Difference between revisions of "Proto-Asuranesian language"

From CWS Planet
Jump to navigation Jump to search
Line 16: Line 16:
! Subfamily Clades!! Description!! Modern Descendants
! Subfamily Clades!! Description!! Modern Descendants
|-
|-
|[[Proto-Littoro-Marianic]] ||The ancestor of the largest branch of Asuranesian, ranging from south and west Nagu right across the Asura Ocean. Branched into [[Marianic]] and xxxx||[[Itongpe]], [[Hangaian]], [[Amakane]], [[Gaam]]
|[[Proto-Littoro-Marianic]] ||The largest branch of Asuranesian, ranging from south and west Nagu right across the Asura Ocean. Branched into [[Marianic]] and xxxx||[[Itongpe]], [[Hangaian]], [[Amakane]], [[Gaam]]
|-
|-
|[[Proto-Cananganamese]] ||The ancestor of a branch widely spoken in southern Nagu||[[Cananganamese]]
|[[Proto-Cananganamese]] ||A branch widely spoken in southern Nagu with the oldest known written records of Asuranesian languages||[[Cananganamese]]
|-
|-
|[[Proto-Camic]] ||The ancestor of a small but internally diverse branch spoken in the highlands of southwestern Nagu||xxxx
|[[Proto-Camic]] ||A small but internally diverse branch spoken in the highlands of southwestern Nagu||xxxx
|-
|[[Proto-Yurija-Yakuna]] ||A small continuum of related language varieties spoken in southeastern Nagu||[[Yurija]], [[Yakuna]]
|}
|}


Line 115: Line 117:


