Focus in Malagasy

Download as pdf or txt
Download as pdf or txt
You are on page 1of 22
At a glance
Powered by AI
The passage discusses the focus system found in Malagasy and argues that it was likely present in Proto-Austronesian based on similarities with other Austronesian languages and the improbability of independent development.

The author argues that the focus system must have existed in Proto-Austronesian based on its presence in conservative Formosan languages like Paiwan. It is also assumed to have existed in the Barito languages spoken by the ancestors of the Malagasy people.

The author provides the similarities between the Malagasy focus system and those found in other Philippine, Minahasan and Formosan languages as evidence. It is also argued that the complex identical systems could not have developed separately in such widely separated areas.

FOCUS

IN

MALAG A S Y A N D
Otto

1.

P R O TO - A U S TRO N E S I A N

Chr . Dahl

I NTRODUCT ION

1 . 1 The Malagasy verb has a focus system very similar to that found in many
Phi lippine , Minahasan and Formosan languages . We shall here study in detail the
morphology and syntax of the Malagasy system , comparing it with the grammar of
other languages and with what may be supposed to be Proto-Austronesian .
1 . 1 . 1 The phonetic development and the vocabulary of Malagasy are so similar
to the languages of the SE Barito subgroup in Kalimantan that it undoubtedly
be longs to this subgroup ( see Dahl 1 9 7 7 ) . The ancestors of the Malagasys seem
to have migrated to Madagascar about 400 A . D . ( Dahl 1 95 1 : 366-369 ) .
1 . 1 . 2 The only Bornean language of this subgroup from which more than wordlists
has been published , is Ma ' anyan . In this language there is some literature
available ( see Dahl 1 9 5 1 : 24-2 5 ) and an outline of a grammar ( Sundermann 19 1 3 ) .
From this gramma r and the texts it is clear , however , that Ma ' anyan does not
possess the focus structure . It has active forms with affixes that are re cog
nisable in Malagasy , and a pass ive form less easily comparable .
But the distance between Madagascar and the northern islands of western
Austronesia is so great that a separate development of a complicated system with
nearly identical forms in each of these widely separated areas must be regarded
as impossible . We are therefore forced to assume that SE Barito had the focus
system at the time of the emigration towards Madagascar , and that these languages
have lost it during the intervening 1600 years .
Such change s are by no means surprising. The Romance and mos t Germanic
languages have lost the old Indo-European case system in the same or even shorter
time than is assumed here . The case system is still present in southern German
and in I ce landic , that is to say on the fringes of the Germanic area.
It should
accordingly come as no surprise to find the focus system in the periphery of
Austronesia , since this merely i l lustrates the general tendencies of language
families to deve lop more rapidly in central areas than in the ir more conservative
fringes .

1 . 1 . 3 Since phonetically conservative Formosan languages l ike Paiwan possess the


focus system, it is reasonable to assume that this system be longed to PAN grammar .
1 . 2 . 1 However , before studying the Malagasy focus system in detail it is neces
sary to consider some important features in the phonetic development of the
language after its arrival in Madagascar . The Barito languages have both con
sonants and vowe ls in final position like so many other AN languages . Indeed ,

Paul Geraghty , Lo is Carrington and S . A . Wurm , eds FOCAL I :


papers from the Fourth International Conference on Austronesian
Lingui s ti cs , 2 1- 4 2 .
Pacifi c Linguis ti cs , C-9 3 , 1986 .
otto Chr . Dahl
Dahl, O.C. "Focus in Malagasy and Proto-Austronesian". In Geraghty, P., Carrington, L. and Wurm, S.A. editors, FOCAL I: Papers from the Fourth International Conference on Austronesian Linguistics.
C-93:21-42. Pacific Linguistics, The Australian National University, 1986. DOI:10.15144/PL-C93.21
1986 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative.

21

22

OTTO CHR . DAHL

this characteristic is so widespread that it is believed to have existed in PAN .


In Malagasy all words have only vocalic finals . Other phonetic changes , similar
to the phonetic development of neighbouring Bantu languages , indicate that the
immigrants found and absorbed a Bantu population in Madagas car . This Bantu sub
stratum then influenced the Austronesian language of the colonists . The Bantu
language s of East Africa have only vocalic finals , and the change is supposed
to have occurred because the substratum found i t difficult to pronounce final
consonants .

-I ,

In all dialects
- s and - h have been apocoped . To - k , - t and - r a final
vowel has been added , but -t and - r have merged into an affricate . In Merina
the result is - t ra , in Sakalava - t s e . Final -p has mostly merged with - t , less
frequently wi th - k . These changes are only word finally .
In Merina the final nasals have all merged into - n , which has added a final
-a l ike - ka and - t r a . In Sakalava the final nasals have been apocoped , l ike
- s and - h ( see Dahl 195 4 , especially pp . 34 3 - 3 4 4 ) .

-I ,

When a suffix with initial vowel is added , no changes in the wordbase are
required by the structure of the substratum. Be fore such suffixes the final
consonant of the wordbas e is therefore often maintained in the shape it now has
in intervocalic position . But be fore the suffixes - ko ' 1st pers . sg . ' and - n y
' 3rd pers . ' the n o f - n a i s also de leted.

1 . 2 . 2 Ma ' anyan has a non-phonemic penultimate stress . When a final vowe l was
added in Malagasy , the number of syllables increased by one . But the accent has
remained on the syllable that was penult before the lengthening . In words ending
in - ka , - t ra or - n a in Merina it now there fore falls on the antepenult . Like
vowels that have come into contact by loss of a consonant , have been contracte d .
The s ame has occurred when a suffix with initial vowel has been affixed to a
word ending in the same vowel . This reduces the number o f syllables . But the
accent remains on what was the penultimate vowel be fore the contraction . There
fore the accent may now also fall on the ultimate syllable .
As a result of these changes Malagasy places
a phonemic accent on one of
l
the l ast three syllables of the word , e . g . Mer t a nana hand arm PMP * t a a n
id . ) , t a n a n a vil lage town * t a n a - an < PMP * t anaq land + * - a n ) , mana l a t o take
away , mana l a take away ! ( imper . < *ma - n - a l a- a < PAN *ma - - a l a + - a ) .

2.

ACTOR FOCUS

As a rule languages wi th focus structure have four different focus forms


generally cal led actor focus ( AF) , ob j ect focus ( OF) , referent focus ( RF ) , and
instrument focus ( IF ) . All these are found in Malagasy.

2.1

Actor focus , which has the performer of the action in focus (mainly as
sub j e ct ) , is often formed with the infix - um- in the Phil ippines and Formosa .
In Malagasy dialects there are still some very few cases of this . In Sakalava
we have l - om- a0 2 to swim < PAN * + - um- a u i , t - om- ay to cry < PAN * t 2 - um- a i t ,
h - om-ehe to laugh , c f . Old Javanese kaka l id. , l - om- ay to run. 3 These forms are
present tense . In the past tense the infix is replaced by the prefix no- :
no- l a o , no- t ay , no-hehe , no- l a y .

2 . 2 In Sakalava we also have h - 6m- a to eat < PAN *k- um- a ?an . This word is
pre sent in Merina too , in the form h - 6m- ana , with the regular development of

FOCUS IN MALAGASY AND PAN

23

final n into na . In this word the crasis of a + 8 has already taken place i n
Kalimantan , cf . Mny kuman id . The word has thus come t o Madagascar in this form ,
with the stress on the U . The fact that the wordbase was so short explains the
abnormal accentuation of the infi x . The word has consequently not been regarded
as a form with infix , but rather as a wordbase used as verb with no separate
form for the past tense .
In Merina , homa na is the only word of thi s shape used as a verb . The first
three forms above are also considered as wordbases in Merina , and form AF wi th
the prefix m i - : m i - l omano to swim , m i -tomany to cry , m i - h omehy to laugh . So
does the fourth one , but this has been trans formed by popular etymology into
m i - o l o ma y to run with haste , lit . to act as a burning person.
-

2.3

The regular formation of AF in Malagasy is wi th the prefixes ma- + nasal


substitution or accretion , generally transitive , and m i - ( probably < PAN *ma y - ) ,
often intran s i tive . Besides l om a Qo to swim ( intr . ) mentioned above , Sakalava
has ma n - d a Qo to cros s swimming ( tr . ) . Other examples from Merina : mamabo or
mam - b a b o to capture> seize as prey , mame r i n a to send back ( tr . ) , m i - ve r i na to
come back ( intr . ) . A few wordbases form AF with ma- wi thout nasal substitution
or accretion , e . g . MIg ma- h i ta to see < PAN * k i t , a ? i d . Similar verbal forms
are found in Ph ilippine languages .

