Sentences

[Wallace Chafe]

This is one of the places where looking at how people actually talk makes a
significant difference. There is, first of all, the difference between
grammatical sentences and prosodic sentences. The latter certainly reflect
something interesting about language processing. Beyond that, I've
suggested in various places that sentence closure (of either kind) is often
decided opportunistically on-line, while people are talking, and doesn't
necessarily reflect the boundaries of cognitively relevant units, although
it may. One kind of evidence I find particularly interesting appears in
repeated verbalizations of (more or less) the same content, where sentence
boundaries may be distributed differently in the different tellings. I
talked about this at GURT in February, but see, for example, chapter 11 of
Discourse, Consciousness, and Time, and my article Things we can learn from
repeated tellings of the same experience. Narrative Inquiry 8: 269-285
(1998). I'm tempted to suggest that linguists' preoccupation with sentences
comes above all from writing and grammatical traditions derived from
written language. To show that I'm wrong about this, one would have to
examine spoken language more carefully than most people have as yet done.