(Note: this page uses several Unicode characters, but those that may not display properly are explained)

The language of the Kai (older Kei) Islands in eastern Indonesia was described by Fr. H. Geurtjens in two extensive works-- a grammar, and a dictionary (publ.1921)-- but the spelling system he devised is extremely confusing.  I seem to recall a statement, perhaps in the grammar, to the effect that he sought to “make the Kei words immediately pronounceable by a Dutch speaker”.

Consonants, for the most part, are clear.  He comments only “de v en w zijn dikwijls niet scherp uitgesproken en zweven tusschen dese twee in, en gewestelijk komt in vele woorden de een o andere meer uit.”  Both v and w reflect original *b; v is never used initially, and rarely between vowels; it occurs in variation with w in final position (i.e. in forms originally *CVbV, where the final V has been lost-- thus we find alternants like tāv ~tawoen (Ford. tavu), but also taw ‘stab’ ~tawak ‘insert a sharp object into’ < *taba( ).  One has the impression G. was trying to limit -w to its diphthongal value, but either he or the printers slipped up in many cases. In a few cases, we find -f for expected -v/w < *-b(V), probably simple oversight/mishearing on G’s part.  It is possible too, that in the many years he spent in the islands, notes from the early days were not amended as he developed a better sense of the language, and variant readings thus found their way into the finished work.

It is also worth mentioning, that the series in which these works appeared is notorious for sloppy printing and proof-reading.  As I have commented elsewhere, when the non-native reader can spot errors in the Dutch text, one wonders what is slipping by in the native-language material.

Here is his list of the vowel sounds of Kai (pp. 5-6 in the dictionary) with their Dutch (or other) equivalents:

          a = a, als in pad

          ā (a-macron) = aa als in haar

          ă (a-breve) = a, als in ‘t Eng. black (gebroken a) (Comment: presumably [æ], but the sound is rare, and occurs only for /a/ in some obvious Malay loans)

          e = e als in bed

          ē (e-macron) = e als in heet, bede

          ê (e-circumflex) = ê als in ’t Fransche être (See the comment at “ô” below)

          ĕ (e-breve) = e als in de (gebroken e) (Comment: presumably [ə] (schwa), also rare but found mostly in alternate readings of initial stop+liquid, or as onset to other initial C-clusters)

          i = i als in Piet

          ī (i-macron) = i als in pier (Comment: this strikes me as an ill-chosen example-- at least in Engl., the -r affects vowel quality.  Would not Du. schieten or gieten more clearly demonstrate the “long-i” sound?  Is there in fact a difference in the sounds of Piet, pier, schieten?)

          ĭ (i-breve) = i als in pit (gebroken i) (My understanding is that in Du., this sound is close to [e]-- perhaps one could call it “lowered [i]” or “raised [e]”.  It has a clear origin in Kai-- AN *ə /__C{i,u}, vs. Kai “e” < AN *ə elsewhere. But the conditioning final Vs have been lost in Kai; I need to check, but I think there are i : ĭ minimal pairs.

          o = o als in pot

          ō (o-macron) = o als in boot, loten

          ô (o-circumflex) = a als in ‘t Eng. walk (gebroken o) (Comment: this is described as “tusschen o en a, met overgang van o op a”, thus something like [o] with an a- or schwa-like offglide, or perhaps a more open o (XSAMPA [O]?) with offglide.  In closely related Fordata, “oa” corresponds. Similarly, it is also probable that “ê” has an offglide; Fordata has “ea” in this case.)

          oe = oe als in boek

          u = ö als in ‘t Duitsche hören (Comment: this can safely be ignored-- occurs only twice, once in “dum”  ‘thumb’ < Du. duim, and once in “ruk” ‘already’ (a particle probably usually unstressed-- the cognate form “roak” in Fordata suggests this should have been written “rôk”)

          aa = a-a (twee lettergrepen met hiaat ertusschen) (Comment:  there are also forms with double “ee” and “oeoe”.  The question is:  what does he mean by “hiaat”? Glottal stop? or simply  transition (the so-called spiritus lenis).  Generally in Geurtjens’ time, Indonesianists used the dieresis, or an apostrophe, to indicate glottal stop between vowels...but not always, and not always consistently.)

If the simple vowels, as given above, can be assigned phonetic (if not phonemic) values, Geurtjens’ “diphthongs” are quite another matter--

          ai = ai als in ‘t Fransche paille (I interpret this as [E] with an i-offglide, though I suspect [ai] is intended.  Cf. “ei” below)

          āi (a-macron+i) = aai als in fraai (I think there are Kai words written “aai” too-- the same?)

          aw = aw als in lawaai (but isn’t -w- consonantal here? There are other problems with his use of w.  There is at least one case of “āw” (a-macron+w).  Is the problem that there are no Du. words with a true [aw] diphthong? What about “flauw”?)

          ei = ei als in leiden, doch in open lettergrepen korter uitgesproken (I believe this is properly [E] with i-glide and if so, how different from Fr. paille?, but perhaps [ai], which it sometimes seems to replace (see below).  “ei” is fairly rare, and in at least some cases seems to represent a borrowed or dialectal /e/.

          ēw (e-macron+w) = eeu als in leeuw (Is this [le:u] or [le:y] in Du? In Kai, it should be the former)

          oei = oei als in loeien

          oi = oy als in ‘t Engl. boy

          ōi (o-macron+i) = ooi als in kooi

          ow = ou als in kous (My understading is that this is [Ou], but I think he intends [aw]; indeed, the loanward kous ‘stockings’ is written “kaws” and “ou” may occur in conditioned environments-- following labials.)

He neglects to mention “iw”, but in the few examples, it reflects *-iwV# with loss of the final V. e.g. siw ‘nine’ < *siwa.

The origin of these diphthongs is interesting, so a few words about Kai historical phonology are in order.

The language has lost most final vowels from *CVCV and final *-VC from many *CVCVC bases, but in derived/possessed forms the original final V shows up (cf. tāv ~tawoen and taw ~tawak above). Most of the basic forms, thus, are monosyllabic (and I am inclined to view “long” vowels in such monosyllables as automatically lengthened, though the data do not bear me out).  Stress, according to G, is final in polysyllables, so we assume tāv ~tawoen = /ta:v/ ~/tavún/. Further, intervocalic *k is lost (probably via a glottal-stop stage), and *l is lost between all sequences involving *a and *u (i.e. a-u, a-a, u-a, u-u).  Also, *s > h initially and intervocalically-- and that led to an interesting development. (It is possible these changes are ordered in some way; e.g. not all *CVCV forms lost their final V.)

Thus the word for ‘dog’ *asu > Kai “yahaw”.  The y- is a regional development; but clearly the -asu developed as follows:  *-asu > **ahu, then due to a mis-phasing of the transition from low-back [a] to high-back [u], an intrusive echo of the [a] developed-- **ahu > **ahau.  With final stress, the [u] was reduced to a glide.  This development is quite consistent for all sequences of **ahi and *ahu, where cognate or proto-forms are known.  There are exs. for **ehu (note esp. lehēw ~lehoen), probably for **ohi (ohoi ‘village’ if cognate with Roti osi ‘gardens near the village’ or Nias (?) ose ‘house’, **eha (mehe ‘single, alone’ if < AN *m/esa ‘one’), and a possible dialect form for **ihu (Kai Besar sehiw, Kai sis ‘hiss’; elsewhere **sisu).

In a few cases, we find “ahei” instead of “ahai”, which leads me to suspect that “ei” is simply a variant of “ai”.

It seems that when *-k- had shifted to glottal stop, it was affected by the same “echo-vowel” rule, for we find forms like ngoe/rāi ‘grime’ < *daki, evidently via **raʔai, and lēw ‘bend’ < *leku, via **leʔeu, and yāw ‘I’ < *i+aku-- indeed this last is written “ya’aw” in a recently published Indonesian study, suggesting there is a glottal stop (unfortunately, however, the analysis in this study is not reliable, in my view).

The “ōi” diphthong would seem to have a similar derivation **-oki-. but the origin of /o/ in Kai and its relatives is obscure; almost none of the many forms have cognates outside the area.

The loss of *l in *a_u produces only a “aw” diphthong-- faw ‘hammer’ < *palu, and note wow ‘eight’ < *walu (here I believe “ow” is conditioned by the initial /w/; cf. also Jav. wòlu). There is also the somewhat inconsistent (?) waawn ‘like; likeness’ ~wa/waawk ‘image, picture’ presumably <*walu( ) on the basis of Ford. and Yamd. walu ~walut ‘image’.

Many instances of the “ai” diphthong derive from *-aya# with loss of the final V. Here I would assume a simple [ai] pronunciation rather than the French [Ei] = “-aille”, but it is possible the [a] is fronted and perhaps raised to [E] by the following i-glide