Syllables in: fairy, ferry, and very

<<I am not sure whether this schwa is included to indicate the presence of a phonetically real off-glide, or else as a typographical convention by which the vowel may be given a separate symbol. In my own speech, I do not think I hear a true schwa there, as DX does. Rather, I agree with Lazar, that I hear an "R-ness that blends with the vowel". I take the transcription epsilon-schwa as merely a traditional designation of this r-coloring. For me, I would say that the quality of this vowel is monophthongal, somewhere between the "a" of "face" and the "e" of dress.>>

Yes; the transcriptions [E@`], [I@`] are just a standardized way that I can transcribe the R-colored vowels. The exact phonetic effect of the R-coloring varies from speaker to speaker, and it's often some combination of length, closeness, etc., not always "diphthongization", strictly speaking. (For example, my values for [E@`] and [I@`] tend to start off a little closer to [e] and [i].) My basic thesis is that these R-colored vowels need their own symbols.

<<For my part, it is very difficult for me to imagine a pronunciation of "fairy" (and therefore "fair") as well with the vowel of "dress". It would come out "ferr", a kind of clipped, bizarre sound. When I hear mergers from the midwest speak, I tend to hear "fairy" for "ferry", as Lazar's transcription suggests, and not "ferry" for "fairy".>>

Yes, that's the thing. When I first read about the standard transcription that would write GenAm "fair, spear" as ["fEr], ["spIr], this struck me as really odd. I mean, if I take the vowels that I use in "ferry" and "spirit" and stick an R on them, to arrive at ["fEr], ["spIr], I get a weird, incredibly short and lax sound, which doesn't sound like what I hear from North Americans.

When I hear people from other parts of the country, on TV and in real life, I so often notice that they pronounce words like "irresistable", "spirit", "ferry", and "America" differently than I do. Conversely, I hear no difference between their pronunciations and mine when it comes to words like "fair" and "spear".

This is why I think the standard transcription doesn't make sense. The standard transcription says that the mergers involve a change in the vowels of "fair" and "spear" but *not* in the vowels of "ferry" and "spirit", which I find to be completely counterintuitive - the complete opposite of what I actually hear.

>>
<<For my part, it is very difficult for me to imagine a pronunciation of "fairy" (and therefore "fair") as well with the vowel of "dress". It would come out "ferr", a kind of clipped, bizarre sound. When I hear mergers from the midwest speak, I tend to hear "fairy" for "ferry", as Lazar's transcription suggests, and not "ferry" for "fairy".>>

Yes, that's the thing. When I first read about the standard transcription that would write GenAm "fair, spear" as ["fEr], ["spIr], this struck me as really odd. I mean, if I take the vowels that I use in "ferry" and "spirit" and stick an R on them, to arrive at ["fEr], ["spIr], I get a weird, incredibly short and lax sound, which doesn't sound like what I hear from North Americans.<<

That is my impression as will, simply because my usual [E] and [I] are both rather centralized, whereas the vowels in these words are not centralized. In the case of "fair" and "ferry" I would use [e_o] ([e] for short) here, as stated before. As for "spear" and "spirit" (both syllables), I would transcribe them with [I], but the [I] in "spear" and the first syllable of "spirit" is actually different from that in the second syllable of "spirit" (the default realization of /I/ IMD) in that unlike my usual [I] and like my usual [i] it is not centralized at all and it is slightly higher than my usual [I].

>>When I hear people from other parts of the country, on TV and in real life, I so often notice that they pronounce words like "irresistable", "spirit", "ferry", and "America" differently than I do. Conversely, I hear no difference between their pronunciations and mine when it comes to words like "fair" and "spear".<<

I note that dialects closer to General American often have a *slightly* laxer vowel than I have for "fair" or "for", [e_o] or [O_c] (you could also say [o_o] if you wish, but this vowel is slightly more open than the [e_o] in "fair", and calling it [O_c] allows it to be abbreviated to [O], conveniently distinguishing it from both [o] from the usual /o/ and the significantly more open [Q] from /Q/, corresponding to GA /O/). Even then, the difference really is quite slight and I have to really listen for it myself to notice it.

