Professional Documents
Culture Documents
Voice in Malay/Indonesian
KUINLI
st
August 21 , 2016 , 12:30-14:15
I. Purpose
• Compare the voice system in Standard Malay/Indonesian (SI) with that
found in several regional dialects of Malay/Indonesian.
• We shall see that the voice systems found in a variety of colloquial dialects
of Malay/Indonesian display varying degrees of deviation from the SI
system.
In contrast, adjuncts can be extracted (and they can even be extracted long distance,
see Cole and Hermon 2000):
2
• In the object voice
o the verb appears in bare stem form
o the patient is the surface subject, appearing typically in clause initial
position
o the agent is found immediately to the left of the verb
o the agent is preceded by negation and auxiliaries:
• The linear position of the agent after auxiliaries and negation and before the
verb suggests that its surface position is Spec of vP, the position in which
agent arguments would be expected to be generated:
3
(16) [TP buku ini [tidak [akan [vP saya [VP baca t ]]]]]
book this not will 1SG read
‘This book won’t be read by me.’
(16’) TP
NP T’
Neg ‘
Neg AdvP
tidak
Adv,
Adv vP
akan
NP v’
saya
v VP
baca t
• Agents in the object voice can antecede a reflexive surface subject (the
patient), Arka and Manning (1998).
4
(17’) TP
NP T’
dirimu T AdvP
Adv,
Adv vP
mesti
NP v’
kamu
v VP
∅
OV V’
V NP
serahkan
ti
5
‘Who has admired the flowers?’
Passive:
• This voice is formed by prefixing di- to the verb stem.
• The patient appears as the surface subject.
• The agent is postverbal.
• The passive voice marker di- is in complementary distribution with the active
voice marker MENG-.
The presence of oleh is obligatory when the agent is not immediately adjacent to the
verb:
(24) a. Tas yang ada di meja ini telah di-ambil oleh Amir.
bag that exist LOC table this PFCT PASS-take by Amir
‘The bag which was on this table has been taken by Amir.’
6
The agent in di- passives cannot be the antecedent for a reflexive in subject position.
(Arka & Manning 1998)
• A currently popular one is the phase edge analysis along the lines suggested
by Rackowski and Richards for Tagalog.
7
We will not go into details here, but such analyses try to model the fact that in
Philippine-type voice systems, the agent is not “demoted” to adjunct in passive-
like constructions.
• In the object voice the agent always occurs to the right of negation and
auxiliaries.
• This is the neutral word order and does not imply focus on the
auxiliaries/negation
• This is in contrast to the active, in which the agent typically occurs to the left
of negation and auxiliaries
• Negation/auxiliaries only occur to the left of the agent in the active if they are
focused
(29) Active
Kami tidak akan membaca buku ini.
1PL not will MENG-read book this
‘We will not read this book.’
This is also in contrast to the passive, in which the agent occurs postverbally and the
passive subject precedes negation and auxiliaries in the neutral word order:
(31) Passive
Buku ini tidak akan di-baca (oleh) Siti.
8
book this not will PASS-read by Siti
‘This book will not be read by Siti.’
(32) Passive
*Tidak akan buku ini di-baca (oleh) Siti.
not will book this PASS-read by Siti
‘This book will not be read by Siti.’
9
Summary SI
• Active
o obligatory nasal prefix (MENG-)
o neutral order of agent-neg/aux-MENG-verb
o agent = subject can also follow VP
• Passive
o di- prefix on verb
o prefix in complementary distribution with MENG-
o neutral word order of patient-neg/aux-di-verb-agent
o patient = subject can also follow VP
o agent optional
o optional preposition preceding agent
• Object Voice
o neutral order of patient-neg/aux-agent-bare verb
o position of neg/aux-agent-bare verb fixed
o patient = subject can also follow VP
o agent obligatory (excepting “pro-drop”)
• Extraction
o only subject arguments can be extracted
o apparent object extraction shows object voice word order
o There is no need for a constraint against extraction over a nasal
prefix verb. This follows from the overall system.
• Problem:
o Is SI a natural language or simply the result of language planning?
