Talk:Cot-caught merger

Is this really mainly just a North American thing? I'm from northern England and I'm fairly certain that I pronounce cot /kɑt/ and caught /kɑːt/. Any comments? GCarty

I take it from your description that you do have two different phonemes, and that you can tell them apart, but you believe that they differ only in length, not in point of articulation. By phonemes, you mean you have pairs of words that you can tell apart, by the fact that one has a long vowel and the other a short one.

For North Americans with the merger, the two words are indistinguishable. Moreover, the phonemic distinction in the speech of others is lost. There is a raft of British "a" and "o" and "aw" vowel sounds that are apparently distinct to most British ears, but which all fall into the generous space left by the merger as they are heard by North Americans. American attempts at describing or imitating what they hear on the lips of British speakers often wander far astray, because what they hear maps into a different system.

Your remark does raise a number of interesting issues. Obviously, North American English relates somehow to the English of the British Isles. There were several different waves of British colonial emigration to the Americas, and these settlers settled in different places. To overgeneralize a great deal, the religious fanatics tended to come from London, Norfolk, and urban areas on the eastern coast of England, and the slavemaster planters from all over southeastern England; they settled on the coasts. The criminals and paupers tended to come from the north or from Scotland, and they came later, and settled inland.

Getting from history to American dialectology is a complicated and rather vague business. There are a number of Scotticisms, like Canadian raising, that are preserved in some varieties of North American English. I know also that the vowel system of Scots differs strongly from that of southern English with far fewer vowel phonemes. Moreover, various varieties of northern English gradually approach Scots (again, a vast oversimplification). I am wondering whether the cot-caught merger might be another Scots or northern English feature that has gotten into NAm English. -- Smerdis of Tlön 01:14, 5 Nov 2003 (UTC)

I am from (and live in) the geographic center of the coterminous USA, and I (and almost everyone else I know here) pronounces "cot" as "kaht," and "caught" as "kawt." Any comments as well?

The following statement is outright wrong:

only a few areas, most of which are along the east coast of the continent, continue to observe the distinction.

Throughout the midwest, everyone I've spoken to observes the distinction, and my friends from the south do as well. As far as my experience goes, the only dialect I've heard where there is no distintion is from the Boston area. CyborgTosser 01:34, 30 Apr 2004 (UTC)

I've added a link to a dialect map that shows the area where the merger occurs. Apparently the Midwest is divided, and on the American side of the Great Lakes region the merger does not occur. Smerdis of Tlön 17:48, 2 May 2004 (UTC)

Thanks, Steve (aka Ihcoyc/Smerdis of Tlön) for the addition. We can all get this article tweaked a little to further facilitate it. BTW, I have spent my evening hours tonight reading your excellent articles; keep up the good work. —Catdude

What factual accuracy is disputed? The existence of the merger, or its geographic distribution? -Branddobbe 03:00, May 4, 2004 (UTC)

Its geographic distribution. The merger surely exists in some English-speaking areas. Thanks for the question. —Catdude

I've lived in California almost my entire life, and I differentiate between the two words. And definitely, Southern speakers have a difference. This article is a little suspect. RickK 04:58, 6 May 2004 (UTC)

I've lived in California my entire life, and I definitely DON'T differentiate between them, and furthermore pretty much everyone I've heard who does is from other parts of the country (or world), usually the East Coast. /O/ doesn't even exist in my idiolect. The cot-caught merger is alive and well. -Branddobbe 05:06, May 6, 2004 (UTC)

The map suggests that the San Francisco or Oakland area may contain a partial exception to the merger. Resistance to the merger apparently is linked to two other phenomena; either the Northern Cities Shift, or the Southern conversion of the sound of /æ/ into a complex diphthong. Some areas of California may have been settled by South Midland speakers with complex /æ/. -- Smerdis of Tlön 14:14, 6 May 2004 (UTC)

I have added yet more wafflements about the geographic distribution of the merger. The people at the Phonological Atlas site believe that resistance to the merger relates to other NAmE features like the northern cities vowel shift (that article needs work also) and the Southern diphthongization of /æ/, so I made reference to those findings as well. Smerdis of Tlön 17:20, 6 May 2004 (UTC)

Thanks for the additions, Steve (aka Smerdis of Tlön). BTW, I think that, at this point, the "factual accuracy dispute" message I appended to the associated article can be removed, which I went ahead and did. I especially appreciate the distribution map you added. —Catdude

---

Some varieties of North American English still have both the vowels [ɑː] and [ɔː], but [ɔː] is a conditioned variant that only occurs before certain sounds, particularly /r/ or /l/, and does not count as a separate phoneme.

