Beruflich Dokumente
Kultur Dokumente
21
The Causatives of
Malagasy
Charles Randriamasimanana
ACKNOWLEDGMENTS xv-xvi
GENERAL INTPODDCTICN 1
Introduction 2-7
Conclusions 195-200
Footnotes 201
CAUSATIVES 202-229
Introduction 202-203
Griffiths 203-205
Parker 205-206
Ferrand 206-208
Malzac 208-210
Dahl 210-213
Rajaona 213-221
Rabenilaina 221-222
Conclusions 222-224
Footnotes 225-229
viii
Introduction 230-236
Assumption 235-236
Section Is Reflexivization and
Granmatical Relations 236-249
Section 2: Reflexivization and
Pronominalization 301-320
Conclusions _ 321-323
Conclusions 414-416
Footnotes 417
Introduction 418
of Oblique 477-490
Bgui-1 495-509
Equi-2 509-525
Raising-to-DO 525-539
Raising-to-Su 539-554
Passive 580-582
Conclusions 588-590
xi
CONSTRUCTIONS 591-674
Introduction 591-592
Izany-Replacement 631-637
Pronominalization 637-639
Izy-Substitution 650-654
Passivization 669-672
Conclusion 672-673
Footnotes 674
BIBLIOGRAPHY 681-683
TABLES
Trigger 385
Table 10: Verb Classes 427
Table 11: Distribution of the Different Verbal
Affixes 449
DO and 10 661
FOREWORD
simple sentences).
mar.
Bernard Comrie
ACKNOWLEDGMENTS
and help this book would have never been published in English:
Noam Chomsky and Kenneth Hale, both of MIT, who expressed in many
1984, 1985 and 1986. I would also like to take this opportunity
of the manuscript.
January, 1986
GENERAL INTRODOCTION
ctions.
CHAPTER ONE
0.1 Introduction.
to propose a set of six factors which will account for all the
(Section 4),
(Section 5) ,
structions:
Jeanne."
punished."
In (8)a. and (9)a., Clefting has not applied, and a special in-
applied.
Section 1
Animacy
Animate;
Furthermore,
Animate;
Animate;
where fusion takes place (see Section 5) correlates with the fol-
prefix, i.e. ampa, aha, ana, or anka, meets one of the following
requirements:
the latter must be in the Affixal Passive form with the perfect-
predicate and where the sequence with the lower predicate strip-
9
2.2.2.4).
The Causative prefix amp (a) can enter into three types
of constructions:
- "neutral" Directive, when both the higher and the lower Sus are
Animate;
lower Su;
reading:
(intended reading)
reading:
11
(intended reading)
it/*he/*she him/her/*it
it *he/*she it/*him/*her
- Abilitative, when both the higher and the lower Sus are option-
ally Animate, i.e. both the higher and the lower Sus are Animate,
only the higher Su is, both the higher and the lower Sus are non-
depending cm the Animacy of the Causer and the Causee and given
V
- Both the higher and the lower Sus are not Animate, as in
Animate:
predicate:
16
logical predicate:
tafa:
punished."
behind."
17
types of constructions:
higher Su is Animate and the lower Su, not Animate (however, see
sentence (5)a.):
"It was because of the talk that Paul was fed up."
broken."
past-caus-be-deaf-of-ear Paul
b. I Paoly no n-ank-arenin-tsofina.
and ank(a), can have the Causal reading whose main characteristic
conditions:
ment NP
Thus, in:
fed-up Jeanne
i Jeanne is not an Agent since it does not have any of the prop-
erties listed:
*T»foka (izy).
fed-up (he/she)
ny fitaratra.
the glass
"Paul was deliberately smashing the glass."
no tonga i Jaona.
part arrived John
"Paul was in the process of breaking the glass when
John arrived."
25
(same as (31')a-)
upset Paul
not possible to have the sequence in (37) b. with the Agent in the
the sentence in (36) b., with the Agent in the genitive case, is
perfectly grammatical. i
28
Table 1
Underlyingly
Higher Su Lower Su
Lower Su = CAUSEE;
+ = Positive;
- = Negative ;
+/- = Optional.
29
Section 2
Control
is based on:
pect marker used or that can be used with the root of the lower
predicate;
ications.
Modification.
"deliberately."
forms voa and tafa, on the one hand, béhave as though they were
they behave like Stative predicates, i.e. the union of the set of
mate Causee (for example (17)a.), which does not involve Psycho-
and an (a) allow voice and aspectual forms of the lower predicate,
allow, in the lower clause, voice and aspectual forms, i.e. voa/-
6.2.9).
J (eanne)."
32
*"P was doing so that the bread was being cut by J."
Jeanne."
gone Jeanne
tired Jeanne
gests that the enbedded clause cannot take any Affixal Passive
with the perfective aspect-marker voa and that of (41)b. with the
ities that can occur and since only a verb in the circumstantial
35
latter is obligatory.
predicate.
past-force/past-let J fut-wash P K
past-force/past-let j fut-wash-pass-by-her
an'i P ny savony i K.
P the soap K
by J."
past-force/past-let P pass^wash-tay J K
past-force/past-let J fiat-sit p
b. Tafa-petraka i J(eanne).
pass-sit J
lasa
(48) a. i J(eanne).
fut gone J(eanne)
"J is afraid."
and the Affixal Passive no.. .ina, respectively; both are gram-
matical. By contrast, in (44) c., where the lower clause has its
despite the fact that the lcwer clause on its own, as shown in
the enbedded predicate has the Affixal Passive form with a. When
the lower predicate has the perfective aspect marker voa or tafa,
tences below.
b. H - i- petraka i Jeanne,
b. H - andeha i Jeanne,
fut-go Jeanne
c. Ho lasa i Jeanne,
fut-sick Jeanne
the active voice are grammatical, from (50) to (54). Hie ungram-
lower clause cannot have its verb in the Affixal Passive, whether
Voa
(58) a. ~ sasa ny lantoa.
ally. "
the root passive latsaka "be in the state of having been dropped"
predicate.
i Paoly.
Paul
ence between (61)b. and (61)d. is that in the latter case, the
Anp (a).
predicate amp (a) and the prefix of the verb, in this case ian,
An (a).
b. N - i - latsaka ny taratasy.
c. Sosotra i Jeanne,
angry Jeanne
c. Reraka i Jeanne,
tired Jeanne
voa or tafa:
fut-go Jeanne
(sane as (68)a.)
tripped up."
an'i Jeanne.
Jeanne
herself sitting."
an' i Jeanne.
Jeanne
herself sitting."
56
nisolifatra).
tripped-up
*"J sat dcwn on the chair (*due to her tripping up)."
ger."
i J ny boky.
J the book
?"It was because of P's anger that the book was being
predicate:
Passive.
stantial voice form or the Affixal Passive with a, which can only
was angry.n
60
Affixal Passive.
4.2.19 and 4.2.20 shows that the embedded predicate cannot have
its absence.
of the Abilitative aha and tafa, the Agent can be dispensed with.
b. *N - an - didy ny mofo.
c. *N - a - tsipy ny rano.
past-pass-throw the water
the final Su, in the first, and in the genitive, in the second.
the genitive. When the Agent is left out, the sentence becomes
dead Paul
"Paul is dead."
predicate.
past-choose P J
pass-choose- (by J) p
J part past-caus-pass-choose P
K part past-caus-pass-choose-by J P
b. Taf - iditr (i P) i J.
pass-enter-by (P) J
"J was actually brought in (by P)."
P part past-caus-pass-enter J
P part past-caus-pass-enter-by K J
In (91) d. and (92) d., the Agent is expressed and the sequences
of the cor re spending root passive—, with the root of the verb
table," i.e.
left out.
68
agent.
the two extreme poles of the continuum, we have, at one end, the
ces, where the Agent must be expressed, and, at the other, the
2.3.0 Intent.
past-go Jeanne
Jeanne."
upset Jeanne
gone Jeanne
The Affixal Passive forms voa and tafa behave like Sta-
deliberately, accidentally":
pass-hit Jeanne
pass-sit Jeanne
upstairs."
upstairs."
Table 2
Control by Causee
Activity State
Deliberate Not Deliberate
Causatives Prefix 1 2 3 4 5 6
CD
"persuasive" 0 * + * * * * * *
Directive "coercive" 0 + * + + * * * *
"permissive" 0 + * * * * * * *
Manipulative An (a) * * S S * * + +
Abilitative Aha * * s s * * + +
Causal Aha * * * * + + + +
Causal An (a) * * * * * * + +
Section 3
Entailment
3_0 Entailment.
there is no Entailment;
tailment; and
ate.
Entailment. There is, however, with the first the proviso that if
is Entailment:
fa tsy n-aha-taf-idina.
fa tsy taf-idina.
but not pass-cone-down
"but was not successfully lowered."
fa tsy voa-tonta-nyj.
but not pass-fall-hirrij
"but was not successfully made to fall*" ì. e
"Paul tried to make the table fall but did not manage.
80
fa tsy n-aha-loka.
fa tsy voa-loka-nyj.
manage."
m (113) and (115), the embedded clauses are not Stative pred-
i.e. "The scale could lift the luggage *but could not."
tailment:
"It was because of the talk that Jeanne was fed up,
se, the root passive under the prefix does not have the sane
Table 3
"coercive" 0 * + N
Directive "permissive" 0 * + N
Manipulative An (a) + * N
Abilitative Aha + + Y
Causal Aha + * Y
Causal An (a) + * Y
Causal Ank(a) + * Y
+ = Possible; or not 7
N = Absence of Entailment;
Y/N = Restriction.
