33 SUBORDINATION

A SUBORDINATE VP is one that has a special relationship with another VP, which we call here INDEPENDENT. The specific kinds of relationships—such as specification, “if-then”, and “cause-effect”—are discussed individually under specific markers. As the label would suggest, a sub­ordinate VP does not occur as an independent sentence;1 that is, it is not a basic VP.

In terms of the semantic relationship of subordinate and independent sentences to their context, Fijian presents a situation somewhat differ­ent from some other languages. Lyons (1968:307) noted that independent sentences “express simple statements of fact, unqualified with respect to the attitude of the speaker towards what he is saying.” However, in Fijian, basic VPs include those with such internal markers as mani, which do indicate the speaker’s attitude towards what he is saying. Therefore, our grammatical classification of Fijian sentences cannot follow these quidelines exactly, and we must rely more on the form of such VPs for the present classification.

Formally, subordinate verb phrases are marked by one of a set of markers: ni, me, kē, sē, and dē. These markers fuse with certain sub­jects, and these contractions will be discussed in the treatment of individual markers.

33.1 NI

There are two types of ni phrases. First, one type specifies the subject or object. The mood is indicative, and the attitude of the speaker is neutral.

Next, ni phrases indicate TIME or REASON. The relationship between the independent and subordinate phrases is thus ATTRIBUTIVE.

Note that in TABLE 33:2 the subordinate phrase precedes the inde­pendent one. With the attributive ni phrase, the order is optional. The order is not optional in the sentence in Table 33:1. Unlike specify­ing NPs, specifying ni phrases do not usually have the option of preceding 427the independent VP to satisfy discourse structure requirements.

We now discuss these two types in more detail.

33.1.1 Ni phrases that specify. A ni phrase is used to show that a VP is serving to specify an object or subject within the independent VP.

33.1.1.1 Object specification

There are semantic and grammatical limitations to this construction. First, since an action/state is naturally singular, inanimate, nonhuman, and non-first or non-second person, the object that represents it is con­fined to third person singular.

Second, the head of the independent verb phrase is usually a member of a class whose common semantic feature is perception (such as rāī ‘see’) or reporting (such as kāī ‘say’). In translation, they are such phrases as ‘I know that …’, ‘I see that …’, ‘I said that …’ The following verbs are a sample of those that operate similarly:

In general, whenever the semantics of a verb allows an action/state (rather than an entity) to be a goal, this action/state is represented in the sentence as a subordinate VP, and usually as a ni phrase, (A me phrase can also specify the object under these circumstances, but with an added subjunctive meaning. See 33.2.)

33.1.1.2 Subject specification. Verbs that allow the subject to be specified by a ni phrase are often verbs of judgment (dodonu ‘correct’), existence (sega ‘is not’) or potential (rawa ‘able’). For example:

In this sentence, the independent VP is e dodonu ‘it is right’. The ni phrase specifies the subject e.

33.1.1.2.1 Idiomatic use of ni phrases. In some common sentences, an analysis of the relationship between independent and subordinate VPs produces unexpected results. For example, although the following sets of sentences look structually similar, they are not:

If, in the first set, we try to find a grammatical function for the ni VP, we fail. For example, the independent VP for the first is

which—if it had meaning at all—would mean something like *‘I am none’. 429And what function does ni kilā have? It cannot specify the subject (for it is in third person), and there is no object to specify.

A similar situation holds for the second sentence. Here, the inde­pendent sentence āū rawa is grammatical, meaning ‘I am able’. But again, the ni phrase cannot specify the subject, and there is no object.

It appears, then, that the surface forms of these sentences are GRAMMATICAL IDIOMS, based on the following underlying sentences:

both of which are grammatical possibilities, but less frequently heard than the idiomatic forms.

An oddity of another sort is the common construction:

Here, the idiom seems unanalyzable, even in terms of an underlying sen­tence. Semantically, it is the first person who is the actor. What role does the ni phrase play then? In terms of meaning, it should function as a specified object. But bese has no formal signs of transitivity.

