Regarding Query: http://linguistlist.org/issues/17/17-2888.html 
 
A few weeks ago, one of the students in my Introduction to General 
Linguistics class raised the question of whether, on the basis of such 
doublets as the following: 
 
Zhe-ge ren mai-le yi-ben shu. 
DET-cl. `person' `buy'-asp. `one'-cl. `book' 
`This person bought a book.' 
 
Zhe-ge mai-shu-de ren 
DET-cl. `buy'-`book'-DE `person' 
`This book-buying person = This person who bought a book' 
 
Chinese could be considered a `polysynthetic' language'. Not being an 
expert in polysynthesis, i was unsure how to respond to this query, so i 
posted it to LINGUIST (17.2888). I am hereby posting a summary of the 
responses i received. 
 
(One correspondent raised the question of how this topic turned up in an 
introductory linguistics course in the first place. To which my answer 
would be partly that i am, among other things, a typologist and i love to 
talk about it, and partly that i have found that, at least to a certain 
level, typology is a concept that students find relatively easy to latch 
onto, and it helps in organizing the vast amount of data i'm necessarily 
throwing at them. Which, of course, is an important part of the function 
of typology.) 
 
First of all, i would like to thank the following correspondents: 
 
Peter T. Daniels 
Bruce Despain 
Christian DiCanio 
Chao Li 
Bingfu Lu 
Ed McDonald 
Vivan Ngai 
Nick Reid 
Elanna Tseng 
Ralf Vollmann 
Adam Werle 
 
Several correspondents spoke of the importance of viewing such categories 
as `isolating', `incorporating', `polysynthetic', etc. as little more than 
arbitrary/conveniently-labelled points on what is really a typological 
continuum. (In my own classes, i try to emphasize that there is probably 
no such thing as a `perfectly isolating' or a `perfectly synthetic' 
language, rather that certain languages are more isolating, or more 
synthetic, or whatever, than others.) And, to some extent, the question of 
whether a (supposedly?) isolating language like Chinese might actually 
qualify as polysynthetic arises from a tendency to reify these labels, and 
thus distort the facts. 
 
A few correspondents noted further that there is as yet no 
generally-agreed-upon definition of `polysynthesis'. Vivian Ngai objected 
to the common practice of treating `noun incorporation' and `polysynthesis' 
as synonymous terms, when according to a definition given by Comrie, 
`polysynthesis' is merely the most extreme form of agglutination, the 
ability `to combine a large number of morphemes, be they lexical or 
grammatical, into a single word'. Elanna Tseng pointed out that 
noun-incorporation is a feature in a wide variety of languages, including 
English (e.g, `babysit'), but English `is certainly not considered a 
polysynthetic language'. Evans & Sasse 2002 have offered a definition of 
`polysynthesis' as involving the possibility of *all* arguments being 
absorbed into a single, one-word predicate which is able to function `as a 
freestanding utterance without reliance on context', which is certainly a 
long way from anything Chinese, pro-drop or no pro-drop, can do. 
 
Skalicka 1979 argues that, rather than being a `true' isolating language, 
Modern Chinese is heavily into compounding. This is certainly true, at 
least judging from my own experience, in spite of the impression created by 
the traditional culture and the writing system that reflects it. And 
compounding can be regarded as a `simple type of synthesis', in DiCanio's 
words. 
 
However, true polysynthesis is supposed to involve the loss of 
independent-word status on the part of the incorporated elements. Bauer's 
proposed definition suggests that, with regard especially to their semantic 
content, the incorporated morphemes exist in a sort of `no-man's land' 
somewhere between the status of derivational morphemes and full words, 
having greater semantic content than the former but less than the latter. 
Furthermore, in a textbook-example polysynthetic language like Sora, 
Greenlandic, or Onondaga the incorporated element is phonologically quite 
distinct from the semantically-equivalent free form. In Chinese, on the 
other hand, `shu' (= `book') is the same whether it is functioning as the 
object of a simple, monoclausal sentence or as an object within a 
noun-modifying relative construction. (I cannot even say with any 
confidence that, in the latter case, it suffers any loss of distinctive tone.) 
 
It is doubtful that the noun-modifying relative construction in Chinese 
even qualifies as a compound, since it allows the verbal element to be 
modified by an aspect marker (such as the perfective marker `-le') 
*intervening* between the verb-stem and its object -- just as would be the 
case in a full sentence -- and the nominal element to take adjectival 
modifiers of its own. Thus, the relative construction `mai-shu-de' in my 
example can be expanded to e.g. `mai-le hen-duo-youqu-de shu-de' (`bought 
many interesting books'). Furthermore, it is *not* possible to put the 
perfective marker after the verb-object string within the relative 
construction (`*zhe-ge mai-shu-le-de ren'), which clearly indicates that 
that string is not recognized by the grammar as a word-level entity, as it 
should be if it were a true compound. 
 
Literature: 
Bauer, Laurie. 2002. Introducing Linguistic Morphology. Edinburgh 
University Press. 
Chi, Hong. 1992. Noun Incorporation: A Chinese Case? Word vol. 43, no. 2 
Evans, Nicholas, & Hans-Jurgen Sasse, eds. 2002. Problems in Polysynthesis 
(Studia Typologica 4). Akademie Verlag. 
Skalicka, Vladimir. 1979. Typologische Studien (Schriften zur Linguistik 
11) Hrsg. v. Peter Hartmann. 
Vollmann, Ralf. MS. Der Wortbegriff im Tibetischen. MS available from 
the author, uni-graz.at>