The Elusive English Schwa

Schwa symbolAn old trick question:  what is the most common vowel in spoken English? Is it the a in cat? The o in top? The ee in keep? In fact, the answer is the puzzling little sound known as schwa.

Schwa was first described to me as “English’s throwaway vowel,” a crude but apt definition. The term refers to the little vowel pronounced in the center of the mouth which you hear at the end of comma, the beginning of afraid, and the middle of salamander. If you speak a non-rhotic accent, as would be the case for many Britons, schwa is also a common realization of “er” at the end of better.  This English learning video offers a simple introduction to schwa (even if you aren’t an English learner):

As indicated, schwa is traditionally represented by the IPA symbol ə (an upside-down “e”), signifying a vowel pronounced smack dab in the middle of the vowel space. As any English phonetician will tell you, however, this a tad misleading as far as English is concerned. Our particular brand of schwa in fact represents a number of vowels which differ depending on the words they appear in.

To use a well-worn example, Americans make a slight distinction between the “schwa” at the end of Rosa’s and the “schwa” at the end of roses. The vowel in the former is pronounced relatively close to the “classic” schwa sound (i.e. right in the center of the mouth).  The e in “Roses,” however, is pronounced much closer (or “higher”), with a vowel more or less equidistant between the “ee” in feet and the “oo” in goose (i.e. IPA [ɨ]).

And there are “alternate” schwas even less apparent to the naked ear. For example, this 2007 study finds that the schwa in the first syllable of begin is closer to the “i” in American kit, while the schwa in the second syllable of probable is closer to the “oo” in the word foot*. English “schwa” is perhaps a blanket term describing any number of unstressed vowels, even if these actually vary in pronunciation.

Why is it we don’t notice these variations?  To look at one hypothetical analogy, if the “e” sound in dress were occasionally pronounced with the “a” sound in father, such a difference would be quite obvious.  But English schwa is a very, very short vowel.  To give a frame of reference, this study on the English of Drogheda (a city on the East Coast of Ireland) found that schwa was nearly 1/2 the duration of the next shortest vowel (the “i” in kit). This vowel is often pronounced so quickly that it takes the sharpest of ears to notice its precise phonetic properties.

Of course, many foreign languages have no schwa-like vowel, the lack of which results in a disproportionate share of pronunciation difficulties for English learners.  The standard schwa sound is not particularly difficult to make (just grunt and you’ve basically got it), but understanding how to use schwa in English is hardly intuitive.  If I were Spanish, for example, I’d probably have a hard time associating the “o” in renovate with such an “un-o-like” sound.

Given the topic of this site, I’ll conclude with something I’m curious about with regards to schwa:  how does it differ from accent to accent?  The 2007 study I linked to above indicates some slight differences within American English (for example, Roses and Rosa’s are actually merged for some Americans).  What other ways does schwa vary from one accent to another?

*Note to the phonetically inclined:  I’m extrapolating here based on their statement about the average F2 of this vowel.  Looking at the chart they provide, I’d say the schwa in “probable” would be more accurately rendered [ɤ].

Share

About Ben Trawick-Smith

Ben Trawick-Smith launched his dialect fascination while working in theatre. He has worked as an actor, playwright, director, critic and dialect coach. Other passions include linguistics, urban development, philosophy and film.
This entry was posted in Miscellaneous Accents and Dialects and tagged , . Bookmark the permalink.

39 Responses to The Elusive English Schwa

  1. Sarang says:

    Isn’t there a pretty pervasive difference between (say) RP and its relatives, in which the short i in “begin” and “houses” is treated — by most dictionaries — as distinct from the schwa, and Am. E. in which it usually isn’t?

    • trawicks says:

      Yes, although much of what I’ve read suggests more contemporary varieties of RP make less of this distinction. In older varieties of RP, the vowel in “housEs” was essentially part of the KIT set (i.e. the word is realized as [haʊzɪz]). This was especially salient in U-RP, because the KIT vowel was slightly closer to [i] than in contemporary RP, hence you may hear U-RP speakers in old films say something that sounds like “houz-eez.”

  2. Andrej Bjelaković says:

    Do you think that there’s grounds to call the vowel in American ‘begin’ and ‘roses’ a centralized version of KIT, rather than a closer version of schwa.