In treating these constructions as idioms, I suggest that they are handled as whole units that are not analyzable in terms of the function of their constituents.

33.1.1.3 Layering of ni phrases. Dependent on the meaning of the individual verbs, a series of ni phrases is possible, producing a hier­archical structure. The following are examples:

The hierarchical structure is as follows:

Another (hypothetical) example is:

33.1.2 Ni phrases that modify. In the second main type of ni phrase, the subordinate VP modifies the independent VP. For example:

with ni:430

An alternate meaning for this sentence is:

The alternate glosses for the examples above shows that the meaning of ni seems to be on a continuum from ‘when’ to ‘because’.

On the surface, the similarity in form of the two types suggests that there must be something about the semantics of the independent VP that signals which type the ni phrase will be. However, the second type of ni phrases differ from the first in that it can be fronted (with the appropriate intonational clues):

33.1.2.1 Ni phrases that indicate cause. When added to a VP with an S1 verb, such as

the nominalized ni phrase refers to the agent, which can be interpreted as reason, or consequence. For example:

Here, the ni phrase seems to be on the functional border between specify­ing the object and modifying the independent VP. That is (in terms of the meaning), the ni phrase can elaborate on the subject e, or serve in an adverbial function.

Even closer to a modifying function is a ni phrase that occurs after verbs of emotion, such as mārāū ‘happy’, which involve a human experi­encer. Things or conditions are not mārāū or rarawa ‘sad’, reki ‘joy­ful’, cudru ‘angry’, rere ‘afraid’, ninini ‘excited’; however, feeling beings are. When such statives are used in the main VP, the following ni phrase indicates the reason or agent of the state.

This sentence could be interpreted as ‘The reason I am glad is that he went’, or ‘I am glad as a consequence of the fact that he went’.

As a matter of fact, a variation on this sentence, with very little change in meaning, uses a causative construction: 431

Here, the main VP is e vaka-mārāū-taki āū; na no-na lako is the specified subject.

Other examples:

For these sentences, the ni phrase cannot be considered as the specified subject, since the subject is animate and can be specified only by an NP referring to an animate being:

These ni phrases seem different from other modifying ni phrases in that fronting the phrase would change the meaning somewhat. Taking the example just above, for instance:

33.1.2.2 The meaning of modifying ni phrases. When a subordinating ni phrase precedes the main phrase, it conveys the sense of both time and consequence.

In constructions like this, it is difficult to separate time from consequence, since the second action both follows the first chronologi­cally, and is—in a sense—the result of the first.

In all these sentences, the action/state indicated by the main VP could not have taken place if that in the ni phrase had not preceded it.

In the sentences in group 2, the order of the phrases may be re­versed2 with no change of meaning; there is no chance that the ni phrase might be thought to specify the object (if there is any). For instance, in the second example, there is no grammatical object, but kāū—as a two referent verb—already has a noun modifier—kākana. And even if the main VP were:

the nature of the verb itself precludes interpreting the ni phrase as a specified object. In other words, one might know that so-and-so hap­pened, or tell it, but one could not carry it.

In some instances, the ni phrase could be interpreted as both indi­cating time and consequence, and specifying the subject:

The different grammatical forms of the possible translations illus­trate two separate grammatical interpretations: for the former, the ni phrase represents time and consequence: the state of the weather both precedes and causes the condition indicated by the main verb phrase. For the latter, the ni phrase is interpreted as a nominalization that func­tions to specify the grammatical subject e. I prefer this interpretation, because otherwise there would be nothing in the sentence or the context to explain what the subject e refers to.433

33.1.3 Morphophonemic changes. There are certain changes connected with ni phrases. First, for all third person and first person inclusive subjects, the e is deleted. Thus:

And so on for the remainder of these sets.

