LINGUIST List 4.20

Mon 18 Jan 2093

Disc: Iconicity and Functionalism

Editor for this issue: <>


  1. , iconicity, functionalism, generative grammar

Message 1: iconicity, functionalism, generative grammar

Date: Mon, 18 Jan 1993 15:58:34 iconicity, functionalism, generative grammar
From: <>
Subject: iconicity, functionalism, generative grammar

In a recent article (Language 68.4:756-796), Frederick Newmeyer discusses
functionalist work on iconicity in language and its relation to generative
grammar. Since debates between generativists and functionalists are rare in
publications and virtually non-existent at conferences, maybe LINGUIST would
be a good place for such discussions. I liked Newmeyer's article, but I would
like to make a few comments on it from a functionalist point of view.
 Briefly, Newmeyer distinguishes three basic claims in iconicity research:
(i) grammatical structure is an iconic reflection of conceptual structure;
(ii) iconic principles govern speakers' choices of structurally available
 options in discourse; and
(iii) structural options that reflect discourse-iconic principles become
He then argues that (i) has been built into standard versions of generative
grammar, that (ii) is irrelevant to generative grammar, and that (iii) poses
no challenge to the autonomy thesis of generative grammar.
 I basically agree with what Newmeyer says, but I don't think that his
apologetic enterprise is successful, because he only shows that the generative
approach is POSSIBLE in view of iconicity, whereas functionalists have claimed
that their approach is SUPERIOR.
 The reason why functionalists consider iconicity so important is that it is
an extremely general extragrammatical principle which accounts for a wide
variety of facts and achieves a higher level of explanatory adequacy than
purely grammatical principles. Therefore Newmeyer's insistence that standard
generative grammar 'captures' or 'represents' iconicity is beside the point:
This only proves that it is descriptively adequate, but the generalization
that the standard mechanisms of grammar can be derived from iconicity has
been missed.
 Of course, "performance iconicity" is irrelevant to generative grammar, but
the reverse of this, the fact that generative grammar is also irrelevant to a
lot of performance questions, is not a point in favor of it. On the contrary,
it seems clear that functionalist theories which account for performance and
competence using the same principles (e.g. Hawkins 1990 in LI) are more
general and therefor superior in terms of explanatory adequacy.
 Regarding the grammaticalization of discourse principles ('grammars code
best what speakers do most'), Newmeyer admits that functionalist theories are
on the right track, while generative grammar has nothing to say on this
matter. But he should also draw conclusions from this, e.g. exhort his fellow
Chomskyans to follow his example and study the functionalists' work carefully
to avoid looking for intragrammatical explanations of phenomena that have been
satisfactorily explained in extragrammatical terms.
 While all but a few extreme functionalists would certainly subscribe to the
weak version of the autonomy thesis that Newmeyer defends, and all but a few
extreme generativists would admit that at least some functionalist explanations
are valid and of interest, the main problem is that each side wants to MAXIMIZE
the phenomena that fall in its area of competence, instead of trying to
cooperate and figure out which phenomena can only be treated
grammar-internally, and which phenomena are susceptible to a deeper
extragrammatical explanation.

Martin Haspelmath (Free University of Berlin)
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue