R. Blust A note on hypercorrection in Mongondow

In: Bijdragen tot de Taal-, Land- en Volkenkunde 139 (1983), no: 4, Leiden, 459-464

This PDF-file was downloaded from http://www.kitlv-journals.nl

Downloaded from Brill.com09/30/2021 12:37:26AM via free access KORTE MEDEDELINGEN

ROBERT BLUST UNIVERSITY OF LEIDEN

A NOTE ON HYPERCORRECTION IN MONGONDOW The phenomenon of hypercorrection is a form of analogy well-known to linguists. Like all analogies it is expressible as a four-part proportion of which three parts are known and the fourth is inferred, e.g. black: white night: —— What distinguishes hypercorrection from other forms of analogy is the necessary precondition that variants exist within a language of which one is widely regarded by speakers as correct and the other(s) as corrupt. To cite just two examples 'from my own experience, Standard Brazilian Portuguese [f] corresponds to [h] in the colloquial speech of many dialect areas, including Pernambuco in the northeastern part of the country. Thus Standard [ratu] (written rato) is colloquial Per- nambucano [hatu] "rat". When a television newscaster in Recife ([hesifi]) once described the Hong Kong flu as gripe Rong Kong he was paying misdirected, but unmistakeable lip service to [f] as a marker of the Standard ("correct") language in contradistinction to [h] as a marker of the local ("corrupt") dialect. Similarly, the phoneme /f/ in Malay is confined entirely to loanwords (mostly from Arabic), and is replaced by /p/ in ordinary speech: e.g. fikir (educatéd speech): pikir (colloquial) "to think". Given this situation we can understand why a Malay friend in Sarawak some years ago attributed the deteriorating morality of his town to the brazenness of the Chinese "frostitutes". Both the Brazilian newscaster and the pious Moslem were hypercorrecting in conformity with the following four-part schema: Model "corrupt" "correct" BP hatu ratu Malay pikir fikir Generalization BP Hong Kong Rong Kong Malay prostitute frostitute Although it often operates on loanwords, hypercorrection can also alter native language material. This note is concërned with a type of hypercorrection affecting native words in Bolaang Mongondow (BM), an Austronesian language of northeastern , which together with its closest relatives in probably subgroups with certain languages of the Philippines. The data are drawn from Dunnebier (1951).

Downloaded from Brill.com09/30/2021 12:37:26AM via free access 460 Korte Mededelingen

In a small number of words with a known etymology BM exhibits intervocalic /r/ as the reflex of what until now has been reconstructed as zero: A 1) *Seyup (> iup) irup "to blow" 2) *pia piacv>pira "good" 3) *guyud (>giud) girudooirud "drag, pull" That this "intrusive" /r/ is not a product of regular phonological change is clear from other etymologies that contain a similar vowel sequence: 4) *diaq dia? "no, not" 5) *duyurj diurj "dugong" 6) *Riaw giow "clamor, outcry" 7) *iap ïap "count" 8) *liuS liu "turn or circle round" 9) * si-ia sia "he/she/it" 10) *tiaN sian "abdomen" 11) *ciap siap "peep (of chicks)" 12) *siar siar "send out beams of light" 13) *siaw siow "nine" A few remarks on the validity of etymologies 1) — 3) are perhaps in order. Cognates of irup and pira are found both in closely related languages ( hiipo, Kaidipang uyupo "to blow"; Kaidipang pia "good") and in more distant relatives (Cebuano Bisayan huyüp "blow air", Malay t-iup "blowing, puffing", Maranao pia "good, well, plentiful", Buli f ia "good"). Prototypes of both cognate sets have been suggested in previous publications: *iup (Dempwolff 1934-38), *S5eyup (Dyen 1965) "to blow", *pia (Blust 1970) "good". Known cognates of girudcoirud are not widely distributed, but Zorc (ms.) reconstructs "Proto-Philippines" *guyud "drag, pull". The contraction of original -uy- to -i- in BM is independently attested in *duyurj > diurj "dugong", and must have occurred prior to r-insertion in girud, since the change *u > i is otherwise unreported. Although examples of the "intrusive" /r/ are not numerous, then, they are significant since a) the etymologies in which they occur are sound and b) the appearance of a consonant for expected zero poses a serious problem to reconstruction. What could be the source of this /t/l Consonants have been added in the history of several (e.g. Kelabit -h, Palauan rj-, -rj), but these generally are statable as regular phonological developments. As we have seen al- ready, the. "intrusive" /r/ in BM is not. A second possibility is that the correspondence in question reflects a previously overlooked Proto-Austronesian phoneme that has been retained as /r/ in BM but which was lost in all other known witnesses. However, no independent support is known for such a hypothesis, and until independent evidence is found for it this solution must be re- garded as unmotivated.1

Downloaded from Brill.com09/30/2021 12:37:26AM via free access Korte Mededelingen 461

