Plural semantics, reduplication, and numeral modi? cation in Indonesian Mary Dalrymple and Suriel Mofu Centre for Linguistics and Philology University of Oxford Patterns of plural marking and numeral modi? cation in Indonesian provide an interesting testbed for theories of the semantics of numeral classi? ers and plurality. Crosslinguistically, the presence of numeral classi? rs in a language is strongly connected with the absence or optionality of plural marking; this generalisation is the basis of Chierchia’s (1998a, 1998b) Nominal Mapping Parameter, and also accords with established typological generalisations (Greenberg 1972, Aikhenvald 2000, Corbett 2000). In Indonesian, both plural marking as reduplication and classi? ers in numeral modi? cation constructions are optional, and bare (non-reduplicated) Indonesian nouns are best analysed as exhibiting general number (Greenberg 1972, Corbett 2000, Carson 2000), rather than corresponding to the unmarked member of a singular-plural opposition.
Unlike many languages with general number, Indonesian exhibits no masscount distinction: notionally “mass” and notionally “count” nouns do not diyer in their grammatical behaviour, and participate equally in reduplication and numeral modi? cation constructions. We provide an analysis of the semantics of reduplication, classi? ers, and numeral modi? cation in Indonesian which rests on the lack of a mass/count distinction and explains the strong dispreference for numeral modi? cation of reduplicated nouns.
Introduction
Indonesian has what has been analysed as plural marking, realised as reduplication: (1) Pulau-pulau Bali, Lombok dan Sumbawa terletak di sebelah timur island-REDUP and TER. lie at side east pulau Jawa. island Java. ‘The islands of Bali, Lombok, and Sumbawa are located east of Java. ’ (Chung 2000, citing Sneddon 1996) The noun pulau ‘island’ is reduplicated in (1), and refers to more than one island. Unlike plural morphology in English, reduplication is not necessary in referring to more than one individual. Bare nouns can also refer to pluralities; nonreduplicated telur in (2) can be interpreted as plural: (2) Saya merebus telur. I ME. boil egg ‘I am boiling eggs. ’ Indonesian also has numeral classi? ers, though their use is optional. In (3), the classi? er orang is glossed CL: (3) lima (orang) guru 5 CL teacher ‘? ve teachers’ It is often assumed that only count nouns, and not mass nouns, can appear with plural marking and in numeral modi? cation constructions. We show that Indonesian does not have a mass/count distinction: reduplication and numeral modi? cation, with or without classi? ers, are available for notionally “mass” as well as notionally “count” nouns in Indonesian.
These patterns are problematic for previous analyses of Indonesian reduplication and numeral modi? cation (Chung 2000, Carson 2000, Sew 2007), which rely on a mass/count distinction. 2 Numeral classi? ers Indonesian has three numeral classi? ers in common use: orang ‘person’ for people, ekor ‘tail’ for animals, and buah ‘fruit’ for everything else. (4) a. lima orang guru 5 CL teacher ‘? ve teachers’ b. lima ekor sapi 5 CL cow ‘? ve cows’ c. lima buah meja 5 CL table ‘? ve tables’ For all numerals other than se- ‘one’, which is a pre? x attached to the classi? er, the classi? r is optional1 (Sneddon 1996, Chung 2000). Alongside (4), the phrases in (5) are also completely acceptable: (5) a. lima guru 5 teacher ‘? ve teachers’ (2000) provides an analysis of classi? ers in Malay, which she uses as a cover term for the closely related Western Austronesian languages spoken in Indonesia (Bahasa Indonesia, or Indonesian), Malaysia (Bahasa Malaysia), Singapore, and Brunei. All of her informants are speakers of Bahasa Malaysia. She claims that classi? ers are obligatory in numeral modi? cation constructions in Malay; however, examples of numeral modi? cation without classi? rs are plentiful in Indonesian as well as other varieties of Malay, as we show below. 1 Carson 2 b. lima sapi 5 cow ‘? ve cows’ c. lima meja 5 table ‘? ve tables’ Although the classi? er is obligatory with the pre? x numeral se- ‘one’; we believe that this is best treated as a morphological fact and not a fact about the semantics of the classi? er system. The examples in (4) and those in (5) are acceptable in the same contexts, and there is no obvious meaning diyerence between them. Examples (6) and (7) show that use of the classi? er can disambiguate ambiguous nouns: (6) korban ‘sacri? e/oyering’/‘victim’: a. lima orang korban 5 CL (person) victim ‘? ve victims’ b. lima buah korban 5 CL (thing) oyering ‘? ve (religious) oyerings’ (7) madu ‘honey’/‘co-wife’: a. seorang madu 1. CL (person) co-wife ‘one co-wife’ b. sebuah madu 1. CL (thing) honey ‘one (jar/comb/amount of) honey’ A successful analysis of the semantics of numerals and classi? ers must allow for numeral modi? cation either with or without the classi? er, and for the classi? er to contribute some semantic information about the type of individual being counted when it appears. 3 Reduplication
As in many classi? er languages, plural marking is not obligatory in Indonesian: both reduplicated and nonreduplicated nouns can refer to a plurality. There do not seem to be any additional semantic eyects associated with plural marking; reduplicated nouns can be de? nite or inde? nite, and are compatible with both collective and distributive readings. Reduplicated nouns can also appear with numeral modi? ers, but this is rare and dispreferred.
Optionality of reduplication
In example (1), the noun pulau ‘island’ is reduplicated, and refers to more than one island.
However, reduplication is not obligatory for the expression of plurality: bare nouns can also refer to pluralities. In example (2), telur ‘egg’ can be interpreted as plural, though it is not reduplicated. Similarly, nonreduplicated buku in (8) can refer to more than one book: (8) (Apa isi peti itu? ) Buku. what contents box that book ‘(What is in that box? ) Books. ’
Semantic eyects of reduplication
Optional plural marking has been claimed in some languages to have additional semantic eyects such as de? niteness or collectivity. For example, Li (1999) shows that in Chinese, a language in which morphologically unmarked nouns an refer to a plurality, the plural marker -men not only requires reference to more than one individual, but also expresses de? niteness. 2 This is not the case in Indonesian, in which reduplication is compatible with both inde? nite and de? nite readings, and with both collective and distributive readings. The reduplicated noun emas-emas ‘gold’ in example (9) is compatible with the demonstrative determiner, and has a de? nite interpretation: (9) Tidakkah engkau memilih emas-emas itu? not. QUES you MEN. choose gold-REDUP that ‘Don’t you choose those [pieces of] gold? ’ (indonesia. eartnsouls. com/cerita/n/c1372. shtml) In contrast, the reduplicated noun anak-anak ‘children’ in example (10) has an inde? nite interpretation; this is the ? rst mention of children in the text: (10) Saat malam datang, anak-anak bermain sepak bola di tepi when night come, child-REDUP BER. play kick ball in banks sungai Musi ini. river Musi this.