• Solution:
o Look at “real” dialects of Malay/Indonesian and see what voice
systems they instantiate.
o Do any dialects have a system like that of SI?
o What systems do we find?
10
• Results in a nutshell:
o There are dialects that are very similar to SI
o Many dialects show a breakdown of the SI voice system
o The breakdown is most extreme in one contact variety of
Malay/Indonesian (a subdialect of Kuching Malay)
o Considerable breakdown is also found in some traditional dialects
o Effects seemingly due to single, coherent system in SI must have
idiosyncratic explanation when they persist in resultant systems
o Idiosyncratic features are subject to loss.
• Contact varieties
o No unbroken chain of speakers back to protolanguage
o Formed by contact among speakers of various languages who used
Malay for inter-communal communication
• Spoken in the villages of Jernih (Jeenas in BS) Muda and Jernih Tua, in the
district of Air Hitam (Ayik Hap), northeast of the town of Sarolangun.
• Voice System
o Actives:
• As in SI nasal prefix is required (N-, not MENG-)
11
b. hendra maŋkuŋ pəɲɒhɒnt pakay kayuw
Hendra N.hit PEN.bad use wood
‘Hendra hit the thief with a stick.’
Object Voice based on the bare verb stem (auxiliaries before agent)
12
b. *sayor koh awaʔ lah bəliy diy pasar
vegetable this 1SG PFCT buy LOC market
‘I have bought this vegetable at the market.’
In the active it is ungrammatical for the auxiliary to precede the agent (except when
the auxiliary is focused etc.)
(41) Agent deletion not possible with bare verb P2, di-passive allows agent deletion:
13
c. *sepida ko didu beli.
bicycle this NEG buy
‘This bicycle was not bought.’
14
• Extraction (WH questions)
o WH in situ
o construction in which a headless relative clause is the subject of a WH
predicate
(43) WH in Situ
a. WH Object
sitiy ɲəmpunt puraŋ?
Siti N.pick.up who
‘Who did Siti pick up?’
b. WH Object
sitiy məliy apu ?
Siti N.buy what
‘What did Siti buy?’
c. WH Subject
apu di-bəliy sitiy diy pasar ?
what PASS-buy Siti LOC market?
‘What was bought by Siti the market?’
15
(44) a. WH + Headless Relative Clause - Grammatical Subject Relativization
(44) a’ CP
apuj C’
C TP
NP NP
tj
(NPi) CP
OPi C’
C TP
naŋ
16
(44) b’ * CP
puraŋ C’
who
C TP
NP NP
t who
(NP ) CP
(person)
OPi C’
C TP
naŋ
that
sitiy ɲəmpunt ti
Siti N.pick.up (that person)
Apparent object relativization is possible when the verb is employed in stem form:
17
a. Auxiliary Precedes Agent
awaʔ didu təntuw [apu [naŋ ndɒʔ kah huroh]]
1SG not know what that want 2SG ask
‘I don't know what you will ask me to do.’
Conclusion:
The voice system in BS is essentially the same as that in SI.
18
VII. Our analysis
The so-called “subject only constraint” follows from a system in which the voice head
agrees morphologically with the nominal argument that moves out of the VoiceP
(little vP in our trees below), e.g. to specifier of TP. This prevents the movement of
other nominals out of the phase because the morphological markers are subject to a
morphological uniqueness requirement. The movement of any other nominal
through the edge of the phase would require morphological agreement with Voice (or
the v head), but this would violate the uniqueness requirement on the Voice head. As
a result, only a single DP (the one triggering agreement) is able to move into the edge
of the phase, Spec of VoiceP. Since the EPP must be satisfied, the unique nominal
that moves out of the VoiceP (or vP) phase must also be the surface subject (specifier
of TP). This leads to the result that only subject nominals can be extracted, but the
extraction of non-nominal arguments and adjuncts is unconstrained since adjuncts do
not trigger agreement. While non-nominals (like nominals) must pass through the
edge of the phase, they are not subject to the morphological requirements that allow
only a single nominal constituent to escape from the Voice Phrase.