86
Section 4
Productivity
4.0 Productivity.
(see (5)c.);
class, i.e. adjectives, and one closed class, i.e. root passives
with no restrictions;
one open class, i.e. adjectives, and one closed class, i.e. root
first two, i.e. the lower verb must be passivized with the
ives ,and one closed class, i.e. root passives, both being sub-
adjectives.
c. H - itsangana i Jaona.
fut-stand-up John
can be embedded under the Directive amp (a). One such case is the
sequence (4) a.
90
ive verb and again Affixal Passive cannot apply after application
under 2.1,3):
91
b. H-ivory ny vahoaka.
4.1.6 The "Neutral" Directive amp (a) and Root Passives in the
Lower Clause.
b. Taf(a)—akatra ny seza.
under Manipulative amp (a), as in (132) c., the output becomes un-
Verbs.
Paul."
94
ive Verbs.
generalization.
95
b. Potika ny fitaratra.
Predicates.
jectival clause:
After Clause-Union.
The relevant reading for the i^ prefix of the embedded verb in the
b. Latsaka ny taratasy.
Haw, although the cor re spending predicate voa/tafa exists and can
fut-go Paul
fut-stand-up Paul
soever .
Causative an(a):
99
"The brush was what Paul was using to paint the wall."
tence in isolation:
Although the root passive in (144) b. can take the voa prefix on
its own, only the form without the perfective aspect-marker can
fut-go Paul
of (145)b. demonstrates. Leaving the Causee out does not help re-
meaning.
b. Tsara i Jeanne,
good Jeanne
b. Ratsy i Jeanne,
bad Jeanne
by Affixal Passive.
John," i.e.
has to operate.
fut-go Jeanne
Leaving out the Causee NP i Jeanne from the embedded clause does
bread."
After Clause-Union.
aspect:
(sane as (116)b.)
(sane as (116)a.)
c. Adala i Paoly.
crazy Paul
"Paul is crazy."
107
rflie simplex sentence (153)c. has been embedded under the Manipul-
is passivized.
gone Jeanne
and large syntactic and not semantic in nature since no root pas-
tain adjectives:
108
fat-sick Jeanne
fut-strong Jeanne
"Jeanne is strong."
(155)b. And this turns out to be the case. Since Affixal Passive
upstairs."
to be used is voa.
Clause.
ition and Fronting of the sane NP separated from the rest of the
Lower Clause.
the lower ^clause provided the latter is in the passive form with
past-go-up Paul
"Paul went upstairs. it
pass-go-up Paul
former the perfective aspect-marker tafa has been left out. This
4.2.4 The Causal Causative Aha and Root Passives in the lower
Clause.
lower clause:
"It was because of the wind that the boat got sunk."
b. Rendrika ny sambo.
"It was because of his job that Paul's hair got white.11
Clause-Union.
after Clause-Uhian:
verbs, as in (159) d.
4.2.8 The Causal Causative Amp (a) and Transitive Verbs in the
Lower Clause.
chosen."
114
Paul."
ity of (163) c., where the Agent has been deleted from the lower
Agent.
tafa:
115
behind."
meaning.
4.2.10 The Causal Causative Amp (a) and Foot Passives in the
Dower Clause.
enfoedded clause:
floor."
b. Latsaka ny taratasy.
lower clause:
117
Clause-union.
voa for the first and tafa for the second, Clause-Union takes
is made to apply.
118
at all in Malagasy.
Lower Clause.
verb:
Lower Clause.
Causal an (a):
Clause-Union.
concerned."
Hie Causal Causative ank (a) does not allcw Affixal Pas-
and (179)c. with the transitive verb m-i-kapoka "to strike (lit-
ical.
Lower Clause.
the perfective aspect prefix voa, like latsaka "in the state of
Lower Clause.
of adjectives:
(183)a., has been embedded under the Causal Causative anka and
126
Clause-Union.
of a word) and the second, the initial of the Passive suffix ina;
the usual way, i.e. the final vcwel of ina deletes. In this par-
way whatsoever.
127
Causal Causatives amp (a) and aha to oily one for the Causal Caus-
ative anka and even then the set of predicates that can be embed-
and amp(a), the lower verb must be passivized with the perfective
the two classes it can embed are subject to the restriction that
Table 4
A B C D E F
Manipulative an(a) * * * + + +
Abilitative aha * * * * + +
Causal ank(a)
Affixal Passive
E = Root Passive
F = Adjective.
129
Section 5
Fusion
sion;
exclusively negative.
130
Three);
and
amount.
5.0.3 Preview.
ation.
grammatical. Likewise, they can all have Tsy "not" in the lower
has also applied to the embedded verbs in (187) a. and (187) b.,
0f the rule were met before Equi-1 deletes the lower Su:
himself."
In (188)a., the verb manao "to do" is in the active voice in the
past tense and the sentence sounds slightly less natural than
at the stage the process applied, both the trigger and its victim
fusion since the higher and the lower predicates are separated by
Su is the trigger:
i P(aoly) i .
P(aul) .
ization are not net. On cycle SI, Equi-2 deletes the lower Su
Construction.
between the higher and the lower predicates since the enfoedded Su
fut-go Jeanne
vtei it gets embedded under the Causative amp (a), as in (191) b.,
136
dipthongization or coalescence.
and (193)d., the lower verb is in the passive voice, first with
ing with the final vowel i^ of the verbal root didy, in isolation,
5.2.4 Pronominalization.
short for Ranona "x," an entity different from Paul but whose
tial with the higher Su. Now, if we substitute ny tenany for azy,
DO position:
139
clause and its victim, the DO of the embedded clause, i.e. Re-
tion be met, with the higher Su as the trigger and the embedded
the passive voice, shows that Affixal Passive cannot apply after
root of the verb starts with an open e or, in careful speech, the
b. Voa-idina ny saina.
pass-go-dcwn the flag
As shown in (202) b., the lower clause on its cwn with the per-
soever .
5.2.11 Pronominalization.
the latter case that the sequence can undergo Passive, as shown
tenany cannot have the reading Where the compound verb refers to
sounds more natural with the adverb tokoa, whereas this is not
between the higher and the lower predicates since the embedded Su
Passivization, as in (207)c.
placement in the lower clause despite the fact that the negative
Clause.
5.2.18 Pronominalization.
"Jeanne is good."
after Clause-Union:
of the NP i Jeanne.
sion between the higher and the lower predicates since the embed-
chy:
be-able-to-be-carried Jeanne
i
Hie predicate loka "with-a-hole," as seen in (216)a. and (216)b.,
the version with voa gets embedded under the Abilitative Causa-
shown in (216)c., where the Su i Paoly has not been fronted, or,
5.2.25 Pronaminalization.
in (218)c., where the Abilitative Causative aha has been left out
in (214) a.
lower clause:
(same as (221)a.)
lower clause:
pass-choose J(eanne)
output.
5.3.4 Pronominalization.
independent pronoun:
res a Marked word order, with the Su preceding the verb. Further-
tenany is possible:
Ihe two sequences (225)a. and (225)b., the first with the Causa-
tive prefix and the second without it, have the sane cognitive
meaning.
after Clause-Union:
b. *I J no n-aha-voa-fidi-n' ny bika-ny.
"It was because of his stupidity that Paul did not go."
behind."
b. Tafa-janona i Paoly.
pass-stay Paul
voa and the sequences (7)b. and (7)c. with a root passive in the
5 3.11 Pronominalization.
an independent pronoun:
n-iala t-eo.
past-leave past-there
spot."
past-caus-pass-hit Paul^
Malagasy:
164
In (233)a., the fact that the Su NP ny tenany has not been front-
The behavior of the Causal Causative amp (a) in (234) a., which is
evident in (234)b.
the presence of the Affixal Passive prefix voa, and then reap-
(236)c., we have the lower clause embedded under the Causal Caus-
clause does not oonprise the negation Operator Tsy and the sen-
pass-touch Jeanne
marker voa. If the latter gets embedded under the Causal Causa-
5.3.18 Pronominalization.
"It was because of the fact that Jeanne did not take
n-i-karakara-in' i Jeanne,
past-circ-take-care-of-by Jeanne
"Paul^ was upset because Jeanne did not take good care
of hirrL."
In both (239)a. with the Causal Causative an(a) and (239)b. with-
out the Causative, the NP i Paoly and the independent pronoun azy
signed the reading "it was with you-idiot that Jeanne was fed
up," whereas (241)b. would then mean "Jeanne was fed up with you-
idiot."
c. Sahirana i Jeanne,
concerned Jeanne
hierarchy:
"It was because of the talk that Paul got fed up."
The Causal Causative ank (a) does not allow Affixal Pas-
b. Voa-latsaka ny taratasy.