    Also worth mentioning that in the antipodean varieties, unstressed KIT and schwa are completely merged, so that ‘Lenin’ sounds exactly like ‘Lennon’, ‘rabbit’ and ‘abbot’ rhyme etc.

    • trawicks says:

      I think there is grounds for that. I’d be surprised if there weren’t some very heated debates about just that topic!

      I might argue that, in American English, the “schwa” in “roses” is an allophone of KIT while the “schwa” in “rosa’s” is an allophone of STRUT. When I slow down both words, as if I were trying to communicate to someone with very bad hearing, “rosAs” is clearly assigned /ʌ/, while “roses” gets /ɪ/. Schwa’s phonemic status, of course, has been debated quite a bit throughout the years.

    • Egad says:

      > ‘Lenin’ sounds exactly like ‘Lennon’, ‘rabbit’ and ‘abbot’

      Both of those, or something very close, can be heard in the US. Don’t know how regional it is there, but I’ve heard Lenin/Lennon inside the DC Beltway and rabbit/abbot in South Texas.

      • Austin says:

        In my experience, in merged accents in the US, both Lenin and Lennon sound like the former in unmerged accents, whereas elsewhere, they both sound like the latter.

  3. dw says:

    I’m surprised that you say most Americans contrast Rosa’s and roses. Although I (RP-ish) do make such contrasts, my understanding is that most do not. The contrast is certainly not shown in most American dictionaries.

    There is pretty strong evidence for this so-called weak vowel merger in many song lyrics. To take one example that Wells mentions, there is a song from the “Wizard of Oz” that rhymes “dinosaurus” with “fores(t)”. Those two words wouldn’t rhyme in either syllable for me. I’ve noticed hundreds of others. Just one more: Stephen Pinker, in “The Language Instinct”, says something about “this rip” being potentially homophonous with “the srip” if English didn’t prohibit initial /sr/ clusters, but it still wouldn’t sound the same for me.

    • trawicks says:

      I make a very slight distinction between Rosa’s and roses, although it certainly is nothing like the traditional RP treatment of those two vowels. The split may be conditioned by region, age or various other factors (that article I linked to mentions that, for some reason, the Pacific Northwest seems to lack the split).

      In my own experience, I’d say there are at least traces of the split in Gen Am, as accents with advanced weak vowel merging (as in Australian and Irish English), always stick out for me as different from my own in this regard.

      • Austin says:

        I’m an American who makes the distinction too. For me, the “schwa” in “roses” sounds like an allophone of KIT and the “schwa” in “Rosa’s” sounds like an allophone of STRUT, just as you said.

        It’s interesting how you mention “houz-eez” for houses, with a closer variety of KIT in the second syllable. I actually associate this with Cockney accents in England. But that doesn’t surprise me, as I’ve noticed a number of features that Cockney and U-RP have in common, e.g., Smoothing in words like fire and power, closer variants of TRAP, DRESS and KIT and r-tapping. Maybe it’s just because they’re both old, (mostly) southern English accents. In America, I associate the “houz-eez” pronunciation with older, rural Midland accents and Southern accents. I’ve heard some Midlanders say damage in a way that to me sounds like “dam-eej”. It could actually be [ɨ] they’re using, which to my ears sounds very [i]-ish.

    • Ellen K says:

      Regarding rhymes, songs and poems do sometimes use near rhymes instead of exact rhymes. So, two words rhyming doesn’t mean the vowel is the same, just that it’s close.

      • dw says:

        Fair enough.

        But I personally find all these rhymes extremely unsatisfying: indeed I would say that, for me, they don’t rhyme at all.

        Obviously other people do find that they rhyme. So there is at least a difference of degree in the vowel distinctions.

  4. A2dez says:

    Roger Lass has a theoretical article on the SEVEN kinds of schwa in his, I believe, New York accent. It’s in the same volume as Flemming’s article, although I haven’t been able to get my hands on it yet.
    http://us.macmillan.com/phonologicalweaknessinenglish
    In my (southside) Dublin accent, I recognise distinctions between belGIUM, belGIAN and belTON (when I yell them) and these correspond to Flemming’s observations about having a three-way micro-system of distinct schwa vowels: one mid, and two high (one front, one back).
    I’ll get back to you about northside schwa as soon as I can

    • trawicks says:

      Interesting! The study I link to above basically makes the same observation: there’s a mid schwa, and then two high variants (front and back). Although it seems there are a few allophones in between as well.

