This articleshould specify the language of its non-English content using{{lang}} or{{langx}},{{transliteration}} for transliterated languages, and{{IPA}} for phonetic transcriptions, with an appropriateISO 639 code. Wikipedia'smultilingual support templates may also be used.See why.(October 2021)

Proto-Austronesian (commonly abbreviated asPAN orPAn) is aproto-language. It is thereconstructed ancestor of theAustronesian languages, one of the world's majorlanguage families. Proto-Austronesian is assumed to have begun to diversifyc. 4000 BCE – c. 3500 BCE inTaiwan.^[1]

Lower-level reconstructions have also been made, and includeProto-Malayo-Polynesian,Proto-Oceanic, andProto-Polynesian. Recently, linguists such asMalcolm Ross andAndrew Pawley have built large lexicons for Proto-Oceanic and Proto-Polynesian.

Proto-Austronesian is reconstructed by constructing sets of correspondences among consonants in the various Austronesian languages, according to thecomparative method. Although in theory the result should be unambiguous, in practice given the large number of languages there are numerous disagreements, with various scholars differing significantly on the number and nature of the phonemes in Proto-Austronesian. In the past, some disagreements concerned whether certain correspondence sets were real or represent sporadic developments in particular languages. For the currently remaining disagreements, however, scholars generally accept the validity of the correspondence sets but disagree on the extent to which the distinctions in these sets can be projected back to proto-Austronesian or represent innovations in particular sets of daughter languages.

Below are Proto-Austronesianphonemes reconstructed byRobert Blust, a professor of linguistics at theUniversity of Hawaii at Manoa.^[2] A total of 25 Proto-Austronesian consonants, 4 vowels, and 4diphthongs were reconstructed. However, Blust acknowledges that some of the reconstructed consonants are still controversial and debated.

The symbols below are frequently used in reconstructed Proto-Austronesian words.

• *C:voiceless alveolar affricate
• *c:voiceless palatal affricate
• *q:uvular
• *z:voiced palatal affricate
• *D:voiced retroflex stop
• *j:palatalizedvoiced velar stop
• *S:voiceless alveolar fricative
• *N:palatalizedalveolar lateral
• *r:alveolar trill
• *R:uvular trill

*D only appears in final position, *z/*c/*ñ only in initial and medial position, while *j is restricted to medial and final position.

The Proto-Austronesianvowels are a, i, u, and ə.

Thediphthongs, which are diachronic sources of individual vowels, are:

• *-ay
• *-aw
• *-uy
• *-iw

In 2010, John Wolff published his Proto-Austronesian reconstruction inProto-Austronesian phonology with glossary.^[3] Wolff reconstructs a total of 19 consonants, 4 vowels (*i, *u, *a, *e, where *e =/ə/), 4 diphthongs (*ay, *aw, *iw, *uy), and syllabic stress.

The following table shows how Wolff's Proto-Austronesian phonemic system differs from Blust's system.

In 2025, Kye Shibata made several changes to reconstructed Proto-Austronesian phonology based on careful examination of Formosan reflexes, resulting in a contrast between dental and retroflex phonemes.^[5] In addition to that, when compared to Blust's reconstruction, Shibata's avoided phonemes that aren't supported by Formosan evidence, namely *c, *D, *r, and *ñ.

According to Malcolm Ross,^[6] the following aspects of Blust's system are uncontroversial: the labials (p b m w); the velars k ŋ; y; R; the vowels; and the above four diphthongs. There is some disagreement about the postvelars (q ʔ h) and the velars g j, and about whether there are any more diphthongs; however, in these respects, Ross and Blust are in agreement. The major disagreement concerns the system ofcoronal consonants. The following discussion is based on Ross (1992).^[6]

Otto Dempwolff's reconstruction of Proto-Malayo-Polynesian from the 1930s included:

• Dental t d n l
• Retroflex ṭ ḍ ḷ
• Palatal t' d' n'
• Palatal k' g'

Dyen (1963), including data from the Formosan languages, expanded Dempwolff's set of coronal consonants:

• t split into t and C
• n split into n and L/N
• d' split and renotated as z and Z
• t' split into s₁ and s₂
• ḷ ṭ ḍ n' k' g' h renotated as r T D ñ c j q

Tsuchida (1976),^[7] building on Dyen's system:

• Further split d into D₁ D₂ D₃ D₄. He also believed that Dyen's c (Dempwolff's k') could not be reconstructed for Proto-Austronesian (he also split Dyen's w into w W and q into q Q, which were not accepted by later scholars.)