Next, when ni is used with first person singular subject, the u allomorph is used rather than āū: ni-u, phonologically nīū.3 Table 33:3 (and subsequent tables for the other subordinate markers), shows the phonological results of different combinations of marker + subject. Por­tions bounded by periods indicate accent measures (see CH 41).

33.1.4 Ka ni. In the literary style of Fijian, the combination ka ni is sometimes used in the sense of ‘because’. An example is:

33.1.5 The intonation of ni phrases. Similar to NPs that specify the subject or the object, specifying ni phrases are rather tightly phonologically bound to the preceding VP. For example, shorter sentences with verbs like sega ‘no’, rawa ‘able’, bese ‘refuse’ in the main VP are usually included in one phonological phrase:

In contrast, modifying ni phrases are less tightly bound, and the individual phrase peaks are distinctly marked.

33.1.6 Structural differences between types of ni phrases. PG (2/79) noted a structural restraint that separates specifying and modi­fying ni phrases: the former can begin with a fronted NP, as in:

This sentence is a variant of:

However, one cannot front gone in a time, consequence phrase:

33.2 ME

Me marks a VP that serves in a SUBJUNCTIVE relationship to the inde­pendent VP. Thus, the VP it is used in represents a situation that does not yet exist—in relation to the independent VP.

Me phrases have a wider range of functions and meanings than those for the other markers. The following examples show me phrases function­ing to show purpose:

Note that in each of the last two examples above, the me phrase is an identifying VP—one that uses a noun.

Along with the sense of ‘serving as’ something, me is used in the sense of ‘changing into’ something:

Note here that it is an identifying VP used with a subordinate mar­ker. Another example from the same text:

Me is also used to propose or suggest, fitting with a notion put forth elsewhere that elocutionary acts in Fijian must be accompanied by the proper grammatical trappings. For example:

The most common use of me for this purpose is with imperatives. Imperatives can occur with both second and third person. For example:

The following is an example in third person:

In some uses, me seems indistinguishable from ni. Note the follow­ing:

The important difference between me and ni is the subjunctive meaning of me: something that is definitely contrary to fact. Note the following minimal pair:

Milner (1972:62) illustrated the difference between the two markers with the following minimal pair:

This particular contrast can be illustrated in English by the pair ‘can’ vs. ‘may’: E rawa ni means ‘can’; E rawa me means ‘may’.

From a list of verbs that take ni phrases, SN (2/79) suggested that the following would take me:

33.2.1. Me vakā. The marker me often occurs before the verb vakā ‘resemble it’. Here, the whole construction has the sense of ‘such as’, ‘like’, ‘for instance’:

33.2.2 Morphophonemic changes. Me is somewhat different from the other subordinate markers in its morphophonemic behavior with subjects. Note, in Table 33:5, the phonological result of combining me with second person subjects.

33.3 KĒ

A kē phrase represents a potential cause, and an independent phrase the effect:

Note, in each of the following examples, that the situation referred to by the kē VP is a conditioning factor for that referred to by the independent VP.

33.3.1 Relationships outside the sentence. The summary description of kē shows the potential cause-and-effect situations represented by verb phrases within the same sentence. However, the situations may also be represented, within a discourse, by different sentences:

In each of the pairs above, the kē phrase could be interpreted as one part of a truncated sentence. For example:

33.3.2 Kē marking both phrases. Grammarians have noted4 that both verb phrases can be marked with kē:

One way to look at this construction is to consider that both VPs represent situations that are hypothetical. In each of these examples, since the verb phrases are not distinguished, phrase order must identify the first phrase as potential cause, and the second as potential effect. This situation is often reflected by ‘if … then’ in the English trans­lation.

