grammaticalization and complexity

Peter Harder harder at
Wed Mar 16 17:24:37 UTC 2011

Dear Wolfgang -

Thank you taking up this discussion of principle (which David and I believe

is important!) Thank you also for your clarifying comments, which show that

the disagreement is different from what I thought. I think the issue comes out

best in the following comment

(PH1:) These affordances/constraints may be more or less

complex for the encoder to match - obvious examples

of extra complexity being elaborate agreement systems -

(WS comment:) Honestly said, I do not fully understand this point. Sure, a L2-learner will happily turn to say Haitian Creole after having struggled with Navajo, but why should this hold for L1-learners?

PH2: you take 'complex' to equal 'difficult' - and that was not the intended meaning. The point I wanted to make was that there was an inherent complexity in the language, e.g. in the agreement system  (which you implicitly agree with in your comment above) - which need not have anything to do with cognitive difficulty, your native Navajo being a plausible illustration of this. (The issue of complexity and difficulty is dealt with in Dahl 2004:39).

Based on the implicit agreement, it appears we should also be able to agree that you can measure complexity of language (as a phenomenon!) provided you take care not to draw simplistic conclusions about cognitive capacity or difficulty from such descriptions.


Professor, dr.phil.
telf. +45 35 32 86 09
Inst.f. Engelsk, Germansk og Romansk/Dept of English, Germanic and Romance Studies
University of Copenhagen
DK-2300 Njalsgade 130
Copenhagen S
Fra: Wolfgang Schulze [W.Schulze at]
Sendt: 16. marts 2011 17:49
Til: Peter Harder
Cc: Funknet
Emne: Re: [FUNKNET] grammaticalization and complexity

Dear Peter,
these are really helpful remarks. Let me nevertheless comment upon some of them (sorry to bother you all again with my horrible English!):

Your argument presupposes that linguistic complexity is

identical to cognitive complexity.

No, that's not what I have tried to say. Starting from your formulation this would mean that linguistic complexity reflects parallel complexity in cognition and vice versa. This is just what I wanted to question. First of all, it is rather obscure to my what should be meant by 'cognitive complexity'. The term would only make 'sense' if correlated with the assumption that where would be some kind of 'cognitive simplicity'. Such a categorization would perhaps make sense in case you compare say the cognition of humans with that of a jellyfish (if ever). But in our case, we deal with human cognition only. Thankfully, the assumption of human cognitions that would differ as for complexity has since long been abandoned. The only instance, where such a term will perhaps make sense seems to be developmental psychology. But even in this case, one might wonder what's the use of the term. In my eyes, the cognitive system of any animate being is (more or less) accommodated to its actual environment and shaped by assimilating crucial features of its actual environment. It hence is a an expression of adequateness. A linguistic 'system' reflects only parts of the cognitive 'world' of humans (much of language in a diachronically distorted manner, that is not in terms of immediate, synchronic causal/symbolic relations, but by reflection of older relations mediated over times and from generation to generation). As I have said before: I think that 'linguistic complexity' is just another heuristic tool to capture the degree of explicitness in language. There is no need to correlate this term to the extremely problematic term of 'cognitive complexity'.

 After fifty years of cognitive science, this is a

natural assumption to make, and this also makes it

is worth putting a question mark against it:


In addition to being the possession of an individual,

a language is also a set of social affordances for

and constraints on making yourself understood.

Sorry to say, but here I cannot follow you. First, language (according to my humble opinion) is nothing that can be possessed (that is an 'object'). Language is always and only given in the individual, not in terms of an 'object', but in terms of a network of symbolic routines, emergent processes and schematic cognitive events. Eben if we look at it 'from the outside' (that is as linguistics), language first of all is a phenomenon, not an object. Its characterization (and construction!) heavily depends from the viewpoint of the observer and his/her experiential horizon (often formulated in terms of 'theories'). The process of 'reification' enables us to take language as an 'object' and to make it describable, but we must not transfer this secondarily construed 'object' into the cognition of an individual. Second, I do not think that language is conditioned by "a set of social affordances for and constraints on making yourself understood". Many sociological models suggest that the feature of 'being understood' is not governed by language, but mainly by social norms and habitual attitudes/practices represented by a potential perceiver. According to my understanding of language, it is first of all conditioned and structured to express 'cognitive needs', regardless whether there is an audience immediately addressed or not (language is a system of 'cognitive cries', if you want). Only secondarily, the corresponding knowledge system becomes (by learning) socialized and integrated into the set of norms and behavioral 'rules' present in a given speech community. These norms etc. naturally have a strong impact on the use of a language by its speaker, but they do not figure as a primary part of its ontology.

