ON THE S Y N TA X O F yang IN I NDON E S I A N

John W . M . Verhaa r , S.J.

I NTRODUCT I O N
The particle y a n g i n Indonesian ( and older Malay) has , i n the pas t , been
duly treated by grammarians , but it does not look as if yang has been di scussed
much within a framework that has notable relevance across languages , or even
fits more naturally into a more hol istic view of Indones ian ( and Malay) syntax .
What has been noted about y a n g has been mainly two characteristics : first , that
it may l ink an attribute to a noun , notably ' adj ectival ' (as d i st inct from
' nominal ' ) attributes ; second , that one of these attributes is the relative
clause . In dealing with the latter construction , grammarians have often dubbed
y a n g a ' relative pronoun ' . I will not now review earl ier treatments of y a n g
( Kaswanti 1981 has a brief survey ) , for those treatments have , on the whole , not
focussed on the larger syntactic i ssues I am address ing myself to in the present
paper . But most of those ear l ier descriptions have been sound enough ,
especially in that they affirmed the bas ic ' sameness ' of y a n g , no matter whether
it relativises or not , even though that ' sameness ' has been largely implicit in
those descriptions . On the other hand , the description of yang as a ' re lative
pronoun ' has not , on the whole , been one of principle in regard to the
' pronominal ' character of y a n g , and thi s name has been given to y a n g perhaps
rather thoughtlessly . One recent example of such thoughtlessne s s appears in my
own paper on alienable and inalienable possession in Indonesian (Verhaar 1978 ) .
In any event , in the present paper I wil l argue , among other things , that y a n g
i s not pronominal .
More particularly , I want to go into that ' sameness ' I mentioned : y a n g
l inking any ( non-nomina l ) attribute to a head noun (as wel l as a few nominal
one s ) . Also , I shall discuss the use of y a n g without a head , as wel l as the
' de f initis ing ' character of y a n g in that use , and in some other syntactic
constructions . Then I wi l l discuss y a n g as only one particular kind of
' l igature ' , d i f ferent from those which are genuine ( relative ) pronouns . Next
I wi l l consider what conditions must be ful f il led for y a n g to relativise a noun ;
those conditions wi l l be shown to be of three kinds , a l l of them involving some
particular k ind of ' co-referentiality ' with the head . I will then briefly
develop the nature of those co-referentiality conditions as rather c lose to

Arnran Halim, Loi s Carrington and S . A . Wurrn , eds Papers from the
Third Interna tional Conference on Austronesian Linguistics , vol . 4 :
Thema tic varia tion , 4 3 -7 0 . Pacific Lingui stics , C- 7 7 , 1983 .
© John W . M . Verhaar , S . J . 43
Verhaar, J.W.M. "On the syntax of yang in Indonesian". In Halim, A., Carrington, L. and Wurm, S.A. editors, Papers from the Third International Conference on Austronesian Linguistics, Vol. 4: Thematic variation.
C-77:43-70. Pacific Linguistics, The Australian National University, 1983. DOI:10.15144/PL-C77.43
©1983 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.
44 JOHN W . M . VERHAAR , S . J .

those obtaining for participial attribute s , i n languages that have them .

Finally , I will raise some questions about non-restrictiveness of certain types
of yang c lause s .
Most of these discuss ions concern relative clauses . But first I propose
to go into the basic ' sameness ' of relativis ing and non-relativis ing yang .
This I will do in the following section .

H I E RARCH I CA L ' SAM EN ESS ' OF yang AS A ' L I GATURE '

What all uses of yang have in common is that this particle ' welds ' a ' head '
and an attribute together , except , of course , for ' headless ' yang , which I wi l l
discuss in the next section . This ' welding ' function of tiny particles of
various forms , in l inking attributes to nouns , has been known for a long time
to l inguists as widely occurring in a number of Austrones ian languages . The
first study ( to my knowledge ) which has placed the description of such noun
phrases in a perspective more widely relevant across languages (with the
ambition even to reveal language universals of attributive structure ) , is a
portion (Chapter 2 ) of Foley ' s dissertation on Austronesian syntax ( Foley 1976 ;
more succinctly restated in Foley 1980) . The principal thesis Foley advances
i s that of what he calls the ' bondedne s s hierarchy ' control ling the use of
those part icles , which he calls ' ligatures ' . S ince that thesis checks out
almost perfectly for Indonesian (which is not in Foley ' s sample ) , I wi l l review
Foley ' s hypothes i s here briefly .

As is we ll known in Phil ippine studie s , many languages in that nation use

a variety of ' l igatures ' in noun phrases . Here follow some examples from
Tagalog , in ( 1 ) through ( 4 ) (Foley 1980 )
(1) ma t a b a - n g ma runo-ng tao
fat LIG wise LIG man
fat3 wise man
(2) ma ram i - ng bata
many L I G child
many chi ldren
(3) ku l ay n i - i yon
colour LIG that
that co lour
(4) a- ng babae - n g nag-babasa n a - ng d i ya ryo
TP LIG woman LIG AF IMP read P LIG newspaper
the woman reading a newspaper
( LIG =l igature ; TP topic marker ; AF
= actor focus ; IMP
= imperfect ; =

P =patient ) . The l igature here connects a noun with one or more adj ectives
« 1 ) ) , or with a quantifier « 2 ) ) , or with a de ictic « 3 ) ) , or with a relative
clause « 4 ) ) . The form varies : - n g , - n i , and there are others . Foley
d i stinguishes nominal attributes (which he says are not relevant to his
hypothes i s ; however , I will show that to a certain extent they are ) , from
' non-nominal ' ones , adj ectival , therefore , which he calls ' ad j uncts ' ; and he
dist inguishes seven kinds of such ' adj uncts ' , which may be seen l i sted in
Figure 1 ( numbering of the levels is mine ) , reflecting also a hierarchy of
' bondednes s ' .
 ON THE SYNTAX OF YANG IN INDONESIAN 45

7 . noun + article
6. noun + deictic
weaker 5. noun + interrogative
4. noun + quantifier
bondedness 3. noun + adj ective
2. noun + participle
l . noun + relative clause

Fi g u re 1 : The ' bondedne s s h i era rchy '

The hierarchy is from high ( level 7 ) to low ( level 1 ) , and the lower one gets
down the hierarchy , the more some syntactic formative device becomes necessary
to hold noun and attribute together . Going down the hierarchy , at j ust which
level the l igature becomes neces sary depends on each particular language and i s
a language-specific matter . However , what i s valid across language i s that , i f
a ligature i s necessary a t level Z , such that Z i s the highest level , i n that
language , at which the ligature is used , then it will be necessary also at a l l
leve l s below Z. Negative ly , according t o Foley ' s hypothes is , n o language wil l
have a ' gap ' level-wise i n l i gature use below Z ; and above Z , n o level wil l
employ the l i gature . Tagalog i s a good example o f a language that has a
l igature even at level 7 . Therefore , a l l other levels require i t .
Foley ' s sample comprises the following language s : Tagalog , Palauan , 1 locano,
Toba Batak , Tolai , Wol io , and Malagasy . Figure 2 maps the use of l igatures in
those languages , showing that the bondedness hierarchy is val id at least for
them.

Tag . Pal . 1 10 . T.B. Tol . Wol . Mlg .

7 . article X
6. deictic X X
5. interrogative X X X
4. quanti fier X X X X
3. adj ective X X X X X
2. participle X
l . relative c lause X X X X X X

Fi g u re 2 : L i gatures i n the l a ngua ges of Fo l ey ' s s ampl e

The ' gaps ' at level 2 are only apparent : none of these languages , except Palauan
and Wol io , have participles in their verbal system. Wol io alone lacks the
l igature with participle attributes , but there that lack is no gap , since only
level 1 has a l igature there . Finally , Malagasy is relevant in that at level 1
it may have a l igature ( i z ay ) , but it appears to be optional , hence it is not
ticked o f f in Figure 2 . I will not reproduce Foley ' s data here , nor , in
particular , the forms l igatures may take in the d i fferent language s . Those
forms are all o f the kind that cannot be assigned any ' categorial ' status ; that
46 JOHN W . M . VERHAAR , S . J .

i s to say , no spec i fic word class membership can be assigned to them . Also ,
these l igatures lack ' argument ' status : they do not ' fi l l ' any such ' slots ' as
' subject ' , ' obj ect ' , etc . , as arguments to the verb , not even lower level ' slots '
l ike those of attributes . Furthermore , the only sequential rule that holds for
these l igatures is that they must be in between noun and attribute , and the
sequential order that controls those constituents is therefore irrelevant . The
bondedness hierarchy i s , in other words , neutral as between left-hand or right­
hand position of the attribute in respect to the head noun . ( In fac t , in
Foley ' s form of Figure 1 , the attribute is to the left ; I have placed it to the
right , for convenience ' sake , because that is the order in Indones ian , which
language is rigidly VO , also infraclausally . )
Foley does not make much of the ' categorial ' status (or , rather , the lack
of it) , of the l igatures . That issue would be irrelevant for the point he i s
making , which i s supposed t o b e val id across language s , and which therefore
obtains also for those l igatures which do have categorial status : for relative
pronouns , in languages that have them . In English , for example , at a l l leve l s
higher than 1 , bondedness i s enough for the noun phrase to hold together without
a l i gature of any kind . Foley shows that such bondedness , without any l inking
device , may be destroyed by right-extraposition of an attribute , making the
phrase ungrammatical , while such an extraposition is still well formed when
there is such a device . Consider Foley ' s examples , here numbered ( 5 ) and ( 6 )

(5) the book ove r t h e re nea r John wh i c h w a s g i ven t o me

(6) * t he book ove r t he re nea r J o h n g i ven to me
In both examples the attribute (wh i ch wa s ) g i ve n to me is extraposed , but in ( 6 )
extraposition triggers unwel l formedness because there i s no ligature i n this
language , at level 2 , that of the participle , whereas there is at level l , that
of the relative clause .
Foley ' s bondedness hierarchy hypothesi s is we ll ver i fied for Indonesian ,
for the l igature yang . At level 7 (on one particular interpretation of that
leve l ) it is forbidde n ; at leve l s 6 through 3 , it is optional ; final ly , at
level l , it is obl i gatory . ( Level 2 should be ignored , as Indones ian has no
participles . ) I now propose to substantiate this briefly .
According to traditional analysis , Indonesian has no article . The
determiner i t u that is deictic wherever it does not function as a topic marker
(a point I cannot discuss here ) . With i t u , it would seem that yang is optional,
as may be seen in ( 7 )
(7) mej a (yang ) i t u
table LIG that
that table
However ( as Kaswanti 1982 points out) , the deicticity of i t u is of two kinds :
' endophoric ' ( utterance-internally ; i . e . merely anaphorically ) ; and ' ec tophoric '
(referring to something utterance-external : ' deictic ' in a more widely used
sense ) . I t i s , on the strength of the examples taken from Foley ' s sample ,
straightforward to consider level 2 of his hierarchy as only ' ectophorically '
deictic . It matters l ittle whether the ' endophoric ' use of i t u in Indonesian
i s called an ' article ' or not , but , considering several characteristics o f
articles i n languages that have them , that would hardly b e a n esoteric label .
For testing purposes of the bondedness hierarchy for Indonesian let me call the
' endophori c ' use of i t u that of an ' article ' , of level 7 , therefore . Then ,
consider Kaswant i ' s examples ( 8 ) and ( 9 ) .
 ON THE SYNTAX OF YANG IN INDONESIAN 47

