Question about functional approaches

[Mark Durie]

Dear David,

some comments on your Funknet posting:

I think the notion of 'functional' vs 'dysfunctional' is confusing and
needs clarification for your distinctions to be meaningful.

I think we need to make a distinction between adequate functioning within
the time-frame of use, and attributions of function as part of causal
explanations.  Working with the former sense, one might conclude that a
structural property such as wh-islands contributes to adequate functioning,
hinders functions, or is afunctional, making no difference to linguistic
functionality.  This is thus a 3-way distinction:  your two-way distinction
is not enough.  Indeed things might be more complex:  a structure might
both help and hinder functioning, but in different ways, at the same time.

Clearly this sense of function as process is orthogonal to the innateness
question.  Some innate things are functional (e.g. legs), some are
dysfunctional (e.g. appendices), some are neutral (e.g. -I suppose- the two
little ridges joining the nostils and the lips), and many if not most
exhibit a mixture of functionality, afunctionality and dysfunctionality.

The second sense of function as causal explanation is the critical one for
the formal-functional contention.  This is the idea that something is the
way it is because of how it functions (in the first sense above).  An
example would be a claim that cars have their structure because of what
they are used for. When Chomsky has argued that the function of the heart
in no way offers an explanation for its structure, he is rejecting a causal
functional explanation, not the idea that the heart is 'functional' in the
first sense that it functions adequately to pump blood.  Clearly these two
senses of functional are related in complex ways:   a particular causal
function may have played a role in determining the evolution of an
organ--sense 2--but no longer play a part in its contemporary
functioning--sense 1--(an obsolete function), or an actual functioning role
of something may have made no contribution to its evolution (an
appropriated function).

The term 'dysfunctional' as you use it as an opposite for 'functional' is
misleading.  'Dysfunctional' is only used, as far as I know, in the first
sense, of inadequate functioning, not for lack of a causal explanation.
I.e. it is not an explanatory term.  One would not what to report Chomsky's
discussion of hearts as a claim that they are 'dysfunctional'.

As I understand it, some generativists' position on syntax is
anti-functional in both senses:   not only is the attempt at functional
(sense 2) explanation regarded as misguided, but syntax is actually claimed
to be significantly dysfunctional (or perhaps afunctional) in the first
sense.  This is the idea that language is like a Rube Goldberg device,
stuck on a stick and used for a sundial.  (Taken from an anecdote in Daniel
Dennett's 'Darwin's Dangerous Idea').

Of course any causal explanation, to avoid teleology, must refer an account
in terms of a temporal sequence of cause-and-effect, so functional
explanations, if they have any scientific substance, are inevitably
diachronic or phylogenic.  They are indeed Darwinian explanations of one
kind or another, for Darwin's achievement was to take the teleology out of
evolutionary theory.

Keeping this in mind, I believe most functionalists would regard syntax as
only partly optimized for function, due to the constant nature of
linguistic change, and the existence of competing functional constraints.
They would thus acknowledge the omnipresence of dysfunction in varying
degrees.  Indeed the very existence of dysfunctionality is essential the
functional theory of competing motivations.  Or to put it another way,
adaptation involves functional compromise.  So I don't think it is quite
right to say that functionalists are really of the 'everything is
functional' camp.

With regard to the learnability issue, I would prefer to remain almost
silent.  I will say however that any purely formal explanation begs the why
question - the question of the ontology of form.  The innatenes hypothesis
is one approach to addressing the ontology problem.  One advantage of the
this hypothesis, from a formal perspective, is that it has allowed
linguists to maintain a maximal autonomy for formal structuresm especally
if one isolates these innate principles from the effects of Darwinian
selection, by assuming that the evolution of language in humans was a
freakish accident, a bolt from the blue, as some generativists have
suggested.  A more plausible intermediate position is that the innate
capacity for learning language involves selected (i.e. functional) aspects,
as well as perhaps some features that are due to accidents of phylogeny.

The fundamental issue is not innate vs. 'functional', but what is the
contribution of adapative selection (through diachrony and phylogeny) in
accounting for the nature of language, versus the role of accidents of
evolution (either diachronic or phylogenetic).  Has language evolved, or
has it just stumbled, or been coopted, into existence?  And likewise for
individual languages.

With regard to syntactic issues - functionalists have indeed paid much less
attention to the kinds of structural syntactic properties you refer to.  R.
D. Van Valin might be a good person to contact as he has been doing work on
functional approaches to wh-islands.  There are some syntactic issues where
interaction have taken place.  For example in functionalist research on
binding, and work on linear precedence constraints.

Of course the issue of whether there are 'language-specific' learning
principles begs the question of what 'language' is that these principles
would be specific to it.  Here there are major disagreements about where to
draw the lines.  My impression of some streams of generative syntax is that
what is seriously 'in' syntax is only taken to be that to which the poverty
of the stimulus argument can be applied.  I.e. syntax is quintessentially
UG.  In parallel fashion, funtionalist grammarians have tended to focus on
syntactic properties which are more amenable to functional explanations,
especially diachronic ones, and these may tend to appear more peripheral
from the perspective of some generativists.  So, as you suggest,
grammarians of the two families tend to focus on quite different structural
properties.  I am hopeful that more interaction will come in time.  After
all, there was a period when the current rich interactions between formal
and functional work in phonology/phonetics may have seemed beyond our
horizon.

Your group might find Dennett's 'Darwin's Dangerous Idea' useful.

Mark Durie

------------------------------------
From:  Mark Durie
       Department of Linguistics and Applied Linguistics
       University of Melbourne
       Parkville 3052

       Hm  (03) 9380-5247
       Wk  (03) 9344-5191
       Fax (03) 9349-4326

M.Durie at linguistics.unimelb.edu.au
http://www.arts.unimelb.edu.au/Dept/LALX/staff/durie.html