I don't think this is true — my English (I'm from Tennessee) is merged, and in it as well as all the merged English I'm familiar with, [ɑː] and [ɔː] do contrast before r (but not l): far and for, star and store, car and core, etc., are not homophones, but use [ɑː] and [ɔː] respectively. The choice of the [ɔː] or [ɑː] vowel here is neither conditioned nor allophonic, though the vowels are merged to [ɑː] in all other contexts. —Muke Tever 04:45, 16 Jun 2004 (UTC)

This would appear to be a different vowel entirely; historically, the vowel in core, store, and so forth is /o:/, the same vowel in "stone" and "boat," not /ɔː/ or /ɑː/. Apparently some southern US speech does in fact raise /o:/ to /ɔː/ before /r/. Smerdis of Tlön 13:59, 16 Jun 2004 (UTC)

Interesting — but that doesn't make the statements as they stand on the page any more true, as they still have both [ɑː] and [ɔː]. In the examples given most may have had /oː/ originally, but not the case of far/for, nor other [ɑː]/[ɔː] pairs such as barn/born, tart/tort. —Muke Tever 16:54, 16 Jun 2004 (UTC)

But in this case, [ɔː] is not a phoneme, it's merely an allophone of /oː/. So while that accent or dialect may have [ɔ], it's not as its own phoneme, so the article still stands. I'll put the necessary distinction in the article. -Branddobbe 17:09, Jun 16, 2004 (UTC) [[ I put in the missing semicolon in 'may have [&#596]' Kesuari]

That makes more sense, thanks. —Muke Tever 14:33, 18 Jun 2004 (UTC)

People seem to have their phonetics wrong...

There is a severe misconception that Americans pronounce the "augh" in "caught" as /O:/. Think about it this way: if the word "bore" is pronounced /bO:r/ (and this is wrong too, it should actually be /bo:r/), and if "bought" is prnounced /bO:t/, then technically "bore" should be "bought" with an /r/ instead of a /t/ at the end, and "bought" should be "bore" with an /t/ at the end instead of an /r/. Are they? Not even close. The correct pnoetic transcription of "caught" should be /kQ_ct/ (with a low back rounded, less rounded) for modern American English, or /kQ@t/ for some northern American dialects (New York, New Jersey, Philadelphia). -Jordan

Thanks for your suggestion. It seems to be contrary to all the phonetic descriptions of English I've ever seen, including in particular those given in dictionaries. Do you have any evidence that isn't original research to support your claim? Nohat 19:13, 19 Feb 2005 (UTC)

See for example the chapter 13 pf the Atlas of North American English (http://www.ling.upenn.edu/phonoatlas/Atlas_chapters/Ch13/Ch13.html), which discusses the short-o configuration of various American accents. Something like /O:/ is used in some regions, especially New York, New Jersey, and Philadelphia—/O:/, /O@/, /Q@/, and even /U@/ are all used in that region, with some variation based on education, social class, etc.; but a lower vowel like /Q/ is more common across most of the U.S. AJD 19:49, 19 Feb 2005 (UTC)

The received wisdom about the merger is that it's /ɔ/ and /ɑ/ that have merged, not /ɒ/ and /ɑ/. The problem is of course that the notion of the exact phonetics described by particular symbols varies. For phonological purposes, it often makes sense to use as few symbols as possible to describe classes of sounds, and the symbol traditionally used for the sound of the vowel in "caught" (for unmerged dialects, including British) is /ɔ/. Now of course there is a cardinal vowel definition for that sound, and it is probably true that for a wide swath of AE speakers the actual vowel used is closer to the phonetic description of cardinal vowel /ɒ/ than /ɔ/. However, this fact doesnʼt override the other fact that there is a long history and tradition of using the symbol /ɔ/ to represent the vowel sound of "caught", regardless of how the symbol /ɔ/ might be defined in other contexts.

I have made some changes to the page based on these comments and the content of the link above (thanks Ajd). I couldn't figure out how to word the bit about /ɔ/ in ??court??, so I took it out for now. Please put it back if you can make it make sense and be correct.

As a note, Itʼs been my experience that General AE speakers with the cot-caught distinction don?;t have a rounding distinction at all, but a height/frontness distinction: that caught is /kɑt/ and cot is /kat/ or /kɐt/. Unfortunately the Phonological Atlas chapter you cite doesnʼt use IPA at all but their own very confusing phonetic transcription. Is there a page that describes what they mean by /o/, /oh/, /ae/, /aeh/, /aw/, etc.?