==== Fricatives ====
==== Fricatives ====
PAs had a fairly rich fricative system, consisting of labial *f, coronal *s *z *sʷ *zʷ, velar *x *xʷ, and glottal *h. The exact qualities of the labialised alveolars *sʷ *zʷ are contested (see previous section) but in modern interpretations are often held to be dental [θ ð] or [θʷ ðʷ], or lateral [ɬ ɮ] or [ɬʷ ɮʷ]. It is also possible that the fricatives had alternately voiced allophones in clusters due to such reflections in daughter languages, but evidence from other daughter languages suggests otherwise, particularly Proto-[Chinese-Limburgish] where fricatives universally shifted to /h/ in clusters (possibly originally /h/ from voiceless fricatives and /ɦ/ from voiced ones), creating aspirated stops and voiceless sonorants such as PAs *sʔluʔá-n “fire” > P-[Chinese-Limburgish] *l̥uan (thwaà /θwa:˦˨˧/ in [Limburgish]). In addition, Early-PAs had a voiced velar fricative *ɣ that developed into *j or *x in Late-PAs. It has been suggested however that the Early-PAs rhotics *r *rʷ (see section on approximants) may have actually been voiced velar fricatives [ɣ ɣʷ], matching the voiceless velar set *x *xʷ, that then later merged into either a single rhotic /r/ or fricative /ɣ/. Evidence for this relies primarily on the behaviour of *r in PAs as only fricatives were permitted to occur before stops and nasals in onset clusters yet *r appears to do so in a number of roots, as well as occurring in some prefixes without an epenthetic vowel. Phonological evidence in descendant language reflexes is limited however as *r is near universally reflected as a rhotic, and where it occurs initially, it is either deleted (as in XX), undergoes metathesis (as in PLM), or leaves a predictable cheshirisation effect to *r in other positions (as in P-Unnamed). This has led to the suggestion that a true rhotic may have been present in PAs as well and that *r merged with it, as well as that Early-PAs *ɣ may have actually been an approximant /ɰ/ or palatal fricative of various quality: /ç/, /ʝ/, /ʃ/, /ʒ/, /ɕ/ or /ʑ/. However, a distinction between /ɣ/ and /ɰ/ is typologically unusual and unstable and so often taken as evidence against this theory, although conversely such a system would be expected to collapse into a more typical one which can be observed in the changes /ɣ/ > /r/ and /ɰ/ > /j/, providing evidence in favour of the theory. The palatal reconstruction is based on the same evidence but favours a typologically more usual and stable system where a palatal fricative was present instead, although the proposed quality varies by author from pure palatal /ç/ to palato-alveolar /ʃ/ or alveolo-palatal /ɕ/. Whether *ɣ was voiced in this theory ranges by author as well, with some favouring the interpretation that it was originally voiced and devoiced to *x around voiceless consonants, whereas others suggest it was originally voiceless and voiced intervocalically and around voiced consonants to *j, although this theory seems less likely on the basis that word-initial *ɣ is always reflected as *j unless preceding a voiceless consonant, as well as the fact that no other fricatives show a development of intervocalic voicing. Interestingly however is that P-[Chinese-Limburgish] reflects voiceless sonorants and aspirated stops, suggesting a shift of *r > *h~ɦ in these clusters, in accordance with the widespread shift of PAs *x > *h and all fricatives being debuccalised when preceding consonants in onset clusters. Additionally, some instances of PAs *r are reflected as *ɦ in the same debuccalisation pattern as *x > *h, suggesting that PAs may have originally made a distinction between two rhotics that other branches merged. On this basis, some authors have proposed a divide in the Asuranesian family between the so-called “R1” and “R2” languages where R1 = 1 rhotic reflected, and R2 = 2 rhotics reflected. Others have suggested that P-Unnamed and other similar languages reflecting R2 patterns may have split from the earlier stage of PAs and that the family should be redrawn as Asuro-Unnamed, but there is no clear evidence and the widespread reflection of *r as a rhotic may have simply been an early areal shift or indeed a reflection of an original rhotic pronunciation as the current reconstruction suggests.
PAs had a fairly rich fricative system, consisting of labial *f, coronal *s *z *sʷ *zʷ, velar *x *xʷ, and glottal *h. The exact qualities of the labialised alveolars *sʷ *zʷ are contested (see previous section) but in modern interpretations are often held to be dental [θ ð] or [θʷ ðʷ], or lateral [ɬ ɮ] or [ɬʷ ɮʷ]. It is also possible that the fricatives had alternately voiced allophones in clusters due to such reflections in daughter languages, but evidence from other daughter languages suggests otherwise, particularly Proto-[Chinese-Limburgish] where fricatives universally shifted to /h/ in clusters (possibly originally /h/ from voiceless fricatives and /ɦ/ from voiced ones), creating aspirated stops and voiceless sonorants such as PAs *''sʔluʔá-n'' “fire” > P-[Chinese-Limburgish] *''l̥uan'' (''thwaà'' /θwa:˦˨˧/ in [Limburgish]). In addition, Early-PAs had a voiced velar fricative *ɣ that developed into *j or *x in Late-PAs. It has been suggested however that the Early-PAs rhotics *r *rʷ (see section on approximants) may have actually been voiced velar fricatives [ɣ ɣʷ], matching the voiceless velar set *x *xʷ, that then later merged into either a single rhotic /r/ or fricative /ɣ/. Evidence for this relies primarily on the behaviour of *r in PAs as only fricatives were permitted to occur before stops and nasals in onset clusters yet *r appears to do so in a number of roots, as well as occurring in some prefixes without an epenthetic vowel. Phonological evidence in descendant language reflexes is limited however as *r is near universally reflected as a rhotic, and where it occurs initially, it is either deleted (as in XX), undergoes metathesis (as in PLM), or leaves a predictable cheshirisation effect to *r in other positions (as in P-Unnamed). This has led to the suggestion that a true rhotic may have been present in PAs as well and that *r merged with it, as well as that Early-PAs *ɣ may have actually been an approximant /ɰ/ or palatal fricative of various quality: /ç/, /ʝ/, /ʃ/, /ʒ/, /ɕ/ or /ʑ/. However, a distinction between /ɣ/ and /ɰ/ is typologically unusual and unstable and so often taken as evidence against this theory, although conversely such a system would be expected to collapse into a more typical one which can be observed in the changes /ɣ/ > /r/ and /ɰ/ > /j/, providing evidence in favour of the theory. The palatal reconstruction is based on the same evidence but favours a typologically more usual and stable system where a palatal fricative was present instead, although the proposed quality varies by author from pure palatal /ç/ to palato-alveolar /ʃ/ or alveolo-palatal /ɕ/. Whether *ɣ was voiced in this theory ranges by author as well, with some favouring the interpretation that it was originally voiced and devoiced to *x around voiceless consonants, whereas others suggest it was originally voiceless and voiced intervocalically and around voiced consonants to *j, although this theory seems less likely on the basis that word-initial *ɣ is always reflected as *j unless preceding a voiceless consonant, as well as the fact that no other fricatives show a development of intervocalic voicing. Interestingly however is that P-[Chinese-Limburgish] reflects voiceless sonorants and aspirated stops, suggesting a shift of *r > *h~ɦ in these clusters, in accordance with the widespread shift of PAs *x > *h and all fricatives being debuccalised when preceding consonants in onset clusters. Additionally, some instances of PAs *r are reflected as *ɦ in the same debuccalisation pattern as *x > *h, suggesting that PAs may have originally made a distinction between two rhotics that other branches merged. On this basis, some authors have proposed a divide in the Asuranesian family between the so-called “R1” and “R2” languages where R1 = 1 rhotic reflected, and R2 = 2 rhotics reflected. Others have suggested that P-Unnamed and other similar languages reflecting R2 patterns may have split from the earlier stage of PAs and that the family should be redrawn as Asuro-Unnamed, but there is no clear evidence and the widespread reflection of *r as a rhotic may have simply been an early areal shift or indeed a reflection of an original rhotic pronunciation as the current reconstruction suggests.