5.3.25 Pronominalization.
The Causal Causative ank (a) allows Pronominalization to
sentence of both the independent pronoun izy and tenany, the sec-
ization to tena:
which present only minimal fusion. Falling between these two ex-
Causative predicate amp (a) or an (a) and the verb prefix, as was
appear to be crucial:
more basic. Hiis confirms the view that there is also maximal
Table 5
N N "c" 0 0 + 0 + + + * * 0
N N "l" 0 0 + 0 + + + * * 0
N N "n" Anp(a) 0 0 + * * + + * *
Y N Manipulative An (a) 0 0 + * * *
+/* + +
Y Y Abilitative Aha 0 0 + * * * +/* + *
Y Y Causal Aha 0 0 + + + + + * *
Y Y Causal An (a) 0 0 + * * + + * *
3 = Pronominalization,
4 = Reflexivization to ny tenany
5 = Reflexivization to tena /
Table 5 (Cont'd)
N = Negative,
Y = Positive,
y/N = Either Y or N
"p" = persuasive,
"c" = coercive,
0 = Not Relevant,
+ = Grammatical,
* = Ungrammatical,
+/* = Either + or *.
178
Section 6
Markedness
6.0 Markedness.
5. Clefting is obligatory;
structions, i.e. aha, amp (a), an (a), and ank (a) are Marked;
will be seen that the Manipulative Causatives amp (a) and an (a)
since:
ces like (l)a. and (l)b. with verbs of the type of mandeha "to
ed, show that the Emphatic Pronoun criterion does not apply;
ive Causatives.
Causative constructions:
counterpart;
sa "will wash," prove that the lower predicate does not have to
be Psychological;
and (3) involving no Clefting shows that the latter is not oblig-
atory; and
(55));
3. there is no Entailment (see (110));
ent Chapter;
construction is Unmarked.
(5)b., and (5)c., where Clefting has not applied, indicates that
atory; and
is Marked.
gamation.
Causal Causative an(a) but where Clefting has not applied, there
"Jeanne got fed up with Paul," but also something like "Paul took
is Amalgamation.
(7)c. and (176)c., for example), Clefting applies, there are in-
stances like (80) b., where it does not apply, yet the output is
is obligatory; and
Causative ank(a), i.e the sentence (116)a., where the lower pred-
the last criterion since, as was shown under 2.1.1, the lower
requires that the main verb manao "to do" remain in its active
form.
Constructions.
criterion "7." This confirms the view provided under 6.1.1 and
struction with amp (a) with the lower verb in the active voice,
ction:
189
out.
construction with amp (a) meets criterion "7" and thus confirms
walk."
(254)b. indicates.
suasive" construction:
the luggage."
(256) b., where the original Su NP of (256) a. has been left out.
relevant factor.
our above claim holds. This confirms what was found under 6 . 1 . 7 ,
can be embedded without any change under the matrix clause of the
6.1.10.
Table 6.
194
Table 6
Markedness
Causative Prefix 1 2 3 4 5 6 7
"persuasive" 0 + * * * * * *
"coercive" 0 * * * * * * *
Directive "permissive" 0 * * * * * * *
Abilitative Aha * * + * * * *
Causal An (a) * + + + * + +
3 = Entailment,
4 = Emphatic Pronoun,
5 = Obligatory Clefting,
6 = Amalgamation,
7 = Embedding.
7^2. Conclusions.
both the Causer and the Causee are typically Animate, as explain-
ed under 1.0, except for the first subtype, where the Causee can
ative construction, both the Causer and the Causee are optionally
(114); and
sive, with either voa or tafa, as in the case of the Causal Caus-
atives aha and amp (a) in (121) and (122); however, if there is
(2.1.4.3) loses not only its tense but also its Stative prefix,
•ina, as in (44) d.
verb, provided the latter is not Stative ((54)a. and (54)b,; also
4.1.3).
be made between amp (a) and an (a). With the first, the embedded
with the Causative an(a), the embedded clause may include a root
(4.2.9); with an (a) and ank (a) — t o a certain extent—' the embed-
^at can occur in the lower clause becomes more and more res-
tricted. Thus, there is only a very limited set that can co-occur
ative construction with amp (a). This argues for the necessity of
ctions show maximal fusion of the higher and the lower predicates
structions, i.e. with aha (6.1.7), aitp(a) (6.1.8) (also 6.2.7 for
200
both), an (a) (6.1.9 and (6.2.8), and ank (a) (6.1.10 and 6.2.9),
which are clearly Marked and, on the other hand, by the "coer-
6.2.2), the "neutral" Directive anp (a) (6.1.3 and 6.2.3) and the
its enbedded clause (6.1.1) and when the latter has its verb in
construction (6.2.4).
201
2
References to examples in the text will be made between paren-
tion thereof.
0.0 Introduction.
anp (a), although nearly all of them in their works provide nume-
rous examples showing that aha, an (a) and ank (a) can also serve
along with amp (a), but dismissed ank (a) on morphological grounds.
amp (a): it was only with Rajaona (1972) that it became relatively
Sxit not least, his criteria are not always syntactic in nature.
which the author did not analyze into their constituent parts; he
following:
verb.
to arrange,"
irretrievably ungrammatical.
into its constituent parts. Thus, he had sequences like the fol-
lowing:
(7) m - amp - i
m - airp - iha
m - amp - aha
m - amp - a
m - aitp - an
m - aitp - ana
m - anp - anka.
206
way classification:
sive with the prefix voa and the one with the prefix tafa: the
ment with examples, but the following pair of verbs from my data
b. voa - petraka
he gave:
(91) m - an - drava
pres-caus-denolished
pres-caus-blind
pres-caus-(fut-)sick
pres-caus-miserable.
208
Ferrand, by the way, did not say anything at all about the mean-
constructions. Thus:
(15) Mankahery
pres-caus-strong
matic study of this. Thus, Malzac wrote that "active verbs with a
thened" ;
tened" ;
made to worship";
his book:
210
i.e. the past and the future —and as such could be left out— he
did not divide the passive forms into root and affixal passives.
prefix."
always has for its subject whatever phrases expresses time, pla-
ce, or instrument."
although he had examples suggesting that ana, an, aha, anka could
ship with one of its direct objects" (see Dahl, p. 193). He de-
fined direct object in purely formal terms, i.e. the word which
212
voa and that with tafa both denote Perfectivity, the only diffe-
Deliberate concept, on the one hand, and Boot Passive and the
gtilike all the authors mentioned above, who touched upon the
integrated theory.
house or not, nor whether the said house is being built or not.
pretation exclusively.
214
like:
In (23), "the judge ordered Rabe to speak" and the sentence does
(24) correlates with the fact that, in the first instance, the
215
indicates a state.
does not entail that "the child (ren) is (are) asleep" since it is
ry zareo.
they
(25) suggests that the embedded State predicate can have a Per-
Thus, we have:
216
4
1. the agentive voice ;
ina suffix.
notes: (a) that the agentive voice is different from the Causa-
since with the first, the verb has the an prefix and it is not
a Stative Causative.
218
clusive. Subsequently, within one given verb form, there can csily
be one and only one prefix" (English). But, immediately after his
pres-caus-pref-verb radical
where amp signals the Causative voice, and an marks the agentive
that "by and large, in Malagasy, a verb form can have one or more
sarily of the first degree, since they can only have one voice-
marker on the verb; and on the other hand, voices like the Caus-
Q
ative, which are not of the first degree .
219
degree and beyond are, for the most part, difficult if not impos-
llius:
pres-caus-recip-pref-verb radical
element. But, as Rajaona points out, there exist two major types
pres-caus-fat
ctivity: there are certain forms with the nank prefix which do
pres-caus-fur ious
the majority of the forms with a na prefix do not allow ank; thus
and if aha and anka are not Causatives, it follows that there
Anong others:
"whiten" "white"
pres-caus-angry
du Bara. also finds that the Bara dialect of Malagasy has a Caus-
ative prefix ag, corresponding to the Merina amp(a), with the two
On the other hand, when the embedded verb is transitive, the pre-
particular:
what pres-caus-difficult it
8.1 Conclusions.
noticed. Ank (a) is the least productive and none of the tradi-
gifted any root passive and most adjectives. But, he felt that
Kajaona 2.2.40. And because this was so, he could not view it as
most traditional grammarians did hot even suspect the exact mean-
lower clause.
224
is not.
2. or an aspect phenomenon;
tic form.
other, although his data clearly suggest that aha covers the
equative sentence type and amp (a) is the Causative prefix accom-
those made in the present study (see Chapter One, Section 7).
225
3
1he original text is in French, the translation into English is
peindre le mur je
b. M - an - doko aho
5
Rajaona (1972) opposes "voix objective" to "voix agissive." in
his 2.2.7-, p. 139, he states that the first has the following
the endpoint of the action. For the second, he has the following
features: (a) its passive verb has the a form; (b) no agent is
writes:
g
Rajaona (1972) recognizes that the circumstantial voice is a
complex one. Judging from his examples in 2.2.26-28, it encom-
(English)
latabatra.
(English)
sion." Et>r him, the "voix instrumentive" has the following fea-
tures: (a) the instrument being the Su, the verb has the a
the ana suffix and the presence of two NPs, one of which is an
Q
Rajaona (1972) eliminates an (a) as à possible Causative prefix
REFLEXIVIZATION
0.1 Introduction.
Grairmatical relations,
Semantic roles,
Cyclic convention,
Boundedness,
Command,
Unlike-Su Constraint.
231
Q 2 The Process.
faces as ny tenany.
no trigger; and
(4)
*Mahita azy^ tenany.^ aho.
Saw hiitu body-of-him.^ I
*"l saw him^ himself
(5)
*Nanome boky ny tenako i Paoly.
gave book the body-of-me Paul
*"paul gave (a) book(s) to myself."
