Schwa-schwi merger

[Travis B.]

Typically English is described as having separate schwa (/ə/, unstressed only if one considers /ʌ/ and /ə/ to be unified as a single phoneme) and schwi (unstressed /ɪ/, typically realized as [ɨ]) phonemes.

However, from paying attention to my own dialect, I have noticed that there are practically no unstressed environments in which I can form a clear contrast between [ə] and [ɨ] in everyday speech. In every environment either only one of the two can be found or the two are in free variation. Apparent exceptions invariably involve some degree of stress, or are word-initial in more careful speech, or are actually a matter of spelling pronunciation.

The rules determining their distribution vary from very straightforward ([ə] before /l/ or syllabification of the /l/, [ɨ] before /n/ or syllabification of the /n/) to completely non-obvious (always [ə] in /l_v/ but [ɨ] otherwise (I think) before /v/ except word-initially, where [ə] and [ɨ] are in free variation before /v/). I could work out a whole table of what is found where, but I will not do so at this moment.

So is anyone else familiar with loss of phonemic distinction between schwa and schwi in the English they are familiar with?

[Rory]

I've only seen the "schwi" in descriptions of American English. I don't believe Wells mentions it in his American section of the classic Accents of English. For what it's worth, I've never been able to figure out what the distinction is supposed to be of.

[Pabappa]

It's the "Rosa's roses" merger. If you pronounce them the same then you have the merger .

[Imralu]

Rosa's roses.
Georgia's boxes and George's boxers.
I have the merger.

I was very shocked to find that British people say <wanted> as /ˈwɒntɪd/ ... it sounds ridiculous to me, when I pay attention to it.

[ol bofosh]

Rosa's roses.
Georgia's boxes and George's boxers.

In which case I don't have the merger. Wanted, bridges, roses, boxes, etc. have schwi for me.

Words like electric and relax I usually pronounce with schwa, sometimes schwi, or even something in between. Could that be part of it? A partial merger perhaps?

[AnTeallach]

It's the "Rosa's roses" merger. If you pronounce them the same then you have the merger .

It's more complicated than that. There are British English accents (and I'm a speaker of one of them) which retain unstressed /ɪ/ in words like "rabbit", "effect" and "Lenin" (cf "abbot", "affect" and "Lennon" with schwa) but have schwa in -es and -ed endings, so that "wanted", "roses" etc. have schwa. (Do I pronounce "roses" and "Rosa's" the same? I'm not sure; I have a feeling that the schwa in "Rosa's", where it's before the morpheme boundary, is somehow "stronger" and perhaps more open, but I'm not convinced I could test that in a remotely scientific way. At least I think there's a strategy there for distinguishing "Rose's" and "Rosa's", which might be necessary in some contexts.)

In general, I think you might find that there's quite a lot of variation among speakers who retain a distinction about which words have which vowel.

[ol bofosh]

Slightly off the theme, but connected, because I discovered it through this thread: I think I have a stressed [ɨ] after /r/.

[ˈtʃɹʷˠɨŋ.kʰɨʔ] for trinket, where before I thought it was [ˈtʃɹʷˠɪŋ.kʰɨʔ]. Tree I thought was [tʃɹʷˠɪi̯] but apears to be [tʃɹʷˠɨi̯].

My vowels are really messed up (me and most of the Anglosphere, lol). Of course that's nothing compared to what happens before /l/.

[Travis B.]

It's the "Rosa's roses" merger. If you pronounce them the same then you have the merger .

For me this is not so simple. I always have [ə] morpheme-finally in stems and suffixes. (No such restriction is imposed on final vowels in prefixes, though.)

Hence I still distinguish Rosa's and roses, but at the same time this does not mean that the distribution of [ə] and [ɨ] is any less predictable IMD.

[finlay]

see I still think I have [ɐ] in Rosa, which means that I don't have this merger but for a different reason than what it's meant to test.

[Drydic]

Exactly. This is probably a solely (north?) american distinction. Bigger post inc later, I have an interview to bullshit my way through before I can finish it.

[finlay]

No, I think RP has the distinction. It's just I think I usually have [ə] for -es and -ed, and for mid-word unstressed syllables, but [ɐ] for -a.

[Travis B.]

It's the "Rosa's roses" merger. If you pronounce them the same then you have the merger .

