Dans un courrier dat=E9 du 15/09/99 16:39:52 , David a =E9crit :
> At 4:37 PM -0400 9/14/99, Boudewijn Rempt wrote:
> >(I've quite forgotten what exactly this thread was about,
> =20
> It started with the old chesnut of verb-only or noun-only languages. (No
> one mentioned Tom O'Breton's ALLNOUN, I don't think). The valid point was
> raised that any such language would still have to cover the same semantic
> ground if it was to be useful, and the claim was made that all-noun-ism o=
r
> all-verb-ism must be impossible.
> =20
i think everybody here thinks it is possible as long as you call a relation
between nouns or between an actor and a verb a semi-colon.
> The reply was that noun and verb are syntactic categories, and that a
> syntactic distinction can be eliminated without neceaarily imporverishing
> the semantics. I consider "adjectival words" to be a typical case, as
> langauges can have a separate syntactic category like many European
> languages, or express these functions by nominal _or_ verbal roots.
> =20
the fact is that even languages using nominal or verbal roots
to make adjectives are aware of the specificity of such state
verbs or attributive nouns.
japanese "verbal adjectives" or "nominal adjectives"
are well defined categories. indonesians readily know
when to use ber- ("have") or "me-" ("do") to make verbs.
but neither feel these deserve more specific a category than,
say "inalienable attributes" in IE langs.
do we european readily make a specific category for nouns like
"one's head", "one's cause", etc ? these are major syntactic devices
in many natlangs. and when delving deeper, one could also tell results from=20
instruments, natural features from acquired ones, whole from parts, unit fro=
m=20
collectivity, etc. (all that conlangers know well). doing so you slowly
drift from syntax to semantics.
> Thus the semantic function of expressing qualities of objects can be
> expressed by several syntactic functions in different languages.
> =20
and v.v. (ditto).
> We _need_ not draw a distinction between syntax and semantics, but in fac=
t,
> such a distinction has explanatory power in discussing and examining the
> similarities that we do in fact notice across languages. This need not be =
a
> "watertight" distinction like the one the early structuralists claimed to
> make (though they were far too sophisticated to let it get in the way of
> their actual work). Subsequent schools (including the generativists, to
> whom I am not very sympathetic) are also criticised on these same grounds
> by some.
> =20
> Without being dogmatic, it's productive to make a distinction between
> syntax and semantics, even though they are not independent (how could the=
y
> be? The syntax exists to express the semantics, which is the goal of the
> process.
> =20
goal. process. if only.
> For computer geeks, the distinction between lexical and grammatical
> analysis of programming languages is somewhat analogous. These are not
> always completely independent (esp. in the monstrous C++, but even in
> Java), and the distinction is theoretically unnecessary (since the set of
> legal programs is the same however you describe the grammar).=20
really ?=20
Yet, it's
> meaningful and useful to make this split.
> =20
> >but
> >tangentially to other posts: anyone who wants to neatly divide syntax
> >and semantics into two separate water-tight compartments might take a
> >look at Wierzbicka's _The Semantics of Grammar_. If there's a grammatica=
l
> >distinction in a language, it will in all probability exist to encode a
> >meaning, is her contention, and the different grammatical distinctions
> >are not divided equally in different languages. She's Polish/Australian,
> >by the way.)
> =20
> This point is a good one, but doesn't actually address the utility of a
> two-level analysis. Phenomena at the syntactic level are meaningful at th=
e
> semantic level. However some things, like syntactic agreement rules, or
> gender systems, have only loose, undependable reflexes in semantics, whil=
e
> they exhibit strong regularity at the syntactic level. Most of the detail=
s
> of agreement and government phenomena can be explained very well at the
> syntactic level (in terms of word classes, etc.), and not so well at the
> semantic level (in terms of semantic categories).
> =20
> -- David
mathias