Dahl reduced Tsuchida's consonants into:

• D₁ D₂ D₃ D₄ into d₃ d₂ d₁ d₃ (with the new d₃ reflecting the combination of the old D₁ and D₄) and combined Dyen's S X x into a single phoneme S. He did accept Dyen's c but did not accept his T D. (He also renotated a number of phonemes in ways that were not generally accepted by later scholars.)

Blust based his system on a combination of Dyen, Tsuchida and Dahl, and attempted to reduce the total number of phonemes. He accepted Dahl's reduction of Dyen's S X x into S but did not accept either Tsuchida's or Dahl's split of Dyen's d; in addition, he reduced Dyen's s₁ s₂ to a single phoneme s. While accepting Dyen's c, he was hesitant about T and D (more recently, Blust appears to have accepted D but rejected T, and also rejected Z).

Ross likewise attempted to reduce the number of phonemes, but in a different way:

• He accepts Dahl's d₁ d₂ d₃ and also Z (eventually rejected by Blust). He notes that the distinction between d₁ and d₂ d₃ is only reconstructable for the Formosan language groupsAmis, Proto-Puyuma and Proto-Paiwan, and only Proto-Paiwan has a three-way distinction among d₁ d₂ d₃; contrarily the distinction between Z and d₁ is reconstructable only for Proto-Rukai and Proto-Malayo-Polynesian, but not any of the previous three groups. However, he still believes (contra Blust) that the distinction among these phonemes is an inheritance from Proto-Austronesian rather than an innovation in the respective groups.
• He notes that d₁ occurs only morpheme-initially, while r occurs only morpheme-non-initially, and as a result combines the two.
• He does not accept the phonemes c z ñ in Proto-Austronesian, and asserts that none of them are "readily reconstructable" outside ofProto-Malayo-Polynesian. Furthermore, while he believes that ñ was a general innovation in Proto-Malayo-Polynesian, c and z "are reflected differently from PMP [Proto-Malayo-Polynesian] *s and *d only in a fairly limited area of western Indo-Malaysia and appear to be the results of local developments".
• He also reconstructs the coronals somewhat differently. He believes that C S l d₃ were all retroflex (respectively,/tʂ/;/ʂ/or/ʃ/;/ɭ/or/ɽ/;/ɖ/), and s and L (Blust's N) were dental /s/ and /l/, as opposed to Blust's reconstruction as dental and palatal, respectively. According to Ross, this is based on their outcomes in theFormosan languages andJavanese; although their outcomes as dental/palatal is geographically more distributed, it occurs only inMalayo-Polynesian, which represent a single clade with respect to the Formosan languages.

As Proto-Austronesian transitioned to Proto-Malayo-Polynesian,Proto-Oceanic, andProto-Polynesian, the phonemic inventories were continually reduced by merging formerly distinct sounds into one sound. Three mergers were observed in the Proto-Austronesian to Proto-Malayo-Polynesian transition, while nine were observed for the Proto-Oceanic to Proto-Polynesian transition. Thus, Proto-Austronesian has the most elaborate sound system, while Proto-Polynesian has the fewest phonemes. For instance, theHawaiian language is famous for having only eight consonants, whileMāori has only ten consonants. This is a sharp reduction from the 19–25 consonants of the Proto-Austronesian language that was originally spoken onTaiwan.

Blust also observed the following mergers and sound changes between Proto-Austronesian and Proto-Malayo-Polynesian.^[2]

However, according to Wolff (2010:241),^[3] Proto-Malayo-Polynesian's development from Proto-Austronesian only included the following three sound changes.

