1. The WordReference Forums have moved to new forum software. (Details)

Canadian Raising in other Languages

Discussion in 'Etymology, History of languages and Linguistics (EHL)' started by JuanEscritor, Dec 10, 2012.

  1. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Are there any other languages (or their dialects) with the /aɪ/ or /aʊ/ diphthong phonemes that exhibit raising of the nucleus in short-vowel environments?

    A while back I was researching Canadian Raising, and came across a paper by Elliot Moreton and Erik Thomas, in which the authors claim that the motivation behind Canadian Raising is purely phonetic and that it is not related to the Great Vowel Shift, as most sources seem to believe.

    Despite a regular full-time job, I've put quite a bit of my time into researching Canadian Raising, including measuring vowel lengths of a few speakers—though I've lost some of the measurements—in different phonetic environments where CR occurs. To avoid being lengthy, I will say that I have come to the conclusion that CR is undoubtedly a residual side-effect of the GVS acting on diphthongs with varying lengths in different morpho-phonetic contexts. I believe the evidence is overwhelming to this point, but will not present it unless the thread takes that direction.

    For now I will return to Moreton's and Thomas's claim that CR is the result of natural phonetic motivation. They admit no knowledge of languages other than English exhibiting CR, claiming that the phonological process responsible for it (what they call 'Asymmetric Assimilation') has so far only been found in English. However, they also claim that "there is some reason to expect it in other languages".1

    And thus the purpose of this thread: to find out whether there really are other languages that exhibit Canadian Raising of /aɪ/ or /aʊ/ (Moreton and Thomas focus only on the former) when the diphthong occurs in phonetic environments similar to those which trigger its appearance in English CR dialects.

    Are there any examples of this outside of English?

    JE
    __________
    Moreton and Thomas (2004 p.16)
    __________
    Moreton, E., and Thomas, E. (2004) "Origins of Canadian Raising in voiceless-coda effects: A case study in phonologization" (available online: http://www.unc.edu/~moreton/Papers/MoretonThomasLabPhon2004.pdf)
     
  2. berndf Moderator

    Geneva
    German (Germany)
    For what it's worth, the opposite occurs in Swiss German (the effect being a bit more subtle, though). In standard German, the diphthongs /aɪ/ and /aʊ/ by processes similar to those of the GVS: /i:/>/ɪj/>/ɛj/>/aɪ/ and /u:/>/aʊ/ (I don't know the intermediary stages here); examples: vîle>Feile (=file; as in nail file not as in computer file), hûs>Haus (=house). The standard realizations are [ɐɪ] and [ɐʊ]. Only Swiss speaker regularly realize them [aɪ] and [aʊ]. For a Standard German speaker that sounds a bit like [a:ɪ] and [a:ʊ]; this is because [ɐ] is also the standard realization of the monophthong /a/ while [a:] is the standard realization of /a:/.
     
    Last edited: Dec 10, 2012
  3. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Very interesting, berndf!

    Are there any dialectical variations with these vowels conditioned by their environment?

    JE
     
  4. berndf Moderator

    Geneva
    German (Germany)
    I've been thinking but until now haven't found any conditions. The problem is that Germans generally don't recognize their /a/ and /a:/ vowels are qualitatively different and this whole issue is rarely discussed in the literature.
     
    Last edited: Dec 10, 2012
  5. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Fair enough. Are there any length variations that you are aware of with the /aɪ/ and /aʊ/ diphthongs?

    JE
     
  6. Youngfun

    Youngfun Senior Member

    Pekino, Ĉinujo
    Chinese/Italian - bilingual
    This may be interesting. In a forum about my native dialect, Wenzhounese (dialect of Wu Chinese), I've found this comparison between the vowel shift in English, and the one that occurs from Middle Wu to today's Wenzhounese.

