criticisms of grammaticalization

Wolfgang Schulze W.Schulze at
Sun Feb 26 08:30:04 UTC 2006

                                         Caminante, no hay camino / se 
hace camino al andar (...) Caminante, no hay camino / sino estelas en el 
mar  (Antonio Machada)

Dear Colleagues,
Maybe that Suzanne's quote, namley:

> Epiphenomenon:  A phenomenon that is a trivial and accidental 
> byproduct of truly significant processes. 

reflects the actual use of the term epiphenomenon in very many 
instances. However, I think, we can break down the term to a more 
specific notion (as I try to do in my Cognitive Typology /Radical 
Experientialism framework). Accordingly we should start from the term 
'phenomenon' itself. In many pholosophical traditions (from which the 
term is taken), a phenomenon  is opposed to something 'really being', 
compare the Old Greek dictum 'kai ontes kai phainomenoi' (those which 
are and which appear to be [in a free translation]). Hence an 'object' 
said to be a phenomenon does not have properties itself (reflected in 
the qualification of the phenomenon), but it acquaires its properties 
through perception (and cognitive construction). A phenomenon thus 
presupposes the relation between an 'object' and its perceiver. Now, it 
seems out of the question that linguistic structures (in the broadest 
sense) are phenomena, perceived differently in space and time etc. In 
fact, Linguistics turns out to be a phenomenology rather than a 
scientific technique to describe the ontological status of language 
itself, even though the ultimate goal of Linguistics should always aim 
at unveiling the ontology of language lying 'behind' the 
phenomenological layer. The epistemological dilemma of Linguistics (we 
use language to describe Language), however, renders it doubtful that we 
will ever reach this goal (even if we try to substitute (!) the 
descriptive layer by a formal apparatus taken from say Natural 
Sciences). Perhaps it is more important to analyze the way the 
(possible) 'object' of Language is perceived and construed in terms of 
phenomena by both speakers and scientists (then turning linguistic data 
into some kind of meta-phenomenology).

An EPIphenomenon naturally relates to the same processes of perception. 
If we again start from the Greek term (epi-phainomai, again the 
passive), we get the notion of 'showing up as' etc. The epi-segment 
suggests that this type of phenomenon presupposes the existence 
(construal) of another phenomenon, without which the epiphenomenon would 
not have come into existence (would not be constructed). Now, if 
Language is an epiphenomenon (sharing its basic features with 
phenomena), the phenomenological substrate should be given in just that 
entity that enables Language, that is Cognition. In other words: 
Cognition (as a phenomenon) supervenes Language (as an epiphenomenon). 
If we observe changes in Language, this should be related to changes in 
the cognitive (functional!) apparatus. Likewise, variations in the 
synchronic 'substance' of languages(s) illustrate nothing but variations 
in (habitualized) experiential strategies to construe 'objects' in terms 
of phenomena (in a communicative perspective). The main point, however, 
is that human beings (better: cognitions) always try to make sense of 
the cognitive processes they 'live by'. In this sense, (epi)phenomena 
are turned into 'real objects' and manipulated/interpreted  accordingly. 
As a matter of fact it is crucial in linguistic analysis to decide 
whether a given process is related to the experience of language as an 
'object' or to the underlying (epi)phenomenology of language.

In this sense, we should distinguish at least two types of 
grammaticalization: a) processes that originate in cognition-driven 
changes in the mode to communicative experience (on a phenomenological 
level) and b) processes "that occur as language is used", as Joan has 
put it. Admittedly, it is not always easy to clearly distinguish these 
two types, but that problem may be conditioned by our yet unsufficient 
tools to identify all the (epi)phenomenologcal layers of language.

Personally, I would not go so far to relate grammaticalization processes 
to evolution (be it in a metaphorical sense). We do not have any 
evidence that would reveal to us the 'dark age' of language, that is the 
gap between language evolution and the earliest reconstructable layers 
of language (in a conservative estimate, the 'dark age' covers at least 
50.000 years or so, more likely much more than 100.000 years). Whenever 
we reconstruct earlier layers of language, we get just what we have, but 
in another phenomenology (perhaps this is also true because we cannot 
reconstruct by comparison but a variant of what we start from in our 
comparison). In this sense, grammaticalization is nothing but a repeated 
shift in the relation of concept and (articulatory) symbolization. For 
instance, the emergence of a 'near future' or present inchoative through 
the grammaticalization of GO verbs does not necessarily mean that on the 
conceptual layer, the notion of near futureness hasn't prior been 
existent. The only question is, to which degree this concept had been 
symbolized before. What we must not do is to infer from the 
non-existence of a grammatical 'form' (symbol/sign) to the non-existence 
of the 'corresponding' phenomenon on the conceptual layer. So it may 
well be that grammaticalization is unidirectional on the 'linguistic' 
layer, but surely not on the conceptual layer.

Naturally, the question remains whether grammaticalization itself is a 
phenomenon shaped in cognition. Personally, I would claim that 
grammaticalization again is an epiphenomenon that takes shape in 
language. it is supervened by cognitive processes that by themselves 
have nothing to do with grammaticalization, but with more general 
procedures of varying patterns of communicative perception and 
experience (e.g. Di(h)airesis, Zipf, the so-called Perception 
Action/Information Cycle, memory routines, metaphorization/metonymy, 
blending etc.). In addition, communicative (pragmatic) routines stemming 
from the layer of language objectification' mentioned above supply these 

Best wishes,

Prof. Dr. Wolfgang Schulze
Institut für Allgemeine und Typologische Sprachwissenschaft  (IATS)
[General Linguistics and Language Typology]
Department für Kommunikation und Sprachen / F 13.14
Ludwig-Maximilians-Universität München
Geschwister-Scholl-Platz 1
D-80539 München
Tel.:     ++49-(0)89-2180 2486 (secretary)
             ++49-(0)89-2180 5343 (office)
Fax:     ++49-(0)89-2180 5345
E-mail: W.Schulze at

More information about the Funknet mailing list