Tom Limoncelli has passed on to me a query from a friend of his:
I have found myself running syllables together in unexpected ways:
instead of “hobo beans” I might say “hobob eans”
or instead of “Jon Bon Jovi” “Jon Bonge Ovi”
or instead of “soup and sandwich” “soups and which” (which is another set of problems, perhaps)
Do you or any linguists of your acquaintance know of this phenomenon?
It happens only orally, and not in writing.
The short answer is: metanalysis, a.k.a. recutting, though the third example seems to involve omission of and followed by recutting. But there’s more to be said here.
A number of different phenomena have been treated together under these headings. First, there are cases where an expression is recut morphologically: hamburger interpreted as ham + burger, leading to the creation of nouns like cheeseburger and to a free-standing noun burger.
Then there are phonological recuttings, resyllabifications, both between words and within words. The major tendency here is to move segments from a less accented syllable to a more accented syllable, as in these between-word examples I noted on ADS-L back in 2003:
final /s/ in this and las’ (for last) moved into the second word in some two-word time expressions, as in this morning, this evening and last night;
public radio announcer Bob Edwards pronouncing his name with the final /b/ of Bob moved to the beginning of Edwards;
public radio announcer Sandip Roy pronouncing his name with the final /p/ of Sandip moved to the beginning of Roy;
an ad for the movie Mystic River in which the final /k/ of Mystic is moved to the beginning of River.
The Bedwards, Proy, and Criver productions were particularly easy to detect, because of the allophonics involved: Bob Edwards tends to have ingressive variants for syllable-initial (but not syllable-final) /b/; and /p/ and /k/ have aspirated variants at the beginning of accented syllables (but not at the end of syllables) position.
(The resyllabifications that Limoncelli’s friend reported would not be so easy to hear, and at least two of them move a consonant from a more accented to a less accented syllable, so I’m a bit suspicious of the report.)
These resyllabifications take place in expressions that are familiar to the speaker (if only by repetition, as in the Misty Criver case) and are likely to be treated as phonological words.
The familiarity effect shows up within words as well. As Dennis Preston claimed in ADS-L, resyllabification is “how you can identify real cheeseheads: Us wis-con-sin, Them wi-scon-sin”. Again, the difference is easy to hear, because We have an aspirated variant of the /k/, while They have an unaspirated variant.
The resyllabifications I’ve looked at so far are phonetic readjustments. But sometimes they can be lexicalized; that is, they can result in changes in the phonological content of lexical items.
Consider the English indefinite article, which has two shapes, a and an, the choice between them depending on the following phonological context, and which also forms a prosodic unit with the following material. As a result, it can be unclear whether an /n/ belongs to the article or to a following noun, and English has seen phonological re-shapings of a number of nouns, going in both directions: apron has lost an initial n, while newt has picked one up.
(Some writers use the term metanalysis to refer only to historical changes.)