In (3) and (4), the victim is a DO; in (5) and (6), it is an io;
Genitive.
past-lie-selfi Paul/hei
Both sequences (11)a. and (11)b. show the Unmarked VOS order,
order is SVO with the Su i Paoly to the left of the verb and the
0.6 Assunption.
b. M - i - sasa ny lamba.
d. *M - i - sasa i Jeanne,
stative-wash Jeanne
the bi-clausal source for manasa "to wash" that it cannot take a
Section 1
1.0 Introduction.
or an Experiences
237
^y Restrictions on Reflexivization.
reflexivization in Malagasy.
by anie•. as in (14)d.
1.1.2 Restrictions.
ny tena-nyi i jaona.
i Paoly^ i jaona.
Paul^ John
matical.
selfi."
i Jeanne^] i paoly]]/
Paul
on the lower cycle yields (19)c. while Egui-2 deletes the lower
Chapter Five.
241
Kuno (1975).
"You^ are the only one who should take care of your-
self."
"(You know) Jeanne, you^ are the only one who should
requested ^ will-take-care ^ p
(20) a. and (20 )b. shew the two possibilities for the speaker to
word order is the Unmarked VOS, as opposed to the same clause em-
comes Marked, i.e. turns into SVO; otherwise, the sequence is un-
Raising-to-DO):
for", i.e
tena
(25) a. *Niandry izay i Jeannehikarakara i i Paoly.
Paul
expected conp Jeannei will-take-care-of selfi
clause and then Passive can apply on the matrix verb as in (24)b.
(25)a. and the grammaticality of (25)b. Last but not least, izay
thing like "Paul is waiting for Jeanne who will take care of her-
self."
root passive verbs like tia "be liked," zaka "be bearable," atao
245
«be done," arobara "be said," aleo "be preferred," and tadidy "be
considered marginal.
ny tenany^.
the body-of-him
miserable" i.e.
ny tenanyi i Paoly
(26), but in the active voice with the Causative prefix an yields
passive.
ny tenany^?
the body-of-him^
ny tenany^i
the body-of-him
" (Well) Paul^ knows all too well what causes him^ to be
miserable1"
ny tena-ny^.
the body-of-him^.
ny tenany^.
the body-of-him^
be miserable."
248
that the two clauses have a sentential Su. Note that these are
/she". The sentence (30)b. with the clitic form of the pronoun,
i.e. -ny "by him/her," confirms the view that it is indeed in the
Genitive case.
shown that the latter is more basic. The following sentence with
i Jeanne ny tenany^.
"The fact that Jeanne did not take care of hirtu was not
Section 2
2• 0 introduction.
Constructions
250
initially,
as ny tenany.
where the trigger and its victim are clausemates, the latter goes
possible:
niatonta.
fell
"As for Paul^, he^ killed himself^ whereas Jeanne took
off and John fell."
under the Causal-Oblique has the SVO order; and in (38), we have
253
time. Compare this with the intended reading provided for each of
possible for the DO to go not only into tena but also into ny
or indirect Causation.
the grammaticality of (39) and (40), both with the SVO word order
and the no particle inserted after the Su. Furthermore, there are
(40): on the one hand, (39) holds true of a situation where Paul
prevent burglars from entering his hare and then at a later date,
In this case, T 1 and T 2 can be very far apart and not oily that
lowing tests (For further details see Chapter Six, Section 8):
burden."
256
i J(aona).
J(ohn)
reward" and "John let Paul look for a house for you-idiot."
259
up as tena.
Oblique respectively:
herselfi>"
victim must be tena in such a case, and this correlates with the
fact that the victim is in the same clause as its trigger. The
that initially tena had its trigger within the same clause, i.e.
261
the one embedded under nitetika. Notice that in both (52) a. and
(53)a. the victim tena occupies the DO position.
The following sequences show cases where the victim is
further details):
262
by Paul/', i.e.
hit."
since the lower clause is in the passive voice with the perfect-
which contrasts with the Marked order SVX (X being the Agent)
SVO, as in:
pattern, with a strong pause just before the particle no. The
"he/she."
264
Fran (56) through (58) above, the victim was a do. i<he
que, respectively:
reward."
for himself^."
2.3 Summary.
ways clausemates initially and that the victim goes into tena if
Section 3
3>0 introduction.
gest that the maximal number is three. All of this leads one to
amp (a), whereas this does not seem to be the case with the rfenj_
system of Malagasy.
3.1 Preview.
there is at least one case where they do, i.e. with the "neutral"
3.1.1 Clausematiness.
ly, as in (60)a. When they are, the victim goes into tena if it
initially the trigger was the Su i Jeanne and the victim tena, a
Jeanne Paul.
l
"Pauli had Jeanne hand him± the letter."
268
Jeanne", i.e.
i Paoly^.
Paul
The simplex sentences in (61) a. and (62) a. are the lower clauses
tive Construction.
the ungrammatically of (61) b. and (62) b., each with four argu-
ank (a) — with the proviso that the entoedded clause be of the
"Paul is good."
c. ?N - i - sasa i Jeanne,
past-stative-wash Jeanne
"Jeanne was clean as a result of washing."
crazy Paul
"Paul is crazy."
successful Paul
verbal predicate, the one shown in (64)c., but without the Sta-
is adjectival.
(68)b. Now, in the ungrammatical (68) c., the lower Su has the
If, on the other hand, the Causer and Causee are different, the
(sane as (69)a.)
higher Su. This is not the same as the grammatical (69) b. whose
Section 4
4.0 Introduction.
follow Clause-Union.
(same as (53)a.)
clause, the output feeds Equi-1, which then yields (72)b. Now, if
age, namely (73) a. and (73) b., although this does not appear to
past-order fut-kill Pi 0i J
a future tense marker ho. Somehow either one of them will be fil-
tered out or the two will merge. The word order in the embedded
precedes Clause-Union.
279
4.1.3.2 Postposition.
literally signifies "to think"; and in (83), manao means "to do".
matical.
the change in word order, from the Unmarked VS to the Marked SV.
281
b. Ny tena-ko mahay.
c. *Mahay ny tena-ny.
b., and those in c. and d. proves that only the first person ny
tena-ko "I (ny humble person)" is possible, but not *ny tena-ny
raised into the matrix clause, one obtains (86)b. If, on the
follow Clause-Union.
Amp(a) Construction.
Jeanne has now been demoted from Su to DO. (See Chapter Six, Sub-
Constructions.
as well as the Abilitative aha type since in each one of the em-
precede Reflexivization.
4.4 Boundedness.
is clause-bounded; and
contrasts with (95)b., where the trigger and its victim are in
is Paul^ himself"
Su: in (96)b., the verb has the nonperfective aspect marker no..-
in the first, the trigger and its victim, as shown in the under-
"Paul^ was angry because of the fact that he^ was not
successful."
Oblique. This suggests that the latter is not part of the clause
tamina, has the sane cognitive meaning as (101)b. with the cor-
it follows that the same may apply to (101) a. And indeed, (102),
Causative and not the other way around. It is, therefore, a piece
b. Sosotra i Paoly.
angry Paul
(104) b.
proposed derivation.
her^"
only show up in the next lower clause, in the case of the Causal-
lower clause.
1 P 1 J
S3] S2J i SlV-
P Rj Ji
1 P 1 J
Bj S2l i ]/ -
P Rj Ji
Yet, when the trigger and its victim are in two adjacent clauses,
four NPs since it only has three. Furthermore, the Unlike-Su Con-
straint is respected since the three NPs do not have the same
300
embedded:
of (60)b.
Section 5
5.0 Introduction.
- when both the trigger and its victim are clausemates, either
-when both the trigger and its victim are clausemates, either
-when the trigger and its victim are not clausemates, either only
details).
5.1 Assumptions.
the following type of structure has been factored out since the
less basic.
both the trigger and its victim are clausemates, either initially
Abilitative Causatives.
ctions :
deceived-self, Paul-
l l
"Paul, deceived himself^"
The victim goes into tena when both the trigger.and its
ctions; and
306
- optionally, when the trigger and its victim are not clause-
Causative Construction:
tenany is possible.
307
verb:
In (122) a. and (122) b., the word order of the embedded clause is
this contrasts with (122)c., where the embedded clause has the
Malagasy.
tive constructions.
the above data, where the victim precedes its trigger except in
constructions.
Causative constructions.
ed," i.e.
Pi," i.e.
"J forced P i to kill himself^"
P-,"
l i.e. "J expected P-
i toJ kill himself.."
where the trigger never commanded its victim at any stage of the
ny tenany to lift it from the lower into the main clause, the
of (l39)b., where the trigger does not command its victim. Final-
issue here.
Malagasy.
(Indirect Causation)
past-caus-kill sel^ Pi J
past-caus-kill Pi selfi J
the Abilitative and the Causal amp (a), aha, an (a) and ank (a). In
A Raising rule would then turn ny tenany into the Su of the high-
tion.
b. N-itsangana izy.
past-stand he
c. *N-itsangana-ny.
past-stand-clitic
c. *N-itadi-ny i Paoly.
Alternation.
the minimal pair made up of the sentences (39) and (40), the use
the Agent only provides the initial impulse while in the second,
Clause-Union can turn the two items into derived clausemates, and
depending on the degree of fusion between the higher and the low-
tena, the trigger and its victim show up in the same clause,
Causee cannot have the same referent as the Causer, although the
ce and despite the fact that the "neutral" Directive amp (a) can
and the Manipulative amp (a), but does not apply to Causal aha,
an (a), and ank (a), judging from the grammatically pattern found
Oblique. Hiis suggests that the latter does not belong in the
PRONCMXNALIZATIQN
0.1 Introduction.
Zero-pronoun,
Hay + pronoun,
pronoun, and
Demonstrative,
0.2 Assumptions.
b. *Rakotoi izyi.