>>This is why I think the standard transcription doesn't make sense. The standard transcription says that the mergers involve a change in the vowels of "fair" and "spear" but *not* in the vowels of "ferry" and "spirit", which I find to be completely counterintuitive - the complete opposite of what I actually hear. <<

To me at least, the vowels in GA in "ferry" and "spirit" are somewhat tenser than the usual [E] and [I], especially in the case of "ferry", and also uncentralized, but at the same time the vowels in "fair" and "spear" are lowered in GA as well (moreso than IMD in the case of "fair").

Thanks, Lazar, for confirming what I have always felt. But the transcription of "fair" as f-epsilon-r does not necessarily indicate that "fair" changes to sound more like the "ferr" of "ferry": maybe it indicates nothing more than the arbitrary decision to use epsilon for the vowel of "fair": in which case.... what symbol should we use for the vowel of "ferry"??

I think the use of epsilon in "fair" and all SQUARE words is borrowed for some reason from RP. Maybe those speakers hear or heard the American pronunciation of SQUARE as closer to their "ferry" than to their SQUARE?

But do look at the Wikipedia page I mentioned: some people were absolutely puzzled by what I was saying (under the name "Gheuf"): they felt that "fairy" (and the for them homophonous "ferry") had unambiguously the vowel of DRESS.....

Travis, thanks for your comments, too, but I have difficulty interpreting your notation of "_o" and "_c". I have no doubt that it is standard in whatever system you are using but it might be easier to read if you simply wrote "lowered /e/", "raised /e/", etc.

<<But the transcription of "fair" as f-epsilon-r does not necessarily indicate that "fair" changes to sound more like the "ferr" of "ferry": maybe it indicates nothing more than the arbitrary decision to use epsilon for the vowel of "fair": in which case.... what symbol should we use for the vowel of "ferry"??>>

That's the thing; if you were *only* concerning yourself with merged dialects, then you could simply use [E] for "fair". (You could make that "arbitrary decision" for simplicity's sake.) The downside is that this transcription is incompatible with unmerged dialects; because the unmerged pronunciation of "ferry" is unquestionably ["fEri], and the [E] of unmerged "ferry" is just too short and lax to be used in "fair".

However, there is an alternate solution, which is used by the Random House Webster's Dictionary and by m-w.com : this solution is to show syllabification for every word.

Unmerged:
"ferry" ["fE.ri]
"fairy" ["fEr.i]
"fair" ["fEr]

Merged:
"ferry" ["fEr.i]
"fairy" ["fEr.i]
"fair" ["fEr]

They've made the arbitrary decision to represent what I call [E@`] as [Er.], which constrasts with [E.r]. The self-evident downside is, of course, that you have to show syllabification for every word.

One note: when I said [O_c] I really meant [O_r], as the diacritic _c really means "less rounded" not "closed", and _r really is what is meant if one wants a raised vowel in X-SAMPA. On that note, in X-SAMPA the diacritic _o means "lowered" or "opened".

Here is a speaker (http://www.alt-usage-english.org/mmm_bc.wav) whose pronunciation of "Mary" sounds to me like "merry".

Yeah, the vowel he uses there is on the opener side of the spectrum; his "Mary~merry~marry" could probably pass for an unmerged "merry". But there are a lot of merged speakers, like Travis, who do use a closer value there.

Lazar,

What about "zero"? I, along with many Americans pronounce it "zee row" /zi.roU/ sharing the same vowel as in "feet" as opposed to the vowel I have in "spear". I hence find [I] necessary to show the contrast between the vowels in "zero" and "hero".

"zero" [ziroU] "zee row"

"hero" [hIroU] "hear o"

<<What about "zero"? I, along with many Americans pronounce it "zee row" /zi.roU/ sharing the same vowel as in "feet" as opposed to the vowel I have in "spear".>>

Yeah, I've heard this pronunciation (and it's attested by m-w.com); I've also heard /hi.ro/ for "hero" (I'm almost certain that m-w.com listed this one, as well, in the past, but it doesn't anymore). But in your speech the two are different, so they provide a great minimal pair.

