Tải bản đầy đủ - 0trang
Chapter Six. The Verb Phrase and Related Issues
1 6 0 6. The Verb Phrase and Related Issues
the coreferential clitics in itself does not guarantee that the predications
constitute a single clause, however. Also recall that the coreferential
clitics are never used for first and second person singular referents,
regardless of the degree of linkage between predications.)
6.1.1. Same-subject infinitival complements
As discussed in Section 3.11, the clitic jfy- (COR) can be used in certain
constructions if subjects of successive verbs or predications are coreferential. Alternatively, one of the verbs might have no Set I clitic. Use of other
Set I clitics on both verbs in such constructions would be interpreted as
indicating non-coreferential subjects (except for first or second person
singular referents). One such construction involves same-subject (SS)
infinitival complements. These complements are marked with the
infinitival/participial nominalizer -jada or -janu (depending on dialect).
The complement can precede or follow the main verb whether or not jfyis used:
'She/he is now beginning to walk.'
'To sing she/he wants.'
'She/he wants to eat the fish.'
SS infinitival complements, as in (1-3), contrast with sequences of SS
predications, as in (4). In (4b, c), the non-coreferential Set I clitic sa- is
used rather than zero or jfy- to refer to the same participant referred to in
(4a). Out of context, use of sa- on the three verbs in (4) could be ambiguous: it could refer to one, two, or three different participants. This correlates with the fact that the predications are all finite in form and they do
not evidence the same degree of conceptual unity as do main verb and
complement in (1-3) above. (4a-c) will be interpreted as encoding three
different actions, whereas (1) will be interpreted as encoding two different
6.1. Verbal nexus 161
facets of a single action or state of affairs. In (2) and (3) the infinitive
expresses the goal of wanting.
3SG-terminate-EMPH-3SG savage there
(a) 'HeĂ threw a spear, (b) He¡ threw a spear,
(c) He/terminated the enemy there.' (TW032-034)
If tense is marked in constructions like (1-3), it can only occur on the
finite verb and may have scope over both predications. Tense interpretation in (2) and (3) need not be the same between main and complement
predications (the singing could be future to the wanting), but tense could
not be marked on the complement.1
There is no one well-defined set of aspectual morphology in Yagua.
Some second position clitics, and verbal locational, iterativity, movement,
completive, and imperfectivity suffixes all have aspectual meanings (cf.
Section 6.5). Certain iterativity or distributive formatives, at least, may
occur on infinitival complements. However, in all such cases that I know
of, the iterativity or distributive suffix forms a well-lexicalized stem with
the verb root and does not have scope over the finite verb, as in (5-6):2
'He wants to study (a problem).'
Compare: jaachipớớtya 'remember'
'He likes to dance.'
Compare: rỗtỗy 'jump'
Aspectual formatives may occur in the finite predicate and do have
scope over the infinitival predicate. In (7), for example, the jumping
would most likely have to be taken as iterative or as a customary habitual
162 6. The Verb Phrase and Related Issues
action in the past, given occurrence of -núúy 'imperfective' in the finite
verb. Iterative and habitual are types of imperfective aspect.
'He used to like to jump.'
SS infinitival complements are different from other infinitival clauses
and other complement clauses (Section 3.11) in that SS infinitives may
intervene between the finite verb and its subject (though the infinitive
need not occur contiguous to the finite verb as illustrated in (22) below).
This is the only construction in which a non-adverbial, non-clitic element
may occur between the verb and its post-verbal subject:3
'The bird(s) now began to get agitated in front of
Example (9) shows that second position clitics may intervene between
the main and infinitival complement predications:
3PL-begin JIITA march-INF
'They all began to march.' (CLS063)
To summarize, SS infinitival complements are different from independent clauses which have coreferential subjects as in (4) above, in that the
former cannot take independent tense and aspect. SS infinitival complements also require either coreferential clitics, or no Set I clitics. (If the
subjects are first or second person singular, coreferential clitics cannot be
used. But no Set I clitic need occur.) SS infinitival complements differ
from indirect quote complements in that the latter can have independent
tense and aspect even though the coreferential clitics are employed
(Section 3.11.6). SS infinitives are different from infinitival adverbials
(Section 3.11.7) in that the former can intervene between the finite verb
and its subject. This last fact also distinguishes SS infinitival complements
6.1. Verbal nexus 163
from nominal object arguments of finite verbs. The latter cannot intervene
between the verb and its subject. Thus, there is evidence that SS infinitival
complements form a more tightly knit unit with the main verb than do
other types of nominal and verbal complements. However, placement of
second position clitics as in (9) recognizes that they are still separate
constituents from the main verb.