‘When the night came, children played football on the banks of the Musi river. ’ (http://wn. com/Dusk in 10 Ulu Senja di 10 Ulu) Reduplicated nouns are also possible in existential constructions: (11) Kunci-kunci perkara ghaib itu ada lima dan lue-REDUP matter magic that be 5 and ‘There are ? ve clues to that magic matter and ’ http://almuslimah. wordpress. com/2008/03/31/ dukun-dan-tukang-ramal-penciduk-agama-dan-harta/ Both collective and distributive readings are also possible with reduplication. Example (12) exhibits a collective reading with a reduplicated subject: 2 Thanks to an anonymous reviewer for helpful comments on these issues. 4 (12) Siswa-siswa berkumpul dan melihat bagaimana instruktur student-REDUP BER. gather and ME. saw how instructor mendemonstrasikan penggunaan senjata, dengan melakukan MEN. emonstrate. KAN PEN. use. AN weapon, with ME. do. KAN berbagai tindakan pengamatan various act. AN PEN. observe. AN ‘Students gathered and saw how the instructor demonstrated using weapons, by using various observation techniques. ’ (http://cky78. multiply. com/journal/item/33/Metode Demonstrasi) Examples (13-14) also have reduplicated subjects, but exhibit distributive readings: (13) Ada kegiatan snacktime di mana anak-anak membawa snack dari there activity snacktime in which child-REDUP MEN. take snacks from rumah. home. ‘There is a snacktime activity where children take snacks from home. (http://edukasi. kompasiana. com/2010/04/01/hemat-plastik-yuuk/) (14) 135. 000 anak-anak membawa sebuah pistol atau senjata lain ke 135. 000 child-REDUP MEN. take 1. CL pistol or weapon other to sekolah school. ‘135,000 children took a pistol or other weapon to school. ’ (http://abasse. blog. binusian. org/) Note that example (14) involves numeral modi? cation of the reduplicated noun anak-anak ‘children’ with no classi? er. In most contexts, reduplicated nouns are intuitively felt to refer to a relatively large number of instances of the noun.
We found no instances of dua orang-orang/2 orang-orang ‘two person-REDUP’ or dua siswa siswa/2 siswasiswa ‘two students-REDUP’ in a web search, and in fact such phrases are intuitively found to be unacceptable. This is puzzling if reduplication has a semantics like the English plural and simply is used when referring to more than one entity. A full analysis of the semantics of reduplication must also encompass this aspect of its meaning. 3. 3 Reduplicated nouns and numeral modi? cation It is di? cult, though not impossible, for reduplicated nouns to appear with numerals and classi? ers.
Chung (2000) provides one example of this construction, and in a web search, we found several more. Reduplicated nouns can appear with numeral modi? cation with no classi? er: 5 (15) Tim ini terdiri dari empat siswa siswa yakni Eka A? dah, team this TER. self of 4 student REDUP namely Duratu Abdiana R, Meliza Istiana, dan Diah Noianah. and ‘This team consists of four students, namely … ’ (http://www. lamongan. go. id/Report/article excel. cfr? articleid=4858) We have also found examples of reduplicated nouns with a numeral and classi? er: (16) Jasa bantuan pemenuhan kebutuhan sekolah ini service aid ful? ment needs school this ditujukan kepada 100 orang siswa-siswa dari 6 PASSIVE. aim. at. KAN to 100 CL student-REDUP from 6 sekolah dasar di atas yang terletak di wilayah Ciumbuleuit. school elementary at above REL located in region ‘The service for the ful? lment of school needs is aimed at one hundred elementary school students from six elementary schools located in the Ciumbuleuit region. ’ (http:/deathrockstar. info/2005/07/) However, the use of the reduplicated form in numerical modi? cation is uncommon and dispreferred relative to the use of the nonreduplicated form, to the extent that some authors (e. . Carson 2000) have claimed that the reduplicated form is ungrammatical in numeral modi? er constructions. A full analysis of reduplication/plural formation and the semantics of classi? ers should predict that examples of numeral modi? cation of reduplicated nouns are rare, but are sometimes found.
4 The mass vs. count distinction in Indonesian It has been claimed that all languages distinguish mass nouns from count nouns (Doetjes 1997, Chierchia 2010). Given this claim, we might expect that only certain nouns in Indonesian – count nouns, and not mass nouns – could be reduplicated or modi? d by a numeral. However, this is not the case: there are no mass/count distinctions among determiners; notionally “mass” nouns can be reduplicated and modi? ed by a numeral, with or without a classi? er; and both mass and count nouns participate in the same way in mensural constructions. 3 Although other languages besides Indonesian have been claimed to lack a mass/count distinction, it is important to note that this is not always correlated with the presence of classi? ers or the absence of obligatory plural marking. Wilhelm (2008) shows that there is a well-motivated mass/count distinction in D? e Su lin? , which has no plural marking, and Cheng & Sybesma (1999) e ? e show that the classi? er language Chinese has a mass/count distinction, though plural marking is not obligatory. A fully general treatment of plural marking and classi? ers cannot, then, rely on the presence or absence of a mass/count distinction. 3 Carson (2000) also claims that there is no need to make a mass/count distinction in Malay, but on the basis of the incorrect claim that both need classi? ers and cannot combine with numerals directly. 6 4. 1 Determiner distinctions
Chierchia (1998a) lists a number of criteria to distinguish between mass and count nouns, including the use of diyerent determiners. In English, certain determiners are reserved for count nouns: (17) a. a book/*a mud b. each book/*each mud Others must be used with mass nouns: (18) a. little mud/*little book b. much mud/*much book The same is not true in Indonesian; all determiners may be used with all nouns, regardless of whether they are notionally “mass”: (19) a. banyak buku much book ‘many books’ b. banyak lumpur much mud ‘much mud’ (20) a. beberapa buku some book ‘some books’ b. eberapa lumpur some mud ‘some mud’ The naturally-occurring example (21) involves modi? cation of a notionally “mass” noun by the determiner setiap ‘each’, which is also used to modify notionally “count” nouns: (21) Setiap air yang dipompa dari kolam penampungan atau each water REL PASSIVE. pump from pond reservoir or kolam pengendapan masuk ke kolam pembesaran, . . . pond PEN. evaporate. AN enter to pond PEN. big. AN ‘each (amount of) water which is pumped from the reservoir pond or evaporating pond to the main pond . . . ’ (http://www. seafdec. org. ph/pdf/Best Management Practices-Bahasa. pdf) 7 4. 2 Reduplication of “mass” nouns
Reduplication of notionally “mass” nouns like air ‘water’ gives rise not only to the interpretation ‘kinds of water’, but also to the interpretation ‘(speci? c) amounts of water’. Chung (2000) denies that such an interpretation is possible, but we have found a number of naturally-occurring examples illustrating this possibility. Example (22) involves reduplication, and refers to multiple speci? c quantities of water: (22) Mereka telah kemasukan air laut terlalu banyak dan They have KE. enter. KAN sea water excessive many and air-air itu sudah berhasil dikeluarkan. water-REDUP that have successfully PASSIVE. xit. KAN ‘They have ingested too much sea water, and those [amounts of] water have successfully been taken away. ’ (http://www. detiknews. com/index. php/detik. read/tahun/2008/bulan/ 02/tgl/04/time/152231/idnews/888917/idkanal/10) Example (23) refers to multiple quantities of minyak ‘oil’: (23) … minyak-minyak itu muncrat dari manhole kapal dan membeku oil-REDUP that stream from manhole ship and MEN. freeze setelah membentuk seperti sabu dan mengotori pantai sekitar. then MEN. form like bubble and MEN. dirty beach around ‘The [streams of] oil streamed from the manhole of the ship and solidi? d, and then formed bubbles and polluted the beach. ’ (http://www. posmetrobatam. com/index2. php? option=com content&do pdf=1&id=1401) Examples (24–25) refer to multiple quantities of lumpur ‘mud’: (24) Air yang segar menyentuh kulit, membersihkan lumpur-lumpur water REL fresh MEN. touch skin, MEN. clean. KAN mud-REDUP yang menempel.