==== Sonorants ====
==== Sonorants ====
Sonorants in PAs consisted of the nasal and approximant consonants *m *n *ŋ *r *l *j *ʟ. The nasal consonant series was typical, with labial *m, alveolar *n, and velar *ŋ corresponding to the labial, alveolar and velar series of stops respectively. There was however no corresponding nasal for the labiovelar stop in the most common reconstruction, despite attempts by some linguists to reconstruct one. In addition, there was a rhotic *r, two laterals, alveolar *l and velar *ʟ, and a palatal approximant *j. The rhotic *r has been variously interpreted as a trill [r], approximant [ɹ], or fricative [ɣ] and it has also been suggested that there may have even been two rhotics: *r₁ /r/ and *r₂ /ɣ/, based on evidence from the syllable structure of PAs (see previous section). Additionally, some have proposed that Early-PAs *ɣ (from which Late-PAs *j arose) may have been a velar approximant [ɰ] rather than a fricative [ɣ], but there is no direct evidence beyond a crosslinguistic tendency for voiced non-sibilant fricatives to lower to more approximant-esque manners of articulation. Other suggestions include palatal fricatives of varying qualities (see previous section) so the topic remains contested.
Sonorants in PAs consisted of the nasal and approximant consonants *m *n *ŋ *r *l *j *ʟ. The nasal consonant series was typical, with labial *m, alveolar *n, and velar *ŋ corresponding to the labial, alveolar and velar series of stops respectively. There was however no corresponding nasal for the labiovelar stop in the most common reconstruction, despite attempts by some linguists to reconstruct one. In addition, there was a rhotic *r, two laterals, alveolar *l and velar *ʟ, and a palatal approximant *j. The rhotic *r has been variously interpreted as a trill [r], approximant [ɹ], or fricative [ɣ] and it has also been suggested that there may have even been two rhotics: *r₁ /r/ and *r₂ /ɣ/, based on evidence from the syllable structure of PAs (see previous section). Additionally, some have proposed that Early-PAs *ɣ (from which Late-PAs *j arose) may have been a velar approximant [ɰ] rather than a fricative [ɣ], but there is no direct evidence beyond a crosslinguistic tendency for voiced non-sibilant fricatives to lower to more approximant-esque manners of articulation. Other suggestions include palatal fricatives of varying qualities (see previous section) so the topic remains contested.


