AFLA XIV Tomoko ISHIZUKA
May 5, 2007
Internal Structure of the DP in Javanese* Tomoko Ishizuka UCLA
[email protected]
1. INTRODUCTION
The goal is to describe and analyze some aspects of the DP in Javanese
1.1. Javanese1 Javanese (the Malayo-Polynesian subfamily of Austronesian family) is spoken by more than 75,5,00,000 people in the central and eastern part of the island of Java, in Indonesia. Three Registers: Ngoko (informal), Madya (polite), Krama (polite formal) Map from www.unicef.org
1.2.Fieldwork Data collected at UCLA: field method class. A native speaker of Javanese East Javanese (Malang), and Central Javanese (Jogjakarta) Javanese-Indonesian bilingual; fluent in Arabic and English 1.3. Previous studies of the syntax of the Javanese DP Java Rijkhoff 1990, (cited in Cinque 2005: footnote 19) Typologically common order: N-Adj-Num-Dem Davies & Dresser 2005 Jogjakarta Relatively free ordering of the constituents (Adj, RC, and Num) Malang Head movement analysis:N0 to D0 Adjectives are X0 s that adjoin to N0 1.4. Highlights of the present study The constituents inside the DP are rigidly ordered (N-Adj-D-Num-RC-Dem-RC) Different positioning of Num, Adj, séng RC associate with different semantics/syntax The relative ordering among adjectives is also rigid (mirror image of universal hierarchy) Derived by phrasal movements in Cinque’s (2005) sense. Needs modification
Cinque (2005) derives attested orders in languages from the universal order of Merge Dem > Num > Adj > N from phrasal movement, pied-piping and spec. extraction
DemP due to the presence 2 of possessor and D 2 NumP Dem 2 2 AP Num 2 2 NP A 2 2 N
*Special thanks to Rachmi Diyah Larasati, the Javanese language consultant of this project, for her help, insightful comments, and patience. Many thanks to Hilda Koopman for invaluable suggestions and support. I also thank Kie Zuraw, Asia Furmanska, Katie Schack, Jieun Kim, Vincent Homer, Chacha Mwita, Dominique Sportiche, Christina Kim, Ben Jones and Ted Gibson for suggestions and support. All errors are my own. 1 Ngoko is informal speech, used between friends and close relatives. It is also used by people of higher status to people of lower status. (adopted from http://en.wikipedia.org/)
1
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
2. GENERAL SYNTACTIC PROPERTIES OF JAVANESE Head initial (Spec, Head, Complement) (e.g., VO, [Prep DP], [Possessee-(n)e Possessor]) Extensive use of non-verbal predication / no visible agreement Pro drop / Topic drop language Extensive use of left dislocation and right dislocation Rich Voice system (e.g., Active ‘ng-’, Theme Voice ‘di-’ (3rd person), ‘taq’ (1st), ‘toq’(2nd), Accidental voice ‘ke-’) Relativization is limited to syntactic subjects and possessors of subjects (see Davies 1993) => the relativizer séng attracts a subject to its specifier2. 2 XP 2 (1) a. * aku m-ôtjô buku séng putri n-ulés I AT-read buku REL putri AT-write séng TP (intended: I read books that Putri wrote.) 2 XP b aku m- ôtjô buku séng di-tulés putri I AT-read book REL TT.write Putri 'I read books that were written by Putri.' or ‘I read books that Putri wrote.’ (2)
bejo ke-temu botjah lanang séng bapaq-e {m-ati / n-angés} Bejo ACCV.meet child boy REL father-ne AT-die / AT-cry ‘Bejo (accidentally) met a boy whose father {died / cried}.’
3. DPS No systematic plural marking (e.g., buku (books, a book), bukubuku (pl. distributive)). Constituents of DP: Nouns, prenominal Numerals ([Num-ng CL]) and postnominal Numerals, Adj., Definite determiner ‘-ne’, Possessors, Dem, and séng RC3. The definite determiner ‘-ne’ implies previous mention in the discourse. ‘-ne’ also appears in possessor constructions. (3) a. klambi-ne garéng clothes-ne dry ‘The clothes are drying.’ (‘Clothes’ has to be previously mentioned.) b. aku seneng {krandjang (antik) -e} siti I like basket old -ne Siti ‘I like Siti’s {basket / old basket}’
2
[N-ne]
[N-ne/N-A-ne Possessor]
Abbreviations: AT active voice; TT theme voice; ACCV accidental voice; REL relativizer; DEM demonstrative. The spellings used in this handout are adopted from Horne (1974: 1961). 3 There are four types of demonstratives in Javanese: iku/kui (that/those;closer), kae (that/those;far), and iki (this/these). Iku (kui) and kae are also used anaphorically, referring to pre-established NP.