    Vowel shift in English:
    ex. ME > 1600 > 1700 > today
    time /iː/ > /əi/ > /ai/ > /aɪ/
    see /eː/ > /iː/ > /iː/ > /iː/
    sea /ɛː/ > /eː/ > /iː/ > /iː/
    make /aː/ > /ɛː/ > /eː/ > /eɪ/
    stone /ɔː/ > /oː/ > /oː/ > /əʊ/
    food /oː/ > /uː/ > /uː/ > /uː/
    house /uː/ > /əu/ > /au/ > /aʊ/

    *ME=Middle English

    Vowel shift in Wenzhounese:
    C MW > ?? > 1900 > today Modern Mandarin
    李 /iː/ > /iː/ > /ɪi/ > /ei/ /i/
    天 /eː/ > /ie/ > /iɪ/ > /i/ /i̯ɛn/
    猜 /ɛː/ > /eː/ > /ɪː/ > /ei/ /aɪ̯/
    相 /aː/ > /ɛː/ > /ie/ > /i/ /iɑŋ/
    可 /ɔː/ > /oː/ > /oː/ > /ʊ/ /ə/~/ɤ/
    火 /oː/ > /uː/ > /uː/ > /ɯ/ /u̯ɔ/
    走 /uː/ > /əu/ > /ɐu/ > /ɐɯ/ /ɤʊ̯/

    C=Chinese character
    MW=Middle Wu Chinese

    As comparison I've also added the corresponding vowel in Modern Mandarin, which are historically more conservative (even more than Middle Wu).
    The bold line seems the closest to Canadian Rising.
    /iː/ > /ɪi/ is also very similar to many dialects of English.
    The vowel [ɑʊ̯] of Mandarin usually becomes [ou] in Cantonese, and a monophtong in most Southern Chinese dialects.
     
    Last edited: Dec 11, 2012
  7. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    So these changes only affected long vowels, creating qualitative distinctions where before there had been quantitative ones? Is vowel length still phonemic in Modern Mandarin or has it been lost as a result of these shifts as it was in English?

    Now that is interesting! Are there any conditions on this, or does it happen in all environments?
     
  8. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    I'm struggling to grasp exactly what you mean by short-vowel environments? If you've got a diphthong then surely that environment is not a short-vowel environment?
    Or are you not talking about anything more than clipping-environments? Sorry, just want to make sure I follow what you're asking.

    [Edit] - Aha, I took that line of "____" to be your signature and ignored what was underneath it but I looked again and realised it was a short list of references and one had a link. Question answered, it is voiceless (clipping) environments. I think calling it "short-vowel" environments is potentially quite confusing (but as with all things, when you know what actually is intended it becomes a lot more understandable).
     
    Last edited: Dec 11, 2012
  9. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    The raised diphthong is most certainly shorter than the unraised one in CR dialects; furthermore, the environments in which the shorter raised form appear also trigger shorter versions of other vowels. So the word cider has [ʌi] while sider retains [aɪ]; at the same time the vowel in cedar is shorter than the vowel in seeder, though here there is no qualitative change in the vowel. In CR, we see a quantitative change triggering a qualitative change in the /aɪ/ vowel.

    We can describe the variations in terms of morpho-phonemic environments:

    V → v/ 's-.dɚ
    V → V/ 's-.d#ɚ

    Any vowel inserted will behave according to these rules; any vowel that is subject to qualitative variation depending on its length (such as /aɪ/) will exhibit the appropriate changes in the relative contexts. For CR, the restrictive rule will keep [ʌi] in the short-vowel environment and [aɪ] in the longer vowel environments.

    Thus CR is triggered by vowel length and vowel length is triggered by the morpho-phonemic environment.

    I'm trying to find whether there are other languages that exhibit this same pattern with the two diphthongs in question (or either one of them) to evaluate the claim made by Moreton and Thomas that these are purely phonetically-motivated changes unrelated to the English GVS that would have also left the same distributional pattern.

    JE
     
  10. berndf Moderator

    Geneva
    German (Germany)
    Unfortunately not. This is an observation from listening not from measuring. And as I am a Standard German native speaker I perceive [a] always as longer than [ɐ] (you would probably transcribe it [ʌ] because [ɐ] belongs phonologically to /ʌ/ in English) irrespective of real length.

    It must be then because /aɪ/ and /aʊ/ are phonemically certainly long. Maybe you should look at Scots or Scottish English because it is at an earlier stage of the GVS and house is depending on dialect either /hu:s/ or /həʊs/.
     
