More on parts of speech

[Masayoshi SHIBATANI]

Many thanks for responses and reactions to our earlier query concerning the grammatical correlates of different part of speech systems.

It looks like our question has aroused some confusion. To make things clear,
we repeat here our proposal in more detail.

First, there is the distinction NOUNY versus VERBY languages (Wetzer 1992,
Stassen 1997). This distinction is based on the strategy followed for
predicative encoding of adjectives. Languages where predicative adjectives
are encoded in the same way as predicative verbs, are said to be verby,
while languages were predicative adjectives are encoded in the same way as
nouns, are nouny. This classification has formed the subject of some
typological research. Locker (1951) has tried to relate the above
distinction to the person-number-gender parameter.  Stassen (1997) manages
succesfully to relate the same distinction to the 'tensedness parameter'.
All these approaches share the characteristic that they stress the
structural and behavioural similarity between adjectives and one of the
other two parts of speech (nouns or verbs).
   First of all, we are looking at the structure of the parts of speech. Word classes have to be defined for the language as a whole, we cannot set them up looking at one single construction. It is the structural behaviour of  lexical items in different functions (predication, reference, attribution) or even their morphological structure, that will
decide whether they are classified as a verb,  a noun or an adjective.
   Suppose we would look only at one single construction, let's say
intransitive predicates. Then we might come across languages which encode
intransitive predicates always with a verbal strategy (Stassen 1997 gives
Malagasy and Tagalog as examples of this case). Since the verbal strategy
dominates here, these languages will be classified as verby. But from the
point of view of the organisation of word classes, these data don't help us
any further: when there are no formal differences, it makes no sense to set
up a different word class. In languages like Malagasy and Tagalog, the
intransitive predicate does not give a clue about the word classes we have
to set up.

   Our proposal is fundamentally different from the verby-nouny distinction in  the way that we do NOT stress the similarity between verbs and adjectives or the similarity between nouns and adjectives. What we call NOUN-prominent languages, isolate nouns form other parts of speech. For example, Japanese treats nouns separately from verbs and adjectives in that nouns have no inherent predicational function while verbs and adjectives do. What we call VERB-prominent languages isolate verbs as a unique category, while nouns and adjectives can be grouped together, as in English.
   We are contemplating whether this distinction reflects the two functional components of a proposition, namely, referencing (identifying a referent) and predication (saying something about it). We believe that languages in general set up two parts of speech, nouns and verbs, precisely because the propositional act involves these two functional components.  We are hypothesizing that languages differ as to which of these two components of a propositional act is made prominent. Some languages make  the referential function central, and thereby set nouns apart from other parts of speech, while some others isolate verbs as a reflection of the weight they place on the predicational act.
   Thus, we expect noun-prominent languages to have a grammatical device, e.g. topic constructions/markers, that elaborates the referential function of nouns. One the other hand, agreement seems to be a typical manifestation of verb prominence. Verb-prominent languages tend to be more strict about the relationship between the verb and its argument, this relationship being overtly expressed by agreement, and the government relation is more local such that no double subject constructions and topic-comment constructions are allowed in this type of language.
   On first sight, this correlation seems to work for languages such as
Japanese and Ainu, Sino-Tibetan languages and Austro Asiatic languages
( nouns being set apart, topic marking, double subject construction etc) and
Indo-European languages ( verb-prominent, agreement languages).
   We are specifically looking for the cases that disconfirm or hopefully confirm the above types of correlation as well as those cases in which our notion of noun-prominence and verb-prominence does not obtain: cases where adjectives share properties with both verbs and nouns, and cases where adjectives share properties with neither verbs nor nouns also exist.
   Thank you for reading and your help.

Matt and Pieter