It's more complicated than that. There are British English accents (and I'm a speaker of one of them) which retain unstressed /ɪ/ in words like "rabbit", "effect" and "Lenin" (cf "abbot", "affect" and "Lennon" with schwa) but have schwa in -es and -ed endings, so that "wanted", "roses" etc. have schwa.

For these I merge rabbit and abbot (/r/ aside), effect and affect v., and Lenin and Lennon, and have [ɨ] in all of these along with -es and -ed, except for having [ə] and [ɨ] in free variation* in effect and affect v.

* When I speak of [ə] and [ɨ] being in free variation I actually mean any phones between the two; I typically have a vowel higher than [ə] but lower than [ɨ] in such positions.

[Imralu]

Words like electric and relax I usually pronounce with schwa, sometimes schwi, or even something in between. Could that be part of it? A partial merger perhaps?

For me, there's a whole lot of words with <e> in an unstressed first syllable, such as 'delay', 'before', 'remember', that can either be pronounced with a schwa when speaking normally or /iː/ (shortened to [ i ]). So 'before' is usually [bəˈfoː] and can be [biˈfoː] but can never be [bɪˈfoː]. Sometimes it comes out a bit like [pfːoː].

There's a funny group of words which I was surprised to discover are pronounced with /ɪ/ in RP, words like 'explode'. For me, they begin with /e/ or /ə/.

see I still think I have [ɐ] in Rosa, which means that I don't have this merger but for a different reason than what it's meant to test.

I have [ɐ] in Rosa but [ə] in Rosa's. /ə/ becomes [ɐ] at the end of a word (whether its <a> or <er>), but if you add -s or -ed or anything, it's the schwa.

'Remember'
[ɹəˈmembɐ] (or occasionally [ɹiˈmembɐ] or even [ɹiˈmembœː] as ultra-emphatic)

'Remember it'
[ɹəˈmembəɹət̚]

'Remembered it'
[ɹəˈmembəɾət̚]

'Remembers', 'remembered'
[ɹəˈmembəz] [ɹəˈmembəd]

'Ted had edited it.'
[ˈtʰeɾəɾˈeɾəɾəɾət̚]

'A little bit illiterate'
[əˈlɪɾəɫˈbɪɾəˈlɪɾəɹət̚] [əˈlɪɾuˈbɪɾəˈlɪɾəɹət̚]

'Wanted'
[ˈwɔn(tʰ)əd̚]

[marconatrix]

I'm originally from the English East Midlands, which is definitely not RP, but I have this distinction pretty consistently. Funny thing is, I've never really noticed that others don't, apart from the fact that someone once pointed out to me that it was characteristic of Australian.

How do you guys pronounce "marriage, carriage, garage" etc.? For me they all have initial stress and some kind of /ı/ in the final syllable.

[Imralu]

Funny thing is, I've never really noticed that others don't

I never really noticed that others do until I started teaching English and came across some stuff that made no sense to me in pronunciation resources. I find it really hard to guess where RP has /ɪ/ a lot of the time. I pretty much just guess any unstressed <i> or <e> is /ɪ/, but then most British dictionaries list "cousin" not as /ˈkʌzɪn/ but as /ˈkʌzən/, so I'm all like WTF?! Apparently in America 'cousin' is generally with /ɪ/? And in both Britain and America, 'explode' begins with /ɪ/ but 'enlarge' begins with /ɪ/ in Britain and /e/ in America (or, rather, their equivalent: /ɛ/). Obviously these unstressed vowels have such a low importance that we don't actually hear these differences so obviously when we hear another accent.

How do you guys pronounce "marriage, carriage, garage" etc.? For me they all have initial stress and some kind of /ı/ in the final syllable.

-age is [ɪdʒ] most of the time
Marriage [ˈmæɹɪdʒ]
Carriage [ˈkʰæɹɪdʒ]
Cabbage [ˈkʰæbɪdʒ]
Luggage [ˈlägɪdʒ] [ˈbægɪdʒ]
Language [ˈlæŋgwɪdʒ]
Message [ˈmesɪdʒ]'

Cf. Knowledge [ˈnɔlɪdʒ]

But in newer borrowings from French, it's [äːʒ] (or for some people, [äːdʒ])
Garage [ˈgæɹäː(d)ʒ] [gəˈɹäː(d)ʒ]
Massage [ˈmæsäː(d)ʒ] [məˈsäː(d)ʒ]

The latter pronunciation of the above two words, with stress on the second syllable, sounds distinctly American to me. And putting 'garage' in the first group is sometimes done here to be funny, but I've only heard it seriously from Brits.