There are good reasons to believe that epenthetic /r/ in BM is in f act a historical by-product of hypercorrection. Many forms in Dunne- bier that contain initial or intervocalic /r/ are cross-referenced to variants with/y/(written/): B 1) daragex>dayag "yellow" 2) korotcvskoyot "very much/many" 3) urotcvmyot "tremble" 4) puraoopuya "red" 5) rabutcsiyabut "pull out" 6) rakupooyakup "scoop up with hands together" 7) raru?c>3yayu? "far"

Where intervocalic /r/ follows /i/ and precedes an unlike vowel the variation is between /r/ and phonemic zero: 8) irapcv>iap2 "blink" 9) rjirurjoorjiurj "nose" 10) piraoopia "good" In two known cases similar variation is found in the environment a _ i: 11) parirjc\3pairj "return, go back" 12) raritcN^yayitooyait "spread out" Charles (1974:466) notes that "Proto-Philippines" *-d- and *-j- regularly became BM -y- and that in some cases this change was generalized to initial position by analogical levelling through prefixed forms in which the stem-initial consonant became intervocalic in the word. Although Charles does not mention the free variation of /r/ and /y/ in some BM words, Noorduyn (1982:259) does: "In several cases of the correspondence of Gorontalo / : BM r there is a BM doublet with y, which may point to a development in BM of r to y." Attention to the available etymologies supports the claim that Proto-Austronesian *d/D, *z/Z, *j and *r (= Charles' Proto-Philippines *d and *j) did not simply become BM -y-, but rather became -r-, -r-oo-y-, -y- or, rarely, zero:3 -r- C 1) *bujirj "charcoal" burirj "dark brown" 2) *zarirjaw dorirjo "Calamus" 3) *kurapu kurapu "marine fish sp." 4) *kururj kururj "chicken coop" 5) *kerud korut "scratch, scrape" -r-oo-y- 6) *bujaq bura?c\Dbuya? "fóarh" 7) *buDak burakcsDbuyak "blossom" 8) *buruk burukcNsbuyuk "rotten" 9) *DaDuSa- dorowaoodoyowa "two" 10) *guDarj gurarj "elders, parents"cvo guyarj "old, older" 11) *puDa puracsspuya "red"

Downloaded from Brill.com09/30/2021 12:37:26AM via free access 462 Korte Mededelingen

-y- 12) *qazay ayoi "chin" 13) *Sajek ayok"kiss" 14) *qiDek iok "armpit" 15) *kuDen kuyon "cooking pot" 16) *quZan uyan "rain" 17) *qapeju opoyu "gall" 18) *pajey payoi "riceplant" 19) *qajerj p-ayorj "charcoal" 20) *tuDuq tuyu? "leak" 21) *seDaq toya? "fish" 22) *sejem toyom "ant" 23) *seDep toyop "to set, of the sun" -0- 24) *Suaji ai-ai "younger sibling" Proto-Austronesian (and historically secondary) *y > rooy, y or, rarely, zero, and so merged with the preceding consonant series in BM: -r-co-y- D 1) *Dayuq raru?cs3yayu? "far" 2) *Seyaq ora? cvsoya? "shame" •y- 3) *kaSiw (> kayu) kayu "wood, tree" 4) *layuG layug "to fly" 5) *sayad tayad "to slice" -0- 6) *layaR leag "sail" Hunggu Tajudin Usup, a native speaker of Kaidipang, informs me that y-variants in BM are now associated with the grandparental gen- eration, r-variants being favored by younger speakers. However, this almost certainly is a relatively recent development from an earlier situation in which the y-variants.were innovative, since y is less likely than r to dèvelop directly from a voiced obstruent. Noorduyn's view that the BM rcoy variation points to a development *r > y thus appears to be essentially correct. Moreover, it is clear that the change was not abrupt, but produced variants one or the other of which has been lost in some items, but both of which have been retained in others.4 For some generations the knowledge that r was the older and more "correct" variant must have been preserved, thereby setting the stage for hypercorrection to occur — as it evidently has — both with /y/ and with the similar non-phonemic semivowel between •/[/ and a following unlike vowel (predicted, but non-attested variants are given in parentheses):5 Model "corrupt" "correct" proto-form English buya? bura? *bujaq foam g-iurj rj-irurj *ijurj nose

Downloaded from Brill.com09/30/2021 12:37:26AM via free access Korte Mededelingen 463

Generalization yayu? raru? *Dayuq far oya? ora? *Seyaq shame (iup) irup *Seyup to blow pia pira *pia good (giud) girud *guyud drag, pull How are we to know that these developments in BM are not simply due to the sporadic change of [y] to [r] ? First, for obvious method- ological reasons sporadic change is to be admitted only as a last resort, and then preferably in connection with a known mechanism of secon- dary change (e.g. assimilation, dissimilation, analogy). Second, [r]cv> [y] variation in BM provides a clear model for the sporadic insertion of /r/. Finally, although the sample is small, [y] (= phonemic /y/ and zero) appears to change to -r- in only a minority of cases (5 of 19; i.e. Al-3, Dl-2), whereas *r changes to -y- in a considerable majority (19 of 24; i.e. C6-24). These facts thus suggest that *r > -y- is a product of primary sound change, but that *y > -r- is an epiphe- nomenon. What is intriguing about hypercorrect /r/ in BM is the evidence it provides that consonant epen thesis — which is attested as a product of regular phonological change in several Austronesian languages — can also result from mechanisms of secondary change. Without such an understanding our attempt to account for the phonological correspond- ences that relate BM to its sister languages and to their common ancestor might well go astray.