(8) buku ( 1'yang ) i t u

book LIG the
the (aforementioned) book
(9) buku ( yang ) i t u
book LIG that
that book (over there)
Phrase ( 8 ) exemplifies the ' article ' u s e of yang , and it is forbidden . In
phrase ( 9 ) , which represents the (ectophor ical ly) deictic use o f yang , it is
optional .
Phrases of levels 5 through 3 are found in ( 10 ) through ( 14 )
( 10 ) a l a t (ya n g ) mana ?
tool LIG which?
which too l ?
( 11) k u r s i (yang ) empat [ . . . ]
chair LIG four
four chairs
( 1 2 ) anak (yang ) banyak [ . . . ]
chi ld LIG many
many chi ldren
( 13 ) r umah (yang ) i ndah
house LIG beautiful
beautiful house
( 14 ) pohon (yang ) be s a r
tree LIG big
big tree
At all these levels , yang is optional , although optionality depends on other
constraints more generally (on which see below) , and on constraints of a quite
particular nature for (de finite and inde finite) quantifier s . A few words must
be said on the quantifier problem firs t , because more general principles are
involved .
I merely recall first a few general principles about quanti fiers which are
generally known . The first is that of ' floating quanti fiers ' . In many languages ,
quantifying constituents , even in NP form, may well formedly take various
positions sequentially ( in a language l ike Japanese they may even become either
' nominal ' or ' adverbial ' depending upon sequential position) within the entire
clause . Also , in languages with a rigid sequential orde r , ' attributive '
quantifiers may ( unl ike a l l other attributes ) freely either precede or follow
the head noun (even when accompanied by numeral classifiers ; the c las s i f ier then
stays with the quantifier ) . Indones ian is such a langauge : one may say k u r s i
empa t ( b u a h ) ( chair four NUM . CL ) four chairs , but in this case without a
ligature . In Javanese , which also has this permutation pos sib i lity , there could
be a ligature in the former case : l i ma - n g ku rs i five chairs ; and k u r s i 1 i ma
( these express ions are not interchangeable , but their difference does not a f fect
the point being made here ) . In a number of languages where this permutation is
poss ible , right-hand pos ition makes the entire phrase definite , l e ft-hand
position inde finite (Greenberg 1978 : 284) . In I ndonesian the construction with
yang is possible , but only with that definiteness , on condition definitising
i t u follows , or some other definitiser ( hence the brackets in ( 11 ) and ( 12 » ,
but then , interestingly , yang is obligatory . I call this interesting for a
spec ial reason : obligatory yang itse l f is o ften definitising ( see below) .
48 JOHN W . M . VERHAAR , S . J .

Kaswanti and Sudaryanto (pers . comm . ) , who suggested to m e some o f the

data of thi s section , inform me that phrase ( 1 2 ) , anak yang banyak may be due
to interference from Javanese ( anak s i ng akeh ; s i n g is one of the l igatures in
Javane s e ) , but that X + yang banyak * ( i t u ) (with i t u , again , obligatory) has X
such that anak seems not to be a member of it . The examples they give have as
members of X ke l ompok group , bag i a n part , portion , t umpukan pi le : all of them
' collective ' noun s , so X might be j ust that . However , the interest of such
data is that Indonesian often ' co l lapses ' a ' group ' with any ' member ' or ' part '
of that group : in se-genap wa rga , for example ( se- is one , genap is who le , and
wa rga is member) what is signif ied is ( from a ' translationese ' standpoint , to
be sur e ) a mixture of the who le membership and each member.
Quanti fiers raise special problems for the l inguist in many languages .
Foley 1976 also has problems with them , in direct relevance to his ' bondedness
hierarchy ' hypothes i s . The trouble with quanti f iers is that they sometimes
seem to behave l ike nouns , sometimes l ike non-nominal attributes , and this i s
important to Foley because h i s hypothe sis i s supposed to b e val id only for the
non-nominal attributes which he calls ' adj uncts ' , and not for the noun + noun
constructions , which , Foley claims , have no l igatures . They do , however , in
Indonesian , and I wil l return to this below.
The compl ications with quantifiers apart , and ignoring special constraints
holding for other levels as well ( see below) , it appears that at level s 5
through 3 yang is optional .
Finally , at level l , that of the relative clause , yang is obligatory ;
consider ( 15 )

( 15 ) O rang * ( yang ) d a t a n g t e r l amba t t i da k bo l e h ma s uk

person LIG come late not may enter
Those who are late may not enter
where yang is indispensable . There are some instances , however , of standing
phrases of high phrasal compactness where relativ i s ing yang may be omitted , as
in ( 16)

( 16 ) Ruma h-nya d i - rampas o l eh o rang tak d i - kena l

house-his PM p lunder by person not PM know
His house was p lundered by person (s ) unknown
( PM =pass ive marker ) , but such phrases are not numerous , though some may be of
comparatively high text frequency .
It appears , then , that Indonesian yang confirms the bondedness hierarchy
hypothe s i s , but also that the Indonesian data may re fine that hypothes i s .
There is first the deixis distinction d iscussed above , the pos s ible application
of which to the languages of Foley ' s sample and to other languages i s l ikely to
produce new rules . However , probably the single most promising analysis would
be that of the ' optionality ' of l igature s . For example , does it really make no
d i f ference in Malagasy whether or not relativis ing clauses are introduced by
i zay? My own hunch is that the answer would be negative ; I have never yet
encountered examples of true ' free variation ' except phonemica l ly .
In any event , the optionality problem for Indones ian yang is important .
To call yang ' optional ' i s not to say that its use or non-use i s merely ' free
variation ' . An exhaustive discussion of thi s problem would probably be lengthy ,
and I wil here only note a few points which are of some interest .
__
 ON THE SYNTAX OF YANG IN INDONESIAN 49

First of all , some phrases already ' welded ' together because of their
semantic content , which cannot be explained from mere attribution , cannot have
yang . Examples are found in ( 17 ) ( Fokker 1951 : 186 )
( 17 ) i l mu pas t i {science certain} mathematics
seko l a h menengah {schoo l middle} High Schoo l
j a l a n bun t u {road b locked-up} dead end s treet
gunung be r-a p i {mountain PREFIx-fire} vo lcano
Also , yang may emphasise the attribute , either for the sake of contrast or
non-contrastively , as in ( 18 ) ( Fokker 1 95 1 : 187)
( 18) mu r i d * ( yang ) bo d oh
pupil LIG uninte l ligent
uninte l ligent pupi l
or in ( 19)
( 19) Say a s uka r umah * ( yang ) besa r , bukan rumah * ( ya n g ) kee i l
I like house LIG big not house LIG sma l l
I like a big house, not a sma l l one
but this presupposes that the phrase concerned cannot be ' close ' enough without
ya n g , which it sometime s can , e . g . in ( 2 0 ) , even though there is an obvious
contrast ( Sudaryanto , pers . comm . )
( 2 0 ) Bukan ape l (ya n g ) me rah ke s u kaan- nya , me l a i n ka n ape l ( y a n g ) h i j a u
not apple LIG red liking his but apple LIG green
What he likes is not red app les, but green app les
where yang could easily be di spensed with . Furthermore , yang is obl igatory
with extraposed attributes , as in ( 2 1 ) and ( 2 2 ) (Fokker 1951 : 188)

( 2 1 ) b i n i - n ya * ( yang ) b i j a k s a n a
wife-his LIG prudent
his prudent wife
(22) i bu - nya \', (yang ) t ua
mother his LIG old
his old mother
( Fokker distinguishes various constructions , whic h , however , a l l fit one species ,
due to extraposition , which he does not mention . ) Finally , when the attribute
itself is composed , either serially ( ( 2 3 » , or because the adj ective has an
adverbial coconstituent ( ( 2 4 » , yang is likewise obligatory ( Fokker 1951 : 189) :

( 2 3 ) anak * ( yang ) raj i n dan pan da i

chi ld LIG industrious and inte l ligent
industrious and inte l ligent chi ld
( 2 4 ) ka l i ma t * ( yang ) k u rang j e l a s
sentence LIG less c lear
not-sa-clear sentence
and sometimes yang is obl igatory because the ' same ' phrase without yang already
has a speci fied meaning , as in ( 24 ) and ( 2 5 ) ( Fokker 1 95 1 : 189)

( 2 5 ) o ra n g * ( yang ) t ua
person LIG o ld
a ld man/woman [ o rang t u a parents ]
50 JOHN W . M . VERHAAR , S . J .

( 26 ) men te r i * ( yang ) muda

minister LIG young
young minis ter [ me n t e r i muda vice-minister ]
where the phrases in brackets are of the kind i l lustrated in ( 17 ) , above .

It seems necessary to add that , though contrast entai ls emphas i s , emphas i s

need not necessarily b e contrastive . For example in ( 2 7 )

( 2 7 ) h a s i l "' (yang ) bagus

resu lt LIG beautiful
a terrific result
I t i s not at all necessary that the result be contrasted with a disappointing
result . Emphasis , of course , by its very nature , ' loosens ' the ' bondedne s s ' of
noun and attribute , since the attribute tends to be , of itse l f , semantically
what it i s syntactically : ' subordinated ' . For example , in Eng l ish , as in many
other languages , emphasised attributive adj ectives are stressed . This wil l be
easier when the head noun has already been mentioned , but when there is a first­
time occurrence , in the discourse , o f a head noun , the ' main ' stress on the
emphas i sed adj ective may wel l be watered down to something l ike ' even stress ' .
I t is , for example , somewhat difficult to emphas i se good in a good c a r in case
c a r fal l s out of the blue as a new sal iency point in the discourse ; and the
speaker may wel l resort to relativisation (a ca r t h a t i s ( rea l l y ) g oo d ) for the
sake of emphas i sing the attribute , in e f fect needing a l igature ( t h a t ) .
Consider also attributes to ' heads ' that are ' heads ' only syntactically but not
semantical l y , as is the case with numeral classifiers . A comfo r t a b l e pa i r o f
s hoes wil l get by , but ," a fa t h e a d o f ca t t l e will not ; this i s because the
paratactic relationship conflicts with the semantic relationship .
I wil l assume , for the moment , that the use or non-use of yang where it i s
' optional ' from a purely ' formal ' point of view i s almost certainly a matter o f
bondedness also , weak or strong respective ly , and therefore perhaps also
distingui shable hierarchically . It would be worth while to test this for
languages having l igatures of a morphemically free form (as is the case with
Indonesian yan g ) ; this would entail expanding Foley ' s sample , most l igatures
of which are morphemically bound .