We argue that this approach is superior to the approach to Phillipine type languages
developed earlier in the literature by Rackowski, A., Richards, N., 2005. ( Phase edge
and extraction: a Tagalog case study., Linguistic Inquiry) .
In R&R the edge of the vP is defined as the highest specifier of vP. In this view,
lower specifiers of vP are not considered as instances of the vP edge, so nominals in
lower specifiers are not extractable. The result of this system is that only a single DP
can move out of vP to a higher position, regardless of whether the position to which it
moves is an A-position or an A’-position. A separate UG requirement, the EPP,
demands that the surface subject position (presumably, specifier of TP) be filled or
the sentence will be ill-formed. Thus, the single DP that escapes from the vP must be
the DP selected to satisfy the EPP, that is, the surface subject.
(a) [CP Whati will [TP youj [vP ti [vP tj [VP see ti in Montana]]]]]
19
R&R only allow the extraction of a single NP from a Phase Edge, while we will show
that in some Indonesian languages we need to allow extraction from positions that are
not in the phase edge. We therefore introduce a morphological constraint and show
that once the morphology is defective in certain regional dialects extraction becomes
allowed from non subject position. The morphological condition and its interaction
with Agree also explains why adjuncts can always be extracted.
According to the VAH, in a Philippine-type voice system the voice system has two
related but independent properties.
In English (and in all non Phillipine type languages) adjunction to vP and VoiceP and
agreement with v are presumably required if an object is to move out of vP, but the
morphology of English voice does not reflect this agreement. Rather, active voice
reflects, roughly, the fact that the agent is generated as specifier of vP and passive
reflects the alternative scenario in which the agent is not generated as specifier of vP.
The same is true of the European type passive di- in Indonesian.
20
(c1). Mengapa mereka belum menjawab surat saya?
why they not.yet answer letter 1SG
‘Why haven’t they answered my letter?’ (Sneddon (1996: 319)
The structure for (c) is presumably as follows ( see modified structures with VoiceP
rather than TP below):
(c1)’ [CP mengapai [VoiceP merekaj belum [vP ti [vP tj menjawabk [VP tk surat saya ti]]]]]
why they not.yet answer letter 1S
In (c)’ both the agent, mereka ‘they’, and the question word, mengapa ‘why’ have
moved out of vP. According to the EPH, this derivation should not be possible
because only the highest specifier of vP should be accessible from outside the vP.
Let us see how regular object extraction is blocked in this system but subject
extractions is not. If an object is extracted (such as in OBJ relativization) we get a
conflict and an ugrammatical sentence:
21
(d’) tree format
VoiceP
sayai
+Nom VoiceP
OPj
+Acc VoiceP
*voice vP
meng-belik
+Nom
+Acc v vP
ti
tj VP
V NP
tk tj
On the other hand the subject position can be relativized without a Case Conflict:
(e) Siapai yang membeli buku ini?
OPi
+Nom VoiceP
*voice vP
meng-belik
+Nom
v VP
ti
V NP
tk buku ini
22
VIII. Overview: SI-like Voice Systems
• BS:
o Roughly the same as SI.
o Extraction constraints appear derivable from voice system.
What we see in the next group of dialects is that the use of bare stem for
actives in one special environment (relativization of objects in actives) has
been generalized to all active environments.
o In TR and MD we find bare verbs in the following environments:
§ imperative
§ deverbal noun
§ object voice
§ active with object relativization
§ active declaratives without relativization
b. la di-ambɪʔ oraŋ
PFCT PASS-take person
‘It was stolen by someone.’
23
(79) duwɪt tu di-bawaʔ, ndaʔ tukar dŋan bras
money that PASS-bring want change with uncooked.rice
‘The money was brought; he wanted to exchange it for rice.’
In the active the verb can occur in either bare stem form or preceded by the nasal
prefix:
The nasal prefix occurs much more frequently than the bare stem, but both are
grammatical:
• In usage TR narratives resemble SI, BS, SL and MBB, in that active sentences
typically employ the nasal prefix.
24
• Unlike SI, BS, SL and MBB, this is a strong tendency in TR rather than a hard
and fast rule.