2
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
3.1. The linear ordering of the constituents in the DP Num-ng – CL – N – A – ne – Possessor - Num – séng RC(A) – Dem – séng RC LINKER POSS/D FOCUS NON-RESTRIC The same constituents seem to appear twice (Num, Adj, séng RC). Does this mean the order is free? NO: Different positions associate with different semantics/syntax.
Two Numerals ([Num-ng CL] and bare numerals) [Num-ng CL] is for mass nouns, whereas postnominal bare numerals are for count nouns.
(4)
rini tuku {telong kilo buku / buku telu} Rini buy 3-ng kilo book book 3 ‘Rini bought {3 kilos of books (i.e., used paper) / 3 books}.’
Two séng RCs. When a séng RC follows DemP, it denotes additional information, resulting in a nonrestrictive RC.
(5) a. aku ndeloq asu telu iku séng I AT-see dog three DEM REL 'I saw these three dogs, which are white' b.
[Mass/Count uses]
putih white
[Num-Dem- séng RC] NON-RESTRICT
aku ndeloq asu telu séng putih iku I AT-see dog three REL white these 'I saw these three dogs which are WHITE.' (white is focused)
[Num-séng RC-Dem] RESTRICTIVE FOCUS
Two Adjectives (bare A, and in ‘séng RC’). => will be discussed in the next section
Table 1: illustration of the word order Num-ng CL
a. aku tuku telong kilo 3-ng kilo I buy b. c. aku ndeloq d. I AT.see
N
A
D
Posses’
Num
séng RC
Dem
séng RC
iku iku
séng putih
séng putih
DEM
REL
apel apple
asu asu asu dog
putih white
-e
(Putri)
ne
Putri
a. I bought 3 kilos of apples. b. I saw these three dogs, which are white. c. I saw these three dogs which are WHITE. d. I saw Putri's three white dogs.
3
telu telu telu 3
REL
white
[Non-restrictive RC] [Restrictive RC and Focus]
white
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
4. ADJECTIVES AND THEIR RELATIVE ORDER DP-internal adjective positions: between N and D, or in séng RC: N – A* – ne – séng RC All adjectives can be used as main predicates. (6) Siti (ora) tuwô Siti (NEG) old ‘Siti is (not) old’ There are restrictions on recursion of bare adjectives: maximum of two. (7) aku ke-tamu wông djôwô (*lanang) tuwô-ne I ACCV.meet person Javanese male old-ne 'I met the old Javanese person.'
[N-A-(*A) - A-ne]
When an NP is modified by three Adjs, one appears in séng clause and receives a focus reading. (8) aku ke-tamu wông djôwô lanang-e séng tuwô [N-A-A-ne séng RC] I ACCV.meet person Javanese male-ne REL old 'I met the Javanese man who is OLD.' Question: How to analyze stacking of bare adjectives? In what order the stacked adjectives appear? Analytical tool: Universal hierarchy of adjectives (e.g., Sproat and Shih 1988:1991, Cinque 2003, Shlonsky 2000, among others). [subjective comment > size > age > shape > color > nationality > material] Example sentences used to elicit the relative ordering among adjectives: (9) a. aku {nemu / ndeloq} krandjang abang tjiliq-e I AT.find / AT.see basket red small-ne ‘I {found / saw} the small red basket.’ (Context: ‘NP-A-A-ne’ (a small red basket) is already mentioned before this target utterance.) b. kertas tuwô gedè-ne duwèku box old big-ne possession-me ‘The big old box is mine.’ (Context: I saw a child carrying a box, and claimed to my friend that it was mine.)