2 . 4 In the past tense the in itial m of all these prefixes is replaced by n , and
in the future tense by h , e . g . n a - h i ta saw , h a - h i ta wi l l see . The origin of the
n is probably that the prefix has got the infix - i n - : m- i n - a - , as seen in some
AN language s . Thereafter the initial syllable has undergone aphesis : m- i n - a - >
n a - . In other languages this - i n - more often seems to indicate perfective aspect
than past tense . But a relation between perfective and past is qui te under
standable .
The origin o f no- in the past tense of the above-mentioned Sakalava verbs
is probably the same . In some AN languages there is a prefix mu / mo- with the
same function as the infix - um- . I suppos e that the prefix has been the original
form , and that the infix has come into being by metathesis wi th the initial
consonant of the wordbase : mu- C . . > C - um- . . . Sak no- is then *m- i n - u - > n u
4 We shall see that in
following the same process of deve lopment a s above .
Malagasy this no- has developed a broader function as formative of the past
tense , even in forms that do not contain - um- .
.

2.5

To the h- of the future tense I have not found any parallel in AN language s .
I n other verbal forms the morpheme o f the future i s ho . I t is possible that thi s
is due t o the Bantu substratum . In neighbouring Bantu languages a pre fix k u / h u
is part of the future morpheme . The substratum , which had a verbal system with
tense s , may have felt the nece ss ity of a future tense in the verb , which the
language of the AN immigrants lacked ( see Dahl 1954 : 3 5 5-360 ) , and it has also
trans formed the AN perfective aspect into a past tense .

3.

OBJECT FOCUS

3 . 1 The morpheme s of ob j ect focus in Phi lippine and Formosan languages are
reflexes of PAN * - an . In Malagasy too we find the same with regular phone tic
reflexes : Mer - i n a , Sak - e . In Sakalava the reflex of PAN *8 is e in all pos i
tions . In Merina i t is e in the accentuated syllab l e and in the preceding ones ,

24

OTTO CHR . DAHL

but in the syl lables fol lowing the accent i t is i . Since PAN *-an was in the
ultima , i t was not accentuated in Barito , and therefore neither in Malagasy , e . g . :
( 1 ) teh e n
- i n a ny 1 a k a n a
push forward OF the canoe
the canoe is pushed forward (with a staff)
< PAN * t 2 aka +

- an , cf . Mer teh i n a staff, stick ( cf . 1 . 2 . 1 and 1 . 2 . 2 above ) .

3 . 2 But i f the wordbas e had final e in Proto-Malagasy the two vowels in contact
are contracted to an accentuated e , e . g . Sak t e re to be mi Zked , Mer te rena id . ,
to be pressed , from * te re-en < PMP * tad 1 at ' + - a n (with apocope of a final s ) .
I f , however , the wordbas e had final i in Proto-Malagasy , Merina shows contraction
into ( , e . g . Mer f i d ( n a but Sakalava f i l ( - e to be chosen from PAN * p i l i q + - an
( apocope of q already in Barito) . The rule of contraction of like vowels has
thus had i ts e ffect both before and after the change of PMlg *e > Mer i in syl
lables following the accent .
3 . 3 In th e future tense all dialects have ho- be fore consonant and h- before
vowe l , c f . 2 . 5 above . In the past tense Merina has no / n - in harmony with this .
But in Sakalava we often find in the past tense the same as in phi lippine lan
guages : the suffix is omitted , and instead we have the infix - i n - , e . g . to
f i l ( - e is chosen we have f - i n - ( l y was chosen , and to l a Q 6s - e to be swum in,
across we have l - i n - a Qo uns swum i across . This correlation between -an and
- i n- seems to be old in AN . But it is also possible in Sakalava to prefix n i
before the whole form with - e , e . g . n i - f i l ( - e was chosen.
3 . 4 This form has in focus the obj ect suffering the action , and this is the
subj ect of the claus e , see example ( 1 ) above .

4.

RE FE RENT FOCUS

4.1

The morpheme of referent focus in Philippine and Formosan languages is - a n


in almost a l l languages where i t occurs , and this has also been supposed t o be
i ts PAN form . Starosta , Pawley and Reid have , however , reconstructed it as PAN
* - a n a , b as ed on Oceanic , Malagasy and Tsou ( 1982a : 16 3 , 1982b : 104) .

4 . 1 . 1 It i s correct that - a n a is the morpheme of RF in Merina and some other


Malagasy dialects . But the f inal a in this suffix is an innovation in Malagasy
due to the transition from consonantal to vocalic finals , see 1 . 2 . 1 above . That
this is so appears c learly in Sakalava and some other dialects , where the RF
suf fix is -a . Here 0 is the regular reflex of final n . The deve lopment of - a n a / a
from PMlg * - a n i s exactly t h e same a s t h e development o f O F - i n a /e from PMlg *-an .
If the suffix had been * * - a n a in Barito , the penultimate a would have been
accentuated , and would have continued to be stressed in Malagasy . But Mer - a n a
is unaccentuated, except when there is contraction with a final accentuated a in
the preceding syllab le of the wordbase , see 1 . 2 . 2 above , e . g . s o r a t - a n a is wri tten
(on) < PAN * t ' u r a t 1 + -an , a l a n a is taken away < a l a - a na < PAN *a l a + - an . S
Mer - a n a can therefore not be used as an argument for PAN * * - a n a . Compar
ative linguistics is diachronic linguistics , and a merely synchronic comparison
wi thout attention to the historical background may be mi s leading .

FOCUS IN MALAGASY AND PAN

25

Ma ' anyan , which now has n o focus forms , has a suffix - a n , never * * - a n a ,
that is used in derivative forms wi th di fferent meanings . And as far as I know
no other language s in Western Austronesian have the suffix in the form - a n a .
There i s therefore no valid argument from Western Austronesian for a PAN * * - a na .

4. 1 . 2

The three authors ' argument concerning Tsou depends on Tung 1964 : 174- 1 7 5 .
Under the heading " derivational suffixes " Tung writes :
I - a n a l , attached to certain conj oined words ( being place
and clan names in mean i ng ) , is very much like the English
suffixes ' -place ' , ' -town ' , ' -man ' , ' smith ' and so on in
function .

The combination with clan names that may also be derived from names meaning
human profe ssions , rai ses the question as to whether the function of the suffix
is really locative . At any rate , it is not a morpheme of RF. In Tsuchida 1976 :
102- 103 we find that " the location focus marker is - i " .
Tsou has , like Kanakanabu and Saaro a , vocalic finals . Tsuchida says (p . 88 ) :
" In word final position a morphophonemic form ending in a consonant or stressed
vowel is re ali zed with a supporting vowel" .
We cannot there fore discount the possibility that Tsou - a n a may have devel
oped from * - a n . However , as supporting vowels Tsuchida ci tes i , U and a , but not
a in modern Tsou . A deve lopment - a na < * - an is thus not certain . However , on
balance , the argument for PAN * - a na from the derivational suffix - a n a is far from
convincing.

4 . 1 . 3 Starosta , Pawley and Reid do not give any detai ls about - a n a in Oceanic ,
either where i t occurs or i ts function , and I have not had the opportunity to
study i t clos ely . If it does occur sufficiently often there , * - a n a may be con
structed as a Proto-Oceanic innovation , but not as PAN . From the very frequent
occurrence of the form - a n of the RF suf fix in Formosa, the Phi l ippines , and in
Proto-Malagasy I consider * - an to be the mos t l ike ly PAN form .
4.2

Before studying the use of - a n a / a in Malagasy we have to note a phonetic


feature in Merina . If the final vowe l of the wordbase is or has been e , Merina
has crasi s of e + a into e , but Sakalava has generally not , e . g . ome 6 to give +
- a ( na ) is Mer omen a , but Sak ome a to be given.
-

4.2 . 1

What is focused with the form wi th - a na/a in Malagasy is not so uniform


as with - i na /e . It may be the place where the action is located , e . g .
( 2 ) Mer t o t 6 f - a n a t a n y n y l avaka
fi l l RF earth the ho le
The ho le (in the ground) is being fi l led with earth.

Here the direct ob j e ct of the action is the earth , t a n y is constructed as such ,


and the hole is the location of the action and subject of the clause .

