LINGUIST List 12.1980

Wed Aug 8 2001

Disc: Semantics and Pragmatics of to-contraction

Editor for this issue: Karen Milligan <karenlinguistlist.org>

Directory

1. Dan Everett, response to hudson and mills on to-contraction
2. Stephen Wilson, More on wanna contraction

Dick Hudson's reply, namely, that 'wanna' has the same distribution as
'want to' may be right. I never claimed that it did not have the same
distribution and my original posting provided an analysis that assumed
that it did. On the other hand, there is something going on here that
is clearly not syntactic, at least 'clearly' according to the data I
am aware of.

As to Mills reply that he 'apologizes' to me for not getting the
judgments, why on earth would anyone think it would bother me that he
or anyone else didn't get these judgments? The point is that some
people do. For the people to whom all examples are acceptable, there
is probably an easy solution. But those are not the dialects in
question.

The original data was noticed by going through many spectrograms (not
systematically for this particular problem, I admit) in the CMU SPHINX
data base and noticing that 'get +to' almost always comes out with far
less reduction of the 'to' than 'got + to', etc. There is a robust
fact here. And I have tested the judgment with many natives speakers,
mostly from the Midwestern US. Again, though, not systematically. That
certainly needs to be done, following something like the methodology
suggested in Wayne Cowart's book, Experimental Syntax.

So, if Dick Hudson is right, that 'we're still in the stage where
"wanna" is an alternative to "want to"' then we are also still in the
stage where a syntactic account seems less than promising.

Dan Everett

A few comments on some of the interesting postings on
to-contraction...

I think Angus Grieve-Smith's suggestion (LL 12.1971) that a fccount
following Bybee's approach is probably the best way to capture most of
the data.

A frequency-account can also capture the following contrast (in my
dialect at least):

(1) I wanna go.
(2) ?? I intenna go. (<< intend to)
(3) ?? I planna go. (<< plan to)

Structurally these ought to be identical and semantically they are
quite similar. Yet only the high-frequency 'want' readily allows
contraction.

However, Dan Everett has written to me that in fact (2) and (3) are
grammatical for him. Likewise, Carl Mills (LL 12.1975) finds many of
the starred examples (e.g. I getta go) acceptable, along with the
following classically ungrammatical sentence:

(4) * Who do you wanna go?

In this connection, it's interesting to note that Crain & Thornton
(Investigations in Universal Grammar, 1998, ch. 21) discuss an
experiment testing children's knowledge of the putative constraint
against contraction across traces. Remarkably, 3 of the 14 children
included in the final analysis (ages 3;6 - 5;5) produced 1 or more
sentences like (4). One child produced 3 of the 6 elicited utterances
along the lines of (4) with contraction.

It's not clear how to reconcile the trace-based analysis with this
finding (C&T offer no explanation), or with Carl Mills' observations.

The variability in judgments suggests that no single synchronic factor
is going to provide a nice answer. Rather this seems to be
grammaticalization in progress.

But Dick Hudson (LL 12.1975) observes that 'wanna' is not a modal
because it takes 3sg agreement and do-support, unlike true modals.

These observations do establish that 'wanna' has not joined the
existing category of modals. However, Bybee & Dahl (Creation of tense
& aspect systems, 1989) point out that "new grams [=grammatical
words/morphemes] are rarely added to existing closed classes, rather,
as they grammaticize, they create new closed classes" (p. 60). Krug
(Frequency, iconicity, categorization, 2001) calls 'wanna' an
"emerging modal", i.e. a new grammatical category. A possible
candidate for the same category is "hafta" which also shows 3sg
agreement and do-support. As Krug points out, following Bybee & Dahl,
the fact that true modals do not show do-support follows from their
"strong entrenchment" which made them immune to this change. We
wouldn't expect new modal-like categories to lack do-support, so the
behavior of 'wanna' is expected.

So while 'wanna' has not gone as far as the true modals, I think the
fact that it allows contraction much more readily than most other
verbs suggests that it has become grammaticalized as an "emerging
modal".

best,

Stephen Wilson
Department of Linguistics, UCLA