Javanese adjectives exhibit rather fixed relative ordering, which is the mirror image order of the universal order of adjectives (see Table 2 for examples). [material – nationality – gender – color – shape – age – size - subjective comments]
4
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
Table 2 Bare adjectives material<
nationality
emas ’golden’ kaju ‘wooden’
djôwô ‘Javanese’ pranjes ‘French’
Must be in séng RC subjective < shape < age < size < comments bunder anjar gedè aju (people) ‘round’ ‘new’ ‘big’ ‘beautiful’ kotak tuwô tjiliq apik (things) ‘square’ ‘old’ ‘small’ ‘beautiful’ dôwô enam lemu djôwô ‘long’ ‘young’ ‘fat’ ‘generous’ Most often in séng RC *NP-A-ne NP-A-A-ne is still possible in séng A Prefer to be in séng RC
< gender
< color
wadon putih ‘female’ ‘white wedok ireng ‘female’ ‘black’ lanang abang ‘male’ ‘red’ NP-A-A-(*A)-D (easily stackable) in séng RC, Adj receives Focus NP amas pranjes- e NP gold French-NE NP djôwô wedok-ne NP Javanese female- NE NP wadon ireng-e NP female black- NE NP abang dôwô -ne NP red old - NE NP bunder anjar-e NP round new-NE NP tuwô- gedè –ne NP old big - NE NP gedè -ne séng apik NP big-NE REL beautiful Other properties of bare Adjectives… Coordination of bare adjectives is not possible, whereas in séng RC coordination is possible (10) a. médjô abang (*lan) bunder-e duwèku [N-A-(*and)-A-ne] table red and round-ne possession-me ‘The round red table is mine.’ b. aku ndeloq médjô -ne séng bunder lan I AT.see table-ne REL round and ‘I saw the table that was round and red.’
abang red
When an Adj. is intensified, it appears in a séng RC. (11) wông wedok *(séng ) tuwô tenan teka nèng person female REL old very come at ‘A woman who is very old came to my house. Question:
[N-séng A and A]
[N-*(séng ) Adj very] omah-ku house-me
Why are there restrictions on recursion (coordination or intensified bare adjectives)? Koopman and Szabolcsi (2000) and Koopman (2002) propose "complexity filters", which apply to the output of the syntactic derivation at the point of spell out and are sensitive to the depth of embedding. 5
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
4.1. The syntax of Adjectives and deriving NP-A-A-D-(Possessor): In earlier analyses, noun-initial order is derived via head movement, whereas in more recent analyses by Roll-up/Phrasal movement A phrasal movement derivation directly accounts for the distribution of bare adjectives. High adjectives like aju ‘beautiful’ must be in séng RC: why? => they are too high in the hierarchy, thus NP cannot roll up. Same goes for Numerals.
A subjective comments séng RC Too high A size Cannot roll-up
A age N-A-A-ne séng RC “Possible roll-up”
A shape A color A gender
A nationality N-A-A-ne “Roll-up”
A material NP
Proposed derivation DP qo NP3 D' 3 3 NP2 2 -ne r 2 A2 DP NP1 2 French (poss) ring A1 golden
Derivation: (see Cinque 2005) Predicate A1 merges with NP1 NP1 moves to Spec NP2 A2 merges with NP2 NP2 rolls up to Spec,NP3 NP3 rolls up to Spec,DP Deriving the surface order NP-A1-A2-ne-(Possessor)
NP3 2 NP2 2 A2 NP2 2 NP1 2 A1 NP1
Questions: (Output constraints: Koopman 2002) - Why can only 2 (out of 7) adjectives precede D –ne? The restriction of the determiner –ne: -ne can attach to only phonologically light XP - Why is coordination not possible? Probably too heavy. (or AP and AP: too high to roll up in the structure?) - Why are intensified adjectives not possible? Probably too heavy since modification involves more structure (or already focused?) 6
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
5. PUTTING STRUCTURES TOGETHER 5.1. Deriving NP-A-D-Num-Dem from the universal order; Problem with D -ne Simple roll-up does not give us the right order. Rolling up Dem>D>Num>A>NP NP-A-Num-D(ne)-Dem (Cinque 2005) DemP 2 2 DP Dem 2 2 NumP ne 2 2 AP Num 2 2 NP A 2 2 N
This would yield the wrong order [NP-A-Num-D(ne)-Dem]
Proposal: [NP-AP-D-Num-Dem] (table 5) Num does not appear between NP –D ;[*NP-Num-D] Roll-up is possible up to the adjectival field, no more. High Adj and Num are too high in the structure. AP rolls up to Spec,NumP then no more Roll-up. AP continues to move up to SpecDP via Spec extraction and then roll up of DP takes place to Spec,DemP. DemP 2 2 DP Dem 2 2 NumP ne 2 2 AP Num 2 2 NP A 2 2 N
Head Movement derivation would not yield the right order => N-A-Num-D-Dem Head movement is only possible into the adjectival field: the rest has to be XP movement.
5.2. Restrictive and Non-restrictive séng RC Assumptions: Kayne (1994) Restrictive RC: [DP-séng RC- Dem] - An RC is always introduced by D (silent in ] Javanese) DemP - When an RC is in the scope of 'D", it gives 3 rises to a restrictive reading, whereas when DP 3 an RC is not, the reading is non-restrictive. Dem ‘DP’ 3 3 Derivation ‘D’ CP Merge DP with AP(Predicate) 3 Merge C séng DP 3 séng promotes subject DP 5 C FocP Relative clause DP rolls up to N-A-ne-Num séng 2 Spec,DemP (DP always does: Dem is final) r The séng RC is in the scope of the Foc SC silent D. 2 DP 7
AP
D.