    Last edited: Dec 11, 2012
  11. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    Regarding this being related to morphemes (just like ŋg -> ŋ / V_# in Early Modern English), are there examples in the research you've done that have maintained the change and aren't just potential formations that fit the environment? With something like 'sid#er' I can see that you might get a distinction based on a novel analysis which could be overridden in other more common words that have been passed down orally (like 'cider'). If that's the case then I'd give a lot more credence to your analysis and find it pretty promising. If not, then it still seems a good hypothesis, but something that would for me, always need more data to make firm conclusions regarding potential motivations given other explanations of how those differences might have come about (as I mentioned before).

    Given that you said "any vowel inserted will behave according to these rules" I imagine there must be a fairly lengthy list of examples which allows such a wide-ranging claim to be made and some of these must be able to starkly show a common history and frequency that shows a morpheme environment to be the deciding factor. The idea does make a lot of sense, because that is quite similar to what happened with NG-coalescence in English. I'm just curious about examples that don't need to resort to non-words to provide evidence for that claim.
     
    Last edited: Dec 11, 2012
  12. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    There is no such thing as phonemic vowel length in English. Vowel length is always predictable. While it is true that diphthongs are longer than monophthongs, this does not mean that all diphthongs are long. To say that a diphthong is short does not mean it has the duration of a short [ɪ], it just means it is shorter than other comparable diphthongs. The vowels in bead and bid are considered 'long' even though they are not the same length. They are considered this because each one is longer than itself in other contexts (for example, the words beat and bit).

    Likewise, the vowel in cider is shorter than the vowel in sider; and the vowel in cedar is shorter than the vowel in seeder, as the morpho-phonemic conditions of each pair are identical to those of the other pair. From my own speech:

    Word Length (sec)
    cider 0.156095
    sider 0.206047
    cedar 0.101830
    seeder 0.144864


    The raised diphthong is most definitely shorter than the unraised one by a degree comparable to the difference between other vowels in the same morpho-phonemic environments.

    Yes, CR is directly related to the vowel alterations that occur with the Scottish Vowel Length Rule as both show the raised forms in short-vowel environments and the unraised form in other environments. I think they are both historically linked as being vestiges of the GVS.

    JE
     
  13. berndf Moderator

    Geneva
    German (Germany)
    If you restrict your analysis to modern North American dialects, this can reasonably be claimed. But I don't think this can be said of all dialects. In the phonology of RP most authors maintain the notion of phonemic vowel length.
    That is no contradiction. In languages with phonemic vowel length, long and short varieties of a vowel are normally accompanied by qualitative differences, i.e. long "i" = [i:], short "i" = [I].
     
  14. Alxmrphi Senior Member

    Reykjavík, Ísland
    UK English
    Not of Australian English, I've read.
     
  15. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Contrastive minimal pairs would be interesting to see. I have taken a look at Australian English, and Wikipedia gives some contrastive pairs of syllable rimes, but I don't see any discussion on minimal pairs or how AuE handles vowel length in loanwords. As for RP, I see nothing on the Wikipedia entry to convince me that there is a phonemic distinction between vowels based on their lengths, though the AuE article does mention other English dialects that have the feature.

    But this is by no means a requirement. One might call the type of phonemic vowel length to which you are referring as somewhat 'defective'—i.e., the distinction between the vowels is not based entirely on vowel length. But I am talking about purely phonemic vowel length. I am not aware of any dialects of English that exhibits CR (including Scottish ones) which make phonemic vowel length distinctions.

    The vowel length distinction in CR is purely allophonic, and accompanied by a qualitative change not seen in other vowels. (That is, the qualitative change only affects the CR diphthong(s), and not other vowels even when in the same morpho-phonemic environments.)

    JE
     
  16. Youngfun

    Youngfun Senior Member

    Pekino, Ĉinujo
    Chinese/Italian - bilingual
    To give a satisfactory answer, I should be more proficient in Middle Chinese at least.
    But in Modern Mandarin (I would say most other Chinese languages too) the phonemic length distinction doesn't exist anymore.
    Actually I don't know if in Middle Chinese the distinction short/long vowel existed or not...

