Only if and except when

I’ve been scribbling about triggering environments for diphthong raising in Canadian English for the last few days, because I keep hearing examples in my own speech that illustrate a complication in the process. The canonical pattern is that the low-nucleus diphthongs acquire a raised central nucleus before voiceless consonants (in the same word). Thus (roughly) [aw] and [aj] become [ʌw] and [ʌj]; the contrast is observable in pairs like ride [rajd] vs. write [rʌjt].

The process is called Canadian Raising because it applies to both diphthongs throughout what Chambers (1973) calls “Heartland Canadian” (Ontario westward), and (as far as I’ve been able to discern) in the remainder of contemporary Canadian English as well. (A little bit of digging into it reveals anecdotal stories of a similar process applying to one or the other of the diphthongs in New England, western Pennsylvania, Virginia, and the colder parts of the Midwest).

The ‘famous’ part (first noted in Joos 1942) is the opacity introduced by intervocalic flapping, such that the diphthong is raised even before the phonetically voiced variant of /t/ in writer, pronounced [rʌyɾɚ] by speakers of the dialect. A separate issue (and I believe one no less deserving of the same degree of teeth-knashing) is the interplay of stress in the application of the rule.

It works like this: Chambers (1973) notes that if the syllable following the one with the diphthong bears relatively more stress, the process is blocked. Thus, the diphthongs remain low in words like typology [tàjpáləǰi], isosceles [àjsásəliz], and psychology [sàjkáləǰi]. If the morphology forces a different stress pattern (namely, a reduced vowel following the diphthong), the same vowels will raise as expected, as in typological [tʌ̀jpəláǰəkəl], isotherm [ʌ́jsəθɚm], and psychological [sʌ̀jkəláǰəkəl]. (btw my judgements match Chambers for this set; we’re both speakers of Cdn. Eng.).

Chambers also notes an issue with cyclops, citing four subjects as well as himself as people who do not raise the diphthong in this word, despite not having relatively more stress on the following syllable. He does acknowledge that others raise the diphthong; I’ve mumbled the word to myself too many times since January to have any reliable judgement one way or the other. (An interesting account offered by Chambers is that perhaps the exceptional form is learned from Classics professors).

Nevertheless there is an additional set of words (for me at least) in which the stress in the second syllable can be relatively less than that of the diphthong, and raising again fails to apply. Some examples include icon [ájkàn], biceps [bájsɛ̀ps], and Ticonderoga [tájkàndərógə] (forgive the stress marks in the last one; I only want to indicate that the first syllable can be more prominent than the second). NB Chambers counts icon as raised – I’d be curious of other judgements going either way.

I’m tempted to lump Cyclops with icon and say that any degree of stress in the following syllable is sufficient to block raising. Then the process basically applies only if the diphthong and voiceless consonant are in the same foot, thus unifying the failure of raising in icon and typology.

But along comes nitrate, which does have a raised diphthong, and which Jensen (1993), another speaker of the dialect, uses to argue that the foot domain cannot be part of the structural description. Presumably nitrate is two feet, parallel to icon. So I don’t know what to do with this; removing the foot restriction ruins the story for typology and icon. Maybe nitrate is actually one foot, because of its chemical morphology or something.

Anyway, it’s an awfully complicated set of restrictions. Even aside from the nitrate problem, it seems like a pretty cool example of combining ‘do something only if’ and ‘do something except when’ … raise the diphthong only if it’s followed by a voiceless consonant (or remnant thereof), except when the following syllable has any degree of stress.

Chambers, JK. 1973. Canadian Raising. CJL/RLC 18. 113-35.
Jensen, J. 1993. English Phonology. Benjamins.
Joos, M. 1942. A phonological dilemma in Canadian English. Language 18. 141-44.
Mielke, Jeff, Mike Armstrong, and Elizabeth Hume. 2003. Looking through Opacity. Theoretical Linguistics 29.1-2: 123-139.

8 thoughts on “Only if and except when

  1. Jenny McGregor

    Hi, does anyone know why “spider” undergoes Canadian Raising, at least for me? Could there possibly be an underlying /t/ here, which undergoes flapping? Otherwise, this would seem bizarre.

    I am an American native speaker of English from western New York state and California and the west coast, by the way, and /ai/ raises for me, not /au/. It seems to me that many Americans raise /ai/, but may not be willing to admit this.

    Also, in response to Bob’s question, for me, raising does not occur in icon, biceps, and Ticonderoga, but is does occur in cyclops, psychology, isosceles, and typology.

  2. Jenny McGregor

    The canonical pattern is that the low-nucleus diphthongs acquire a raised central nucleus before voiceless consonants (in the same word)

    This is not always true. A counter-example would be high school (as opposed to high score), where the /ai/ raises, caused by the voiceless /s/ in the following word. Of course, it could be argued that high school could be considered one word, but then would one also consider middle school or elementary school one word? I don’t think so; but maybe since high has only one syllable, we classify high school as a single word?