• The difference between TR and SI, BS, SL and MBB is reflected in judgments
of TR speakers, and not just in textual frequencies.
• Speakers uniformly judge active sentences using the bare verb stem to be
grammatical.
• Thus, there is no discord between grammaticality judgments and occurrence in
texts (except that frequency cannot be determined by grammaticality
judgments).
25
e. joko di-bagɪ duwɪt daʔ?
Joko PASS-give money NEG
‘Did he give Joko money?’
b. *ayam ko la kesot
chicken this PFCT slaughter
‘This chicken is already slaughtered.’
26
c. ayam ko la di-kesot
chicken this PFCT PASS-slaughter
‘This chicken is already slaughtered.’
b. *motor tu jual.
motorcycle that sell
‘The motorcycle was sold’
c. motor tu la di-jual.
motorcycle that PFCT PASS-sell
‘The motorcycle has been sold.’
Summary:
• Passive voice
• Object voice (same properties as the passive and object voice in SI, BS, SL
and MBB)
• Two forms of the active, actives with the nasal prefix and bare stem actives.
Extraction:
27
(90) a. *CP
apoj C’
C TP
NP NP
tj
NP CP
OPi C’
C TP
naŋ
28
(91) b. CP
Siapoi C’
C TP
NP NPi
NP CP
OPj C’
C TP
naŋ
29
(92) a. CP
apoi C’
C TP
NP NP
ti
NP CP
OPj C’
C TP
naŋ
b. CP
apo C’
C TP
NP NP
NP CP
OPj C’
C TP
naŋ
In a few instances, however, extraction over a nasal was accepted by some speakers.
30
This constraint seems to be in a state of flux in TR.
Alternations in word order like that observed in (93) are not restricted to first and
second person agents:
As in SL and MBB, informants express a preference for the object voice order
(94b), but they are adamant that the order in (94a) is well-formed and could be heard
in speech in the village.
Summary of TR
(95) a. Active Voice: N- or bare verb stem
b. Passive Voice: di-
c. Object Voice: bare verb stem; when present, negation/auxiliary precedes
agent
d. The ban on extracting over a nasal is a strong preference, but not absolute.
31
There are many relative clauses in which it is impossible to distinguish object
voice and active voice:
Examples like those of (96) are ambiguous between an object voice analysis and a
direct extraction from bare active analysis:
(97) [motor tu [CP OPi naŋ [TP ti [vP kanti aku jual ti] rusaʔ
motorcycle that that friend 1SG sell broken
‘The motorcycle that was sold by my friend broke.’
(98) [motor tu [CP OPi naŋ [TP kanti aku [vP jual ti] rusaʔ
motorcycle that that friend 1SG sell broken
‘The motorcycle that my friend sold broke.’
The fact that (97) and (98) share both the same linear string of words and the same
semantic interpretation might be expected to contribute to the erosion of the
distinction between the two constructions.
Mudung Darat (MD) is the second dialect spoken across the river from Jambi City.
MD is about 6 kilometers northeast of TR. MD and TR are to a very great extent
mutually comprehensible. However, although they share both lexicon and most
grammatical morphemes, they differ grammatically in important ways.
In the active voice nasal prefix verbs and bare stem verbs alternate:
32
1SG kick ball LOC front house
‘I kicked the ball in front of the house.’
As for TR, in MD for most verbs the preponderance of tokens of transitive verbs
employ the nasal prefix:
• The frequency of active transitive verbs with the nasal prefix is roughly
the same in the two Jambi dialects.
• Active verbs with the nasal prefix are much more frequent than actives
with the bare stem.
• However, both forms are clearly grammatical.
• As in the case of TR, speaker judgments are in complete conformity with
text counts.
33
Is there object voice in Mudung Darat?
(104) a. In the object voice, negation and auxiliaries obligatorily precede the agent.
b. The agent cannot be omitted in the object voice.
Furthermore, the order of negation/auxiliaries vis a vis the agent is generally very free
in MD:
34
d. kau bli buku ko la
2SG buy book this PFCT
‘You have bought this book.’