Rakotoi he^
"Hei is Bakotoi."
Marked and the pronoun izy and Rakoto are coreferential. The se-
quence in (l)b. is also a simplex one but the vrord order is Un-
Section 1
1.0 Introduction.
bution. Specifically:
dency;
when the trigger is not animate, the victim goes into Ilay +
pronoun.
In all of the sentences from (3) to (12), the trigger and its
ungrammatical from (3) to (7), while from (8) to (12) the b. ver-
When both the trigger and its victim occupy the IX)
or a Pronoun:
"Paul looked for the person.^ but did not see him^."
"Paul looked for the person^ but did not see hiir^."
In both (13) and (14), the trigger and its victim are simul-
taneously DOs. Both the a. version with the Zero-pronoun and the
1.1.3.0 Both Trigger and Victim Are Neither Sus Nor DOs.
When both the trigger and its victim do not occupy the
Su or a DO; last, we will take up those where the trigger and its
against him^."
332
against hiiiu."
matical.
accept."
money."
obligatory.
yet happy."
against hiitK
a book."
a book."
dress."
as-soon-as past-talk (h
b. N-andefa ny enta-n' i J. i p
b. N-andefa ny enta-n1 i J. i p
matical.
344
Table 7
Anaphoric Pronominalization:
Human Trigger
Trigger Victim
SO DO 10 OBL GEN
10 *0; Pro *0; Pro *0; Pro *0; Pro *0; Pro
OBL *0; Pro *0; Pro *0; Pro *0; Pro *0; Pro
GEN *0; Pro *0; Pro *0; Pro *0; Pro *0; Pro
clitic;
* = ungramraatical;
Absence of * = Grammatical.
2. Trigger = Human;
3. Trigger = Singular.
Pronoun, except:
346
- or both the trigger and its victim are neither Su nor DO,
so pass-wait-for-by-me well
iti carefully."
it^ carefully."
in all of the above sentences, both the trigger and its victim
is not embedded within the matrix, as is the case with (41) and
victim are Sus, but that it can sometimes go into Ilay + Pronoun
but pres-afraid 0^ 0j
"P did not feed the dogi and did not care about it^"
"P was supposed to help the horse^ and not just look at
(him^)."
"P was supposed to help the horse^ and not just look at
him^."
"P was supposed to help the horse and not just look at
it. "
"Paul gave the dogi some rice since he^ was hungry."
"Paul gave the dog^ some rice since he^ was hungry."
"Paul was hiding from the dogi for hei was ferocious."
"Paul was hiding from the dog. for he^ was ferocious."
"Paul was hiding from the dog^ for it^ was ferocious."
"p chased the dogif but the latter^ did not yield."
"P did not chase the dog^, but gave him^ rice."
"P did not chase the dog^, but gave hinii rice."
"P did not chase the dog^, but gave it^ rice."
"P did not chase the dog^, but hid from hirn^."
"P did not chase the dog^, but hid from him^."
"P did not chase the dog^ but hid from iti."
356
"P did not chase the dog^ but touched hisi tail."
"P did not chase the dog.,, but touched his.^ tail."
the tail-of it
i
"P did not chase the dog , but touched its^ tail."
i
Except for (53), in the sentences from (52) to (56), oily the
(56), with the victim in the Genitive case, both (56)b. and
(53), which has its victim in the DO position with the predicate
threw-stone it,
ifc
i
"P stepped on the doge's tail and kicked iti-"
359
grammatical.
the DO position, then the victim can optionally surface into Ilay
+ pronoun:
it.
l
"P gave rice to the dogi after giving iti water."
360
tamin'ilay izy^.
with it^
ny vilia-n'ilay izy^.
"P hid from the dogi but did not give himi food."
ifc
i
"P hid from the dogi but did not give iti food."
"P hid from the dogi but did not go with himi."
tamin'ilay izy^.
with it.^
"P hid from the dogj^ but did not go with iti."
362
ny rambo-n'ilay izyi.
the tail-of i^
"P hid from the dog^ and stepped on its^ tail."
fut-give rice Oi
"P stepped cxi the dog^s tail when he/she was about
"P stepped Cxi the doge's tail when he/she was about
niafina tamy-(h.
hid from-Oj
niafina tami-ny^.
hid from-hiitij
"P was angered by the barking of the dog^ and so, went
"P was angered by the barking of the dog^ and so, went
n-itsaha-ny ny rambcMK.
"p was angered by the barking of the dog^ and so, step-
n-itsaha-ny ny rambo-ny^.
"P was angered by the barking of the dog^ and so, step-
"P was angered by the barking of the dog^ and so, step-
In (61), both the trigger and its victim occupy the 10 position
in (64) have their trigger in the Oblique and their victim in the
ation.
Table 8
Anaphoric Pronorainalization:
Trigger/' Victim
SO DO 10 OBL GEN
OBL *0;*Pro *0; Pro *0; Pro *0; Pro *0; Pro
—here Io "this."
In (70) and (71), both the trigger and its victim are Sus and the
fa tsy naharodana
!
b. TNanandrana nandrodana ny trano^ ry Paoly
"P was looking for the car^, but did not find itj."
"P was looking for the carif but did not find it^"
"P was looking for the car v, but did not find itj."
In (72) and (73), both the trigger and its victim are DOs and as
In (74) and (75), only the trigger is a Su, whereas in (76) and
(77) only the trigger is a DO. Only the c. sequences with Ilay +
was shaky."
was shaky."
when sought CK
nor a DO, then the victim goes into Ilay + Pronoun although when
, neg accepted gave the rent-of the housej Paul & assoc.
neg accepted gave the rent—of the house. Paul & assoc.
rent for the housej since they did not want to be in-
rent for the housej since they did not want to be in-
associates
and (85)d. Last but not least, notice that when both the trigger
granmatical.
Table 9
Anaphoric Pronominalization:
Non-Animate Trigger
Trigger
/
/ SU DO
Victim
10 OBL GEN
10
io io io io
io io io io
3. Trigger = Singular.
Section 2
2.0 Introduction.
herj."
"P gave her^ some advice and at the same time repri-
manded Jj."
"P gave herj some advice and at the same time repri-
manded J.."
393
"P was hoping that Jj would come, but shej did not."
b. Nanantena ny fahatongava-ny^ i p
"P was hoping that Jj would come, but shej did not."
c. *Nanantena ny fahatongava-ny^ i p
"P was hoping that Jj would come, but shej did not.
herj."
her^"
an'i J. intsony i P.
J^ no-longer P
"p used to go out with CT although he no longer sees
her.,."
sound right."
J^ sound right."
J^ sound right."
(109)a.
dia hampahatsiahy azy^ ilay resaka.
of the talk."
b.
397
^ the talk P
"Whenever P gives heri (a) book(s), he will remind ^
of the talk."
Sus; this obviously does not relate to the issue under consider-
(same as (98)b.)
(same as (93)b.) —
(same as (94)b.)
before the main clause Su, toward the end of each sentence. All
(112)b., (113)b., and (114)b., such a pause comes right after the
400
holds true of (112)b., (113) b., and (114) b. But again, this is
nominalization.
whenever the trigger does not command the victim, the sentence is
ungrammatical:
him. "
401
happen to Jeanne^."
lower clause and, therefore, does not command the victim. All of
reading.
Ny-complementizer:
"It was the giving JN money which was told herj by P,"
money."
spective between (103)a., on the one hand, and (103)b. and (103)
c., on the other; the first has a "de re" interpretation, whereas
the last two have a "de dicto" reading. Since the latter is ac-
matical.
405
Section 3
3.0 Introduction.
r
Section 3 will attempt to shew that:
c. *N-ank-adala i jeannei.
past-caus-crazy O i Jeanne^
Both (123)b. and (123)c. derive from the same underlying repres-
ical sequence, which is not the case at all. The other interpre-
where the Zero-pronoun and the NP _i Jeanne of (123) c. are not co-
hearer or both.
rules:
(same as (124)b.)
this."
410
this."
(same as (109)a.)
(same as (109)b.)
3.4.2 Derivations.
(133')c.
4.0 Conclusions.
Marked structures, not only of the type presented under 0.2 but
(134) Tonga 0?
perf-arrived 0
whose main characteristics are (a) that it only has a victim, but
or to the hearer with whom the speaker does not wish to have a
examples given.
that the four possible forms (0.1 above) of the victim under Pro-
ditioned since it can only occur if both the trigger and its vic-
the least Marked, whereas this is not the case when it has either
strative ireo.
as opposed to when both the trigger and its victim do not occupy
ation where both trigger and victim are DOs, in which case we can
3
In all of the b. sequences in 2.2.1, there is a strong pause
PASSIVE ZATION
0.0 Introduction.
gasy verbs;
Constructions in Subsection 2.
419
Section 1
0.1 Introduction.
b. voa;
c. a;
d. tafa; and
Hierarchy (1977).
Table 10.
classes, the first two of which are open, all the others being
closed. Classes 1 and 2 are the cries which give verbs that take
aspect, and between voa and tafa in the perfective aspect; and
latter is a root (this holds for class 2.M), it can take no...ina
fix;
prefix;
perfective aspect, then the latter takes the voa prefix; and
according to:
in the appendices.