    Now that is interesting! Are there any conditions on this, or does it happen in all environments?[/QUOTE]
    I would say, just different phonetic evolutions, without an apparent rule.
    For example the Mandarin "jiao" [tɕiaʊ̯] has [au] vowel in Cantonese (probably pronounced [kau]), but Mandarin "hao" [xɑʊ̯] is pronouced [hou] in Cantonese.
    In my dialect (Wu) the word corresponding to Mandarin [tɕiaʊ̯] is [ko], while Mandarin [xɑʊ̯] is [hø].
    So, still not the same thing as Canadian Rising.

    My understanding is that Mandarin is more conservative in vowels (big change k>tɕ), while Cantonese is more conservative in consonants.

    Just to be clear, my post may have confused you, because I have used "becomes" and "dialects".
    But here I'm not talking about "dialect" in the English meaning (American English, British English, etc.), but rather the German meaning (Hochdeutsch, High Saxon, Low Saxon, Allemanisch, etc.).
    Mandarin, Cantonese, Wu, Min and so on are all evolutions of Old Chinese and Middle Chinese...

    While if you talk about Mandarin pronounced with heavy accent, that's another story.
    In my region, older people including my grandpa would pronounce [o] instead of [ɑʊ̯], so they would pronounce "ni hao" /ni xɑʊ̯/ (hello) as [ni ho]; and pronounce [e]~[ei] instead of [aɪ̯], so they would pronounce "guo lai" /ku̯ɔ laɪ̯/ (come) as [ku le] or [ku lei].
     
    Last edited: Dec 12, 2012
  17. berndf Moderator

    Geneva
    German (Germany)
    I don't know of any language with phonemic vowel length where this "defect" does not occur. I am thinking of
    Classical Greek - η vs. ε; ω (o-mega) vs. o (o-micron).
    Latin - Archetype of a systematic long-short vowel system: /a:/ is low central, /i:/ is high front, /u:/ high back, /e:/ mid-high front, /o:/ mid-high back (large triangle); short vowels are shifted towards centre and lowered compared to their long counterparts except /a/ (small triangle). (Chart)
    Tiberian Hebrew - The only precisely overlaying long and short vowels are Kamatz katan/gadol and even there it might be just because we don't know the difference
    Arabic - difficult to say because of the simple vowel system and therefore large number of allophonic differences but still, short and long variants have different "favourite" allophones.
    German - Essentially the same as the Latin system with additional complication of the umlauts. Slight qualitative difference between /a:/ and /a/ which is largely ignored by native speakers but plays a role in the perception of length.
    Swedish - Very similar to German; the qualitative difference between long and short a is stronger ([ɑ:] vs. [a]).
    That's why I think it is phonetic and not etymological.
     
    Last edited: Dec 12, 2012
  18. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Well, that is the purpose of this thread. If the phenomenon is phonetic, then we should find it in other languages; but the distributional pattern is exactly what we would expect if the phenomenon exists as a side-effect of GVS (and aside from Moreton and Thomas, this is a typical analysis in the literature), and so this is the best explanation if a phonetic explanation cannot be supported.

    So far I have seen no support for a phonetic explanation. Even within a single analysis the phonetic explanation usually fails to account for variations seen; for example, most AE speakers with raised /aɪ/ do not have raised /aʊ/, an issue never investigated by Moreton and Thomas.

    JE
     
  19. Youngfun

    Youngfun Senior Member

    Pekino, Ĉinujo
    Chinese/Italian - bilingual
    Some phoneticians, for example Luciano Canepari, denies that phonemic vowel length distinction exists in English. He considers the "long i" /Ii/ and the "long u" /ʊu/, i.e. diphthongs.
     
  20. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Yes; but even besides that, for almost all varieties of English that exist today, vowel length is not phonemic. What folks are calling long vowels versus short vowels are actually qualitative distinctions more than quantitative. In the proper phonetic context, long /ɛ/ can be equal to or longer than short /e/. And I don't even see evidence of actual phonemicity of vowel length in AuE, since no minimal pairs have been presented.

