criticisms of grammaticalization

Tom Givon tgivon at
Sun Feb 26 04:48:46 UTC 2006

The *epiphenomenon* epithet is just that. All biologically-based, 
adaptively-evolved systems can thus be dismissed as 'epiphenomena', 
because they are the product of interaction between serendipity (random 
mutation) and teleology  (not ID,but the teleological behavior of 
organisms during selection). Does that mean that evolutionary biology 
has not come up with explanatory principles that account for why 
synchronic forms are the way they are? No serious biologist would agree 
with that. Does it mean that evolutionary biology has succeeded in 
explaining everytring? No serious biologist would be so dumb as to claim 
that. Science is gradual and cumulative (not revolutionary, as Kuhn 
would have us believe), you don't solve all problems with a neat new 
theory. Not even Einstein did.

Up to a point, linguistic *diachrony* is rather similar to 
bio-evolution. Both are ultimately adaptively-driven, thus (Labov & 
Newmeyer notwithstanding) functionally motivated. All you have to do is 
study the rich variational data a bit more carefully. Both begin with 
rathe local, low-level *variation* that eventually may engender rather 
global, often strartling consequences. Both begin with *functional 
extension* (thus early *functional ambiguity*) of an existing structure 
(or lexeme). Both are profoundly uni-directional. In both, the 
directionality (and its governing motivating principles) are *never 
100%*, but is nonetheless  quite robust. And as Ernst Mayr said, what 
distinguishes biology from physics/math/logic  is precisely 
that--less-than-100% generalization--but generalization nonetheless. And 
finally, in both the reason for less-than-100% generalization is the 
same: *Multi-variant environment*, *competing motivations*, and the 
availability of alternatives. For example, there are at least 7-8 major 
ways (plus lots of minor ones) for grammaticalizing the 'passive' 
function (agent suppression), or the REL-clause fiunction.  And  indeed 
quite often this multiplicity of courses ius found in a single language, 
so that multiple alternative solutions (constructions) *compete* for the 
same (or rather similar) function. Why eventually one alternative is 
chosen over the others to be statistically dominant depends on those 
multiple other factors.

 Here is another similarity with bio-evolution and diachrony--the very  
same initial popullation can re-fashion the same source-organ towards 
different target (think of the mammal forelimb; or the 
gframmaticalization of 'go', 'take', 'come' etc.). That is, 
*one-to-many*. Likewise, *many-to-one* is found is found in both 
grammaticalization (massively; that's the essence of 
grammatical-typological diversity) and in bio-evolution, although much 
less less commonly in the latter. (Think e.g. of the main metabolic 
pathways to energy production: anaerobic sulfur bacteria, oxygen-burning 
organisms; the latter plus photosynthesis; the way different organs may 
be recruited for doing respiration in different phila).

One of the main differencs between bio-evolution and diachrony has to do 
with  the *source of the serendipity* (randomness). We have no real 
equivalent in diachrony to random mutations, since morpho-syntax gets 
forever re-cycled, rather than genetically coded. But everyday 
communicative behavior of individiuals (as Joan Bybee says) in a way 
apes the randomness of DNA mutation, by producing--in the communal 
pool--multiple variants that then compete for selection. Conversely, the 
'*selection*' part of diachronic change is much more socially dependent 
than selection in biology; although in many social species there begins 
to be an element of *social transmission* of individual innovative 
behavior, which becomes part of the overall mechanism of selection.

Finally, if I were to hazard a guess, I'd say Newmeyer, Joseph and Janda 
have been fighting the same old rear-guard war agains viewing 
grammaticalization as a *natural phenomenon*, rather than a bizarre 
artifact ('epiphenomenon'). And of course, their work is part and parcel 
of what Chomsky has been trying to do over a lifetime; that is, viewing 
language as a unique phenomenon that is not subject to selective 
pressures (viz his recent, most intriguing, foray into evolution--of 

I think it behooves us all to take biology a bit more to heart.

Best,  TG


  Likewise, there are many targets that can be colonized by the same 
source (think of how 'go' can grammaticalize

>On Sat, 25 Feb 2006 hilpert at wrote:
>>1. Unidirectionality, if it exists, is an even greater problem for
>>functionalism than if it turns out to be false. Developments that span
>>centuries would have to be explained independently of speakers, who only
>>have access to three generations of other speakers. (I attribute this one to
>>Janda 2001.)
>So how is this different from the argument that says that the eye could
>not be the result of evolution through natural selection, since that
>would require the organism to teleologically look many generations
>down the road to see the culmination of the process?
>Scott DeLancey
>Department of Linguistics
>1290 University of Oregon
>Eugene, OR 97403-1290, USA
>delancey at

More information about the Funknet mailing list