Non-phonemic diphthongs, "r" and Spanish place names

Lazar   Friday, May 13, 2005, 04:47 GMT
<<My accent undergoes final consonant devoicing.>>

That's interesting. Where are you from, Free Free?
Kirk   Friday, May 13, 2005, 05:00 GMT
I know something that's long been common in English dialects everywhere is that in most places final (voiced) consonants devoice at least partially, but not completely (as in standard German). I haven't heard of an English dialect that completey devoices final (originally voiced) consonants.
Lazar   Friday, May 13, 2005, 05:06 GMT
I think English is the only Germanic language that preserves final consonant voicing. Maybe this is due to French influence.
Kirk   Friday, May 13, 2005, 05:49 GMT
I think the Scandinavian languages largely preserve final-consonant voicing. And, on a technical picky phonetic level, English's word-final voiced consonants can often devoice partly, but not fully like in German, where the words "rad" and "rat" sound the same. Of course, lengthened vowels before the (semi)-voiced consonants in English keep the difference clear.
Travis   Friday, May 13, 2005, 07:30 GMT
Kirk, if it is true that the North Germanic languages preserve to a significant extent final-consonant voicing, then it is very well possible that it is actually North Germanic influence on English that is responsible for the lack of full devoicing of final consonants in English. Remember that the North Germanic languages have had a significant amount of influence on the basic nature of the English language, whereas French influence has been more superficial.

As for final stops, there are other factors that allow distinguishing of them in English even if they were devoiced. First, final /p/ and /k/ may be aspirated, whereas final /b/ and /g/ do not. Secondly, in my own dialect, final /t/ may be realized as [?], whereas final /d/ does not become such; final /t/ may also aspirate when speaking forcefully or carefully, unlike final /d/, as well. And then there is the preceding vowel length thing that also would help distinguish such consonants. Hence, there is a whole range of clues which would help distinguish voiced from unvoiced stops even with full word-final devoicing. This is similar to how intervocalic /t/ and /d/ can still be distinguished, even though they both are realized as [4], due to the length of the preceding vowel differing, and in the dialect I speak at least, /aI/ changing to [@I] before /t/ but not /d/ in such conditions, even though /t/ and /d/ themselves are indistinguishable from each other.
Kirk   Friday, May 13, 2005, 08:24 GMT
Very good points and insight, Travis. I hadn't considered North Germanic's possible influence on English in that respect.

In my dialect intervocalic /t/ and /d/ are indistinguishable, as my vowels before them don't have differing qualities from each other. It makes sense, tho, since you have at least some traits of Canadian raising in your dialect. For me:

"rider" /raId@r/ --> [r\aI4@`]
"writer" /raIt@r/ -->[r\aI4@`]

So they're both the same, even tho they have underlying differences with /t/ and /d/.
Travis   Friday, May 13, 2005, 08:33 GMT
For me, it's not just the Canadian Raising, but also vowel length, as

"rider" : /"raIdr=/ --> ["raI:.4@`]
"writer" : /"raItr=/ --> ["r@I.4@`]

Even though the /d/ in "rider doesn't end up in the coda of the first syllable, it still has enough influence to noticably lengthen the vowel of the first syllable of "rider" versus the first syllable of "writer". Of course, the effects of Canadian Raising are *far* more noticable than the effects on vowel length in this case, as Canadian Raising results in obviously dissimilar vowel realizations, whereas the vowel length difference is much more subtle than it.
Travis   Friday, May 13, 2005, 08:35 GMT
A minor correction to the above post is that the instances of [r] in the realizations of "rider" and "writer" above should have been [r\] instead.
Franz   Friday, May 13, 2005, 13:37 GMT
Thank you so much for your answers. I have learned something new about NAE pronunciation. I find it most interesting that an American would use /o/ for Spanish "los" and "dos", where I as a German would take /O/. Especially so because for me the diphthong quality is important, which does not fit to the Spanish pronunciation.

One question that has not been picked up however is the question about dictionaries that use the SAMPA version without /ou/ and /ei/. Any hints would be appreciated.