      It’s also worth noting that similar sets of allophones have been found in some other languages with schwa (for example Dutch).

    • Dw says:

      Are these separate phonemes, or just allophones?

  5. Luke Pederson says:

    The only schwa for me is the schwa in Chutzpah.

    Or the schwa in the Schwartz.
    May the Schwartz be with ya – I mean, with thee!
    :)

  6. m.m. says:

    I remember reading about the weak vowel merger with [ɨ] schwi and [ə] schwa, which I then learned I was [mostly] not merged, and that some people merged towards [ɨ].
    I wouldn’t call it a ‘slight’ distinction though. Distinctions are very audible to the non-merged, as you say, they can sound almost like they would belong to /ɪ/ and /ʌ/. I imagine that’s how non cot-caught merged people hear the low back vowels, as belonging to separate/distinct sounds.

    hehe, [ʊ̈] schwu

  7. boynamedsue says:

    The pronunciation of houses would be “haʊzɪz” in most northern English varieties (except North East, where it’s “haʊzəz”).

    There’s a definite difference between northern and southern/estuary varieties in the value of the final schwa in words like “brother”. The southerners pronounce it forward than RP, northerners further back. I’ve seen this represented in writing in the comic Viz (written by Geordies, whose schwah sounds very close to RP to me), the characters Paul Whicker the Thin Vicar, and Cockney Wanker say “muvva(h)” and “teacha”, whereas yorkshireman 8 Ace says “muthuh” and “wankuh”.

    • Ed says:

      The boxES and startED vowels are subtle ones. I think that this is why they don’t vary much according to class: they vary according to region.

      As you say, they take a schwa in the north-east. They also have it in an area of Yorkshire south of Leeds (roughly Wakefield district plus South Yorkshire) and in East Anglia. These areas don’t all join up and there are areas inbetween these three zones that use the KIT vowel.

      For the record, I use schwa in boxES and startED but not in happY (last one takes schwa in broad East Anglian accents).

      • boynamedsue says:

        I lived 2 miles north of Wakefield for 18 years, and never realised this.

        Thinking about it, though our accents were very similar to Wakefield, we could tell people who “sounded Wakey”, perhaps it was this we were picking up on. When I think of my friends from the south side of Wakefield, they definitely use schwah, while I don’t, except occasionally in boxes.

        I’ve flitted, but my natural accent is very broad, and my family don’t use schwah in these words either, so I’m pretty sure I must have grown up right next to this isogloss without knowing it.

        • Ed says:

          Did you live in Outwood? That’s where my grandma lives. She definitely uses schwa.

          I’m from Ossett originally, and almost everyone in Ossett uses the schwa. Same in Dewsbury and Batley. I live in Leeds now, and everyone uses the KIT vowel. Class just doesn’t come into it.

          This is mentioned in the Yorkshire Dictionary by Arnold Kellett towards the start. He says that this feature starts south of Halifax Road in Bradford, which is avery precise line to draw.

          It’s funny how the boxES and startED vowels have patches of each variant across the country rather than there being a straight north-south or east-west split on it.

        • boynamedsue says:

          Hi, didn’t see your answer. I actually come from Ardsley, just next to Outwood on the Wakefield-Bradford road. It’s very close to Wakey but has been influenced very strongly by Morley and Leeds over the past 100 years, due to local government boundaries. I expect the residual schwah in “boxes” means we used to pronounce all of them as schwah.

  8. trawicks says:

    @m.m.

    By “slight” I just mean that the distinction carries a low functional load. That being said, I know what you’re saying about the merger being quite salient to non-merged speakers. The typical Australian pronunciation of “limited” (which usually has a schwa-like vowel for the latter two syllables) has always stood out to me for this reason.