Sonorants in PAs could cluster quite heavily and it has been suggested that traditional complex clusters may have actually involved syllabic sonorant nuclei, as in the word for “knife” *sntro-n, traditionally reconstructed as /sntron/ but alternatively suggested to have been /sn̩trón/. Evidence for this comes primarily from the reflexes of clustered sonorants in many descendant languages as having epenthetic vowels inserted, a development commonly associated with syllabic consonants. For example, PLM reflects *sntro-n as *santroón with an epenthetic vowel *a inserted before the nasal *n. Likewise, PYY reflects *sənatréi and PCAM *šən’ts’an, both with an epenthetic vowel *ə having been inserted before the nasal. Additional evidence comes from the placement of the word pitch in different case inflections of the noun, with the pitch remaining on the first vowel despite an expectation that it would shift to the second: nominative *sntro-n and accusative *sntróga-n vs expected *sntrogá-n. In the old analysis, there was an assumption that case markers simply were not accented and that only the word stem could be with some exceptions, but more recent analyses have revealed a far more regular pattern of nominal pitch placement based on mora count, so these patterns are unusual. From this, it can be reconstructed that syllabic sonorants counted as morae and therefore contributed to pitch placement. The internally reconstructed older stage of PAs, Early-PAs appears to reflect historical schwa *ə where clusters now exist, suggesting this vowel was either deleted or absorbed by the sonorant - both of which are crosslinguistically very common changes that would result in the formation of syllabic sonorant nuclei. This pattern is only reflected in some daughter languages however. Notably, PCAM regularly reflects the accusative form of *sntro-n as if it were pitched on the second vowel: *sntrogá-n > *šən’ts’əgan. It is unclear if these developments are innovations by means of levelling of the original pitch placement through analogy, or if they reflect the original state of affairs and later shifts in pitch placement in languages such as PLM reflect an innovation in mora count by counting clustered sonorants as well. Likewise, conflicting evidence for the syllabic hypothesis comes from several “UNNAMED (formosan grouping)” languages that appear to reflect clusters, with nasal consonants dropping and resulting in nasalisation or prenasalisation on following vowels or consonants respectively. For example, UNNAMED reflects *sntro-n as -rtâ¹â² /ʈɑ̃:˥˧/ with nasalisation of the original PAs vowel in place of the epenthesis present in PLM and other descendant languages. Likewise, Proto-Unnamed reflects nasal clusters as geminate consonants: *nl *nʔ > *nn, and its later descendant Proto-[Chinese-Limburgish] displays further cluster reduction with the development of prenasalised stops and voiceless sonorants: *nt > *nd and *sn > *hn.
Sonorants in PAs could cluster quite heavily and it has been suggested that traditional complex clusters may have actually involved syllabic sonorant nuclei, as in the word for 'knife' *''sntro-n'', traditionally reconstructed as /sntron/ but alternatively suggested to have been /sn̩trón/. Evidence for this comes primarily from the reflexes of clustered sonorants in many descendant languages as having epenthetic vowels inserted, a development commonly associated with syllabic consonants. For example, PLM reflects *''sntro-n'' as *''santroón'' with an epenthetic vowel *a inserted before the nasal *n. Likewise, PYY reflects *''sənatréi'', and PCAm *''šən’ts’an'', both with an epenthetic vowel *ə having been inserted before the nasal. Additional evidence comes from the placement of the word pitch in different case inflections of the noun, with the pitch remaining on the first vowel despite an expectation that it would shift to the second: intransitive *''sntro-n'' and accusative *''sntróga-n'' vs expected *''sntrogá-n''. In the old analysis, there was an assumption that case markers simply were not accented and that only the word stem could be with some exceptions, but more recent analyses have revealed a far more regular pattern of nominal pitch placement based on mora count, so these patterns are unusual. From this, it can be reconstructed that syllabic sonorants counted as morae and therefore contributed to pitch placement. The internally reconstructed older stage of PAs, Early-PAs appears to reflect historical schwa *ə where clusters now exist, suggesting this vowel was either deleted or absorbed by the sonorant - both of which are crosslinguistically very common changes that would result in the formation of syllabic sonorant nuclei. This pattern is only reflected in some daughter languages however. Notably, PCAm regularly reflects the accusative form of *''sntro-n'' as if it were pitched on the second vowel: *''sntrogá-n'' > *''šən’ts’əgan''. It is unclear if these developments are innovations by means of levelling of the original pitch placement through analogy, or if they reflect the original state of affairs and later shifts in pitch placement in languages such as PLM reflect an innovation in mora count by counting clustered sonorants as well. Likewise, conflicting evidence for the syllabic hypothesis comes from several “UNNAMED (formosan grouping)” languages that appear to reflect clusters, with nasal consonants dropping and resulting in nasalisation or prenasalisation on following vowels or consonants respectively. For example, UNNAMED reflects *''sntro-n'' as ''-rtâ¹â²'' /ʈɑ̃:˥˧/ with nasalisation of the original PAs vowel in place of the epenthesis present in PLM and other descendant languages. Likewise, Proto-Unnamed reflects nasal clusters as geminate consonants: *nl *nʔ > *nn, and its later descendant Proto-[Chinese-Limburgish] displays further cluster reduction with the development of prenasalised stops and voiceless sonorants: *nt > *nd and *sn > *hn.