4.2.2

In other cases the person profiting from the action , he who receives the
direct obj ect of the action , is in focus and constructed as subj ect , e . g .
( 3 ) Mer to l 6 r
- a na fa nome z a n a ny vah fny
hand over RF gift
the guest
The guest is presented with a gift .

26

OTTO eHR . DAHL

The gift is the direct obj e ct .


two ob jects :

I n AF both these verbs may be constructed with

( 2a ) Mer ma not ot ra t a n y n y l a vaka IZY


He fi l ls the ho le with earth. O R

He fi l ls earth into the ho le.

( 3 a) Mer mano l o t ra fa nomezana ny va h ln y IZY


She gives the guest a gift , or

( 3b ) Mer mano l o t ra fa nomezana ho a n ' ny vah l n y IZY


She presents a gift to the gues t , with one obj ect and one complement .

4.3.3
(4)

But in many case s a direct ob j ect seems to be in focus , e . g .

Mer mama fa
toko t a n y IZY
AF sweep courtyard he
He sweeps the courtyard.

( 4b ) Mer f a f a n a
ny toko t a n y
sweep RF the courtyard
The courtyard is being swept .
An explanation for the apparent anomaly may b e that i n this case the courtyard

is both ob j e ct and location of the action .


the s ame interpretation , e . g .
(5)

And there are other verbs that al low

Mer s o ra t - an a ny t a r a t a s y
wri te R F the paper
there is written on the paper or the letter is written

But in
(6)

Mer so r a t - a n a n y t e n i - ny
wri te RF the word his
his words are written down ; only a direct obj ect is in focus .

4.3.4

AF of ma n - ome can take two obj ects , and both may be focused by

(7)

ome na :

ma n - ome n y vah l n y
ny s a kafo i an a o
you
A F give the guest (s) the food
You give the gues t (s ) the food.

( 7b) Mer omen


- a o ny vah l n y
ny s a ka fo
give RF you the guest (s) the food
The food is given (to) the guest (s) by you.
(8)

RF

Or

Mer manome s a ka fo ny v a h l n y i a nao


You give the guest (s) food.

( 8b ) Mer omen ao s a k a fo ny vah l n y


The guests are given food by you.
-

In the latter case the focus is benefactive , but in the former there is no trace
of benefactive or locative .

4.3.5
(9)

Many forms with - a n a / a have only the direct obj ect in focus , e . g .
a r la n a n y f a ko f a ko
the rubbish
throw away
The rubbish is thrown away .

FOCUS IN MALAGASY AND PAN

27

In such cases -ana has the same function as - i na , and i t is often impossible to
understand why - a n a is chosen instead of - i n a .

4.4

I n Merina forms wi th - a n a
no/n- , future tense ho/h - . In
- a , but never the i nfix - i n - .
forms with - a n . The name of a
a t the stone where things have
fand r a ka chise l .

5.

have the same tense prefixes as - i na , past tense


Sakalava we mostly find n i - and ho- combined wi th
However , Malagasy must earlier have used - i n - with
certain town in Betsi leo is Am- ba t o- f- i n - a nd r a h - a n a
been chise l led or a t the chisel led stone , from

I N ST RUMENT FOCUS

5 . 1 . 1 In many Phi l ippine languages there is a form wi th i - focusing the instru


ment or the means used to perform the action . In Formosan languages the prefix
is generally s i - , but in Bunun i s - . From these reflexes I have reconstructed the
prefix as PAN *S i - ( Dahl 19 7 3/76 : 119) .
5 . 1 . 2 However , Starosta , Pawley and Reid do not find my reconstruction suffi
ciently motivated , and prefer to reconstruct i t as PAN * i S i - with the fol lowing
motivation :
Dahl . . . reconstructs this form as *S i - for PAN , in spite
of the fact that this would be expe cted to produce h i - in
Tagalog , rather th an the ? i - that is actually attested
In Bunun , there is a similar form , but it is i s - rather than
s i - , and marks future AF as wel l as I F .
. . . the recons truc
tion of * i S i - provides a better explanation of the reflexes
in Bunun and Philippine languages than does * S i - . Bunun
i s - can be accounted for as a result of vowel loss rather
than metathe sis , whereas Phil ippine ? i - forms can be assumed
to have developed by reduction of the Phil ippine reflex * i h i
to * ? i - . Northern Philippine languages which reflect PAN *S
as glottal stop ( or zero) would have reduced * i ? i - to * ? i - .
A few Philippine languages still show h i - rather than ? i as the IF prefix .
( Starosta , Pawley and Reid 1982a : 16 5 )
.

5 . 1 . 3 But a s far a s I know n o language other than Bunun has the sequence i s - ,
and no language has re flexes of all the three phonemes in * * i S i - . I f this was
the original form, we should expect the form * * i h i - in some Philippine language .
In pr ivate correspondence R . D . P . Zorc has given me the following survey of
th e reflexes of the prefix in Phi l ippine languages :
There is no Philippine and no other Formosan evidence for
a PAN * i S i - , only *S i - . The only Philippine language that
gives clearcut evidence for *S i - is Tausug with a h i
punctual instrument and a h i pag- durative instrumental
prefix , i . e . , some form of h i - ( with h < *S ) is retained
throughout the gramma ti cal system . Samar-Leyte has ma h i
and n a h i - in the potential instrumental sys tem , but simply
? i - in the punctual and durative ; similarly , A}- lanon has
an accidental instrumental prefix h i - ( future , in contrast
with a h a - past < PAN * S a - ) , but otherwise uses ? i - as the
normal ins trumental prefix in the durative and punctual

28

OTTO eHR . DAHL

systems .
( See Zorc 19 7 7 : 117-118 , for a description of
the durative vs punctual systems . ) Obviously , the Aklanon
and Samar-Leyte evidence is conflicting ; Ak l seems to
indicate a split of PAN * S i - into an irregular ( i . e . loss
of h < * S ) form normally used in the system, wi th the
accidental form ( if from the same PAN *S i - ) retaining the
h . Same problem wi th S-L.

5.1.4

I know no case of PAN intervocalic *S into Phi lippine ?


The intermediate
form * * i ? i - is therefore very hypothetical . In initial position , however , some
Phi lippine languages have cases where ? seems to reflect PAN *S ( see examples in
Dahl 1981 : 4 5- 46) . I f we as sume that the proto-form of the prefix was *S i - , it
thus fits better wi th the reflexes in Phil ippine than * * i S i .
Zorc mentions Tausug and Aklanon as languages with an uncomposed h i - . And
both i n these languages and in Samar-Leyte we have h i - in composed prefixes .
There are there fore reasons to believe that the Proto-Philippine form of the
prefix was * h i - , and that the h was later lost in most Phi lippine languages . The
h h as the weakest articulation of all consonants , and a regular or irregular los s
of i t is therefore frequent in the hi story of language s . French orthography shows
that this language has lost h twice . Malagasy is now losing it for the third time
in i ts hi story : 1) PMP *h < PAN *S , 2 ) SEBarito h < Barito 5 < PAN * t ' , 3) and now
the Merina dialect is losing h < PAN * k . That h is lost in a prefix , even irreg
ularly , is not very surpri sing . A syl lable at some distance before the accentu
ated one has often a feebler articulation , and its frequent use further weakens
it . Pronounced distinctly or not it is always understood from the context .
When this initial h had been dropped , the i - was in initial posi tion . Many
Phi lippine languages have developed glottal onset to initial vowels , and auto
matically the IF i - must also be articulated in the same way . I therefore assume
that the glottal stop here does not directly reflect PAN *S , but has developed
secondar i ly as a normal part of the articulation of initial i .