The very free word order instantiated above would not provide speakers with any way
to distinguish object voice from the bare active. Thus, we suggest that object voice
should not be posited for MD, and that sentences that appear to be instances of object
voice are, in fact, bare actives.
The optionality of the agent in putative instances of object voice provides further
evidence that these sentences are not in fact instances of object voice:
• These examples illustrate the free ordering of the agent and negation/auxiliaries
in sentences with bare verbs in MD.
• Of particular note is (109e) and (110), which illustrates the fact that the agent is
optional.
35
• We conclude that in MD the object voice has merged with the bare active.
• This merger results in a word order in MD that is markedly freer than that in
TR.
• Such a state of affairs is explained if, as the result of the merger of the active
and the object voice, speakers perceive the various word orders found in what
in other dialects constitutes object voice as variants on the bare active.
• That is to say, orders that are indicative of object voice in SI are instances of
scrambling in MD.
Our analysis predicts that the ordering of negation and auxiliaries will in general be
freer in MD than in TR (in all voices):
• Although MD displays a markedly freer word order than TR with respect to the
position of negation and auxiliaries, in other ways the grammar is quite
constrained.
36
• As in TR, the fronting of objects to the beginning of the clause is
ungrammatical:
MD Voice System:
(115) Active: N- and bare stem
Passive: di-
Object Voice: does not ocur.
37
c. *hape tu sayo malɪŋ
cell.phone that 1SG N.steal
‘I stole the cell phone.’
Relativization:
Extraction over a nasal prefix verb is blocked
38
(119) a. siapo naŋ joni baŋkʊŋ di pasa?
who that Joni hit LOC market
‘Who did Joni hit at the market?’
• In MD the voice system no longer consists of three voices (active, object voice
and passive) but rather of two (active and passive).
Active is marked by either a nasal prefix N- or a ∅ prefix. The N- obeys the CCC and
hence no extraction over a nasal is allowed. The ∅ prefix, however, is defective and
hence is compatible with both an agent or a theme (both Nom and Acc +N in Spec of
vP or VoiceP). Thus, in MD the occurrence of the feature matrix in (e) would be
well-formed when the voice head bears the null prefix:
The examples of (118)-(119) show that relativization respects the constraint that
extraction over a nasal prefix verb is ungrammatical. While in the case of SI and BS
this morphosyntactic constraint is general to N- and the object voice, in MD the voice
system no longer consists of three voices (active, object voice and passive) but rather
of two actives and a passive (N- active, null prefix active and passive). Although the
null prefix in MD indicates agent rather than patient extraction from vP, it exhibits the
same property seen in SL: the null prefix is exempt from the morphological
constraint blocking more than one extraction from vP. Thus, in MD the occurrence of
the feature matrix {∅ N Nom Acc} would be well-formed when the voice head bears
the null prefix:
We hypothesize that constraints that apply only to a single member of a class (the N-
prefix in our case) are more likely to be subject to diachronic loss than constraints
applying to a complete natural class. Thus, it would be expected that a voice system
like that instantiated by MD would be unstable, at least with respect to the
maintenance of the CCC. We shall now turn to evidence that such instability in fact
exists. There is suggestive evidence that speakers have started to lose the constraint
against extraction over N-. In some cases speaker judgments exhibit variability
regarding whether extraction over N- is genuinely ill-formed (although with a clear
preference for the bare form when extraction occurs). In other cases, such sentences
were judged consistently to be minimally grammatical, but still less acceptable or less
usual than some other form. Some examples that were judged to be jaraŋ (rarely
used) rather than ungrammatical are provided.
(119’) Extraction Over a Nasal Prefix Verb, judged as jaraŋ
??[naŋ baru mli maʔe t] budi ɲjual tu
that new N-buy mother-3 Budi N-sell that
‘Budi sold the one that his mother just bought.’
39
(120) Extraction Over a Nasal Prefix Verb, evaluated as grammatical, but less
preferred
than the example with a bare verb
??hen neŋoʔ budaʔ naŋ anjɪŋ lagi ŋgɪgɪt
Hendra N-look kid that dog PROG N-bite
‘Hendra is looking at the kid that the dog is biting.
