parts of speech
Alan R. King
mccay at REDESTB.ES
Thu May 20 06:57:35 UTC 1999
The question asked by Matt Shibatani and his students would seem to require
a wide-ranging typological survey in order to obtain empirical answers.
Pending such, we can at least reflect on the meaning and implications of
Most languages are recognised as having at least nouns and verbs as
distinct categories. Prototypically, verbs are specialized in the function
of predication (i.e. that is their typical function), while the typical
function of nouns is referential, the characterization of arguments. In
both cases there are mechanisms for changing these functions: nouns can be
used as predicates, and verbs can be used attributively, as in relative
clauses. In English, the copula (which is a verb) is used for the former
operation, and a relative conjunction such as _that_ in the latter:
--- SYSTEM 'A' ---
2) is N
4) that V
I believe it was Jespersen who pointed something like this out. This kind
of organisation involves having parts of speech that are distinguished
precisely by these characteristic functions. Therefore in the case of
verbs the predication function needn't be explicitly indicated
syntactically because it is already lexically inherent, whereas the
predication of a noun is may need to be so marked (in English by the
copula); and vice-versa for verbs and the reference function. We could
also envisage a system in which lexemes do not have these inherent
functions and there is no such part of speech distinction; rather, the
difference between predication and reference functions would be marked
syntactically in all cases. The single "general" part of speech that
resulted, let us call it G, might then configure a single set of
functional-syntactic options, which we might portray in shorthand thus:
--- SYSTEM 'B' ---
1) is G
2) that G
I understand that Jürgen Broschardt proposes something like this for
Tongan, but I am unable to fullly judge the validity of his claim. In any
case, most languages seem to have been considered by most linguists to
conform to "System A" above or something approaching this.
The thing about adjectives is that they have two equally typical functions,
since they are usually fully "at home" in both predicative and attributive
uses. Therefore, in the abstract, it might seem sensible for adjectives to
display something like System B. However, what this assumption fails to
take into account is that the language probably already has a System A "up
and running", so to speak, for nouns and verbs. It is therefore probably
more economical for the whole system if adjectives are just assimilated to
one of the two main types. And that is where different languages make
their choice to assimilate adjectives to verbs as in Japanese or to verbs
as in Latin.
This argument can be extended to other parts of speech besides those
already mentioned. Quantifiers, for example, may also be predominantly
verb-like in some languages but nominal in others.
Now I'm not sure that the assimilation of in-between words like adjectives
to the category of verbs in a language makes the language "verb-dominant".
Somehow that term doesn't seem to me to capture what is essential about the
issue. The essential question seems to be how each language handles the
two main functions of predication and reference, and of course how
individual lexical items are "plugged into" the resulting syntactic framework.
Thus I would not be too surprised to find that there is some sort of
correlation between the way languages deal with adjectives, on the one
hand, and the way "copular" sentences are treated in the languages, on the
other, or also the way relativization is handled.
Another related question of typological interest is, in languages with
verb-like adjectives, just how completely verb-like they actually are (or
are not); I'm sure some universal tendencies can be found there. And
mutandis mutandi for noun-like adjectives in other languages.
Apologies for a very rushed reply.
More information about the Lingtyp