There seem to be some situations where the environment forces unstressed vowels to become [ɪ]. For example ...

Barnacle [ˈbäːnɪkəɫ]
Hydrogen [ˈhɑedɹɪdʒən]

... so /ɪ/ is not completely absent from unstressed syllables. I think it's at least partly environment dependent. For example, I can't think of any word that has /əkəl/. Monocle is, for me, [ˈmɔnɪkʰəɫ] and when I first came across the word 'coracle' I was sure it was [kʰɔɹɪkʰəɫ]

'Clerical' and 'maniacal' also have [ɪ], but a word like 'principle'/'principal', which has /p/ instead of /k/.

I think it might have something to do with /k/ and /dʒ/ because I can't think of any words that have /ək/ or /ədʒ/ other than initially ('occasion', 'adjourn'). I worked with a guy called Marek, and a friend of a friend of mine is called Marek too. The friend of a friend is Polish and everyone calls him /ˈmäːɹek/ with, maybe tertiary stress on the second syllable. My old colleague is Australian, of Polish parents, and everyone calls him /ˈmæɹɪk/, with an unstressed final syllable. Hmm, although a nasty politician who was around a while ago is called Phillip Ruddock, and I would say that as [ˈfɪləp ˈɹäɾək], but the name Derrick is [ˈdeɹɪk], so possibly there's more to it, at least in my head ...

So, possibly we do have a distinction, at least in some situations but not all, it's just that the majority [məˈdʒɔɹəɾi] of unstressed [ɪ]s have become schwas ...

I just looked up parthenogenesis in a couple of dictionaries to see what they list as the pronunciation. Decoding a rather horrible made up pronunciation system, an American dictionary gave me: /ˌpɑrθənəʊˈdʒenɪsɪs/ and in nice, shining IPA, a British dictionary gave me /ˌpɑːθɪnəʊˈdʒɛnɪsɪs/. They disagree on the second syllable. For me, it's /pä:θənəˈdʒenəsəs/.

[Nortaneous]

Where in America does 'enlarge' not have the same first vowel as 'explode'?

NE: actually I feel weird about saying 'enlarge' with anything but a syllabic n but that's a different issue

[Yng]

Yeah, I've met people with the merger in the UK, but only people with strong northern English accents (and there it was noticable). It's definitely a pretty common distinction here.

[Travis B.]

Where in America does 'enlarge' not have the same first vowel as 'explode'?

NE: actually I feel weird about saying 'enlarge' with anything but a syllabic n but that's a different issue

I have slightly different vowel qualities in enlarge and explode, specifically [ɨ] and [ë]* respectively.

* When I have spoken of something "varying between [ə] and [ɨ]" I really mean this.

[Ser]

Barnacle [ˈbäːnɪkəɫ]
Hydrogen [ˈhɑedɹɪdʒən]

Talking about dissonances between orthography and pronunciation, let us not forget all those nouns and adjectives with [ɪt] represented as "-ate": palate, vertebrate, etc. This doesn't mean that there aren't lots and lots of nouns and adjectives with [eɪt] instead, probably more likely the less common they are (in which case they would arguably be reading pronunciations, e.g. "that apostolate" as [ðæɾəˈpɑstəleɪt]). Though who knows, this dictionary gives me [ɪt] in "permanganate"... [ɪt] seems to be strongly associated with nouns and adjectives, but it's not compulsory. If a compound noun's second element in -ate actually comes from a monosyllabic word ending in [eɪt], compounds seem to retain it: "classmate" [ˈklæsmeɪt].

Verbs that end in "-ate" have [eɪt] without exceptions for sure, however. (Ha! And then they say sound changes can't be restricted to certain morphological categories...)

I think I should open a new thread about this [ɪt] - [eɪt] business... "A desolate place", "to desolate the victims"; "the French delegates", "to delegate responsibilities"...


Also: somebody in these forums once told me that "bias" is apparently pronounced [baɪɪs] (BYE-iss) in NAm English.

[----]

Also: somebody in these forums once told me that "bias" is apparently pronounced [baɪɪs] (BYE-iss) in NAm English.