Perhaps more convincing than the existence of “noise” in speaker judgments is the
occurrence of occasional examples in our naturalistic data in which there is extraction
over a nasal prefix verb. Argument WH questions are typically instantiated by WH in
situ, but examples occur in which a WH operator has been fronted in its own clause
despite the fact that this requires it to cross a verb with a nasal prefix:1
Although examples like those of (123) occur in our data, both educated and
uneducated speakers typically reject extraction over a nasal prefix, and, almost all
examples of object extraction conform to these speaker judgments and have a bare
verb. We suggest, however, that the occasional violations of the constraint against
extraction over N- may be innovations that are entering the language.
Summary:
40
• Tanjung Raden and Mudung Darat appear to be undergoing a process of
simplification of their voice systems in comparison with SI and BS.
• In addition to allowing relativization of objects (as in SL and MBB), the Jambi
dialects both permit the use of the bare verb as an active.
• In the case of TR, the bare active is in addition to the continued use of the bare
verb for the object voice.
• In MD, however, the object voice has been lost and the active (together with
the imperative) has become the primary use of the bare verb.
• While the facts are not sufficiently clear to be certain, it also appears that the
constraint against long distance extraction over a verb bearing the nasal prefix
is starting to disappear.
Object relativization over a verb with the nasal prefix is usually judged by speakers
as ungrammatical:
41
(134) a. *Apai [CP yang [TP anak itu ngebaca ti ]]?
what that child that N.read
'What is the child reading?'
Long distance extraction of the subjects of complement clauses (with a bare verb in
the main clause) is well-formed as well:
b. Siapai [CP yang [TP dia lapor-in/*ngelapor-in [CP [TP ti ngebaca buku ]]]]?
who that 3SG report-APPL/N.report-APPL N.read book
'What did she report the child was reading?'
c. Anaki [CP yang [TP dia lapor-in/*ngelapor-in [CP [TP ti ngebaca buku ]]]]
child that 3SG report-APPL/N.report-APPL N.read book
pinter banget
smart very
'The child that she reported was reading a book is very smart.'
• Loss of constraint against extraction over nasal predicted for same reasons as
for Jambi dialects.
• A few examples may suggest that this process has started, or they may simply
be speech errors.
(137) a. Nah itu [OPi yang kita sendiri belum bisa ngantisipasi ti] itu
well that COMP 1PL self not.yet can N.anticipate that
42
'Well, that, (the one) that we have not yet been able to anticipate is that.'
(A-A1)
While no regularity appears to be derivable from the data for all the speakers, subsets
of speakers do appear to have adopted a regular system.
• For one group of speakers, which includes our primary informant and others in
her age group and social class (late 20s, college educated), the apparent chaos
regarding when the nasal prefix is employed indicates that there is in fact no
synchronic rule prefixing transitive verbs with N-.
o Rather, for some verbs N- has been incorporated into the verbs stem
and the etymological stem form is no longer recognized as the verb
stem.
o For other verbs, the etymological stem continues to be the verb stem.
o In the absence of a synchronic rule affixing N- to the verb stem in the
active only the stem form is available.
o Not surprisingly, for some verbs doublets are available.
o One form exists in which the nasal has been reinterpreted as part of the
stem while in another form the etymological form exists without the
nasal.
When the nasal has been reinterpreted as part of the stem, it is predicted that the
passives of verbs taking the nasal prefix in the active will be:
(148) di-N-verb
• This subdialect of KM has gone a step further than TR, MD and JI in the
process of losing the marking of the active voice by a nasal prefix.
• For these speakers, the process of loss of the rule is now complete and the
remaining prefixes are etymological vestiges of a rule that no longer exists
synchronically in the language.
43
• Implications of the loss of the nasal prefix rule on other aspects of the
grammar of the dialect.
• In TR and MD the loss of the obligatory marking of the active by the N- prefix
is accompanied by an incipient loss of the constraint against extraction of
arguments over N-verbs.
