Journal of the Southeast Asian Linguistics Society JSEALS 19.1 (2026): clv–cxcii Data/Notes Article ISSN: 1836-6821 DOI: https://doi.org/10.21313/10524/52559 University of Hawai‘i Press A GRAMMAR SKETCH OF URUANGNIRIN1 Eline Visser Uppsala University eline.visser@lingfil.uu.se Abstract This paper presents a grammatical sketch of Uruangnirin, an Austronesian language of Indonesia. It provides an overview of word classes, personal pronouns, nouns and noun phrases, verbs and verbal inflection, clausal syntax, valency-changing strategies, complex predicates, clause coordination and complement clauses. A brief overview of other Uruangnirin materials and publications, notably the audio-visual corpus, is included. Keywords: Indonesian languages, Austronesian, language description, endangered languages ISO 639-3 codes: urn 1 Introduction Uruangnirin is an Austronesian language spoken in eastern Indonesia, with an estimated speaker population of approximately 400 individuals.2 It is used in four villages Tarak, Tuberwasak,3 Faur, and Kiaba on the Karas Islands. The geographical position and all six villages are shown in Figure 1. The Uruangnirin-speaking islands are green. Uruangnirin names differing from Indonesian names are given in italics. Figure 1: Location of Karas Islands 1.1 Genealogy and other languages in the area The Karas Islands are also home to Kalamang, a Papuan language belonging to the Greater West Bomberai family (Usher and Schapper 2022). Geographically, the islands lie just offshore from the Bomberai Peninsula, an area with a total of 16 languages. Figure 2 presents a map of the languages spoken on and around the Bomberai Peninsula (known in Indonesian as Semenanjung Onin ‘Onin Peninsula’). The language boundaries shown are based on SIL’s 2003 Peta bahasa Papua ‘Map of Papuan languages’,4 and the Austronesian languages represented all belong to the Seram–Tanimbar–Bomberai subgroup (Grimes & Edwards 2026). These include Onin, Sekar, and Uruangnirin (Bomberai Tip), as well as Arguni, Mbaruan (also known as Bedoanas), Yarikman (also called Erokwanas), and Irarutu (East Bomberai; see Jacobson Holmström 2025 for an alternative classification and Milano 2025 for discussion of mutual intelligibility among Arguni, Mbaruan, and Yarikman). Three languages of the Greater West Bomberai family are spoken in the area: Kalamang, Iha, and Mbaham. The remaining Papuan languages5 are Mor (an isolate), Kemberano (South Bird’s Head), Tanahmerah, also known as Sumuri (isolate), and Buruwai and Kamberau (Asmat–Kamoro). While I cannot confirm the accuracy of the area labeled as uninhabited on the map, there is at least one settlement there – Malakuli – which is marked in Figure 1. This village was established several decades ago by the Indonesian government as an accessible administrative center for Karas District and is inhabited by speakers of Kalamang, Uruangnirin, and Buruwai. The closest language to the west that is not shown on the map is Geser-Gorom, an Austronesian language of the East Central Maluku group, spoken on and around Gorom Island (also known as Goran or Gorong), approximately 150 km west-southwest of Karas. Figure 2: Languages spoken on and around Karas Uruangnirin is grouped with Onin and Sekar as Bomberai Tip under the Tanimbar-Bomberai subgroup. It has long been clear that these lects are very similar to Uruangnirin. While much less data is available for Onin and Sekar, recent documentation efforts (Wijaya 2025; Visser 2025a) enabled a lexical and phonetic distance study between the three lects. Kaplan (2026) concludes that Bomberai Tip is best seen as a language chain. The geographic distribution of the Tanimbar–Bomberai languages is illustrated in Figure 3. Figure 3: The Tanimbar-Bomberai languages and Banda. (Courtesy of Grimes and Edwards, 2026) 1.2 Level of endangerment and sociolinguistic situation According to Ethnologue (Eberhard et al. 2020), Uruangnirin is assigned an EGIDS rating of 6b (Threatened, Lewis and Simons 2010), implying that language transmission still occurs in at least some child-bearing generations. Based on my own observations, however, there appear to be no fluent speakers born after the year 2000. For this reason, I classify Uruangnirin as EGIDS 7 (Shifting), corresponding to the category of Definitely Endangered on UNESCO’s scale. A shift from Uruangnirin to Papuan Malay is currently underway, with a pronounced generational divide among individuals born approximately between 1990 and 2000. Most speakers born before 1990 remain fluent in Uruangnirin and use it regularly with other speakers across a wide range of everyday contexts. All Uruangnirin speakers are bilingual in Papuan Malay, and some additionally speak Bahasa Indonesia, the standardized national variety of Malay. Papuan Malay is typically employed when a non-Uruangnirin speaker participates in a conversation, which commonly occurs during events such as village meetings or wedding speeches. Nevertheless, Uruangnirin continues to be used at such gatherings in smaller-group interactions and in ritual contexts. Papuan Malay is also generally used when addressing people born after approximately 2000. Older speakers report that they did not acquire Papuan Malay or Indonesian prior to entering school. In addition to Papuan Malay, many residents of Kiaba (the village where most of the fieldwork was conducted) have a strong command of Geser-Gorom. Kalamang is also widely spoken, as many inhabitants of Kiaba have relatives in Mas, a Kalamang-speaking village. With the exception of Papuan Malay, none of these languages are currently transmitted to children. In mixed marriages, communication between spouses takes place in Papuan Malay or Uruangnirin. Uruangnirin does not have a written tradition and is not used for administrative purposes. However, when prompted, speakers readily write words and texts in Uruangnirin using Indonesian orthography, which is well suited to the language’s phonological system. With the increasing availability of affordable smartphones and internet access, some use of Uruangnirin can now be observed on social media platforms such as Facebook. 1.3 Data collection and database The data underlying this sketch were collected during three fieldwork trips conducted by the author in the villages of Kiaba and Tarak. The first trip took place in the winter of 2022/2023 and lasted three and a half months; the second occurred in early 2024 and lasted nearly two months, and the third, in 2025, lasted three weeks. All materials are archived at the Endangered Languages Archive (Visser et al. 2023). Examples cited in this sketch are accompanied by identifying tags, which can be used to find the corresponding items in the archive. 2 Phonology Uruangnirin phonology was sketched in Visser and Stickl (2025) and is therefore treated only summarily here. There are 18 consonant phonemes: /p b t d c k g m n ŋ r f s h w j l/, and 5 vowel phonemes: /i e a o u/. The consonant phoneme inventory is given in Table 1 and the vowels are given in Table 2. There is variation in the realization of especially /e/, which is extensively discussed in Visser and Stickl (2025:xxix–xxx). Table 1: Consonant phonemes (orthography, where different from IPA, is given in angle brackets) bilabial labiodental alveolar palatal velar glottal plosive nasal p b m t d n c y k g ŋ trill r fricative f s h approximant w j w lateral l Table 2: Vowel phonemes i u e o a The minimal syllable consists of a single vowel, while the maximal syllable structure includes a vowel that may be preceded by one or two consonants and followed by a single consonant. Most lexical roots are disyllabic, with monosyllabic roots occurring less frequently. Across the corpus, the most frequent root pattern is CVCVC. Roots beginning with a vowel are generally rare. All consonants except /ŋ/ are found in onset position. The only unvoiced stop that cannot occur in coda position is /c/. The other voiceless stops (/p, t, k/) are unreleased in this position. Voiced stops (/b, d, ɟ, g/) do not occur in coda position. Finally, /h/ cannot occur in coda position in words other than interjections. This phoneme is otherwise only used in loan words. While CC-onsets are allowed, they are mostly found in verb stems and typically avoided by adding a vowel-final subject index prefix (e.g., ftobir ‘search’ ∼ ta-ftobir ‘we search’). Second, words with CC-onsets seem to be the result of vowel deletion, and usually have a variant with a vowel (e.g., ftobir ∼ fatobir ‘search’). Vowels may occur in any position. Vowel sequences are common and there are few restrictions. Stress is typically on the penult (ˈfodan ‘tikal leaf’), with some exceptions where stress is on the last syllable (kaˈsuk ‘man’). Some of the words with final stress are loans from Kalamang (saˈnoŋ ‘thatched roof’). Gunnarsson (2025) presents a first analysis of Uruangnirin intonation, with a description of three prosodic domains: the phonological word, the accentual phrase and the intonational phrase. There are two morphophonological processes that are important for this grammar sketch. Both pertain to verbs: vowel assimilation and vowel deletion and prothesis. First, certain verbs exhibit vowel assimilation in the first syllable when their stem begins with an onset /p/, /t/, or /k/, followed by /a/ and a consonant. In these verbs, inflection with a subject index triggers a change in the stem’s initial /a/ to /i/ whenever the subject index itself contains /i/ What was described as vowel harmony earlier (Visser & Stickl 2025) is better described as vowel assimilation, since the process does not affect the whole word, as vowel harmony typically does. (1) a-kabom ‘3-be.wet’ ∼ i-kibom ‘1SG-be.wet’ Second, at least 50 lexemes show vowel deletion and prothesis in the first syllable. It occurs mainly on verbs, and mainly involving the consonants /f/ and /m/ and the vowel /a/. Many words with these phonemes in the first syllable show alternations (for example between am- and ma-), or they allow one of the forms. Many of these words seem to carry more or less fossilized remnants of proto-Austronesian *ma- ‘stative’ and *pa- ‘causative’ (Blust 1993:371), reflected as am-/ma- and af-/fa-. (2) amdufak ∼ madufak ‘to forget’ (3) afreka ∼ farkea ‘to remember’ The first person plural exclusive and second person plural subject indexes can also undergo vowel deletion and prothesis. The variation appears to be without function. Uruangnirin shows sociolinguistic variation primarily in phonology and morphology, broadly dividing speakers into older/more conservative and younger/more progressive groups, with the conservative pattern largely absent among the youngest speakers. Conservative speakers retain final /n/ where progressive speakers use /ŋ/, and they also preserve word-initial /ŋg/ onsets that progressive speakers simplify to /g/. In verbal morphology, conservatives use an irregular third-person prefix n-, whereas progressive speakers reanalyze this element as part of the stem and instead apply the regular prefix a-. Additionally, vowel assimilation occurs only in conservative speech, while progressive speakers generally do not apply it. 3 Word classes Verbs can be active or dynamic (denoting events) or stative (denoting states or properties). Only active verbs can carry subject indexes, and subject indexes, though common, are not obligatory. Property words like amnipis ‘thin’ are also analyzed as verbs. Some verbs may take or show remnants of the following verbal morphology: causative f-, transitivizer -k and passive m- or ta(r)-. There are regular and irregular verbs. The main difference is that regular verbs are inflected with a- in third person, while irregular verbs are inflected with n-. Further properties of verbs and verbal inflection and derivation are described in §6. §8 addresses valency-changing strategies. Multi-verb constructions are described in §9. Nouns are unmarked for case regardless of their syntactic function. They can be modified by NP modifiers like attributively used verbs, quantifiers, demonstratives, and other nouns or noun phrases. NPs may also occur in predicate function in nominal or equational clauses, see §7.2. Nouns can be reduplicated for various reasons (§5.1). Nouns and NPs are further described in §5. Pronouns differ from nouns in one respect: they cannot be reduplicated. Uruangnirin pronouns distinguish between first, second and third person and singular and plural number. There is an inclusive and an exclusive first-person pronoun. Personal pronouns are treated in §4. Quantifiers can be subdivided into two groups: numerals like fat ‘four’ and non-numeral quantifiers like fisik ‘some’ and demtotir ‘all’. Quantifiers are adnominal modifiers in the NP. If the referent is clear from the context, they can also be the NP head. For a further description, see §5.2.2. Pronominal demonstratives take the slot of the noun; adnominal demonstratives occur after the noun and identificational demonstratives occur in copula and non-verbal clauses. Uruangnirin uses the same three basic forms in all three uses: proximal ke and distal ka/ko. They are further illustrated and described in §5.2.3. Adverbial modifiers can be divided into functional groups such as modal adverbials or manner adverbials. Most of the modifiers are clause final. The most common adverbial modifiers have to do with aspect and negation: they are som ‘already’, tai ‘no; not’, efi ‘not yet; still’ and boit ‘yet; still’. They are described in §7.5. There are three indigenous prepositions, all derived from verbs, and one loan from Papuan Malay. The indigenous prepositions are locative amina ‘in; at’, ablative afara ‘from’ and comitative/instrumental nora ‘with’. The borrowed preposition is sampe ‘until’. They are not discussed separately in this sketch, but they are illustrated in several examples throughout the text, like (16), (18), (35), (38) and (94a). The distribution of amina ‘in; at’, ablative afara ‘from’ and comitative/instrumental nora ‘with’ and their predicative counterparts is analyzed in Rodehav (2026). Question words are nominal, adnominal, quantifier and adverbial question words. Their main shared feature is that they structurally and obligatorily replace other constituents, which is why they are grouped together as a word class. There are four question word roots: be ‘where; which’, fira ‘how much/many’, se ‘who’ and suai ‘what’, some of which can be combined with other elements. Table 3 lists the question words with their English translations, the entity they question, and their syntactic identity. Question words are illustrated in §7.3. Table 3: Question words questions syntactic identity afbe why reason adverbial afbe how manner adverbial be where place adnominal/adverbial be which thing adnominal fira how much/many quantity quantifier se who person pronominal suai what generic pronominal Conjunctions include multi-functional bo that translates mostly as ‘but’ or ‘and’. The second-most common conjunction is jadi ‘so; then’, a loan from Papuan Malay. It is used to relate clauses sequentially or to express consequence and reason. Bi ‘or’ is a disjunctive coordinator. Clause coordination and subordination is described in §10. Interjections and fillers have not been explored properly yet. Examples for interjections are neno, used to express surprise, m, a confirmative answer and o ‘EMPH’ as illustrated in (36). Fillers including the placeholder suai ‘what’ and hesitators like ah. 4 Personal pronouns The personal pronouns are the same regardless of syntactic function. They are given in Table 4. Table 4: Pronouns SG PL 1 lau INCL: ita EXCL: ami 2 o imi 3 ia sina The first-person singular pronoun lau is illustrated in S, A and O function in examples (4)–(6). (4) lau i-fer nong ke 1SG 1SG-come_from sea PROX ‘I come from the sea-side here.’ [a-petikrica_103] (5) lau baru i-suan ain fira bi laf mo 1SG newly 1SG-plant plant how_many or only SOFT ‘I just planted only I-don’t-know-how-few plants.’ [c-sp-lk-bk_68] (6) sin-teni a-bantu lau 3PL-three 3-help 1SG ‘They three helped me.’ [s-pear-skm_20] Possessive pronouns are given in Table 5. They all share some material with the basic pronouns and typically have a nasal (/m/ or /n/) plus a thematic vowel added. The 1pl.excl, 2pl and 3pl forms all contain a reduplicated VC-syllable. The 1SG and 1PL.INCL forms have extra material. These are also often shortened in a myriad of ways. For example, common short forms of the first-person singular are anung and lanung, while common short versions of the first person plural inclusive are itnen and tanen. The third person singular form ini also surfaces as ni and in. Possessive constructions are described in §5.2.6. Table 5: Possessive pronouns SG PL 1 lanunggi INCL: itanen(d)a EXCL: amami 2 omu imimi 3 ini sinina Lastly, there is a set of pronouns that translate as ‘pronoun + alone’ or ‘pronoun + self’. They are given in Table 6. Forms consist of the basic pronoun plus (C)(V)mVs(V), which could be related to mancia ∼ mansia ∼ mencia ∼ mincia ‘person; people’. Compare also Arguni misíer (Milano 2025:38) and Mbaruan masíar (Jacobson Holmström 2026:48), which mean ‘person’ or ‘people’ but are also used as a numeral classifier with pronouns. The only exception to this pattern is itamasita which has an extra /t/. Table 6: ‘Alone/self’ pronouns SG PL 1 amasalau INCL: itamasita 1 EXCL: amasami 2 omoso imisimi 3 imisia simisina This paradigm seems on the way out, however, as younger and less conservative speakers use the basic pronoun plus imisia to make the ‘alone/self’ construction. Hence, for those speakers imisia is reanalyzed as meaning ‘alone/self’ without being marked for person. The two strategies are contrasted in (7) and (8). (7) bisa itamasita ta-funa som can 1PL.INCL.alone 1PL.INCL-do IAM ‘We can do it ourselves.’ [c-kampung-sl-lk_910] (8) ita imisia ta-suan 1PL.INCL alone 1PL.INCL-plant ‘We plant [it] ourselves.’ [c-kampung-sl-lk_987] 5 Nouns, nominal modifiers and noun phrases In this section, I describe some aspects of nouns, nominal morphology and the (modifiers in the) noun phrase (NP). Example (9) shows a subject noun in an intransitive clause. (10) shows a subject and an object noun in a transitive clause. There is no case, gender or number marking on nouns. However, the subject can be indexed on the verb with a prefix. For third person referents, only number is indexed. For first and second person referents, both person and number are indexed. Subject indexing is not obligatory, but it is common. Objects cannot be indexed on the verb. Verb inflection is described in detail in §6. (9) wat a-irit coconut 3-fall ‘A coconut falls.’ [e-pron-nk_16] (10) kasuk n-ois kawat man 3-cut wire ‘A man cuts wire.’ [e-trnstv-kky_62] In §5.1, I discuss the only derivational device available for nouns, reduplication, and some related issues like generic and plural nouns. In §5.2 I show the different ways in which nouns can be modified. I described the adjectival use of verbs in §5.2.1. Quantifiers and demonstratives are treated in §5.2.2 and §5.2.3. I sometimes venture away from the noun phrase to discuss some other traits and functions of these word classes. Nominal modifiers are the topic of §5.2.4. What appear to be relative clauses are described in §5.2.5. Finally, possessive constructions are described in §5.2.6. 5.1 Reduplication, generics, plurals There is no nominal derivational morphology in Uruangnirin except reduplication. Reduplication can be used to derive nouns from verbs. (11) shows some examples. The process is not rare, but it does not seem to be very productive either. (11) a. sula ‘to dance’ → susula ‘dance’ b. mian ‘to be; to live’ → mimian ‘living’ c. suan ‘to plant’ → susuan ‘plant’ d. kutar ‘to tie up hair’ → kutkutar ‘bun’ e. labat ‘to be wounded’ → lablabat ‘wound’ f. to ‘to descend’ → toto ‘descent’ Reduplication can also be used for plurals or for collective reference, but it is quite rare. It seems to be restricted to referents that one often encounters in masses (bones, leaves) and kinship terms. Three examples are given in (12). Note that in minlen, it is not the first but the last syllable that is reduplicated. This is probably because of the final stress in this word (§2). Kasuk ‘man’, which also has final stress, behaves similarly: kasuksuk ‘men’. (12) a. nurin ‘bone’ → nurnurin ‘bones’ b. tatan ‘grandparent; grandchild’ → tatatan ‘grandparents; grandchildren’ c. minlen ‘woman’ → minlenlen ‘women’ A minor pattern of reduplication with nouns is generalization (an analysis first noted in Sibakov 2024). The reduplicated forms are not plurals or collectives, but a semantically related but more generic referent. (13) a. bebin ‘fruit skin; nut shell’ → bebebin ‘waste’ b. numit ‘algae’ → numnumit ‘fungus’ Three kinship terms have forms that are used as plurals or for generic forms that are different from their singular or specific forms. The forms seem related, but they are not pluralized following the same pattern. They are given in (14). Forms that seem cognate with either the singular or the plural forms in this example can be found in several Tanimbar-Bomberai languages (Mettler and Mettler 1997; Milano 2025; Jacobson Holmström 2026, 2025; Wijaya 2025; Visser 2025a). (14) a. keka ‘child’ → kamkau ‘children’ b. marwana ‘man’ → ma(n)rwanas ‘men’ c. mopata/mapata ‘woman’ → matafasi ‘women’ The forms kamkau, ma(n)rwanas and matafasi can mostly be analyzed as plurals. They are typically used to refer to any men, women and children in the village, for example when describing tasks at a feast (‘when the women had cooked, they served the food’). (15) and (16) show generic uses that are actually singular. (15) was given by a speaker to illustrate the use of kamkau in the dictionary. He refers to his former child-self with the generic form. In (16), the speaker describes the rituals related to the arrival of a new bride to the village. The story describes a hypothetical case, so the speaker uses generic matafasi instead of the specific mopata/mapata. (15) lau kamkau tai som 1SG children NEG IAM ‘I’m not a child anymore.’ [dictionary_kamkau] (16) adat rwang dike matafasi afar di_luar a-ma custom village PROX women from outside 3-come ‘The custom of this village is, [when] a woman from outside comes...’ [n-ppbaru-kky_3] 5.2 Noun modification Nouns can be followed by one or more modifiers. In (17), the inherently generic/plural noun marwanas ‘men’ is modified by the numeral nua ‘two’ and the proximal demonstrative ke ‘PROX’. In (18), the noun kem ‘tent’ is modified by the stative intransitive verb perperi ‘new’ (functioning as adjective) and the numeral sa ‘one’. That clause also contains an oblique argument introduced by a preposition: the noun tawanang ‘side’, modified by ka ‘DIST’. In (19), the noun mancia ‘people’ is modified by the quantifier fisik ‘some’. (17) [marwanas nua ke]NP a-fnaka a-mpeni bola men two PROX 3-run 3-play ball ‘These two men run playing football.’ [e-se2a-kky-bk_44] (18) a-fun [kem perperi sa]NP am tawanang ka som 3-make tent new one LOC side DIST IAM ‘One [should] make a new tent on that side.’ [c-sp-lk-bk_7] (19) [mancia fisik]NP n-amat waras people some 3-bring rope ‘Some people bring rope.’ [n-perahu-kky_85] 5.2.1 Adverbs Words describing properties and qualities are analyzed as stative non-agentive verbs because they function as the predicate of the clause in the vast majority of cases. The same goes for dynamic non-agentive verbs. Some of these verbs start with am-, ma- or m-, which seem to be reflexes of pAN *ma- ‘stative’ (Blust 2013:371). Examples are amdekar ‘dry’ and manawas ‘long’. A few passive and/or anticausative forms can be derived from transitive verbs with am- or m-, for example amnesik ‘unplugged’ from nesik ‘to unplug’. Ta(r)- also derives passive and/or anticausative forms, like tarpu- pis ‘opened’ from pupis ‘open (tr.)’. These are rather minor patterns – see for further description §8. Other stative intransitive verbs seem monomorphemic, like koin ‘elder (of relatives)’. Yet others are partially reduplicated, like petpetin ‘white’ and other colors, and wapwapar ‘ripe’, without the root having a meaning on its own. Some of the partially reduplicated stative intransitive verbs are derived: wirwiris ‘stringy’ comes from the noun wiris ‘string’, somsomut ‘boiled’ comes from the verb somut ‘to boil’ and komkoma ‘black’ comes from the verb koma ‘dark’. Compare also the compound lafi koma ‘ash’, where lafi means ‘fire’. Examples (20)–(23) show intransitive verbs modifying nouns. In other words, these verbs are used attributively here, essentially functioning like adjectives. (20) n-ois [ai maresresar]NP 3-cut tree very_hard ‘One cuts very hard trees.’ [n-perahu-kky_113] (21) kanambo a-pasang [gunting manawas ke]NP a-sa then 3-install scissor_beam long PROX 3-ascend ‘After that, one installs the long “scissor beam”.’ [a-rumah_59] (22) [seir narnaran ka]NP a-num [seir petpetin ka]NP fish red DIST 3-swallow fish white DIST ‘That red fish swallows that white fish.’ [e-fishfilm-kky-bk_11] (23) [keka lalak]NP bebar child small hungry ‘The small child is hungry.’ [e-val-kky_4] 5.2.2 Quantifiers Numeral and non-numeral quantifiers illustrated in (17)–(19) always follow the noun they modify. When the referent can be retrieved from the context, they may also be used predicatively. The main numerals are given in Table 7. Some numerals are building blocks in creating higher numbers. The place where a numeral must be inserted is indicated with a plus (+). When there is an exception, it is also given in the table. For example, 100 is raca and 300 is rateni, but the other hundreds are created with the building block *rati followed by another numeral. Table 7: Numerals 1 sa 2 nua 3 teni 4 fat 5 nima 6 nem 7 taransa 8 terinua 9 saputi 10 puca 11–19 puca resin + 10s puti + 21–29, 41–49 etc. puti + resin + 30 puteni 100 raca 100s rati + 300 rateni 1,000 ripi sa 1,000s ripi + 10,000 ripi puca, ripuca 10,000s riputi + 100,000 ripi raca 100,000s ripi rati + 300,000 ripi rateni 1,000,000 juta The ordering of the parts of numerals is from larger to smaller parts, as illustrated in (24)–(27). In the latter example, ripi ‘thousand’ is left out because it is understood from the context. The example, like (25) and (26), is about money, and for bigger sums thousands need not be mentioned.6 (24) raca puti nua resin teni hundred ten two NUM_LNK three ‘123’ [dictionary_raca puti nua resin teni] (25) kapala sa ripi rati nem house one thousand hundred six ‘600,000 per house’ [a-petikrica_77] (26) juta fat ripi puti nem million four thousand ten six ‘1,460,000’ [n-taruhharta-kky_89–90] (27) juta sa rateni million one [thousand] three_hundred ‘1,300,000 [lit. one million three hundred]’ [c-pancinganak-el_229] Uruangnirin, unlike other languages of the region (Klamer 2014:111, Levin and Polinsky 2019) and the subgroup (Milano 2025:38, Jacobson Holmström 2026:48), has no obligatory numeral classifiers. There is one noun, ai(n) ‘tree; plant’, that is sometimes used as one, as illustrated in (28). (28) sayang ain raca rati-nua nutmeg plant one_hundred hundred-two ‘one or two hundred nutmeg plants’ [c-kampung-sl-lk_223] 5.2.3 Demonstratives The three basic demonstratives are proximal ke and distal ka/ko. The main opposition is between proximal ke and distal ka. The form ko is much less used. What governs the choice between ka and ko is unclear to me. In elicitation sessions with the help of a demonstratives questionnaire (Wilkins 2004), speakers sometimes offered both ka and ko for the same distal item. This is reminiscent of Kalamang, the Papuan neighbor of Uruangnirin, and to Yamdena, belonging to the same subgroup. Kalamang has a basic opposition between proximal wa and distal me, and a less common and less flexible form owa, which is analyzed as a far distal (Visser 2022:230). Yamdena has a basic opposition between proximal ye and distal ne, and a far less common form no, which is analyzed as a far distal (Visser 2024:85) (Mettler and Mettler 1997:140). There is no evidence, however, that the Uruangnirin form ko functions as a far distal.The demonstratives have three uses: pronominal, adnominal and identificational. The same forms are used regardless of function. I illustrate the adnominal use for each demonstrative in (29)–(31). (29) sayang ke tubi bobar afke nutmeg PROX grow quickly like_this ‘This nutmeg grows this quickly.’ [c-kampung-sl-lk_262] (30) kasuk ka toning som man DIST want IAM ‘That man (already) wants [it].’ [c-tokipala-kb_41] (31) ari ko o a-mtontonang person DIST EMPH 3-sit.CONT ‘That man is sitting.’ [e-pron-ab-kky_18] Demonstratives are used pronominally when the referent can be retrieved from the context. When the demonstrative is used as one of the NPs in an equational clause (such as ‘this is a banana’), that use is called identificational (see §7.2). There are longer forms of the three basic demonstratives. Their functions are not immediately clear and need to be investigated further. These longer forms are made with a prefix and a suffix. The prefix seems to be derived from the relative marker da. It may or may not show vowel assimilation with the demonstrative. With proximal ke we find dike, deke and dake. The only attested prefixed form of ka is daka. With far distal ko we find both dako and doko. The forms that show vowel assimilation are more common than the ones that don’t. The form dike is (much) more common than the form deke. Examples are given in (32) and (33). (32) keka dike a-fun keka doko in tangan labat child PROX 3-do child DIST 3SG.POSS arm wound ‘This child is hurting that child’s arm.’ [e-trnstv-kky_50] (33) Dahlia wanggi seir daka Dahlia give fish DIST ‘Dahlia gave that fish.’ [a-masak_610] The demonstrative ke ‘PROX’ may also be lengthened by suffixing me, while ka ‘DIST’ can be lengthened with either me or mese. The suffix -me(se) is not attested with far distal ko. The demonstrative affixes can only be combined on ke: dikeme is attested once in elicitation and once in a translated text. Table 8 shows the possible combinations. A thorough corpus analysis, possibly combined with some elicitation, is needed to understand what conditions the choice between these forms. Eriksson (2026) makes an attempt at finding relations between form and function. Although he finds some correlations (for example that the simple forms are more likely to occur adnominally than the complex forms, and that daka is more common as a pronoun), it seems impossible with the current data to determine the exact use and meaning of the different forms. Table 8: Possible combinations of demonstratives and affixes PROX de- ke -me de- ke di- ke -me di- ke da- ke DIST da- ka ka -me ka -mese FDIST da- ko do- ko The manner demonstratives are afke ‘like this’, afka ‘like that’ and afko ‘like that’. While afka is very common (with 238 occurrences) afko only has four occurrences in the current corpus, one of which is given in (34), from a text about a landslide. The meaning of distal ka in this example is not quite clear, but prosodically it goes together with pertama (one can take a break after ka, not after pertama). I come back to discourse-ordering functions at the end of this section. (34) jadi pertama ka ia afnaka afko ma kawer so first DIST 3SG 3-run like.that come soil ‘So first it ran this way like that, the soil.’ [c-longsor-bk_150] The basic demonstratives can also be combined with locative am(in) and ablative (a)far(a): am ke ‘here’, far ke ‘from here’, am ka/ko ‘there’ and far ka/ko ‘from there’. These constructions may be combined with the nouns nan/nana ‘land’, non/nona ‘sea’, nas/nasa ‘top’ and pap/papa ‘bottom’. The latter two have variants which are used in other contexts and constructions than the ones described here, but the distribution is not yet clear: nasin ‘top’ and papan ‘bottom’. Shorter versions of ‘land’ and ‘sea’, namely na and no, feature in the verbs mna ‘go landwards’, tna ‘go landwards’, mno ‘come seawards’ and tno ‘go seawards’. An example of the template ‘preposition + demonstrative + relational noun’ is provided in (35). This template can be extended with a shorter version of the relational noun, as illustrated in (36).7 (35) ami tno amin ka nona 1PL.EXCL go_seawards LOC DIST sea ‘We went to that seaside.’ [n-kerjapp-an_841] (36) anung beda nan daka nana o 1SG.POSS axe land DIST land EMPH ‘My axe is there on the landside.’ [a-tanampala-kky_18] It is possible that the distal demonstratives ka and ko used to have something to do with landside and seaside, since they have the same thematic vowels /a/ and /o/. However, in current Uruangnirin it is perfectly possible to refer to a distal object on the landside (nan) with ko, and to a distal object on the seaside (non) with ka. Finally, there is the anaphoric demonstrative peta. Unlike ke, ka and ko, it cannot carry a dV- prefix or -me(se) suffix, it cannot be turned into a manner demonstrative with af-, and it cannot be preceded by a preposition or combined with relational nouns like nan(a). Its typical use is as a noun modifier, indicating that the noun it modifies was mentioned earlier in the discourse or should be considered part of the common knowledge. For example, in (37), the speaker describes a short video clip which features a toy car. The speaker had already seen this toy car in an earlier clip, followed by clips featuring other things and people. He uses peta to indicate that it’s the same toy car as a few clips ago. Similarly, in (38), the speaker has mentioned kapuni ‘bamboo’ in utterance 38 and 44, and then refers to it again in utterance 50 to indicate it’s still the same bamboo he is talking about. (37) [mobil lalak peta]NP a-fnaka car small ANA 3-ride ‘That small car [that we’ve seen earlier] is riding.’ [e-se2a-kky-bk_67] (38) a. ia n-osir o o m-ois kapuni ke teni [...] 3SG 3-say QUOT 2SG 2SG-cut bamboo PROX three ‘He said: “You cut three stems of this bamboo.”’ [n-iblismesjid-bk_37–38] b. ia sobang a-seik wer a-to ah kapuni nanam [...] 3SG order 3-pour water 3-descend HES bamboo inside ‘He ordered to pour water inside the bamboo stems.’ [n-iblismesjid-bk_43–44] c. i-namat nora [kapuni peta]NP 1SG-hold with bamboo ANA ‘I held it with that bamboo.’ [n-iblismesjid-bk_49–50] Peta is also a temporal adverb that means earlier. As such, it can be used at the start or end of a clause. Peta is similar in function to Kalamang opa (Visser 2020) and Indonesian tadi (Sneddon et al. 2012:135). The different demonstrative forms seem to have many other functions than the ones described in this section. One indication of this is the fact that demonstratives forms are frequently combined, as in (115), or appear clause-finally. Some of these functions seem to have to do with discourse organization, but this is a topic for future investigation. 5.2.4 Nominal modifiers Nouns can modify nouns, whereby the second noun functions as a modifier of the first noun. So, in (39), the noun ruang ‘village’ modifies mancia ‘people’, narrowing its scope. In (40), the noun ai ‘tree’ is further specified with the name of a species: sagum. (39) [mancia ruang ke]NP a-bana Pakpak people village PROX 3-go Fakfak ‘The people from this village went to Fakfak.’ [n-basunat-kky_38] (40) m-ois [ai sagum]NP laf 2SG-cut tree sagum just ‘You just cut the sagum tree.’ [a-tanamrica_238] It is yet unclear whether or constructions that can be analyzed as ‘noun + nominal modifier’ should be separated from ‘noun + noun’ compounds. 5.2.5 Relative clauses Relative clauses can be formed with relativizer da, as illustrated in (41) and (42). (41) keka da a-pake sungsungga fisik a-tewa keka da a-pake sungsungga manawas child REL 3-wear pants short 3-hit child REL 3-wear pants long ‘The child that wears shorts hits the child that wears long pants.’ [e-da-kky-bk_8] (42) minlen da a-ti a-sobang woman REL 3-go 3-order ‘It was my wife who went to order.’ [n-rumah-kky_113] Examples like (41) are rare and are all elicited. And as is clear from (42), da could also be analyzed as a topic marker. In (43), the relative clause reading is not possible anymore and should be analyzed as a topic marker. (43) se da amin Kaimana who TOP LOC Kaimana ‘Who is in Kaimana?’ [a-petikrica_243] Da is also a very common conjunction for temporally sequential events, as described in §10.1. 5.2.6 Possessive constructions In possessive constructions, NPs can either function as the possessor or the possessed NP. In (44), the NP headed by keka ‘child’ is the possessor, and the NP topi ‘hat’ is the possessed. The possessor precedes the possessed NP. The possessed NP is preceded by the third person possessive pronoun ini. The possessive pronouns are listed in Table 5 in §4. (44) [[keka lalak ka]poss’or [ini topi]poss’ed]Subj a-irit child small DIST 3SG.POSS hat 3.fall ‘That small child’s hat fell.’ [s-pear-skm2_35] Example (45) illustrates a possessive construction with a possessive pronoun as the possessor, and rau ‘canoe’ as the possessed. (45) m-ur ami ta-soin [[amami]poss’or [rau]poss’ed]Obj nomba 2PL-help 1PL.EXCL 1PL.INCL-push 1PL.EXCL.POSS canoe first ‘You help us push our canoe first.’ [n-perahu-kky_78] Possessor-possessed is the typical order for languages in Wallacea, the linguistic area to which Uruangnirin belongs (Schapper and Gasser 2023). The place of the possessive marker in this area varies, but most Austronesian languages have a suffix or enclitic on the possessed noun. Uruangnirin patterns with languages in Ambon, Western Seram, Southwest Maluku and eastern Timor by having a proclitic on the possessed noun. Note that Kalamang – which is Uruangnirin’s closest neighbor and a Papuan language – patterns unlike other Papuan languages of Wallacea in having a suffix on the possessed noun, perhaps influenced by Austronesian languages. Though spelled as a word, ini is a proclitic that is often shortened to in or ni. Possessive clauses can be nested. In (46), the head kapala ‘house’ is modified by Rens in laman ‘Rens’ father’. More levels of embedding were elicited in (47), where the head sukmatan ‘door’ is modified by kapala ‘house’, which is in turn modified by popa ‘mother’s brother’ and laman ‘father’. These examples also show the (seemingly unmotivated) variation in form of the possessive marker. (46) a-puti amin Rens in laman in kapala 3-gather LOC Rens 3POSS father 3POSS house ‘They gather at Rens’ father’s house.’ [c-minang-wn-nk_4] (47) ini laman ini popa in kapala in sukmatan 3POSS father 3POSS mother’s_brother 3POSS house 3POSS door ‘his/her father’s uncle’s house’s door’ [notes2022-11-09_2] Possessive constructions without a possessive pronoun optionally occur in teknonyms (the custom to refer to adults by the name of their first child or grandchild, in Uruangnirin combined with a word like ‘(grand)father’ or ‘(grand)mother’. So, one can say Rens in laman ‘Rens’ father’ with in ‘3POSS’ as in (46) above, or simply juxtapose the name and kinship term: Rens laman ‘Rens’ father’ (illustrated in 54 and 55 in §7.1). Possessive classification is also very common in Wallacea but has so far not been found in Uruangnirin. A common type of possessive classification differentiates between alienable and inalienable nouns (Schapper and Gasser 2023). (48) illustrates that both patuan ‘stomach’ (a typical inalienable noun) and pena ‘pen’ (a typical alienable noun) occur in the same possessive construction. (48) a. anunggi patuan sokar 1SG.POSS stomach hurt ‘My stomach hurts.’ [e-d3-kky_13.3] b. anunggi pena a-fata som 1SG.POSS pen 3-break IAM ‘My pen is broken.’ [e-val-nk_23] 6 Verbs and verbal inflection This section treats verbs and verbal inflection. In §6.1, I describe regular and irregular verb inflection. §6.2 describes how this inflection is sometimes omitted and what conditions this differential subject realization. 6.1 Regular and irregular verb inflection Regular verbs and irregular verbs differ in their subject indexing and whether they are obligatorily inflected. They may carry a third person index a- or remain uninflected (in any person). Irregular verbs carry a third person index n- and cannot remain uninflected. While they can only be inflected with consonant-only subject indexes for the persons where this is available (2SG, 1PL.EXCL, 2PL), regular verbs may take a whole range of index variants for these persons. All irregular verbs have a vowel-initial stem, but not all vowel-initial stems are irregular. The properties of the verb classes are summarized in Table 9. The twelve irregular verbs in the current corpus are listed in Table 10. Table 9: Subject indexes of the two verb classes regular irregular 2SG index mi-, m- m- 1PL.EXCL index ama-, am-, ma-, mi-, m- m- 1PL.INCL index ta-, t- t- 2PL index im-, mi-, m- m- third person index a- n- obligatorily indexed? no yes phonotaxis any #V Table 10: Irregular verbs stem translation amat hold; bring arat bite an eat ana get edat ask edin pluck emin drink iat touch; hit ofak disappear ois break; cut or(a) bring; join; follow; be with; use osir say; follow Table 11 illustrates all subject indexes. It shows the inflection of three regular verbs (one obstruent-initial, one vowel-initial, and one nasal-initial) and an irregular verb. The variation is commented on in §6.2. Table 11: Subject indexing on regular verbs and an irregular verb. regular regular regular irregular bana ‘go; walk’ ut ‘pull out’ nen ‘load’ osir ‘say; follow’ 1SG lai-bana, lai-ut, lai-nen, lai-osir, li-bana, li-ut, li-nen, li-osir, i-bana i-ut i-nen i-osir 2SG mi-bana mi-ut, mi-nen m-osir m-ut 3SG a-bana a-ut a-nen n-osir 1PL.EXCL ama-bana, ama-ut, ama-nen, m-osir am-bana, am-ut, am-nen, ma-bana, ma-ut, ma-nen, mi-bana mi-ut, mi-nen m-ut 1PL.INCL ta-bana ta-ut, ta-nen t-osir t-ut 2PL im-bana, im-ut, im-nen, m-osir mi-bana mi-ut, mi-nen m-ut 3PL a-bana a-ut a-nen n-osir Younger and less conservative speakers have reanalyzed the irregular verbs as regular ones. The third person index n- is reanalyzed as part of the verb stem, and hence they inflect the verb in the same way as regular verbs that start with /n/, like nen ‘load’ in Table 11. Table 12 shows how such speakers inflect an irregular verb. Table 12: Subject indexing on irregular verb osir ‘say; follow’(left )and re-analyzed regular form (right) irregular regularized 1SG lai-osir, li-osir, i-osir lai-nosir, li-nosir, i-nosir 2SG m-osir mi-nosir 3SG n-osir a-nosir 1PL.EXCL m-osir ama-nosir, am-nosir, ma-nosir, mi-nosir 1PL.INCL t-osir ta-nosir 2PL m-osir im-nosir, mi-nosir 3PL n-osir a-nosir One verb that is irregular but behaves differently from the other irregular verbs is uan ‘to eat’. It only occurs in first person singular as (la)iuan, as in (49), and in second person and first-person plural exclusive as muan, like (50) and (51).8 The much more common an ‘to eat’ is used with all persons and numbers. (49) i-uan seir nora kokok patin 1SG-eat fish with chicken egg ‘I eat fish with an egg.’ [n-kenari_590] (50) ma m-uan som m-uan m-ukur omu tenaga come 2SG-eat IAM 2SG-eat 2SG-measure 2SG.POSS stamina ‘Come eat! Eat and measure your stamina!’ [n-kerjapp-an_390] (51) m-uan namnamin punit kembali masikit 1PL.EXCL-eat food finish return mosque ‘After eating, [we] returned to the mosque.’ [n-maulud-ak_26] Another type of irregularity on verbs pertains to the verb stem, which in a few cases shows vowel assimilation in the first syllable of the stem with the subject index. This is only used by older speakers and was described in §2 above. Uruangnirin has no infinitive verb form, but the third person form is used by most people as the citation form and is also used in impersonal constructions. 6.2 Differential subject realization Uruangnirin has differential subject realization, which means that there is variation in how the subject is expressed or referred to depending on the properties of the subject. This pertains both to the expression of the subject in the clause (noun, pronoun, zero) and to the marking of the subject on the verb (index or no index). A detailed account of differential subject marking in the clause and on the verb is Faghiri and Visser (2025). Here, I summarize the findings for differential subject indexing, i.e., the indexing of the subject on the verb. In the two-way choice between indexing or not indexing the subject on the verb, four parameters play a role: • speech act participants are more likely to be indexed than non-speech act participants • subjects expressed as zero pronouns or noun phrases are more likely to be indexed than subjects expressed with a pronoun • referents with low topic persistence9 are more likely to be indexed than those with high topic persistence • patients are more likely to be indexed than agents Other variables that were tested are transitivity, semantic role, referential distance10 and serial verb constructions, but these do not seem to play a role in the choice between indexing the subject on the verb or not. However, the choice is not only between having and not having a subject index on the verb. When a speaker opts for indexing the subject on the verb, there is still sometimes a choice between different subject prefixes. All subject indexes are given in Table 13, together with the pronouns. Some of the variation in indexes can be explained by morphophonological rules. The short prefixes m- (2SG, 1PL.EXCL, 2PL) and t- (1PL.INCL) are only used on vowel-initial verb stems, and on irregular verbs as described in §6.1. These prefixes are therefore printed in gray. If we ignore these, there is seemingly free variation in index choice for 1SG, 1PL.EXCL and 2PL. Table 13: Pronouns and subject indexes. Morphophonologically restricted indexes that are printed in gray pronoun subject index(es) 1SG lau lai-, li-, i- 2SG o mi-, m- 3SG ia a- 1PL.EXCL ami ama-, am-, ma-, mi-, m- 1PL.INCL ita ta-, t- 2PL imi im-, mi-, m- 3PL sina a- For 1SG, the prefix that is preferred by far is i-, used in about 70% of the cases. Lai- and li- are equally common. For 1PL.EXCL, ma- is the preferred form (70%), followed by am- (20%). Mi-, which is also used for 2SG and 2PL, and the longer form ama- is quite rare. For 2PL, mi- is used in roughly 60% of the cases and im- in about 40%. This variation is partly explained by the phonotaxis of the verb stem, with vowel-final verb indexes used to break up consonant clusters, as mentioned in §2. The other two factors that play a role in the choice of index are referential distance and topic persistence. Roughly speaking, the lighter or shorter the form, the lower the referential distance and the higher the topic persistence. In other words, the heavier and longer forms are typically used when the referent has been mentioned a long time prior and is not very topical. This is in line with Givón (1983) and Ariel (1988, 2000, 2014). 7 Clausal syntax A clause is a grammatical unit that consists minimally of a predicate. A predicate is an element that takes a subject to form a clause, and it expresses something about this subject. The predicate is usually a verb, but can also be a noun, a demonstrative, or a quantifier. Verbal clauses are described in §7.1 and non-verbal clauses in §7.2. 7.1 Verbal clauses Each verbal predicate licenses a number of arguments. An intransitive verb requires one argument: a subject NP. A transitive verb requires a subject and an object NP. Ditransitive verbs require an indirect object NP in addition to a subject and object NP. Uruangnirin also has peripheral arguments, such as locations, comitatives or instrumentals, which are typically but not necessarily introduced by prepositions (given in §3). Complement clauses (§10.3) are another type of peripheral argument. The structure of a basic intransitive, transitive and ditransitive clause is presented in (52). (52) a. intransitive: S V b. transitive: A V O c. ditransitive: A V IO DO The subjects of active intransitive and transitive verbs are typically (but not obligatorily, see §6.2) indexed on the verb by means of a prefix. The subjects of stative intransitive verbs and objects are not indexed. When arguments are expressed in the clause, subjects precede the verb, while objects follow the verb. Typical intransitive and transitive clauses are given in (53)–(55). (53) stative intransitive [ai sa ka]S ririn som tree one DIST tall IAM ‘That one tree is tall already.’ [s-pohon-kky_175] (54) active intransitive [Senja laman]S a-ma som to Senja father 3-come IAM right ‘Senja’s father has already come, right?’ [c-tokipala-kb_65] (55) transitive [Amin tatang]A a-sei [sanong]O Amin grandparent 3-sew thatched_roof ‘Amin’s grandmother sewed the thatched roof.’ [n-rumah-kky_15–16] The only syntactically ditransitive verb is wanggi ‘give’. The theme follows the recipient, and neither are marked with case or adpositions. An example is given in (56). Other verbs that are trivalent (i.e., requiring three participants, but not all of them expressed as core arguments) are for example fasa ‘buy’, ftobir ‘send’, (f)penik ‘sell’ and sobang ‘send; order’. They are monotransitive and introduce the oblique participant with a preposition. (56) ditransitive [ia]A a-wanggi [sina]IO [wawa puan teni]DO 3SG 3-give 3PL mango fruit three ‘He gave them three mangoes.’ [s-pear-kky_67] The basic valency of verbs can be changed with the help of affixes and other strategies. These are discussed in §8. 7.2 Non-verbal clauses Non-verbal clauses have a predicate that is not a verb. This clause type is common in Uruangnirin, since nouns, demonstratives and quantifiers can act as the predicate. (57) shows two subsequent utterances from the same text. If we take the position in the clause as the parameter for determining which word acts as the predicate, (57a) has a nominal predicate (ai tultulis ‘tultulis wood’), and (57a) has a demonstrative predicate (dike ‘PROX’).’ (57) a. ai ke in nggaran ai tultulis wood prox 3SG.POSS name wood tultulis ‘This wood, its name is tultulis wood.’ [n-ujungkampung-in1_28] b. tultulis da dike tultulis REL PROX ‘It’s tultulis that’s here.’ [n-ujungkampung-in1_29] Examples (58) and (59) illustrate clauses with quantifiers as the predicate. In (58), speaker A asks about the number of boats that arrived to an event, using a structure akin to English ‘how many boats were they?’. The answer by speaker B contains two numeral quantifiers as predicates: first sa ‘one’ to indicate the number of boats, and then teni ‘three’ to indicate the number of people. In (59), both speakers use the non-numeral quantifier dum as the predicate of their clause. (58) A: sina rau fira 3PL canoe how_many ‘In how many canoes did they come?’ B: rau sa laf sina teni canoe one only 3PL three ‘Just one canoe, they were three [people].’ [c-kampung-sl-lk_12–13] (59) A: puan dum bi tai fruit many or neg ‘Are there many fruits?’ [a-panjatpala_130–132] B: ino puan dum wana wow fruit many friend ‘Wow, there are many fruits, friend.’ [[a-panjatpala_130–132] Many more examples of clauses with non-verbal predicates can be found in the archived questionnaire [e-cop-kky]. 7.2.1 Similative clauses Similative clauses are a subtype of non-verbal clauses. They are formed with pon(a) ‘like’, illustrated in the short conversation in (60). (60) A: manik ka pon kokok rorang bird DIST like chicken too ‘That bird is like a chicken.’ B: manik pon bebak ka to bird like duck DIST right ‘A bird like that duck, right?’ [a-tanamrica_353–354] 7.3 Interrogative clauses The question words, listed in Table 3 in Section 3, are illustrated in (61)–(65). These are all content questions (also called wh-questions or information questions). The question words fill the slot of the constituent they replace. (61) fauar afbe news how ‘How are you doing?’ [e-d1-kky_3] (62) sebenarnya sina fer be da ma actually 3PL from where REL come ‘Actually, where did they come from?’ [s-pear-skm_47] (63) im-amat pingan fira 2PL-bring plate how_many ‘How many plates did you bring?’ [a-tanampala-kky_240] (64) se da amin Kaimana who REL LOC Kaimana ‘Who is in Kaimana?’ [a-petikrica_243] (65) sina a-fun suai 3PL 3-do what ‘What are they doing?’ [e-pron-ab-kky_1] Polar questions are made with bi ‘or’. In (66), the speaker asks which way they should walk to the gardens. (66) fer nas bi fer pap from bottom or from top ‘Via the bottom or via the top?’ [a-tanamrica_33] Polar questions can also be made with bi tai ‘or not’, like in (67). (67) Ayu nenang o fonak kacang panjang erik bi tai Ayu mother.3POSS 2SG store bean long seed or NEG ‘Ayu’s mother, did you store long bean seeds?’ [a-petikrica_171] 7.4 Negation Clauses are negated with clause-final negator tai. (68) fisik a-ma tai some 3-come NEG ‘Some didn’t come.’ [n-taruhharta-kky_36] Non-verbal clauses are negated in the same way. Non-existence is expressed with negation of the existential verb mian ‘be there; live’. Impossibility is expressed with eran tai ‘cannot’. Eran cannot be used in positive contexts and is therefore not analyzed as a separate morpheme. (69) illustrates eran tai ‘cannot’. (69) kalo timur ke ia a-sa eran_tai if east PROX 3SG 3-ascend cannot ‘If [the] east [wind blows] it [the boat] cannot come up.’ [a-masak_167] For the interaction of the negator with aspectual modifiers, see §7.5. 7.5 Adverbial modifiers and phasal polarity There are many kinds of adverbial modifiers: there are words changing the mood, aspect or mode of a predicate or clause; or that specify the manner, temporal setting, degree or other characteristics of the state or event expressed by the verb, such as repetition or exclusivity. Most of the modifiers are clause-final. I only treat the most common ones here. Three common adverbs are laf ‘just; only’, panik ‘also; again’ and fayoi ‘well’. Laf ‘just; only’ ex- presses exclusivity and is illustrated in (70), which is directed at people speaking Malay entering the conversation. Panik ‘also; again’ expresses repetition. It is illustrated in (71), which is from a story about fishing techniques, where the speaker talks about going back and forth between different parts of the sea to employ different techniques. Fayoi ‘well’ is given in (72), modifying the verb tubi ‘grow’. (70) am-farak Wangnirin laf 1PL.EXCL-speak Wangnirin only ‘We speak Wangnirin only.’ [a-petikrica_167] (71) kanambo ma-lalaf ma-mna ma-tua panik then 1PL.EXCL-return 1PL.EXCL-come_landwards 1PL.EXCL-sinkerfish again ‘Then we return landwards and fish with sinkers again.’ [c-pancinganak-el_437] (72) ia tubi fayoi 3SG grow well ‘It grows well.’ [a-kebun-cam2-ak_113] There are three aspectual modifiers boit, efi and som, expressing the meanings ‘still; yet’, ‘not yet’, ‘already’ and ‘not anymore’. Words with these and similar meanings have received varying analyzes in the languages of eastern Indonesia. ‘Still’ is labelled as continuative and progressive; ‘not yet’ as incompletive, continuative negative, and nondum (van der Auwera 1998), and ‘already’ as perfective, perfect, non-progressive, iamitive (Olsson 2013, for a warning against overapplying this term, see Krajinović et al. 2025), and as realis. Only for ‘not anymore’ I haven’t found other terminology. Awaiting a more detailed semantic analysis of these words in Uruangnirin, I stick to the English translations in this section. However, I gloss som in examples with the catch-all gloss IAM to indicate that it may contain features of a perfect aspect, of ‘already’, and of a change-of-state. There are various possible combinations that do not logically follow from their individual meanings. Boit (often pronounced as bot, rarely as boti,11 and by some speakers as bota) is seldom used alone. If so, it means ‘still’. More than 95% of the time, it is combined with efi, meaning either ‘still’ or ‘not yet’. Efi used on its own also means either ‘still’ or ‘not yet’. With a negator in the clause, boit efi and efi also means ‘not yet’. Negated som means ‘not anymore’. The meanings and corresponding expressions are listed in Table 14. Table 15 sorts the same data by expression. Table 14: Aspectual markers boit, som and efi in affirmative and negative clauses: function → form. function form still boit Predicate Predicate efi boit Predicate efi not yet Predicate efi boit Predicate efi boit Predicate NEG efi Predicate NEG efi already; perfect Predicate som not anymore Predicate NEG som Table 15: Aspectual markers boit, som and efi in affirmative and negative clauses: form → function. form function boit Predicate still Predicate efi still; not yet boit Predicate efi still; not yet boit Predicate NEG efi not yet Predicate NEG efi not yet Predicate som already; perfect Predicate NEG som not anymore Figure 4 is a simplified version of the graphs used in van der Auwera (1998) to illustrate phasal polarity systems in the languages of Europe. The arrow illustrates time progressing from left to right. The polarity of the state referred to is given below the timeline. Above the timeline, the constructions and glosses are given. Figure 4: A schematic representation of the adverbial modifiers and polarity (73) boit ‘still’ ia boit a-mariri 3SG still 3-stand ‘He was still standing.’ [s-pear-un_27] (74) efi ‘still’ kantong amka efi to bag there still right ‘The bag is still there, right?’ [c-sp-lk-bk_63] (75) efi ‘not yet’ unang a-to efi to rain 3-descend not_yet right ‘It doesn’t rain yet, right?’ [n-jualan-an_79] (76) boit efi ‘still’ tompat ka boit a-mian efi container DIST still 3-exist still ‘That container was still there.’ [s-pear-skm_38] (77) boit efi ‘not yet’ daka bot n-edin efi deke DIST still 3-pluck yet PROX ‘That isn’t harvested yet, this.’ [c-kampung-sl-lk_1209] (78) boit NEG efi ‘not yet’ kalo boit punit tai efi if still finished NEG yet ‘If it isn’t finished yet...’ [n-perahu-kky_24] (79) NEG efi ‘not yet’ nenmatan stapak mian tai efi road concrete exist NEG yet ‘The concrete road wasn’t there yet.’ [c-longsor-bk_294] (80) som ‘already; perfect’ rau a-ngguret som canoe 3-strand IAM ‘The canoe is stranded.’ [e-trnstv-kky_75] (81) NEG som ‘not anymore’ sina a-tawar tai som 3PL 3-haggle NEG IAM ‘They didn’t haggle anymore.’ [n-minang-kky_73] Som is also an interjection used to soften imperatives. It is not unlikely that boit or efi had another meaning than they have now. Efi seems to be bleached in meaning, since it can be used on its own meaning both ‘still’ and ‘not yet’. Boit, on the other hand, must mean ‘still’ when used on its own. When combined with efi, it may mean ‘still’ or ‘not yet’. It might be currently used as a reinforcer of efi. It seems strange that there are so many options to express the meaning ‘not yet’. It is especially puzzling that the negator tai is optional. Without context, it seems impossible to know if (boit) + Predicate + efi should be interpreted as ‘still’ or ‘not yet’. Perhaps telicity or some other property of the predicate plays a role (Persohn 2024). A systematic review of the corpus examples, combined with an interpretation test with native speakers, is needed to make a more precise analysis. 8 Valency-changing strategies and related issues There are several strategies for changing the valency of a verb stem, some of which are by means of non-productive bound morphology, while others are analytic. To decrease the valency of a transitive stem, there is one strategy: • passive/anticausative constructions with ta(r)- and m(a)- To increase the valency of an intransitive stem, the following strategies and constructions are attested: • serial verb constructions • causative constructions with f(a)- (and sometimes -k) In addition, one can: • use the same verb stem for both transitive and intransitive constructions • use a different verb stem for the transitive construction than for the intransitive construction None of the four valency-changing morphemes (ta(r)-, m(a)-, f(a)- and -r) are productive. I only gloss them as separate morphemes when they attach to a stem that can also occur without the valency-changing morpheme or with another valency-changing morpheme instead. When this is not the case, such as for stative intransitive verbs like amdirin ‘cold’ (there is no stem dirin or causative form fa-dirin), I consider the form to be unanalyzable. Valency-decreasing prefixes are typically preceded by a-. Because valency-decreased constructions often have third-person subjects, it is un- clear whether this a- is part of the valency-decreasing prefix or whether it is the third person subject index. I analyze it as the latter. The three valency-changing prefixes all seem to be reflexes of proto-Austronesian: causative f(a)- seems to be a reflex of *pa- ‘causative of dynamic verbs’, passive/anticausative ta(r)- seems to be a reflex of *taR- ‘spontaneous or accidental action’ and passive/anticausative m(a)- seems to be a reflex of *ma- ‘stative’ (forms and functions from Blust 2013:371). Reflexes of these proto-Austronesian forms are widely attested across the Austronesian language family. I am not sure of the origin of the suffix -k. I will discuss valency-decreasing and valency-increasing strategies in turn in §8.1 and §8.2. 8.1 Valency-decreasing strategies There are two non-productive prefixes that decrease the valency of a verb by one: ta(r)- and m(a)-. They are typically employed to remove the agent/subject and promote the object/patient to become the subject. (82) and (83) are descriptions of cut & break clips (Bohnemeyer et al. 2001), where I asked the speaker to describe both the event and the result. (82) a. ia a-ris watpaning 3SG 3-rip cloth ‘She rips the cloth.’ [e-cut-bk2_36.1] b. watpaning a-m-ris cloth 3-PASS-rip ‘The cloth is ripped.’ [e-cut-bk2_36.2] (83) a. ia a-sanak botal in suman 3SG 3-open bottle 3SG.POSS cap ‘She takes the cap off the bottle.’ [e-cut-bk2_44.1] b. botal in suman a-ta-sanak som bottle 3SG.POSS cap 3-PASS-open IAM ‘The bottle’s cap is taken off.’ [e-cut-bk2_44.2] These constructions fill all the requirements for passives in Siewierska (2013), except that in Uruangnirin the agent of the active verb cannot be expressed as an oblique phrase to the passive verb, while in prototypical passive constructions this is an option. Constructions with ta(r)- and m(a)- can also be used as anticausatives, denoting spontaneous events without agents (Zúñiga and Kittilä 2019:41). (84) was used to describe a video of a rope that snaps in two without a visible agent. (85) is from a description of events preceding a landslide. There is no agent that broke the soil.12 (84) waras imisia a-m-ois fisik nua rope 3SG.alone 3-PASS-cut part two ‘The rope was cut in two on its own.’ [e-cut-bk2_46] (85) kawer a-ta-pawar soil 3-PASS-break ‘The soil cracked open.’ [c-longsor-bk_17] Both m(a)- and ta(r)- can be analyzed as middle markers following Inglese (2022), who uses this term as an overarching term for functions like passive and anticausative. In any case, constructions with ta(r)- or m(a)- make the patient prominent at the expense of the agent. I will gloss them as passives for the sake of simplicity, but keep in mind that they are non-prototypical passives. There are no roots that can be inflected with both ta(r)- and m(a)-. It is unclear at this point whether there is a difference in meaning between the two prefixes, or if the choice is lexically specified. 8.2 Valency-increasing strategies 8.2.1 Causative with serial verbs The most common valency-increasing strategy is a serial verb construction with funa ‘do; make’ or a movement verb like to ‘ascend’. In (86), a causer is added to the normally intransitive verb barsi ‘clean’ with the help of funa ‘do; make’. (86) a. ta-desil sampe nanam barsi 1PL.EXCL-plane until inside clean ‘We plane until the inside is clean.’ [c-minang-wn-nk_99] b. sina funa ia barsi 3PL make 3SG clean ‘They cleaned it.’ [n-kerjapp-an_621] Rwa is an intransitive verb that means ‘to lie down’ or ‘to sleep’. To add an object, both funa ‘do; make’ and to ‘descend’ are used, as illustrated in (87). (87) a. kasuk a-rwa man 3-lie_down ‘The man is lying down.’ [e-trnstv-kky_26] b. ta-funa kasuk a-to a-rwa 1PL.EXCL-make man 3-descend 3-lie_down ‘We laid the man down.’ [e-trnstv-kky_27] Many other examples are with movement verbs like to ‘descend’ (as in 88) and sa ‘ascend’, but there are plenty of examples with other intransitive verbs such as akabom ‘wet’, tubi ‘grow’, amnipis ‘thin’, lalaf ‘return’, and pura ‘flee; run’, the latter illustrated in (89). (88) sina a-fun namnamin a-to 3PL 3-make food 3-descend ‘They served the food.’ [n-taruhharta-kky_105] (89) iyo kokok ka peta sina funa ia pura am ka bo yes chicken DIST earlier 3PL make 3SG flee LOC DIST but ‘Yes, earlier they shooed away that chicken, but... [now it has come back].’ [c-pancinganak-el_301] SVCs with funa ‘do; make’ are also used to make reciprocal constructions, see §8.3. Other complex predicates are discussed in §9. 8.2.2 ‘Applicative’ with serial verbs Uruangnirin has no applicative morphology, i.e., no morphology to introduce a peripheral argument into the clause as an object. Peripheral arguments are expressed by one of the three prepositions: afara ‘from’, amina ‘LOC’ or nora ‘with’. Serial verb constructions can be employed to introduce peripheral arguments, for example the goal of a verb of motion. Serial verb constructions are also used to introduce goals and locations. Motion verbs like ti ‘go’, to ‘descend’ or sa ‘ascend’ are used as the second verb in these constructions. These constructions are most common in the data where speakers are asked to describe short video clips, specifically the put videos (Bowerman et al. 2004) and the Surrey pronominal marking clips (Fedden et al. 2010). In (90) and (91), a goal is added to transitive verbs with ti ‘go’ and to ‘descend’, respectively. (90) minlen ka a-fonak gambar a-ti patpatak woman DIST 3-put picture 3-go wall ‘That woman puts a picture on the wall.’ [e-put-kky_6] (91) a-bantu ia a-seik ini wawa a-to ni anggara 3-help 3SG 3-fill 3SG.POSS mango 3-descend 3SG.POSS basket ‘They helped him fill his basket with mangoes.’ [s-pear-kky_60] For these examples, the result (the picture is on the wall, the mangoes are in the basket) would be described with the locative preposition amina. In (92), a location is added to the verb sandar ‘lean’. This verb is intransitive with an inanimate subject and reflexive or transitive with an animate object. (92) a. neran a-sandar a-ti patpatak stairs 3-lean 3-go wall ‘The stairs lean against the wall.’ [e-trnstv-kky_87] b. a-sandar ia a-ti in laman 3-lean 3SG 3-go 3SG.POSS father ‘He leans on his father.’ [e-pron-ab-kky_10] The constructions in (92) can be replaced with a preposition. Consider the examples in (93), which are semantically very close but introduce the location with (a short form of) the locative preposition amina instead. (93) a. fafan a-sandar am dinding plank 3-lean LOC wall ‘The plank leans against the wall.’ [e-trnstv-kky_87] b. a-sandar ia amin patpatak 3-lean 3SG LOC wall ‘He leans against the wall.’ [e-pron-ab-kky_19] While applicatives in general function to include topical arguments into the core of the clause, it is not yet clear what governs the choice of an ‘applicative’ SVC versus a preposition in Uruangnirin. However, the fact that they are particularly common when speakers are asked to describe short video clips, suggests that topicality plays a role. The video clips have no context and display only one event, and so every argument is equally topical. But even these situations do not force the use of an ‘applicative’ SVC: (93b) is a description of a Surrey pronominal marking clip. Finally, note that objects are often left out in ‘applicative’ SVCs. Constructions with seik ‘fill’ and to ‘descend’ (like 91) are found more than thirty times in the corpus, of which at least twenty are from renditions of the Pear story (Chafe 1975). In most of the constructions, the object (the fruit) is very clear from the context and therefore left out. 8.2.3 Causative f(a)- and transitivizing -k The two valency-increasing morphemes are f(a)- and -k. F(a)- is a causative that is attested on two verb stems and a noun: riri ‘stand’, tonang ‘sit’, and masi ‘smoke (n)’.13 -k seems to be a transitivizer. It is only attested on two verb stems and a noun, all of which end in -i: fini ‘choose; pick’, riri ‘stand’ and masi ‘smoke (n)’. Note that two of these also occur with causative f(a)-, and that f(a)- and -k then must occur simultaneously. Neither of the valency-increasing affixes are productive. Example (94) shows the verb stem tonang ‘sit’. In (94a) there is only a subject (and valency-decreasing m-, discussed in §8.1), and in (94b) there is a subject and object, and the verb is marked with causative f-. (94) a. lau i-m-tonang gigira ka nor Ayu nenang 1SG 1SG-PASS-sit yesterday DIST with Ayu mother.POSS ‘Yesterday, I sat with Ayu’s mother.’ [c-minang-wn-nk_99] b. ta-tna ta-f-tonang wai 1PL.INCL-go_landwards 1PL.INCL-CAUS-sit thing ‘We went landwards to put down things.’ [n-iblismesjid-bk_7] Riri ‘stand’ is typically used with one argument, and is then inflected with m(a)- (discussed in §8.1). When it is used with two arguments, it carries both causative f(a)- and transitivizer -k. The transitive variant of riri can only be used with an inanimate object. (95) a. minlen ka a-ma-riri a-sanak ni topi afar puatin woman DIST 3-PASS-stand 3-take_off 3SG.POSS hat from head ‘That woman is standing and takes off her hat from her head.’ [e-put-kky_47] b. a-f-riri-k kapuni teni peta amina masikit tawanang 3-CAUS-stand-TRNSTV bamboo three ANA LOC mosque side ‘He stood those three bamboos next to the mosque.’ [n-iblismesjid-bk_59] Neither tonang nor riri can be used without a valency-changing morpheme. The forms with m(a)- function as normal intransitives; they are not passives or transitivizer forms. There are no examples of stems that are unmarked for valency in the intransitive construction and then causativized with f(a)- and/or -k in the transitive construction. The verb fmasik ‘smoke-dry; smoke-cook (v)’ seems to be derived from masi ‘smoke (n)’. There is no verb masi. (96) illustrates fmasik ‘smoke-dry; smoke-cook (v)’ and also shows it can be inflected for first person singular. (96) li-f-masi-k sayang amin ber 1SG-CAUS-smoke-TRNSTV nutmeg LOC smokedry_place ‘I smoke-dry nutmeg in the smoke-dry place.’ [dictionary_afmasik] Note that (fossilized remnants of) the morphemes fa- and (a)f- are attested on many other verbs, some of which show alternation between the two forms. This is discussed in §2 and more elaborately in Visser and Stickl (2025). 8.2.4 “Suppletion” Finally, there are a few sets of completely different roots for the transitive and intransitive variant of the same meaning. An example is ‘sharp’. The intransitive is mangan. There are three transitive counterparts, none of which are derived from or similar to the intransitive: asapit, asinggar and alimar, all of which mean ‘sharpen’. This is somewhat comparable to verbs with m(a)- and ta(r)- which can be opposed to verbs with the same root or to verbs with a different root, as discussed in §8.1. 8.3 Reciprocal and reflexive constructions Reciprocal constructions are made with funa ‘do; make’ and reduplicated verbs. (97) sina-nua a-fun pang∼pangga 3PL-two 3-do RECP∼hit ‘They two hit each other.’ [e-rec-ab-kky_7] There are a few inherent reflexive verbs: these must be followed by an object pronoun that is coreferential with the subject when used reflexively. They are terik ‘to rest’, paning ‘turn’ and serin ‘change clothes’. The latter two are illustrated in (98) and (99). In the case of a third person subject, the construction is ambiguous between a reflexive and a transitive reading. (98) ta-paning ita ta-tno som 1PL.INCL-turn 1PL.INCL 1PL.EXCL-go_seawards IAM ‘Let’s turn around and go seawards.’ [n-iblismesjid-bk_36] (99) baba Haji a-bana a-serin ia father Haji 3-go 3-change 3SG ‘Father Haji went to change.’ [n-iblismesjid-bk_72] Otherwise, reflexive-like constructions can be made with a restrictive focus pronoun imisia ‘3SG.alone’, which might be translated with ‘self’ or ‘own’ in these cases. When presented with pictures of people combing, shaving, washing and otherwise grooming or doing things to others and themselves, my consultant often offered a clause with imisia for the reflexive actions, although it seemed optional in all cases. (100) mapata ke imisia a-sisir puk woman PROX 3SG.alone 3-comb hair ‘This woman combs her (own) hair.’ [e-refl-in_10] 9 Complex predicates In this section I describe verb-based complex predicates, which are constructions “serving as a predicative constituent to a single syntactic subject, and involving at least two lexical items belonging, either synchronically or historically, to the class of verbs” (Vanhove et al. 2021). A typical construction captured by this definition is given in (101), with amriri namat ‘stands holding’. (101) marwana ka a-mriri n-amat ini pingan man DIST 3-stand 3-carry 3SG.POSS plate ‘That man stands holding his plate.’ [e-se2a-kky-bk_77] Not captured by this definition are constructions with two contiguous verbs that express sequential events, like amriri abana ‘stands up and walks’ (102). (102) kasuk sa a-mtonan amin patpatak langsung ia a-mriri a-bana man one 3-sit LOC wall directly 3SG 3-stand 3-walk ‘A man sits at a wall. Immediately, he stands up and walks.’ [e-pron-ab-kky_144] Some of the constructions in the following sections may involve verb forms that have been termed co(n)verbs, light verbs, or auxiliary verbs in other works, but I do not wish to go into terminological detail here. Likewise, I make no analysis of whether some or all of these constructions would count as serial verb constructions following definitions such as those given in Aikhenvald and Dixon (2006) or Haspelmath (2016). The focus of this section is on the functions of all verb-based complex predicates. For now, I distinguish seven functional subtypes of complex predicates. There may be more subtypes, such as instrumental complex predicates, but this requires a more in-depth analysis of the data. In the following subsections, I give at least one example for each subtype and briefly comment on its morphological and syntactic properties. In the description, I use V1 for the first verb and V2 for any subsequent verb. Most complex predicates contain two verbs, but more is not uncommon either. Several examples are provided elsewhere in the text, e.g., (111) and (114). I comment little on argument sharing between the verbs, because this issue has not been investigated in detail yet. The same goes for polarity, but in general I can say that the negator tai must come after the second verb to negate the complex predicate. 9.1 Aspect and modality Completive aspect is expressed with punit ‘be finished’, illustrated in (103). (103) ia sofa punit ia namnamin 3SG bathe finish 3SG eat ‘After he finished bathing, he ate.’ [c-longsor-bk_20] Punit is always the V2. An object may precede the V2. It is unclear to what extent arguments are shared between the two verbs. The expression of completive aspect with punit is extremely productive and is found with a wide range of verbs. Punit is a verb that can be used on its own (illustrated in 104), but like other stative intransitives (§7.1) it is never inflected for person. (104) nanti fan hidup ka punit ma-pake jarutu panik later bait living DIST finished 1PL.EXCL-use sinker again ‘Later when the living bait is finished, we use sinkers again.’ [c-pancinganak-el_146–147] Other verbs that could possibly be involved in forming aspectual complex predicates are otin ‘continue’, terus/tarus ‘continue’ (< Malay), aftok ‘stay’ and anggait ‘arrive’. These need to be investigated further. All or some of the words14 bisa ‘can’, mau/mo ‘want’, harus ‘must’, eran tai ‘cannot’ and pora ‘PROH’ can be said to enter into complex predicate constructions. Of these, all except eran tai ‘cannot’ can be inflected for subject. Eran tai ‘cannot’ (discussed in §7.4) only occurs in negated clauses and should perhaps be analyzed as an adverb and not as a verb. Bisa ‘can’ and mau/mo ‘want’ are the clearest examples of modal complex predicates and are illustrated in (105) and (106). (105) kanambo bisa ita tafeda jaring irok then can 1PL.INCL throw net shark ‘Then we can throw shark nets.’ [c-pancinganak-el_285] (106) kasuk dike a-mau a-sa in Haji man PROX 3-want 3-ascend 3POSS Hajj ‘This man wants to go on Hajj.’ [a-tanampala-kky_272] The object follows both verbs in constructions with bisa ‘can’ and mau/mo ‘want’. Bisa seems to have a variable placement: in (105) it comes before the subject, but it may also follow it, like in (107). (107) ia ma kanambo ita bisa mariri 3SG come then 1PL.INCL can stand ‘When she comes, we can stand.’ [c-minang-wn-nk_1118] Constructions with bisa ‘can’, mau/mo ‘want’, harus ‘must’, eran tai ‘cannot’ and pora ‘PROH’ are very productive. Mau/mo seems to have taken over the function of the clause-final future marker ni. Ni ‘FUT’ is sometimes used in combination with mau/mo and sometimes on its own, but has a much lower frequency than mau/mo. 9.2 Resultative Resultative complex predicates combine an action (V1) and its result (V2). This is an uncommon type in the corpus, but (108) illustrates it quite well. While V1 wanang ‘to fell’ expresses an ongoing action, V2 amrepa ‘felled’ expresses a state, the result of felling (see §8.1 for comments on the passive). (108) lau da i-wanang ai ka amrepa 1SG REL 1SG-fell tree DIST felled ‘It was me who felled that tree [until it] was felled.’ [s-pohon-kky_159] In resultative complex predicates, the object comes between V1 and V2. The object of V1 is the subject of V2. It’s unclear how productive this type is. Most corpus examples are with stative intransitive verbs as V2. The exception is a-tonggar n-ois ‘pulls breaks’, but this example could be interpreted as sequential. It is illustrated in (109), which also shows the result with mois ‘be cut’. (109) a. ia n-ana kabas a-tonggar n-ois fisik nua 3SG 3-get thread 3-pull 3-cut part two ‘He gets a thread and cuts [it] in two (by pulling it).’ [e-cut-bk2_38.1] b. kabas a-mois fisik nua thread 3-be_cut part two ‘The thread is cut in two.’ [e-cut-bk2_38.2] 9.3 Valency-changing Valency-changing complex predicates create causative and applicative constructions. They are also described in §8.2.1 and §8.2.2, respectively. Causatives are formed with funa ‘do; make’ as the V1, followed by the introduced object and the V2. These could be seen as a type of resultative complex predicates, since the V2 expresses a result. Unlike the examples found in that type, however, the V2 in causative complex predicates may be active intransitive, like to ‘ascend’ in (110). In that example, the object ‘the fruits’ is elided, but it would go in between the verbs. In (111) the object is expressed and the V2 is a stative verb. The object of V1 is the subject of V2. (110) ia fun a-to anggara 3SG do 3-descend basket ‘He put [the fruits] in a basket.’ [s-pear-un_2] (111) jadi goni fat li-mau i-fun ia amdekar so sack four 1SG-want 1SG-do 3SG dry ‘So those four sacks, I want to dry them.’ [n-pala-nk1_8] Applicatives are made with the V2 as the applicativizing verb, followed by the applied object. illustrates an applicative construction, where a location is introduced with the help of the V2 bana ‘go; walk’. The fact that the V2 isn’t inflected for person and is shortened to ban might indicate that this verb in this use is developing into a preposition. On the other hand, elision of subject inflection is quite common, and final /a/ deletion, too, so the evidence is not conclusive. (112) berati ma-fnaka ban Tonggarai doko that_means 1PL.EXCL-sail go Tonggarai F_DIST ‘That means we sail to Tonggarai over there.’ [c-pancinganak-el_264] It is yet unclear how productive this type is, and which V2’s can introduce an applied object. Verbs that are found so far are bana ‘go; walk’ and ti ‘go’. Other possible candidates are resi ‘exceed’ and nora ‘follow; join; bring’. 9.4 Direction In directional complex predicates, the V2 specifies the direction in which the V1 is performed. This is an extremely common type, and it is found with at least the following directional verbs: ma ‘come’, ti ‘go’, mno ‘come seawards’, mna ‘come landwards’, tno ‘go seawards’, mna ‘go landwards’, wara ‘go uphill’, poti ‘go downhill’, sa ‘ascend’, to ‘descend’, bana ‘go; walk’ and lalaf ‘return’. These are all verbs that can be used on their own, too. (113) is a transitive example, where the object of the V1 is the subject of the V2. (114) is intransitive and has two V2’s, with perhaps a lexicalized meaning. (113) ia sobang a-seik wer a-to kapuni nanam 3SG order 3-fill water 3-descend bamboo inside ‘He ordered to fill the bamboo with water.’ [n-iblismesjid-bk_43] (114) ia a-fika ti a-ma 3SG 3-look go 3-come ‘He looked back and forth.’ [e-pron-ab-kky_109] 9.5 Manner Manner complex predicates consist of two verbs that express simultaneous events, whereby the V2 expresses the manner in which V1 is performed. They are typically formed with stative intransitive verbs like tuari ‘be long’ dum ‘be many’ and manonang ‘be true’, which can all be used as independent verbs. The second verb specifies the action of the first verb, and can’t be said to have a shared subject. Remember also that stative intransitives are not inflected for subject in Uruangnirin (§6.1). The order of transitive manner complex predicates is V1 V2 O. An example (without object) is given in (115). (115) katimun ke ka fian ia lakir olat cucumber PROX DIST good 3SG crawl far ‘This cucumber [plant] is good, it crawls far.’ [a-tanamrica_266] 9.6 Lexicalized In complex predicates that are lexicalized, the meaning of the predicate can’t be deduced from its parts anymore. A good example is osir ‘say’ + wanggi ‘give’, which means ‘let know; tell’. Another construction has osir ‘say’ as the second verb, where it can mean ‘follow’, illustrated in (116). Osir ‘follow’ can only occur in complex predicates, never on its own. (116) ia fika n-osir sina 3SG see 3-say 3PL ‘He follows them with his eyes.’ [s-pear-un_41] 9.7 Purpose This is a tentative subtype of complex predicates, where the V2 expresses the purpose of the action expressed in V1. In many cases where it is natural to paraphrase the verbs in English as ‘V1 to V2’ (‘go down to sell’, ‘meet to collect’), one could alternatively analyze them as two sequential predicates ‘V1 then/and V2’ (‘go down and sell’, ‘meet and collect’). Some, however, must be interpreted as one event. (117) has patar ‘stay up late’ and fun jualan ‘make food to sell’. This definitely doesn’t mean ‘stay up late, then make goods to sell’. (117) lai-patar fun jualan 1SG-stay_up_late make food_to_sell ‘I stay up late to make food to sell.’ [n-jualan-an_19] 10 Clause coordination and complement clauses Clause coordination is described in §10.1, tail-head linkage in §10.2 and complement clauses in §10.3. 10.1 Clause coordination The five most frequent clause coordinators are bo ‘and; but’, jadi ‘so’, bi ‘or’, kanambo ‘then’ and da ‘then; so’. Bo is mostly used with the meaning ‘but’, but can also mean ‘and’. These meanings are illustrated in (118) and (119), respectively. (118) bunga dike Arman a-futir a-bana Sorong Pakpak a-bana Makassar bo a-sngangga tai flower PROX Arman 3-search 3-go Sorong Fakfak 3-go Makassar but 3-find NEG ‘This flower [a prospective wife] Arman looked for in Sorong, in Fakfak, in Makassar, but he didn’t find [her].’ [n-minang-kky_46] (119) o ta-ti kapala bo o m-ana kumar sa amka INT 1PL.INCL-go house and 2SG 2SG-get sachet one there ‘Oh, we go to the house and you get one sachet [of tobacco] there.’ [n-taruhharta-kky_129] Jadi ‘so’ is a loan from Papuan Malay or Indonesian and is illustrated in (120). (120) sina am Nusa jadi fatek sina tari lau 3PL LOC Tuburuasa so later 3PL bring 1SG ‘They are in Tuburuasa, so later they’ll bring me.’ [c-kampung-sl-lk_134] Bi ‘or’ is illustrated in (121). It is also used as a question tag. (121) kasuk ke wara som bi efi man PROX go_uphill IAM or not.yet ‘Did this man go up yet or not?’ [a-petikrica_49] Kanambo ‘then’ is used to link clauses sequentially, as in (122). It can be shortened in a myriad of ways: kabo, kam, kambo, kamo, kana, kanam, kanamo and nambo are all attested. There is no evidence for kanambo being multimorphemic, despite it being longer than the other frequent coordinators and it seemingly containing bo. (122) ma-tafeda et kanambo tinggal ma-wewak laf 1PL.EXCL-throw anchor then what_remains_is 1PL.EXCL-fish just ‘We throw the anchor, then all we need to do is fish.’ [c-pancinganak-el_54] Da is used either as a sequential coordinator (‘then’) or as a purposive coordinator (‘so’). These uses are illustrated in (123) and (124). (123) lai-wara i-nois kananggiar ka da li-bana poti fika neno 1SG-go_uphill 1SG-cut bamboo DIST then 1SG-go go_downhill see wow ‘I went uphill to cut that kananggiar bamboo, then I went downhill and saw: “Wow!”’ [c-kasuksuk_29] (124) bana far ke da nawan_afeni walk from PROX so be_short_of_breath ‘I walked from [via] here, so I’m short of breath.’ [a-petikrica_26–27] Borrowed coordinators from Papuan Malay/Indonesian include karena ‘because’, sehingga ‘so that’ and terus ‘then’. 10.2 Tail-head linkage Clauses are also commonly combined with tail-head linkage, the repetition of the last clause or part of a clause at the beginning of the next clause (de Vries 2005). (125a) shows the repetition of the verb and object from (125b). This is the most common type, although it is also possible to repeat the entire clause. (125) a. terus ami undangan ruang taransa then 1PL.EXCL invite village seven ‘Then we invited seven villages. [n-maulud-ak_5] b. undangan ruang taransa ma Kiaba untuk Maulud invite village seven come Kiaba for Maulud ‘Invited seven villages to come to Kiaba for Maulud.’ [n-maulud-ak_6] 10.3 Complement clauses It is hard to determine whether Uruangnirin has complement clauses, because there are no infinitives, subject marking is (almost) always optional, and there are no complementizers. There is one exception: the speech verb complementizer o, illustrated in (126). (126) jadi fisik n-osir o ita a-sa am Goran ita so some 3-say QUOT 1PL.INCL 3-ascend LOC Gorom 1PL.INCL por terlalu dum PROH too many ‘So some said: “[If] we go to Gorom, we shouldn’t be to many.”’ [n-minang-kky_135–136] To introduce speech or thought, osir ‘say’ can be preceded by another speech or perception verb. The same construction is found in Kalamang (Visser 2022:329,392). It seems that osir must be inflected for generic/third person, even though the subject of both verbs seems to be the same. An example is given in (127). (127) da li-pikir n-osir o [...] so 1SG-think 3-say QUOT ‘So I thought: [...].’ [c-kampung-sl-lk_1255] A possible other kind of complement clause is with the verb rasa ‘feel’, which seems to require another clause as a complement, but I do not at this point know enough about the behavior of this verb to conclude. It is illustrated in (128). (128) o mi-rasa be da fian 2SG 2SG-feel which REL good ‘You feel which one is good.’ [a-tanampala-kky_39] 11 The Uruangnirin corpus and other materials All Uruangnirin language materials collected by me are archived at ELAR (Visser et al. 2023). The collection contains: • 15 hours of transcribed recordings of naturalistic speech of different genres, including narratives, conversations and activities filmed with action cameras (more than 8 hours of transcribed segments within these recordings) – all transcriptions are translated to English and Papuan Malay • an additional 47 hours of untranscribed communicative events with ample metadata in English and Papuan Malay • recordings of elicitation sessions on a wide range of topics • hundreds of pictures of different events • an Uruangnirin – Papuan Malay – English dictionary with more than 400 illustrating pictures, also available as Android application • a Fieldworks project file which contains the whole lexicon (3300 lexemes) and all 140 translated and glossed texts, both with extensive notes • sociolinguistic questionnaires with more than 20 speakers and a cultural questionnaire All items in the collection can be filtered by among other things genre (e.g., narrative or notes), topic (e.g., agriculture or history), tags (e.g., fruit or government) and file type. Most files are open access, while a few files can be accessed after creating a user account with ELAR. An annotated text collection of narratives about life and death is Visser (2026). A language snapshot, with a short introduction to the corpus, can be found in Visser (2025b). 