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
Non-restrictive RC: [DemP-séng RC] 'DP' Derivation 3 Merge DemP with AP CP 3 Merge séng 'D' CP séng promotes subject DemP to Spec,CP 3 Relative clause CP rolls up to Spec DP DemP 3FocP 5 C 3 Now the relative CP is out of the c-command domain of the silent ‘D’. N-A-ne-Num-Dem séng 3 Focus SC 2 DemP AP Remaining question: Why is the focused reading only possible in séng RC? => Probably because Focus projection is available only in séng RC Further research: This derivation predicts that [DemP-séng RC] is ambiguous between restrictive and nonrestrictive readings due to reconstruction. This point needs to be tested (see Ishizuka 2006) 6. CONCLUSION The linear order of the constituents inside the Javanese DP NP-A-D-Num-RC-Dem-RC is derived from the hierarchical structure Dem > D > Num > Adj > NP . The linear ordering is quite rigid: different positioning associates with different semantics/syntax. The linear order of adjectives is also quite rigid and exhibits a mirror image of universal adjectival hierarchy. The distributions of adjectives and numerals (high adjectives as well as numerals cannot appear between the noun head and D) can be directly captured by adopting Cinque’s phrasal movement proposal (2005) with the restriction of height of roll-up in the adjectival field. Another restriction, that at most 2 adjectives can follow the head noun and precede the D, can be understood as the property of the D clitic –ne and the restrictions on syntactic derivation constraint of NPAA (Complex Filters: Koopman and Szabolcsi (2000), and Koopman (2002). DPs or DemPs can be merged as subjects of Adjectival predicates, and can be promoted to Spec,DemP or Spec,DP which give rises to restrictive and non-restrictive readings respectively Restrictive relatives: [DemP Dem ['DP' DP 'D' [CP DP séng [DP AP]]] [DemP ['DP' DP 'D' [CP DP séng [DP AP]]] t Dem t ] Non-restrictive relatives: ['DP' 'D' [CP DemP séng [DemP AP]]] ['DP' [CP DemP séng [DemP AP]]t 'D' t ]
8
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
References Cinque, G. (2005) Deriving Greenberg's Universal 20 and Its Exceptions, Linguistic Inquiry 36.2. Cinque, G. (2003). The dual source of adjectives and XP – vs. N-Raising in the Romance DP. UCLA class lectures handout. Corver, N. (2006). Getting the (syntactic) measure of Measure Phrases, Davies, W. (1993). Javanese subjects and topic and psych verbs., Linguistics 31, 239-277. Davies, W. and Dresser, C. (2005). The structure of Javanese and Madurese determiner phrases. UCLA Working Papers in Linguistics, 12, Proceedings of AFLA XII, (eds) Heinz & Ntelitheos. Horne, E.C. (1974) Javanese English Dictionary, Yale University Press: New Haven & London. Horne, E. C. (1961) Beginning Javanese, New Heaven and London, Yale University press. Ishizuka, T. (2006). Restrictive and Non-restrictive relative clauses in Japanese: Evidence for movement. Ms., UCLA. Kayne, R. (1994). Antisymmetry of syntax, Cambridge, Mass.: MIT Press. Cambridge, Mass. Koopman, H. and Szabolcsi, A. (2000). Verbal Complexes, MIT Press, Cambridge, Mass. Koopman, H. (2002). Derivations and filters in Dimensions of Movement, in (eds.) Alexiadou and et al. Benjamins. Moro, A. (1997). The raising of predicates, Cambridge University Press, New York. Shlonsky, U. (2004). The form of Semitic noun phrases. Lingua 114.12:1465-1526. Simpson, A. (2005). Classifiers and DP structure in Southeast Asia., Eds. Cinque, G. and Kayne, R. In The oxford handbook of comprative syntax, 806-838; Oxford University Press: Oxford. Sproat, R. and Shih, C. L. (1988). Prenominal adjectival ordering in English, NELS 18, 465-89. Sproat. R & C. Shih (1991). The crosslinguistic distribution of adjective ordering restrictions, In C. Georgopolous & R. Ishihara (eds.) Interdisciplanary approaches to language. Essays in honor of S. Y. Kuroda, Kluwer Academic Publishers, Dordrecht.
