(A couple of times my PC has trashed some mail and I think this might be one
item that didn't get out. --BN)
[From Bruce Nevin (970910.1326 EST)]
Three kinds of explanation that I have seen in science:
Analogy. Draw parallels between something that is well understood and the
thing to be explained. To the extent that the parallels are valid, the
well-understood thing is a model of the thing being explained. Model
building as in PCT is a refinement of analogical explanation.
Reduction. Show that something that seems complex emerges from
characteristics of simpler things that are well understood. HPCTheorists
sometimes make reductionist claims about social phenomena, some of which
have been modelled (ring and arc formation in crowd behavior), others not
(language).
Generalization. Identify and apply principles, rules, processes, etc. that
marshall the data to be described into perspicuous categories. To the extent
that the description is simpler, or is made an instance of or part of the
description of otherwise unrelated phenomena, the principles, etc. are felt
to "explain" something. Here's an instance from
Avery Andrews 970910.0939 --
Some evidence the third sentence has a processing rather than a
grammatical problem is that `center-embedded' sentences become a lot
more palatable if the scenarios they describe are sufficiently banal:the car the guy the cops are looking for is driving is a red Corolla.
Here's a counterexample, no less banal:
The car the guy the cops caught drove crashed.
I suggest the difficulty has to do with the reduced intonation of a relative
clause. A relative clause generally repeats something known, associating
familiar information with a particular word repeated in the host sentence,
where something new is being said involving that word. The repeated word is
represented by a relative pronoun who, which, that, etc. in the relative
clause. Within the relative clause, we typically lower pitch, perhaps
increase pace, are less precise with pronunciation, and so on. (The main
exception is contrastive stress -- e.g."the guy the cops *caught*".)
A center-embedded relative clause is basically a paratactic interruption. At
its end, we pick up the intonation pattern that prevailed before the
interruption. Repeated center embedding is difficult (I suggest) because we
run out of intonational means for differentiating the boundaries of the
successive relative clauses.
The counterexample sentence is more difficult than Avery's sentence not
because it is less banal, but because we can't use the monosyllabic verbs
"caught" and "drove" as carriers for a resumed intonational contour.
Avery's observation about banality possibly is related to familiarity of the
information in the relative clauses. If the interrupting clauses carry novel
information, we are less free to rush over them in a "you know about this,
I'll just remind you" kind of way.
So Avery's and my accounts are an instance of explanation by generalization.
Martin Taylor 970910 10:25--
I used the word "description" advisedly in the last sentence. In my view,
an "explanation" is neither more nor less than a way of describing a
phenomenon very succinctly, using language that is also available for
describing other phenomena.
Each kind of explanation is a description, but not all descriptions, however
succinct, are explanations. And while a final textbook rendition may be
satisfyingly succinct, brevity isn't necessarily a good guide when
explanations are in the making. (I'm thinking here of C.S. Pearce's
criticism of tidy pyramidal proofs in mathematics, after the fact
misrepresenting how people actually arrive at conclusions.)
Bruce Nevin