I saw this sign outside a storefront in San Francisco’s recently-renovated Ferry Building this past Sunday, and it got me to thinking about the English-as-a-second-language speech of my (Bolivian) Spanish-speaking family, some of whom I was visiting at the time.
I’ve only been casually observing so far, but it appears that English flapping (/t,d/ → [ɾ]) is reanalyzed as Spanish spirantization (/d/ → [ð]), such that both lettuce and let us are pronounced [ˈleðas], both latter and ladder are pronounced [laðeɹ], etc. (What I’m transcribing as [eɹ] seems to vary along a continuum between [eɹ] and [ɚ], an independently interesting but otherwise not-so-relevant point at present.)
Both English flapping and Spanish spirantization are (arguably, at least) examples of lenition, and the distributions of the lenited allophones overlap significantly: English [ɾ] occurs between two syllabic elements the first of which is stressed, and Spanish [ð] occurs after vowels and nonhomorganic consonants.
Two interesting things I’ve noticed. (Maybe others have before me; if so, my apologies — I don’t read much of the L2 phonology literature, even though there’s plenty on Spanish and English.)
First, [ɾ] is an independent phoneme in Spanish and is totally at home in the environment of English flapping. For nonalternating items like lettuce, at least, it seems to me that [ˈleɾas] would be the best adaptation, but it’s not in my family’s speech. (Apparently it is in the L2 English speech of other Spanish speakers — I recently read a cartoon by the great Argentinian cartoonist/artist/humorist Quino in which Shut up! was written SHÁRÁP!)
(I’ll try to scan in the cartoon in my future copious free time.)
Second, the reanalysis of flapping as spirantization is sensitive to the other allophonic variants of /t/. As all students of phonology know, there is a distinction between sought Ed (flapping) and saw Ted (no flapping, aspiration instead). There appears to be the same distinction in my family’s L2 speech: spirantization in the first case (sought Ed = [ˌsoˈðeð]) but not in the second (saw Ted = [ˌsoˈteð]).
With /d/, spirantization applies across the board just as it usually does in Spanish. Even though there’s a distinction in English between crusade or (flapping) and say door (no flapping), there isn’t in my family’s L2 speech: both end in [ˌsejˈðoɹ].
As I said above, this is all casual observation so far. But pretty interesting.
Is it possible that these Spanish speakers choose not use the tap and instead the fricative because they know that this is the lenited pronunciation of a stop, and they choose their own lenition of a stop, which is the fricative and not the tap, which is an independent phoneme?
Alternatively, they may not choose their tap because it isn’t phonetically identical to the English flap, in that the tap is a ballistic raising of the tongue tip from below the alveolar ridge to make brief contact, while the flap is instead an equally ballistic flinging the tongue tip forward from behind the alveolar ridge. The point of tongue contact is also different: the upper surface of the tip in the tap, but the under surface in the flap.
The problem with this second explanation is that I know I pronounce the English flap as a tap in some contexts, including the one in Eric’s example, “lettuce”. Perhaps, the possibility of another articulation is sufficient to sometimes inhibit Spanish speakers from choosing their tap as an equivalent for the English flap.
John’s first hypothesis is interesting. It would entail that L2 speakers have knowledge of the general character of the L2 process (in this case, that it’s a case of lenition) and that their approach to the L2 process is sensitive to their knowledge of phonemic categories in the L1 (in this case, the fact that stops alternate with fricatives, and that the tap is an independent phoneme). Crucially, though, they have to ignore their L1 knowledge that voiceless stops don’t alternate with (voiced) fricatives; that is, the fact that /t/ lenites in the L2 (English) must override the fact that it does not in the L1 (Spanish). Very interesting.
As for John’s alternative hypothesis, I was working under the (apparently mistaken) assumption that “flapping” is a misnomer for the English process, and that the result is always a tap, not a flap. John (or anyone): what are some examples in which the result of flapping is surely a flap?
Regarding Eric’s first comment, I think L2 speakers might be able to infer the relationship from the spelling. I agree that they have to ignore the absence of voiceless stop lenition. But that’s a mere bagatelle.
Regarding his second point, I agree that this sound is actually a tap rather than a flap most of the time. It’s reliably a flap in my speech after a rhotic, i.e. in “sorta” as compared to “soda”. This conditioning is unsurprising, as the tongue tip is retroflexed for the rhotic. So Eric isn’t mistaken; instead I overstated the case for the flap pronunciation. But the argument still goes through: if this sound is always a tap, then perhaps Spanish speakers won’t map it onto their tap. What undermines this account is that the tap and flap are nearly indistinguishable by ear, so how could they tell there are two pronunciations?
Interesting discussion!
Eric, do any of them use a tap for English /r/? If so, maybe they don’t want to neutralize lenited /t/ and /d/ with /r/.
John, would you expect the /t/ in “sorta” do be flapped (not tapped) even if the /r/ were bunched (with a lowered tip) and not retroflex?
My pronunciation of sorta is also flapped — thanks for pointing this out, John. I still have my doubts about this hypothesis, but now I have another way to teach students about flaps … except, perhaps, those with bunched /r/?
To answer Bob’s first question: as far as I can tell, nobody in my family substitutes a tap for English [ɹ] or [ɚ]. Their pronunciation of it may sound forced at times, and as I noted in this post their attempt at [ɚ] may be closer to [eɹ], but it’s definitely English-like.
In fact, during my visit my father explicitly commented a couple of times about his (younger) sister’s pronunciation of margarita (while speaking English, of course). He couldn’t understand why she would pronounce it as if it were an English word (with reduced vowels and two [ɹ]s) rather than as the Spanish word that it “truly” is (with unreduced vowels and two taps).
(I should note that the folks I’m talking about have all lived in the U.S. and/or have spoken English on a regular basis for 40 years or more, since their early 20s.)
Regarding bunched /r/ and whether it induces a flapped pronunciation, a paper by Guenther et al. (1999, in the Journal of the Acoustical Society of America) shows that speakers produce /r/ bunched after lingual consonants and retroflexed otherwise — the bunched articulation requires them to move the articulators a shorter distance following a lingual consonant. All seven speakers in their study show this variation, which suggests that the difference between bunched and retroflexed pronunciations may not be idiolectal but instead contextual. As the preceding segment is a vowel in “sorta” I’d expect a retroflexed pronunciation there.
Thanks for the citation. I just had a look, and it seems like 2 of the subjects show less contextual differences. S4 looks more like a consistent retroflexer and S6 seems like a consistent buncher.
(I’m basing this on the slope of the solid line between the anterior and middle pellets in the figures).
This is consistent with some anecdotal ultrasound observations; I’ve seen other people who seem to have a bunched configuration across the board.