Notes on the position of Rejang

[PLEASE CITE BY URL AND DATE]

Permanent URL: https://ezlinguistics.blogspot.com/p/rejang.html
First posted: 31 May 2019
Last updated: 08 June 2019

Introduction

The position of the Rejang language (spoken on Sumatra in Bengkulu, and in some parts of neighboring South Sumatra) is still debated. In conservative classifications (e.g. Adelaar 2005), it is treated as an isolate branch of the Malayo-Polynesian languages, and I want to advocate here that this still is the safest solution in spite of several proposals to classify Rejang within mid-level subgroups of Malayo-Polynesian.

McGinn's Out-of-Borneo-hypothesis

The most challenging proposal was brought forward by Richard McGinn (2003, 2009), linking Rejang to the Land Dayak languages of Borneo, specifically to Bukar-Sadong. This proposal is accepted by the Glottolog, where Rejang is listed as deeply nested within Land Dayak in a "Sadong-Rejang" branch. The main piece of McGinn's evidence is what he considers parallel reflexes of PMP *a in closed ultimate syllables. In Rejang, PMP *-aC shifted to Proto Rejang *-əC, unless C was *k, *ŋ, *q. In the latter case, this shift was blocked and *a remained unchanged. For Bukar-Sadong, McGinn recorded a similar sound change: *-aC > -ʌC, which was also blocked by following *k, *ŋ, *q. Phonetically, these two changes are indeed very similar, especially with the identical blocking rule. However, phonetic shifts have little weight in subgrouping, unless they also have an effect on the phonological system. Since in Rejang we observe a shift of PMP *a to PR *ə, the real question is: what happened to PMP *a and *ə in PR and Bukar-Sadong? Were they affected by parallel shifts, mergers and splits? Only if the last question is answered positively, we actually have supporting evidence for McGinn's Out-of-Borneo-hypothesis. For this purpose, I will analyze the Rejang and Bukar-Sadong data by McGinn (2003 and 2005)

For Rejang, it is clear that *a and *ə in closed final syllables only merged if the penultimate vowel was *a. In this case, the reflex is PR *ə before non-velars, e.g. PMP *panas > PR *panəs, *taləs > PR *taləs, and *a before velars (i.e. *k, *ŋ, *q), e.g. PMP *daRaq > PR *dalaʔ , *tanəq > PR *tanaʔ.

If the penultimate vowel was *ə, *a and *ə were kept distinct: *a remains unchanged, regardless of the final consonant, e.g. PMP *takəbas > PR *təbas, *təbaŋ > PR *təbaŋ; *ə remains unchanged before non-velars, e.g. *pəRəs > PR *pərəs, but became *o before a velar, e.g. PMP *təktək > PR *tətok.

In the PMP vowel templates *–uCaC and *–iCaC, reflexes of ultmate *a follow the same pattern as with *–aCaC: *ə before non-velars, and *a before velars, e.g. PMP *quzan > PR *ujən, PMP *tulak > PR *tulak. In the PMP templates *–uCəC and *–iCəC, we observe a more complex shift which involves lowering of the penultimate vowel and subsequent total harmonization of *ə to the preceding vowel, resulting in the templates *–oCoC and *–eCeC. E.g. PMP *hikət > PR *eket, PMP *qutək > PR *otok. There are a few cases where PMP etyma with *–uCəC and *–iCəC did not undergo harmonization, but these can be easily explained as loans from Malay (e.g. PMP(?) *buləd > PR *bulət).

As we can see, the phonetic shift of PMP *-aC to Proto Rejang *-əC and the blocking rule before velars are actually part of a complex set of sound changes, which led to the partial conditioned merger of PMP *a and *ə in the final syllable, which only occurred if the penultimate vowel was *a.

In Bukar-Sadong, *a and *ə have not merged at all in closed final syllables, as far as we can deduct from the wordlist in McGinn (2003). *ə remains unchanged, while *a splits into ʌ before non-velars and a before velars. Thus the simililarity of the phonetic sound changes in Rejang and Bukar-Sadong which were brought forward by McGinn is not matched by any shared phonological innovation. We are left with a partially shared retention, viz. retention of the contrast between PMP *a and PMP *ə in the final syllable. Consequently, McGinn's main argument for the deeply-nested inclusion of Rejang in the Land Dayak subgroup cannot be upheld. Together with the extreme sparseness of the lexical evidence, I consider the Out-of-Borneo-hypothesis, as far as the connection to Land Dayak is concerned, as untenable.

Other proposals

In other proposals, Rejang is included in groupings that contain–among other languages–also the Malayic languages. Nothofer (1988) grouped Rejang together with Embaloh and "Iban-Malayan" (=Malayic in current terminology) in what he then called "Malayic". Blust (2010) places Rejang in his "Greater North Borneo" subgroup, together with Moken, Sundanese, the Chamic languages, the Malayic languages, and all non-Malayic languages of Borneo except for the Barito languages. The Greater North Borneo hypothesis is solely based on lexical innovations. Blust further speculates about a "Western Indonesian" subgrup that would include Greater North Borneo, the Barito languages and "all languages of western Indonesia outside Sulawesi".