5 . 1 . 5 I f the original form of the I F prefix was *S i - , we have to explain how it


has be come i s - in Bunun . Metathesis is frequent in this language , especi ally in
the I sbukun dialect . Compare the following forms in Bunun dialects : Metathesis
of consonants : Ttd , Tkb l i s av , I sb s ( l a v Zeaf; Ttd , Tkb qops ( l , I sb x 6s p i l hair.
Metathesis of vowel and consonant : Ttd , Tkb q a 1 6a ? , I sb ? a x 1 6a ants ; PAN
* t 2 a 1 i QaH 2 , Ttd t a i Qa h , Tkb t a ( Qa ? , I sb t a Q (a ear (metathesis of vowe l and con
sonant or of Q with an original 1 that was l ater dropped) . Metathe sis of vowels
occurs in all dialects : PAN * a t ' e Q , Ttd i s ? a Q , Tkb i s ? a : Q , I sb f s ? a Q breath; PAN
* q a S , e l u [ H 2 ] , Ttd qos aoh , Tkb qosa : o? , I sb x osao pestZe ( Tsuchida 197 1 : 4 , 6 , 9 , 1 3 ,
19 ) . Wi th this frequent occurrence in Bunun the hypothes is of metathesis from
*S i - into i s - seems very reasonab le .
5 . 1 . 6 In addition , the morphology of the IF prefix in Bunun gives important
information about i ts history . In i ts past and perfective forms it is combined
with the infix - i n - , and in this form is s - i n - wi thout the initial i ( Ferrell
1972 : 1 2 3 ) . When i n is infi xed , its place is always behind the i ni tial consonant .
When i t i s affixed to a word with ini tial vowel , it is prefi xed . The composed
prefix should thus have been Bun * * i n - i s - if it had initial vowel at the time of
the combination of the two morphemes . The affix i n is a very old morpheme in
Austronesian ( c f . Starosta et al . 1982a : 16 3 ; 1982b : 1 2 1 ) , and the combination of
the two morphemes is therefore likely to have taken place far back in history .
The form s - i n - reveals that the IF prefix had ini tial s when the combination

FOCUS IN MALAGASY AND PAN

29

took place . In my opinion this proves that the original form of the prefix in
Bunun was s i - . A metathesis of i t has thus taken place later , produced by the
trend towards metathesis found in this language .

5 . 1 . 7 In Atayal the I F prefix is 5 - (Egerod 1965 : 26 9 ; Ferrell 19 7 2 : 12 4 ) . 7 I n


this l anguage the vowel fol lowing the initial consonant is very often lost in
the non-active focuses ( Egerod 1965 : 2 5 5 ) , which explains the reduction of * S i
into 5 - . That this *S in Atayal has the reflex 5 and not h , shows that the proto
form of i t was PAN *S 1 '
5 . 2 . 1 I have shown that the modern Malagasy morpheme of thi s focus is not i ,
which would be the normal reflex of *S 1 i - , but i t i s a - . Moreover , this form
has not only the instrument in focus , but even more frequently a moving obj e ct .
wi th the prefix i - the same is seen in some Philippine languages . No pre fix
cognate to *S 1 i - was found in Malagasy when I treated these prefixes , but I
assumed that it had existed earlier in the history of thi s language ( Dahl 19 7 8 ,
especial ly p . 389 ) .
5 . 2 . 2 One of the Malagasy dialects , Antemoro , has a literary tradition , at least
500 years old , written in Arabic script . No texts exist that can be proved to be
so old, because they are written on a locally produced paper which is not suf
ficiently durable . The olde s t texts have therefore been copied several times ,
and may have been ' moderni sed ' by copyists . But magic texts have a more archaic
language than the others . The least change in a magic text may cause the loss of
i ts magic power , and it must therefore be copied more s crupulous ly . In these
texts I h ave now found instrumental forms with i - , which do not exi st in modern
Malagasy , e . g .
( 10 ) s 6 r a t s y h i - t a vo- ny ama h i n6m- i - ny
writing IF anoint he and drink OF he
Writing with which he sha l l anoint himself and which he sha l l drink .
Here h - i - t avo-ny is future tense of IF with i - and with third person pronoun
suffi x , and h- i nom- i - n y is future tense of OF with - i ( n a ) and the same pronoun
suffi x . The written magi c text is suppos ed t o b e dissolved in water and is the
means to be used for anointment and the ob j ect to be drunk .
However , the same texts also contain forms wi th the prefix a - , but these
have a moving obj ect in focus , e . g .
a - 6 roy
a - f l i y
( 11 ) ron6no - n ' 6 1 0n
mi l k
gen . human being bring down in nose
Human mi lk is poured by him into his nose.
( a l l i y is composed of a - I l i + - ny , and a - 6 roy of a locative prefix + 6 ro +
- n y , both with assimilation of the nasals ) ( Dahl 1 983 : 36-38 and 19 7 1 ) .

5 . 2 . 3 I proposed in 1978 as a tentative hypothesis that two prefixes , one with


an instrument in focus , the other with a moving object in focus , had merged and
combined the two functions ( Dahl 197 8 : 389) . Thi s was a gue s s , because no form
wi th i - had then been observed in Malagasy . But now this seems to be corrobor
ated by the discovery of the two forms in Malagasy . How far this is valid only
for Malagasy , or for old Austronesian in general , is an open question .
Starosta , Pawley and Reid have this obj e ction :

30

OTTO CHR . DAHL

Dahl cites Ami s IF s a - as one j usti fi cation for the initial


*S , but it turns out that Ami s sa- is not a regular IF mar
ker in Amis . I nstead , Amis s a - derives ins trumental nom
inalisations which only rarely occur in a construction which
could be analys ed as having an Instrumental subj ect .
It i s somewhat puzzling to find this as an obj e ction in a paper which claims
that the original function of this and other focus affixes was nominalisation
( Staros ta et al . 1982a : 16 5 ; 1982b : 131) . The authors rely on Teresa Chen 1982 :
117 :
Although the verbs in these examples
have instrumental
subj ects , it would be somewhat mi sleading to label them
" Instrument Focus" in the Phi lippine sense because they do
not form a paradigm, and are not marked by any consistent
IF affi x .
.

Ferre l l , however , gives a paradigm with s a - a s morpheme of I F , fol lowing


Ogawa and Asai 1 9 35 : 40 3 , although he also gives examples of the same kind as Chen
( Ferre l l 19 72 : 122-123 ) . It is possible that her investigations and those of Ogawa
and Asai have been made in different Amis dialects . But at any rate Amis s a - does
not seem to indicate a moving obj ect , and is therefore not exactly parallel to
Mlg a - .
In Saaroa , however , focus forms with the prefix saa- (which Tsuchida calls
" special focus" ) may have as subj ect an instrument/means or an obj ect , e . g .
speaking i n a fairytale of a bamboo on which a girl was climbing , i t i s sai d :
am i
muu - ca p i na a f a i n a i s a
( 1 2 ) S a a - l ev e - a
IF
go by means of is-said AF drop to woman her
She came down by means of (it) to her mother3 it is said.
(The function of the final - a , present in some forms but not in all , is not
clear . )
( 1 3 ) S a a - f ama r - a cu
ami
ka t a p u f aceQe
I F burn
already is-said the monkey
(Tsuchida 1976 : 7 5-77 )
(It) was burnt by the monkeY3 it is said.
In ( 1 2 ) the means for cl imbing is in focus , in ( 1 3 ) the obj ect burnt . This
resembles the double use of a- in Malagasy , but the ob j ect is not c learly moving .

5 . 3 When Malagasy verbs in AF have two ob j ects , the direct obj ect is often
moving , and the receiver of it is the indirect ob ject . Both these may be focused ,
the direct one with the a- form , the indirect one with the - a n a form , e . g . Mer
( 14 ) ma no l ot ra fanomezana a n - d ra i - ny IZY
acc . father his he
hand over gift
He presents his father with a gift .
The same may be said in the following ways :
( 14a) t o l o r - a - n y f a nomezana ( ny ) r a i - n y
His father is given a gift by him.

Or

( 14b ) a - t o l o - n y a n - d r a i - n y ny f a nomezana
The gift is given to his father by him .

FOCUS IN MALAGASY AND PAN

31

Man6 1 0 t ra is AF , to 1 6 r - a - ny is RF to 1 6 r - an a + third person suffix , a - t 6 1 0- n y


is I F a - t6 1 0 t r a + third person suffix , from the wordbase t6 1 0 t rn .

5 . 4 Because the IF forms begin with the vowel a , past tense has only n and future
tense only h - , like these tenses in forms wi th - i n a and - an a when the wordbase
has initial vowel .

6.

