core is basically a verbal nucleus affixed with up to two 'core arguments'. 2 I am using 'marked' here and in the paragraphs that follow in the Praguean sense of the word. When used in this sense it will be italicised.
Unmarked
THEME PREDICATE
GIVEN/KNOWN NEW
THEME PREDICATE
NEW KNOWN
FIGURE 11: TWO LEVELS OF PREDICATIONAL STRUCTURE
Examples of these different possibilities are presented in Chapter 3 below .1 The following two sections look at unmarked structures only.
1.3.3.2.2 SEGMENTAL STRUCTURE OF NOMINAL PREDICATIONS As noted above, while nominal predications seem to fulfil a variety of
different 'predicative' functions,2 many of these functions can be conflated in a discourse based grammar where the very general structure is THEME + PREDICATE. However, nominal predications still fall into three distinct types according to the number and
referential status of the terms needed to account for them. One can identify a class of one- referent propositions with two terms, where a referent is assigned to some class (i.e. a definite nominal is classified as an indefinite member of a class, or is equated with a non substantive adjectival noun). I shall call these attributional predications. E.g.
TOPIC PREDICATE
(34) Ike auke. (NWMek)
this dog "This is a dog." TOPIC PREDICATE (35) la loßia-ga. (NWMek) s/he good "He is good."
There is another class of one-referent propositions with two terms, where a referent is identified with some other definite term and so assigned a unique description (i.e. one
1 The given precedes the new in English "unless with good reason - which means here, unless it is a response to a specific question, either asked or implied" (Halliday, 1973: 163).
2 These correspond in a general way to the set of functions labelled
'attributive' in Fawcett's revised version of Halliday's systemic network for the English relational clauses (see Fawcett, 1987: 6.11).
definite nominal is identified with another definite nominal). I shall call these identificational predications. E.g.
TOPIC PREDICATE
(36)
la
I op a.
(NWMek)s/he chief "He is the chief."
Finally, there is a class of propositions with three terms, two of which are referential, and where one referent is assigned a relationship to another (i.e. one definite nominal is related to another definite nominal by means of a relational term). These will be referred to as relational predications. E.g.
TOPIC PREDICATE
(37)
L o p a ,
iu
ama-u.
(NWMek)chief I father-1SG "The chief is my father."
TOPIC PREDICATE
(38)
Ida,
imi
inä.
(WMek)she child mother. 3SG "She is the child’s mother."
These last constructions are often referred to as examples of 'inalienable possession' in the literature of Oceanic linguistics. The assertion of 'inalienable possession' is here treated as the assertion of a determining relation. Otherwise put, an abstract relational class term is made determinate with regard to some already determinate noun or pronoun.
In relational predications the nominal representing the 'object' of the relation is always incorporated in the predicate, while the relational term carries the tonic accent and is thus the focus:
PREDICATE TOPIC I________ FOCUS
(39)
Isa,
imoi
inä.
(EMek)she child mother. 3SG "She is the child’s mother."
All of the above constructions most typically take third person pronominals or demonstratives as their topics, and these regularly undergo ellipsis. So one frequently has to reconstruct the underlying proposition - by adding a definite deictic or anaphoric nominal - in order to account for the person and number of the predicate marking suffixes. Problems of analysis often occur when one ellipsed construction is embedded
in another, and where two predicate marking suffixes are in competition for a single slo t.1
The suffixes that mark the predicate in a nominal predication (and reflect the person and number of some topic nominal, which may be in a subject or an object relation to the predicate) are not discarded after rankshift - after which the erstwhile predicate itself functions as a topic. The same suffixes that functioned as predicate markers then function as specificational topic markers.