 These affordances/constraints may be more or less

complex for the encoder to match - obvious examples

of extra complexity being elaborate agreement systems -

Honestly said, I do not fully understand this point. Sure, a L2-learner will happily turn to say Haitian Creole after having struggled with Navajo, but why should this hold for L1-learners? Navajo speakers did not by large abandon their language before the intrusion of English even though other 'less complex' languages had been available in their region. The present-day preference for English is not necessarily due to the fact that it is less complex that Navajo, but because it has social relevance and social 'marks' that are estimated more profitable for young Navajo speakers. But all this again depends from whether Navajo and English are both (!) learnt by the corresponding individual. A Navajo child will perhaps switch from L1-Navajo to L2-English because English is structurally more 'transparent' than Navajo, it would never do when not exposed to English at all (which sounds trivial). Disregarding the competing existence of English, L1-Navajo speakers are not faced with more problems of language learning and encoding than L1-English speakers - else Navajo would have - since long - been a 'dead language'.

 and that is a different question from the question of how complex the intended message is.

What do you mean by 'message'? As I understand this term, it is immediately related the expression of cognitive states, enriched by aspects of intentionality. I cannot fully see what the criteria of complexity would be here.

 In your terms, language is in itself  a form of 'explicitness', not just a cognitive structure.

Explicitness by itself means that certain ensembles of cognitive concepts, schemas etc. are symbolized in a more fine-grained way than others. But as I have said before: This does not mean that the 'fine-grained' properties of these concepts, schemas etc. are not present and active if not symbolized at all. Rather we have to deal with the typical relation between the micro and macro layers of a given concept. Sometimes, the symbolization of the macro layer suffices to match the conventions, norms and the collective's set of cognitive concepts etc., sometimes the symbolization of its micro layer may have become relevant, for which reason so ever.

Another way of saying this is that you presuppose that encoding comes for free and adds no extra complexity.

I guess you mean symbolization = encoding, right? Naturally, a symbolic system that constantly refers to the micro layer of a concept is more detailed than a symbolic system that constantly refers to the macro layer. But the second system does not 'lose' information. Personally, I have a pronounced preference for gestalt oriented psychology. And in this perspective, we may even claim that the first type of symbolic systems is even less 'complex' than the second one, because it is less informative with respect to the 'gestalt' of the concept.

 Even in a hypothetical case where we assume that an intended message is identically specified for three different potential languages, and the speaker knows all language equally well, the encoding tasks are not the same.

Sure, there is not doubt about this!

Best wishes,

Prof. Dr. Wolfgang Schulze
Institut für Allgemeine & Typologische Sprachwissenschaft
Dept. II / F 13
Ludwig-Maximilians-Universität München
Ludwigstraße 25
D-80539 München
Tel.: 0049-(0)89-2180-2486 (Secretary)
         0049-(0)89-2180-5343 (Office)
Fax:  0049-(0)89-2180-5345
Email: W.Schulze at<mailto:W.Schulze at> /// Wolfgang.Schulze at<mailto:Wolfgang.Schulze at>
Personal homepage:
Diese e-Mail kann vertrauliche und/oder rechtlich geschützte Informationen enthalten. Wenn Sie nicht der richtige Adressat sind bzw. diese e-Mail irrtümlich erhalten haben, informieren Sie bitte umgehend den Absender und vernichten Sie diese e-Mail. Das unerlaubte Kopieren sowie das unbefugte Verwenden und Weitergeben vertraulicher e-Mails oder etwaiger, mit solchen e-Mails verbundener Anhänge im Ganzen oder in Teilen ist nicht gestattet. Ferner wird die Haftung für jeglichen Verlust oder Schaden, insbesondere durch virenbefallene e-Mails ausgeschlossen.

More information about the Funknet mailing list