<<I hence find [I] necessary to show the contrast between the vowels in "zero" and "hero".>>

But that's no problem in my system. ;-) My proposal would be:

zero ["ziroU]
hero ["hI@`oU]

(Just to be clear, I pronounce "zero" and "hero" to rhyme, both with [I@`].)

[hIroU], within my system, could be used in "serious-Sirius"-unmerged dialects for a hypothetical word "hirrow".

<<We distinguishers pronounce them:

"ferry" /"fE.ri/
"fairy" /"fE@`.i/

The non-distinguishers don't allow sequences like /E.ri/, so they merge them as:

"ferry" /"fE@`.i/
"fairy" /"fE@`.i/>>

Well I have a distinct pattern in my speech. I pronounce "ferry" and "fairy" differently, but make no distinction between "ferry" and "furry", both are /f3`.i/.

>>The non-distinguishers don't allow sequences like /E.ri/, so they merge them as:

"ferry" /"fE@`.i/
"fairy" /"fE@`.i/ <<

Hmm... most NAE dialects that I am familiar with don't use [@`] here but rather just plain [r\], with such breaking of vowel-/r/ sequences only being common with the diphthongs /aI/, /aU/, and /OI/. Of course, then, I by far have had the most contact with dialects from the Upper Midwest, but even for most other people that I know who speak relatively GA-like dialects and in most media content such breaking does not occur, at least from what I have noticed. Of course, then, I wouldn't be surprised if such breaking were common for merged /Er/ and /er/ in some areas such as maybe the Northeast.

I also forgot to mention that it seems quite common in NAE dialects for such breaking to only occur when such an /r/ falls in a coda position, with such not occuring intervocalically (so that if one attaches a morpheme starting with a vowel to a word ending in /r/, such breaking is disabled).

<<I pronounce "ferry" and "fairy" differently, but make no distinction between "ferry" and "furry", both are /f3`.i/.>>

Yeah, it's common to merge some, but not all, of the distinctions. (For example, I've read that in the South it's very common to merge "Mary, merry" but keep "marry" distinct. And in Philadelphia they keep "Mary, merry, marry" all distinct, but they merge both "hurry, furry" and "merry, Murray".)

<<Hmm... most NAE dialects that I am familiar with don't use [@`] here but rather just plain [r\], with such breaking of vowel-/r/ sequences only being common with the diphthongs /aI/, /aU/, and /OI/. Of course, then, I by far have had the most contact with dialects from the Upper Midwest, but even for most other people that I know who speak relatively GA-like dialects and in most media content such breaking does not occur, at least from what I have noticed. Of course, then, I wouldn't be surprised if such breaking were common for merged /Er/ and /er/ in some areas such as maybe the Northeast.>>

Well, the thing is, I'm really just using those rhoticized diphthongs as a transcriptional convention. The generally accepted convention *already* is to use those diphthong transcriptions for non-merging dialects like mine, even though in my own unmerged speech I think [E:r\] or [e:_or\] might be a better transcription for my "Mary" vowel, and [I:r\] for my "serious" vowel. Crazy as it may sound, I'm not really intending to make any assertion about diphthongization. The reason for my choice is my thesis that words like "mare, spear" should have the same transcription across merged and unmerged dialects, and that the merged values of "merry" and "spirit" tend (at least for many speakers) to be more similar to the historical "mare, spear" values. I've just arbitrarily chosen [E@`] and [I@`] to represent the whole range of vowels that can be found in "mare" and "spear" - which, depending on the speaker, can differ from basic [E] and [I] by varying degrees of length, closeness, and sometimes diphthongization. (An analogy can be drawn with the use of [E@] and [I@] in British English to represent vowels that, for many speakers, are in fact monophthongal [E:] and [I:].) One of the main reasons why I've chosen these transcriptions is my desire for cross-dialectal compatability in my transcriptions - I find it really easy to make nice, neat, and you could say somewhat overgeneralized phonemic comparisons between rhotic and non-rhotic, merging and non-merging dialects when I use this transcription scheme. It's no coincidence that you find this proposal being made by someone from Massachusetts - you could say it's bias on my part, based on a desire to have a transcription scheme that accomodates Northeastern dialects. (My system isn't perfect, but in my opinion, at least it's an improvement over the traditional system that uses the exact same vowel in merged "fair" as it does in unmerged "ferry".) I mean, I don't want to sound presumptive, but it seems from your phonetic descriptions that your "fair" and "spear" vowels are almost exactly the same as mine (in other words, as I've said, I don't think mine are truly diphthongal either); so if I were transcribing your dialect *at the same level of broadness at which I transcribe my own*, then I would use [E@`] and [I@`] there as well. (My use of diphthongal symbols would simply be an arbitrary decision, and it wouldn't imply the use of true phonetic diphthongs in your dialect any more than it implies it in mine.) It's just that we're aiming for different levels of broadness and cross-dialectal compatability in our transcriptions.
I hope what I've just written makes some kind of sense.

