[Lingtyp] Structural congruence as a dimension of language complexity/simplicity

[Matthew Dryer]

Peter,

The point of classifying the language as SVO is that it behaves like an 
SVO language as far as word order correlations are concerned. Not 
classifying it as SVO means that one would fail to explain the 
correlations. Hawkins’ theory predicts that such a language counts as 
SVO. The class of languages I treat as SVO is defined roughly as those 
languages where the statistically dominant order in usage is AVP. There 
is nothing that the grammars of this set of languages share: these 
languages resemble each other only at the level of usage, not at the 
level of grammar. Hawkins’ theory predicts that the set of languages 
that I classify as SVO should tend to have prepositions. His theory 
predicts that the set of languages that have prepositions need not have 
anything in common in their grammars, only at the level of usage.

Matthew


On 1/19/16 2:58 PM, Peter Arkadiev wrote:
> Then I can't help asking a very naive question, appearing as though I haven't read the relevant literature (I have): if, as Matthew says, "classifying a language as SVO makes no claim about the categories in the language, nor that these categories determine word order even if the language has such categories", what's the point of classifying the given language as SVO in the first place? If the categories of a particular language can be totally at variance with those notions which typologists employ for comparative purposes, then the fact that a given language happens to be classified as SVO appears to be completely arbitrary and non-informative. Even worse, given this stance regarding the correspondence between comparative concepts and language-particular categories, word order correlations just can't follow, let alone be explained. Correlations between, say, OV and NPost in a given language are and have to be stated in terms of the categories relevant for this language, aren't they? And if such language-particular correlations can be mapped on robustly observed cross-linguistic patterns subject to well-articulated processing explanations such as those advanced by Hawkins, then, by necessity, this mapping cannot be just arbitrary, and vice versa.
> Again, I admit that I don't understand something.
>
> Best,
>
> Peter

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20160120/76610d1d/attachment.htm>