On another note, I?m not in principle opposed to changing the article. It clearly needs help. It's just that one must be skeptical about assertions made by anonymous users without any sources. Cheers! Nohat 23:24, 19 Feb 2005 (UTC)

The phonological atlas doesn't use phonetic transcription, or doesn't use it often, anyway. Precise phonetic values are indicated by citing the numerical formant values of the vowels in question. /o/, /oh/, and so on are phonemic symbols, whose phonetic values vary from dialect to dialect. /o/ denotes the phoneme of cot and hop, /oh/ in caught and moss, /ae/ in cat and lap, /aw/ as in house, and so on. /ah/ as in father, and so on. /aeh/ is a phoneme that exists only in dialects like New York and Philadelphia, where words like grass have a different phoneme than either lap or father. The notation is based (loosely) on Labov's analyses of how the "subsystems" of the vowel system shape sound change. There's probably a chapter which explains all the different symbols somewhere in the Phonological Atlas, but I don't remember offhand which one it is. AJD 00:45, 20 Feb 2005 (UTC)

From what I understand, the caught-cot merger describes that Americans use similar, but not equal, vowels in the words "cot" and "caught." In British English, these words are pronounced /kQt/ and /kOt/. In modern American English, these words are pronounced with much more similar sounds: /kAt/ and /kQt/. Listen as I pronounce "cot" and "caught" using the vowels [A] and [Q]:

http://www.geocities.com/jordanekay/cotcaught.wav

We do not pronounce "caught" with an [O]. Do we say this:

http://www.geocities.com/jordanekay/caught.wav

Absolutely not.

And about "bought" and "bore" having the same vowel sound, again, if you believe that to be true after hearing these words pronounced,

http://www.geocities.com/jordanekay/boughtbore.wav

you need to get your ears checked. --Jordan

These kinds of forced productions by linguistically aware individuals are not useful at all for making any valid conclusions about phonetics. You need to analyze the speech of people who aren't conscious of they way they are producing sounds. Second, you've ignoring the essential point here that the symbols are just that, symbols, and what they exactly represent is not expected to be universally agreed upon. The symbol /ɔ/ is used for the vowel in caught because that's the symbol that has always been used for the vowel in caught. Nohat 08:12, 20 Feb 2005 (UTC)

Also, an acoustic analysis of your productions in [1] (http://www.geocities.com/jordanekay/cotcaught.wav) match up quite well to the description of American English /ɔ/ and /ɑ/ in Ladefoged and Maddieson's Sounds of the World's Languages, page 286, Figure 9.5 (left). There is an approximantly 100 Hz difference in F1, accounting for the height distinction, and about a 300 Hz difference in F2, accounting for the fronting distinction in your productions of those vowels. In Ladefoged and Maddieson, the height difference is greater, with a 150 Hz height difference, and the backness distance is smaller, with only a 100 Hz difference. With the differences in F1 and F2 being substantial, it's doesn't really seem tenable to say that the only distinction between these vowels is a rounding difference. There is a distinction in both vowel height and backness, so choosing symbols whose descriptions only distinguish rounding isn't a particularly accurate use of those vowel symbols. Based on this data, I would say your vowel for 'cot' is [ɐ] and your vowel for 'caught' is fully back, with a height between [ɑ] and [ɔ] (just slightly lower than your [ɐ]). Your vowel for 'bore' is clearly much higher and is probably [o], or more likely between [o] and [ɔ]. This is in line with my earlier prediction about the vowels for General AE speakers with a cot-caught distinction. Nohat 10:19, 20 Feb 2005 (UTC)

OK, now I see that we use [O] to denote the broad transcription of the word, thus it has nothing to do with the IPA symbol /O/ (the "augh" in British English "caught"). So there is no need to change this, but I would like to refute your claim that my "ah" and "aw" are pronounced in different positions in my mouth. They are indeed, unrounded and rounded versions of each other, and the vowels I have used are most definitely /A/ and /Q/. And with regard to my being "linguistically aware" of what I am saying, listen as Ben Stiller, who I'm sure has never even heard of the IPA, pronounces "not" with /A/ and "dog" with /Q/:

http://www.geocities.com/jordanekay/linguisticallyunaware.wav

Here they are individually

/A/:

http://www.geocities.com/jordanekay/OpenBackUnrounded.wav

/Q/:

http://www.geocities.com/jordanekay/OpenBackRounded.wav

So this article is fine as it is, but wouldn't it be useful to include the actual IPA representations used in modern American English (which contains the "cot-caught" merger), /A/ and /Q/, instead of meaningless "phonetic" symbols such as [O], which could stand for virtually any back , rounded vowel?

Oh, and I just got finished placing my hand on my whole tongue as I repeatedly said "ah" and "aw," my hand staying completely stationary.