ACKNOWLEDGEMENT I wish to thank J. Noorduyn and H. Steinhauer for critical comments which led to improvements in an earlier version of this note.

NOTES 1 Data on BM's closest relative, Ponosakan, in Niemann (1869) sometimes show intrusive -h- corresponding to BM intrusive -r-: *pia > ma-piha "good", *guyud (> giud) > mo-gihur "pull, drag" (but *Siup > mog-iup "blow on"). Noorduyn (personal communication) points out that Bentenang ma-pigo "good" may show a similar epenthetic fricative (probably [x]). Given their probable subgrouping relationship these three languages must be treated as a single witness. 2 Dunnebier writes a semivowel in this word (iyap), but omits it in e.g. iap "to count". Since it is phonetically impossible to pronounce the continuous vowel sequence i plus unlike vowel without an intervening high front semi- vowel the implied phonemic contrast between -ia- and -iya- appears un- justified. Indeed, Dunnebier himself cross-references such orthographic variants as liuc^liyu "space under the house" and liomojliyom "to swallow". For the sake of consistency the orthographic semivowels in such environ- ments are omitted (as in C14 iok). 3 Reconstructions Cl, Cl and C22 are from Dempwolff (1924-25), and Cll, C14, C21 and D4 from Zorc (ms.). Dl follows from Isneg adayyó, Tagalog ma-ldyo?, Western Bukidnon Manobo me-ziu? "far", together with the BM

Downloaded from Brill.com09/30/2021 12:37:26AM via free access 464 Korte Mededelingen

form. All other reconstructions are from Dempwolff (1934-38), or from previous publications of my own. Dempwolff (1934-38) posited *zaRirjaw (C2) and *buRuk (C8), but BM reflects *r instead (also cf. Zorc *buruk "no good, rotten"). The change *-d > t in C5 is unexplained. 4 It is possible — even likely — that Dunnebier has not recorded all actual variants. A statement such as "one or the other of which has been lost in some items, but both of which have been retained in others" must therefore be understood solely in relationship to the available corpus of published material. 5 In at least one other case /r/ appears to reflect a semivowel which developed after the loss of intervocalic *h: *kahit (Dyen 1953:12) > kait > kayit > karit "hook, to hook" (cp. Gorontalo kayito "to hook"). The divergent treatment of *Seyup (with contraction of -ey-) and *Seyaq (without contrac- tion) is unexplained.

REFERENCES Blust, Robert A. 1970 'Proto-Austronesian addenda', Oceanic Linguistics 9:104-62. Charles, Mathew 1974 'Problems in the reconstruction of Protophilippine phonology and the subgrouping of the ', Papers of the First Inter- national Conference on Comparative Austronesian linguistics, Oceanic Linguistics 13:457-509. Dempwolff, Otto 1924-25 'Die L-, R- und D-Laute in austronesischen Sprachen', Zeitschrift für Eingeborenen-Sprachen 15:19-50, 116-38, 223-38, 273-319. 1934-38 Vergleichende Lautlehre des austronesischen Wortschatzes, Zeitschrift für Eingeborenen-Sprachen, 1. Induktiver Aufbau einer indonèsischen Ursprache, Supplement 15 (1934); 2. Deduktive Anwendung des Ur- indonesischen auf austronesische Einzelsprachen, Supplement 17 (1937); 3. Austronesisches Wörterverzeichnis, Supplement 19 (1938). Berlin: Reimer. Dunnebier, W. 1951 Bolaang Mongondowsch-Nederlandsch woordenboek, The Hague: Nijhoff. Dyen, Isidore 1953 The Proto-Malayo-Polynesian laryngeals, Baltimore: Linguistic Society of America. 1965 'Formosan evidence for some new Proto-Austronesian phonemes', Lingua 14:285-305. Niemann, G. K. 1869 'Mededeelingen omtrent de alfoersche taal van Noord-Oost Celebes', BKI 4:205-51, 399-445, 5:69-119. Noorduyn, J. 1982 'Sound changes in the Gorontalo language', Papers from the Third International Conference on Austronesian linguistics, vol. 2: Tracking the travellers, Pacific Linguistics C75:241-61. Zorc, R. David Ms. Proto-Philippine finder list, Cornell University.

Downloaded from Brill.com09/30/2021 12:37:26AM via free access