An additional question: Suppose one has two different phonetic corpora using different rules which have to be unified. In my case this would be one that uses /ou/ and another which uses only /o/. At certain points this would run into problems. In my example one would not want to map /o/ to /ou/ when /r/ follows. Is there any automatic way to detect such difficulties, any programs that do that? Any literature about it?

Franz
Travis   Friday, May 13, 2005, 14:59 GMT
I just have a slight correction to make to something I said earlier. Actually, at least in the dialect here, there are two lax vowels besides a schwa which can appear in word-final position, /O/ and /{/, even though /{/ only rarely shows up in word-final position (I can only think of one word in which it does show up in word-final position, which is "yeah").

As for /o/ versus /ou/, the realization of /o/ as [oU] in at least North American English is not phonemic, so hence /o/ can be [o] or [oU] depending on context and the particular dialect in question. Whether it is actually realized as [o] is another story, but that does not affect what *phoneme* one is using. In NAE, there is no phonemic diphthong /oU/, one must remember

However, the case that you speak of is a bit more complex, because in most NAE dialects, [o] does not appear before /r/, but rather [O] is used in that position, but in Canadian English, and some more northern American English dialects, such as my own, [o] appears before /r/, and [O] does not. The question here is whether the phoneme in question is /o/, which just happens to be being realized as [O] before /r/ in most NAE dialects but not in Canadian English or some northern American English dialects, or whether it is /O/, which just happens to be being realized as [o] before /r/ in Canadian English and some northern American Englsh dialects.

I of course would be biased towards favoring /o/ in this position, due to the use of [o] here in my own dialect, but there are historical reasons to favor the use of /o/ here in some words, for example "core". In words like "core" and like, it's clear that historically a long vowel was used in this position, even though such would be realized in much of NAE as [k_hO:r\] not [k_ho:r\], and since historical vowel length distinctions have translated into modern tense/lax distinctions, in NAE. Of course, though, this is rendered irrelevant by the loss of the "horse"/"hoarse" distinction, which made it so that there was no distinction between /o/ and /O/ before /r/. Hence, one can say that whether one chooses /o/ or /O/ as the representation of the phoneme used in this position is completely arbitrary, as laxness versus tenseness is simply not distinguished before /r/ in the vast majority of NAE dialects.
Free, Free   Sunday, May 15, 2005, 06:01 GMT
''That's interesting. Where are you from, Free Free?''

Lazar,

I'm from Singapore.

bug - /bV:k/
buck - /bVk/
Doorknob   Sunday, May 15, 2005, 06:13 GMT
''I haven't heard of an English dialect that completey devoices final (originally voiced) consonants.''

Kirk:

I think some speakers of AAVE devoice final consonants, while preserving the originally allophonic vowel length:

bid - /bI:t/
bit - /bIt/
Observer   Tuesday, May 17, 2005, 03:10 GMT
"Well, good thing my life and the way I speak aren't bound by prescriptivists whose narrow definitions of acceptable speech are, surprise, surprise, based off of their own speech! Funny how that works. Mxsmanic, people would take your arguments more seriously if you stopped implying or outright stating that other (native!) speakers' speech patterns were inferior or "substandard.""

I think a lot of this also has to do with him living his life almost solely out of textbooks.
Kirk   Tuesday, May 17, 2005, 03:16 GMT
Thanks for your comments, Free Free and Doorknob. I know several people from Singapore here and come to think of it I think they do devoice final consonants, while retaining the vowel length.
Jim   Tuesday, May 17, 2005, 08:15 GMT
Let me just rephrase Mxsmanic's rantings.

There are only three phonemic diphthongs in Mxsmanic's dialect of English: those heard in "eye", "now" and "toy", respectively ... "respectively"? Many pronunciations of English include a lot of other diphthongs, but they're not phonemic for Mxsmanic and so are not important for ESL.

Many phonetic transcriptions of English use /e/ for the vowel in "get" which is correct in some dialects. This leads to no end of confusion when teaching ESL students who think they know how to use the IPA correctly but in reality have only been exposed to the type of nonsense that Mxsmanic is pushing. These English transcribers who do this usually put /eI/ routinely in place of /e/ which lo and behold is actually correct in some dialects, e.g. RP. The distinction is in the pure vowel in Mxsmanic's dialect but it's the presence or absence of a diphthong in others.