REL MEN. stick. ‘Fresh water touched (my) skin, and cleaned (away) [pieces of] mud which covered (my body). ’ (http://nekopinku. multiply. com/journal/item/38) (25) Hari-hari berlalu. Lumpur-lumpur menyembur. Saya day-REDUP BER. pass. Mud-REDUP streamed. out. SG tersentak ketika televisi menyiarkan tanggul penghalang TER. shock when television MEN. broadcast wall PEN. block lumpur jebol pada hari ke-74. mud collapsed on day ORDINAL-74 ‘Days passed. [Amounts of] mud streamed out. I was shocked when the TV broadcasted that the wall/dam that blocked the mud collapsed on the 74th day. ’ (http://babang-juwanto. blogspot. com/2010/08/tempo-interaktif-cari-anginlumpur. html) 8 And example (26) refers to multiple quantities of tanah ‘soil’: (26) Aliran sungai biasanya akan menyebabkan kerusakan dalam ? ow. AN river usually will MEN. cause. KAN KE. estroy. AN in bentuk longsoran tanah. Tanah-tanah yang berada di tebing sungai form slide soil. Soil-REDUP REL BER. exist in side river akan tergerus pada bagian tepinya dan masuk ke aliran sungai. will TER. erode in part side. NYA and enter to ? ow river ‘The river stream usually causes destruction in the form of a mud slide. The [amounts of] soil along the river banks will be eroded. ’ (222. 124. 230. 101/konservasi tanah/kerusakan tanah. htm) 4. 3 Numeral modi? cation of “mass” nouns Chierchia (2010) claims that the “signature property” of mass nouns is their inability to be modi? d by numerals; however, numeral modi? cation of notionally “mass” nouns is possible in Indonesian, with the non-animate classi? er buah (‘fruit’) or with no classi? er.
Example (27) involves reference to portions of ice cream, a noun for which conventionalised portions can be identi? ed; such examples are also found in English (“? ve ice creams”, “? ve coyees”). Crucially, examples are also found in which the mass noun does not have conventional portions: example (28) refers to geographical areas of sea water, and example (29) refers to pieces of wood. 27) Berapa lama waktu yang diperlukan oleh 5 anak untuk makan How long time REL PASSIVE. need. KAN by 5 child to eat 5 buah es krim? 5 CL ice cream ‘How long does it take ? ve children to eat ? ve (portions of) ice cream? ’ (http://rumiati. wordpress. com/2007/09/11/teki-teki-matematika) (28) Seandainya masing-masing dari dua air laut tersebut memiliki If each of 2 water sea TER. mention have salinitas (kadar garam) yang berbeda atau temperature yang berbeda, salinity (level salt) REL diyerent or temperature REL diyerent apakah keduanya akan bercampur ketika bertemu di satu QUEST ORD. wo. NYA will BER. mix when BER. meet in one tempat sehingga keadaan awal dari masing-masing (salinitas maupun place so condition early of each-REDUP (salinity or temperature) berubah menjadi satu keadaan salinitas dan temperature) BER. change MEN. become one condition salinity and temperature yang baru ? temperature REL new? ‘If each of two (areas of) sea water mentioned has diyerent salinity or diyerent temperature, will they mix when they meet in one place so that the original salinity and temperature change into a new salinity and temperature? (http://seremonia. net/hello/index. php? option=com content&task=view &id=32&Itemid=33) 9 (29) Untuk satu peleg butuh sekitar sepuluh kayu. for one peleg need about 10 wood ‘For one peleg, ten (pieces of) wood are necessary. ’ (http://jawapos. co. id/radar/index. php? act=detail&rid=62414) 4.
4 Mensural constructions Mensural constructions are equally acceptable with notionally “mass” and notionally “count” nouns. 4 (30) lima kilo apel / mangga / tepung / daging ?ve kilo apple mango ? our meat ‘? ve kilograms of apples / mangoes / ? ur / meat’ Examples of reduplicated nouns in mensural constructions can be found, but as with the cases of numeral modi? cation discussed in Section 3. 3, reduplication is rare and dispreferred in mensural constructions. Examples such as the following are possible in principle, and we present some naturally-occurring examples below: (31) lima kilo apel-apel ? ve kilo apple-REDUP ‘? ve kilograms of apples’ However, although we found many examples of the pattern [numeral] kilo [noun] ‘[numeral] kilos of [noun]’ for the unreduplicated nouns ikan ‘? h’, apel ‘apple’, mangga ‘mango’, keladi ‘taro’, pepaya ‘papaya’, jagung ‘corn’, tepung ‘? our’, and daging ‘meat’, no examples of this pattern were found in which these nouns were reduplicated. There were also many examples of the pattern [numeral] gelas [noun] ‘[numeral] glasses of [noun]’ for the nouns kopi ‘coyee’, jus ‘juice’, and teh ‘tea’, but none in which the nouns were reduplicated. Thus, there does not seem to be a diyerence in availability or acceptability of reduplication of notionally “mass” and notionally “count” nouns in the mensural construction; reduplication is rare and dispreferred in almost all cases.
There are two sorts of exceptions to this generalisation. First, we found some examples with reduplication which exhibit a “kinds of” reading, such as the following examples, in which the nouns beras ‘rice’ and rokok ‘cigarette’ are reduplicated: (32) a. … dengan uang Rp 4100,00 kita sudah dapat membeli 1 with money Rp 4100,00 1PL. INCL already can MEN. buy 1 kilo beras beras berkualitas rendah … kilo rice REDUP BER. quality low ‘… with Rp 4100,00 we could have bought 1 kilogram of low quality rice… ’ (http://kitho-waskitho. blogspot. com/2009 01 01 archive. tml) 4 Thanks to Peter Austin for discussion of Indonesian mensural constructions. 10 b. Mereka membawa berbagai replika bungkus rokok-rokok 3pl MEN. bring various relics? packet cigarette-REDUP brand ternama. brand TER. name ‘They brought various well-known cigarette packets. ’ (http://www. harian-global. com/index. php? option=com content&task=view&id=8198&Itemid=54) The availability of a “kinds of” reading under pluralisation is a common property of mass nouns, but our data show that a “kinds of” reading with reduplication is available for both mass and count nouns.
The other exception is the noun air ‘water’. We found several examples of mensural phrases with the reduplicated noun air-air, including this one: (33) Kolam berisi 2,5 juta liter air air (660,430 gallons) Pond BER. content 2,5 million litre water REDUP (660,430 gallons) dibangun berlorong-lorong seolah goa di dasar laut. PASSIVE. build BER. alley-REDUP as. if cave in base sea ‘A pond containing 2. 5 million litres of water (660,430 gallons) was built like alleys as if (they were) caves at the bottom of the sea. ’ (http://forumbebas. com/printthread. php? id=52821) In this example, the reason for the appearance of the reduplicated form of the noun may have to do with the very large quantity of water involved. Such examples were rare; almost all examples of mensural phrases with air-air appeared in a recipe context, such as the following examples: (34) a. Rebus semua bahan dengan 3 gelas air air sampai Boil all material with 3 glass water REDUP until tersisa 1 gelas. TER. remain 1 glass ‘Boil all the materials with 3 glasses of water until the remaining (water) is one glass (only). ’ (Hariana 2004) b.
Daun segar sebanyak 1 genggam dicuci bersih, rebus leaf fresh amount 1 handful PASSIVE. wash clean, boil dengan 3 gelas air air bersih sampai mendidih. with 3 glass water REDUP clean until MEN. boil ‘A handful of fresh leaves is washed clean, (and then) boil (them) with 3 glasses of clean water until (the water) steams. ’ (http://feed. feedcat. net/loveherbal) We have no immediate explanation for the exceptional behaviour of the reduplicated noun air-air in recipe contexts in the mensural construction, other than to note that the recipe register often diyers from other registers, and that this may play a role here.