• PAn *ɬ > PMP *ñ, l, n
• PAn *s > PMP *h
• PAn *h > PMP *Ø

Proto-Oceanic merged even more phonemes. This is why modern-dayPolynesian languages have some of the most restricted consonant inventories in the world.^[2]

Unusual sound changes that occurred within the Austronesian language family include:^[2]

• Proto-Malayo-Polynesian *wor *b >Sundanese c- or -nc-
• Proto-Oceanic *wor *y > p inLevei Khehek
• Proto-Oceanic *ror *R > g​͡ʟ inHiw^[9]
• Proto-Polynesian *lor *r > ŋg (via *ɣ or *ʁ) inRennellese
• Proto-Polynesian *t > k in Hawaiian,Samoan, andOntong Java (after *k > ʔ)^[10]

Proto-Austronesian is a verb-initial language (includingVSO andVOS word orders), as mostFormosan languages, allPhilippine languages, someBornean languages, all Austronesian dialects ofMadagascar, and allPolynesian languages are verb-initial.^[2] However, most Austronesian (many of which areOceanic) languages ofIndonesia,New Guinea,New Caledonia,Vanuatu, theSolomon Islands, andMicronesia areSVO, or verb-medial, languages.SOV, or verb-final, word order is considered to be typologically unusual for Austronesian languages, and is only found in various Austronesian languages ofNew Guinea and to a more limited extent, theSolomon Islands. This is because SOV word order is very common in the non-AustronesianPapuan languages.

The Austronesian languages ofTaiwan,Borneo,Madagascar and thePhilippines are also well known for their unusualmorphosyntactic alignment, which is known as thesymmetrical voice (also known as the Austronesian alignment). This alignment was also present in the Proto-Austronesian language. Unlike Proto-Austronesian, however, Proto-Oceanic syntax does not make use of the focus morphology present in Austronesian-aligned languages such as thePhilippine languages. In thePolynesian languages, verbal morphology is relatively simple, while the main unit in a sentence is the phrase rather than the word.

Below is a table of John Wolff's Proto-Austronesian voice system from Blust (2009:433).^[2] Wolff's "four-voice" system was derived from evidence in various Formosan and Philippine languages.

However, Ross (2009)^[11] notes that what may be the most divergent languages,Tsou,Rukai, andPuyuma, are not addressed by this reconstruction, which therefore cannot claim to be the alignment system of the protolanguage of the entire family. He calls the unit to which this reconstruction appliesNuclear Austronesian.

The following table compares Proto-Austronesian and Proto-Malayo-Polynesian question words.

Currently, the most complete reconstruction of the Proto-Austronesiancase marker system is offered byMalcolm Ross.^[2] The reconstructed case markers are as follows:

Important Proto-Austronesian grammatical words include the ligature *na and locative *i.^[2]

Morphology andsyntax are often hard to separate in the Austronesian languages, particularly thePhilippine languages.^[2] This is because the morphology of the verbs often affects how the rest of the sentence would be constructed (i.e., syntax).

Below are some Proto-Austronesianaffixes (includingprefixes,infixes, andsuffixes) reconstructed byRobert Blust. For instance, *pa- was used for non-stative (i.e., dynamic) causatives, while *pa-ka was used for stative causatives (Blust 2009:282). Blust also noted a p/m pairing phenomenon in which many affixes have bothp- andm- forms. This system is especially elaborate in theThao language of Taiwan.^[2]

A verbal prefix *paRi- is also reconstructed (albeit at the lower level ofPEMPTooltip Proto-Eastern Malayo-Polynesian), for "reciprocal or collective action";^[15] it is particularly developed inOceanic languages.^[16]

CV (consonant + vowel)reduplication is very common among the Austronesian languages. In Proto-Austronesian, Ca-reduplicated (consonant + /a/) numbers were used to count humans, while the non-reduplicated sets were used to count non-human and inanimate objects. CV-reduplication was also used to nominalize verbs in Proto-Austronesian. In Ilocano, CV-reduplication is used to pluralize nouns.

