Phonemic Transcriptions in dictionaries.

Tom   Saturday, November 13, 2004, 20:20 GMT
Let me put it this way: his-her-pl.wav sounds like "see". It doesn't sound like "say".
If this is the reference recording for [e], then [e] is most certainly not an allophone of [eI].
If you say that the vowel in his-her-pl.wav and [i] are very different, then, well, our ears must be very different.
Mi5 Mick   Sunday, November 14, 2004, 01:50 GMT
This is how a Scot pronounces: "Okay mate":
~ o[ke] [met] <- This is the transcription: do you agree? I can't speak for Americans, but it doesn't sound very American to me. Is it an allophone of [eI]?

If I edit out the "O" and don't know it's a Scot speaking, it sounds like: "Care met" ~ [ke:] [met]

I don't hear the [i] you hear in these words. To me, this is an [i]:!s.mp3
Mxsmanic   Sunday, November 14, 2004, 02:13 GMT
I get [okemet] for "Okay, mate." The [t] has no audible release, which is just an observation. The vowel is not diphthonged (and neither is the IPA model for /e/ for American English, surprisingly).

Part of the problem may be that some regional pronunciations aren't right on the mark for the IPA values of /e/ and /E/. The /E/ might be a bit closer than the IPA chart implies, or the /e/ might be a bit more open. Some people may not hear /e/ unless it's diphthonged. This can be a source of mutual incomprehension with other regional pronunciations for short utterances or isolated words. In context, it's easier to correctly guess what is being said, which is why people with different pronunciations still manage to communicate most of the time. But if there are conflicts in the allophones for different phonemes between pronunciations there WILL be misunderstanding; it is inevitable.

If I didn't know that the recording was "Okay, mate," I'd guess that it wasn't English at all. In real life, I'd probably only understand it within an appropriate context.

I agree that I hear [i] in kit_lipa_witals.mp3.
Mi5 Mick   Sunday, November 14, 2004, 02:14 GMT
*Edit: Is it an allophone of [eI] in your American accent, Tom?
Mi5 Mick   Sunday, November 14, 2004, 02:40 GMT
Here's more of the "okay mate" passage to put it into context:
Tom   Sunday, November 14, 2004, 17:30 GMT
I have to say the [e] in Mi5 Mick's recording sounds like an allophone of [eI].
However, the [e] in this recording is very different from the one in Mxsmanic's recording (which, to me, sounds like an allophone of [i]).
Mi5 Mick   Monday, November 15, 2004, 05:08 GMT
Maybe because you perceive the one in my recording as a long vowel, so it matches the long quality of a diphthong. Whereas, the other seems cut a bit short and needs a consonant tacked on the end to make sense in English. But both sound the same to me.
Mxsmanic   Monday, November 15, 2004, 05:37 GMT
They look the same in spectrograms.
Jim   Monday, November 15, 2004, 08:00 GMT
I agree with Tom. The "say" in the "his-her-pl" wave file does sound like "see".

The [e] and [E] in Mick's "eE" wave file both sound like allophones of /e/ to me. However, in his "Okay Mate" wave file I can recognise the [e]s to be allophones of /ei/ but this is mostly due to context. It actually sounds like [oke:me?].

Note: the first vowel certainly sounds long to me making it easier to distinguish from /e/. It does, however make it sound like a /e../ though. I agree with Mick: the "Care Met" file does sound like "care met" if we edit out the "mate" I suppose the "kay" would sound even more like "care".

What should be apparent is that things are never black and white. The phone [e] is an allophone of /e/ or of /ei/ depending on your accent. The phone [e:] is an allophone of /e../ or of /ei/ depending on your accent. You find similar things when you consider other sets of phonemes aswell. There are thus no "correct" IPA symbols until you decide upon a "correct" accent but in truth there cannot be a "correct" accent.

This is the problem with Mxsmanic's insistance that phonemic transcriptions be based on the "correct" IPA symbols. If there are no correct IPA symbols (and as I keep stressing, there aren't), then you can't base a phonemic transcription on them. What's the big deal about the IPA anyway? We have ordinary English orthography we can all agree on the usual spellings. How not use this, as Pitman did in his ITA? Who needs Greek letters when we've got more familiar Roman ones?

At the end of the day, you could probably get away with pronouncing /e/, /ei/ and /e../ all as [e] and still be understood. They don't form that many minimal pairs (or triplets) and where they do, context will usually serve to indicate which one was the intended phoneme. An important part of context, though, is accent; if it's a Scottish accent you're hearing, you might be expecting [e] or [e:] for /ei/ but not if it's an Aussie or Kiwi one. The problem is what are you expecting in a foreign accent?

ESL teachers will tend to find other areas more important to deal with than this I'm sure. If a student were using [e] for /ei/ but could still make him/herself understood, no, I wouldn't dwell on it either. However, I wouldn't go so far as to recomend [e] for I'm aware that this is also an allophone of /e/.
Jim   Monday, November 15, 2004, 08:04 GMT
Typo: "Why not use this," not "How not use this,".
Paul Waters   Monday, November 15, 2004, 23:12 GMT
Quote-''"One can ask if the distinction reported by Paul is phonemic."

Quote-''One can ask whether this fellow is telling the truth at all. For all we know he may not be Welsh and he may not speak like this but what he writes is not complete phantasy''.

Jim, well couldn't you then being doing the same thing about saying that [a:] and [^] having the same position in Australia?

I'm not wrong about ''vane'', ''vain'' and ''vein'' being distinct anymore than you are wrong about [a:] and [^] having the same position.

I don't pronounce [a:] and [^] in the same position, so does that mean I should think that you're not telling the truth when you say you do? I don't think so.

Not everyone has the same accent, Jim. There are different accents out there.
Jim   Tuesday, November 16, 2004, 00:31 GMT
Mxsmanic   Tuesday, November 16, 2004, 06:17 GMT
If you want hard data, look at spectrograms. The IPA assigns similar symbols to sounds with similar spectrograms. What they "sound like" to individual various human ears is not terribly important, since human ears vary so much.
Mi5 Mick   Wednesday, November 17, 2004, 00:25 GMT
I want hard data though I don't know how to go about it with your system. Can you look at the spectro of "egg" in the following? (I hear [eg] or very close to this):

Mxsmanic   Wednesday, November 17, 2004, 05:29 GMT
Looks like a moderately diphthonged vowel: [eig]. For an open vowel, listen to the instances of "there" in the same recording. These are relatively pure [E], although almost all of the vowels in the recording are diphthonged to some extent.