12 Outlook I am working on a grammar of Uruangnirin, but it is unclear when it will appear. In the meantime, I hope this sketch answers some of the most basic questions about this previously undescribed language. Readers who want to know more about a specific topic are encouraged to take a look at the archived materials, in particular the Fieldworks project. You can also contact me directly with any kinds of question you might have. Abbreviations 1, 2, 3 first, second, third person A subject in a transitive clause ANA anaphoric CAUS causative CONT continuative DIST distal DO direct object EMPH emphatic EXCL exclusive F_DIST far distal HES hesitative IAM iamitive INCL inclusive INT interjection IO indirect object LOC locative NEG negation NP noun phrase NUM_LNK numeral linker O/Obj object PASS passive PL plural POSS possessive poss’ed possessed poss’or possessor PROX proximal QUOT quotative RECP reciprocal REL relative clause marker S subject in an intransitive clause SG singular Subj subject TOP topic TRNSTV transitivizer V verb References Aikhenvald, Alexandra Y. and R. M. W. Dixon, ed. 2006. Serial verb constructions: A cross-linguistic typology. Oxford: Oxford University Press. Anceaux, Johannes Cornelis. 1958. Languages of the Bomberai Peninsula: Outline of a linguistic map. Nieuw-Guinea Studiën 2:109–121. Ariel, Mira. 1988. Referring and accessibility. Journal of linguistics 24:65–87. Ariel, Mira. 2000. The development of person agreement markers: From pronouns to higher accessibility markers. In Usage-based models of language, ed. M. Barlow and S. Kemmer, 197–260. Stanford: CSLI. Ariel, Mira. 2014. Accessing noun-phrase antecedents. London: Routledge. van der Auwera, Johan. 1998. Phrasal adverbials in the languages of Europe. In Adverbial constructions in the languages of Europe, ed. Johan van der Auwera and Dónall P. Ó Baoill, 25–145. Berlin/New York: Mouton de Gruyter. Blust, Robert. 1993. Central and central-eastern Malayo-Polynesian. Oceanic Linguistics 32:241–293. Blust, Robert A. 2013. The Austronesian languages. Number A-PL 008 in Asia-Pacific Linguistics. Canberra: Research School of Pacific and Asian Studies, Australian National University, 2nd edition. Revised edition. http://hdl.handle.net/1885/146289 Bohnemeyer, Jürgen, Melissa Bowerman, and Penelope Brown. 2001. Cut and break clips. In Manual for the field season 2001, ed. Stephen C. Levinson and Nick J. Enfield, 90–96. Nijmegen: Max Planck Institute for Psycholinguistics. Bowerman, Melissa, Marianne Gullberg, Asifa Majid, and Bhuvana Narasimhan. 2004. Put project: The cross-linguistic encoding of placement events. In Field manual, ed. Asifa Majid, volume 9, 10–24. Nijmegen: Max Planck Institute for Psycholinguistics. Chafe, Wallace. 1975. The pear film. University of California at Berkeley. https://drive.google.com/file/d/1jF6vtUdOlBN9LFzJ_NujWfP9Ef6ObK9K/view Eberhard, David M., Gary F. Simons, and Charles D. Fennig, ed. 2020. Ethnologue. Dallas: SIL International, 23rd edition. http://www.ethnologue.com Eriksson, Samuel. 2026. Uruangnirin demonstratives: A study of endophoric and non-standard exophoric uses. Bachelor’s thesis, Uppsala University. Faghiri, Pegah, and Eline Visser. 2025. Subject realization in Uruangnirin. STUf 78:573–606. Fedden, Sebastian, Dunstan Brown, and Greville G. Corbett. 2010. Conditions on pronominal marking: A set of 42 video stimuli for field elicitation. Gaby, Alice. 2008. Distinguishing reciprocals from reflexives in Kuuk Thaayorre. In Reciprocals and reflexives: Theoretical and typological explorations, ed. Ekkehard König and Volker Gast, 259–288. Berlin, New York: Mouton de Gruyter. Givón, Talmy. 1983. Topic continuity in discourse. Amsterdam, Philadelphia: John Benjamins Publishing Company. Givón, Talmy. 1983. Introduction, 3–41. Philadelphia: John Benjamins. Grimes, Charles E., and Owen Edwards. 2026. The Austronesian languages of eastern Indonesia and Timor-Leste. Languages of the New Guinea Region. Berlin: Language Science Press. Gunnarsson, Nils. 2025. An initial investigation of intonation in Uruangnirin. Bachelor’s thesis, Uppsala University. https://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-553077 Haspelmath, Martin. 2016. The serial verb construction: Comparative concept and cross-linguistic generalizations. Language and Linguistics 17:291–319. Inglese, Guglielmo. 2022. Towards a typology of middle voice systems. Linguistic typology 26:489–531. Jacobson Holmström, Tilda. 2025. Reevaluating the Classification of Nuta: A study of Arguni, Erokwanas, and Bedoanas within the Central Malayo-Polynesian subgroup of the Austronesian language family. Bachelor’s thesis, Uppsala University. Jacobson Holmström, Tilda. 2026. A sketch grammar of Mbaruan. Master’s thesis, Uppsala University. Kamholz, David. 2014. Austronesians in Papua: Diversification and change in South-Halmahera-West New Guinea. Doctoral Dissertation, University of California, Berkeley, CA. Kaplan, Johanna. 2026. Bomberai tip as a language chain: An analysis of lexical and phonetic distance in Onin, Sekar and Uruangnirin. Bachelor’s thesis, Uppsala University. Klamer, Marian. 2014. Numeral classifiers in the Papuan languages of Alor and Pantar. In Number and quantity in East Nusantara. Papers from 12-ICAL, Volume 1, ed. Marian Klamer and František Kratochvíl, 103–122. Canberra: Asia-Pacific Linguistics. Krajinović, Ana, Jozina Vander Klok, and Kilu von Prince. 2025. When (not) to establish a new category. The case of perfect, ‘already’, and iamitives. Studies in Language 49:546–586. Levin, Theodore, and Maria Polinsky. 2019. Morphology in Austronesian languages. In Oxford research encyclopedia of linguistics. Oxford: Oxford University Press. Lewis, M Paul, and Gary F Simons. 2010. Assessing endangerment: expanding Fishman’s GIDS. RRL LV:103–120. Mettler, Heidi, and Toni Mettler. 1997. Yamdena grammar notes. Ambon. https://www.sil.org/resources/archives/94080 Milano, Pietro. 2025. A sketch grammar of Arguni. Master’s thesis, Utrecht University. https://studenttheses.uu.nl/handle/20.500.12932/49378 Olsson, Bruno. 2013. Iamitives: Perfects in Southeast Asia and beyond. Master’s thesis, Stockholm University. https://su.diva-portal.org/smash/record.jsf?pid=diva2%3A633203&dswid=4595 Pattipeiluhu, John, and Eline Visser. ms. A grammar sketch of the endangered Goran language of Gorom Island, East Seram Regency, Maluku, Indonesia. Persohn, Bastian. 2024. When ‘still’ means ‘not yet’. STUf 77:83–39. Riesberg, Sonja, Maria Bardají i Farré, Kurt Malcher, and Nikolaus P Himmelmann. 2022. Predicting voice choice in symmetrical voice languages: All the things that do not work in Totoli. Studies in Language 46:453–516. Rodehav, Beatrice. 2026. Analysis of the words afara, anora & amina in Uruangnirin. Bachelor’s thesis, Uppsala University. Schapper, Antoinette, and Emily Gasser. 2023. Adnominal possession in the languages of Wallacea: a survey. STUf 76:273–329. Sibakov, Henna. 2024. Reduplication in Uruangnirin. Course paper, Leiden University. Siewierska, Anna. 2013. Passive constructions (v2020.3). In The world atlas of language structures online, ed. Matthew S. Dryer and Martin Haspelmath. Zenodo. https://zenodo.org/records/7385533 Sneddon, James Neil, Karl Alexander Adelaar, Dwi N. Djenar, and Michael Ewing. 2012. Indonesian: A comprehensive grammar. London/New York: Routledge. Usher, Timothy, and Antoinette Schapper. 2022. The Greater West Bomberai language family. Oceanic Linguistics 61:er1–er59. Vanhove, Martine, Walter Bisang, Andrej Malchukov, and Alexandre François. 2021. ComPLETE: Complex Predicates in Languages: Emergence, Typology, Evolution. Presentation of ComPLETE, a project funded by the ANR and DFG (2022-2025). https://complete.huma-num.fr Visser, Eline. 2020. Tracking and recognitional use of Kalamang opa: Demonstrative of cognitive accessibility. In Demonstratives in discourse, ed. Åshild Næss, Anna Margetts, and Yvonne Treis, 103–121. Berlin: Language Science Press. Visser, Eline. 2022. A grammar of Kalamang. Number 4 in Comprehensive Grammar Library. Berlin: Language Science Press. Visser, Eline. 2024. A grammar sketch of Yamdena and an introduction to its corpus. Language Documentation & Conservation 18:67–108. Visser, Eline. 2025a. Patipi Onin (Onim) language recordings. https://catalog.paradisec.org.au/repository/EV2 Visser, Eline. 2025b. Uruangnirin (Indonesia) – language snapshot. Language Documentation and Description 24:1–7. Visser, Eline. 2026. Life and death on Karas. Number 9 in Open Text Collections. Berlin: Language Science Press. Visser, Eline, Kamal Kanabaraf, Rustam Kanabaraf, Salis Maswatu Kanabaraf, and Safrudin Tianotak. 2023. Uruangnirin language documentation. Berlin: Endangered Languages Archive. http://hdl.handle.net/2196/dd459a7c-2493-4d26-9801-cc071a72bd34 Visser, Eline, and Artúr Stickl. 2025. A phonological sketch of Uruangnirin, and endangered Austronesian language of Eastern Indonesia. Journal of the Southeast Asian Linguistics Society 2:xxiii–lxiv. de Vries, Lourens. 2005. Towards a typology of tail-head linkage in Papuan languages. Studies in Language 29:363–384. Wijaya, Dendi. 2025. Sekar language documentation. https://catalog.paradisec.org.au/collections/DW1 Wilkins, David. 2004. The 1999 demonstrative questionnaire: “this” and “that” in comparative perspective. In Manual for the 1999 field season, ed. by David Wilkins, volume 9, 1–24. Nijmegen: Max Planck Institute for Psycholinguistics. Zúñiga, Fernando, and Seppo Kittilä. 2019. Grammatical voice. Cambridge: Cambridge University Press Reviewed: Received 20 September 2025, revised text accepted 16 February 2026, published 1 May 2026 Editors: Editor-In-Chief Dr Mark Alves | Managing Eds. Dr Paul Sidwell, Dr Sigrid Lew, Dr Mathias Jenny 1 This article was written with support from the Wenner-Gren Foundations, Endangered Languages Documentation Program and the Indonesian National Research and Innovation Agency (BRIN). 2 I would like to thank the Kiaba and Tarak communities for their hospitality, especially Badarudin Kanabaraf in Kiaba and Nurlia Patur in Tarak, who hosted me in their homes. Kamal Kanabaraf Yorkuran, Badarudin Kanabaraf, Nur Ain Kanabaraf and Abdul Karim Samay provided much of the data. Safrudin Tianotak, Rustam Kanabaraf and Salis Kanabaraf Maswatu helped with transcribing and translating many recordings. Terima kasi semuanya. 3 Alternative spellings found in official and unofficial sources include Tuberuasa, Tumberwasak, and Tubirwasak. 4 Although this map appears to be an unpublished draft, it is cited, for example, in Kamholz (2014). 5 Subgrouping in this paragraph follows Glottolog. 6 1,000 rupiah is also the smallest bill used in the area, and hence the smallest price one can pay for anything. Though coins of 500 rupiah exist, these are not used in Fakfak regency. 7 Example (36) does not contain a preposition, but all kinds of combinations of ‘preposition + short relational noun + demonstrative + long relational noun’ were elicited in [notes2022-12-16] and [e-dem-in]. 8 First person singular (la)i-uan is only used by conservative speakers, while the form muan is used by all speakers. Progressive speakers use (la)i-nan for the first person singular. 9 Topic persistence is the number of times the referent is mentioned in the 10 subsequent clauses. 0 means it is not mentioned again, 10 means it is mentioned in each subsequent clause. This measurement is typically used to indicate the topicality of a referent (Givón 1983:14 and Riesberg et al. 2022:12–13). 10 Referential distance is a count of how many clauses ago the referent was mentioned last, with no upper limit. A referential distance of 4 means the referent was mentioned four clauses ago for the last time. This measurement is also used to measure the topicality of a referent (Givón 1983:13 and Riesberg et al. 2022:12–13). 11 Geser-Gorom on Gorom Island has boit∼boti used in a construction meaning ‘not yet’ (Pattipeiluhu & Visser, ms.). 12 Compare also Gaby’s (2008:279) definition of deagentive, which is used for events caused by an inanimate force. 13 There are four other verbs with f-, one of them also with -k, that may be derived from other verbs. I do not see the semantic connection at all or not clearly enough, but report them here for reference: poti ‘go downhill’ – fpotik ‘sprout; bud’, resi ‘exceed’ – fresi ‘fight’, rekar ‘cry’ – freka ‘remember’, tara ‘cut’ – ftara ‘be angry’. 14 It is not clear which word class these belong to. For example, bisa, a loan from Indonesian, cannot be inflected but must occur together with another verb. Mau/mo ‘want’ is the only one of the words in the list that can be inflected, but it does not occur independently. The dependence of harus ‘must’, eran tai ‘cannot’ and pora ‘PROH’ is not yet well-investigated. --------------- ------------------------------------------------------------ --------------- ------------------------------------------------------------ Eline VISSER | Uruangnirin Grammar Sketch | JSEALS 19.1 (2026) 3 Copyright vested in the author; Creative Commons Attribution License