Reduplication patterns include (Blust 2009):

• Full reduplication
• Full reduplication plus affixation
• Full reduplication minus the coda
• Full reduplication minus the last vowel
• Full reduplication with vocalic or consonantal change, or both
• Full reduplication with consecutive identical syllables
• Prefixal foot reduplication/leftward reduplication
• Suffixal foot reduplication/rightward reduplication
• CVC-reduplication
• CV-reduplication (marks durative aspect, collectivity, or intensity in Bunun; future in Tagalog)
• CV-reduplication plus affixation
• Ca-reduplication (used to derive human-counting numerals and deverbal instrumental nouns in Thao and Puyuma)
• Extensions of fixed segmentism
• Reduplicative infixes
• Suffixal syllable reduplication

Other less common patterns are (Blust 2009):

• Vacuous reduplication (occurs in Paamese)
• Full reduplication minus the initial (occurs in Anejom of southern Vanuatu)
• Full reduplication plus an initial glide (occurs in Kosraean)
• Partial reduplication minus initial glottal stop (occurs in Rennellese)
• True CV-reduplication (occurs in Pangasinan)
• Rightward trisyllabic reduplication (occurs in theManam language)
• Double reduplication (occurs in Woleaian)
• Triplication (only in theThao language)
• Serial reduplication (only in theThao language)

The Proto-Austronesian and Proto-Malayo-Polynesianpersonal pronouns below were reconstructed byRobert Blust.^[2]

In 2006,Malcolm Ross also proposed seven different pronominal categories for persons. The categories are listed below, with the Proto-Austronesian first person singular ("I") given as examples.^[17]

1. Neutral (e.g., PAN *i-aku)
2. Nominative 1 (e.g., PAN *aku)
3. Nominative 2 (e.g., PAN *=ku, *[S]aku)
4. Accusative (e.g., PAN *i-ak-ən)
5. Genitive 1 (e.g., PAN *=[a]ku)
6. Genitive 2 (e.g., PAN *(=)m-aku)
7. Genitive 3 (e.g., PAN *n-aku)

The following is from Ross' 2002 proposal of the Proto-Austronesian pronominal system, which contains five categories, including the free (i.e., independent or unattached), free polite, and three genitive categories.^[2]

Proto-Austronesian vocabulary relating to agriculture and other technological innovations include:^[2]

• *pajay:rice plant
• *beRas:husked rice
• *Semay:cooked rice
• *qayam: bird (means "domesticated animal" in PMP)
• *manuk: chicken (PMP *manu-manuk means "bird")
• *babuy: pig
• *qaNuaŋ:carabao
• *kuden: clay cooking pot
• *SadiRi: housepost
• *busuR: bow
• *panaq: flight of an arrow
• *bubu: fish trap
• *tulaNi: bamboonose flute

Proto-Malayo-Polynesian innovations include:

• *puqun: base of a tree; origin, cause
• *sumpit: blowpipe
• *haRezan: notched log ladder (used to enter pile dwellings)
• *taytay: bamboo suspension bridge (POc *tete "ladder, bridge")
• *kaka: elder same sex sibling
• *huaji: younger same sex sibling
• *ñaRa: brother of a woman
• *betaw: sister of a man

Proto-Malayo-Polynesian also has several words for house:

• *balay (house, building for public use)
• *Rumaq (house, family dwelling)
• *banua (land, village, house, country, sky, heaven) – hencevanua andwhenua (as intangata whenua)
• *lepaw (granary)
• *kamaliR (bachelors' clubhouse)

Below are colors in reconstructed Proto-Austronesian, Proto-Malayo-Polynesian, Proto-Oceanic, and Proto-Polynesian.^[2][20] The first three have been reconstructed byRobert Blust, while the Proto-Polynesian words given below were reconstructed byAndrew Pawley. Proto-Polynesian displays many innovations not found in the other proto-languages.

The Proto-Austronesians used two types of directions, which are the land-sea axis and the monsoon axis. The cardinal directions of north, south, east, and west developed among the Austronesian languages only after contact with the Europeans. For the land-sea axis, upstream/uphill and inland, as well as downstream/downhill and seaward, are synonym pairs. This has been proposed as evidence that Proto-Austronesians used to live on a mainland, since the sea would be visible from all angles on small islands.^[2]

• *daya: inland (also upstream/uphill)
• *lahud: seaward (also downstream/downhill)
• *SabaRat: west monsoon
• *timuR: east monsoon
• *qamiS: north wind

In Kavalan, Amis, and Tagalog, the reflexes of *timuR mean "south" or "south wind," while in the languages of the southern Philippines and Indonesia it means "east" or "east wind."

