grammaticalization and complexity

Tom Givon tgivon at
Wed Mar 16 23:34:59 UTC 2011

Dear Fritz & all,

I think Fritz has once again raised a beautiful question that, as 
before, can also be  a can of warms. I will try to say something about 
the narrower question he posed in his second message, but I think the 
way several people took the discussion may be interesting enough to 
maybe merit going first.

One way of looking at the topic is comparing the two extremes of 
pre-grammatical (pidgin) communication and grammaticalized language, 
then asking (as some of you have alluded): Is there a difference between 
"complexity of the linguistic signal system as analyzed by the 
linguists" and "complexity of the cognitive task for the speaker/hearer"?

Take grammatical morphology first: Pidgin communication with no 
def/indef articles for nouns & no TAM markers for verbs seems--to 
us--simpler. Fewer coded distinctions. But to the speaker/hearer the 
processing task is more difficult, requiring more scanning & analyzing 
of the context--lexical, propositional, discourse, social, etc. So, when 
a hearer needs to decide whether "horse" is a first introduction (indef, 
presentative) or accessible/predictable (def), s/he cannot rely on 
morphological clues & make automated decisions, but has to spend much 
more time & mental effort on contextual scanning & analysis. And 
likewise, is "run"--past, future, present, habitual etc.?

Same with syntactic construction: In Bambara &Old Hittite, REL clauses 
look like conjoined/chained clauses, only intonation might furnish some 
clues. So the decision how to interpret the information--as asserted or 
presupposed--require contextual analysis, rather than grammatical clues. 
Again, the system as it looks to the linguist is simpler, fewer coded 
clause-types. But the processing task--at least to the hearer--is more 
complex, in a sense that more sources of information need to b e considered.

Lastly, Fritz's narrowed-down question: There is a difference between 
early & late grammaticalization. The former is communicatively-driven & 
create relatively clean morphological systems. The latter is largely 
phonologically driven, with the advent of de-stressing & assimilation 
rules, and creates irregularities, morphophonemics &, eventually, total 
zeroing of grammatical morphology. This is the so-called "cycle". So if 
I can interpret Fritz's narrow question, the answer is that ALL 
grammaticalization in its later stages results in such "complexity".

Best,  TG


On 3/16/2011 11:24 AM, Frederick J Newmeyer wrote:
> Dear all,
> Thanks so much for your replies! As some of them have indicated, I 
> probably did not give the ideal example to illustrate what I am after. 
> One category splitting into two (the example I gave) increases 
> complexity in one way (a bigger inventory of categories results), but 
> perhaps not in other ways, particularly if the new category encodes a 
> coherent semantic class.
> Here's a better example of what I am looking for. A case where the 
> result of grammaticalization is more irregularity and idiosyncracy. As 
> a hypothetical example, say we have one or more verbs or nouns 
> grammaticalizing into prepositions (or whatever), where the resultant 
> prepositions (or whatever) are irregular in some way with respect to 
> other  pre-existing members of that class.
> --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, 15 Mar 2011, Frederick J Newmeyer wrote:
>> Funknetters,
>> I am looking for nice examples of where a grammaticalization-related 
>> change, however motivated it might be from the point of view of the 
>> language user, ends up increasing the overall complexity of the 
>> resultant grammatical system. One example that came to mind is the 
>> formation of the distinct grammatical category of Modal Auxilary in 
>> English out of a subclass of verbs. One might argue that English 
>> grammar is now more complex because there are two categories rather 
>> than one and each have very distinct properties. Can anybody think of 
>> other/better examples from other languages?
>> Thanks! I'll summarize if there is any interest.
>> --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]

More information about the Funknet mailing list