==== Reflexes ====
==== Reflexes ====
Line 128: Line 130:
* Widespread loss of the glottal stop *ʔ, resulting in the formation of long vowels and diphthongs.
* Widespread loss of the glottal stop *ʔ, resulting in the formation of long vowels and diphthongs.
* Proto-Littoro-Marianic vocalised various clustered consonants, such as *ʔ > *å, *r > *aa and *l > *o. Complex clusters were otherwise broken up by the addition of epenthetic vowels across all descendant languages.
* Proto-Littoro-Marianic vocalised various clustered consonants, such as *ʔ > *å, *r > *aa and *l > *o. Complex clusters were otherwise broken up by the addition of epenthetic vowels across all descendant languages.
* Clustered nasals display widespread reflexes with epenthetic vowels inserted, typically before the nasal but also after in some descendant languages. The inserted vowel was often *a (PLM, Cananganamese, PCAM and PYY) but some descendant languages display other reflexes, notably insertion of *u before *m in PLM and epenthesis of *i instead of *a in Proto-XX.
* Clustered nasals display widespread reflexes with epenthetic vowels inserted, typically before the nasal but also after in some descendant languages. The inserted vowel was often *a (PLM, Cananganamese, PCAm and PYY) but some descendant languages display other reflexes, notably insertion of *u before *m in PLM and epenthesis of *i instead of *a in Proto-XX.


[[Category:Languages]][[Category:Asuranesia]][[Category:Extinct languages]][[Category:Proto-languages]]
[[Category:Languages]][[Category:Asuranesia]][[Category:Extinct languages]][[Category:Proto-languages]]

Revision as of 22:16, 11 July 2020

Proto-Asuranesian (PAs) is the linguistic reconstruction of the hypothetical common ancestor of the Asuranesian languages, one of the world's major language families.

PAS is estimated to have been spoken as a single language from 3500 BCE to 2500 BCE during the Neolithic Age, though estimates vary. According to the prevailing hypothesis, the original homeland of the Proto-Asuranesians was in south-west Nagu around the great lakes of Cananganam and the xxxx Mountains. The linguistic reconstruction of PAs has also provided insight into the culture and religion of its speakers.

As Proto-Asuranesians became isolated from each other through migration, the Proto-Asuranesian language developed into regional dialects which then underwent various shifts in phonology and morphology, transforming these dialects into the daughter languages seen today. Today, these languages are spoken across Nagu and Asuranesia by xxxx million speakers.

PAS had a fairly elaborate system of morphology that included inflectional suffixes (analogous to English "life, lives, life's, lives'‍") as well as apophony, reduplication, and accent placement. PAs nominals and pronouns had a fairly simple system of declension, marking only case and number whereas verbs were far more complex, conjugating not only for person and number but also for voice and valency, the latter being marked by apophonic alternation for the three valency categories: intransitive, monotransitive and ditransitive. Adjectives were represented by verbs and were declined accordingly. PAs' phonology, particles and numerals are also well reconstructed.

An asterisk is used to mark reconstructed words, such as *mbuxʷ- 'canoe', *dnlákʷ- 'to swim' (Itongpe no̧n) and *sʷámenl 'I am big, I am harmful (first singular active intransitive imperfective)'. A dash indicates that the word is either a stem and requires an inflectional suffix, such as *mbuxʷ- 'canoe' or that it is an affix, such as *-(i)m '1st person singular verb suffix'.

Descendants

The following are listed by their theoretical glottochronological development:

Subfamily Clades Description Modern Descendants
Proto-Littoro-Marianic The largest branch of Asuranesian, ranging from south and west Nagu right across the Asura Ocean. Branched into Marianic and xxxx Itongpe, Hangaian, Amakane, Gaam
Proto-Cananganamese A branch widely spoken in southern Nagu with the oldest known written records of Asuranesian languages Cananganamese
Proto-Camic A small but internally diverse branch spoken in the highlands of southwestern Nagu xxxx
Proto-Yurija-Yakuna A small continuum of related language varieties spoken in southeastern Nagu Yurija, Yakuna

Phonology

The phonology of PAs has been reconstructed by linguists based on the similarities and differences among current and extinct Asuranesian languages. Because PAs wasn't written, linguists must rely on evidence of its earliest attested descendants, such as Old Cananganamese and xxxx.