' RE P LAC I V E ' yang

I borrow the term ' replacive ' from Downing 1978 , where i t i s used to
characterise relative clauses without a ' head ' . I apply it , then , to any noun
phrase with yang which has no ' head ' , no matter whether the phrase is a relative
clause or belongs to some higher level of the bondedness hierarchy . Since yang
i s frequently used without a head , it merits some discussion here , but it i s
doubtful i f thi s u s e of yang i s still that of a ' ligature ' . Indeed , we may wel l
b e concerned here with something speci f ic for Indones ian , a s compared to the
languages of Foley ' s sample , the l igatures in which are morphemically bound and
can therefore not stand alone .
It appears that all occurrences with yang at d i fferent level s of the
bondedne ss hierarchy may occur without a head , given a suitable context .
Consider examples ( 9 ) through ( 15 ) , which i llustrate those leve l s . It would be
easy to f ind contexts where those examples might read : yang i t u ( i . e . for ( 9 ) ,
not for ( 8 » ; yang mana ? ; yang empa t [ . . . ] ; yang banyak [ . . . ] ; yang i n dah ;
yang be s a r ; yang d a t a n g t e r l amba t . The context would be such that the head has
been mentioned before , or at least is situationally clearly presupposed . Thus ,
 ON THE SYNTAX OF YANG IN INDONESIAN 51

also in ( 19 ) , rumah may be left out in its second occurrence on the strength of
the firs t ; or in both occurrences on the strength of a previous occurrence .
Yang used replacively is invariably obligatory .
It is important to note that replacive yang replaces the noun full y , in
that replacive yang may even be preceded by a preposition , which it would
otherwise never take . This is quite normal at all levels , but with some
complications at level 1 . Thus , the headless phrases g iven j ust now may become
prepositional phrases of one form or another : dengan (�th) yang i t u ; u n t u k
(for) y a n g mana ? ; t a n pa (�i thout) y a n g empa t i t u : �ith that one ; for �hich one ? ;
� thout those four (ones ) ; etc . Replacive yang phrases may occur in all
argument positions , except those which are obligatorily pronominal (on which
see below) .
As noted , for level 1 there are some problems with replacive yang .
Consider ( 2 8 )
( 28) ? Kepada yang t i d a k mau i ku t akan d i - s aj i ka n aca ra lain
to ?LIG not �ant fo l lo� � l l PM offer program other
To those �ho do not �ant to come along some other entertainment
�i l l be offered
The yang clause is yang t i da k mau i k u t . Some speakers would prefer to have a
head there , e . g . me reka they (or some appropriate noun) , and to them Ke pada
me reka yang t i dak mau i k u t would be better . Some very careful speakers ,
however , approve of ( 28) as it stands . The problem seems to be greater when
the head i s inanimate , as in ( 2 9 )
( 2 9 ) ? ? Ten t ang y a n g s udah ka l i a n pe l aj a r i akan ada uj i an
about ?LIG already you [ PL ] s tudy � l l be examination
About �hat you have already s tudied there �i l l be an examination
and many speakers would want to have an appropriate head there , such as bahan
subject matter , or simply the pronominal antecedent a pa �hat , so that the
result would be : Ten tang bahan/apa yang s udah [ ] . However , the doubtful
. . .

wel l formedness of ( 29 ) may s imply be due to the cacophony of the two /-ang/
clusters i n ten tang yan g ; according t o Sudaryant o (pers . corom . ) headless yang
i n ( 2 9 ) would b e a l l right if the prepos ition were not ten tang but , for example ,
mengena i about , concerning , because that would avoid the cacophony . This may
wel l be so ; another reason might be that mengen a i is really a verbal form ( from
kena (be ) hi t , pre fixed with men - and with a focus ending - i ) . However , about
relativising headless yang after a verb some more analysis would have to be
done .
In ( 2 8 ) and ( 29 ) , the gloss LIG is preceded by a question mark , as I
consider the ' ligature ' status in them ( as wel l as in any headless yang phrase)
doubtful . One reason for my doubt is that , of course , it is difficult to
consider yang as l inking an attribute to a noun that is not there ; however , it
would not be whol ly unreasonable to assume that there is a zero noun , which of
course would be a real constituent , albeit in zero form , and would not be the
same as there being no head at all . More importantly , however , headless yang
is invariably ' definitising ' , even though what is being definitised may be
something generic . In contrast , yang with a head need not necessarily be
def inite , even though it often is . The matter is of some importance , so I
want to devote a separate section to it .
52 JOHN W . M . VERHAAR, S . J .

Yang AS D E F I N I T I S ER
Replacive yang raises some syntactic questions , which may be given some
more re levance across languages if confined to relative clauses alone . For
example , in English , headless relative c lauses are possible only with wha t , and
it does not seem outlandish to recognise in that indefiniteness some diachronic
relation with the interrogative origin of that pronoun , but not with wh i ch or
who , as il lustrated in ( 3 0 ) through ( 3 2 )
( 30 ) Wha t you cannot d o w i l l b e done b y o t he r s
( 3 1 ) *Wh i ch you cannot d o w i l l be done by o t h e r s
( 3 2 ) *Who you c a n n o t phone w e w i l l v i s i t
but when ' de finiteness ' i s taken out of the latter two , by adding -eve r , the
results are wel lformed , as in ( 3 3 ) and ( 3 4 )
( 3 3 ) Wh i c heve r you cannot d o I w i l l t a ke ca re o f
( 34 ) Whoeve r cannot he l p s hou l d l e t us know
(on condition , of course , that wh i ch - is used adj ectival ly , i . e . with anaphoric
deletion of the head ) . Even wha t , already pretty much indefinite in ( 30 ) , is
made even more so by adding -eve r , as in ( 35 )
( 3 5 ) Wha teve r you cannot d o w i l l b e done b y o t h e r s
Or , alternatively , the head may b e ' attracted ' into the relative clause itse l f ,
with the relative pronoun a s i t s attribute (except with who , which i s invariably
substantiva l ) , as in ( 36 ) and ( 3 7 )
( 36 ) Wha t ever j ob you cannot d o w i l l b e done by o t h e r s
( 3 7 ) Wh i chever j o b you cannot do w i l l be done by o t he r s
and , although ' indefiniteness ' i n ( 3 7 ) and ( 3 3 ) i s restricted i n that wh i ch ­
selects from a l imited number o f j obs , yet within that l imited number there i s
no definiteness o f any kind . ' Attractions ' l ike these may also b e found i n
Classical Latin , a s i n ( 38 ) ( for which I am indebted to Eceizabarrena , Sophia
University )
( 38 ) Quam q u i s q ue nov i t a r t em , i n hac se exe rceat
RP everyone know ART in this se lf exeraise
Le t everyone who knows an art praatise it
( I suspect that such ' attractions ' are triggered b y the OV t o VO change that
carries the change from prenominal relative clauses to postnominal ones . )
In contrast , replacive yang clauses in Indonesian invariably ' definitise '
whatever it is that is modi fied by the attribute , and indefinitene s s wi l l
preclude the use o f yan g . Consider ( 39 )
( 39 ) Yang m i s k i n pe r l u d i - to l ong
?LIG poor neaessary PM he Zp
Those who are poor mus t be he lped
In ( 39 ) , the reference is not to all the poor , but to those who are poor in one
particular group that is ( contextua l ly ) we ll defined . On the reading that all
the poor nlust b e helped (whoever , wherever) ( roughly , therefore , equivalent to
the reading of English t h e poo r , without a following noun) , ( 39 ) would be
deviant , and instead of yang the collective determiner kaum would have to be
use d ; Ka um m i s k i n [ . J.
The speaker of ( 3 9 ) has the poor members of one
. .
 ON THE SYNTAX OF YANG IN INDONESIAN 53

particular group in mind , and those poor members are identi fiable ; I have tried
to convey that idea of definiteness by the gloss those who are poor. According
to Kaswanti (pers . comm . ) yang in ( 39 ) is even contrastive , i . e . in contrast to
those ( in the group) who are not poor .
Apart from replacive yang , some instances of yang with an antecedent have
yang as necessarily definite . Thi s is certainly so with de ictics , interroga­
tives , for obvious semantic reasons ( the interrogative , of course , inquires
after a definite [ . . . ] ) , and with quantifiers (as discussed above ) . With
adj ectives , whenever contrastive . But there is a special problem with l a i n
other, as pointed out by Kaswanti , in examples ( 4 0 ) and ( 4 1 ) ( I have changed
the latter sl ightly)
( 4 0 ) Be rhubu ng tempat i n i a kan d i - b e r s i h kan , seba i knya k i ta
because p lace this wi l l PM c lean it-is-bes t we [ INCL ]
p i ndah ke t empa t ( "'yang ) l a i n
transfer to p lace LIG other
Because this room is going to be c leaned, we had better move
to some other p lace
( 4 1 ) Tempa t ,', ( yang ) l a i n i t u s u d a h t i d a k kosong l ag i ; padaha l
p lace LIG other that already not empty again neverthe less
tempat i t u tad i ma s i h kosong
p lace that just-now s ti l l empty
That other room is no longer free; but just now it sti l l was
In ( 40 ) , yang is forbidden because the other place the speaker wants to go to
is not definite in his mind ; any other place will do , provided it is available
for use . In contrast , in ( 4 1 ) , yang is obl igatory because the speaker is now
talking about one particular room . Perhaps there should be a special rule for
l a i n , which is idiosyncratic among adj ectives in that it may e ither follow or
precede the head noun ; or , perhaps , if placed to the left of the noun , it is
not an adj ective but an indefinite numeral (other = more ) . I will not go into
this problem now, for there are still other problems with l a i n which would have
to be solved first ( and would take me too far afield) , notably when to use , or
not to use , - n y a after l a i n in postnominal position . To unravel all this would
easily take a whole pape r .
One characteristic of y a n g i n Indonesian i s that it may connect two nouns ,
which is what Foley ( 19 7 6 ) says a l igature never does . First , a few examples ,
in ( 4 2 ) and ( 4 3 )
( 4 2 ) o rang ,', ( yang ) p r o f e s o r i t u
person ?LIG professor that
that man, the professor
( 4 3 ) ad i k- mu ,q yang ) ten t a ra i t u
(younger) brother you ?LIG so ldier that
your brother, the so ldier
One may occasionally hear the opinion that this use of yang is not ' correct '
Indonesian , and that it is an innovation through the influence of the ' article '
in languages l ike Dutch and English. However (as argued in Verhaar 1980 and
Kaswanti 1981 ) , the credentials of yang between nouns in older Malay are
impeccable , and there seems to be no need to blame interference . But yang here
does have something in common with the article in languages l ike English or
Dutch ( as wel l as with other languages having definite articles ) , i . e . its
54 JOHN W . M . VERHAAR , S . J .

