drawring, etc.

[Laurence Horn]

At 11:56 AM -0500 1/23/05, Beverly Flanigan wrote:
>In first language acquisition (and early second also), children typically
>use unmarked 'a' in all contexts, adding liaison /n/ before vowels quite
>late (4 or 5, if memory serves away from my books).  In some "subdialects"
>(I don't like the word, but OK for now) the lack of liaison may persist
>throughout adulthood, esp. in informal contexts (my Baltimorean ex-husband,
>a comp. lit scholar, was a good example of this).  Witness 'a apple', 'a
>egg', 'a angel', etc.  I have a Peanuts cartoon in which Charlie Brown
>tries to instruct Sally in its use, after which she writes in a letter to
>someone or other, "Dear Sir, You are an ucklehead."  And of course, there's
>'a napron' --> 'an apron' and Shakespeare's 'nuncle' (a reverse
>reanalysis), both of which suggest a psycholinguistic awareness of the "a
>to an before vowels" rule dating at least to Early Modern English.  Ditto
>for 'my to mine', 'thy to thine', etc.--even my American grad students are
>amazed when they realize that this is what causes the pronoun variation in
>the Lord's Prayer, not to mention Shakespeare.
>
>Beverly Flanigan

I'm not sure what the metanalyses show about directionality, given
that as you mention there are a number of examples in each direction:

an + V > a + nV

a newt (cf. _eft_)
a nuncle (didn't take)
a n
a (whole) nother thing
a nonce (word)

vs.

a + nV > an + V

an apron
an orange
an adder (the snake)
an umpire

This shows speakers are aware that the "@nV-" sequence for indef.
art. + CN can reflect either phonological structure, but not that
there's an insertion rule vs. a deletion rule for the article.  (In
fact, there's a similar allomorphic alternation for the Greek
privative prefix:  anechoic, anhydrous, anaerobic, anesthetic vs.
amoral, aphasic, asymptomatic.)  The best evidence for deletion
rather than insertion is that it's easier to predict what isn't there
before consonants than what pops up before vowels, especially given
the similarity between the article case and the prefix case, although
the latter is frozen (whence the fixed nature of "anhydrous" vs. the
online alternation between "a/an historical accident" depending on
whether the /h/ is pronounced).  The best evidence for insertion is
the dialectal and developmental evidence noted by Wilson and
Beverly--the frequently attested occurrence of "a apple" and the like
vs. the non-occurrence of "an pear".

(Well, yes, there is the very small group of speakers I mentioned
earlier--including one of my undergraduates--who do have "an" before
consonants, but since this involves a specific semantic value and
prosody ("I'm (just) *an* person", meaning 'only one') rather than a
variant of the indefinite article as such, I think we can discount it
heavily.)

larry