Due to the massive number of descendants, which often vary hugely from one another, there are disagreements on how to properly reconstruct PAs' phonology. The most commonly accepted features are:

  • Nasals at 3 POA's
  • Voiceless-Voiced stop distinction at 3 POA's
  • Glottal stop
  • Labiovelars
  • Labialised fricatives
  • The 5 common vowels (/i e a o u/) + at least 2 others

PAs is traditionally reconstructed to have had the following phonemes. See the article on Asuranesian sound laws for a summary of how these phonemes are reflected in the various Asuranesian languages. Phonemes in brackets are attested in only some daughter branches.

Consonants

The consonant system of PAs was crosslinguistically largely quite typical, with a voicing contrast in the plosives, three nasals, and labiovelar obstruents *kʷ *xʷ. However, there were also some less common features, notably the labialised coronal fricatives *sʷ *zʷ, velar lateral *ʟ and the lack of a labiovelar approximant /w/.

Labial Alveolar Palatal Velar Glottal
plain labial plain labial
Nasals *m *n (*ŋʷ)
Stops voiceless *p *t *k *kʷ
voiced *b *d *g
Fricatives voiceless *f *s *sʷ *j *x *xʷ *h
voiced *z *zʷ (*r) (*rʷ)
Approximant *r
Laterals *l

The table gives the most common notation used in modern publications

Plosive Series

PAs had a simple stop system with a voicing contrast at three places of articulation: *p *b, *t *d, and *k *g. In addition, there was a labiovelar stop *kʷ and a glottal stop *ʔ, although unusually there appears to have been no voiced labiovelar stop *gʷ. Some authors have attempted to reconstruct a phonemic *gʷ but virtually no evidence can be found to support its existence. An allophone of *g [gʷ] may have occurred however around labiovelar *kʷ *xʷ based on evidence from daughter languages reflecting *b, although it is also possible that the labial reflection is merely due to cluster reduction and allophony played no part.

Labials

PAs had three bilabial consonants: *m *p *b, a labiodental fricative *f, and four labialised consonants *kʷ, *sʷ, *zʷ and *xʷ. The labials *m *p *b *f are well attested, as are the labiovelars *kʷ *xʷ, but the labialised sibilants *sʷ *zʷ are more controversial. The evidence for labialisation comes primarily from the labial reflexes of the consonants in Proto-Yurija-Yakuna (PYY): *f *v, the development of PAs clusters CʷC > CuC in Proto-Camic (PCM) through epenthesis of the vowel /u/, and rounding effects on vowels in [INSERT]. However, the exact quality of these consonants is contested, with reconstructions all favouring a fricative manner of articulation but varying from the traditional sibilant reconstructions to dental /θ ð/ and lateral /ɬ ɮ/. Both Proto-Littoro-Marianic (PLM) and Proto-Unnamed (P) reflect *sʷ *zʷ as laterals *ɬ *l̥, and *ɮ *l in PLM and P-Unnamed respectively. This provides supporting evidence as the lateral reflections suggest a lateral or dental interpretation as the lateral quality may have been inherited or have developed from the crosslinguistically attested phenomenon of dental fricatives developing into lateral fricatives. Cananganamese and PCM on the other hand both reflect plain sibilants *s *z which could have easily evolved from any of the proposed phoneme qualities and thus do not favour any one interpretation over another. A shift from lateral fricative to dental fricative is crosslinguistically attested, as is the reverse, but the dental interpretation is the most favoured as a shift of /ɬ/ > /f/ is far less likely than the well attested Th-Fronting /θ/ > /f/. A rounded fricative [ɬʷ] developing into [f] is possible but most evidence seems to point to *sʷ *zʷ having actually been dental /θ ð/, likely with allophonic rounding [θʷ ðʷ]. As this topic remains contested, the traditional symbols *sʷ *zʷ remain widespread in publications and are used here as well.