    @boynamedsue,

    To me, the “e” in “houses” sounds especially pronounced in Northern accents. It may be because Estuary/Southeastern accents seem to have shifted to a considerably more centralized realization in the past half-century, making the Northern vowel more noticeable by comparison.

    • dw says:

      @trawicks:

      Does it carry a low functional load? I’m not so sure.

      Of course, in non-rhotic accents with the merger like those of Australia and New Zealand, there are a substantial number of homophones. Wikipedia lists arches-archers, batted-battered, chatted-chattered, founded-foundered, matted-mattered, offices-officers, sauces-saucers, splendid-splendo(u)red and tended-tendered.

      • trawicks says:

        I’d say the functional load is low for most American English, but considerably less so in non-rhotic Englishes. It’s fascinating how much rhoticity (of the American variety at least) shrinks the list of minimal pairs.

  9. Ed says:

    Dutch has a lot of schwa sounds as well. I find this makes it more difficult to learn as a foreigner, as you never know when a vowel gets reduced to a schwa.

    In German, the only letter that ever gets reduced to a schwa is E. Combined with predictable word-stress, it’s not as difficult to guess the schwas in Germans.

  10. AL says:

    I’m American and I think I pronounce the ‘e’ in roses and houses as I do the ‘i’ in his (or close to that). But I think I pronounce the ‘a’ in Rosa’s completely differently… it’s more like an “uh” sound, which is the schwa, right?

  11. Charles Sullivan says:

    It would be nice if the name for ə were pronounced “schwuh.” Then the sound itself would be in the word. I know I can’t be the first person to think that.

    • dw says:

      Actually “uh” doesn’t suggest schwa for me: it suggests the STRUT vowel which is, for me, distinct.

      Impossible to please everyone :)

      • Charles Sullivan says:

        Okay. Maybe if it were pronounced “schwə”. That should please everyone.

        • Ellen K. says:

          But it can’t be pronounced with a schwa. It’s only one syllable, and it’s a noun, and it just can’t have a schwa. A vowel articulated at the same place as a schwa, yes, but a schwa, no.

        • Charles Sullivan says:

          If the the letter B is pronounced Bee or Bay (depending on your language), and it has the sound contained in the name of the letter, then why can’t the name for ə have the same sound?

        • Ellen K. says:

          Well, Charles, it can if it’s a multi-syllable word. But a one-syllable word that’s a name for something can’t have a schwa for a vowel, because the schwa is an unstressed vowel, and any word that’s a name for something (or, more generally, any word that carries stress) has to have a stressed syllable.

  12. Miruna says:

    Interesting :D In Romanian we have a special letter that is read as the Schwa sound. It’s written like ă Ă. For example, the following words would be trascribed as
    coat = căut
    boat = băut
    window = uindău

    That’s actually how our English teacher taught us how to pronounce in English ^_^

    • Austin says:

      Yes, that’s one of the (many) things I love about Romanian actually. However, and correct me if I’m wrong on this experts, I don’t know if it would actually be considered “schwa” because it can also occur in stressed syllables, correct? That wouldn’t be any problem for them Kiwi’s though :)

      • Ben says:

        As far as I’m aware, “schwa” generally refers to a mid-central vowel but does not specify that it is unstressed. Most dialects of English do not have schwas in stressed syllables (STRUT-schwa merged dialects are the exception) but it doesn’t mean that a stressed mid-central vowel cannot also be called a schwa.

        Apparently, according to Wikipedia, “Sometimes the term “schwa” is used for any epenthetic vowel, even though different languages use different epenthetic vowels (e.g., the Navajo epenthetic vowel is [i]).” This seems silly to me. Why not just call them epenthetic vowels? (Then again, why not just call the schwa a ‘mid-central vowel’ and be done with it? I suppose it only needs a special word for it because it’s a bit different in some languages.)

  13. Pingback: This Week’s Language Blog Roundup | Wordnik ~ all the words

  14. Pingback: Jamaican Patois (And English Schwa) | Dialect Blog

  15. Pingback: "Prime Ministaw:" Jamaican Rounded Schwa | Dialect Blog

  16. I’m not certain where you’re getting your info, however good topic. I must spend some time studying more or understanding more. Thank you for excellent information I was searching for this information for my mission.