I MP E RAT I VE

6 . 1 The only mood di fferent from the forms treated above is the imperative . The
AF imperative i s formed with the suffix - a to AF present tense , e . g . m i - s 6 t ro to
drink , m i - so t r6 - a drink ! Because a syllable is added , the accent is moved to the
new penult . If the wordbase has final - ka or - t r the suffixation of the imper
ative morpheme follows the same diachronic rules as that of - i n a and - ana ( c f .
1 . 2 . 1- 1 . 2 . 2 ) , e . g . m i - pe t raka t o sit , m i - pe t rah - a sit down ! , man6ra t ra t o write ,
mano r a t - a write ! When a wordbase wi th antepenultimate accent has final - n a , the
s ame rule leads to the following consequence : the imperative morpheme replaces
the secondary final a , and this displaces the stress to the penult , e . g . mana ton a
m a n a t 6 n - a draw near ! The only
t o approach , ma na ton + the imperative suffix - a
audible and visib le di fference between AF present tense and imperative is then
the place of the accent .
=

If a wordbase accentuated on the penult has a in final position , the - a of


the suffix contracts wi th the preceding a , e . g . m i - a l a to go away , *m i - a l a - a >
m i a l a get out ! Here too the shift of the accent , now to the ultima , is the only
di fference between the two forms .

6.2

All the non-active focuses have the same imperative suffixe s . In Merina
the regular suffix is - 0 , but when there is an 0 in the wordbase , -y [ i ] is used
inste ad of -0 by a rule of euphony . Before the suffixation of -0 or -y the suf
fixes - i n a and - a n a are de lete d . OF and RF have thus the same form in imperative ,
e . g . t a p a h - i na to be cut , t a p a h - o ny t ad y cut the rope ! , s o r a t - ana to be written ,
s o r a t - y ny t a ra t a s y wri te the letter ! , a - t6 1 0 t ra to be handed over , a - t o 1 6 r - y a z y
ny v 6 1 a give him the money ! The " thing" i n focus is sub j ect of the verb in the
imperative too , and must be in a definite form , here as in non-imperative clauses .
The sub j e ct is not always pronounced , but neverthe less implicit , because i t is
"old information" . I f the obj ect of the action is inde fini te , active imperative
must be used , e . g . mamb6 1 y va ry to p lant rice , mambo l e va ry p lant rice ! We see
that if the wordbase had an original final e , Merina has cras is between this and
the - a , as h as been observed with - a na , while Sakalava has mambo l e - a ( c f . 4 . 2
above) .
In Sakalava the suf fix of the non-active focuses is always - 0 , e . g . s o r a t - o
z a o write this ! When the wordbase has final 0 , the two a ' s are contracted , e . g .
von6-e t o b e ki l led , von6 ki l l ! from von6- 0 .

6.3

The suffix - i i s the morpheme of

imperative i n Atayal too (Egerod 19 65 :


It is used in RF or locative
forms in several Phil ippine languages . Mlg -0 may be cognate to Atayal - a u ,
wh ich is OF subj unctive morpheme in this language (Egerod 1965 : 26 9 ) , and has
simi lar use in other Formosan languages , e . g . Paiwan .
RF

269 ) , and also imperative in Sedek ( Asai 195 3 : 56 ) .

32

7.

OTTO eHR . DAHL

VERB OR NOUN ?

7.1

Si nce Dempwolff , i t has been discussed whether the non-active focus forms
are verbal or nominal . Dempwolff always spoke of das nominale Denken der
Aus tronesi e r ' the nominal thinking of the Austrone sians ' . This is endorsed by
Erin Asai ( 19 36 : 3 7 ; 1 9 5 3 : 6 2 - 6 3 ) and Cecilio Lopez ( 19 4 1 ) who were among his
students . The reason given by Dempwolff was that these forms construct the actor
in the same way as the owner to hi s posse ssion , and should accordingly be con
sidered as nouns .
with my b ackground in Malagasy I could not accept this . I had the feeling
that the forms of non-actor focus were as verbal as AF , expressing actions and
states to the same degree . But fee lings are not scientific arguments , and for a
foreigner semantics is too often influenced by translation to his mother-tongue .
In Indo-European languages we have verbal nouns expressing actions , but
syntactically their function is nominal , even when they , as infini tive , have no
nominal inflection . The decis ive criteria must there fore be found in syntax .
I s it possible to find such criteria in Malagasy syntax?

7 . 2 . 1 In a simple clause where no member is emphasised more than others , the


word order in Malagasy is predicate ( new information) - subj ect ( old information) .
Both predicate and sub j ect may be eith er noun or verb , e . g .
eto
mp-on i n a
the inhahitant here
faY'l1ler
The inhabitants ( P ) here are farmers ( S )

( 1 5 ) mpam- bo l y n y

( 16 ) mam- bo l y

v a r y n y mp-on i n a
cu ltivate rice the inhabitant
The inhabitants ( S ) cultivate rice ( P ) .

( 17 ) mpam- bo l y d a ho l o ny

m-on i n a e t o
faY'l1ler
all
the A F live here
They who live ( S ) here are a l l faY'l1lers ( P ) .

( 18 ) mam- bo l y

v a r y ny m-on i n a e t o
A F cultivate rice the AF live here
They who live ( S ) here cu ltivate ( p ) rice .

In clauses of this type the sub j ect , whether noun or verb , must be in a
definite form in Merina , whether preceded by the article ny , or preceded and
fol lowed by a demonstrative pronoun .
( Proper nouns and pronouns are definite
by nature and do not require these determiners . )
In the examples above the AF
verb mon i na is nominalised by the article , and functions like a participle , but
without any formal change of the finite verb .
( 19 ) mamonj y a i n a ny

ma n - dos i t ra
AF save life the AF flee run away

This may
they who flee
infinitive or
noun or verb ,

be understood in two ways , whether : to flee ( S ) saves ( p ) life , or :


save their lives . The nominalised verb may thus function l ike an
like a participle in Indo-European . In these clauses the predicate ,
cannot be in a defini te form with n y .

In the examples above I have used AF because this form bears no morpho
logical resemblance to nouns . I t takes obj ects and complements in the same way
as active verbs generally do . I t should there fore be pos s ible to agree on their

FOCUS IN MALAGASY AND PAN

33

verbal nature . A s sub j ect with ny w e must consider them a s nominal ised verbs ,
deverbatives , but the ir form is exactly the same as the finite verb .

7 . 2 . 2 In the des criptions above of the three non-active focuses there are several
examples of these forms used as predicate ( no . l , 2 , 3 , 4b , 5 , 6 , 7b , 8b , 9 , 11 ,
14a, 14b) . But they may also be used as subj ect , e . g .
( 20 ) t s y .3z0ko n y l a z a - i n - ao
not understood I the say OF you
I do not understand what you say. ( l i t . the said ( 8 ) by you (is ) not
understood ( p ) by me . )
( 2 1) mba
va k ( - o
n y no- s o ra t - a n - ao
please read imper . the past write RF you
Please read ( P ) what you wrote ( s ) .
( 2 2 ) r a t s y t ok6a ny a - f ( n d ra ho am i n ' (0
l a s y (0
bad
very the IF move to
that camp that
To be moved ( S ) to that camp is very ad ( p ) .

7 . 2 . 3 When the predicate is strongly emphasised , to the virtual exclusion of


all e l s e , the clause has a special structure . The subj ect ( old information) is
in Merina preceded by no ( in other dialects by ro) , particles resembling the case
marke rs in Formosan and Phi lippine languages . It is very often natural to trans
late it in our l anguages as it is X that/who is/does Y.
In Malagasy X is predi
cate and Y is subj ect . After no the article ny cannot be used ( and only rarely
after ro) . Both predicate and subj ect may be noun or verb , and a noun as predi
cate may be in an inde fini te or defini te form , e . g .
( 2 3 ) mpamp i a n a t r a n o t 6mpo- n '
n y t rano
teacher
owner gen . the house
It is a teacher ( p ) who is (the) owner ( 8 ) of the house.
( 24 ) ny mpamp i a n a t r a no t 6mpon ' ny t r ano
It is the teacher ( p ) who is (the ) owner of the house.
(25)

( ny ) vazaha no mamp i a n a t ra t en y f ra n t say


( the) stranger A F teach
word French
It is a/the stranger ( p ) who teaches ( 8 ) French .

But i f a verb is predicate , it is hardly ever possible to use it in an


inde finite form . It must generally be preceded by the article ny . Compare the
fol lowing examples :
( 26 )

( n y ) mpamp i a n a t r a t e n y f ra n t s ay no f a n t a t r - o
(the) teacher
word French
known
I
It is a/the teacher ( p ) of French whom I know ( 8 )

As mpamp i a n a t ra is a noun , it may be in indefinite or de finite form . But


here it is possible to use the verb mamp i a na t ra teach instead of mpamp i a n a t ra
teache r , and then it must be preceded by the article :
( 26a) ny mamp i a n a t r a t e n y f r a n t s a y no fan t a t r - o
the A F teach
word French
known
I
l i t . It is the teaching ( p ) French whom I know ( 8 ) .
Here the verb expresses the actor.