The facts about SS infinitival complements accord well with the notion
of "core juncture" discussed by Foley and Olson (1985). Foley and Olson
distinguish three levels in the clause. In simplex clauses the "nucleus" is
essentially the verb plus its aspectual operators. The "core" is the
nucleus plus those arguments which are subcategorized or selectionally
restricted by the verb (more or less equivalent to what I term the nuclear
predication in Chapters 3 and 7). The "periphery" is the core plus noncore arguments such as locatives and other oblique noun phrases. Operators at the peripheral level include epistemic modals and evidentials. At
any given level tokens of the same type may join together to form complex
constructions, potentially resulting in nuclear junctures, core junctures,
and peripheral junctures. In a nuclear juncture, verbs or verb roots are
joined together (not necessarily phonologically), and share all arguments
equally. In a core juncture, however, the core arguments of each nucleus
(verb) are still selected independently, though certain serial core junctures
require the actors of the two nuclei to be coreferential. The two cores
share a common set of locational and time arguments, as well as tense
and mood specification (though not necessarily aspect). Peripheral
junctures result in conjoined clauses.
Foley and Olson are primarily concerned with types of verb serialization when they propose this schema, and serial constructions may have
either nuclear or core junctures. But they clearly intend that the general
framework should extend to languages which do not have canonical serial
constructions. In Yagua, SS infinitival complements could be said to form
core junctures with their main verbs. One of the arguments is, by definition, coreferential between the two verbs. But other arguments are
selected independently. In (3) above, for example, quiivỗ 'fish' is not an
argument of vỗỗta "want' but only of jimyiy 'eat'. Nevertheless, the infinitival complement cannot have independent tense and person/number specification. Its immediately post-verbal placement also indicates a special
type of juncture with the main verb. This is represented in (10) (adapted
from Foley and Olson).
We might hypothesize that whenever participants within a single C 1 or
C clause are coreferential with one another, the coreferential clitics jtyand -yii will express all but the linearly first mention of the participant.
This would cover the case of infinitival adverbials discussed in Section
164 tí. The Verb Phrase and Related Issues
3.11.7 which cannot have independent tense specification from the finite
verb, and which could be paradigmatically substituted for nominal objects
of postpositions. As with SS infinitival complements, infinitival adverbial
phrases have lost their clausal status and are nothing more than parts of a
simplex C 1 or C clause.
[AUX Vj = V-INF
Where " = " indicates core juncture within the scope
of a single clause. The subject argument S is shared
between V j and V-INF. (The relative order between
O j and C>2, if they occur, and their verbs varies from
instance to instance.)
This hypothesis also accounts for why the coreferential clitics are not
used within relative clauses even though relative clauses share an argument with their main clause (Section 3.11.4). In this case not everything
within the syntactic scope of the higher c V c clause forms a single clause.
Rather, there is embedding of a relative C 1 clause within the higher c V c
clause. The relative clause retains its status as a clause. A single tense or
aspect need not govern both clauses. In order to interpret reference inside
the relative clause, a C 1 boundary must be crossed, but this is not true
with SS infinitival complements and infinitival adverbials.
The notion of core juncture and occurrence within a single C 1 clause
does not account so nicely for use of coreferential clitics in indirect quote
complements. As illustrated in Section 3.11.6, except for use of the coreferential clitics, indirect quote complements are fully independent clauses.
They may have independent tense and aspect and there is no overt
complementizer. The time phrase in example (226) also illustrates that
oblique ("peripheral") elements need not have scope over both verbs. In
sum, occurrence of coreferential participants within a single c V c clause
is a sufficient, but not a necessary nor the only, condition for use of the
6.1.2. Verb serialization
A limited amount of verb serialization in the sense of Foley and Olson
occurs. Only movement verbs may occur as the second member of a serial
complex. These form one phonological word with the main verb as shown
6.1. Verbal nexüs 165
by palatalization and metathesis processes. The movement root immediately follows the other verb root or lexicalized stem.
'He ran out long ago.'
Ί stood and got up earlier today.'
'It sprouted up.'
Unlike movement suffixes (cf. Section 6.5), movement roots can occur
as main verbs. (A classifier serves as a nominalizer on the verb maay
'sleep' in .)