InIlocano,dáya andláud respectively mean "east" and "west," while inPuyuma,ɖaya andɭauɖ respectively mean "west" and "east."^[21] This is because the Ilocano homeland is the west coast of northern Luzon, while the Puyuma homeland is on the eastern coast of southern Taiwan. Among the Bontok, Kankanaey, and Ifugaw languages of northern Luzon, the reflexes of *daya mean "sky" because they already live in some of the highest elevations in the Philippines (Blust 2009:301).

Also, theMalay reflex of *lahud islaut, which means "sea", used as directionstimur laut (means "northeast",timur = "east") andbarat laut (means "northwest",barat = "west"). Meanwhile, *daya only performs inbarat daya, which means "southwest".

Below are reconstructed Proto-Austronesian, Proto-Malayo-Polynesian, Proto-Oceanic, and Proto-Polynesian numbers from the Austronesian Basic Vocabulary Database.^[22]

Note that *lima 'five', ultimately the root for 'hand', is not found for 'five' in some Formosan languages, such as Pazeh, Saisiat, Luilang, Favorlang and Taokas; numerals cognate with Proto-Malayo-Polynesian 6–10 are found in Amis, Basay, Bunun, Kanakanabu, Kavalan, Paiwan, Puyuma, Saaroa and Tsou. Pazeh, Favorlang, Saisiat and Taokas reflect *RaCep 'five'.

Laurent Sagart suggests that this was the PAn root, replaced by *lima in a lineage that lead to the remaining languages, rather than the reverse, because it seems to be retained in proto-Malayo-Polynesian in the forms 7, 8, 9, which appear to be disyllabic contractions of additive phrases attested from some of the western Formosan languages, especially Pazeh: Pazehxaseb-uza 'six' (literally 'five-one');xaseb-i-dusa 'seven' ('five-and-two'), with thebidu cognate with PMP *pitu;xaseb-a-turu 'eight' ('five-and-three'), with thebaturu cognate with PMP *walu;xaseb-i-supat 'nine' ('five-and-four'), with thesupa (< PAn *Sepat 'four') cognate with PMP *Siwa.^[23]

The Proto-Austronesian language had different sets of numerals for non-humans ("set A") and humans ("set B") (Blust 2009:279). Cardinal numerals for counting humans are derived from the non-human numerals through Ca-reduplication. This bipartite numeral system is found in Thao, Puyuma, Yami, Chamorro, and various other languages (however, Paiwan usesma- andmanə- to derive human numerals). In many Philippine languages such as Tagalog, the two numeral systems are merged (Blust 2009:280–281).

Proto-Austronesian also used *Sika- to deriveordinal numerals (Blust 2009:281).

Below are reconstructed Proto-Austronesian, Proto-Malayo-Polynesian, Proto-Oceanic, and Proto-Polynesian verbs from theAustronesian Basic Vocabulary Database.

The following are monosyllabic Proto-Austronesian roots reconstructed by John Wolff (Wolff 1999).^[24]

• Austronesian personal pronouns
• Austronesian alignment
• Fossilized affixes in Austronesian languages
• Proto-Malayo-Polynesian language
• Proto-Philippine language
• Proto-Oceanic language
• Proto-Polynesian language
• Proto-Austroasiatic language
• Proto-Hmong–Mien language
• Proto-Tibeto-Burman language

• Austronesian languages
• Proto-languages

• Articles with short description
• Short description matches Wikidata
• Pages with non-English text lacking appropriate markup and no ISO hint
• Pages with non-English text lacking appropriate markup from October 2021
• Pages with plain IPA
• Articles containing Māori-language text
• Articles containing Ilocano-language text
• Articles containing Puyuma-language text
• Articles containing Malay (macrolanguage)-language text
• Articles containing Paiwan-language text
• Articles containing Tagalog-language text