But to say :

34

OTTO CHR . DAHL

( 26b) * *mamp i a na t ra teny f ra n t s a y no f a n t a t ro


It is a teaching French whom I know.
would not be grammatical . To have an indefini te predicate we must introduce a
noun to which mamp i an a t ra is an attribute :
( 26c) e l ona mamp i a n a t ra t eny f ra n t say no f a n t a t r-o
I
known
word French
teach
person
It is somebody ( li t . a person) teaching French whom I know.
I f , however , the verb expresses the action and not an actor ( c f . example ( 19
it may be without article , e . g .

( ny ) mamp i a na t ra teny f ra n t say n o raha r a h a - n y


I t i s t o teach ( p ) French that i s his/her/their occupation ( 8 ) .

( 27 )

With or without an article the meaning is the same .


With verbs in the non- actor focuses the syntax is the same , e . g .
( 28)

azo- ko
ko no e f a
i r (ny nfinished got I
the past desire OF I
It is what I desired ( P ) that I have got ( 8 ) .

( 29 )

no- s o r a t - a na no ho- va k r na
ny e f a
fut . read OF
the perfective past write RF
It is what has been written ( P ) that sha be read ( 8 ) .

( 30 )

ko no l aza- i - ko am i n - a o
a- e l i ny h you
say OF I t o
the fut . IF diffuse I
It is what I sha diffuse ( p ) that I te ( 8 ) you .

In these cases the verb must be preceded by the article , because it is used
l ike a participle expressing , e . g . , the obj ect of the action . But the non-actor
focuses too may express the action i tse l f , and then ny is facultative , e . g .
( 31)

( ny ) a r l
ana no a n t enona azy
suitab e it
(the) throw away RF
It is to be thrown away ( p ) that is suitabe ( 8 ) for it.

We may say that the most nominal use of the verb expresses the action itself
( like an infinitive or a gerund) and is treated syntactically as a noun . Where
it expresses actor or obj ect ( like a participle ) , it retains more of i ts verbal
character and is treated unlike a noun . Actor focus and non-actor focuses are
treated in the same way , as verbs , not as nouns .

7.2.4

Malagasy has yet another construction which clearly shows that in thi s
language , at any rate , Dempwolff ' s argument is not valid a s proof of nominal
character . Non-active imperative forms may have the short- forms of the second
person pronoun suffixed to them , l ike nouns and the ordinary non-active focus
forms . The appeal is then emphasised . And imperative is incontestably a verbal ,
not a nominal form. Compare the following example s :
i ty
ao
( 3 2 ) t r a n o- n house pas s . you ( sg . ) this
this is your house

FOCUS IN MALAGASY AND PAN

35

( 3 3 ) t a fo- a n - a o n y t ra n o
thatch RF you the house
you are thatching the house
( 34 ) heve r - onao ny t oe t r a ny
think imper . poss . you the si tuation his
Do think ( p ) of his/her/their situation

(5) .

( 3 5 ) F i d i - 0na rea
a n (o i zay ho- tompo- i n - a reo
choose imper . pos . you ( pl . ) today who fut . serve OF you ( pl . )
Do choose ( p ) today whom you ( pl . ) wi L L serve ( 5 ) .

7.2.5

Morphologically noun and verb have owner and actor constructed in the
same way , but syntactically nominal and verbal forms have different constructions .
The l imit between the two categories is , however , not the same as in Indo-European
languages . Used about the action ( like an infinitive ) the verb has nominal char
acter , but used like a participle it is verbal . The participle is an adj ectival
It is often
form , and in Malagasy the ad j ective belongs to the verbal category .
formed with the verbal prefix ma- and has the same tense inflection as the verb ,
past n a - , future ha- . I t has also an imperative form wi th - a like AP , e . g .
ma - d i o c Lean , n a - d i o was cLean , ha - d i o wiL L be c Lean , ma- d i ov- a be cLean !

7 . 2 . 6 In Malagasy it thus seems clear that the non-active focus forms are verbal .
Only a syntactic examination of other l anguages can show whether this is the case
in these languages too . The cri teria have to be chosen according to the syntactic
rules of each language . A comparison of the re sults may inform us about the
character of these forms in modern language s .
7.3.1

What are we able to say today about the situation in Proto-Austronesian?


5tarosta , Pawley and Reid argue that :
*-an , * n i - /- i n - , * - a n a , * - i S i - , and possibly *mu - / - umwere all noun-deriving affixes in PAN , as they still are
to a large extent in the modern languages outside the
Philippine are a , and that they have in fact retained thi s
function t o a previous ly unrecognised extent even within
the Phi l ippine language group . We argue further that
Austrone sian nominalisations in * - an , * n i - /- i n - , * - a n a ,
* i S i - and possibly *mu - / - um- did not develop from original
passive constructions , as concluded by Dahl ( 19 7 3 ) , Wol f f
( 1979) , and Pawley and Reid ( 19 79 ) , but rather that the
nominalis ing function was the original one , and that the
pass ive and verbal focus uses of these affixes in Phi lippine
languages are a se condary development . That is , verbal
focus in Proto-Austronesian was at mos t an incipient
mechanism that was later elaborated and deve loped by the
language s of Borneo and the Celebes .
( 5 tarosta et al . 1982 a : 148)

We shall see that development from nominal into verbal forms has probably
taken place in Malagasy ( see 8 . 3 below) , and such changes are certainly possib le .

7 . 3 . 2 However , if we consider as PAN only the nominal forms with the


the focus system , the focus system i tself must have developed later .
opinion this i s not pos sible . The focus system is found both in some
and in some western languages from the Phil ippines to Madagascar with

affixes of
In my
Formosan
similar

36

OTTO eHR . DAHL

forms and in similar constructions . But the languages where i t is found be long
to different primary subgroups of Austrone sian . The Formosan language s have so
many archaic features which they do not have in common with languages outside
Formos a , that they mus t repre sent the first offshoots from the PAN centre ( Dahl
19 7 3 : 1 2 4 - 12 5 ; 1976 : 12 5 ; 1981 : 15 3 ; Blust 1980 : 1 3 ) . Till now I have not been able
to find innovations common to all Formosan language s . For the time being they
must there fore be considered as belonging to several first-order subgroups of
Austrones ian.
But all the languages outside Formosa have innovations in common , e . g . PAN
*S l ' S 2 ' H 1 , H 2 into PMP *h , and after this PAN * t ' generally into 5 ( Dahl 1981 :
45-62 ) .
In the Formosan languages which have had PAN * t ' > 5 , thi s 5 has merged
with 5 < PAN * S l ( 1981 : 84) . If 5 from both these PAN phonemes had existed sim
ultaneously in the proto-language of the languages outside Formosa , we should
have expected the same merger in at least some of these language s , but this is
found nowhere outside Formosa . The change of PAN * t ' into 5 outside Formosa
must therefore have taken place after the change PAN *S l > h ( Dahl 1981 : 87 ) .
The nasalisation of the first consonant of the wordbase , which has lead to
nasal accretion in Oceanic and nasal substitution in Western Austrones ian , is
also confined to the non-Formosan languages . We must therefore consider all
languages outside Formosa as one primary subgroup , which Blust has called Malayo
Polynesian ( see e . g . B lust 19 80 : 13 ) .

7.3.3

The four-focus gramma tical system exists both in Formosan and MP languages ,
that is to say in more primary subgroups . I f this system did not belong to PAN
but developed later from nominal PAN forms , parallel development must have taken
place in several subgroups after their s eparation . However , not only are the
morpheme s identi cal , but also the syntactical use of the forms . The choice of
focus form permits placement as subj ect words with different relations to the
action or state expressed in the clause . A parallel development of this sophisti
cated system in di fferent subgroups of AN from nominal forms with the same affixes
does not seem possible to me . There are too many simi larities . For instance ,
in Atayal and Malagasy the fundamental features of the four-focus system are
virtually the same in spite of the long separate development of the languages .
The only s igni ficant difference is that the more differentiated modal categories
in Atayal are reduced to two in Malagasy .
To me the possibility of parallel development from verbal into nominal forms
seems much greate r . The uses of the non-verbal forms are not s o similar i n the
di fferent language s . In Malagasy these forms are generally not nouns but adj ec
tives , and thus nearer the quality of the verb . The Malagasy focus forms are not
only used as sub j ect and predicate , but also as quali fiers of nouns , l ike adj ec
tive s , e . g . zava t ra ome na a thing given , in constructions like zava t ra maVe 5 a t r a
a heavy thing. From such constructions to the use of the same affixes to create
adj ectives the way is short , e . g . ra no l ome r - i n a water overgrown with moss from
l emo t ra moss , t a n y va t e - a n a stony earth from v a to stone .
( Neither l omot ra nor
v a t o have verbal forms . )
I f the noun is omitted in such constructions , the
adj ective is nominalised . The next step i n the evolution may then b e a real noun .
We have seen ( 7 . 2 . 5 above) that in an AN language the difference between
noun and verb is not the same and not so sharp as in IE languages . For that
reason , the possibi lity that the same affixes as those forming focus may have
been used to produce nominal forms in PAN i tself cannot be excluded ( cf . Saaroa
5 a a - in 5 . 2 . 3 above) .