' definitis ing ' function ( symbolised by the i n the glosses above ) , making a unique
identification by way of an apposition . This type of yang , given a suitable
context , may also be used replacively , in such phrases as yang p r o f e s o r i t u
and yang t en t a ra i t u . Here yang not only ' definitises ' semantically , but also
becomes a ' determiner ' syntactically , and thus becomes virtually an attribute .
Since that does not seem to be characteristic of yang before a non-nominal
attribute (with or without a head) even where it is definitis ing (which it must
be if used replacively) , it is doubtful whether yang between nouns can be
called a ' l igature ' at all ; hence , again , a question mark to LIG in the glos s .

This is perhaps the place for a short digression on Foley' s claim ( 19 7 6 ) that
' l igature s ' ( in his interpretation of that term , which would certainly inc lude
Indonesian internominal yang ) are tied to the specific relations obtaining
between noun plus ' ad j unct ' , and not to those (often non-specific one s ) between
noun plus noun . Foley ' s main point here seems to be a semantic one . In his
view , the difference between a noun phrase with a non-nominal attribute and one
with a nominal attribute is that , whil e in the former the semantic relation is
basically determined by the attribute , in the latter there may be all sorts of
semantic relations , not dependent on the attributive noun alone . What Foley
refers to is the multiple semantic relations that may hold between the nouns in
phrases l ike My ron ' s s t a t ue ( i . e . a statue symbolis ing Myron , or made by Myron ,
or owned by Myron , etc . ) or John ' s p re s e n t ( i . e . a present given by John , or to
John , etc . ) (Foley 1 976 : 7 9-80 ) .
What Foley feels he has to explain is that sometimes a phrase of the form
noun + l igature + noun seems to be counterevidence to the assumption that noun
+ ' ad j unct ' and noun + noun are bas ically different , among other things in that ,
supposedly language-universally , noun + noun does not need a l igature of any
kind . He adduce s evidence from Chinese and Trukese to show that the exceptions
to that rule are only apparen t ; however , that part of his reasoning is hard to
evaluate as he gives no data . He discus ses also numeral ' ad j uncts ' , which ,
because of the property of many languages that they have numeral classifiers ,
may then take the form of ( numeral plus ) noun ( i . e . the c lassifier) plus noun .
I would like to say that , semantically , the numeral clas s i fier noun is not the
' head ' , even though , in a sense , it might be called that purely syntactically .
This is why "'a fat head of ca t t l e is not well formed , as I noted above in a
s l ightly different context . But what is ( to me ) most interesting is how Foley
deal s with noun + l igature + noun in Palauan , where , for example , we may observe
( 4 4 ) ( Foley 1976 : 84 )
( 44 ) a ? e rm- ek el bab i
ART anima l ISG LIG pig
my anima l, the pig
having the ligature e l in between nouns . The issue here is that there is an
apposition . I do not know Palauan , but the language is OV , and my bet is that
the gloss of ( 4 4 ) is not my anima l, the pig but the pig, my animal : that is to
say not the pig , but my anima l is the apposition ! The reason for my hunch is
that that is the way OV languages deal with ' appositions ' : the quotes I use
here are ' scare quotes ' , for OV languages do not really have ' appositions ' , if
by ' apposi"tions ' be meant paratactic , and not hypotactic , co-constituents to
nouns . Consistent OV syntax cannot have any (unambiguously) non-restrictive
attributes , a point to which I will return below , in regard to relative clauses .
Appos itiona are ( normally) non-restrictive . Thus , in Japanese , in ( 4 5 )
 ON THE SYNTAX OF YANG IN INDONESIAN 55

( 4 5 ) t omodac h i no Ta naka-san
friend SM HON
Mr Tanaka, my friend
has what would be the head in the English equivalent such as that appearing in
the gloss as an a ttribute , a subordinated (or hypotacti c ) constituent , therefore
( SM= subordination marker ; no in possessive constructions marks the possessor ,
but here , of course , there is no question of a possessive relationship) . There
is no non-restrictiveness here . Engl ish sometimes reverses nouns ( i . e . ' head '
and ' attribute ' ) in the way comparable to the construction of ( 4 5 ) , as in ( 4 6 )
( 46 ) a gem of a n i dea [ L e . "an i d ea wh i c h i s ( 1 i ke ) a gem" ]
a p r i nce of a fe l l ow [ L e . "a fe l l ow who i s ( l i ke ) a p r i nce" ]
a he l l of a p ro b l em [ L e . "a p rob l em wh i ch i s ( l i ke ) he l l " ]
(double quotes symbolise paraphrases rather than glosses) .
Now , Foley is concerned to have a ? e rme k in ( 44 ) as an ' adj unct ' . His
argument is that it has to be ( even though it looks like a noun in a l l respects
that matter ) , since the ligature e l is used . But that rather begs the question .
The argument has become c ircular . It is not that the ' nominal ' character o f
attributive nouns cannot b e i n doubt , especially i n Indonesian , i n which
language nouns may be rather ' squishy ' , as pembo ros and pengecut in ( 4 7 ) and ( 4 8 )
(Fokker 1 9 5 1 : 19 1 )
(47) o ra n g ( ya n g ) pemboros
person LIG spendthrift
(aJ spendthrift
( 48) l a k i - l a k i (yan g ) pengec u t
male LIG faint-hearted
(aJ faint-hearted man
are , witness the optionality of yang . However , no such interpretation i s
possible for ?e rmek i n ( 4 4 ) .
Clearly there are still many problems with the language-universal
properties of noun + noun phrases . Foley ' s semantic analysis is probably
basically sound . Perhaps , then , both ' definiteness ' and the kind of syntax
involved ( i . e . OV for alleged ' appositions ' ) must be brought to bear to arrive
at a better view. For such a view I consider Foley ' s approach as a substantial
beginning .

L I GATURES ACROSS LANGUAGES ; yang AS ' CONNECT I V E '

It is time to return to indubitable ligatures . What has been established ,
among other things , by the verification of Foley ' s ' bondedne ss hierarchy ' for
Indonesian yang is the basic ' samenes s ' of this particle at all levels of the
hierarchy at which it occurs , a property which , negatively , entails a lack o f
word class membership ( o r ' categorial status ' ) of this particle and its
inability to ' fill ' a functional ' argument ' position within the relative clause
itse l f ( lack , that i s , of ' argument status ' ) . This , evidently , makes yang
di fferent from relative pronouns , which , as pronouns , have both categorial and
argument status . S ince , therefore , yang is not a pronoun , let me call yang as
a relative clause introducer a ' connective ' ( following Downing 197 8 ) . The term
' l igature ' , then , I maintain as a generic term , comprising pronominal ligatures
as wel l as those of the ' connective ' type .
56 JOHN W . M . VERHAAR , S . J .

Some o f the differences between pronominal l igatures and connective ones

are known . More particularly , relative pronouns invariably in troduce the
relative clause (hence forth : RC ) , and the RC has to be postnominal if so
introduced . In contrast , prenominal RC ' s (which occur typically in OV languages )
never have relative pronouns . I t is not true , however (despite Downing ' s
statement to the contrary : Downing 1978 : 39 4 ) that prenominal RC ' s never even
have any connective either : Palauan , which is OV in structure and has prenominal
RC ' s , l inks those RC ' s to the (postsequent) head with the connective e l or 1 ,
and this connective is typically positioned at the end of the RC and immediately
preceding the head . I have not hunted for other languages having the same
structural characteristics of RC ' s , but there seems to be no reason to assume
that Palauan is extraordinary in thi s regard .
The argument status of relative pronouns ( RP ' s ) in a language l ike Engli sh
is we l l known , and I exemplify a few typical cases of its argument distribution
in ( 4 9 ) through ( 5 4 )
( 4 9 ) t h e m a n WHO came yeste rday
(50) the s a l e s c l e r k WHOM I s aw l a s t week
( 5 1 ) the g i r l (TO ) WHOM C h a r l e s gave the r i ng
( 5 2 ) the foo l THAT he was
(53) ( l ook i ng l i ke ) a baseba l l p l ay e r , WH I C H he a ppea red t o be
( 5 4 ) the a ccoun t a n t W I TH WHOM s he wen t ove r t he reco rds
I n ( 4 9 ) , who is the sub j ect of the RC ; in ( 50 ) , whom is the direct obj e c t ; i n
( 51 ) , ( to ) whom is the indirect obj ect ; in ( 5 2 ) that is the ' nominal part of
the predicate ' ( restrictive , in this case ) ; in ( 5 3 ) , wh i ch is the same (but
non-restrictively) ; finally , in ( 5 4 ) , w i t h whom is an ' adj unct ' , of the
exocentric type , with whdm as the head within that type .
NOw , in contrast , consider the Indonesian phrases in ( 5 5 ) through ( 5 9 )
( 5 5 ) tamu yang d a t a n g t e r l amba t
guest LIG come late
the guest who was late
( 5 6 ) pemba n t u yang menge rj a ka n t ugas i n i
he lper LIG accomp lish task this
the emp loyee who does this job
( 5 7 ) *orang yang saya mengundang
person LIG I invite
the man I am invi ting
( 5 8 ) * t eman kepada yang saya menj e l a s kan ma s a a l a h i n i
friend to LIG I explain prob lem this
the friend to whom I exp lained this prob lem
( 5 9 ) * t eman yang saya menj e l a s ka n masaa l a h i n i kepada-nya
friend I exp lain problem this to him
the friend to whom I exp lained this problem
First , let me assume ( incorrectly , but j ust for the sake of argument) that yang
in these phrases has argument status . On that assumption , grammaticalness and
ungrammaticalness of ( 5 5 ) through ( 5 9 ) could be explained up to a certain extent .
That is to say , yang in ( 5 5 ) and ( 5 6 ) would be the ' subject ' of the RC ; in ( 5 7 ) ,
yang would be the ' obj ect ' but in prepredicate position , which is invariably
 ON THE SYNTAX OF YANG IN INDONESIAN 57