In sum, there seems to be no motivation to make a distinction between mass and count nouns in Indonesian, at least on the basis of patterns of determiner use, reduplication, numeral modi? cation, and mensural constructions. This 11 ?nding goes against some previous claims: Sew (2007) explores the count-mass distinction in Malay (by which he means Bahasa Malaysia, the closely-related variety spoken in Malaysia), and proposes a distinction between Malay count and mass nouns on the basis of reduplication patterns. He marks notionally mass” nouns as ungrammatical if they are reduplicated, claiming that air-air ‘water-REDUP’, udara-udara ‘air-REDUP’, and pasir-pasir ‘sand-REDUP’ are not possible words (Sew 2007: 23). We do not agree with Sew that these are impossible words, and indeed we presented above some naturally-occurring examples of reduplicated air-air ‘water-REDUP’ (cf. example 22 and the examples immediately above). We have also found examples of reduplicated udara-udara ‘air-REDUP’ (example 35) and pasir-pasir ‘sand-REDUP’ (example 36, from the on-line newspaper Malaysian Insider, written in Bahasa Malaysia). Thus, we do not believe that Sew’s judgements re? ct a diyerence between Indonesian and Bahasa Malaysia, since examples like these appear to be common in both varieties. (35) Sedangkan sebagai manusia kita seharusnya tidak While-KAN as human 1PL. INCL SE-must-NYA NEG menghirup udara-udara tersebut. MEN-ingest air-REDUP TER-mention ‘But as a human being we should not ingest those [amounts of] air. ’ (http://skyindo. wordpress. com/… / pencemaran-udara-akibat-transportasi-di-jakarta/) (36) Noh (gambar) yang juga Menteri Pertanian dan Industri Asas Noh (picture) REL also Minister Agriculture and Industry Basic Tani mendakwa KSSB turut memberi kemudahan kredit Farmer MEN-accuse KSSB take. art MEN-give KE-easy-AN credit kepada sebuah syarikat dan lima kontraktor untuk mengorek pasir to 1. CL association and ? ve contractor to MEN-dig sand pasir tanpa perlu membayar deposit. REDUP without need MEN-pay deposit.
‘Noh (pictured) who is the Minister of Agriculture and Farm Industry accused KSSB of involvement in giving easy credit to an association and ? ve contractors to mine [amounts of] sand without paying any deposit. ’ (http://www. themalaysianinsider. om/bahasa/article/ noh-wujud-salah-guna-kuasa-pembelian-pasir-kumpulan-semesta/) Sew also claims that unreduplicated mass nouns as well as reduplicated mass and count nouns cannot appear in numeral modi? cation constructions (discussed in Section 3. 3) and cannot be modi? ed by certain determiners such as beberapa ‘some/a few’ (discussed in Section 4. 1); on this basis, he proposes to analyse reduplicated nouns as mass nouns. However, we have provided many examples showing that both notionally “mass” and notionally “count” unreduplicated nouns appear with these determiners and in numeral modi? ation constructions. We agree with Sew that reduplicated nouns are rare in these constructions, but this is equally true for notionally “mass” and notionally “count” nouns; thus, data from these constructions does not motivate a distinction between mass and 12 count nouns in Indonesian. To summarise the previous sections, a complete analysis of reduplication and numeral modi? cation in Indonesian must account for the following: • Classi? ers are optional in numeral modi? cation with all nouns. • Reduplication is optional in the expression of plurality. The use of reduplicated nouns with numerals (with or without classi? ers) is possible but dispreferred.
• There is no mass/count distinction. 5 5. 1 Classi? ers, numeral modi? cation, and general number General number and Greenberg’s generalisation In many languages, numeral classi? ers are required in numeral modi? cation. Japanese is an oft-cited example of a language with obligatory classi? ers: (37) enpitsu ni-hon pencil 2-CL ‘two pencils’ (Japanese: Matsumoto 1993) The numeral cannot modify the noun directly in Japanese, and a numeral classi? r is required to appear. In fact, the Indonesian pattern, with optional rather than obligatory classi? ers in numeral modi? cation, is crosslinguistically unusual. Like many other classi? er languages, plural marking in Japanese is not required when referring to more than one individual: (38) Otokonoko-ga asonde-iru boy-NOM play-PROG ‘A boy is playing. ’/‘Boys are playing. ’ (Japanese: Nakanishi & Tomioka 2004) Greenberg (1972) presents a foundational typological survey of numeral classi? ers and the morphology of number in classi? r languages like Japanese, and draws an important descriptive generalisation: “Numeral classi? er languages generally do not have compulsory expression of nominal plurality” (Greenberg 1972: 177). 5 Greenberg bases this generalisation in part on the work of Sanches (1973), who makes similar claims; more recent work by Aikhenvald (2000) also supports this generalisation. As Greenberg (1972) states: 5 It should be noted that the term “classi? er” has been used for a wide variety of other ways of indicating the classi? cation of nouns.
In a comprehensive typological survey, Aikhenvald (2000) describes a variety of linguistic devices for indicating noun category which have been referred to as classi? ers: for example, some languages have noun classi? ers, which appear with nouns even when they are not modi? ed by a numeral; a possessed classi? er can mark noun class agreement on a possessor with a possessed noun; and verbal classi? ers appear on the verb and show noun class agreement with one of the core arguments of the verb. Aikhenvald lists a number of other devices which have been referred to as classi? rs. This similarity of terminology for diyerent classi? er types has been a source of confusion in the literature; Greenberg’s generalisation does not draw a connection between the morphological expression of plurality and non-numeral classi? ers. 13 A considerable number of classi? er languages … have what are generally described as plural a? xes. However, closer examination seems to show that in almost every instance the “unmarked” singular is in fact a form which, like the collective in languages with a compulsory plural, is non-committal in regard to number. …
What is hypothesised, then, is that in the usual classi? er language (i. e. without in? ection for number), classi? able nouns in their isolated form, that is when not accompanied by a classi? er or a plural marker, are like collectives in their semantic non-speci? cation of number and in their avoidance of a direct number construction. (Greenberg 1972: 183– 184) Greenberg (1972) and Sanches (1973) use the term general number for the unmarked form of nouns in languages with nonobligatory plural morphology (see also Corbett 2000). General number is common in numeral classi? r languages (see, for example, Nakanishi & Tomioka 2004 for Japanese), but is also found in languages with no classi? ers (see Wilhelm 2008 for D? ne Su lin? ). Care ? e son (2000) and Sato (2008) argue convincingly that Indonesian nonreduplicated nouns exhibit general number, and can refer either to single entities (atoms) or to pluralities (sums). Carson (2000) provides example (39) in support of this view, and against the view that morphologically unmarked nouns are ambiguous and refer either to a single individual (on one reading) or to multiple individuals (on the other reading).
Although in (39) the same property is required to hold of both John and Bill, it may be that John saw one horse and Bill saw several, or vice versa; if the unreduplicated noun kuda ‘horse’ were ambiguous and expressed either a singular meaning or a plural meaning on each occasion of its use, this would not be predicted. (39) John melihat kuda dan Bill juga John see horse and Bill also ‘John saw a horse/horses and Bill did too [saw a horse/horses]. ’ (Carson 2000: 46–47) Similar arguments for the analysis of morphologically unmarked nouns as having general number are given by Rullmann & You (2003) for Chinese and by Wilhelm (2008) for D? e Sulin?. e ? e This view ? ts well with Corbett’s (2000) claim that the function of number marking in languages with general number is to provide additional speci? cation of the number of entities that are referred to. In support of this view, Corbett (2000: 10–11) cites the following data from Bayso (Cushitic), showing that while the morphologically unmarked form has general number and can refer to one or more lions, additional marking can be added to restrict reference to one, a few, or a large number of lions:6 (40) a. l? ban u foofe lion. GENERAL watched. SG ‘I watched (one or more) lion(s). ’ Corbett (2000: 10–11) states that the unmarked form in (40a) is “non-committal as to number” and so is an instance of general number, Corbett & Hayward (1987: 5) claim that this unmarked form, which they refer to as the “unit reference form”, is used “to denote either an individual member or the class of the referent”. 6 Although 14 b. l? ban-titi u foofe lion-SINGULAR watched. 1SG ‘I watched a lion. ’ c. l? ban-jaa u foofe lion-PAUCAL watched. 1SG ‘I watched a few lions. ’ d. l? ban-jool u foofe lion-PLURAL watched. 1SG ‘I watched (a lot of) lions. Pretheoretically, the semantics of number marking for general number nouns involves reference to, or individuation of, a certain number of parts of the denotation of the morphologically unmarked noun. 5. 2 Theories of plural semantics and numeral modi? cation The observation that plural marking tends to be obligatory in non-classi? er languages (like English) but not in classi? er languages (like Japanese) underpins Chierchia’s (1998a, 1998b) in? uential theory of the semantics of bare nominals, the Nominal Mapping Parameter. The Nominal Mapping Parameter establishes a connection between plural morphology, numeral classi? rs, and the ability for bare nouns to appear in argument position. It claims that in languages like Japanese and Chinese, all nouns are mass nouns, plural morphology is generally absent, classi? ers are required with numerals, and bare nouns can appear as arguments to predicates. In languages like English, in contrast, nouns are either mass or count, count nouns are marked as either singular or plural, numerals can appear without classi? ers, and bare mass nouns and bare plural count nouns, but not bare singular count nouns, can appear as arguments to predicates.
