dlevere at ilstu.edu
Wed Dec 23 16:06:50 UTC 2009
Let me add that I agree completely with Tom. I agree that the sentence-discourse dichotomy is largely a false one. In fact, when an MIT linguist's book on antecedent contained deletion came out several years ago, I wrote him to point out that many of the same phenomena operate across sentence boundaries, so that the solution to the problems would have to be discourse-based.
In fact, the hypothesis that I have reacted to most over the past couple of years, in my syntax work at least, has been the sentence-based proposals (since they implicitly assume the Minimalist Program) of Hauser, Chomsky, and Fitch (2002).
My most recent statement on that particular research program was posted to my website yesterday: http://www.llc.ilstu.edu/dlevere/docs/Another%20reply%20to%20Nevins.pdf
And now for a shameless plug. Yesterday NPR joined the London Sunday Times and Blackwell's Booksellers in choosing my book as one of the 'best books of 2009': http://www.npr.org/templates/story/story.php?storyId=121515579
On 23 Dec 2009, at 10:56, Tom Givon wrote:
> Dear FUNK people,
> I thought maybe just a short reply to one item in Dan's summary might be useful.
> An important principle known to all of us, certainly to Dan, is that different languages perform the same (or similar) communicative functions by different structural means. This has been, at least to my understanding, the very gist of cross-language grammatical typology. Now, if we restrict our discussion of "long distance relations" to only those found in sentence-level (i.e. complex-clause-level) constructions, we devalue this principle in two important ways.
> First synchronically: Languages in which V-complements and REL-clauses are still handles paratactically by fiat then "lack long-distance relations". They perform the same communicative functions, observe the same long-distance government-and-control constrains, but across adjacent paratactic chunks (clause-chains). So the study of government-and-control would be deprived of important typological data, which are "outside the pale". This is precisely how Chomsky could possibly view "pro-drop" as typologically weird subtraction of grammar, rather than as what it really is: the most natural & wide-spread communicative device in human (and non-human) communication.
> Second and (to me) more important--diachronically: As far as I can see, all syntactically-complex constructions are diachronic condensations of paratactic constructions that perform the same communicative function--albeit more ambiguously. Two chapters in my "Genesis of Syntactic Complexity"(2009, chs 4, 5) document these processes exhaustively for diachrony, and two others (chs 8,9) for child language. Marianne Mithun's recent work on the early stage of this process of "condensation"--merger of intonation contours (combined with changes in world-level intonation, most commonly de-stressing) is particularly relevant here. What many works on this topic show, I think, is that the change from parataxis to syntaxis in the genesis of complex syntactic construction is gradual and often subtle. So that following Dan's implicit distinction between "sentence level syntax" and discourse/clause-chaining is not going to be all that easy. But more to the point--is it going to be desirable? Illuminating? As Paul Hopper noted in his early response to Dan, the usage frequency difference is often no more than, say, 95% vs. 100%. To give an example: The use of zero-anaphora subject in clause-chained discourse is predictable at the 95% level (co-referent found in adjacent clause). In REL-clauses & (equi-subject) V-complements it is 100%. So while we try to understand (communicatively, cognitively, neurologically, whatever) why 100% predictability of syntacticized constructions is important (i.e. adaptive)--that is, what is the sadaptive impetus for the rise of morpho-syntax?-- it would be counter-productive to draw the kind of hard-&-fast distinction that Dan would have us draw.
> Best, TG
More information about the Funknet