imposs ible in this language , so that ungrammaticalness would be ' explained ' .
Phrase ( 5 9 ) would not be wel l formed , for yang would be an ' indirect obj ect ' ,
which is equally impossible in this language in prepredicate position ; and the
pronominal copy at the end of the phrase would not restore the balance . The
ungrammaticalness of ( 58 ) , however , could not be ' explained ' on s imilar grounds ,
for it would be quite possible to consider kepada yang as not necessarily an
' indirect obj ect ' : it could also be a peripheral constituent in the RC , so that
argument status would not come in at all , or at least it would fail to explain
why yang could not have a prepos ition ( kepada ) with it . (In fac t , headless yang
can , s o why not here ? )
However , rej ection of the argument status o f yang on typological grounds
is more convincing . These grounds are wel l known (especially for Phi lippine
languages) , and I may state them here briefly . In this typology , the princ ipal
constraint on RC ' s is that the focus marking on the verb in the RC ( let me call
this the ' relative verb ' , or RV ) should be such that the head of the RC is the
' target ' of that focus . Thus , in ( 5 6 ) , rnenge rj akan is marked for the ' agent '
focus (by the ' prenasali sation ' rnen - ) , and pernba n t u i s that ' agent ' . In ( 5 5 ) ,
d a t a n g , which happens to be monomorphemic , is unmarked for focus , but the one
who comes is still the. guest. In contras t , the focus marking in ( 5 7 ) through
( 59 ) , where the RV ' s have the same prenasal isation as the RV in ( 5 6 ) , i s
' targeted ' on saya , not o rang o r terna n ; therefore , these three phrases are not
wel lformed . This rule of focus marking on the head could, in itse l f , be
expressed by saying that the l igature yang must invariably be the ' subj ect ' of
the RC . However , in most languages of the typology under review here , the
ligatures under discussion here are not only bound forms morphemically , but the
constituents to which they have been ' welded ' have , in some instances , not even
independent existence as constituents : a good example is Tagalog a n g , whose
divi sion into a - n g makes sense in that - ng is a recurrent item in various
argument positions of nouns having a ligature in that language ; but for a- to
be separated in that language makes no sense synchronically , since it never
occurs independently , or , for that matte r , with any other l igature . That
Indones ian yang happens to be morphemically free makes no difference to its
essential sameness with the l igatures of Tagalog , and of many related languages ,
in regard to its es sentially non-argument character . That in e ffect rules out
the ' explanation ' of yang as necessarily the ' subj ect ' of the RC .
This stated , let me now develop a theoretical framework for relativis ing
ya n g , in a manner that has some relevance across languages , and may be the
framework within which a step forward may be made to more knowledge of yang
within a view that has reasonable relevance across languages . As I am now
dealing with RC ' s with a head , let me speak about that head , or antecedent ,
first . Across languages , the head of a RC i s known to have various properties .
First, the head may be considered according to the place it takes in the clause
in which it is an argument , or a peripheral constituent . (That clause may
itself be a main clause , or a subclause : it makes no difference for the point
being di scussed here . ) Let me call the head considered from that angle the
' main clause head ' (MCH) . Second , the head may be considered as part of the
noun phrase which consists of head plus RC ; considered from that angle , the
head is ' proleptic ' in regard to the RC , and let me call it the ' proleptic head '
(PH) . (Of course , I deal here only with an antecedent head , not with heads
relativised prenominal ly ; Indonesian has only antecedent heads of RC ' s . )
About the MCH I may be brie f . Any full noun NP , whether nuclear or extra­
nuclear in its own c lause , may be relativi sed in Indonesian . Among ful l pronouns,
only those that are preposed Agentives , with no pos sibility of interposition of
58 JOHN W . M . VERHAAR , S . J .

any other constituent between pronominal Agentive and verb al form , preclude
relativi sation , no matter whether the pronoun is cliticised or not (on this ,
see Verhaar forthcoming) .
For the PH , however , there are a number of characteristics relevant to
Indonesian , statable as at least three different forms of ' co-referentiality ' ,
which I shall discuss now .

Yang CLAUSES AND CON D I T I ONS OF CO-RE FERENT I AL I TY

Any RC , first of all , that has a head is tied to that head by some form of
' co-referentiality ' , in the sense that , in any language , a RC is , as Kuno has
phrased i t , "a statement about its head noun" (Kuno 1976 ) . Markings of RC ' s
across languages , according to Downing ( 1978) , are mainly of three kinds . The
first is that of pronominal ligatures , or RP ' s . The second is that of a
' connective ' ligature , of the type found in a number of Austronesian languages ,
as discussed above . The third is some special marking on the RV in some
languages ; the marking is syntactically wholly different from both the relative
pronoun and the connective , but it is irrelevant for the matter in hand here .
The interested reader may be referred to Downing ( 1978 ) .
The question now arises how the PH differs according to relativisation
introduced by a RP , or by a connective . Let me call the PH relativised by a RP
( actually, also by relative adverbs , a point to which I wil l return ) , the
' delegating head ' (DH) . That is , such a DH ' delegates ' the argument position
it wil l take within the RC by co-reference to the RP . In other words , the DH
is itself neutral as to the position of its ' representative ' , the RP , in the
argument s tructure in the RC : the DH ' delegate s ' that position to the RP .
Phrases ( 4 9 ) through ( 5 4 ) , above , illustrate this : in each of those , the RP
who , whom , etc . takes its own argument position in the RC , unrelated to whatever
relation the RV might have to the head . Such a relation is not direct , but only
indirect , through the RP , to which the DH delegates the RC-internal argument
position . What is more , the RP may even be a hypotactically al igned part of an
argument to the RV , as shown in a phrase like ( 6 0 )
( 6 0 ) the sa l e s c l e r k WHO S E wa l l e t she fo und on t he coun t e r
where whose is not an argument to the RV f ou n d , but j ust an attribute to the
obj ect NP (wa 1 1 e t ) of the RV . (This ' delegation ' , therefore , may even be one
to a lower hierarchical level . ) Since I will be mentioning two other
co-referentiality conditions in what fol lows , giving them names , I might as well
label the one discussed here , call ing it ' delegating co-referentiality condition'
(DCC) , but I shall not have occasion to deal with it any more except once or
twice rather in passing .
In contrast , the head of a RC marked by a connective ( l ike y a ng ) does not
' de legate ' anything to the connective , which cannot have any argument status .
Inevitably , then , the co-referentiality is one between the focus of the RV and
the head as its ' target ' . Let me call this condition the ' focus co-referentiality
condition ' (FCC ) . It is the condition discussed in regard to examples ( 55 )
through ( �; 9 ) above .
However , not all RC ' s in Indonesian are characterised by the FCC , and
there is yet a third condition . Consider ( 6 1 ) through . ( 7 0 )
( 6 1 ) Guru I n I yang anak- nya men i ngga 1 . [ . . . ]
teacher this LIG chi ld-his die
This teacher, whose child died, [ . . . ]
 ON THE SYNTAX OF YANG IN INDONESIAN 59

'
(62) Pemban t u yang s a p u - nya h i 1 ang [ . ] . .

jani tor LIG broom-his disappear

The janitor whose broom has disappeared [ . . . ]
( 6 3 ) Saya ma u membe l i rumah i t u , yang a t a p - nya d i - bongka r
I want buy house that LIG roof-its PM tear-down
I want to buy that house, (the one ) of which the roof has been
torn down
( 6 4 ) Saya mau membe l i r uma h i t u , yang d i - bongka r a t a p - n y a
[ gloss same as i n ( 6 3 ) ]
( 6 5 ) '''' G u r u i n i , yang me n i ngga l a n a k - nya [ . . . ]
[ see ( 6 1 ) ]
( 6 6 ) ," Pemba n t u yang h i 1 ang s a p u - n ya [ . . . ]
[ see ( 6 2 ) ]
( 6 7 ) Tema n yang s u d a h l ama saya i ng i n menj e l a s ka n ma s a a l a h i n i
friend LIG already long I want exp lain problem this
kepada - n ya [ ] . . .

to him
The friend to whom I have already wanted for a long time to exp lain
this problem [ . ] . .

(68 ) *Anak yang d u a ha r i yang l a l u A l i memu ku l - nya i t u

child L I G two day LIG past [ name ] beat him that/the
te tangga s a ya
neighbour I
The boy that A li beat two days ggo is my neighbour
( 6 9 ) Ana k yang s ud a h s e r i ng ka l i A l i memu k u l - nya i t u
chi ld LIG already often time [ name ] beat him that/the
t e t angga saya
neighbour I
The boy that A li has beaten already so often is my neighbour
( 7 0 ) In i me r u pa ka n soa l yang s u d a h l ama k i ta i ng i n
this be prob lem LIG a lready long we [ INCL ] want
memb i ca r a ka n - nya
discuss it
This is a prob lem that we have already wanted to discuss for a long
time
In sentences ( 6 1 ) through ( 6 6 ) may be observed the second form of
co-r'eferentiali ty , which I call the ' possessor ' s pro- form co-referential i ty
condition ' (PPCC ) . In contrast , the co-referentiality condition i llustrating
the third kind is found in ( 6 7 ) through ( 7 0 ) , and I label it the ' de layed pro­
form co-referentiality condition ' ( DPCC) . Let me now discuss these types
brie fly .
What makes the PPCC different from the DCC i s that in the PPCC the head of
the RC is not the focus of RV marking . Instead , in the PPCC-control led RC , a
possess ive pro- form -nya which is co-referential with the head is at the same
time such that the posse ssee is the ( formative ! ) subject of the RC : a n a k in
( 6 1 ) , s a p u in ( 6 2 ) , a t a p in ( 6 3 ) , a n a k in ( 6 5 ) , and s a p u in ( 6 6 ) . I have
skipped ( 6 4 ) , in which the RC has no formative subj ect . The subj ect , which
60 JOHN W . M . VERHAAR , S . J .

normal ly i s preverbal in Indonesian , but which may under certain conditions be