Fricatives

PAs had a fairly rich fricative system, consisting of labial *f, coronal *s *z *sʷ *zʷ, velar *x *xʷ, and glottal *h. The exact qualities of the labialised alveolars *sʷ *zʷ are contested (see previous section) but in modern interpretations are often held to be dental [θ ð] or [θʷ ðʷ], or lateral [ɬ ɮ] or [ɬʷ ɮʷ]. It is also possible that the fricatives had alternately voiced allophones in clusters due to such reflections in daughter languages, but evidence from other daughter languages suggests otherwise, particularly Proto-[Chinese-Limburgish] where fricatives universally shifted to /h/ in clusters (possibly originally /h/ from voiceless fricatives and /ɦ/ from voiced ones), creating aspirated stops and voiceless sonorants such as PAs *sʔluʔá-n “fire” > P-[Chinese-Limburgish] *l̥uan (thwaà /θwa:˦˨˧/ in [Limburgish]). In addition, Early-PAs had a voiced velar fricative *ɣ that developed into *j or *x in Late-PAs. It has been suggested however that the Early-PAs rhotics *r *rʷ (see section on approximants) may have actually been voiced velar fricatives [ɣ ɣʷ], matching the voiceless velar set *x *xʷ, that then later merged into either a single rhotic /r/ or fricative /ɣ/. Evidence for this relies primarily on the behaviour of *r in PAs as only fricatives were permitted to occur before stops and nasals in onset clusters yet *r appears to do so in a number of roots, as well as occurring in some prefixes without an epenthetic vowel. Phonological evidence in descendant language reflexes is limited however as *r is near universally reflected as a rhotic, and where it occurs initially, it is either deleted (as in XX), undergoes metathesis (as in PLM), or leaves a predictable cheshirisation effect to *r in other positions (as in P-Unnamed). This has led to the suggestion that a true rhotic may have been present in PAs as well and that *r merged with it, as well as that Early-PAs *ɣ may have actually been an approximant /ɰ/ or palatal fricative of various quality: /ç/, /ʝ/, /ʃ/, /ʒ/, /ɕ/ or /ʑ/. However, a distinction between /ɣ/ and /ɰ/ is typologically unusual and unstable and so often taken as evidence against this theory, although conversely such a system would be expected to collapse into a more typical one which can be observed in the changes /ɣ/ > /r/ and /ɰ/ > /j/, providing evidence in favour of the theory. The palatal reconstruction is based on the same evidence but favours a typologically more usual and stable system where a palatal fricative was present instead, although the proposed quality varies by author from pure palatal /ç/ to palato-alveolar /ʃ/ or alveolo-palatal /ɕ/. Whether *ɣ was voiced in this theory ranges by author as well, with some favouring the interpretation that it was originally voiced and devoiced to *x around voiceless consonants, whereas others suggest it was originally voiceless and voiced intervocalically and around voiced consonants to *j, although this theory seems less likely on the basis that word-initial *ɣ is always reflected as *j unless preceding a voiceless consonant, as well as the fact that no other fricatives show a development of intervocalic voicing. Interestingly however is that P-[Chinese-Limburgish] reflects voiceless sonorants and aspirated stops, suggesting a shift of *r > *h~ɦ in these clusters, in accordance with the widespread shift of PAs *x > *h and all fricatives being debuccalised when preceding consonants in onset clusters. Additionally, some instances of PAs *r are reflected as *ɦ in the same debuccalisation pattern as *x > *h, suggesting that PAs may have originally made a distinction between two rhotics that other branches merged. On this basis, some authors have proposed a divide in the Asuranesian family between the so-called “R1” and “R2” languages where R1 = 1 rhotic reflected, and R2 = 2 rhotics reflected. Others have suggested that P-Unnamed and other similar languages reflecting R2 patterns may have split from the earlier stage of PAs and that the family should be redrawn as Asuro-Unnamed, but there is no clear evidence and the widespread reflection of *r as a rhotic may have simply been an early areal shift or indeed a reflection of an original rhotic pronunciation as the current reconstruction suggests.

Sonorants

Sonorants in PAs consisted of the nasal and approximant consonants *m *n *ŋ *r *l *j *ʟ. The nasal consonant series was typical, with labial *m, alveolar *n, and velar *ŋ corresponding to the labial, alveolar and velar series of stops respectively. There was however no corresponding nasal for the labiovelar stop in the most common reconstruction, despite attempts by some linguists to reconstruct one. In addition, there was a rhotic *r, two laterals, alveolar *l and velar *ʟ, and a palatal approximant *j. The rhotic *r has been variously interpreted as a trill [r], approximant [ɹ], or fricative [ɣ] and it has also been suggested that there may have even been two rhotics: *r₁ /r/ and *r₂ /ɣ/, based on evidence from the syllable structure of PAs (see previous section). Additionally, some have proposed that Early-PAs *ɣ (from which Late-PAs *j arose) may have been a velar approximant [ɰ] rather than a fricative [ɣ], but there is no direct evidence beyond a crosslinguistic tendency for voiced non-sibilant fricatives to lower to more approximant-esque manners of articulation. Other suggestions include palatal fricatives of varying qualities (see previous section) so the topic remains contested.