"had rather," tonga "be in the state of having arrived," and avy
(4) a. Re ny vaovao.
c. *N - a - re ny vaovao i Paoly.
"be seen," very "be lost," lafo "be expensive," and afaka "be
liberated."
1
d. N - ila - in i Paoly ny boky.
prefix a.
"dare."
Table 10
Verb Classes
as Active * * * * OK
Root Used
as Passive * OK OK OK *
Active Derivative OK OK * OK *
Passive Derivative OK OK * * OK
1.1.7 Transitivity.
ther the verb must take a DO, can optionally take one or cannot:
428
have a DO:
prefix and does not have the corresponding intransitive verb with
b. l.A1, the TVP has the i prefix but not the corres-
c. l.B, the TVP has the an prefix and has the corres-
ponding intransitive verb with the i. prefix, but the latter can-
be left out:
a. l.D, the TVP has the an prefix and has the corres-
b. l.E, the TVP has the an prefix and has the corres-
(13) i.);
429
itive with i., which can take a Causal, but not an Instrumental
(see (14)d.);
(16)); and
(17));
which result frcm Object incorporation.
follows:
430
aspect:
(see (11));
c. with Subclass l.F, l.G, l.H, 1.1, and l.J, only the
ive aspect:
d. *N-i-dodona i Jeanne.
past-pref-hurry Jeanne
"Jeanne hurried."
e. *N-a-dodona i Jeanne.
past-pass-hurry Jeanne
a. *N-i-kendry i Paoly.
past-pref-aim-at Paul
*"Paul was aiming at."
d. N-i-pasaka ny ovy.
d. *N - i - loa ny vola.
a. *N - i -telina ny marary.
with Subclasses l.A, l.A', and l.B, only no...ina can be used, as
opposed to that found in (9), (10), and (11), where only the a
d. N-i-tono ny hena.
with iron."
(same as (13)a.)
/*tamin'ny tady.
d. *N-i-ripaka ny fahavalo.
past-pref-destroy the enemy
der 1.1.8 holds. The sentence (13)i. suggests that the apparent
literally "Did deposit with the money Paul," i.e. "Paul made a
(12)b. with no...ina and (12)g. with a is that in the first case,
meat," whereas in the second, "she just put it on the fire and
corporated DOs show up; if the DOs are definite and get promoted
Passive is used.
(
I,
past-pref-disappear Paul
"Paul disappeared."
b. N-i-latsaka ny taratasy.
b. *N-i-rovitra ny taratasy.
past-pref-torn-up the letter
b. *N-i-mainty ny volo.
in (18) and (19) and that between the a. and the b. versions in
(20), (20'), and (21) suggests that these are typically intrans-
pattern shows that only the circumstantial voice with an/i. .^j/-
ana is possible since only the e. versions of (18) and (19) are
above.
fa tsy voa-didi-ny.
fa tsy hita-ny.
cessfully," i.e.
vain."
"but in vain."
cut."
past-pref-stop Jeanne
fa tsy voa-jano-ny.
"but in vain."
"but in vain."
on Table 11:
Table 11
b. when the verb is a root which can take the tafa pre-
Deliberate Activity.
1.2.5 above:
The sentence (27) does not have an Agent, therefore, the bare
"managed to stand up." In (31), with both no and an (a), the verb
punishment."
ruly children."
unruly children."
First, the examples (32) and (33) contrast with (l)a. under 1.1.1
case seen in (29)a. and (31)a., where the affixes are optional
454
differs from ho, which can replace it, in that the latter refers
coded oi the verb itself. First, it will be shown that the Obli-
gasy:
1. Intermediary
2. Instrument
3. Directional
4. Beneficiary
5. Manner
6. Locative
7. Temporal
8. Source
9. Ccmitative
10. Causal
455
thermore, we have:
"Tamatave yesterday."
"with Paul."
457
noho i Jeanne.
because-of Jeanne
"because of Jeanne.ii
(40), we have the verb miaraka "to accompany" in the past tense
(portion of (39))
(portion of (40))
ii
'Jeanne accompanied Paul."
(same as (34))
marker is a verb and the Oblique NP can always show up after the
there must be a pause after the Su, for the utterance to be ac-
ceptable, but even then, as can be seen in (50) b., native speak-
it can oily occur after Su, as can be deduced from the grammati-
the preposition:
clothes."
:
b. (Tamin')io andro io no nanohatra ny akanjo i Jeanne
on this day this part past-try-on the robe Jeanne
"It was that day Jeanne was trying the robes on."
r
b. (Avy tany) Toamasina no niainga i Paoly.
luggage."
465
very good sentence, as can be seen in (56), even when the prepo-
which was available with the noun is encoded in the verb so that
ny akanjo-ny.
the clothes-their
"It was with the mud that the children were dirtying
their clothes."
i Jeanne vola.
Jeanne money
i Paoly ny trano.
Paul the house
"It was for his friend that Paul was painting the
house."
ny hafalia-ny.
the joy-her
"It was with joy that Jeanne was giving bread for
the workers."
i Paoly ny ankizy.
Paul the children
"It was in the room that the children were being
given shots by Paul."
i Jeanne ny akanjo.
Jeanne the robe
"It was on that day that Jeanne was trying on the
robes."
469
luggage."
in (62), the verb has the a form of Affixal Passive and the no.-
tional since both (63) a. without Clefting and (63) b. with Cleft-
go on the verb:
471
(same as (62)a.)
(same as (64)a.)
i Paoly ny trano.
the house."
473
ho an'ny itpiasa.
"It was with joy that Jeanne was giving bread for
the workers."
ho an'ny mpiasa.
"It was with joy that Jeanne was giving bread for
the workers."
ho an'ny mpiasa.
"It was with joy that Jeanne was giving bread for
the workers."
"The room was where the children were being given shots
by Paul."
474
"The room was where the children were being given shots
by Paul."
(see (67)a.)
"It was in the room that the children were being given
shots by Paul."
(see (68)a.)
"That was the day when Jeanne was trying the robes on."
"That was the day when Jeanne was trying the robes on."
(see (69)b.)
(see (70)a.)
(see (70)b.)
(see (71)a.)
luggage."
476
luggage."
luggage."
tion and the verb can take the a form of Affixal Passive, as is
form of Passive, i.e. with the circumfix an...an, shows the cor-
Intermediary Instrument
Instrument
Directional
(same as (35))
"Paul was "kicking" the ball with his head into the
goal."
Directional
478
"Paul was "kicking" the ball into the goal with his
head."
tamin-kafaliana i Jeanne,
with-joy Jeanne
Manner
Manner locative
"Paul was working well at Antsirabe."
Locative Manner
"Paul was working at Antsirabe well."
Temporal Locative
"Paul was studying previously at Antsirabe."
480
Source Temporal
Temporal Source
"Paul heard the news yesterday front Tamatave."
Source Comitative
Causal
Jeanne."
Comitative Souce
Causal
Jeanne."
481
Causal Source
Comitative
John."
Causal Comitative
Source
after the Su. Finally, (89) c. and (89)d. prove that it is not
Instrument Directional
ny baolina.
the ball
"The ball was being "kicked" by Paul with his head into
the goal."
Directional Instrument
483
ny baolina.
the ball
"Hie ball was being "kicked" by Paul into the goal with
his head."
Directional
Beneficiary
kers."
Beneficiary
Directional
tave."
Manner Beneficiary
484
ny nofo.
the bread
(see (85))
"Hie bread was being given by Jeanne with joy for the
workers."
Beneficiary Manner
ny nofo.
/ the bread
/
"The bread was being given by Jeanne for the workers
with joy."
Manner Locative
Locative Manner
Locative Temporal
Temporal Locative
studying."
Teitporal Source
Source Temporal
Source
Source Comitative
i Jeanne.
Jeanne
Comitative Source
i Jeanne.
Jeanne
Comitative Causal
Causal Comitative
Causal Comitative
tered without any pause between the two obliques, whereas in (90)
liques, whereas (92)b. does not; so, it can be inferred that the
clear case where Manner must precede Locative, otherwise the se-
them, the first set having a verb in the active voice, whereas
omaly.
yesterday
ner, the Manner before the Locative, and the latter before the
different subclasses.
for structures with active voice verbs; and with respect to (f)
combination with (c), (d), and (e) above for structures with
Table 12
Active Passive
2.lnstr. (t)amin(a) OK * Opt Opt Opt * Cpt Opt Opt Opt Opt
7.Temp. (t) amin (a) M OK Opt Obi * * * Obl Opt Opt Opt
Properties: a b e d e f g h i j k
Table 13
addendum:
Section 2
2.0 Introduction.
constructions.
and Equi-2;
and Raising-to-Su;
above rules in a way which shows that it follows the Cyclic Con-
vention; and
Clause-Union;
fication for:
1. Equi-1;
2. Equi-2;
3. Raising-to-DO; and
4. Raising-to-Su
will be proposed.
Passive in a way which suggests that this rule follows the Cyclic
rule out this process, as will become evident from the grammat-
2.1.1.0 Equi-1.
In the process of Equi-1, the Su of the main clause
Thus in
to be derived from:
Likewise in
to be derived from
higher Su.
that the NP i Jeanne is the Su of the higher and not the lower
clause:
The sequence (111), where the embedded verb has been passivized,
pretation at all.