33.3.3 Kevakā. The use of kē with vakā might be considered an idio­matic alternate to kē alone. However, there are some structural differ­ences. Strictly speaking, it is a subordinate verb phrase in itself; the main verb is vakā ‘be like it’. The following verb phrase specifies the object, but it itself is not marked for subordination:

439Arms (1985:15) noted an important feature of ke-vakā: that it may introduce a clause beginning with me:

33.3.4 Morphophonemic changes. Table 33:7 shows the rules for kē plus subject:

33.4 SĒ

Phrases marked by sē often occur in pairs, indicating different or contradictory potential situations. If only one sē phrase occurs, the opposing situation is implicit. Grammatically, the phrases specify sub­jects or objects.

TABLE 33:8

Examples are:

33.4.1 Types of verbs in the independent phrase. it was noted earlier that sē phrases functioned as specified subjects or objects. We now give additional examples:

Note that the second phrase is truncated. The following underlying structure might be proposed:440

TABLE 33:9

Here, both these subordinate phrases serve as the specified objects of kilā ‘know it’.

Many of the examples collected are variants of this particular pat­tern, using as the independent phrase [ ] sega ni kilā. The contrast between two (or more) different possible situations need not be made explicit. In the following example, it is implicit.

Another common example of sē phrases used as specified objects involves imperatives with speech-act implications. Note the following:

Other possible verbs in this semantic category are ka-y-a ‘say it’, sau-m-a ‘answer it’, and taro-g-a ‘ask it’.

Sē marking phrases as specified subjects very often occurs in a sub­ordinate relationship to the following dominant phrase:

For example,

This construction is like the previous construction specifying an object in that it can state alternatives explicitly, or state only one, leaving the other(S) implicit.

33.4.2 Semantic relationship between sē and the independent VP. The frequency of occurrence of sē phrases with the examples above—X sega 441ni kilā ‘X doesn’t know it’ and e sega ni macala ‘it’s not clear’—under­scores an important semantic feature of sē: it refers to a situation that is, respectively, unknown or uncertain. This semantic property is high­lighted even further in the use of sē phrases with two different verbs in the dominant phrase, dēī-tak-a ‘confirm it’ and tuku-n-a ‘tell it’. Since their meanings are the antithesis of uncertainty, such verbs domi­nate sē phrases only when the former are set in the negative or the future. For example:

If a sē construction were not set in the negative or future, it would not make sense, for without those restrictions, the uncertainty would vanish. Under such circumstances, a ni phrase is used instead:

33.4.3. Question words in sē phrases. Consistent with the UNKNOWN or UNCERTAIN semantic characteristics of a sē phrase is its use with the interrogative roots cēī ‘who?’, naica ‘when?’, vēī ‘where?’, cava ‘what?’, and vica ‘how many’. Examples (supplied by BK) are:

33.4.4 Morphophonemic changes. The following rules hold for sē plus subjects (see Milner 1972:55):

33.5 DĒ

A Phrase market by dē represents the effect of a cause, represented A phrase marked by de represents the effect of a cause, represented by the dominant phrase. Such a construction differs from the one with kē in that the dē phrase reflects the speaker’s opinion that the effect would be undesirable.

44333.5.1 Adversative vs. subjunctive. Many sentences with dē reflect the attitude of the speaker that (a) one should avoid a particular situ­ation because the effect would be undesirable, or (b) steps have been or should be taken to insure that the undesirable effect does not come about.

The first type often involves the use of kua or kākua (prohibitive):

Examples of the second type:

In both these subtypes, the relationship between the dominant and the subordinate phrases is roughly—in traditional terms—ADVERBIAL. That is, the subordinate phrase serves as a special kind of ATTRIBUTE.

However, in the second principal type of dē construction, the subor­dinate clause specifies a subject or object. For example:

In the first of the two examples immediately above, there seems to be no afflictive meaning to the situation represented by the dē phrase. Instead, it seems to be merely subjunctive: the situation might occur.

444Arms (10/84) pointed out that dē can also be used to introduce principal clauses, with the meaning of ‘perhaps’.

33.5.2. Morphophonemic changes. The combination of dē plus sub­jects results in the forms shown in Table 33.12.