Re the rule-list fallacy

[David Tuggy]

Martin Haspelmath wrote:

Thanks a lot, Brian, for this very lucid explanation of the issues from 
a psycholinguistic point of view! I have long shared your view that "in 
a given usage of a particular form by a given person at a given moment, 
one seldom knows whether rules or lists applied. Only if a clear 
productive overgeneralization occurs, and this is very rare, can one 
know for sure that a rule or gang effect applied" (my view is probably 
due to your influence, however indirectly!). This also makes me quite 
skeptical of "cognitive linguistics" of any sort – the adjective 
"cognitive" sounds great, especially to linguists who don't know much 
about cognition, but it probably promises more than we can deliver as 
linguists. <snip>

and Brian MacWhinney responded:
> <snip> yes, Martin I also sympathize with your wariness of the 
> application of the term "cognitive" as a magic wand for linguistic 
> analysis. I think the hope is that corpora and richer streams of data 
> recording can help us reduce this huge indeterminacy, but I can't see 
> how it would ever vanish entirely, given the complex dynamics of the 
> interplay.
To be sure, "cognitive" can be simply an impressive but empty buzzword. 
It can also be somewhat more legitimately used to describe a linguistics 
that hopes to be at least cognitively plausible and at best responsive 
to all the solid conclusions of cognitive psychology.

In the case at hand, I do not follow Martin's logic, which I understand 
to be going from (a) we can almost never know for sure if something was 
produced by rote or by rule, to (b) we as linguists cannot deliver on a 
"cognitive linguistics" of any sort. Brian's answer, as well, seems to 
suggest that, as long as we cannot make the indeterminacy vanish 
entirely, or at least reduce it greatly, we do not have a linguistic 
analysis .

What about a linguistics that would embrace the indeterminacy? What 
about one that would say precisely "we cannot, in the absence of 
empirical evidence, definitively say if something was produced by rote 
or by rule"? That would refuse to say that "since some Derived Nominals 
are related to their corresponding verbs only in an irregular fashion, 
DN's as a class are not produced by the grammar but rather listed in the 
lexicon " (my paraphrase of Newmeyer 1980 summarizing Chomsky 1967). And 
that would say that when Martin Haspelmath wrote "explanation" in his 
email of 10 June 08 at 1:49 pm, he almost certainly got the word "off 
the shelf" (i.e. from a list), but that without having measured his 
brain functions at that time, in ways we do not yet know how to do, we 
cannot know for a fact that no rules or gang effects were active in the 
process.

In the end the question of whether a structure was produced by rote or 
by rule doesn't seem, in many cases at least, to be crucial for 
communication —a speaker may use either or both and a hearer use the 
opposite mechanism or the same one or both, and adequate communication 
can take place. So why does it have to be crucial for linguistics?

If I have to choose between a linguistics that is determinate 
(absolutely predictive, yielding crisp judgements) and one that leads me 
to expect the indeterminacies that are there empirically in language, I 
prefer the latter.

--David Tuggy