Pragmatic Paragraphs (was:Roehampton or SIL Either/Or)
ButhFam at compuserve.com
Mon Aug 7 13:18:51 EDT 2000
Below is a proof by induction that paragraphs are not SEMANTIC but
Imagine a young child's text. 12 events/propositions. All equally formed
semantically in terms of interpropositional relations.
[[Johnny, what did you do yesterday?]]
.p. "I went to the store."
"Then I looked for candy."
"Then I went to the check-out counter."
"then I went to Judy's house."
"Then her parents called her down to play."
"then we climbed into the tree house."
"then we ate the candy."
"Then I went home."
.p. "Then I told my mother what I did."
"then I watched the Howdy Doody show."
"Then we ate dinner."
"then I to bed."
There is more than one way to CORRECTLY format the text, or to reformat the
text, and to process this as one or more units.
When telling the story, significant intonation breaks would show the
"pragmatic packaging" of the storyteller. There may be more than one
correct 'packaging' with eaxactly the same semantic and syntactic
Indentations in a written text would reveal the "pragmatic packaging" of
the storywriter. Again, there may be more than one such correct packaging.
In both cases the encoder has made a pragmatic choice to package the
overall text into smaller units for ease of processing.
The semantic interpropositional relationships are NOT the decisive
criterion. Ultimately, it is the packaging choice of the encoder that is
decisive. And yes, those choices will be reflected in surface structures.
(e.g. One of the most obvious surface packaging devices in Greek is the
demoting of propositions to participial clauses within a sentence. The old
SIL semantic analyses did not properly incorporate that though that has
somewhat changed by allowing surface structure notes to override "naturally
prominent semantic relationships".)
Having said the above, there are many criteria that will line up as
reinforcing 'isographs' that make a paragraph break 'more
appropriate'/'more probable' than another place. Some criteria, in fact
become almost mandatory (e.g. 'and it happened while crossing the river
...' in Hebrew).
So I find I agree with Cindy in letting a text speak for itself, (looking
for the criteria and allowing them to override sematnics if need be), and I
agree with Stephen in that a paragraph is not 'syntactic' per se,
yet practically I suspect that we all agree in studying and using
linguistic criteria for the marking/be they semantic or 'surface
Finally, for me, paragraphs are clearly pragmatic packages of information,
a third point on a triangle.
PS: (let's see if we can't douse someone's molotov cocktail)
>The first two pages of
>this section reads like 95 theses prepared to be nailed on the door at
>Roehampton. Levinshon cites a litany of scholars including Beekman,
>& Callow, Tomlin, and Givon in support for the notion that paragraph level
>discourse structure is Semantically indicated and that formal language
>features are only a secondary and somewhat unreliable indication of
>paragraph level boundaries.
The last sentence is true, but because paragraphs are pragmatically
indicated and are neither marked by linguistic surface structures nor by
More information about the B-Greek