'He goes up by the ladder (e.g. into the house).'
'He got up out of his sleeping mat/cloth.'
Foley and Olson argue that cross-linguistically, the most likely verbs to
occur in serializing constructions are intransitive verbs of motion, location,
or position (41):
Intransitive verbs, particularly active intransitive verbs of motion,
location, or posture, are favored in a restricted slot to form nuclear
junctures with another verb in an open slot. These are favored because
as active intransitive verbs they introduce no new arguments in the core,
all core arguments being a function of the lexical entry of the verb in the
open slot in the juncture.
166 6. The Verb Phrase and Related Issues
Following Foley and Olson, I hypothesize that the difference between
finite verb plus SS infinitival complements and serial verb complexes in
Yagua is one of "core" versus "nuclear" juncture. In a nuclear juncture,
all arguments of the two verbs must be the same.4 This is represented in
[V l : V 2 ] S O j
Where " : " indicates nuclear juncture and V 2 is a movement verb
Within the verb phrase the majority of adverbs most neutrally follow the
'We're going to eat first.'
When an adverb precedes the verb, it conveys extra pragmatic force or
degree of the quality expressed by the adverb. Compare (18) and (19):
'She is walking fast.'
'She is walking very fast.'
The heightened degree of the quality expressed or the pragmatic force
communicated by preverbal position suggests that the preverbal adverb is
actually occurring in the pragmatically marked PM position (cf. Chapters
3 and 7). However, some adverbs always precede the verb. Mitya 'just' is
one such case. Mitya can be used postverbally, but only with appropriate
pauses as indicated by the commas in (20a). Mitya also has the idea of
'nothing' and that is the sense conveyed in the following case.
6.2. Adverbs 167
NEG someone IRR be.afraid-TRNS-HABIT-lPL
Η . » Λ»
ti , muya,
There wouldn't be anyone who would frighten us, nothing,
3SG-NIY-TCY bother-INF-INAN forest-CU-in
of those who are bothering ones in the forest.' (LX045)
Jááryiy 'really5 and adverbial phrases with jááryiy most commonly
precede the verb. By its very meaning, jááryiy heightens the degree of
whatever quality or action is being expressed; it thus expresses a type of
semantic markedness (Chapter 7).
Jááryiy váneera sa-rupífy.
'She is walking really fast.'
really 3SG-like-INAN hammock-inside sit-INF
'He really likes to sit in the hammock.'
6.3. Subject - object asymmetries: Evidence for a verb phrase
containing the object?
Positing a structural verb phrase constituent containing the verb and
object is one possible way to account for subject - object asymmetries.
This would be particularly motivated if the subject - object asymmetries in
question could be argued to stem directly from a structural difference
where the subject is immediately dominated by the sentence (or clause),
while the object is immediately dominated by the verb phrase.
One subject - object asymmetry in Yagua concerns what can determine
the index of the coreferential clitics jfy- and -yù. As discussed in Chapters
3 and 4, jfy- is part of the Set I clitic paradigm and can refer to a subject,
168 6. The Verb Phrase and Related Issues
a genitive, or an object of a postposition. The clitic -yù is part of the Set
II clitic paradigm and can only refer to objects.
The clitics jfy- and -yù do not have an inherent person/number index,
but take their index from something else within the C1 clause. This index
can be controlled by a linearly preceding subject as in (23), but not by a
linearly preceding object as in the ungrammatical reading for (24).5
Linearity alone does not account for the asymmetry, as both subject and
object phrases can precede jfy- and -yù clitics. Here I illustrate just for jfy-:
Preceding subject (in bold):
3SG-wash Celina COR-house-inside-3SG
'Celinaj washes him/her inside herj house.'
Preceding object (in bold):
3SG-wash-3SG Anita COR-house-inside
'She/hej washes Anita inside his/her^ house.'
*'She/he washes Anitaj inside herj house.'
Within certain frameworks, an explanation for this asymmetry might be
(partially) sought in positing a constituent-command relation between a
preceding subject and the jfy-1-yù clitic, a relation which does not obtain
between a preceding object and the jfy-1-yù clitic.6 This relation does not
hold between the object and the clitic because the object is "lower" in
the structure, occurring inside the verb phrase constituent. Positing SVO
as the underlying basic order would facilitate such an analysis in that the
verb and object are then contiguous, and a verb-plus-object constituent
may be more easily argued for.