• It would be expected that in a language like this subdialect of KM that the
constraint against movement over N-verbs would not hold.
• This expectation is correct (N- indicates etymological N-):
(169) a. WH in Situ
ʃya məlik apa?
Sya N.buy what
‘What did Sya buy?’
b. Moved WH
apa ʃya məlik t ?
what Sya N.buy
‘What did Sya buy?’
c. Clefted Question
apa nok ʃya məlik t ?
what that Sya N.buy
(170) a. WH in Situ
⎧nanam⎫
kitak ⎨ ⎬ apa?
⎩tanam ⎭
2 N-plant what
∅-plant
‘What did you plant?’
b. Moved WH
⎧nanam⎫
apa kitak ⎨ ⎬ t ?
⎩tanam ⎭
what 2 N-plant
∅-plant
‘What did you plant?’
c. Clefted Question
⎧nanam⎫
apa nok kitak ⎨ ⎬ t?
⎩tanam ⎭
44
(171) a. WH in Situ
oraŋ ya ŋambik sapa di skola?
ambik
b. Moved WH
sapa oraŋ ya ŋambik t di skola?
ambik
c. Clefted Question
? sapa nok oraŋ ya ŋambik t di skola?
ambik
As was seen in (169-171) the variety of KM under discussion has lost the constraint
against the extraction of objects over a nasal prefix, a constraint that was seen to
follow as a corollary from the voice system in SI. This constraint was predicted to be
vulnerable to loss in dialects in which the SI voice system has broken down.
Although the constraint has only shown the first signs of incipient loss in JI, loss
appears to be well under way in TR and MD, and to be complete in this dialect of
KM. This would be expected if the basis for the constraint in SI is a system in which
both A-movement and A’-movement are limited to nominals that stand in an agree
relation with v and which are thereby at the edge of vP.
As a result the existence of Ø-verb actives in TR and MD, and the more extreme loss
of the synchronic N- prefix in KM, the relationship between the agreement marker on
v and extraction is obscured. The breakdown of the distinction between Ø-verb as an
indicator of object shift to an edge position, and N- as an indicator a lack of object
shift, has obscured the relationship between agreement marking on v and extraction,
thereby permitting the extraction of objects without the need for special voice
marking.
45
(172) Cline of erosion of the Philippine-type voice system
Agent Theme Locative, goal, Benefactive,
Indonesian
BasA N- ∅
Sarolangun
∅ (defective),
allows +Nom,
+Acc features
(N- is part of
stem, ∅ is part
of stem)
Even in dialects which only preserve one overt agreement marker (Nom, subject
agreement) like MD, the single remaining voice prefix can be viewed as instantiating
a Philippine-type voice system in miniature: the constraint barring movement of a
nominal argument over a verb bearing the nasal prefix can be analyzed as following
from the requirement that N- does not allow conflicting case features. Thus, even
the last remnant of the Philippine-system can be viewed as a synchronic instantiation
of an agreement–based voice system. It is only when this last piece of voice
morphology loses its function as a voice agreement marker that the last remnant of
Philippine voice disappears and multiple arguments can be extracted without the
appearance of a voice agreement marker on the verb.
46
List of Abbreviations:
1. 1 = first person
2. 2 = second person
3. 3 = third person
4. APPL = applicative
5. EMPH = emphatic marker
6. LOC = locative marker
7. MENG = nasal prefix in Standard Indonesian
8. N = nasal prefix
9. PART = discourse particle
10. PASS = passive
11. PE-/PEN- = nominalizer
12. PFCT = perfective.
13. PL = plural
14. PROG = progressive
15. QUANT = quantifier
16. RED = reduplication
17. SG = singular
References
Aldridge, E., 2004. Ergativity and word order in Austronesian languages. Doctoral
dissertation. Cornell University, Ithaca, NY.
Alsagoff, L. S., 1992. Topic in Malay: the other subject. Doctoral dissertation.
Stanford University, Palo Alto, California.
Arka, I. W., Manning, C. D., 1998. Voice and grammatical relations in Indonesian: a
new perspective. In: Butt, M., King, T.H., (Eds.), The Proceedings of the LFG '98
Conference. CSLI Publications , Stanford.