The phonetic research shows consistently (Ladefoged and Maddieson, [2] (http://www.brunel.ac.uk/depts/ee/Research_Programme/COM/QinASRU2003.pdf)) that in American English, there is a substantial difference in F1 for /ɔ/ and /ɑ/, which corresponds to a difference in vowel height. When selecting symbosl from the IPA, it is important to select symbols from the IPA chart whose relationship on the chart corresponds to the acoustic differences that can be measured in those sounds. Since a clear height distinction can be measured in those sounds, symbols that represent a height distinction are selected from the IPA chart. Your particular /ɔ/ may be lower than the average for Americans, the values of which can be found in the acoustic research, but I would think it would be obvious that it's better to select symbols based on analyses of the vowel spaces many speakers, rather than just based your own vowel space. Your demonstrations on Talk:International Phonetic Alphabet for English have already made it clear that your accent is strongly regionally marked, so it's probably unwise to rely on your own productions to draw conclusions about General American vowels. Nohat 21:21, 20 Feb 2005 (UTC)

Merged vowel

This article makes it seem as if all merged speakers realize the merger as /A/. This is false; merged speakers in New England realize the merger as /O/. I, for example, am a merged speaker from New England and I pronounce "cot" and "caught" with a rounded vowel distinct from the unrounded vowel of "father". --Industrius 03:52, 8 Apr 2005 (UTC)

You're right. I've corrected it. I don't think you're right, though, that the New England realization of the merged vowel is as high as [O]; I've described it as [ɒ]. AJD 04:47, 8 Apr 2005 (UTC)

That's a matter of interpretation. is traditionally used by American linguists to transcribe any low to lower-mid rounded vowel, even if the sound in question is phonetically fully open. It's not true, however, that all Canadians merge to ; I think only the ones in the Maritimes do. From Ontario to B.C. I'm pretty sure the merger is to an unrounded . --Angr/_comhrá 05:53, 8 Apr 2005 (UTC)

Yes, well, I hold that we should value accuracy over tradition to the extent possible. Anyway, speakers I know from Ontario definitely have a rounded cot-caught—often more rounded than my own New England one. Labov's ANAE suggests, but does not state, that this is the norm throughout Canada: he doesn't describe the roundedness or lack thereof of Canadian /o/ (his notation), but he does describe it as having moved in the direction of old /oh/; and it's strongly backed and slightly raised, which would tend to have a rounding influence on it.

Also, American English may not have distinctive phonemic length, per se, but those symbols were in brackets, indicating plain old phonetic length, which I'm going to restore. Deleting the length mark makes it look like Boston cot sounds the same as London cot, which it doesn't. AJD 06:29, 8 Apr 2005 (UTC)

But including the length mark makes it look like GenAm cot sounds the same as London cart, which it doesn't. Anyway, vowel length should only ever be marked when it's phonemic (contrastive), because length is relative. A vowel isn't "long" or "short" unless it contrasts with another vowel of the same quality that is shorter or longer, respectively. --Angr/_comhrá 06:54, 8 Apr 2005 (UTC)

Western US cot isn't as different from London cart as Boston cot is from London cot. Anyway, in American English length isn't contrastive per se, in that there aren't vowels with the exact same quality that are distinguished only by length; but length is still relevant and important to the phonological system. In particular, the distributional constraints on short vowels (as in pit, pet, pat, putt, put) are stronger than those on long vowels and diphthongs—short vowels can't occur word-finally, and so on. The vowels of cot, in (the relevant) American dialects, have the same distributional constraints as diphthongs, even if they're monophthongal themselves: that means that, phonologically speaking, they're long vowels. It's not at all true that a vowel isn't "long" or "short" unless it contrasts with another vowel of the same quality; it can still be longer or shorter than vowels of different qualities, and in this case it is relevant enough to be mentioned. AJD 07:28, 8 Apr 2005 (UTC)

I suppose it's a matter of opinion. To my ear, Western US cot is noticeably shorter than RP cart, while the difference between Boston cot and RP cot is more one of vowel quality than duration. As for what is relevant and important to the phonological system, for example in distinguishing vowels that can stand in open stressed monosyllables from those that cannot, I think most phonologists would say it is tenseness rather than length that's critical: lax vowels can stand only in closed stressed monosyllables, while tense vowels can stand also in open ones. While it's true that tense vowels are longer than lax ones in the same environments (beat is longer than bit), vowel length is affected by many other things as well, such as the voicing of the following consonant: (beat is actually shorter than bid). As for the distribution of , it's very strange as it seems to pattern with lax vowels in a lot of ways but with tense vowels/diphthongs in other ways. Like lax vowels it can stand before consonant clusters like [sp, sk, ks, ps, kt, pt] within the same morpheme (wasp, mosque, box, copse, concoct, opt), which tense vowels and diphthongs can't (things like beaks, beeps, leaked, beeped always involve a morpheme boundary). On the other hand, it can stand in stressed open monosyllables (ma, pa, baa, bra, spa), but only in a handful of words that are hypocoristics, onomatopoeias, or recent borrowings. --Angr/_comhrá 08:29, 8 Apr 2005 (UTC)