Even if one were to posit SVO as basic in some underlying sense in
order to facilitate this analysis, existence of a verb-plus-object constituent
would not in itself provide a unified account for the asymmetry in what
can control the index of jfy- and -yù. Not only can the index be controlled
by a preceding subject, but also by a preceding genitive noun as in (25)
and (26), or by the object of a postposition as in (27). That is, the index
can be controlled by any Set I argument.
6.3. Subject - object asymmetries 169
Control by Genitive (bold; genitive NP is bracketed):
[Tomása roorijvyiimu jíchuutaníí.
'In Tom'Sj house hej washed him/herj.'
'She¡ washes her· inside Anita's- house.'
OR: 'She¡ washes herself inside Anita'Sj house.'
Control by Object of Postposition (bold):
lSG-see 3SG-DAT COR-house-inside
Ί saw him/herj inside his/herj house.'
What we need to account for is not the subject - object asymmetry, but
the Set I vs. object asymmetry. Most likely a mixture of pragmatic and
syntactic factors must be acknowledged in order to completely account for
what can control the index of jfy- and -yù? My major point here is to
show that positing SVO as the basic constituent order, such that one can
more comfortably say the object is part of the verb phrase, does not in
itself provide a unified explanation for what can control the indices of
In Section 3.8 I noted another possible asymmetry between subject and
object relative to question formation on arguments of embedded clauses.
The available data are not conclusive as to whether such an asymmetry
exists. But even if it should, there are other possible solutions besides
positing a structural VP containing verb and object.
First of all, subject - object (and even Set I argument - object) asymmetries are not the only ones that need to be accounted for in language.
Second, there is no a priori reason that such asymmetries have to be
accounted for in terms of structural or configurational relations. In
particular, Keenan (1984) discusses the closer semantic ties which hold
between verbs and their objects, as opposed to verbs and their transitive
subjects (A). By itself this semantic tie might predict the existence of
languages where asymmetries also exist between A and S, particularly as
such semantic relations might have historically resulted in differential
grammaticization of A versus S arguments, or differential syntactic
1 7 0 6. The Verb Phrase and Related Issues
behavior of A versus S arguments. Exactly such syntactic asymmetries are,
in fact, claimed to exist in some ergative-absolutive languages (Dixon
1979, T. Payne 1982). However, Keenan's observation in itself does not
account for languages such as Yagua where there may be asymmetries
between objects and intransitive subjects as well. In languages such as
this, why should transitive and intransitive subjects (A and S) be grouped
together as opposed to objects? This is due to the fact that A and S share
certain other properties which motivate grammaticization of a "subject"
category comprised of both S and A. For example, Du Bois (1985)
discusses the functional role which both S and A share in encoding
given/highly topical information in discourse. Such grammaticization may
have consequences in terms of subcategorization of verbs for their objects,
as opposed to their subjects (both S and A). 8 S. Anderson (1984) argues
that in Kwakwala, a verb-subject-object Wakashan language, subjectobject asymmetries can be accounted for by subcategorization relations,
rather than a configurational relation which presupposes a structural VP
consisting of verb and object. A similar account could be argued for in
Yagua (if not in most languages).
Some further suggestive evidence that verb and object do not form a
constituent comes from placement of the second position clitic jffta and
placement of certain adverbs.
In general, second position Yagua clitics follow the first constituent of
the sentence and indicate a variety of aspectual/ modal and discourse
functions. When there is no preverbal element, jffta and all other second
position clitics, such as numaa 'now', directly follow the main verb within
a simple verb phrase. This has been amply illustrated in Chapter 3 where
we discussed second position clitics in C and C 2 . However, if some
constituent other than the verb occurs in first position, jffta always follows
that constituent. This constituent can consist of more than word, as shown
by example 82 of Chapter 3. The point is, these clitics occur after whatever is the first constituent of the C 1 or C 2 clause (depending on the
subclass of the clitic).
Note, now, that jffta always intervenes between a verb and its object. It
does not occur after the verb-plus-object complex - even if there is no
intervening constituent. If the basic order was SVO and if the verb and
object formed a constituent, we might expect that jffta would be able to
occur after the verb-plus-object constituent, particularly when the subject
was preverbal or expressed simply by a subject pro-clitic.
In Chapter 3 we discussed the fact that jffta can alternatively follow
auxiliary-plus-verb complex, or can follow the auxiliary and precede the
verb (cf. Section 3.4.3). We discussed various aspects of this variable
placement of jffta within the verb phrase and suggested that it may have