  3. Bob Kennedy

    Jenny’s comments help illustrate that different dialects have different environments for raising. I’ve also heard high school with a raised diphthong, but from Americans with non-raising dialects (i.e., same person who raises high school does not raise in rice. My diphthong in high school is low.

    Raising in spider is more problematic for the generalization; this I’ve heard from Canadians too. An abstract way out is to claim an underlying /t/ for the word, which triggers raising despite being a flap on the surface (like in vitamin.

    More chemical morphology: I have noticed that micron follows nitrate in having a raised diphthong despite being two feet.

  4. Joshua Guenter

    The words wherein the diphthong raises before a stressed syllable all have consonant clusters in the second syllable’s onset: “cyclops”, “nitrate”, “micron”, (“high school”) while the words which block raising all have single consonants in the same place: “icon”, “biceps”, “ticonderoga.”

    Proposal: raising of the diphthong is conditioned by a voiceless consonant in the coda. Initial single consonants of a stressed syllable are only in the onset of that syllable, and hence do not condition raising. Initial single consonants of an unstressed syllable are ambisyllabic and hence condition raising (note that this is the same environment that conditions flapping as well). Initial consonants of a consonant cluster in stressed syllables are ambisyllabic, and hence also condition raising (they cannot be considered purely in the coda of the preceding syllable, because the /t/ in “nitrate” undergoes palatalization and affrication in dialects which have that, distinguishing it from “night rate.”)

    Why? I don’t know.

    Or maybe somebody can come up with some counter-examples (i.e., diphthongs that raise before a single consonant or don’t raise before a cluster).

    In response to Bob Kennedy: yes, we could just consider “spider” (and “cider” as well. They both rhyme with “writer” for me, not with “rider”) to have /t/ underlyingly (orthography be damned!) and this would solve things synchronically. But this does raise (sorry) a question of how this sound change came about.

    I’ve got nothing there either.

  5. Erika

    I think words like “spider”, “cider”, etc. raise for some people/dialects because there’s no shorter form of the word to illustrate whether there’s a /t/ or /d/ underlying, so they could reasonably raise or not raise.

    Doesn’t explain why I raise the diphthong in ‘tiger’ though.

  6. Geoff

    One possibility that some have suggested is that the rule is no longer postlexical and the vowels have phonologized. I doubt this myself because any Canadian I have taught either phonetics or phonology to can’t hear the difference until several weeks into the semester (and even then, not reliably). It may be that the raising is correlated not with voicing per se, but with short vowels, which are conditioned not only by word-final voiceless consonants but also as the first vowel in a bisyllabic foot. Or at least that it’s moving that way.

  7. Rachel Klippenstein

    Contrary to Chambers and your judgements, I do have a raised diphthong preceding a more-stressed syllable in words such as “typology” “iscosceles” and “psychology”. I’ve also done some research to show that some other speakers do this too. Some speakers also raise /ai/ sometimes before /r/, most notably in the morpheme “fire”, but also I think sometimes in other words.

    I think I do have an unraised diphthong in “icon” and “biceps”, but I would use a raised diphthong in “Cyclops”.

  8. R. Wilson


    I’m from Bucks County, Pennsylvania, USA, not too far from the city of Philadelphia. A couple months ago I had an online conversation with a friend from Nebraska, who said that nine and night basically rhyme, as in they have the same vowel sound. I said “sort of” and we got into a little argument over it. Then I set out to prove once and for all that they were indeed different sounds. Of course, dictionaries list them as the same sound, “long i”, so I tried doing some searches on Google, but I was able to find very little on the topic. I asked several of my friends here at the College of New Jersey, and some other friends from my hometown, how they say “high” and how they say “high school”. For each of them, the vowel in “high school” was raised and was noticeably different than the vowel in “high”. The students here at the College of New Jersey come mainly from all over NJ, so this peculiarity must not have just come from my hometown.

    (My much older brother says “high school” as you would two separate words, but everyone else in my family raises the vowel in “high school”. Because of this, we make fun of the way he says “high school”, it is quite humorous.)

    I came to the conclusion that I/we generally raise the [ai] sound when it comes before a voiceless consonant, and do not when it’s before a voiced consonant (or no consonant), at least for one syllable words.

    Raised: “right”, “ice”, “hike”, “ripe”, “knife”

    Not raised: “ride”, “size”, “tiger”, “vibe”, “dive”, “nine”, “time”, “blithe”, “fire”, “file”, “die”

    So it seems that this is not limited to Canada, western PA, cold parts of Midwest, etc. but may/does extend to some other areas as well…

    Unlike in Canadian English, though, we never raise the [aw] sound, but I bet you’d be surprised if we did.

Comments are closed.