9
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
APPENDIX Other seemingly flexible ordering 2 Numerals (bare Num, and in ‘séng RC’) - When a Num appears inside the séng RC, a partitive meaning becomes available. (1) asu-ne Siti séng telu pôdô turu dog-ne Siti REL 3 all sleep (a) Three of Siti’s dogs are all sleeping. (S has more than 3) (b) Siti’s dogs that are THREE are all sleeping.'(S has only 3) (2)
séng
telu pôdô turu REL 3 all sleep ‘Three of them are all sleeping. => Headless relatives only have a partitive reading.
Why do bare ‘numerals’ sometimes precede bare ‘adjectives’? [Num – Adj] séng (3) aku ndeloq asu {putih telu / telu putih} I AT -see dog white 3 /3 white 'I saw three {white dogs / dogs which are WHITE}.' => Because silent ‘séng ’ is present between Num and post-numeral adjectives (in casual speech, séng can be dropped). Why do bare numerals sometimes follow ‘séng RC’?
jumblah- e t
(4) a. aku n-duwé foto-ne siti séng lutju telu I AT -have picture-ne Siti REL funny 3 ‘I have Siti’s funny picture whose amount is three.’ b. aku I
[séng RC – Num]
*iku DEM
n- duwé [foto-ne siti séng lutju]i [(jumblah-e ti) telu] AT-have picture-ne Siti REL funny amount-ne 3 ‘I have Siti’s funny picture whose amount is three.’
Because there is a silent ‘jumblah-e (its amount)’: jumblah-e can be dropped. If the number in (4a) is in the DP spine, we would expect demonstrative to be possible, but it isn’t. The only flexible constituent is prenominal [Num-ng CL] [Num-ng Cl] can also and quite often does appear postnominally (5) rini tuku {[telong kilo] apel / apel [telong kilo]} [Num-ng Cl. – NP] Rini buy 3-ng kilo apple / apple 3-ng kilo ‘Rini bought 3 kilos of apples. Question: Why is the positioning of prenominal [Num-ng CL] flexible? 10
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
The Syntax of Prenominal Numerals [Num-ng CL.] appears to be a surface constituent (dislocation/coordination) (6) a. [gulô putih-e]i taq tuku [telong sendok t ]i st sugar white-ne 1 .TT. buy 3-ng spoon Lit: ‘The white sugar, I bought 3 spoonfuls.’ b. aku butuh [gulô [telong sendok] utôwô [patang sendok]] I need sugar 3-ng spoon or 4-ng spoon ‘I need three spoonfuls or four spoonfuls of sugar.’ [N [#-ng CL] or [#-ng CL]] [Num-ng CL] serves as a clausal predicate: (7) abôt-e beras-ku rong kilo heavy-ne rice-1sg.poss 2-ng kilo ‘The heaviness (weight) of my rice is 2 kilos’ Proposal: - ClassP (i.e., [Num-ng CL]) is a predicate selecting NP as a subject (see Corver 2006). - The linker –ng, classifiers and numbers are distinct functional heads (see Simpson 2005). - The derivation involves linker –ng attracting NumP to its specifier. predicate inversion in Moro’s sense (1997) Structure for ClassP (8) a. telung sendok gulô 3-ng spoon sugar
Derivation: NP (spoon) and NumP (3) form a small clause Linker –ng attracts NumP and they form a phonological unit The subject DP (sugar) and the predicate ClassP (3-ng spoon) forms a small clause Predicate inversion in Moro’s sense (1997) takes place (triggered by a silent linker)
XP 3 ClassPj SC 3 3 NumP ti Class’ DP (ClassPj ) 5 2 5 three -ng SC sugar 2 NP ti 5 spoon
11
AFLA XIV Tomoko ISHIZUKA
May 5, 2007
Justification… the linker –ng and CL being distinct heads: (Coordination) (9) aku butuh [gulô] [[telong sendok] utôwô [papat]] I need sugar 3-ng spoonful or 4 ‘I need three or four spoonfuls of sugar'
Why is the positioning of ClassP flexible? It is a predicate, and is different from other constituents in the DP spine. Predicate inversion allows the order of Noun and ClassP to alternate. Similar to English Copula constructions: [A picture of the wall] was [the cause of the riot]. [The cause of the riot] was [a picture of the wall]. (Moro 1997; 2)
[N [#-ng Cl] or [#]]
Coordination (Kyle Johnson 2004) XP 3 NumP ti 3 5 Class’ ClassP three 5 3 -ng spoon ClassP orP 2 2 ti Class’ or ClassP 2 3 -ng SC NumP Class’ 2 5 2 NP four -ng SC 5 2 spoon NP 5 ATB spoon
12