postverbal , must be preverbal in RC ' s . It is because the subj ect is postverbal
in ( 6 5 ) and ( 66 ) that the se sentences are not wel l formed ; their wel l formed
sequence is that of ( 6 1 ) and ( 6 2 ) respectively . In other words , in this type o f
RC , permutation of subject and predicate i n the RC is forbidden . Why , then , is
such a permutation al lowed in ( 64 ) ? The reason is that , there , the possessee
is inal ienably possessed by the possessor , i . e . the roof by the house : if the
roof is torn down , then the house itsel f is , at least in part , being torn down .
Such ' inalienabil ity ' is not found in ( 6 1 ) and ( 6 2 ) : the child ' s death does not
entail the father ' s death , and for the broom to disappear is not the same as
for the j anitor to di sappear . However , the permutation of possessor and
posses see in ( 6 4 ) does not cause the subject to move to postverbal position .
Instead , the RC in ( 6 4 ) does not have a subject any more than do RC ' s controlled
by the nec , and the only difference between (64) and ( 5 5 ) - ( 56 ) is that (64) has
a pro-form , while ( 5 5 ) - ( 5 6 ) do not . In conclusion , it seems that the constraints
for the PPCC are that the possessee must be the subj ect in the RC , in preverbal
position , and that that argument may become postverbal (and lose subjecthood in
the proces s ) only when the possession relation is inalienable . The permutation
is possible on the basis of a similar permutabi lity in main clauses (Verhaar
1978 ; in that paper I still , naive ly , accorded to yang argument status , without
us ing that term ; indeed the· entire issue had not occurred to me then ) .
The DPCC is mysterious ( to me ) , and deserves more study ; I am indebted to
Kaswanti for drawing my attention to data which I now try to explain with the
DPCC constraint. In ( 6 7 ) , the co-referentiality is between the head ( t erna n )
and the postverbal - nya , which i s i n direct obj ect position . Note how close
( 6 7 ) is to ( 5 9 ) , which , nevertheless , is not wel lformed ! Kaswanti has analysed
the difference between those two phrases as follows (Kaswanti 1981) . He says
that ( 5 9 ) is ungrammatical , basically because the (what I cal l ) FCC has not been
met . In contrast , though that condition is not met in ( 6 7 ) eithe r , tha t phrase
has a modal verb modifier not found in ( 5 9 ) . Kaswanti ' s example is actually
not ( 6 7 ) (of which otherwise he approves ; pers . comm . ) , but ( 7 0 ) , which has a
similar modal modi fication to the RV . ( In Verhaar 1978 I doubted the wel l­
formedness of ( 7 0 ) , misled as I must have been by phrases of the type of ( 5 9 ) . )
Examples ( 68) and ( 69) I also borrow from Kaswanti ( 1981) , and the wel l ­
formedness evaluation is his . Note the verbal modification in ( 6 8 ) , which is
not wel lformed , is merely one of time (dua ha r i ya ng l a l u ) , and not ' modal ' as
in ( 70 ) . For whatever my command of Indonesian is worth , I have some difficulty
' feel ing ' s udah se r i ng ka l i in ( 6 9 ) as ' modal ' , but the problem may be j ust the
term , and an ' aspectual ' interpretation ( ' iterative ' ) is certainly not far­
fetched : Ali has the habi t of beating the child .
While ' modals ' may wel l be relevant in some such way as Kaswanti suggests
in allowing ( 69) and ( 7 0 ) , what seems to me to be subj ect to no doubt at all
is that the ( comparative) length of the RC must be a highly operative factor in
whatever makes the DPCC apply . It is for this reason that I have called this
type of co-referentiality the ' delayed ' one , since the pronominal copy comes at
the very end of the RC , which is what cannot happen if the PPCC applies ( except
by coincidenc e , as in ( 6 4 ) , because of the inalienability permutation , in a RC
which is very short ) . Perhaps the ' delayed ' pro-form , in RC ' s of greater length,
is triggered by a characteristic of all yang clauses : that they are wel lformed
as main c lauses simply by cutting off yang . This stamps yang clauses as
cons iderably more ' paratactic ' with the head than can be the case in pronominally
introduced RC ' s in languages that have that type of RC ' s . This i s , of cour se ,
as suming ·that there is a whole gamut of intermediate pos sibilities between
 ON THE SYNTAX OF YANG IN INDONESIAN 61

whol ly paratactic and whol ly hypotactic relationships , a point to which I wil l

return once more below . Such a n assumption would rather straightforwardly
explain the DPCC in clauses of this type , for then the ' delayed ' pro-form would
be c loser to the kind of pro-form we find in a more loosely knit di scourse
structure . .The RC then becomes rather a sort of ' afterthought ' .

PART I C I P IAL ATT R I BUTES AND CO-REFERENT I A L ITY

I have said that yang clauses are subj ect to certain co-referentiality
constraints , especially that of ' focus ' ( FCC) . It can hardly be a coincidence
that a similar co-referentiality constraint seems to hold for participial
structures in languages that have participles . Consider ( 7 1 ) through ( 7 8 ) , the
f irst three of which are taken from Schwartz ( 1976 : 52 9 ) , and the others from
Downing ( 1978 : 3 9 5 ) :
( 7 1 ) The boys [ ¢ t h row i ng s tones a t the dog ] ran away
( 7 2 ) *The s tones [ t he boys t h row i ng ¢ at t he dog ] h i t t he ca r
( 7 3 ) *The dog [ t he boys t h row i ng s tones a t ¢ ] wen t mad
( 7 4 ) peop l e pay i ng money
( 7 5 ) *money peop l e pay i ng
( 7 6 ) damage - caus i ng w i nd s
( 7 7 ) *w i nd - c a u s i n g damage
The point Schwartz makes ( in dealing with I locano , which has the same character­
istics of FCC for RC ' s as does Tagalog and Indone sian) is that we may observe a
similar constraint of co-referentiality in participial attributes in English :
( 7 1 ) is all right because the ' source ' of the action expre ssed in t h row i ng is
co-referential with t h e boys as the head of the entire phrase , while in (72) and
( 7 3 ) there is no such co-referential ity . The point Downing makes is somewhat
different from mine , but his data wel l illustrate the point I am making here .
Apparently , only relativisations of DH ' s are such that the relativisation
can be made on the basis of an entirely independent argument structure of the
RC , with the co-referentiality functional load whol ly on the pronominal
character of the l igature as anaphoric . RC ' s with RP ' s are perhaps much more
hypotactic in regard to their heads , language-universally , than are RC ' s with
connectives l ike Indone sian yang , which seems to be more comparable ,
' tactically ' , to participial attributes of the kind i llustrated . This is
perhaps also the reason why attributes higher on the bondedness hierarchy scale
can be co-ordinated with those lower on that scale , in Indonesian , with only a
single yang introducing both , as in ( 78 )
( 7 8 ) o rang y a n g ba i k d a n b i sa d i pe rcaya
person LIG good and can be-trusted
(aJ good and trus tworthy person
where ba i k is an adj ective and b i s a d i pe rcaya a RC . The gloss given illustrates
the co-ordination of the two attributes ; another gloss , equal ly correct as far
as glosses go , shows the difference with English , i . e . a good person ( *andJ who
can be trusted. The adj ective level ( level 3 of the hierarchy ; see Figure 1 ,
above ) is probably the highest that can be combined in co-ordination with level
7 , that of a RC , in this language , as appears from the ungrammaticalness of ( 7 9 )
62 �OHN W . M . VERHAAR , S . J .

( 79 ) *orang yang i n i d a n b i sa d i pe rcaya

this
*orang yang mana dan b i s a d i pe rcaya
whiah [ ? ]
Language- niversally , such comparisons ra ise tantalis ing que stions : perhaps RC ' s
(at least those introduced by yang ; on other RC ' s in this language , see below)
in Indones ian , assuming there is a language-universally valid ' average ' of the
bondedness hierarchy , are closer to some level in between those of adj ectives
and participles? Or , inversely , Indonesian adj ectives like ba i k would , on such
an ' average ' scale , be closer to ( finite ) verbs in a language l ike Engl ish , in
grammatical organisation?

RELAT I V E C LAUSES AND NON -RESTR I C T I VENESS

The distinction between ' restrictive ' and ' non-restrictive ' RC ' s is wel l
known and needs no explanation here . Let me symbolise these types as RRC and
NRRC .
I believe ( though I have never heard or seen it stated) that prenominal
RC ' s are necessarily restrictive . I have yet to see counterevidence to this
assumption . Indeed , this may almost be expected a priori : a NRRC is like some
kind of ' afterthought ' , and , if one were to have that prenominally , one would
have to have it , so to speak , by way of ' forethought ' . In any event , there are
indications of a syntactic nature . For example , in Japanese , where all RC ' s are
prenominal , the topic marker wa cannot normally appear , and the ' subj ect ' marker
ga must be used instead ( i f there is a ' subj ect ' in the RC at a l l ) : to introduce
a new topic in a RC is , of course , to have a NRRC . Another test would be to try
to introduce a ' sentence adverb ' l ike inaidenta l ly into a RC in this language :
this appears impossible . Needless to say , even OV languages have ways around
such constraints , by way of parenthetical clauses , or certain arrangements of
topic chains , or in general in ' afterthought ' anacolutha , which surely are
possible , at least colloquially , in all languages . In fac t , such ' afterthought '
structures , and right-extraposed formations more generally , are probably the
explanation why some very rigidly OV languages have postnominal RC ' s (as , for
example , in Persian ) . But such developments trigger changes from OV to VO
syntax ; indeed , they are those changes .
My purpose in making these fairly general comments is a special one . The
impossibility of having prenominal NRRC ' s (or , for that matter , any unambiguously
non-restrictive attribute ) in OV syntax constitutes an ' asymmetry ' with VO
syntax , where NRRC ' s are definitely possible . Thus , ' left-branching ' and
' right-branching ' syntax are not in all respects one another ' s ' mirror image ' .
I mentioned , in passing , another ' asymmetry ' above : the apparent impossibil ity
of having ' paratactic ' attribute s , i . e . ' appositions ' , in OV phrase syntax . In
contra s t , such appositions are a common feature of VO organisation of grammar .
I helve called the typical VO ' asymmetry ' as compared to OV the " expansion
to the right" (Verhaar 1980) , and I wish to summarise the major points here .
The theory is based on the use of ' copies ' , of which the best-known type is that
of pro-forms . I hypothesise that most pro-forms are anaphoric , and that
prophoric: forms are relatively rare , unless followed by an ' afterthought topic '
( He we n t away - the k i n g ) , and perhaps largely confined to hypotactic
construc1:ions ( Af t e r he l l e f t , F rank l beg a n to recons i de r ; compare with this
,"H e 1 l e f t , and F ra n k l beg a n to r e c o n s i d e r )
. More generally , pro- forms have , in
a number of languages exceeding chance frequency , been transferred from ' topic
 ON THE SYNTAX OF YANG IN INDONESIAN 63

agreement ' to ' subj ect agreement ' , a development which is wel l known ( e . g . Giv6n
1 9 7 6 ) . However , I wish to suggest that such pro-forms are more typical of VO
syntax than of OV syntax ; compare ( 8 0 ) with ( 8 1 ) , from Indonesian and Japane se
respectively
( 8 0 ) Seko l a h i t u I I ha l aman- nya l ua s
schoo l TM ground its large
This school II i ts grounds are large
( 8 1 ) Ano gakkoo wa II g u rando ga h i ro i de s u
that schoo l TM grounds 8 M large PoMa
That schoo l II i ts grounds are large
(TM = topic marker ; 8M sub j ect marker ; PoMa
= pol i teness marker ; i t u in ( 80 )
=

is not deictic ; II functional pause) . Note that ( 80 ) has the pro-form - nya ;
=