FOCUS IN MALAGASY AND PAN

37

7 . 3 . 4 I f the focus system be longed to PAN , some Formosan and many MP languages
have lost i t , among others Malagasy ' s nearest relative s , the Barito languages .
Here about 1600 years of separation has been sufficient to produce the difference .
PAN was probably spoken around 5000 B . C . (Blust 1 9 80 : 13 ) , and a loss of funda
mental gramma tical features during these 7000 years is not extraordinary . Modern
European languages have lost the fundamental case structure of Proto- Indo
European in a much shorter time , replacing it with a set of prepositions formed
from old material . In languages which have lost the focus system , there may be
new forms replacing categories in the focus system , here also using old material .
It may be worth examining modern gramma tical sys tems with thi s in mind .

8.

C I RCUMSTANT I AL FOCUS

8. 1

Malagasy has also a fifth focus whi ch is formed by a circumfix where the
suffix is always Mer - a n a , Sak - a , and the prefix is any AF pre fix deprived of
i ts ini tial m- . The form has thus initial vowel and therefore n- and h- in past
and future tenses like I F . In most dialects it forms i ts imperative according
to the s ame rules as the other non-active focuses . But in Tesaka the imperative
of this focus always has the suffix -y ( Deschamps 19 3 8 : 20 ) . We have already seen
that - i forms the imperative of RF in Atayal ( 6 . 3 above ) , and in other languages
it is a locative suffix . Because CF has the suffix - a n a , it has some resemb lance
to RF . This may be the reason why - y i s the imperative suffix here , and this may
be the origin of the alternative -y in Merina too .
This focus was called relative voice by the old gramma rians ( Cousins 1894 :
Howeve r , since this form
has no simi larity to the accustomed use of relative in grammars ( relative pronoun ,
relative clause) , I prefer another term : circumstantial focus ( CF ) .
48) because i t has in focus any re lation to the action .

8 . 2 Any circumstance having a relat ion to the action or state expressed by the
verb may be focused by thi s form : place , time , cause , intention , reason , means ,
ins trument , bene ficient , e . g .
( 32)

( pe t r a ka t o sit)
i - p e t r a h - a - ny n y s t h a
he the ohair
sit CF
The ohair ( S ) is where he is sitting ( P ) .

( 33 )

( a ngona to assemble go to ohuroh )


i - a n g6n - a n a n y a l a ha d y
assemb le CF the Sunday
Sunday ( S ) is when going to ohuroh ( p ) .

When CF is use d , there is often so great an emphasis on the circumstance


that the construction with no is preferred :
( 34 )

a n - dos ( r- a - ny
( 1 6s i t r a flight )
t a ho t r a n o n past flee CF
he
fear
It was for fear ( P ) that he fled ( S ) .
hamot s ( - a n a ny t rano no i l a - ko s okay
want I lime
conj . fut . white CF the house
It is in order to whitewash ( p ) the house that I want ( S ) lime .

( 3 5 ) mba

( The wordbase o f h amot s i a na i s f6 t s y white , AF mam6t s y , and of i l a ko ( l a , AF


m- ( l a to want . Because the AF prefix is only m- , which must be deleted , CF of
this verb has no prefix . )

38

OTTO CHR . DAHL

( 36 ) s o ka y no h - amo t s f - a - ko n y t ra n o

It is with lime

( p)

that I sha l l whitewash ( S ) the house .

ants i - nao no anapah- 0


ny t a d y
CF imp . the rope
the knife gen . you
out
Cut ( S ) the rope with your knife ( p )

( 37 ) n y

( 38) ny

ray
ama n - d rEfny no h ano 1 6 r - a n - ao n y v6 1 a
fut . present CF you the money
the father and
mother
It is to the parents ( p ) that you sha l l give ( s ) the money .

The CF may also express an action on only a part of the sub j ect .
the following clause s :

Compare

( 39 ) vono- y

n y a k6ho- ko
ki l l OF imper . the ohioken my
ki l l my ohio kens (al l of them)

( 40 ) amono-y ny a k6ho- ko

ki l l some of my ohiokens

( 41 ) amono - y r6a ny ak6ho- ko

ki l l two of my ohiokens

Ny a kohoko is the sub j ect of all the three clauses , r6a in the last one is
ob ject of amonoy . The command in the last one may also be expres sed in AF , but
in a less e legant manner :
( 41 a) mamon6- a

r6a a m i n ' n y a k6ho- ko


AF ki l l imp . two among the ohioken my

The fact that the part in AF is expressed wi th a complement ( am i n ' ny a - ) ,


and accordingly as a circumstance in relation to the verb , explains the use of
CF w i th the same meaning .
Preceded by the article the CF may also mean the action itself nominalised ,
e . g . ny i - pet r a h - a n a the sitting, the aotion to sit .
The CF form makes the language very flexible in that it allows any part of
a statement to be emphas ised . This is necessary because more than ordinary stres s
cannot b e used to emphasise single words . Focus thus offers a suitable means o f
achieving emphasis .

8. 3

I have found no clear parallel


seems to be a Malagasy innovation .
created , I have found some forms in
nouns formed with the circumfix pa
or with p i - + - a n . The meanings of
itse l f , the place , time , instrument

to CF in any other AN language . It there fore


Looki ng for material from whi ch it has been
Ma ' anyan . This language has some abstract
with nasal substitution or accretion + - a n
these forms are the following : the action
( Sundermann 19 1 3 : 2 19 - 2 2 1 ) .

Ma ' anyan p has become Mlg f , and in Malagasy we have the same forms wi th f - :
f a - with nasal substitution or accretion + Mer - a n a , Sak - a , and f i - + - a n a / a .
The meanings of the Mlg forms are the same as in Ma ' anyan plus most of the
meanings of the CF . But the Mlg forms with f - and the Mny forms with p- add to
these meanings the notion o f habitual . Compare the fol lowing :
fanaov - a n a a z y
( from tao t o do, make)
the doing
it
The habi tual, general way of doing it (always) .

( 42 ) ny

39

FOCUS IN MALAGASY AND PAN

( 42a) n y a naov- a n a a z y
the way o f doing it (in the actual si tuation) .
There are also forms wi thout - a na which have the same habitual meaning , e . g .
f - omba custom habit from omba to accompany .
It
meaning
What is
For the

is therefore pos sible that the old forms with p / f + - a n /a n a and habitual
have eliminated the f - , and thereby removed the connotation of habitude .
left is the form of CF , and to this still other meanings have been added .
t ime being I consider this to be the most l ike ly hypothesis .

It is worth noting that the nominal form with f - has an obj ect , j ust l ike
the verbal CF form without f- . This shows that in Malagasy the diffe rence between
noun and verb is not so sharp as in Indo-European . It would be interesting to
know if this is true of other AN languages .

9.