Chierchia’s theory makes interesting predictions about correlations between morphology and syntax, and these seem to accord with established typological generalisations. Chung (2000) observes that the fact that Indonesian has classi? ers and also allows plural formation via reduplication seems to run counter to the Sanches/Greenberg/Aikhenvald generalisation about numeral classi? er languages and plural expression, and also seems to present a problem for Chierchia’s Nominal Mapping Parameter. In particular, she addresses the connection in Chierchia’s theory between the presence of classi? rs and the absence of plural morphology, claiming that Indonesian has classi? ers but also expresses plurality as reduplication. If the connection between the presence of classi? ers and the absence of plural morphology does not hold up, Chierchia’s transparent connection between morphological expression and semantic interpretation cannot be maintained in a simple manner. We believe Chung’s analysis to be problematic in two respects. First, the optionality of classi? ers in Indonesian is surprising in light of Chung’s claims about Indonesian as a typical classi? r language. Chung points out that the classi? er was more frequent in earlier stages of Indonesian, but this does not mitigate the fact that the classi? er is optional in contemporary Indonesian. It should be noted, of course, that this pattern is also problematic for Chierchia’s 15 claims about the strict correlation between language type and the requirement for classi? ers to be used in numeral modi? cation. A full analysis of Indonesian numeral modi? cation must account for the fact that numerals can combine with nouns either directly or in combination with a classi? er.
Second, Chung’s criticism of the Nominal Mapping Parameter goes through only under the assumption that Indonesian reduplication has exactly the same semantics as plural formation in a language like English. In fact, we believe that this is not the case: the optionality of reduplication to refer to pluralities is problematic for Chung’s characterisation of Indonesian as a numeral classi? er language with plural morphology. Indonesian bare nouns are best classi? ed as exhibiting general number rather than being the unmarked, singular pair in a singular/obligatory plural opposition.
The equal acceptability of reduplication with notionally “mass” and notionally “count” nouns also distinguishes Indonesian reduplication from English plural marking. In the following, we summarise the Nominal Mapping Parameter and some subsequent proposals which aim to provide a more crosslinguistically adequate account of the semantics of plural marking. We then propose a semantics for Indonesian reduplication and numeral modi? cation which builds on this and related work on plurality and numeral modi? cation in languages with classi? ers and general number. 5. 2. 1 Chierchia 1998a,b: Classi? ers and the mass/count distinction
Chierchia (1998a,b) analyses singular count nouns in English as denoting sets of atoms, and pluralisation (Chierchia’s PL) as an operation that produces the denotation of a plural noun by mapping a set of atoms into the set of pluralities constituted by those atoms. (41) PL(A) = *A ? A where *A is the closure of A under ? (the set of all sums of elements of A) (Chierchia 1998a: 59–60) The eyect of PL on a set of elements A (corresponding to the denotation of a singular noun) is to form all sums of the elements of A, and then to remove A, so that the denotation of the plural noun does not include the denotation of the singular noun.
In (42), the denotation of singular table is the set of individual tables {table1, table2, table3}. The denotation of plural tables is obtained via the operation PL, which applies to the denotation of singular table to give the set of sets containing more than one table: (42) table1 {table1,table2,table3} tables = PL(table) ? table {table1,table2} {table1,table3} table2 {table2,table3} table3 In contrast to count nouns like table, mass nouns like furniture neutralise the distinction between singular and plural – Chierchia analyses them as, in a sense, both singular and plural: 6 {table1,table2,table3} (43) {table1,table2} table1 {table1,table3} table2 {table2,table3} table3 ? ? ? ? ? ? ? ? ? ? ? furniture Chierchia’s PL does not apply in a sensible way to mass nouns, because there is no distinguishable set of individuals analogous to the denotation of singular table for PL to operate on. This means that mass nouns cannot be pluralised via PL. To capture the generalisation that nouns in numeral classi? er languages tend to resist pluralisation, Chierchia proposes that all nouns in numeral classi? r languages are mass nouns. This means that pluralisation via PL is not possible in such languages, and that numeral classi? ers are required for numeral modi? cation, since the classi? er is necessary to individuate relevant parts of the mass noun to enable counting.
This proposal seems to ? t reasonably well with the Indonesian data we examined above, since Indonesian has a classi? er system, and does not have a mass/count distinction. However, it is not appropriate for many other classi? er languages, including Chinese and D? ne Sulin? since these languages do e ? e distinguish mass and count nouns. 5. 2. 2 PL and general number Rullmann & You (2003) and Wilhelm (2008) recast Chierchia’s (1998a, 1998b) proposal for the semantics of plurality with reference to Greenberg’s general number. Their proposals allow a treatment of unmarked nouns in Chinese and D? ne Sulin? as similar to English mass nouns like furniture, but without dise ? e carding the semantic mass/count distinction that is needed in these languages. Rullmann & You (2003) propose an alternative de? ition of PL for general number nouns (see also Link 1983, Carson 2000, Wilhelm 2008): (44) Pluralisation (Rullmann & You 2003: 5): PL(N) = *N ? At where *N is the closure under union of N and At is the set of atoms. This de? nition diyers from that of Chierchia (1998a,b) in that the set of atoms is subtracted from the result of forming all sums of the elements of N, rather than the set N itself. Rullmann & You (2003) provide the following diagram as an illustration of the diyerence between Chierchia’s PL and theirs: (45) Rullmann & You’s PL: {table1,table2,table3}
PL(table) table table3 {table1,table2} table1 {table1,table3} table2 {table2,table3} This de? nition works for Rullmann & You’s analysis of Chinese, since the operation of pluralisation can appeal to a well-de? ned set of atoms in the denotation 17 of the nouns to which it applies. But it is still not what we want for Indonesian, which does not have a mass/count distinction; we have seen that it is possible to reduplicate mass nouns like air ‘water’, which do not have distinguishable atoms in their denotation. Instead, our analysis requires that numeral modi? ers, numerals + classi? rs, and reduplication in Indonesian individuate parts of the denotation of the noun as well as specifying the number of individuals involved. 5. 2. 3 Inclusive and exclusive plural readings Another wrinkle in the interpretation of the Indonesian plural relates to the distinction between what Farkas & de Swart (2010) call the “inclusive plural” vs. the “exclusive plural” sense of plural nouns. Building on work by Krifka (1989), among others, Farkas & de Swart explore the “inclusive plural” reading of plural nouns in English and Hungarian in certain contexts, in which the denotation of a plural noun ranges over atoms as well as sums.