    Now it is true that because of the way English handles voicing on final consonants that vowel length still carries a reasonable functional load, but that does not make it phonemic; vowel length is non-phonemic in PDE, with rare exception (that I am taking on faith, having yet to see evidence of phonemicity). And as you mention, most of the tense vowels in English (and in many AE dialects, this is also becoming true of lax vowels as well) are diphthongs, which only serves to further create distinction based on quality instead of quantity, a process English has been undergoing since we first have records for it.

    JE
     
    Last edited: Dec 12, 2012
  21. berndf Moderator

    Geneva
    German (Germany)
    I think it is time now for you to explain why you think an etymological approach explains the phenomenon better. I am not really seeing your point yet.
     
  22. berndf Moderator

    Geneva
    German (Germany)
    Burn-bun. In all non-rhotic accents that have the phoneme /ʌ/, the distinction between /ə/, /ʌ/ and /ɜ:/ depends crucially on length (reduced-short-long).
     
  23. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    What makes you think these vowels are of identical quality?
     
  24. berndf Moderator

    Geneva
    German (Germany)
    /ɜ:/ ranges from [ə:] to [ɜ]; /ə/ from [ə] to [ɐ]; /ʌ/ occupies a space in the triangle [ɜ]-[ɐ]-[ʌ]. There is simply too much overlay to allow a stable phonemic distinction without the support of quantity. This is like /ɪ/ and /e:/ in Latin which were simply to close qualitatively to allow continued phonemic distinction when vowel length became non-phonemic and the two merged and we find Latin became di in Italian and de in French.
     
  25. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    ME /i:/ went through a stage during the GVS of something like /ʌi/ (the same diphthong seen in CR). As the final stage of the Great Vowel Shift, this vowel underwent a further shift down to its present Standard form of /aɪ/. But as we know, GVS acted only on long vowels.

    In most dialects, the rules for determining vowel length were not in their present predictable form, and so /ʌi/ was a long vowel as it had historically been when it was /i:/. In some dialects, however, present vowel-length rules were developing (e.g., short vowels precede voiceless consonants) and so any instance of /ʌi/ that satisfied the requirements for being short became short, and so failed to undergo the final GVS shift: creating a distributional pattern of [ʌi] in short-vowel environments and [aɪ] in long-vowel environments--a distributional pattern still seen today in CR dialects.

    If you are convinced that the matter is phonetic, then your evidence consists of showing that the process is universal (short /aɪ/ becomes [ʌi] in all languages where the proper conditions are met), which might be difficult to do since even different dialects of English don't observe the CR rule even though they may make /aɪ/ phonetically shorter in specific environments.

    I'm not convinced the phonetic explanation has to be wrong, but I am convinced that there has not yet been any evidence presented in its favor; the GVS explanation, on the other hand, is swimming with evidence to its credit.

    JE
     
  26. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Vowels can have phonetic length differences without having phonemic length differences.

    But, I feel this is going a little off-topic with the discussion of phonemic vowel length in English. It is enough to point out that the raised diphthongs of CR occur in environments that trigger short-vowels; that they are shorter than their unraised counterparts; and that this distributional pattern is the same for all CR dialects.

    The issue is: why does CR always trigger in short-vowel environments? Is it because, as Moreton and Thomas claim, diphthongs are 'dominated by the offglide' when short, or is it because the phenomenon is related to the historic GVS?

    JE
     
  27. berndf Moderator

    Geneva
    German (Germany)
    Agreed. Let's drop it.
    Ok.
    Short vowels in general have a tendency towards to centre, i.e. towards [ə]. That is what we can learn from Latin and German, I think.

    If it had an etymological background then you should find similar conditions in Scots. Scots isn't really in my centre of expertise, so I am hesitant to make a statement. But I am not aware that there are any conditions for the realizations [əʊ] or [aʊ]. I think, it is simple dialectal variation.
     
  28. berndf Moderator

    Geneva
    German (Germany)
    Your formulation suggests you assume it is a retention of an older state rather than a new development subsequent to the completion of the shift to [aɪ] in the 18th century. I wouldn't take this as a matter of course and it should be investigated.
     