CONCLUS I ONS

Malagasy has the four-focus gramma tical system which is also found in
Formosan , Phil ippine and Minahasan languages , and wi th affixes that are present
in other focus languages too . I t is found here that the PAN forms of these
affixe s are AF *- um- , OF * - an and - i n - , RF * - a n and IF * S , i - ( perhaps also * S a - ) .
Malagasy shows regular reflexes of all these . The imperative suffixes present
in Malagasy are AF -a < PAN * - a , in the non-active focuses - i < PAN * - i and - 0 <
PAN * - a u . Instead of PAN * - um- Malagasy mostly uses reflexes of *ma - + nasal
accretion or substitution , or of *ma y - .
Since Dempwolff the question of whether the non-active focuses are verbal
or nominal h as been a moot point . Malagasy syntax shows that in this language
they are verbs - in spi te of the construction of the actor be ing in the same form
as the owner of the noun expre ssing his pos session . The non-active imperatives may
suffix the second person short form , l ike nouns . But imperatives are inconte st
ably verbal forms . Thi s shows that such construction is no proof of nominal
character . To settle this question for the focus languages in general a syntac
tical examination of the function of focus in these languages is neede d .
Today t h e focus system is found i n several first-order subgroups o f Austro
nes ian ( see 7 . 3 . 2 above ) . The similarities are so great that a parallel devel
opment of the system in these languages must be excluded. Focus must therefore
have belonged to the PAN grammatical system . Malagasy has , however , developed a
fi fth focus whi ch must be a local innovation .
The language s without the focus system today must have lost i t . Instead
there are sometimes found new forms replacing categories in this system , formed
at least partly wi th old material . A further study of this in Austronesian
languages is needed.

NOT ES
1.

The following abbreviations are used : AN


Austronesian , I sb
I sbukun ,
Mer
Merina , Mlg
Malagasy , Mny
Ma ' anyan , MP
Malayo-Polynesi an , P
predicate , PAN
Proto-Austronesian , PMlg
Proto-Malagasy , PMP
Proto
Malayo-Polynesian , S
subj ect , Sak
Sakalava , Tkb
Takbanuad , Ttd
Takitu?duh .
=

40

OTTO CHR . DAHL

2.

In Malagasy orthography 0 is the symbol for the vowel [ u l , and y is written


for f inal [ i l .
I mostly take my examples from Merina , which is the base of
the official l iterary Malagasy , but also f rom Sakalava , and only occasion
ally f rom other dialects , when they give us information about the historical
deve lopment of the language .

3.

There are two pos s ible proto-forms for this wordbase : PAN * l a y i u and * l a i a t ' ,
b oth meaning to run , and both with irregular development of the last vowel .
The two h ave probably merged , because in imperative we have both l - om- a i - a
and l - om- a i s - a , cf . 6 . 1 .

4.

That there has been a prefixed form wi th mu- i s corroborated by the causa
tive form of these verbs . The ordinary causative prefixes in Malagasy are
mamp a - + nasal accretion or sub stitution , or mamp i - , corresponding with the
But in
AF prefixes ma- + nasal accretion or substitution or wi th m i - .
Sakalava the causative prefix of the verbs with -om- is mampo- , e . g . mampo
l ay to oause to run . See also mu- in Formosan languages ( Dahl 197 3/76 : 1 19) .

5.

For the deve lopment o f f inal nasals , contraction o f vowe ls and accent in
Malagasy , see Dahl 1 9 5 1 : 62-65 and 84-9 1 .

6.

ome

7.

Ferrell h as mi sunderstood two forms in Egerod ' s paradigm .


In private corres
pondence Egerod has informed me that perfective OF should be . q - n - a l u p and
RF q - na l u p - a n .
In perfective IF the form Egerod has written. ( i ) n s q a l u p i s
so r ar e in h i s material that he does not consider i t a s certain .
Ogawa and
Asai do not h ave this form in their paradigm ( 1 9 3 5 : 3 0 ) .

< PAN * b a y a i wi th fos s i l i s ed - um- : * b - u m- a y a i > *wumee > ome .


loss of w before u is irregular .

Only the

RE FERENCES
ASAI , Erin
19 36

A s tudy of the Yami language , an Indonesian l anguage spoken on Botel


Tobago Isl an d . Leiden : J . Ginsberg .

1953

The Sedik l anguage o f Formosa .


Kanazawa .

Kanazawa : Cercle Linguistique de

BLUST , Robert A .
1 980

Austronesian e tymologie s .

Oceanic Lingui s ti cs 19 : 1-189 .

CHEN , Teresa M .
1985

Verbal cons tructions and verbal classification in Nataoran-Amis .


( Ph . D . dissertation , University o f H awai i , 1982 . )
PL , C-85 .

COUS INS , W . E .
1 89 4

A concise in troduction to the s tudy of the Malagasy l anguage a s


spoken in Imerina .
Antananarivo : The Press o f the L . M . S .

DAH L , Otto Chr .


19 5 1

Malgache e t maanjan .
Instituttet .

Une comparai son l i nguis ti que .

Oslo : Egede

FOCUS IN MALAGASY AND PAN

41

1954

L e substrat bantou en ma1gache .


1 7 : 3 2 5 - 36 2 .

19 71

La forme locative du nom ma1gache .


Sprogvidenskap 24 : 1 5 4- 16 2 .

1973

Scandinavian Institute of Asian Studies ,


Proto-Austronesian .
2nd revised edn ,
Monograph Series No . 15 .
Lund : Student1 itteratur .
1976 .

1977

La subdivis ion de 1a fami11e Barito et 1a place du ma1gache .


Orien tal ia 38 : 7 7 - 1 3 4 .

1978

The fourth focus .


In S . A . Wurm and Lois Carrington , eds Second
International Conference on Aus tronesian Lingui s t i cs : proceedings ,
3 8 3 - 39 3 . PL , C-61 .

1981

Earl y phonetic and phonemi c changes in Aus tronesian . Oslo : The


Institute for Comparative Research in Human Culture B 6 3 .
Universitets for1age t .

1983

Norsk Tidsskrift for Sprogvi denskap


Norsk Ti dsskri ft for

Sorabe revel ant l ' evol ution du di al ecte antemoro .


T.P.F .L.M.

Acta

Antananarivo :

DE SCHAMPS , Hubert
1936

Le di alecte Antaisaka .

Tananarive : Pitot de 1a Beauj ardi e re .

EGEROD , Sren
1965

Verb inflection in Ataya1 .

Lingua 15 : 2 5 1- 2 82 .

FERRELL , Raleigh
1972

Verb systems in Formosan languages .


In Jaque 1ine M . C . Thomas and
Lucien Bernot , eds Langues et techniques , na ture et soci e t e , vo1 . 1 ,
1 2 1- 1 2 8 .
Paris : K1inck s ieck .

LOPEZ , Ceci1io
1941

A manual of the Phi l ippine national l anguage .


Bureau of Printing .

3rd edn .

Mani la :

OGAWA , Naoyoshi and Erin ASAI


1935

The myths and tradi tions o f t h e Formosan native tri bes .


Taihoku Imperial Univers ity .

Taihoku :

SNEDDON , J . N .
1978

Proto-Minahasan : phonology , morphology and wordl i s t .

PL , B- 5 4 .

STAROSTA , Stan ley , Andrew K . PAWLEY and Lawrence A . REI D


1982a

The evolution of focus in Austrones ian .


In Amran Ha1im , Lois
Carrington and S . A . Wurm , eds Papers from the Thi rd International
Conference on Austronesian Lingui s t i cs , vo1 . 2 : Tracking the
travel l ers , 1 4 5- 17 0 . PL , C- 7 5 .

1 9 8 2b

The evolution o f focus i n Austronesian .


of 1 9 8 2 a .

MS .

An expanded version

SUNDERMANN , H ermann
19 1 3

Der Dia1ekt der " Olon Maanj an" ( Daj ak ) i n Su d-ost-Borneo .


tot de taal- , l and- en volkenkunde van Nederl andsch-Indi e

Bi jdragen
66 : 2 0 3 - 2 36 .

42

OTTO CHR . DAHL

TSUCH I DA , Shigeru
19 76

Recons tructi on of proto-Tsouic phonol ogy . Study of Languages and


Cultures of Asia and Africa Monograph Series No . 5 .
Tokyo .

TUN G , T ' ung-ho


1 9 64

A descripti ve s tudy of the Tsou l anguage , Formosa .


Institute of
History and Philology special publication No . 4 8 .
Taipe i : Academia
Sinica .

ZORC , David Paul


197 7

The Bisayan dialects o f the Phil ippines : s ubgrouping and recons truc
t i on . PL , C-44 .

Dahl, O.C. "Focus in Malagasy and Proto-Austronesian". In Geraghty, P., Carrington, L. and Wurm, S.A. editors, FOCAL I: Papers from the Fourth International Conference on Austronesian Linguistics.
C-93:21-42. Pacific Linguistics, The Australian National University, 1986. DOI:10.15144/PL-C93.21
1986 Pacific Linguistics and/or the author(s). Online edition licensed 2015 CC BY-SA 4.0, with permission of PL. A sealang.net/CRCL initiative.

You might also like