Re: Evolution of Applicatives

Tom:
> From:    And Rosta <a.rosta@...>
> > Tom:
> > From:    John Cowan <cowan@...>
> > > > Eh?  Are you confused by my use of "topic"?  I don't mean it in the
> > > > linguistic sense, but rather as a synonym for "subject matter".
> > >
> > > Ah, well, we *were* talking about linguistics on a rather formal
> > > level, so I just assumed... but I still can't get any reading other
> > > with Tuesday as the patient, and not as the day.  I'm pretty sure
> > > that most English speakers will agree with me in this respect.
> >
> > I agree with John. Most English speakers massively overgenerate
> > false negatives, when it comes to acceptability judgements.
>
> I certainly won't deny this to an extent.  When I was TA an intro
> class, there were a few students who rejected topicalizations like
> "Him, I know".
And don't I know it! (A syntactician friend of mine's *wife* refused
to accept it as a possible sentence, which reduced him to apoplexy
despite his uxoriousness.) A good teacher would have the patience to
get such students to see that they're mistaken. I lack the patience,
though -- on constitutional rather than ethical grounds, I hate
performing the act of persuasion.

> But when there are many, many people, especially
> those who are trained to be introspective about their language use
> (under what conditions they do and don't use some construction)
> who reject something, I would have to say this represents a real
> feature of language variation.
But experience teaches that this is not so, when we are dealing with
constructions that are grammatical but subject to (discourse- or
lexical-) semantic constraints. Even experienced syntacticians
differ drastically in their ability to introspectively contextualize
sentences in such a way as to render them acceptable, and the
average ability level is pretty crap.

> As I mentioned in my last post to
> John, I asked around the department and sent out a survey to our
> departmental list, and out of more than 10 responses, all said
> that "Tuesday was being written on" would be ungrammatical in the
> meaning I was using it for (where "Tuesday" is a time of the weak).
I agree, fwiw. But prep passives, at least -- or especially --
prep passives with 'extraction'/promotion out of an adjunct, are
subject to a semantic constraint having to do with affectedness.
If "Tuesday was being written on" is no worse than "August is
normally gone on holiday during" (which for me is reasonably
acceptable), then that is explanation enough. If it is worse,
though, -- as I suspect it may well be -- then some further
explanation is required; my intuition is telling me that there
is a problem with stranding a bleached preposition adjunct.

> > > Again, this isn't relevant. *"The NEA was given money to by
> > > liberal activists" is grossly ungrammatical, and that's the analogous
> > > structure you're invoking.
> >
> > It's usually considered ungrammatical, but in fact I've collected
> > quite a corpus of actual nonerror examples,
>
> What criteria did you use to make sure they're not errors? I'm not
> trying to say you're wrong (I don't know), just that it's not clear
> to me how one would be systematic about this.  Wouldn't one have to
> keep a careful record of an individual's actual "error" rates and
> compare a given example to that?
I used the criterion of interrogating the utterer, who, for reasons
having to do with accessibility of data, is often me.

> Again, I must repeat my concern about statistical corpus studies.
> They can be valid, but by virtue of abstracting over speakers, it's
> not clear exactly *what* they are describing sometimes.
Oh aye, give me the intuition of a good intuiter any day, in
preference to corpus data. On the other hand, corpus data is
better than the intuition of a crap intuiter: nowadays, when
somebody says to me "One can't say that in English", I can
straightforwardly bash them with a couple of hundred googled
examples (if the construction is amenable to a google search
string), without having to fall into the apoplexies of yore.

--And.