As detailed below in Sections 1.4 & 1.4.1, all non-verbally functioning lexical items are treated here as nominal expressions, and so one can have any of the following as nominal predicates: proper names (individual nouns), class nouns, non-substantive (adjectival) nouns and non-finite 'verbs', as well as all kinds of pronominal expressions (a class that includes deictics, specific and non-specific, and quantifiers, definite and indefinite). The following classes of nominal predications have been identified:
a) Identification. This constitutes assignment to a one-member class: "He is the c h ie f . This category differs merely in terms of class size, i.e. in degree, from the next one.
b) Attribution. This constitutes assignment to a class of things that share a given property, i.e. to a 'basic level' class2 ("That is a dog"), or assignment of the property itself, here expressed as a nominal predicate ("The tea is hot", "She is kind").
c) Determination. This constitutes the determination of one thing in relation to another, both of which must be definite ("This is my hand", "She is your mother"). The function here called determination is expressed by means of relational predications.
d) Quantification. This amounts to counting and/or asserting the quantity of things: "The coconuts are few/many/five/etc", "It is much", "They are many". Nominal predicates frequently carry a determining suffix indicating the person/number of the subject-topic, though there are exceptions to be detailed below. These suffixes are listed exhaustively in Section 2.1.2.1.1. As I am here concerned only
1 See Petrie (1980: 51-52) who recognised the same problem for Roro. Starosta, Pawley & Reid (1981: 27-28) touch on this problem in relation to Philippines and Formosan languages.
2 The cognitively 'basic' level of classification has been found to correspond with the genus level of Linnaean classification; it is a 'middle' level ("in the middle of taxonomic hierarchies" acc. to Lakoff, 1987: 46). More recent work by Rosch and associates verifies these findings. Classes are culture-specific.
to describe the discourse function of predicates (as opposed to any grammatical and/or referential functions the suffixes may have in themselves) the suffixes can be regarded as predicate markers.
Predicate markers are obligatory when the topic has a first or second person referent.
(40)
Oi
apala-mu.
(EMek)you bad-2SG
"You are bad."
(41)
Isa
lau
ama -u.
(EMek)s/he
I
father-1SG "(He is) my father."Also for all person/number combinations in the subject-topic of a relational predication:
(42)
Eke, im-babie-ga
imä.
(NWMek)that child.female-3SG hand.3SG "That (is) the little girl's hand."
But for third person singular topics of attributional predications the predicate marker is optional, and its presence can be regarded as emphatic.
(43)
Tsi
mekia.
(EMek)tea sweet "The tea is sweet."
(44)
Tsi
mekia-i)a!
(EMek)tea sw eet-A SS.3SG 1 "The tea is sweet indeed!"
However, in the plural the appropriate predicate marker is obligatory:
(45)
Imi
abala-tsi!
(WMek)child bad-3PL "The children are bad!"
1 1 add ASS (for ASSERTION = predicate marker) because it is optional here. One thinks here of the POC nominalising suffix *-(a)l)a, though Ross (1988: 390) sees its use as a predicate marker in Bali as an innovation. However, the Mekeo predicate markers are in fact determining suffixes marked for person and number.
In this secondary, emphatic function the 3SG predicate marker is found even on some topics which do not seem to have undergone rankshift. In this generalised use it indexes these items as marked topics. The form
-ga
is for example attached to first and second person emphatic pronouns (e.g. EMeklau-ga
"as for me"), as well as to the third person singular and plural forms (EMekisa-ga
"as for them"; see Section2.2.1.2).
As noted above, there are two possibilities for mapping topics and predicates, which are structural functions, onto given/known and new (this was diagrammed in the last section).1 These can be illustrated for nominal predications as follows:
UNMARKED: GIVEN/KNOWN > NEW
(46)
Inaina
amuV-ga!
(EMek)this dog-ASS.3SG "This is a dog!"
(47) A u
a pa la!
(EMek)man bad
"The man is bad!"
(48)
Eke
kuma!
(NWMek)that pig "That is a pig!" MARKED: NEW > KNOWN
(49)
Amu?e,
inaina!
(EMek)dog this
"It's a dog, this is!"
(50)
Apala,
au-ga!
(EMek)bad one-3SG
"It's bad, this man/one is!"
(51)
Kuma,
eke-ga!
(NWMek)pig that
"It's a pig, is that!"
The predicate marking suffix on demonstrative pronouns is obligatory in EMek after rankshift, i.e. in nominal groups, but optional in the other dialects under the same conditions. It is optional in EMek before rankshift, adding an emphatic nuance to the predication (see (46) above).