I pronounce it like that but with a /z/ at the end (BYE-is). How else would you pronounce it?

[kodé]

Barnacle [ˈbäːnɪkəɫ]
Hydrogen [ˈhɑedɹɪdʒən]

Talking about dissonances between orthography and pronunciation, let us not forget all those nouns and adjectives with [ɪt] represented as "-ate": palate, vertebrate, etc. This doesn't mean that there aren't lots and lots of nouns and adjectives with [eɪt] instead, probably more likely the less common they are (in which case they would arguably be reading pronunciations, e.g. "that apostolate" as [ðæɾəˈpɑstəleɪt]). Though who knows, this dictionary gives me [ɪt] in "permanganate"... [ɪt] seems to be strongly associated with nouns and adjectives, but it's not compulsory. If a compound noun's second element in -ate actually comes from a monosyllabic word ending in [eɪt], compounds seem to retain it: "classmate" [ˈklæsmeɪt].

Verbs that end in "-ate" have [eɪt] without exceptions for sure, however. (Ha! And then they say sound changes can't be restricted to certain morphological categories...)

I think I should open a new thread about this [ɪt] - [eɪt] business... "A desolate place", "to desolate the victims"; "the French delegates", "to delegate responsibilities"...

I think this is just a reflection of the different stress patterns of nouns and verbs, as in pairs like <rebel (n.)> ["rE.b@l] vs. <rebel (v.)> [r@."bEl]. In nouns, the final syllable must be unstressed: <delegate (n.)> ["dE.l@.g@t]; while in verbs, it can get stress: <delegate (v.)> ["dE.l@.%geIt]. As far as the "schwa/schwi" difference goes, I think it's only contrastive word-finally (or stem-finally). The reason for the apparent different distribution of schwa and schwi probably lies in the surrounding consonants: phonetic studies have shown that schwas have no fixed target in the vocal tract.

[Imralu]

Barnacle [ˈbäːnɪkəɫ]
Hydrogen [ˈhɑedɹɪdʒən]

Talking about dissonances between orthography and pronunciation, let us not forget all those nouns and adjectives with [ɪt] represented as "-ate": palate, vertebrate, etc.

[...]

Also: somebody in these forums once told me that "bias" is apparently pronounced [baɪɪs] (BYE-iss) in NAm English.

All with a schwa for me. I only became aware that those -<ate> words have [ɪt] in other dialects a couple of months ago from some ESL materials.
For me: [ˈpʰælət̚] [ˈvœːɾəbɹət̚] [ˈbɑe̯jəs]
Bias sounds the same as "Buy us ..." as long as 'us' is unstressed and schwa-ified.

I can't find 'bias' with /ɪ/ in any dictionary, only with the schwa.

Also: somebody in these forums once told me that "bias" is apparently pronounced [baɪɪs] (BYE-iss) in NAm English.

I pronounce it like that but with a /z/ at the end (BYE-is). How else would you pronounce it?

With /s/, as I'm sure you are actually aware. Incidentally, if I pronounced 'bias' with /z/, it would merge with 'buyers'.

I have a friend who pronounces 'us' with /z/. She also doesn't have the Australian /æ/ /æː/ distinction and pronounces 'planet' and 'plan it' the same way, both with a short /æ/. For most Australians, 'planet' and 'plan it' are /ˈplænət/ /ˈplæːnət/. I haven't got her to say "Mrs Manning is manning the front desk" yet. I grew up pronouncing 'length' and 'strength' with /ɪ/ until I learnt that everyone else says them as /e/ (it always sounded like /ɪ/ to me) and I also grew up pronouncing 'onion' as /ˈʌŋjən/, which I got from my mum. My brother pronounces <ing> as /iːŋ/ [iŋ] and no one's ever noticed (until he wrote something phonetically). Everyone's misheard some things at some time, analysed something into a neighbouring phoneme and grown up pronouncing something a bit differently. That's one of the things that causes language change.

Because of the schwa-schwi merger, and non-rhoticness, Australians have an interesting minimal quadrant with voicing as the only distinction.

surface [ˈsœːfəs] service [ˈsœːvəs]
surfers [ˈsœːfəz] servers [ˈsœːvəz]

I think some of my students heads exploded the day I discovered this. Most of them are fine with the /f/ /v/ thing in the middle, but hearing voiced and voiceless consonants at the end of a word is a problem for most.