  29. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    It's not centered; it's raised. Both parts of the diphthong are raised in the vowel space, and the offglide of the raised diphthong is even further from the center than the offglide of the unraised diphthong.

    I have formant values that I can post this weekend.

    And we do. Look at the variation in the same phoneme that happens as part of the Scottish Vowel Length Rule. The variation is also reported in other dialects of English (which I can list when I get to my sources).

    As the Wiki article says, the raised versions of both of these vowels occur in short-vowel environments.

    JE
     
  30. berndf Moderator

    Geneva
    German (Germany)
    C'mon. You know perfectly well what is meant by "towards [ə]", the big spider in the centre of the cobweb.;)
    Than we have to clarify what "version" of the CR were are talking of. The realizations are not uniform.
    (Wiki). The transcriptions [ʌɪ] and [ʌʊ] do not indicate rasing of the second part.
    Would be interesting. Even more interesting would be the recordings. (I hope not your own voice, that would be a grave methodological error; the speaker should not know what you are trying to analyse, otherwise hyper-corrections are all but inevitable; but you certainly know that -- just making absolutely sure).
    Thanks. That is an important clue. Do you know if the occurrence of the phenomenon in Canada is connected with immigration from Scotland?
     
  31. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    By centre I assume you mean 'center'. If you meant something else, then please tell us what that was.
    Wiki is wrong. I have sources on this and will post them when I find them (and the time); then I will edit the Wiki to reflect the sources (the sourcing in the article is really poor and it needs cleaning up).
    I don't have all the recordings, but I do have numbers. Some of this is stuff I collected as an undergrad and it has been misplaced or is incomplete. I'll see what I can find.

    Phoneticians analyze their own speech all the time when formulating hypotheses. I am not publishing books with the information; I think for informal analysis self-recordings are fine. I do also have recordings of other speakers that I am eager to check for this when I get a free moment; I also have some length data scratched out on some paper from other recordings I collected a few years back. It's informal, but will do for a forum discussion. The information I've analyzed (from myself and others) conforms to the hypothesis I've laid out in this thread.

    I'd have better information on this if my life revolved around linguistics as I'd like it to. :(

    It has been only somewhat investigated and I have read some of those investigations, but I don't remember all the conclusions. It is certainly a possibility that these are connected, but it is by no means a requirement. Once the distributional pattern for CR is established in a group of speakers it can spread into other dialects which are otherwise unrelated to the original one. (Indeed, this is true of all dialectical variations.) If the variation originates from the final stages of GVS it could be impossible to find a single geographical origin (if one ever even existed). This may be especially true given the fact that vowel length rules differ in the various CR dialects and the process of creating a CR distributional pattern from GVS involves two independent steps that could be completed to different degrees in different and unrelated dialects.

    JE
     
  32. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Or that I made a generalization. The number of environments that exist for fully studying this is small.

    Either way, the link between vowel length and CR is borne out in several examples, such as higher vs. hire, high school vs. highschool; the latter of each set being (potentially) raised.

    Non-words? The words are unusual, but by no means 'non'.

    JE
     
  33. berndf Moderator

    Geneva
    German (Germany)
    Raising of [a] is a move towards the centre of vowel space. And that happens to short vowels in may languages: low vowels are raised, high vowels are lowered, front vowels move backwards and back vowels move forward ([u:] vs. [ʊ], [i:] vs. [ɪ], and, by the same logic, [a:] vs. [ɐ]).
    Thank you. If it is too much effort, then don't. I was just curious. There are enough samples on the net.
    Sure. The question was, if immigration of Scots can be traced as the seed of the phenomenon.
    I understand; we all have day-time jobs.
     
  34. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    But the [ɪ] is also raised, putting it at . One value of Moreton and Thomas's research is the extensive acoustic analysis they did on CR for /aɪ/. They found that (1) the diphthong is raised in its entirety instead of the /a/ portion merely moving closer to /ɪ/; (2) the raised diphthong is more diphthongal than the unraised one. Their analysis demonstrates that CR is not an occurrence of typical English vowel reduction/centering. In the measurements I have done, I have only been able to confirm their results.

