classifiers (and ligatures)

Alan R. King mccay at REDESTB.ES
Fri Nov 6 09:42:48 UTC 1998

Hideaki Sugai wrote:

>In Hokkien, there is a
>classifier which is homophonous to the possessive marker [...]
>But the same kind of  homophonous pair cannot be found in other
>dialects such as Mandarin and Cantonese.

Hang on a minute!  As far as I know that is true for Mandarin but I am not
so sure about Cantonese.

In Cantonese the possessive relation may be expressed by placing between
the possessor and the thing possessed either (a) the possessive marker GE,
or (b) a classifier.  Classifiers (the same ones) are also used to link
numerals, and other quantifiers and determiners, to a following noun.  GE
is also sometimes used to link attributive adjectives to a following noun.

The most common, hence unmarked, classifier is GO.  And guess what?  The
possessive (etc.) marker GE is etymologically related to (and perhaps a
weakened form of) the classifier GO!

However, I'm reluctant to believe that there is a direct connection between
numeral classification and the possessive relation.  Instead, I have an
alternative proposal.  It may be that the key notion in all the functions
of GE and/or GO listed above is embodied in the word "link".

We're looking at a basic syntactic (macro-)function here.  It just so
happens (as so often) that it is not a function that is prominent in the
better-known European languages, which I guess makes the concept, de facto,
"exotic" if you like; but I think it is a fairly widespread phenomenon,
once we start looking for it.  We may have to look hard because the written
grammars of different "exotic" languages tend to diverge wildly in their
terminology for such unfamiliar concepts, but the term *ligature* (or
*linker*) comes to mind.  Let me quote Trask's dictionary:

"ligature (also linker).
A morpheme which in certain languages, notably Austronesian languages, is
required to link certain specifiers or modifiers to a head noun within a
noun phrase." (p. 160)

While some Austronesian languages have provided the best-known or paradigm
instances of grammars in which the ligature concept is prominent, I believe
that phenomena comparable to or at least reminiscent of this recur
elsewhere, probably fairly frequently (this assertion does not rest on a
statistical count!).  Apart from paradigm cases like Tagalog, there are
arguably ligature morphemes in at least one Indo-European language
(Albanian), in Afro-Asiatic (Hausa), and in certainly Micronesian (a part
of Austronesian with some Asian [areal?] features, including fully
developed systems of numeral classifiers!).  And that's just for starters.

In contrast to more specific functions familiar to us all, such as
"genitive", which cover fairly coherent ranges of relationships, what seems
to me to characterize ligatures is their largely "formal" function: a
ligature merely signals that there is some "attributive" relation or other
between the two elements it links.  The actual relation in question could
be adjective-noun, determiner-noun, quantifier-noun, possessor-noun,
noun-noun, relative-clause-noun, etc. (the ordering is irrelevant to the
question!).  Typically, it seems, in languages where the ligature concept
is prominent, the same ligature morpheme (or set of ligature forms), or
else phonologically and/or etymologically very similar ligatures (or sets
of ligatures), occur in constructions denoting SEVERAL of these relation
types, not just in one (in which case we would simply have a boring old
genitive marker, relative marker or whatever).  I think it is the throwing
of several of these relations into the same bag, as far as marking is
concerned, so to speak (actually that's a Euro-centric way of viewing it,
of course), that is the hallmark of ligatures.  (Tangential thought:
perhaps the "construct" construction found in Semitic and elsewhere might
be seen as a variation on the ligature theme; there certainly seems to be
some possibility for overlap.)

In some of the "ligature languages" I know of, furthermore, the use of the
ligature (which seems usually to have to occur - now this is iconicism for
you - in between the two "linked" items) is even independent of the
ordering of the two items linked.  So with an adjective modifying a noun,
you could perhaps get the two possible orders:


something which struck me as pretty odd until I started trying to think in
terms of the ligature concept rather than Latin!

Now while in some languages the ligature may take the form of a single
invariable morpheme or a set of contextually alternating allomorphs, in
quite a few cases the ligature varies according to some sort of agreement
principle or other.  In languages with gender, for example, it may vary for
gender of one of the "ligata"; and so on.  I would also venture to say that
there is often SOME variation in the form or behaviour of the ligature
depending on the kind of relation involved (e.g. the
ligature-cum-relative-marker may be only partially similar to the
ligature-cum-adjective-marker or the ligature-cum-possessive-marker).

So now back to Chinese (and Asia).  If in a numeral-classifier language the
classifiers also have ligature-like functions, this would be fairly
comparable to a gender/number language in which there is a ligature-like
morpheme which varies for gender/number.  Right?

R.L. Trask, _A dictionary of grammatical terms in linguistics_, Routledge,

More information about the Lingtyp mailing list