Page 57 of the Dec. 25, 2006/Jan. 1, 2007 New Yorker, in a piece by Julian Barnes called “The past conditional“:
In the car on the way back to London, we had an–to me–even more peculiar exchange about my niece and her boyfriend.
Two theories:
- For Julian Barnes or an editor, a/an allomorph choice can skip over a parenthetical.
- It’s just an error; Barnes and editors would have changed it if they’d noticed it. Maybe Barnes wrote an even more peculiar exchange and inserted to me later, neglecting to change an to a.
Still, it makes me wonder. Are there English speakers for whom the choice between a and an can (or must??) ignore, in some circumstances, what immediately follows? If so, what are the syntactic or prosodic conditions?
And if not–if all English speakers would consider the above example to be an error–how common are speech errors in which the choice between a and an gets locked in before the speaker decides to insert some more material? Does the error’s frequency vary as a function of syntax or prosody? I pose these as serious questions: maybe someone has looked at this, if not for a/an then maybe for a similar case.
What do you think of a/an before um and other hesitations? I can’t decide what I think about these (imagine the following as fairly fluent utterances):
- It’s a(n), um, strong argument.
- It’s a(n), um, uneven surface.
(I can imagine at least three possibilities: always use a before um; always use an before um; always act as though the um weren’t there, assuming you’ve already got the next word lined up in your speech plan.)
Unfortunately I cannot provide a comment on the English. But here is a similar (?) example. It is claimed that in Florentine Italian raddoppiamento fonosintattico skips the parenthetical (even if it’s a long one). Cf. the following quote (Loporcaro, Michele (1996) Lengthening and raddoppiamento. In The Dialects of Italy, ed. Martin Maiden & Mair Parry, p. 44)
te grede:vi d’Esse bbO:no a kkome ddi ffrega ttutti ‘you thought you could – how to say? – trick everybody’ (intercalated ‘kome ddi’ does not block RF)
That is, Loporcaro claims that the doubling in ffrega is due to the fact that but for the parenthetical, it would happen thanks to the preceding a (where it is completely regular). Unfortunately, this particular example appears flawed, since the preceding ddi is lexically of course dir(e), and Loporcaro himself insists on distinguishing (morphophonological) raddoppiamento and (postlexical) sandhi consonant assimilation. However, I guess the comment in parentheses did not appear out of thin air.
Sounds like it’s time for some kind of experiment, Kie … and I wonder if somebody’s done something along these lines already.
Just to add something extra into the mix: sometimes in my writing, I find reason to parenthesize a (superfluous) modifier — as I just did. But if I’m using a/an as the determiner and the modifier and noun each begin with a different type of sound, I can’t decide which to use (a or an) and so I rephrase what I was writing in order to avoid it! So:
a (superfluous) modifier — OK
an (extraneous) adjective — OK
?? a/an (superfluous) adjective
?? a/an (extraneous) modifier
Pavel, the raddoppiamento case is very intriguing, though as you point out the example given is confusing. I wonder if raddoppiamento can happen even after a parenthetical that ends in a consonant?!
And Eric, that is interesting. It never even occurred to me to be uncomfortable about these cases (I just go with the surface word order). Do you think you’d have the same hesistation in speech?
I’m about to post a follow-up…
Shoot, I meant to respond to your spoken examples. Before ‘um’, I definitely need ‘an’, but if the following noun begins with a consonant, I’m not sure. I asked Karen, and she pointed out that you can pronounce ‘a’ as [eɪ:] (as opposed to [ə]) and get away with it before ‘um’, which is what I suppose I’d do. (But needless to add, it’s hard to tell these things introspectively.)
I strongly suspect that written examples may suffer from either one of two issues that you may or may not be interested in. First, the writer/editor may (like me) just have some quirky sense of what ‘sounds’ (‘looks’?) right. Second, the writer may have added the parenthetical after having written an initial version without it, and failed to spot it on a later proofreading. (I’ve found examples of this sort of thing in my own writing when I’m proofreading, so I imagine I might sometimes miss an example or two.)
Now on to read your follow-up…
I think I agree with Karen’s intuition, though my intuitions feel very unreliable here (if that meta-intuition can be trusted…). Strange that it works before um but not before a real word: *that’s [eɪ:] interesting point there.
I’m sure you’re right about the written cases–insertion of the parenthetical in a later draft can just produce an error. In the case of the parenthesized modifiers you mention, I wonder if uncertainty about the written string’s relationship to its spoken form also plays a role. We often write things like as many authors (Smith 1950, Jones 1960) have pointed out that are awkward to read aloud; I’m not sure in that case if I would pronounce the material in parentheses or not.