    It's effort all right; but fun and worth the trouble!

    If you know of a source for these samples, a link would be greatly appreciated.

    It maybe can be, but I don't think it has been.

    JE
     
  35. berndf Moderator

    Geneva
    German (Germany)
    I just browsed in historical dictionaries to find the standard pronunciation (the then equivalent to RP) of these diphthongs around 1800 as a historical reference. Walker's transcription for /aɪ/ is a2e1 and for /aʊ/ o3u3. Translated into IPA this would be /ai/~/ɑi/ (can't be distinguished in his system; note that the second part is /i/ and not /ɪ/) and /ɔʊ/, respectively. There is no indication of variants. About and bound have the same transcription of the diphthong.
     
  36. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Moreton and Thomas also transcribe the phoneme as /ai/, but they do not actually think the second element of the unraised form is /i/, as evidenced by their distinction in phonetic transcription and the formant values they collected.

    Transcriptions can be tough things to trust for such specific distinctions.

    JE
     
  37. berndf Moderator

    Geneva
    German (Germany)
    Walker explicitly says it is /i/. It is not just sloppy transcription. But that doesn't mean he's right, of course. Walker was the Jones of his days and his dictionary had some normative influence.
     
  38. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Can I ask which particular text you're looking at?
     
  39. berndf Moderator

    Geneva
    German (Germany)
    I am sorry, I thought "Walker" didn't need any explanation (like "Webster"). But maybe it is only well known among students of British English historical phonology: John Walker: A critical pronouncing dictionary. There are various editions available online from the 1790s until the 1850s. Here is one.
     
  40. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Oh I found plenty about Mr. Walker, I just couldn't find much specifically about the transcription you mentioned. I figured if we were working from the same text it might help the discussion.

    As for his renown in the U.S., you might be right. I think his name only came up a couple of times in all my linguistics courses.

    Regarding the document in your link, I am not sure how his transcription here can be too helpful. He doesn't seem to give us any indication as to the actual phonetic values of the phonemes he has transcribed. If I were giving a phonemic description (what dictionaries give) of my own CR speech, I would transcribe the /aɪ/ phoneme the same regardless of whether or not it was raised in a particular context.

    If we are looking to historical sources for guidance, my alma mater has an old book that discusses the raised value of the diphthong. I don't remember exactly what it says, but I can see if the book is still around and post the relevant portion.

    And now to make a list of all the stuff I need to find for this thread...

    JE
     
  41. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    For another interesting example of CR being conditioned by morpheme boundaries, CR dialects (typically) phonetically distinguish I scream and icecream. The former is [aɪskɹḭ:m], the latter [ʌiskɹḭ:m].

    JE
     
  42. berndf Moderator

    Geneva
    German (Germany)
    Maybe, maybe not. The concepts of phonemic vs. phonetic descriptions hadn't been developed at the time.

    But I gave you this information for a different reason, viz. as the reference realization in the early 19th century. It shows an asymmetry between the two diphthongs: The CR of /aʊ/ in short contexts could be conjectured to be a not a raising but an inhibited lowering in long contexts and the comparison with Scots makes this conjecture the more plausible. In the case of /aɪ/ is is somewhat less clear because we don't really know which sound Walker's a2 represents exactly.
     
  43. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Which is why I am hesitant to interpret his guide as being purely phonetic and having no phonemic influence, especially since we all know that people 'hear' in phonemes, not typically in phones (unless well-trained to make such distinctions).

    I don't see the distinction you're seeing in the text you linked me to. Walker uses one transcription for /aɪ/ and one for /aʊ/.

    Absolutely! We can describe the phenomenon in either set of terms. The only reason to prefer one over the other is based on which of the variants we take to be the phonemes and which the allophones; typical analysis sets the phonemes up as the 'elsewhere' variants and the allophones as the 'restricted' variants.

    JE
     
  44. berndf Moderator

    Geneva
    German (Germany)
    Here and here.

    I cannot detect and raising of the second part in those sample, but I haven't measured the formants yet. The raised versions sound pretty much like the corresponding German diphthongs (weißes Haus).
     