Blake, F. R., 1925. A grammar of the Tagalog language. New Haven: American
Oriental Society.
47
Blust, R., 2002. Notes on the history of ‘focus’ in Austronesian languages. In: Wouk,
F. and Ross, M., (Eds.), The History and Typology of Western Austronesian
Voice Systems. Pacific Linguistics 518, pp. 63-80.
Cole, P., Hermon G., 1998. The typology of WH movement: WH questions in Malay.
Syntax 1, 221-258.
Cole, P., Hermon G., 2000. Partial Movement in Universal Grammar. In Lutz, U.,
Mueller, G. and A. von Stechow (Eds.), Wh-Scope Marking, 101-130. John
Benjamins Publishing Company, Amsterdam/Philadelphia.
Cole, P., Hermon G., 2005. Subject and non-subject relativization in Indonesian.
Journal of East Asian Linguistics 14.1, 59-88.
Cole, P., Hermon G., and Yanti. 2008. Voice in Malay/Indonesian. LinguaVolume
118, Issue 10, pp. 1500-1553
Chung, S., 1976b. On the subject of two passives in Indonesian. In: C. Li (Ed.),
Subject and Topic. Academic Press, New York, pp. 57-98.
Chung, S., 1978. Stem sentences in Indonesian. In: Wurm, S. A. and Carrington, L.,
(Eds.), Second International Conference on Austronesian Linguistics:
Proceedings. Pacific Linguistics, Series C, No. 61, pp. 335-365.
Guilfoyle, E., Hung, H., Travis, L., 1992. Spec of IP and Spec of VP: two subjects in
Austronesian Languages. Natural Language and Linguistic Theory 10, 375-414.
Huang, C. T. J., 1982. Logical Relations in Chinese and the Theory of Grammar.
Doctoral dissertation. MIT, Cambridge, Mass.
48
Kroeger, P. 1993. Phrase structure and Grammatical Relations in Tagalog. CSLI
Publications, Stanford CA.
Kroeger, P., 1998. Nouns and verbs in Tagalog: a reply to Foley. Paper presented at
the Workshop on Voice and Grammatical Functions in Austronesian Languages,
LFG '98 conference, Brisbane.
http://www.sultry.arts.usyd.edu.au/LFG98/austro/download/download.htm
Rackowski, A., Richards, N., 2005. Phase edge and extraction: a Tagalog case study.
Linguistic Inquiry 36.4, 565-599.
Ross, M. D., 2002. The history and transitivity of western Austronesian voice and
voice-marking. In: Wouk, F. and Ross, M., D. (Eds.), The History and Typology
of Western Austronesian Voice Systems. Pacific Linguistics 518, pp. 17-62.
Ross, M. D., 2004. Notes on the prehistory and internal subgrouping of Malayic. In:
Bowden, J., Himmelmann, N., (Eds.), Papers in Austronesian Subgrouping and
Dialectology. Canberra: Pacific Linguistics, pp. 97-109.
Saddy, D., 1991. Wh scope mechanisms in Bahasa Indonesia. In: Cheng, L. and
Demirdash, H. (Eds.), More Papers on Wh-Movement. MITWPL 15, pp. 183-
218.
Tadmor, U., 2002. Language Contact and the Homeland of Malay. Talk delivered at
the Max Planck Institute for Evolutionary Anthropology, Leipzig, 10 July 2002,
http://lingweb.eva.mpg.de/jakarta/docs/homeland_handout.pdf.
Wolff, J., 1996. The development of the passive verb with pronominal prefix in
western Austronesian languages. In: Nothofer B., (Ed.), Reconstruction,
Classification, Description: Festschrift in Honor of Isidore Dyen. Hamburg:
Abera, pp. 15-40.
49
Questions
1. How was the subject only condition accounted for traditionally and/or in other
frameworks?
3. Can you illustrate this condition with more than one construction from your own
language or dialect?
4. What is the system in a GB/Minimalism framework that would account for these
dialects and for the apparent “subject only” condition in SI and some traditional
varieties?
50