cited earlier.

    You keep invoking MYL, as if I'm trying to disagree with him. I'm not; I was merely responding to an American commenter who brought up the 'Rosa's roses' example. After reading the comment, I said the phrase to myself without any preconceived notion of the what the result would be, and I instantly discerned that the two words sounded very different in my accent. You keep saying that I may well be imagining the difference. I'm certain that I'm not. Beyond citing other evidence out there that my pronunciation does indeed exist, there isn't much I can do.

  92. Pflaumbaum said,

    August 1, 2013 @ 7:03 pm

    Fair enough. When I say them aloud they sound different too. But that's the inherent problem with pulling words out of natural speech and saying them aloud to yourself. That will also get me a vowel that sounds like KIT, but in fluent speech it is reduced (as I assure you – for what it's worth! – is yours) and centralised. That wiki version sounds different because it's not only stressed but extended.

  93. Levantine said,

    August 1, 2013 @ 7:24 pm

    I am willing to believe that it becomes reduced, but not into anything resembling a /ɨ/ — at least not in my speech. If any reduction is going on, it is minimal, perhaps to the point where the difference between the resultant sound and the I of 'kit' is inaudible to most listeners. I can't hear it, though i can clearly hear the difference between how I would say 'roses' and how a Brummie would.

  94. Levantine said,

    August 1, 2013 @ 7:37 pm

    And to address your point about natural speech versus saying things aloud to yourself, the sound in question seems unreduced to me even in audio clips like those on the BL website I just mentioned. These clips are of respondents speaking naturally as they answer someone's questions. Perhaps my ears aren't as phonetically attuned as yours are.

  95. Jongseong Park said,

    August 2, 2013 @ 6:27 am

    To make things clear, the symbol /ɨ/ at least as I have been using it (and is also the convention on Wikipedia) is to represent the reduced vowel that may be realized as either [ɪ] or [ə]. But this is just a useful shorthand to represent variable pronunciation, and it doesn't necessarily mean that the vowel represented by /ɨ/ is actually pronounced as the cardinal close central unrounded [ɨ] in IPA. Reduced [ɪ] does have a wider range of realization than an unreduced [ɪ] of course and may approach something that could be represented by [ɨ], but that's not what is meant by our use of the symbol /ɨ/.

    Levantine: The undotted Turkish I sounds to me like the I of 'cousin', which I pronounce as a schwa.
    I am a native speaker of a language (Korean) that uses the close back unrounded vowel /ɯ/, which is also the pronunciation of ı in Turkish. I pronounce 'cousin' with a syllabic nasal, as I think most English speakers do. Syllabic nasals do sound like they 'carry' the vowel /ɯ/ to me, probably because of the similarity in tongue position—phonetically [ɯn] and [n̩] are very similar, with only a tiny difference in timing. An actual schwa in 'cousin' would sound different, and while it won't be 'wrong', it would sound slightly unnatural to most speakers.

    There are environments where an actual schwa (not a syllabic consonant) becomes closer and backer and somewhat approaches /ɯ/ (e.g. 'contain'), but this is a separate issue—typically for a reduced vowel, the schwa has a very broad range of realizations, including the very open [ɐ~ʌ] you find in final positions in words like 'comma'. You will note that the schwa in 'comma' sounds nothing like the Turkish ı.

    Pflaumbaum: The only language I know that has both /ɨ/ and /ə/ as phonemes (apart from arguably British English) is Romanian.
    Korean has /ɯ/, which was traditionally the more central [ɨ], and /ʌ/, whose long version was traditionally [əː]. So we're close—even in contemporary pronunciation, [ɨ] and [ə] are acceptable variants of /ɯ/ and /ʌ/. The Southeastern dialect merges these two phonemes, incidentally.

  96. Levantine said,

    August 2, 2013 @ 6:58 am

    Jongseong Park, thank you for your comments. You're right: the I of 'cousin', which I thought of as a schwa and which sounds to me like the undotted Turkish I, is indeed quite different from the schwa of 'comma'/'Rosa'. I suppose that bears on the question of whether there are distinguishable kinds of reduced vowels in English. However, that's not really a question I've addressed in my comments. I've limited myself to the observation that the vowel I and other speakers of similiar accents produce in the second syllable of 'roses' sounds to me like the unreduced vowel of 'kit' (I think you noted something to the same effect in an earlier comment discussing RP speakers). Perhaps, as Pflaumbaum is arguing, this vowel is invariably reduced in the context of 'roses', but if this is so, the effect on my ear is absolutely minimal, and the sound is nothing like the /ɨ/ as Pflaumbaum was defining it (since s/he said it sounded like the undotted Turkish I, s/he was clearly assigning to it a particular phonetic value, which is different from how you're using it).

  97. Jongseong Park said,

    August 2, 2013 @ 7:49 am

    It would be nice if I could post pictures of the vowel charts given in the Longman Pronunciation Dictionary. Rather than locating vowels as points on the vowel charts, you have clouds (like the electron clouds of atoms) representing the possible range in realizations.

    In the vowel charts for Received Pronunciation, the strong /ɪ/ and reduced /ɪ/ overlap, hence the same symbol for the two in the dictionary; however, the reduced /ɪ/ is represented by a bigger cloud. The cloud for /ə/ is still bigger. Yet the clouds for reduced /ɪ/ and /ə/ are clearly distinct in the chart.

  98. Matt McIrvin said,

    August 2, 2013 @ 12:52 pm

    Now that people mention it, I vaguely remember first seeing "Mickey D's" in TV ads for McDonald's that were clearly targeted at black American audiences. (I was unfamiliar with it at that point, though it was immediately clear what it meant.) If it's an AAVE-ism, presumably they got it from existing usage.

    I think I've only heard "Macca" as an affectionate nickname for Paul McCartney. If I spent more time in the UK that would probably change.

  99. J G Stringer said,

    August 3, 2013 @ 8:04 pm

    True fact: I used to know a lady here in Australia by the name of Alice Dallas.

  100. dw said,

    August 4, 2013 @ 7:20 am

    @J G Stringer

    That name would certainly rhyme in Australia.

  101. J.W. Brewer said,

    August 5, 2013 @ 2:48 pm

    The Dallas/Alice rhyme is found in canonical American literature of the second half of the 20th century:

    "I been warped by the rain, driven by the snow
    I'm drunk and dirty don't ya know, and I'm still, willin'
    Out on the road late at night,
    Seen my pretty Alice in every head light
    Alice, Dallas Alice"

    — L. George (1945-1979). In terms of regional accent, he himself was born and raised in Los Angeles, and wrote this song from the first-person POV of a truck driver working in the southwestern quadrant of the U.S. ("Tucson to Tucumcari, Tehachapi to Tonopah"). So a lack of merger found in RP speakers would not be salient to either writer or character.

  102. m.m. said,

    August 11, 2013 @ 3:05 pm

    i never noticed the different vowels, until i became really exposed to southern hemisphere englishes, where most if not all aussies [with an [s] not [z] here] and kiwis ive heard are categorically schwa users, which always gets my attention when they dont use a schwi where i do, including mickdonalds. the americas and english englishes tend to be mixed

RSS feed for comments on this post