in contrast , ( 8 1 ) has no pro-form; ga is not anaphoric . The reason for the

difference is clear : the possessor in ( 80) ( s e ko l a h ) is to the left of the
pos sessee (ha l aman ) , which is all right in the interclausal relationship that a
topic has to the rest of the sentence , but not intraclausally , so that the
sequential order balance has to be restored , with - nya , placing the possessor to
the right . In ( 8 1 ) , however , where intraclausally the possessor would have to
be to the left anyway , its interclausal position to the left does not require a
pro-form to restore the balance intraclausally . Of course , the kind of topic
agreement i llustrated in (80) is not the same kind as that which gives r ise to
agreement on the verb , but its discourse role i s nevertheless significant . (I
would suspect that verbal agreement would be rare i n verb-initial languages , but
I have not researched that in any depth . ) What is important , for my topic , in
(80) is that we are looking here at the ' intraclausalisation ' of an interclausal
relationship , and this is exactly what happens when topic-comment becomes
subj ect-predicate .
Taking off from that kind of process , what I wish to suggest is that the
pro- forms discussed here are not the only ' copies ' which are typical of ' after­
thought ' follow-ups in speech , and , more particularly , that VO syntax has special
strategies for this not found in OV syntax . (Again , of course , speakers of OV
languages can also add ' afterthoughts ' and then ' intraclausalise ' them, but
then , to the extent that such strategies become prevalent , they will deviate
from OV alignment , and develop towards VO organisation of the grammar . )
Consider the many verbs in Indones ian which , although they are already ful ly
trans itive , still add ( optionally) prepositions as transitional forms from V to
0 : me ng u ra i ka n ( te n t a n g ) to exp lain (about) ; membenc i ( a ka n ) to hate (in regard
to) ; membedakan (an t a ra ) to distinguish (between) ; me nj e l a s ka n (menge na i ) to
c larify (concerning) ; etc . These transitional prepositions are ' semantic copies '
of what is already fully contained , lexically , in the verb , even though it is
not neces sarily the entire semantic content of the verb that is being ' copied ' .
A s imilar ' semantic copying ' may be observed in certain verb serialisations in
this language (actually , mengena i in the last example i s a verbal form) : j a t u h
b e r g u g u r a n to fa l l to b e scattered; t e rpen t a l t e rg u l i ng - g u l i ng to be flung away
to ro l l away ; uj a rnya menamba hkan he answered added; etc . ( resp . to fa l l in
great numbers ; to be flung ro l ling away ; he added in rep ly ) ; also , a similar
' copying ' may be observed in dem i u n t u k for-the-sake-of for- the-purpose-of;
ama t s a n g a t very very ; pun p u l a also a lso ; etc . The preposition da r i of may
often be placed between posses see and pos ses sor , even though that sequential
order is already enough to express possession : r umah ( da r i ) Pak Ahmad Mr Ahmad ' s
house , and da r i even may become da r i pada : r umah ( da r i ( pada ) ) Pak Ahmad . But
perhaps the most interesting instances of such semantic ' copying ' are found in
expressions of the type i llustrated in ( 8 2 )
64 JOHN W . M . VERHAAR , S . J .

( 8 2 ) s u ra t ( ke t e ranga n ) j a l a n
letter information trip/trave l
trave l permission document
cerpan (susunan) A l i
short story wri ting
A li 's short stories
ruang ( t empa t ) rapa t
room p lace meeting
mee ting room
of which there are many . The optional forms are a l l nouns , and they ' copy ' , in
terms of lexical semantic content , part of the semantic content of the first
noun . What is characteristic of such phrases in regard to their syntactic
organisation is that a step-by-step IC analys is will not work . The opening
noun is certainly the head , the closing one is certainly the attribute . The
optional noun is certainly the head of the clos ing noun , but it is not an
attribute to the opening noun . In fac t , barring coincidences (because of
context ) , the opening noun plus the optional noun together make no sense , and
to the extent that they might , they would not be paraphrases of the expressions
including the clos ing noun .
The very form of nominal attributes in Vo syntax has characteristics not
found (to my knowledge) in OV al ignment . Consider a phrase like the f l owe rs on
the t a b l e ; once we know that is a phrase , we know , of course , that on the t a b l e
is the attribute (or prepositional form) . But such an attribute could also be
an adj unct of place in a sentence : for example , I s aw t h e f l owe rs on t h e t a b l e
is ambiguous ; either the table was the place where I saw the flowers , or the
flowers are , attributively , identified as those on the table . It is character­
istic for OV syntax that such ambiguities cannot occur there , as i llustrated by
( 8 3 ) and ( 84 ) , from Japanese
( 8 3 ) Tee bu l u no ue de wa hana 0 m i ma s u
table of top on TM flower OM see
On the tab le I see the flowers

(84) Tee bu l u no ue no hana 0 m i ma s u

AM
I see the flowers on the tab le [ on the tab le as attribute ]
(TM = topic marker ; OM obj ect marker ; AM
= attribution marker) . In ( 8 3 ) , de
=

marks what precedes it as an adj unct (wa follows it to present it as topic ) ; in

contrast , (84) has no ( its second occurrence , after ue) as an attribution marker
( no may also be j ust of, but hardly as occurring after ue ; for a test to this ,
see Verhaar 1980 : 6 1 ; however , that issue does not affect the point being made
here ) . Japanese distinguishes sharply between attributes and adj uncts ; the
former are invariably marked with the postposition no , which places the
attribute in a clearly hypotactic relation to the head ( no matter what the
semantic relation between the two nouns ) , and adj uncts have place , time , etc .
markers .
I suggest that the reason for attribution markings in a way identical
with that: of adj unct markers i s not a coincidence , and is es sentially tied
to VO syntax , i . e . with the attribute to the right . Adjuncts are peripheral
constituE!nts which , unless fronted for topicali sation reasons , are added on
to the argument structure of the verb , to the r ight ; such ' extranuclear '
constituents can then easily be ' intranuclearised ' , in that they become
attributE!s to arguments of the nuclear structure . Once this process has
 ON THE SYNTAX OF YANG IN INDONESIAN 65

established results , then of course such time- or place-marked attributes may be

attached to any noun , nuclear or not . But the origin is that of an afterthought
which is then admitted into the core structure of the clause .
I believe this general characteristic of VO syntax is the appropriate
basis for understanding non-restrictiveness of RC ' s . Of course , NRRC ' s are
genuine attributes , but their relations to their heads are much less ' hypotactic'
than is the case with RRC ' s . In English , for example , a NRRC has opening who
or wh i ch more as and he/she/they and as and this/[ etc . ] than as a co-referential
introduction to a ' l imiting ' attribute which identifies uniquely . Thi s is a
language-universal characteristic of non-restrictive attributes in general , and
such attributes are uniquely characteristic of VO syntax , as compared to OV
syntax .
Indonesian yang clauses are not different , and confirm the general hypoth­
esis j ust phrased . Both RRC and NRRC are pos s ible , and the di fference may
appear , in appropriate context , in the place of the determiner . Compare phrases
( 8 5 ) and ( 8 6 )
(85) ternan saya i t u II yang d a t a n g kerna r i n
friend I DET LIG come yes terday
that friend of mine II who came yesterday
(86) ternan saya yang da tang kema r i n i tu
friend I LIG come yesterday DET
that friend of mine who came yes terday
( DET= determiner ; actually , depending on context , i t u in ( 8 5 ) and ( 8 6 ) could
also be a topic marker ; II functional pause) . I t u appears before the NRRC ,
=

and follows the RRC ; the NRRC is an ' afterthought ' , which does not affect the
determiner of ternan saya ; whereas the RRC is an integral part of the phrase
determined by i t u . (Exactly the same analysis obtains , muta t i s mutandis i f i t u
is topic marker . )
Relativising yang is not itself a ' copy ' , or part o f a ' copy ' , of the
antecedent , since only pronominal l igatures , but not connective ones , could be
that . Nevertheless , yang ' codes ' the ' copy ' element of whatever the co­
referential element is (depending on whether the FCC , the PPCC , or the DPCC
appl ie s ) . Yang is , in discourse communication , the signal for the hearer that
something is going to be added , as a ' statement ' about the noun immediately
preceding i t . Before giving examples i llustrating thi s , let me show an
analogous characteristic in adverbial relativisations in this language . Compare
the last example of ( 8 2 ) , here repeated as ( 8 7 ) , with ( 8 8 )
(87 ) ruang ( t ernpa t ) rapat
room p lace meeting
mee ting room
(88 ) r uang "' ( ternpa t } karn i rnen gadakan rapa t i t u
we [ EXCL ] ho ld
the room where we wi l l ho ld the meeting
In ( 8 7 ) , t ernpa t is optional , a partial semantic ' copy ' of ruang . In contra s t ,
i n ( 88 ) , t ernpa t i s obl igatory ( for the same reason relativising ruang ( j ust as
where is in English , as in the g loss to ( 8 8 ) ) ) . Note that even t ernpa t obeys
the bondedness hierarchy for Indonesian : at level l , that of the RC , it is
obligatory , as in ( 88 ) , whereas at the higher level i llustrated in ( 8 7 ) , that
of a phrase , it is optional . It is , of course , di fficult to assign a leve l
' number ' to the ( 8 7 ) phrase , since Foley ' s bondedness hierarchy does not provide
66 JOHN W . M . VERHAAR , S . J .

for noun + noun ; nevertheless , an attribute which i s not clausal must be higher
than one that is clausal , as in ( 8 8 ) . Tempa t in ( 8 8 ) differs from yang only in
that the head ( r uang ) ' delegates ' its involvement in the RC to a ' proxy ' , i . e .
tempa t , and i n so far there i s a measure o f ' analogy ' with pronouns introducing
a RC . But the analogy is very l imited : relativis ing t empa t can onl y be
' adverbial ' in the internal structure of the RC , and cannot be an argument to
the RV (as can all genuine relative pronouns) , l ike ' subj ect ' ( see ( 8 9 » , or
' object ' ( see ( 9 0 »
(89) * r uang tempa t me rupakan tempa t ba i k u n t u k rapa t
room p lace be p lace good for mee ting
the room which is a good p lace for the meeting
(90) * r uang tempa t saya mempe r s i a p ka n u n t u k rapa t
prepare
the room which I prepared for the meeting
In ( 8 9 ) , t empa t functions as the ' subj ect ' of the RC , but t empa t , being
peripheral to the argument structure of the RC , cannot be that . Hence it cannot
be the ' obj ect ' either , which is why ( 9 0 ) is not wel l forrned . (Of course , an
obj ect cannot be prepredicate in this language anyway . ) Note that relativis ing
tempa t has thi s in cornmon with relativis ing yang that both are ' peripheral ' in
the argument structure of the RC , although the difference between the two is
that t empa t as a relativiser is at least adverbial in its relation to the
argument structure of the RV , whereas yang is not even that . In other words ,
while tempa t does the ' copying ' ( it is co-referential with the antecedent ) ,
yang only codes the ' copying ' ( yang only ' signal s ' some kind of co-referential i ty
between head and the RC it introduces ) .
As place is relativi sed , so also time . Compare ( 9 1 ) and ( 9 2 )
(91) j am (wa k t u ) ke r j a
hour time work
working hours
(92) se l ama j am-j am * (wa k t u ) kamu beke rj a
during hour-RED you work
during the hours when you worked
( RED = reduplication ; I have a persistent feeling that relativis ing wa k t u can
only relate to the past ; this may have to be verified ) . The ( 9 1 ) phrase is o f
the ( 8 2 ) type , and strictly parallel to ( 8 7 ) : wa k t u i n i t is a semantic ' copy ',
and it is optional . In contrast , in ( 9 2 ) , wa k t u is obligatory , and all that
has been said about the syntactic properties of t empa t in ( 8 8 ) holds also for
wa k t u in ( 9 2 ) .
Now consi�er the following examples with yang clauses , ( 9 3 ) through ( 9 6 )
(93) Saya ma u membe l i r uma h i t u , ma k s u d saya yang a t a pnya d i bong ka r
meaning I
I want to buy that house, I mean the one of which the roof has
been. torn down [ see ( 6 3 ) , above ]
(94) Seba i knya kamu membaca l ag i buku i n i , yang saya k i ra ada d i
��t-is-bes t you read again book this LIG I think be in
pe rpus t a kaan pus a t , (yang ) ba rang ka l i akan sanga t memba n t u-mu
Ubrary centre LIG probab ly wi l l very he lp you
 ON THE SYNTAX OF YANG IN INDONESIAN 67