Finally, I have a question for people who speak those British dialects with final /ɪ/ in words like 'city', and who also pronounce the vowel in -es and -es (when there is one) as /ɪ/. Is there a merger between 'taxes' and 'taxis' or 'studded' and 'studied'? If not, how is it resolved?

[Ser]

I think this is just a reflection of the different stress patterns of nouns and verbs

No, because of noun/adjective counterexamples, such as, apparently, "apostolate" or "̈italianate".

I grew up pronouncing 'length' and 'strength' with /ɪ/ until I learnt that everyone else says them as /e/ (it always sounded like /ɪ/ to me)

You just wrote "discriptive" in the other thread:

My mum struggles to recall the proper words for things at times and just uses a discriptive phrase.

Related?

[Imralu]

I grew up pronouncing 'length' and 'strength' with /ɪ/ until I learnt that everyone else says them as /e/ (it always sounded like /ɪ/ to me)

You just wrote "discriptive" in the other thread:

Oops. That happens.

The 'length' 'strength' thing is surely not related to the schwa-schwi thing because these are stressed vowels. I think it's probably just some allophonic thing caused by /ŋ/, which nudges the vowel a little bit higher in people's mouths and just over the position where my brain has put the ɪ-e perceptual boundary. It probably indicates that my ɪ-e boundary is a little bit lower than most other people's.

My mum struggles to recall the proper words for things at times and just uses a discriptive phrase.

Related?

To what? (Well, obviously I'm related to my mum but ...) Are you asking if my mum's tendency to use descriptive phrases could have caused my poor spelling, or if it has caused the Australian Schwa-Schwi merger? Or are you asking if the schwa-schwi merger is the cause of my spelling mistake? If it's the latter, then quite probably. The fact that around half of the vowels I pronounce are schwas probably doesn't help me remember spelling.

Tonight, or rather very early this morning, while very tired at a friend's place, I couldn't spell 'electrocute'. I thought it was 'electricute', which, now, looks completely, ridiculously wrong, but at the time, looked right. That's probably partly due to association with the words 'electric' and 'electricity' and partly down to the fact that I pronounce it [əˈlek̚tʰɹ̥ɪˌkçʉːt̚]. I asked my friend how he pronounced the 'o' and, interestingly, he wasn't quite sure whether he said [ɪ] or [ə]. He said he was aiming for /ə/ but it came out more like [ɪ]. I feel like I'm aiming for /ɪ/. We both agreed that it doesn't matter which one you aim for, there, as it comes out of our mouths the same way ... which to me, is completely non-scientific but interesting evidence of a true merger there. I'm starting to notice that completely unstressed vowels seem to me to be more like [ɪ] before post-alveolar and velar consonants, ie. /k/ (but not /kw/) /g/ /dʒ/ /tʃ/ /ʃ/ /ŋ/ and more like [ə] before consonants further forward in the mouth. For some reason, /kw/ doesn't seem to be included in this, as 'elegant' seems to have [ɪ] but 'elephant' and 'eloquent' seem to have [ə]. Of course, this doesn't tell you how I actually say these sounds, but just how I perceive them, and this might not be universal among all Australians, as the difference from my friend shows.

The only thing that seems to go against my consonant theory is the fact that words with -ock, like Ruddock and bullock definitely seem to have [ə], which may mean the distinction is actually only preserved before /k/ but neutralised in all other environments. I can't think of any words that have a post-tonic schwa before /g/ or /ŋ/, so I can't tell if it's all velar consonants. Again, though, the business with the 'o' in electrocute seems to hint that the merger is lost before /k/ ... so I'm not sure. There's a lot of maybes here. Another one: maybe the distinction is kept in final syllables before /k/ but not in non-final syllables. What I'm sure about, though, is that the merger is complete before /s/ /z/ /t/ /d/ /f/ /p/ /θ/ and /ð/. It feels very artificial to pronounce unstressed /ɪ/ before them. </unscientific analysis of own speech>

[TzirTzi]

Also: somebody in these forums once told me that "bias" is apparently pronounced [baɪɪs] (BYE-iss) in NAm English.

I pronounce it like that but with a /z/ at the end (BYE-is). How else would you pronounce it?

/"bAI%as/ or /"bAI%{s/, with secondary stress and an unreduced vowel in the second syllable.