  45. berndf Moderator

    Geneva
    German (Germany)
    No, he transcribes the modern /aʊ/ as /ɔʊ/ which I hear here in two of the 18 samples: benzo and GordonRugg.
     
  46. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Sorry; I see what you mean now. The 'i' diphthong is transcribed with a single symbol, so I didn't consider checking the parts of the 'ou' diphthong with their corresponding monophthongs.

    Thank you for those links. I listened to some of the samples. It is hard to always hear raising of the off-glide, but a formant check should reveal whether it is there or not. It is hopefully very clear that the raised version is considerably shorter than the unraised one.

    If you are going to measure the formants, could you tell me how it is you will get the sound files into the measurement program? :) I've never had any luck getting embedded sounds downloaded and opened in my software. :(

    JE

    ABE: From your first link:

    So while the question of who brought the pattern to southern Ontario (and ultimately to western Canada) is hard to answer, historical linguists do agree on one thing: that the pattern is a fossil of the Great Vowel Shift that occurred in England in the 15th and 16th centuries. The Great Vowel Shift refers to the rearrangement of the entire English vowel system from Middle to Modern English. Prior to the shift, words like five and house were pronounced [fi:v] and [hu:s], with high vowels. The Great Vowel Shift lowered their vowels to their current low-vowel pronunciation, [fayv] and [haws]. It is believed that the diphthong-raising pattern is inherited from certain middle-English dialects in which the lowering of [i:] and [u:] stopped at the mid-vowel height in some words.

    A little over-simplified as there is no mention of the conditioning involved (i.e., vowel-length environment), but this should support my initial claim that most linguists are in agreement as to the origin of CR. Moreton and Thomas clearly disagree with this idea, but their hypothesis makes certain predictions that even they admit are not validated elsewhere (such as the appearance of CR in other languages).
     
    Last edited: Dec 14, 2012
  47. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    Here I have analyzed a few items from a New Hampshire female, early 20s. The words analyzed were: beat, bid, Rosa, bud, spot, spite (blue), spied (red).

    The length of vowels in the following word: bit, bid, spite, spied, doubt, (en)dowed. Not all of these items appear on the graph.

    As can be seen, the vowel in spite is raised in the nucleus and trajectories toward the /i/; the vowel in spied is not raised in the nucleus and trajectories toward the /ɪ/.

    NHF_LongI.jpg

    Unlike the CR studied by Moreton and Thomas and the CR in my own speech, the raised diphthong is not more diphthongal than the unraised one. The length values are as follows:


    Word Length (sec)
    bit 0.07
    bid 0.16
    spite 0.14
    spied 0.38
    doubt 0.19
    endowed 0.35


    The last two items are interesting, because Moreton and Thomas's claim that shortened diphthongs assimilate the nucleus to the offglide would predict a CR variation for this phoneme as well in this speaker given the length differences. However, there were no appreciable differences measured, the longer diphthong has F1 and F2 values of 513,2442→565,1252, while the shorter one has values of 515,2372→578,1146.

    I think this alone should be sufficient to conclude that CR is not a phonetically motivated process. Moreton and Thomas's proposed cause will work for CR dialects with raised /aʊ/, but it also predicts a raised /aʊ/ in CR dialects that don't have raising for this diphthong; prediction of this sort so solidly falsified can only spell invalidated for their phonetically-motivated CR explanation.

    Values from my own speech to come...
     
  48. berndf Moderator

    Geneva
    German (Germany)
    Sorry, I can't see any raising of /I/ in the graph, only of /a/. The /I/ is fronted but not raised.
     
  49. JuanEscritor

    JuanEscritor Senior Member

    Minnesota
    English - AE
    I never claimed it was raised. You will see in my own graph that there is definite raising of the offglide. For our NH female the offglide of the raised diphthong is brought into the vowel space by fronting; in either case, the historical value of the diphthong is produced: [əi] (or [ʌi]).

    JE
     
  50. berndf Moderator

    Geneva
    German (Germany)
    Well, you did ("the vowel in spite is raised in the nucleus and trajectories"), but never mind, your point is understood now.:)
     
    Last edited: Dec 16, 2012

Share This Page