They present the examples in (46) to illustrate this pattern: (46) a. Have you ever seen horses in this meadow? b. If you have ever seen horses in this meadow, you should call us. c. Sam has never seen horses in this meadow. (Farkas & de Swart 2010: 3) The response to (46a) is “yes” if the addressee has seen even one horse; in (46b) the addressee is expected to call even if only one horse is seen; and in (46c) Sam is claimed never to have seen even one horse.
The analyses of plurals proposed by Chierchia (1998a,b) and Rullmann & You (2003) do not allow for the inclusive reading of plural nouns, since Chierchia’s (1998a, 1998b) PL operator produces a plural noun whose denotation does not include individuals, and Rullman & You’s (2003) PL operator removes the set of atoms from the denotation of the plural noun. Chierchia (2010) acknowledges that these data present a problem for his earlier proposal for PL, and proposes the following de? nition of pluralisation, which assumes that plurals include atoms in their denotation: (47) Pluralisation (Chierchia 2010: 114): PL(N) = ? . ?Q. Q? N ? x = ? Q {table1,table2,table3} PL(table) {table1,table2} table1 {table1,table3} table2 {table2,table3} table3 table This proposal and related proposals (Sauerland et al. 2005, Farkas & de Swart 2010, Spector 2007) require an additional principle to ensure that plural nouns do not refer to atoms in examples such as “We saw horses in the meadow”, which requires that more than one horse was seen. We return to a discussion of inclusive/exclusive plural readings and the semantics of plurality in Section 5. 4 below. 18
Somewhat surprisingly, Indonesian reduplicated nouns also exhibit an inclusive plural reading in these contexts. This is illustrated in example (48), where anak-anak ‘children’ is in the scope of negation, and the denotation of anak-anak includes single children: (48) Kami tidak melihat anak-anak di sana. Kami hanya melihat we NEG ME. see child-REDUP in there. We just ME. see ribuan dan ribuan orang muda, laki-laki dan wanita dari thousand and thousand people young, man-REDUP and woman from banyak warga negara. many citizen country. ‘We did not see children there.
We just saw thousands and thousands of young people, men and women from many countries. ’ (http://lucky77khoe. wordpress. com/… /pewahyuan-di-neraka-06/) This example means that no children were seen, not even one; it does not describe a situation where no more than one child was seen. Inclusive plural readings can also be found in questions; examples (49-50) can be answered ‘yes’ if the addressee has one relevant child, and (51) can be answered ‘yes’ if the addressee has bought one book: (49) Apakah anda memiliki anak-anak yang bermasalah?
Quest 2SG MEN-have child-REDUP which BER-problem ‘Do you have children who have problems? ’ (http://anaksuper. schools. o? celive. com) (50) Apakah Anda memiliki anak-anak yang tinggal di rumah Quest-KAH 2SG MEN-have child-REDUP which live in house Anda? 2SG ‘Do you have children who live in your house? ’ (http://solusibangunanmu. com/index-mod-info-opt-vwInfo-id-116. html) (51) Apakah anda telah membeli buku-buku islam untuk memahami Quest 2SG have MEN-buy book-REDUP Islam to understand islam? Islam ‘Have you bought Islamic books to understand Islam? ’ (http://blog. aldohas. om/muhasabah-harian-setiap-muslim/) Inclusive plural readings are also found in downward-entailing environments such as the restrictor of a universal quanti? er like setiap “each”; the requirement to prepare a party holds even for parents with one child in (52), and example (53) refers to people who buy at least one precious stone: (52) Setiap orang tua yang mempunyai anak-anak usia sekolah Each person old REL MEN-have child-REDUP age school kadangkala juga mempersiapkan sebuah pesta ulang tahun untuk often also MEN-prepare 1. CL party birthday for anak-anak mereka, … hild-REDUP 3PL 19 ‘Each parent who has school-age children has to prepare a birthday party for their children, … (http://http://kontesseopertama. blogspot. com/… /meriahkan-pesta-ulangtahun-bersama. html) (53) Setiap orang yang membeli batu-batu permata, maka Each person REL
MEN-buy stone-REDUP jewelry, then orang-orang Banjarlah yang menjadi juru person-REDUP Banjar-FOC REL MEN-become expert taksirnya. interpret-3SG. POSS ‘Every person who buys precious stones, the Banjar people who will be the interpreter of the price. ’ (http://bubuhan-bubuhan. blogspot. om/) 5. 3 Numerals and classi? ers To set the stage for our analysis of Indonesian reduplicated nouns, we ? rst propose a semantics for Indonesian numeral modi? ers and classi? ers, following Wilhelm’s (2008) proposals for the semantics of numerals and classi? ers in English and D? ne Su lin?. e ? e Wilhelm (2008) proposes that numerals in diyerent languages can make different semantic contributions depending on the presence or absence of classi? ers in the language. Wilhelm gives the following semantics for numerals in D? ne e Su lin? , which does not have classi? rs, and English (Wilhelm 2008: 55; see also ? e Krifka 1995): (54) English ‘three’, D? ne Su lin? taghe ‘three’: e ? e ? P.? x. [P (x) ? OU (x) = 3] a function from a set P of atoms and sums onto that subset of P containing the sums of three object units/atoms Wilhelm’s analysis, following a suggestion by Krifka (1995), assumes that numerals incorporate a classi? er OU which counts the number of “object units” or atoms contained in a plurality. On her view, OU does not create atoms, but instead accesses the atoms or minimal units in the noun’s denotation.
This analysis is similar to the proposal by Cheng & Sybesma (1999) for the treatment of Chinese count-classi? ers for count nouns, which “name the unit in which the entity denoted by the noun naturally occurs” as opposed to what they call massi? ers, which create a unit of measure. This works well for Wilhelm’s analysis of D? ne Sulin? , since in that language e ? e there is a clear distinction between mass and count nouns, demonstrated by the fact that numerals combine only with count nouns (which have accessible atoms in their denotation) and cannot appear with mass nouns.
However, it is not adequate for Indonesian, since Indonesian does not exhibit a mass/count distinction. We propose the following semantics for Indonesian noun phrases in which a noun is modi? ed by a numeral and by a numeral + classi? er:7 7 Carson’s (2000) proposal for the semantics of the numeral + classi? er + noun combination is broadly similar to (55), but she proposes a diyerent compositional semantics, based on her claim that the classi? er is obligatory in numeral modi? cation. As we have seen, this claim does not match the Indonesian data we have observed. 0 (55) Numeral modi? cation: lima meja ‘? ve tables’: ? x. [table(x) ? CL(x) = 5] the pluralities constituted by tables which comprise 5 “portions” of table, where “table portions” (individual tables) are speci? ed conventionally or, if no conventional speci? cation is available, contextually by the generic classi? er CL (56) Numeral + classi? er modi? cation: lima buah meja ‘? ve tables’: ? x. [table(x) ? buah(x) = 5] the pluralities constituted by tables which comprise 5 “buah” of table, as speci? ed conventionally or, if no conventional speci? ation is available, contextually by the classi? er buah Unlike Wilhelm’s OU, which counts the number of atoms in a plurality, CL in the examples above acts as a massi? er, individuating portions of the denotation of the noun, creating rather than accessing atoms. This view is compatible with Chierchia’s (1998b: 347) view of numeral classi? er and measure phrases as serving to “map mass noun denotations into sets of atoms”, and is similar in spirit to Borer’s (2005) characterisation of cardinal numerals in some languages as both “counters” and “dividers”.
