Language Hat recently linked to Abecedaria, a newish blog by Suzanne McCarthy devoted to writing systems. McCarthy already has quite an archive built up, and I was intrigued to find an entry on a Caroline Islands syllabary (CIS). She links to a proposed Unicode table on Michael Everson’s website for symbols in the syllabary, and to CarolineIslandScript.com, a site devoted to the system that offers a lengthy discussion regarding the its possible historical origins. Its owners (Dan and Andy Koch) really seem to be keen on the notion that CIS is a Woleaian invention, contrary to Reisenberg & Kaneshiro (1960).

The Koches’ site includes an FAQ area that starts with the following background:

The name Caroline Island Script (here CIS for short) refers to a writing system which originated in the Central Pacific, and only in a small group of islands in that part today called Micronesia ( and the part which was once called the Caroline Islands). The writing system is comprised of over 100 individual signs (“graphemes”), representing some 100 syllables , and which linguists have divided into 2 general categories ( “Type 1” and “Type 2”), depending on the signs ultimate origin. The script was (and may be still) used to write of concepts and ideas within the native dialect, today called Woleaian, which is the name for the native language used among these islands.

NB “these islands” means the Woleai group, a cluster centered around Woleai atoll, and not the entire Carolines. I’m familiar with Woleaian from H.M. Sohn’s grammar (1975); it’s a language spoken on Woleai atoll, along with a few nearby islands, including Faraulep, Ifaluk, Elato, Lamotrek, and Eauripik.

My first glance at the Unicode character table told me something was off slightly. Lest you think I object to this implementation, I do not – Everson has done thorough work. The issue is with the accompanying phonetic values, which Everson obtained either from R & K (1960) or from one of their sources, Smith (1956). I’ll refer to these transcriptions as the R&K values. Anyway, the script does not have a mechanism for marking segment length, although length is contrastive for both consonants and vowels in Woleaian. (Turns out that this simply a shortcoming of the syllabary, as Reisenberg & Kaneshiro discuss). Further, probably because of this, the “phonetic values” given for each symbol do not quite match either the modern Roman-based Woleaian orthography (as used in Sohn 1975), and viewable in action here), or the true phonetic values Sohn associates with the modern system.

I can’t put the CIS symbols in this post, but I can again link to Everson’s table. The following table shows the R&K values arranged C x V:

For contrast, here are the consonants of Woleaian, in Sohn’s transcription, with phonetic values noted where different. I give oral consonants in singleton and geminate rows; note several columns in which geminates differ from singletons by some secondary feature.

And here are the vowels:

Curiously, eo and oa only occur long.

Now if you compare the R&K values with Sohn’s segment inventory, some segments are missing, and other symbols seem to stand for sounds the language doesn’t use. For example, the R&K values have no [l], or anything that would line up with “g”/[x]. The vowel digraphs are also different. There was only one way to settle this … to the library!

Turns out that Reisenberg & Kaneshiro tell a fascinating story of why who knew how to do what when. They describe how a missionary named Snelling was lost or forced ashore in 1905 at Eauripik, an island in the Woleai group, with a crew of Trukese sailors. They had knowledge of a Roman-based alphabet developed near Truk (to the east of the Woleai group) which they taught to the Woleaian-speaking locals on Eauripik, who adopted it as a syllabary. R&K refer to this system as Type 2; its symbols represent word-initial vowels and all consonants + [i]. The Type 2 symbols can be seen in the rightmost two columns of Everson’s Unicode table, and their Roman origin should be clear.

The castaways made their way back to Truk, stopping at various places, including Woleai (where Snelling died) and Faraulep. Either because of this trip or for independent reasons (or both), Type 2 diffused through the area. At Faraulep, the system was expanded into a larger syllabary with novel intricate symbols (developed possibly in a rebus manner) called Type 1, whose symbols represent consonants in combination with other vowels. Type 1 then spread throughout the region, back to places where Type 2 had already been learned. There is no overlap in the sounds that the symbols represent, and the two systems were used in tandem in written texts. By the time missionaries got back to Woleai atoll in 1913, it looked as if an indigenous writing system had been found.

Anyway, not only do Reisenberg & Kaneshiro provide a comprehensive summary of the inventories of symbols provided by individual informants, they go over quite a bit of detail regarding the phonetic values associated with each symbol. The non-overlap between the spelled-out values of the symbols and inventory given by Sohn seems to be attributable to the lack of encoding for the length contrast.

Thus, any symbol whose syllable is spelled with a “b” in the chart lines up with “b” or “bb” (the velarized labials) in current orthography. Likewise any “n” stands for [l] or [nn] (the latter being the long alternant of the former), and any “k” stands for [x] or [kk] (again, where the latter is the long alternant of the former). For example, the GA symbol could be used for [xa] (spelled ga) or [kka], while the NA symbol could be used for [la] or [nna]. Moreover, all the vowel symbols (singletons and digraphs) stand for different vowel qualities, disregarding length. So “i” refers to [i] or [ii]. The criticial ones are UU = iu = [ü], OE = eo = [ö], and AE = e = [e].

So, just in case any other Woleaianists came across the script and scratched their heads, hopefully this clears it up.

Reisenberg, S.H., and S. Kaneshiro. 1960. A Caroline Islands Script. Bureau of American Ethnology Bulletin 173, 269-333. Washington DC: The Smithsonian Institute. (This was difficult to find – E51 .U6)

Sohn, H.M. 1975. Woleaian Reference Grammar. University of Hawaii Press.