Three: 1.1 and 3.1.1) are that: 1. both the trigger and its vie-
498
tim are in the same clause; and 2. that the trigger is necessari-
ly a Su.
originated in the lower clause but has been lifted into the Su
position of the main clause. This is not the case since the main
(see 2.1.4.7). This is not the case with an Equi-1 verb since the
valence is irrelevant.
(not interpretable)
the latter case, the embedded verb retains its active form;
grammatical sequence;
an ungrammatical output;
tial with the main clause Su, even in a Marked structure with hoe
"quote...unquote" complementizer.
matical:
502
ence between the two is that the first comprises the complemen-
tizer ny while the second does not. If, for the moment, it is as-
where the three dots represent the position vacated by the sen-
tential DO.
grammatical sequences:
to be derived from
has the verb of the embedded clause in the active voice and is
perfectly grammatical.
505
clause:
(see (121))
mitetika "to make plans," manandrana "to try," milofo "to stri-
ve," mikatsaka "to aim at," mikely aina "to take great pains in
se."
cyclical rule:
"It was to get his name on the ballot that Paul tried
to do."
to be derived from
[ e ho-fidi-na i Paoly G ].
s
0 0
pass-choose Paul
0i i Paoly
2.1.2.0 Equi-2.
to be derived from
to be derived from
clause.
with the DO, but not with the Su of the higher clause, as can be
of doubtful grammaticality;
and
part.
2.1.2.2 Complementizers.
(142)a. M-ianara!
pres-study
"Study!"
"Please, study!"
matical.
2.1.2.4 Postposition.
Postposition of the lower clause often yields sequences
of doubtful grammaticality:
taken place, with the position vacated by the embedded clause in-
dicated by three dots in each case, and the results are sequences
of doubtful granmaticality.
past-force P pass-wash-by J X
to be derived from
lower clause.
517
past-cajole P pass-wash-by J X
to be derived from
past-cajole Pj fut-wash Pj J X
c. [gg ho-sasa-n' i J i P ].
pass-wash-by J P
past-cajole Pj pass-wash-by J Pj X
past-cajole Pj pass-wash-by J Oj X
clause. But, the latter can only refer to the DO of the main verb
tive case. This Agent R was initially the Su of the main verb in
is irretrievably ungrammatical.
within the same clause, i.e. the main clause, a pronoun prefe-
and (151)c. with tena coreferential with the main Su are perfect-
i.e. in the first instance, one of the Equi-2 type and in the
tial with the main clause DO, the output becomes irretrievably
ungrammatical:
520
gasy, the trigger must be a Su (see Chapter Three: 1.1) and be-
following type:
himselfi/herselfif"
its victim.
521
past-force X fut-go Oj X
b. H-andeha i J.
fut-go J
(intended meaning)
b. H-andeha ny raharaha.
double passives:
522
active sentence with an Equi-2 verb and its double passive coun-
terpart:
past-force J fut-wash P X
b. No-tere-n-dR ho-sasa-in' i J i P.
pass-force-by-R pass-wash-by J P
"P was being forced by X to be washed by J."
elected," i.e.
election,"
to be derived frcm
R 1 J 1 p
S0Í iSll
Deletion yields
h-ilatsaka ho-fidi-na O^ i
1 J
i Sl] 1 p
S2Í'
J p
i
°i SO) 1 P
S2Í •
°i Sl] 1 J
i S2^'
°i Ji
2.1.3.0 Kaising-to-DO.
dicated by the three dots, has been postposed after the matrix
3. Postposition is possible;
present;
complementizer is mba;
clause;
equative sentence.
2.1.3.2.1 Complementizers.
The sentence (167) has the sane cognitive meaning as (166) al-
rence between the embedded Su, deleted in the course of the deri-
erence between the embedded Su, deleted in the course of the de-
is optional when both the main and embedded verbs are in the
active voice. Here, in (168), we have a case where both the high-
2.1.3.4 Postposition.
verbs:
bread."
531
complementizer fa is present:
most exactly the same as (170) a. except for the absence of the
Clause.
Note that the ooreference reading between the pronoun azy and the
like manantena "to expect" can refer back to the main clause Su,
construction.
ny tenany:
533
The sentence (172)a. has its verb in the active voice, as opposed
to that of (172)b., which has its verb in the passive. The pos-
an-Rasoa/ny fiara].
Rasoa/the car
Rasoa/the car."
an-dRasoa/ny fiara].
Rasoa/the car
equivalent.
535
clause is non-equative:
In both (176)a. and (177)a., the verbs are in the active voice,
manaiky "to accept," mitady "to seek," and milaza "to consider,
tenany:
circumstantial voice:
(see (163)a.)
(see (163)b.)
passive has the same cognitive meaning as (179) a. with its verb
Convention:
[
S0 h o voa
-karakara °i S0] Sl] 1 P
S2]'
comp pass-take-care-of-by P
to be derived from
ny raharaha R g Q ] i p g l ] i j g2]/,
the affair X P J
[gQ ho voa-karakara s q] iP .
camp pass-take-care-of P ]
the unspecified Agent since the relevant form here is the per-
2.1.4.0 Raising-to-Su.
coning."
form izy:
bread."
Jeanne."
2.1.4.0.2 Raising.
the bread."
the bread,"
where the three dots represent the position vacated by the embed-
ties:
clause;
543
matical sequence;
first of which has an embedded verb in the active and the second,
rearranged;
tion.
another negation:
leaving."
not be coming."
bread,"
where the complementizer fa has been deleted and where the higher
The only difference between (190)a. and (190)b., on the one hand,
and (190)c. and (190)d., on the other, is that the first pair has
sentences.
2.1.4.4 Postposition.
jective, as in:
546
tive, as in (191)a.
grammatical sequence:
tial Su. The reading in which we are interested is the one where
the matter with you?" to which the interlocutor would use (195)b.
to say "(X am) angry because Paul did not take care of himself."
cture like (181)a., where the embedded clause has its verb in the
in:
(see (117))
icate:
Su position:
(see (183))
come."
i J tsara ny ankizy. S Q ] g l ] g2 ],
to be derived from:
1 J 1 p
sol" Sl3 S2l/-
J P
"It was evident that the children were taken good care
of by J."
complementizer fa underlyingly.
under the complementizer fa, but not under the complementizer ny.
pect to the type of structure they can co-occur with, i.e. equa-
will not be relevant for the rule of Affixal Passive and will be
set aside.
555
the n£ —the two structures, one with two active verbs and the
(205) below.
car (s)."
to be derived from
to be derived from
past-plan fut-go P
going."
from that of the NP with the subscript j and where R, short for
mentizer fa, whereas this is never the case with the complemen-
tizer ny:
(see (212)a.)
retrievably ungrammatical.
sent by Jeanne."
the student(s)."
the past tense with n or in the present tense with the zero pre-
more, when the same verb mihevitra "to think, to consider" takes
sequence.
alalana "to give permission to," and manampy "to help." Yet, they
Since the verbs in both (220) and (221) belong in the Equi-2 sub-
posited under lying ly for the entire set of verbs of this sub-
Thus:
to be derived from
after Equi-1:
567
matrix:
Since ny, ho, fa, and mba are the only words in the
mba cannot co-occur with the entire set of verbs belonging in the
and fa.
underlyingly.
ment-Causative amp (a) can also take the Affixal Passive in its
lower clause; and, on the other hand, that Affixal Passive inter-
a cyclic rule.
to be evolved from:
to be derived from:
(227)c.
underlyingly
tired Paul
to be derived from
voa, or tafa;
to be evolved from
The sentence (230)a. has a verb which has undergone Affixal Pas-
tical. Hie same holds for (231) and (232). Therefore, it can be
fixal Passive.
Paul,"
letter,"
c
" /"SO kanaka n Y taratasy] noho i Paoly]]/
dropped the letter because-of Paul
ative predicate amp (a) can indeed embed a root passive or an ad-
576
Passive with the perfective aspect voa or tafa, but not one with
ed,"
c
- /ttgoNb-saz-ina i Paoly][slnoho ny ditra-ny]]/
pass-punish Paul because-of the mischief-his
578
been embedded under the higher Causative arrp(a) and the output is
• ina, as seen in (237) a., and if this is embedded under amp (a),
(same as (233)a.)
c
- /[[ggLatsaka ny taratasy][g^noho i Paoly]]/
dropped the letter because-of Paul
white,"
Passive with voa or tafa, but not one with no...ina. Frcm this,
/[
S1 N-aha-[g0 voa-sazy i Paoly] ny ditra-ny]/.
past-caus pass-punish Paul the mischief-his
(same as (236) a)
(242)b., confirms the view that the Causative aha can indeed
c
- /[[gQH-arary ì Paoly] [S]_amin' ny sakafo]]/
fut-sick Paul because-of the food
(same as (238)a.)
Causatives.
cally transitive verb. This contrasts with the case where the em-
Table 14
Equative Non-Equative
Anp(a) OK OK OK
Aha OK OK OK
An (a) OK OK *
Ank(a) OK * *
(250)a., the verb is in the active voice and the sentence is per-
pretation whatsoever.
3.1.1 Conclusions.
a. root passives;
c. affixal passives.
ive predicate.
point.
tive, Whereas the Causative an (a) can only embed a root passive
or an adjective.
operate
cyclic rule:
tion; and
CAUSATIVE CX3NSTR0CTIONS
0.1 Introduction.
tial sources.