This formulation privileges conventionally speci? ed individuation of the noun denotation over contextual speci? cation. 8 We must allow for the numeral or numeral + classi? er to be interpreted relative to contextually speci? ed portions in examples like (28), repeated here, since there are no conventional ways of dividing sea water into portions that can play a role in the interpretation of this example:9 (57) Seandainya masing-masing dari dua air laut tersebut memiliki If each of 2 water sea TER. mention have salinitas (kadar garam) yang berbeda atau temperature yang berbeda salinity level salt) REL diyerent or temperature REL diyerent ‘If each of two (areas of) sea water mentioned has diyerent salinity or diyerent temperature … ’ However, when conventional portions exist, the numeral or numeral + classi? er construction must be interpreted as referring to them, rather than to ad-hoc, contextually speci? ed portions. An example like (58a) means ‘? ve (individual) apples’; it cannot mean ‘? ve groups/kilos/piles/bags of apples’, even if this is strongly suggested by the context. Similarly, example (58b) means ‘? e teachers’, not ‘? ve groups of teachers’: (58) a. lima buah apel / lima apel ? ve CL apple ? ve apple ‘? ve apples’ (not ‘? ve groups/kilos/piles/bags of apples’) are grateful to Peter Svenonius for discussion of this issue. that the contextual speci? cation of relevant units involved in numeral modi? cation is diyerent from the “granularity” parameter which Champollion (2010) shows to be relevant for the interpretation of mass nouns, though granularity may also be speci? ed contextually: although example (57) involves a contextually speci? d division of sea water into two portions, sea water has a much smaller granular structure, since a very small portion of sea water still counts as sea water. 9 Note 8 We 21 b. lima orang guru / lima guru ? ve CL teacher ? ve teacher ‘? ve teachers’ (not ‘? ve groups of teachers’)
Our analysis depends on the following semantic contributions for numerals and classi? ers:10 (59) lima ‘? ve’: ? C.? P.? x. [P (x) ? C(x) = 5] optionally: ? x. CL(x) [where CL is the generic classi? er] (60) buah (classi? er): ? x. buah(x) Indonesian numerals combine with a classi? r meaning and a noun meaning, and also optionally contribute a generic classi? er meaning CL to individuate the denotation of the noun. That is, unlike D? ne Su lin? , English, or Japanese, e ? e a default classi? er may be contributed by the numeral itself: if no overt classi? er is speci? ed, the default generic classi? er contribution is used. This is similar to Champollion’s (2010) unpronounced [MANY] contribution in numeral modi? cation constructions, which combines with the numeral to produce a cardinality requirement, though we do not assume that the optional classi? r contribution occupies a position in the phrase structure tree. Wilhelm (2008) observes that the assumption that numerals in English and D? ne Sulin? include a classi? er contribution correctly predicts that numerals in e ? e these languages can be used pronominally, since numerals in these languages do not denote just a number, but a particular amount of some substance. In Chinese, the numeral + classi? er combination can be used pronominally, but numerals alone cannot. In Indonesian, either numerals without a classi? er or the numeral + classi? r combination can be used pronominally. This is as expected if Indonesian numerals contain a optional default classi? er. (61) Saya meminjam sepuluh buku dan mengembalikan… I MEN. borrow. KAN ten book and MEN. return. KAN ‘I borrowed ten books and returned… ’ a. … satu buah 1 CL b. … satu 1 c. … dua buah 2 CL d. … dua 2 It is also possible to use numbers predicatively in Indonesian, as shown in example (11), repeated in (62). This is unsurprising given their relatively rich semantics:
10 Separate obligatory and optional semantic contributions of numerals are easily speci? ble in the “glue” semantic approach described in Dalrymple (1999, 2001), though we do not provide an explicitly worked out treatment here. 22 (62) Kunci-kunci perkara ghaib itu ada lima dan … clue-REDUP matter magic that be 5 and … ‘There are ? ve clues to that magic matter and… ’ http://almuslimah. wordpress. com/2008/03/31/ dukun-dan-tukang-ramal-penciduk-agama-dan-harta/ 5. 4 Reduplication, mass nouns, and inclusive/exclusive plurality We now turn to the semantics of reduplicated nouns. We treat reduplication as a massi? er, individuating portions of the denotation of the nonreduplicated noun in a way similar to the classi? r: reduplicated air-air ‘water-REDUP’ refers to individual amounts of water, just as reduplicated buku-buku ‘book-REDUP’ refers to individual books. Our semantics for reduplication must also account for the “inclusive” plural readings of reduplicated nouns in certain contexts, allowing reduplicated nouns to have atomic reference as well as sum reference. Finally, we must account for the fact that reduplicated nouns in upward-entailing contexts generally refer to a relatively large number of individuals, where the relevant amount is ontextually determined (somewhat like “several” or “many” in English). Many previous accounts of plural semantics have assumed that the exclusive reading of plural nouns arises in competition with the meaning of singular nouns. Sauerland et al. (2005) and Spector (2007) assume that plural nouns can refer to individuals as well as sums, and that a principle such as Maximise Presupposition applies to strengthen the interpretation of a plural noun to exclude atomic reference in many contexts, since if atomic reference had been intended, the singular form would have been used.
Farkas & de Swart (2010) also provide an account based on competition, which attempts to correlate morphological markedness with semantic markedness; they present a set of OT constraints which have the eyect that nominals with plural morphology must include sums in their denotation, while the interpretation of an unmarked form in competition with a marked plural form is atomic reference. A serious problem that the Indonesian data present for these approaches is that the unmarked, unreduplicated Indonesian noun has general number, and does not compete with the reduplicated noun in the same way as in English.
That is, the fact that the reduplicated noun cannot refer to individuals is not functionally motivated by competition with an unmarked form with exclusively singular reference. Farkas & de Swart’s claim that morphological markedness is correlated with semantic markedness, and their reliance on a competition-based view of the interpretation of marked, plural nouns, make it particularly di? ult to explain why both unmarked and marked/reduplicated Indonesian nouns can have sum reference, while the morphological contrast between English marked and unmarked nouns gives rise to a singular/plural distinction; that is, the morphological basis of their account makes it di? cult for them to distinguish unmarked singular nouns from nouns with general number. We agree with these authors that a principle is needed to resolve the competition between inclusive and exclusive senses of plural nouns, even in Indonesian.
We propose the following disjunctive semantics for reduplicated nouns in Indonesian, and follow Farkas & de Swart (2010) in appealing to the Strongest Meaning Hypothesis (Dalrymple et al. 1998) in selecting the appropriate reading for a reduplicated noun in in a given context: 23 (63) Reduplication: meja-meja ‘tables’: ? x. [table(x) ? CL(x) = N (? N is relatively large)] the pluralities constituted by tables which comprise N “portions” of table, where “table portions” (individual tables) are speci? d conventionally or, if no conventional speci? cation is available, contextually by the generic classi? er CL. Optionally, N is speci? ed as a relatively large number in the given context. As above, CL is a massi? er, individuating portions of the denotation of the noun. Reduplicated nouns refer to a relatively large number of individuals, where the relevant amount is contextually determined; our semantics for reduplication optionally includes this aspect of its meaning. This optionality corresponds to the inclusive vs. xclusive sense of the plural: with the optional component, the denotation of the reduplicated noun ranges only over (relatively large) sums, while without the optional component, there is no requirement for N to be large or, indeed, greater than one. Exactly as in Farkas & de Swart’s approach, inclusion of the optional meaning component N is relatively large produces a stronger meaning in upward-entailing contexts, since the meaning with the speci? cation that N must be large asymmetrically entails the meaning without this speci? ation.