da l am meng had a p i uj i a n bu l a n depan

in confront examination month front
You 'd better read this book once more, which I think is in the Main
Library, and which wi l l probab ly he lp you very much in preparing
yourse lf for the examination next month
( 9 5 ) Saya i n g i n mempe r ken a l kan Sauda ra dengan ad i k saya -
I want introduce you with (younger) brother I
yang memang , sepe r t i s u d a h saya ka t a ka n kema r i n , (ya n g ) mungk i n
LIG indeed as already I say yesterday LIG perhaps
b i s a meno l ong Sauda ra -
can he lp you
I should like to introduce you to my brother, - who, as I already
said yes terday, may be able to he lp you -
(96) PLN menye bu t ka n s ua t u a n g ka , ya i t u I i ma be l a s
[ name power company ] mention a figure name ly fifteen
j uta Rup i a h , yang t e rnya t a s e t e l a h d i l a kukan pendeka t a n
mi l lion appear after be-done approach
dengan ma sya raka t , b i aya i tu t i d a k b i sa d i kumpu l ka n
with society expense that not can be-co l lected
The PLN (power company ) named a sum, namely 15 mi l lion Rupiah, which,
as it turned out after there had been consu ltation with the peop le,
i t would be impossib le to co l lect
Sentence ( 9 3 ) i s the same as ( 6 3 ) , except for the addition ma k s u d saya I
mean , which makes the following yang clause a NRRC beyond any doubt ; it could
even be argued that yang thereby becomes ' replacive ' . Sentences ( 9 4 ) and ( 9 5 )
are from Kaswanti (pers . comm . ) . Note that there are two yang ' s i n each
sentence , the second occurrence of which is optional , and it seems that a
speaker would tend to use it to the extent ( s ) he feels that a topic rather
remotely to the left needs to be picked up again ; non-use of the second yang ,
of course , shows how almost perfectly paratactically such afterthoughts can be
added on ; in ( 9 5 ) , the first yang clause is not even finished before a second
yang may be used to ensure continuity of the entire afterthought , continuity of
which , in thi s instance , may be threatened by the parenthetical sepe r t i s u d a h
s a y a ka t a ka n . Note also that serial RC ' s can be con j oined without any overt
co-ordination device such as dan and, a device which a language l ike Engli sh
would need at least before the last of the serie s , as i llustrated in ( 9 7 )
(97) Let m e i n t roduce y o u t o m y b ro t he r , who , i n c i d e n t a l l y , wa s a s t uden t
a t Ha rvard , ( a n d ) (who ) i s a famous phys i c i s t now , "' ( a n d ) (Who) w i l l
p roba b l y be a b l e to he l p you
where the RP i s optional , but not co-ordinating and before the last serial i sed
RC . In Indones ian , dan is normal ly also needed before the last item of an
enumeration , but not before the last RC in a seriali sation o f RC ' s . I regard
this as typical for the ' paratactic ' nature of NRRC ' s in this language .
Finally , ( 96 ) (which I found in a newspaper , Kompas , 23 January , 198 1 , page XI I ,
col . 3 ) , first introduces a NRRC with yang , and then ( after an embedded c lause
[ se t e l a h . . . masy a r a ka t ]) picks up the head a full phrasal form again ( b i aya i t u ) ,
with anaphoric i t u ! The interesting feature here i s that , i f (relativi s ing)
yang were to be replaced by ( co-ordinating) dan and, the entire sentence would
68 JOHN W . M . VERHAAR , S . J .

be perfectly we ll formed ( according to some , even better ) . I t may be said that

( 96 ) is rather col loquial (as are ( 9 3 ) , ( 9 4 ) , and ( 9 5 ) ) , and prescriptively
minded ( Indones ian) linguists would (as I have heard from themselves ) l ike to
turn down a sentence l ike ( 9 6 ) . Whatever one ' s ideas may be about the relation
between formal and informal Indones ian , it is certain that a sentence l ike ( 9 6 )
is perfectly in l ine with the typology o f this language .

Again , note how the paratactic strategies i llustrated in RC ' s in a

consistently Vo language l ike Indonesian are also uncharacteristic of a VO
language l ike Japanese , where interclausal co-ordination is virtually
impos sible , and where CLAUSE + a n d + CLAUSE i s out ; this language does not even
have a word for English and . The preceding clause is invariably in strict
hypotactic relation to the fol lowing clause in a compos ite sentence , and the
latter is strictly ' main ' claus e . Such OV syntax cannot accommodate afterthoughts
except anacoluthically . And its syntax lacks the features due to ' intraclausal­
isation ' of interclausal relationships , and those due to the ' intranuclearisation '
of peripheral constituents .
In conclus ion , perhaps a word or two must be said about interrogatives
functioning as relativisers in Indonesian . One may find s i a pa , which i s normal
as interrogative pronoun (who ? ) , in RC- i ntroducing capacity ; l ikewise , one may
find d i ma n a (which normally means where ? ) used as would Engl ish relativising
whe re . The consensus among l inguists is that these uses of interrogatives as
relatives are due to interference from Dutch and English . This is probably true.
Neverthele s s , one may venture the tentative prediction that such relative use
of interrogatives may gain the day in the long run . My reason for thinking so
is not that I feel such interference is nece s sarily insuperable , but that the
interfering languages themselves have gone through the same process . Relatives
in many languages that have them have been derived from interrogatives , at least
in part (another part would be derivation from demonstratives , probably
originally used for ' topic agreement ' in the ' comment ' ) . These derivations are
so widely spread that they might develop in a language l ike Indones ian also .
The reason for such developments may we l l be the increasing role of written and
more formal language , in the history of the language . Soc iolinguistically ,
written language tends to be organised more along hypotactic l ines , spoken
language more along paratactic line s . This itself has , I bel ieve , very l ittle
to do with a language having VO or OV syntax . VO syntax will develop its own
devices for parataxis , but OV does the same in terclausa l ly ; in spoken Japanese ,
for example , what would be a complicated multiple embedding ' period ' , is often
colloquially broken down into topic strings , which may go quite a long way
before the main clause makes its appearance . That , too , of course , is parataxis .
What characterises paratacticisation in VO syntax i s that it is largely intra ­
c lausal , and continues to characterise spoken language .
 ON THE SYNTAX OF YANG IN INDONESIAN 69

B I BL I OGRAPHY

DOWNING , Bruce T .
1978 Some universals of relative clause structure . In Greenberg 1978 :
375-418 .
FOKKER, A . A .
1951 Inl eiding tot de st udi e van de Indonesische syntaxi s . Groningen/
Jakarta : J . B . Wolters .
FOLEY , Will iam A .
1976 Compara t i ve syntax in Austronesian . Ann Arbor , Michigan : University
Microfilms .
1980 Towards a universal typology o f the noun phrase . Studies in Language
4 : 17 1-199 .
GIV6N , Talmy
1976 Topic , pronoun , and grammatical agreement . In Li , ed . 1976 : 149- 188 .
GREENBERG , Joseph H .
1978 Generalizations about numeral systems . In Greenberg , ed . 1978a:
249-295 .
GREENBERG , Joseph H . , ed .
1978a Uni versals of human language , vol . 3 : Word structure . Stanford ,
Cal ifornia : Stanford University Press .
19 78b Uni versals of human language , vol . 4 : Syntax . Stanford , California :
Stanford University Press .
KASWANTI PURWO , Bambang
1981 Kata yang dalam bahasa Indonesia . Manuscript .
1982a Deiksis dalam bahasa Indonesia . Ph . D . Dissertation , Universitas
Indonesia , Jakarta . Publi shed as : 1983 , Deiksi s dalam bahasa
Indonesia . Jakarta : Balai Pustaka .
1982b Bahasa Indonesia dalam rangka tipologi Li dan Thompson .
In Kridalaksana and Moel iono , eds 1982 : 1-16 .
KEENAN , Edward L . and Bernard COMRIE
1977 Noun phrase acce ssibil ity and universal grammar . Linguistic Inquiry
8 : 63-99 .
KRIDALAKSANA , Harimurti and Anton M. MOELIONO , eds
1982 Pelangi bahasa . Kumpulan esai yang dipersembahkan kepada Prof.
J . W. M . Verhaar, S . J . Jakarta : Penerbit Bhratara Karya Aksara .
KUNO , Susumu
1976 Subject , theme , and the speaker ' s empathy - a reexamination of
relativization phenomena . In Li , ed . 1976 : 4 17-444 .
 70 JOHN W.M. VERHAAR , S . J .

LEHMANN , Winfred P .

1973 A structural principle of language and its impl ication . Language

4 9 : 47-66 .
Li , Charles N . , ed .
1976 Subject and topi c . New York : Academic Press .
MORAVCSIK, Edith A .

1978 Agreement . In Greenberg , ed . 1978b : 33 1- 3 7 3 .

SCHWARTZ , Arthur

1976 On the universality of sub j ects : the I locano case . In Li , ed .

197 6 : 519-54 3 .
VERHAAR, J . W . M .

1978 Syntactic ( in ) al ienability in Indones ian . In Wurrn and Carrington ,

eds 1978 : 3 1 7 -3 2 5 .
1980a Tipologi struktural dan bahasa I ndonesia . In Verhaar 1980b : 3 2 - 7 1 .
1 980b Teori l ingui stik dan bahasa Indonesia . Yogyakarta : Penerbitan
Kanis iu s .
fie Syntactic ergativity in Contemporary Indonesian . (Paper presented at
the Third Eastern Conference on Austronesian Linguistic s , May , 198 3 ,
Ohio Univers ity , Athens , Ohio ) .
WURM , S . A . and Lo i s CARRINGTON , eds
1978 Second Interna tional Conference on Austronesian Linguistics :
proceedings . Fascicle 1 : Western Austronesian . PL , C-61 .

Verhaar, J.W.M. "On the syntax of yang in Indonesian". In Halim, A., Carrington, L. and Wurm, S.A. editors, Papers from the Third International Conference on Austronesian Linguistics, Vol. 4: Thematic variation.
C-77:43-70. Pacific Linguistics, The Australian National University, 1983. DOI:10.15144/PL-C77.43
©1983 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.