Rule-List Fallacy

Matthew Dryer dryer at
Tue Jun 17 20:05:15 UTC 2008

I'm joining this discussion late, after being off of email for over a week.

I want to remark on Fritz's comment

"I would think that for any semiotic system involving discrete infinity, 
the existence of rules (schemas, constructions) would be the null 

I would agree with this argument (though I recognize that not all would) to 
the extent that it means that there must exist SOME rules.  But a variation 
of Martin's claim that lists are the null hypothesis is to take it as a 
claim not in general but in any particular instance.  For example, I have 
argued in a talk I have given in a number of places that in many if not 
most languages, there is no word class of prepositions or postpositions, 
but just a set of separate constructions that speakers store separately 
without generalizing across them.  In other words, in this particular 
instance, the null hypothesis is that there is just a list, and the burden 
of proof is on anyone who wishes to claim that speakers recognize them as a 
class.  But the general assumption of linguists of various stripes has 
usually been that they form a class, i.e. that there are rules across them, 
and the idea that there might be a need to justify that assumption rarely 

I was persuaded many years ago that lists are the null hypothesis by Bruce 
Derwing's 1973 book "Transformational grammar as a theory of language 
acquisition: A study in the empirical, conceptual and methodological 
foundations of contemporary linguistics".  The first half of the title is 
somewhat misleading.  I would say that the idea that lists are the null 
hypothesis was one of its main themes.

Matthew Dryer

--On Tuesday, June 10, 2008 12:24 PM -0700 Frederick J Newmeyer 
<fjn at> wrote:

> I would think that for any semiotic system involving discrete infinity,
> the existence of rules (schemas, constructions) would be the null
> hypothesis.
> I don't pretend to have read all of the literature on formulaic language.
> But my impression is that those who put such language on centre stage (1)
> focus almost exclusively on language production and all but ignore
> comprehension and (2) show no interest at all in language users' ability
> to make judgments of well-formedness of sentences that they have never
> heard. It seems self-evident to me that once comprehension and judgment
> data are brought into the picture, the need for rules (schemas,
> constructions) becomes indispensable.
> Let me stress that I am NOT offering an argument for 'innateness' here. I
> am not even offering an argument for generative grammar, as opposed to,
> say, cognitive grammar or construction grammar. Just an argument for
> rules (schemas, constructions).
> --fritz
> Frederick J. Newmeyer
> Professor Emeritus, University of Washington
> Adjunct Professor, University of British Columbia and Simon Fraser
> University [for my postal address, please contact me by e-mail]
> On Tue, 10 Jun 2008, Martin Haspelmath wrote:
>> It seems to me that Fritz Newmeyer's appeal to the
>> Rule-List Fallacy in the context of the argument
>> about formulaic language overlooks a crucial
>> asymmetry between rules and lists:
>> While lists are a necessary component of all
>> semiotic systems, rules are not. All languages must
>> at least have lists of morphemes, and then in
>> addition they may have rules. But the burden of
>> proof is on those who want to claim that they have
>> rules (or schemas, or constructions). In general,
>> the evidence for rules has been considered
>> overwhelming (in all languages), so almost everyone
>> accepts them.
>> Now I think Fritz's argument doesn't go through: If
>> one could show that it is in fact possible to
>> explain speakers' behaviour by claiming that their
>> knowledge of language consists of a simple list of
>> morphemes (or formulas), then this would indeed be
>> a powerful argument against the existence of rules.
>> In other words, the null hypothesis should be that
>> languages have no rules, and if not enough evidence
>> can be found to reject this hypothesis, we should
>> assume that they don't.
>> Notice that this doesn't work the other way round:
>> The null hypothesis cannot be that languages have
>> no lists, but only rules -- languages must have
>> lists. So if one discovers rules, this does not
>> mean that the same phenomena are not also stored as
>> lists. The Rule-List Fallacy is unidirectional.
>> But while I think that this particular argument is
>> invalid, Sandy Thompson and Paul Hopper will need
>> to do a lot more to convince linguists that no
>> rules (or schemas, or constructions) are needed to
>> explain speaker behaviour. Strictly speaking, they
>> are defending the null hypothesis, but in actual
>> practice, almost all linguists (regardless of their
>> ideological preferences) find that they need rules
>> for their work.
>> Martin Haspelmath
>> Frederick J Newmeyer wrote:
>>> Let me start by calling attention to what Ron
>>> Langacker has called the 'Rule-List Fallacy'.
>>> Ron noted, completely correctly in my opinion,
>>> that it was a fallacy to assume that lists have
>>> to be be excised from the grammar of a language
>>> if rules that subsume them can be established.
>>> The converse of this fallacy is equally
>>> fallacious: that rules have to be be excised
>>> from the grammar of a language if lists can be
>>> established. Even if it were the case that a
>>> huge percentage of language users' output could
>>> be characterized by lists (formulas, fragments,
>>> etc.), that would not exclude their also have a
>>> grammar composed of rules (or their notional
>>> equivalents) that allow hearers to analyze
>>> unfamiliar collocations and assign to them
>>> structure and meaning.
>> --
>> Martin Haspelmath (haspelmath at
>> Max-Planck-Institut fuer evolutionaere
>> Anthropologie, Deutscher Platz 6	D-04103 Leipzig
>> Tel. (MPI) +49-341-3550 307, (priv.) +49-341-980
>> 1616
>> Glottopedia - the free encyclopedia of linguistics
>> (

More information about the Funknet mailing list