According to the Strongest Meaning Hypothesis, the optional meaning component is included to strengthen the meaning in upward-entailing contexts. In downward-entailing contexts, the meaning without the optional component (corresponding to the inclusive plural sense) is stronger, and is chosen according to the Strongest Meaning Hypothesis in these contexts. 5. 5 Reduplication and numeral modi? cation The fact that reduplicated nouns are less acceptable with numerals and classi? ers, discussed in Section 3. , follows from a compositional semantic treatment: examples in which a reduplicated noun is modi? ed by a numeral involve redundant individuation of the noun’s denotation. An example such as lima buah meja-meja ‘? ve table-REDUP’ would have the following compositional meaning: (64) ? x. [table(x) ? CL(x) = N (? N is relatively large) ? buah(x) = 5] Reduplication contributes the information that the number of “table portions” is some (relatively large) N, while the numeral and classi? er contribute the information that the number of “buah” of table is independently speci? ble as ? ve. This involves individuation of the same “table substance” by two diyerent means, CL and buah, and we believe that this is the source of the perceived unacceptability, and the rarity, of such examples. 6 Conclusion Indonesian provides evidence that numerals, classi? ers, and plural morphology in diyerent languages can have diyerent semantics: the semantics of plural morphology in English is quite diyerent from the semantics of reduplication in Indonesian, and the semantics of numerals in a language in which classi? rs are required is diyerent from the semantics of numerals in a language like Indonesian, where classi? ers are optional. Even given a cross-linguistically relatively 24 uniform semantics for noun phrases in which a noun is modi? ed by a numeral, the semantics of nouns, number marking, and numeral modi? cation must be examined carefully in each language to determine their particular individual contributions. Authors’ address Mary Dalrymple and Suriel Mofu Centre for Linguistics and Philology Walton Street Oxford OX1 2HG UK mary. dalrymple@ling-phil. ox. ac. uk suriel. mofu@ling-phil. ox. ac. uk
Acknowledgments For very useful comments and discussion, we thank I Wayan Arka, Mark Donohue, Manfred Krifka, and the audience at AFLA XV, Sydney, 2008; Peter Austin, Peter Svenonius, and the audience at LAGB09; audiences at the Philological Society, January 2010, and the Cambridge University Linguistic Society, March 2011; Vera Hohaus, Sarah Murray, and the audience at SULA-bar, Manchester, 2011; Gennaro Chierchia, Eytan Zweig, and the audience at the workshop “Current Issues in Semantics and Pragmatics”, York, June 2011; and two anonymous reviewers for the Journal of Semantics.
References Aikhenvald, Alexandra Y. 000. Classi? ers: A Typology of Noun Categorization Devices. Oxford: Oxford University Press. Borer, Hagit. 2005. Structuring Sense, Volume I: In Name Only. Oxford: Oxford University Press. Carson, Jana C? cile. 2000. The Semantics of Number in Malay Noun Phrases. e Master’s thesis, University of Calgary. Champollion, Lucas. 2010. Distributivity as a Bridge Between Aspect and Measurement. Ph. D. thesis, University of Pennsylvania. Cheng, Lisa Lai-Shen & Rint Sybesma. 1999. Bare and not-so-bare nouns and the structure of NP. Linguistic Inquiry 30(4), pp. 509–542. Chierchia, Gennaro. 1998a.
Plurality of mass nouns and the notion of “semantic parameter”. In Susan Rothstein (editor), Events and Grammar , pp. 53–103. Dordrecht: Kluwer Academic Publishers. Chierchia, Gennaro. 1998b. Reference to kinds across languages. Natural Language Semantics 6, pp. 339–405. Chierchia, Gennaro. 2010. Mass nouns, vagueness and semantic variation. Synthese 174, pp. 99–149. 25 Chung, Sandra. 2000. On reference to kinds in Indonesian. Natural Language Semantics 8, pp. 157–171. Corbett, Greville G. 2000. Number . Cambridge, UK: Cambridge University Press. Corbett, Greville G. & R. J. Hayward. 1987. Gender and number in Bayso.
Lingua 73(1–2), pp. 1–28. Dalrymple, Mary (editor). 1999. Semantics and Syntax in Lexical Functional Grammar: The Resource Logic Approach. Cambridge, MA: The MIT Press. Dalrymple, Mary. 2001. Lexical Functional Grammar , volume 34 of Syntax and Semantics. New York, NY: Academic Press. Dalrymple, Mary, Makoto Kanazawa, Yookyung Kim, Sam A. Mchombo, & Stanley Peters. 1998. Reciprocal expressions and the concept of reciprocity. Linguistics and Philosophy 21(2), pp. 159–210. Doetjes, Jenny Sandra. 1997. Quanti? ers and Selection: On the Distribution of Quantifying Expressions in French, Dutch, and English.
Ph. D. thesis, Leiden University. Farkas, Donka & Henri? tte E. de Swart. 2010. The semantics and pragmatics e of plurals. Semantics and Pragmatics 3, pp. 1–54. Greenberg, Joseph H. 1972. Numeral classi? ers and substantival number: Problems in the genesis of a linguistic type. In Working Papers on Language Universals, volume 9, pp. 1–39. Reprinted in Greenberg et al. (1990). Greenberg, Joseph H. , Keith Denning, & Suzanne Kemmer. 1990. On Language: Selected Essays of Joseph H. Greenberg. Stanford, CA: Stanford University Press. Hariana, H. Arief. 2004. Tumbuhan Obat dan Khasiatnya.
Niaga Swadaya. Krifka, Manfred. 1989. Nominal reference, temporal constitution and quanti? cation in event semantics. In R. Bartsch, J. van Benthem, & P. van EmdeBoas (editors), Semantics and Contextual Expressions, pp. 75–115. Dordrecht: Foris Publications. Krifka, Manfred. 1995. Common nouns: A contrastive analysis of Chinese and English. In Gregory N. Carlson & Francis Jeyry Pelletier (editors), The Generic Book . Chicago: University of Chicago Press. Li, Yen-Hui Audrey. 1999. Plurality in a classi? er language. Journal of East Asian Linguistics 8, pp. 75–99. Link, Godehard. 1983.
The logical analysis of plurals and mass terms: A latticetheoretical approach. In Rainer Bauerle, Christoph Schwarze, & Arnim von Stechow (editors), Meaning, Use, and Interpretation of Language. Berlin: de Gruyter. Matsumoto, Yo. 1993. Japanese numeral classi? ers: A study of semantic categories and lexical organization. Linguistics 31, pp. 667–713. 26 Nakanishi, Kimiko & Satoshi Tomioka. 2004. Japanese plurals are exceptional. Journal of East Asian Linguistics 13, pp. 113–140. Rullmann, Hotze & Aili You. 2003. General number and the semantics and pragmatics of inde? nite bare nouns in Mandarin Chinese.
URL http://semanticsarchive. net/Archive/jhlZTY3Y/Rullmann&You2003. pdf. Semantics Archive, deposited 9 December 2003. Sanches, Mary. 1973. Numerical classi? ers and plural marking: An implicational universal. In Working Papers on Language Universals, volume 11, pp. 3–22. Sato, Yosuke. 2008. Radical underspeci? cation, general number, and nominal denotation in Indonesian: An exo-skeletal approach. URL http://129. 242. 176. 75:9091/lingbuzz/@dceLbyOMOVaptzzt. LingBuzz archive, CASTL, Tromso. Sauerland, Uli, Jan Anderssen, & Kazuko Yatsushiro. 2005. The plural is semantically unmarked.
In Stephan Kepser & Marga Reis (editors), Linguistic Evidence. Berlin: de Gruyter. Sew, Jyh Wee. 2007. Reduplicating Nouns and Verbs in Malay. University of Malaya Press. Sneddon, James Neil. 1996. Indonesian: A Comprehensive Grammar . London: Routledge. Spector, Benjamin. 2007. Aspects of the pragmatics of plural morphology: On higher-order implicatures. In Uli Sauerland & Penka Stateva (editors), Presupposition and Implicature in Compositional Semantics. Basingstoke, UK: Palgrave Macmillan. Wilhelm, Andrea. 2008. Bare nouns and number in D? ne Sulin?. Natural Lane ? e guage Semantics 16, pp. 39–68. 27
