Here’s an informal and amusing observation about something I keep noticing at Starbucks. I’m a fan of the Venti (= large) Latte. Whenever I order a [vɛnti lɑteɪ], it seems that nine times out of ten, the barista repeats it back to me as [vɛnteɪ lɑteɪ]. And this is from different baristas at many different locations around northern California. Interestingly, I’ve yet to hear anyone say *[vɛnti lɑti]. I don’t usually go for the chocolate espresso, but my hunch is that if I were to order a [vɛnti mokə], the barista would probably repeat it faithfully as [vɛnti mokə], as opposed to *[vɛnteɪ mokə] or, even less likely, *[vɛntə mokə].
the barista would probably repeat it faithfully as [vɛnti mokə], as opposed to *[vɛnteɪ mokə]
Just for the sake of argument, maybe she would indeed use [vɛnteɪ mokə]? That is, maybe she doesn’t like [i] in an unstressed (or non-primary-stressed) syllable following unflapped /t/.
If you go order one you can send me the bill, because I want to find out.
May I suggest a hypothesis not involving vowel harmony? Specifically, I think this may be “model-driven hypercorrection.” Here is my (somewhat convoluted) chain of reasoning.
(a) English speakers are aware of final [eɪ] in words from French whose orthographic representation would lead them to expect [i] (negligee, soiree, puree).
(b) As a result of this and other factors, final [eI] in English is a marker of foreignness. In support of this I cite the common hypercorrect pronunciation of “lingerie” with final [eI], despite the [i] in the original French.
(c) To turn to the Starbucks case, one might say that the barista’s unconscious internal monologue for “latte” goes like this: “if this were a normal word, I’d expect [lɑɾi] or [læɾi]. But in fact it’s a fancy foreign word, so I will act accordingly and say [lɑteɪ] instead.”
(d) The inner monologue continues, “‘venti’ is even more exotic than ‘latte’ is. If I’m going for the foreign pronunciation of ‘latte’, it would be inconsistent not to do the same for ‘venti’.” So “venti” ends up hypercorrected to [vɛnteɪ] (the same hypercorrection seen in “lingerie”).
If this is right, one would expect [vɛnteɪ mokə] to occur, though perhaps less frequently than [vɛnteɪ lɑteɪ], the “model factor” not being present.
You can find more on the phonology of hypercorrection if you Google the oeuvre of Rich Janda.
Thanks, Bob and Bruce. Those are much more plausible explanations. Now I’m curious about the venti mocha hypothesis. I’ll let you know after I’ve ordered enough of them to find out (which may take a while).
When I was little, I thought “Reese’s Pieces” was supposed to be pronounced [risiz pisɨz], but I found that very hard to say, so I always said [risiz pisiz]. That’s going the opposite direction from your observation, though.
[risiz pisiz] looks like a case of aggressive reduplication (see Kie Zuraw’s work). Perhaps [vɛnteɪ lɑteɪ] could be as well, although in this case there is less similarity between the two parts.
I find the vowel harmony argument compelling, but there also may be an assimilatory effect from the common Spanish lexical entry “veinte”; that is, many an American 1st-language-English-speaking teenager is certain to have learned Spanish numbers up to 20, or so, and could be exchanging final [eI] for [i] to match the stored Spanish form. Granted, that doesn’t explain why your barista would correctly produce the vowel of the first syllable (rather than [eI]), but maybe they didn’t? Aside from the common error of assimilating the first consonant to English /v/, it’s not uncommon for non-native Spanish speakers to diphthongize where they shouldn’t (e.g., the final vowel in this case), and fail to where they should (e.g., the first vowel of Spanish [beInte]–or [bejnte]). But definitely try the mocha next time!