Co-occurrence Restrictions,
Scope Ambiguity,
Transportability of Adverbs,
Reflexivization.
Exclamation Formation,
Izy-Substitution.
592
Section 1
Bi-Sentential Source
tence between:
verb;
adverb.
simplex sources and since in each case we can have only two dif-
have complex sentence sources and that they have two clauses
tial sources.
593
pres-here Paul
"Paul is here."
b. jí-any i Paoly.
pres-there Paul
"Paul is there."
past-here Paul
b. T-any i Paoly.
past-there Paul
that can appear on such a place adverb and that of the verb in
at A(ntsirabe)."
at A(ntsirabe)."
"It was the food which caused the child(ren) who was/
"It was the food which caused the child(ren) who is/
"there" in (6) must belong in the lower clause. The same holds
apparently we only have one verb and yet, in (9) and (10), which
different from that on the place adverb. The same remark applies
amp (a), (13) and (14) with the Manipulative an (a), (15) and (16)
with the Abilitative aha, (17) and (18) with the Causal aha, (19)
with the Causal an (a), (20) and (21) with the Causal ank(a). in
All the fusion sequences in a., from (25) to (33), show only two
marker on the place adverb, which does not normally co-occur with
there," i.e.
Since the place adverb XS-any "present-there" and the time adverb
complex source. And since (34)a. and (34)b. involve only two dif-
embedded predicate:
In (37)a., we have two different tense nodes, the (h) being op-
Sinple Clauses.
aspect-marker on the verb and that on the same time adverb in the
past-caus-go-pass-by P J pres-before
P pres-there at-school
narrow scope and belongs in the same clause as the embedded verb
5.1.2 for the relevant examples), the sequence has two clauses.
the examples mentioned: for example, in (42), the time adverb has
sentence, the past tense t on the adverb' triggers the past tense-
ing exairples:
607
since the zero tense-marker can only have the value "present" or
then that in the wide scope interpretation, the manner adverb has
the final position in the sequence with only a wide scope reading
can occupy the same final position, in which case it also has a
that in the narrow scope interpretation, the time adverb has been
Hie manner adverb in the a. sequences from (51) to (55) can only
also occupy the same position as above in the sequence with the
same narrow scope reading, i.e. the adverb ranges oily over the
lower predicate:
past-caus-good really P J
shows up between the compound verb and its DO, to the position
construction:
In each case, the three dots to the left represent the vacated
manner adverb has the effect of lifting the latter from the lower
from the above sentences suggests that the absolute final posi-
so occupy the same final position in the sequence with the same
wide scope reading, i.e. the adverb ranges over the higher pred-
icate:
fut-caus-enter M P in-future
fut-caus-enter M in-future P
the higher clause into the lower clause, at least, in one of the
(same as (88))
(same as (89))
fut-caus-enter M in-future P
(same as (90))
(same as (91))
(same as (92))
tafa-latsaka.
perf-dropped
ly) be posted."
(same as (93))
fut-caus-done that
voa-tatitr'i P.
perf-carry P
(same as (94))
h-aha-finaritra an'i P.
fut-caus-happy P
happy."
(same as (95))
an'i P.
tired."
(same as (96))
h-anka-leo an'i P.
fut-caus-fed-up P
fed up."
izany koa in Malagasy, where toraka means "like, such as," izany
following examples.
also past-there to M
with the NP i P; both NPs are Sus and we know from Pronominali-
zation that when both the trigger and its victim occupy the Su
1.4 Izany-Replacement.
rated from its predicate by the question particle ve, which in-
(same as (106))
fut-do that J
fut-do that J P
a. *H-anery/H-amela an'izany i P.
fut-force/fut-let that P
(same as (108))
fut-caus-do that J
fut-caus-do that J P
(same as (109))
fut-do that M
(same as (110))
fut-do that J
fut-do that J P
an'izany aza i P.
that neg P
believe this."
"Pi does not believe that he^ will be happy with his."
"P^ does not believe that he^ will be happy with this."
aza i J.
neg J
believe that."
aza i J.
neg J
"J^ does not believe that P will get fed up with her^."
"J^ does not believe that P will get fed up with heri."
tically of the sequence in (124)a. with the verb manery "to for-
ny, in which case the trigger and its victim are not in the same
at the very best marginal when the target is a DO while their de-
occurs when the trigger and its victim are not clausemates (see
"Jeanne was looking for Paul^ but did not find him^."
nary conversations.
the other.
ing with the Fusion parameter and its interaction with the Caus-
the Abilitative construction also allows the latter; that all the
Causal Causatives allow both. Otiis being the case, it can only be
Section 2
Single S Output
2.0 Introduction.
that there is, in fact, only one such Su NP. In this section, it
Fusion (see Chapter One, Section 5). The assumption here is that
b.
643
go in thereI"
in there!"
stairs."
b.
N-aha-finaritra an'i Paoly anie io (§)!
(153)a.
*N-aha-finaritra anie an'i Paoly io (e)!
b.
646
of the Su NP:
stage of the derivation where the embedded Su has just been deno-
stage, then, we seem to have two Su NPs; the first being that of
2.3.2 Izy-Substitution.
and that there is only one Su. Given that izy is Su for "he, she"
The above are the same sentences as those used in 2.3.1, except
/she." All of our examples from (166) to (172) clearly prove that
Genitive position.
2.4.0 Preview.
the b. sequence, the NP an'i Jeanne follows the NP an'i Jaona and
The first pronoun azy io in (175) can only refer to the NP ny ta-
Now, inverting the order of the two pronouns of (175) does not
show that each has a pattern of its own since a DO comprising the
nite article n£. Cn the other hand, the sequence (179)b. is gram-
2.4.3 Bo-Placement.
its left, but when it occupies the 10 slot, then the result is
grammatical:
that in the second, the implication is that "Paul and Jeanne were
659
Thus, we get:
past-fall it yesterday
(184) is the basic sentence from which all the others are deri-
ved. In (182), the definite DO ny boky "the books" has been fron-
has been fronted in exactly the same manner as the DO, but the
tinction between the Su and the DO is that only the first can
cal sequence (see (186)). It appears then that oily the DO can be
Table 15
DO 10
of them."
664
Without any exception, all three diagnostic tests prove that the
(198) suggests that we are dealing here with an 10, but the semi-
which has been promoted to Su, was originally an 10 and not a DO.
Fronting tests:
668
straight DO.
2.6 Passivization.
shown, has one Su and at least one DO, we new have to demonstrate
that there is only one verbal predicate, the lower predicate be-
least, partially)
"Peter had "Paul give the book to Jeanne."
least, partially)
J
N-ana-tsara an'i Paoly i Jaona.
(218)a. past-caus-good Paul John
"Jbhn caused Paul (to appear to be) good."
"John praised Paul."
whereas in the b. version the Causer has Control over the Causee.
possible oily with the Manipulative amp (a) (see 5.2.13 of that
3 Conclusion.
It appears, that:
tense-marker on the place adverb and the verb (1.1.1 and 1.1.2),
the place adverb and the time adverb (1.1.4), and the tensemarker
on the verb and that on the time adverb (1.1.5) which hold in a
ctions;
involving Fusion;
one;
1
I am precluding situations where Paul just began studying and,
2
See explanations relative to (48), (49), and (50) for the
adverb.
3
The construction type where the manner adverb cranes as an
factored out.
4
These constructions comprising four arguments are rare. Native
a cluster of four NPs with the use of the passive verb asaina
The main purpose of this work was to show that all the
sentential sources.
indicate that there is only one Subject Noun Phrase in the final
sequence.
676
leans slightly toward the negative pole of the continuum and the
sentential sources.
APPENDIX A
LIST OF ABBREVIATIONS
VP = Verb Phrase
V = Verb
Caus. = Causative
Part. = Particle
NP = Noun Phrase
Art. = Article
Su = Subject
ED = Direct Object
10 = Indirect Object
OBL = Oblique
GEN = Genitive
Pref./prf = Prefix
Prep. = Preposition
APPENDIX B
CLASSES OF VERBS
Verbs like mitondra "to bring, to carry" and roanka "to take,"
different classes:
fire;" eso "irony," maneso "to make fun of;" fafa "sweeping," ma-
rahina "be followed;" foy "which can be given up," manoy "to give
"lying on the ground," mandavo "to throw onto the ground," nola-
voina "was thrown onto the ground;" loka "with a hole," mandoka
"to put a hole in,™ noldhaana "was put a hole in;" pepo "with a
buirp," manapepo "to make a bunp in," nopepohina "was made a bunp
norovitina "was torn up;" tery "tight," manery "to tighten," no-
Class _4: hita "seen," mahita "to see;" lafo "expensive, been
Class 5: mila "to need," ilaina "is needed;" sahy "to dare,"
both class 1 and class 2 can take the Stative prefix i-:
ritra "to limit," mifaritra "to have its limits clearly delineat-
bump."
BIBLIOGRAPHY
Press.
Oslo, Norway.
England.
161-196.
Prentice-Hall.
Tananarive, Madagascar.
London, England.
Tananarive-Bordeaux, Madagascar-France.
Fianarantsoa, Madagascar.
University of Hawaii.
CHARLES RANDRIAMASIMANANA received his Ph.D.
from the University of Southern California in
1981 for the study of the syntax of Malagasy, his
native tongue. He is now assistant professor of
linguistics at the University of Melbourne, Park-
ville, Victoria, Australia.