Yesterday I threatened my introductory phonetics class by telling them that by taking my class, they were consigning themselves to a torturous life. All of a sudden, anything anyone says will become fodder for observation and analysis! I told them that they’d find themselves sitting on airplanes, thinking about things like what I’m about to write.
Say the word “continental”. Quick! Write it down in IPA before moving on!
{blank space here to distract you}
Now say the word “sentimental”, and write that one down in IPA too.
I flew on Continental Airlines 3 times this summer. During the second flight, I realized that the flight attendant was saying [kɑnɪnɛnəl]. I tend to say that word as [kɑntɪnɛntəl], but I understand that “Continental” is probably very a frequent lexical item for the flight attendant, and her pronunciation made sense.
But then it occurred to me that while one can say either [kɑnɪnɛnəl] or [kɑntɪnɛntəl], it would be very awkward to say or hear [kɑnɪnɛntəl] or [kɑntɪnɛnəl]. The same thing goes for “sentimental”, although this word is somewhat more frequent for me and I say [sɛnɪmɛnəl], without the released /t/s.
At first I was a little puzzled about why /t/ had to either appear in both places or be deleted from both, but now I’m attributing it to the stress environment. Most likely it’s possible to either delete the /t/ or release it in these unstressed positions, but whatever rule is applied, it must be applied consistently. It’s interesting, of course, that frequency also appears to play a role in what rule is chosen.
Well, once they take phonology too, they’ll not only torture themselves by noticing these comparisons, but hopefully they’ll also be able to figure out the right analysis.
Bert Vaux and I have a paper (well, kind of) on iterativity & optionality that touches on this issue. I think this would be an example of what we call type 2a optionality, where there is an option but it applies over the whole word rather than each circumstance individually.
I, a native speaker of Californian, would probably use [kɑʔnɛnl], especially in compounds like intercontinental. Gotta keep those ns from running together. Sentimental, however, remains glottal-stop-less.
Are you sure the reduced variants are not nasal flaps? In my own speech, sentimental [sɛɾ̃əmɛɾ̃l̩] has two nasal flaps (I agree with Bridget that flapping just one occurrence would be quite awkward at the very least). In continental, the medial n is typically syllabic, which triggers glottalization of the first /t/ (the “button” rule) ([kɑnʔn̩ɛɾ̃l̩]). The fact that it’s the syllabicity of the intervening /n/ that matters can be shown by a minimal pair: Centinela (monomorphemic) is eligible for a syllabic n, and the first /t/ is typically glottalized: [sɛ̃nʔn̩ɛlə]. By contrast, the name Santa Nella has a boundary, and remained unreduced or gets a nasal flap: [sæɾ̃ənɛlə].
I think the optionality must be controlled at some higher level than the flappinɡ rule itself, though. For me, at least, the syllabic nasal/glottalization can apply without flapping ([kɑnʔn̩ɛntl̩]), and both can be suppressed in very careful speech ([kɑntɨnɛntl̩]), but flapping T2 without the syllabic nasal/glottalization of T1 is awkward at best (*[kɑntɨnɛɾ̃l̩]).
I guess the /nt/ sequences could be realized as nasal flaps (although that really doesn’t seem right for my speech*), but in any case, the fact that you also agree that both nt1 and nt2 have to be realized the same way is what’s really interesting, I think.
*It doesn’t seem right because I get a distinct /I/ quality for the second vowel in continental, not a syllabic nasal. And, as I mentioned in the initial post, I actually prefer the pronunciation with released /t/ in my casual—not just careful—productions. Not so for ‘sentimental’, however.
Bridget and I did indeed discuss this in our optionality paper, and I presented the relevant facts at the LSA in 2003. In my idiolect (and, I think, that of many other Americans) it is possible to optionally get a nasal flap as the outcome of either of the /nt/ sequences in Continental. This is a case of what I called “sequential optionality”, and what Jason RIggle in a subsequent paper calls “local optionality”.
If your (= Lisa’s and Adam’s) grammars in fact link the treatment of the two sequences, then it could be connected to the phenomenon of all or nothing application that we find in e.g. Warao labial voicing (Howard 1972:87).
And regarding Centinela, an interesting pair is potent (with glottalized t before syllabic n) vs. impotent (no glottalization before syllabic n). I discussed this in an LSA paper on aspiration, flapping, and glottalization about 5-6 years ago.
Nobody has commented on the fact that ‘Continental’, in at least a few of the transcriptions, lacks a nasalized vowel in the first syllable. This is a general process in casual American English–in words like ‘sentence’, ‘mountain’ and so on. For me, for example, they are
[sɛʔn̩s]
and
[maʊʔn̩]
Although I’ve been aware of this for quite a while, I’ve never seen it discussed anywhere.
Geoff