Sonorants in PAs could cluster quite heavily and it has been suggested that traditional complex clusters may have actually involved syllabic sonorant nuclei, as in the word for 'knife' *sntro-n, traditionally reconstructed as /sntron/ but alternatively suggested to have been /sn̩trón/. Evidence for this comes primarily from the reflexes of clustered sonorants in many descendant languages as having epenthetic vowels inserted, a development commonly associated with syllabic consonants. For example, PLM reflects *sntro-n as *santroón with an epenthetic vowel *a inserted before the nasal *n. Likewise, PYY reflects *sənatréi, and PCAm *šən’ts’an, both with an epenthetic vowel *ə having been inserted before the nasal. Additional evidence comes from the placement of the word pitch in different case inflections of the noun, with the pitch remaining on the first vowel despite an expectation that it would shift to the second: intransitive *sntro-n and accusative *sntróga-n vs expected *sntrogá-n. In the old analysis, there was an assumption that case markers simply were not accented and that only the word stem could be with some exceptions, but more recent analyses have revealed a far more regular pattern of nominal pitch placement based on mora count, so these patterns are unusual. From this, it can be reconstructed that syllabic sonorants counted as morae and therefore contributed to pitch placement. The internally reconstructed older stage of PAs, Early-PAs appears to reflect historical schwa *ə where clusters now exist, suggesting this vowel was either deleted or absorbed by the sonorant - both of which are crosslinguistically very common changes that would result in the formation of syllabic sonorant nuclei. This pattern is only reflected in some daughter languages however. Notably, PCAm regularly reflects the accusative form of *sntro-n as if it were pitched on the second vowel: *sntrogá-n > *šən’ts’əgan. It is unclear if these developments are innovations by means of levelling of the original pitch placement through analogy, or if they reflect the original state of affairs and later shifts in pitch placement in languages such as PLM reflect an innovation in mora count by counting clustered sonorants as well. Likewise, conflicting evidence for the syllabic hypothesis comes from several “UNNAMED (formosan grouping)” languages that appear to reflect clusters, with nasal consonants dropping and resulting in nasalisation or prenasalisation on following vowels or consonants respectively. For example, UNNAMED reflects *sntro-n as -rtâ¹â² /ʈɑ̃:˥˧/ with nasalisation of the original PAs vowel in place of the epenthesis present in PLM and other descendant languages. Likewise, Proto-Unnamed reflects nasal clusters as geminate consonants: *nl *nʔ > *nn, and its later descendant Proto-[Chinese-Limburgish] displays further cluster reduction with the development of prenasalised stops and voiceless sonorants: *nt > *nd and *sn > *hn.

Reflexes

Some of the changes undergone by the PAs consonants in daughter languages are the following:

  • Proto-Littoro-Marianic, Proto-Yurija-Yakuna, Cananganamese and Proto-Unnamed lost the labiovelars *kʷ *xʷ, either though delabialisation *kʷ *xʷ > *k *x and k x (P-Unnamed, PYY and Cananganamese), or a shift to a labial place of articulation *kʷ *xʷ > *p *f (PLM and PYY).
  • Proto-Littoro-Marianic and Proto-Unnamed reflect lateral reflexes of the labialised fricatives *sʷ *zʷ: *ɬ *ɮ and *l̥ *l respectively, whilst Proto-Camic and Cananganamese reflect plain sibilants *s *z and s z respectively.
  • Widespread loss of the glottal stop *ʔ, resulting in the formation of long vowels and diphthongs.
  • Proto-Littoro-Marianic vocalised various clustered consonants, such as *ʔ > *å, *r > *aa and *l > *o. Complex clusters were otherwise broken up by the addition of epenthetic vowels across all descendant languages.
  • Clustered nasals display widespread reflexes with epenthetic vowels inserted, typically before the nasal but also after in some descendant languages. The inserted vowel was often *a (PLM, Cananganamese, PCAm and PYY) but some descendant languages display other reflexes, notably insertion of *u before *m in PLM and epenthesis of *i instead of *a in Proto-XX.