In his tour de force classification of the "Western" Malayo-Polynesian languages, Smith (2017) accepts Blust's Greater North Borneo hypothesis (except for the inclusion of Moken), but splits Blust's "Western Indonesian" based on reflexes of PMP *j into 1. the "Sumatran" languages (=Batak–Barrier Inlands + Nasal), which reflect PMP *j > *g (with subsequent shift to g, x, h, k or zero in indivdual languages), and 2. his narrowed down version of "Western Indonesian" (including Greater North Borneo, thus Rejang), which according to him is characterized by the merger of PMP *j and *d.

As a "Celebicist", I agree with Smith that reflexes of PMP *j have a key role for high-order subgrouping. Taken by itself, this is definitely not a sufficient criterion, but appears to be a good necessary criterion which then needs to be corroborated by other pieces of evidence.

In the case of Rejang, Smith (2017:480) erroneously presents it as having merged *j and *d, and consequently lists it among the Western Indonesian languages. McGinn (2005:20–21) however has shown that PMP *j is consistently reflected by a velar stop in all dialects except for Rawas. In the latter, medial *j also has a velar reflex, but final *j is reflected by t (pusət < *pusəj, ulət < *qusəj, dalət < *laləj), which McGinn takes as evidence that Proto-Rejang has retained a distinct *j in final position. I rather suspect Malayic influence here, which is also betrayed by the vocalism of the first two items (cf. poso[g/k], olo[g/k], dalə[g/k] in the other dialects of Rejang). In any case, the velar reflexes in Rejang are not compatible with its inclusion wthin Western Indonesian and Greater North Borneo, unless we give up the notion that reflexes of *j have any value in the high-order subgrouping of Malayo-Polynesian.

Conclusion

I have shown here that both the Out-of-Borneo-hypothesis and the Greater North Borneo/Western Indonesian-connection cannot be upheld as solutions for the classification of Rejang. The main counter-argument for the Greater North Borneo-affiliation, viz. the velar reflex of PMP *j, may serve as a starting point for searching into a new direction. Based on this reflex, Rejang would actually better fit into Smith's "Sumatran" branch, together with Nasal and the Batak–Barrier Islands languages. This conclusion is, however, premature. As I have shown in an earlier paper (Zobel 2002), the Batak–Barrier Islands languages share important grammatical innovations with Lampung, Javanese, Malayic languages, Celebic languages, South Sulawesi languages and Chamorro, viz. preposed actor pronouns (e.g. Nias u-halö 'I take [it]') and affix combinations such as *maN- -i. None of these are found in Rejang, which is in fact characterized by the total lack of suffixes, even though it has retained prefixes and infixes which are in employed in a simple two-voice system. At least typologically, Rejang is indeed somewhat closer to the languages of Borneo.

To sum up: at the current state of research, I believe the best approach to the Rejang question is to treat it as an isolate branch within the Malayo-Polynesian languages.


Bibliography

Adelaar, Alexander (2005). "The Austronesian languages of Asia and Madagascar: a historical perspective". In Alexander Adelaar & Nikolaus P. Himmelmann (eds.) The Austronesian Languages of Asia and Madagascar , pp. 1–42. London: Routledge.

Blust, Robert (2010). "The Greater North Borneo Hypothesis". Oceanic Linguistics 49: 44–118.

McGinn, Richard (2003). "Raising of PMP *a in Bukar-Sadong Land Dayak and Rejang". In John Lynch (ed.) Issues in Austronesian historical phonology, pp. 37–64. Canberra: Pacific Linguistics.

McGinn, Richard (2005). "What the Rawas Dialect Reveals about the Linguistic History of Rejang". Oceanic Linguistics 44: 12–64.

McGinn, Richard (2009). "Out-of-Borneo subgrouping hypothesis for Rejang: re-weighing the evidence". In Alexander Adelaar and Andrew Pawley (eds.) Austronesian historical linguistics and culture history: a festschrift for Bob Blust, pp. 397-410. Canberra: Pacific Linguistics.

Nothofer, Bernd (1988). "A discussion of two Austronesian subgroups: Proto-Malay and Proto-Malayic". In Mohd. Thani Ahmad & Zaini Mohamed Zain (eds.) Rekonstruksi dan cabang-cabang Bahasa Melayu induk, pp. 34–58. Kuala Lumpur: Dewan Bahasa dan Pustaka.

Smith, Alexander D. (2017). "The Western Malayo-Polynesian Problem". Oceanic Linguistics 56: 435–490.

Zobel, Erik (2002). "The position of Chamorro and Palauan in the Austronesian family tree: evidence from verb morphosyntax". In Fay Wouk & Malcolm Ross (ed.) The history and typology of western Austronesian voice systems, pp. 405–434. Canberra: Pacific Linguistics.