I myself tend to favor transcriptions of my own speech which do not seek to be crossdialectal but rather seek to be very narrow (to the point of marking both vowel and consonant length and nasalization consistently), primarily because I do not trust my own judgement with respect to the phonologies of other dialects as my actual real-life experience with other dialects is rather narrow, having lived in southern Wisconsin my entire life and having primarily come into contact with other Upper Midwesterners until recently (due to getting a job where many people are from outside the Midwest). At the same time, I have in the last few years paid close attention to dialect and idiolect features locally, and thus have become rather well-acquainted with the specifics of the speech varieties here (aside from AAVE, which I really do not come into sufficient contact with for such). For similar reasons, for transcriptions of things such as General American I tend towards relatively broad transcriptions, albeit ones that do admittedly tend to be biased towards my dialect in ways (for example, by using [e] and [o] rather than [eI] and [oU] and by not marking breaking of monophthongs before /r/, and in general representing more conservative and GA-like dialects in the Midwest) except where I am sure that a more narrow transcription is broadly applicable.

However, I do omit certain aspects of my speech which would only really be transcribed in very narrow transcription such as the difference between [e] and [e_o] (especially since in informal speech my [e] can be rather low to begin with at times and because the two are never potentially distinctive), the difference between the weakly laminal and strongly laminal allophones of /s/ IMD, the distinction between fortis and and voiceless lenis consonants (which is very often nonexistent, especially in the case of sibilants, and is far more strongly marked by vowel length), the glottalization of postvocalic fortis stops (the distinction of such with devoiced stops is also unambiguously marked by vowel length), and occasionally the distinction between dental and alveolar consonants IMD (such as [n] from /n/ and [d] from /d/ versus [n_d] and [d_d] from /D/). Note, though, that these are more exceptions for the sake of succintness and avoiding overwhelming the reader with excessively complex transcriptions more than anything else. The reason why I mark vowel length and nasalization consistently, however, is because one can readily form minimal pairs based on them IMD, even if they are often not "stable" (in that they often break down across registers or when words are used in context rather than in isolation); even still, there are a few cases like "breath" [br\ET] versus "breadth" [br\E:T] and "with" [wIT] versus "width" [wI:T] which could be potentially taken as true (if marginal) minimal pairs for vowel length phonemicity.

Interesting discussion :) Do you know any websites where they have recordings of people who distinguish "higher" and "hire" (as [haI@`] and [haIr]) for example? I just can't distinguish them myself, so it would be interesting to see if I can hear the difference.

I've always found the schwa and r to be a bit confusing. For example is the [@`] sound as in "tiger" or "water" really a schwa + r? Some people say [@`] and [3:`] (as in hurt) are practically the same, the only difference being that the latter is stressed. And again same people say the schwa is similar to [V] (as in cup), but the [V] is stressed and schwa isn't. So are they really different as to where your tongue is in the mouth (so much that it's reasonable to say that their different phonemes and not just allophones of the same vowel?) And again I would agree that the schwa sounds a lot like to [V] and [@`] sounds a lot like [3:`] but I wouldn't say that [V] and [3:`] sound similar ([3:`] doesn't seem like a V+r to me at all, and I think most people would agree).

Also, some dictionaries don't have [3:`] and [@`] at all, and instead they have just [r] or [r:] instead. Do you think that makes sense?