Marshallese Spelling Reforms

The public renunciation by several media giants of spelling reforms promulgated in Germany less than a decade ago has generated some discussion in the blogosphere, notably on Rainy Day, Crooked Timber, and a Fistful of Euros, the last two with wide-ranging comment threads.

To take the discussion a little farther afield, I'd like to add a glimpse of what teachers and writers of two Micronesian languages are up against. In Marshallese and Yapese, spelling reforms promulgated in the early 1970s have yet to take hold. (I use "promulgated" to mean 'imposed by specialists whose expertise is unimpeachable, but whose vision is clouded by thoroughly impractical ideals'.)

In both cases, the new orthographies suffer from two major drawbacks. (1) The only major literature written in each language has been the Bible. One tampers with holy scripture at one's peril. Just witness how many Christians still stick to the King James Bible or to Latin liturgy. (2) Linguistic experts were overzealously committed to the "one phoneme, one symbol" principle of orthography design. Among all the languages I've dabbled in, Marshallese, Yapese, and possibly Nauruan seem the most resistant to any orthography that places that principle above all others.

Here's a bit of a glimpse at Marshallese. Yapese will follow in another post.

Marshallese can be analyzed as having only four vowel phonemes that differ by height, but whose roundness (oh-ness vs. eh-ness) or backness (uh-ness vs. eh-ness) depend on their neighboring consonants. For instance, the vowel phoneme /e/ can sound like eh (open e), uh (schwa), or oh (o). In the textbook Spoken Marshallese (1969) the vowels are written i, e, a, & (yes, ampersand, but it was later replaced with an ę). The linguist Mark Hale refers to these four phonemes as cup of coffee, telephone, yinyang, and soccer ball, presumably because each word or phrase contains the varying sound values of the respective abstract phoneme.

Marshallese consonants distinguish only three main positions of articulation: lips (p, m), tongue tip (t, n), and tongue back (k, ng). Voicing (t vs. d, p vs. b) is not distinctive, but three secondary articulations are: "light"/palatal (py, my, ly), "heavy"/velar (p, m, l), and rounding (kw, ngw, lw). The parenthetical examples are not orthographic, but only intended to hint at pronunciation differences. One solution is to write the "heavy" consonants as if they were voiced: b, d, g vs. p, t, k, but that doesn't help with the nasals: m, n, ng (the latter also written g).

The "light" consonants front the neighboring vowels (e > eh), the "heavy" consonants back them (e > uh), and the round consonants round them (e > oh). Two different consonant types on either side can pull the vowel in two different directions, creating dipththongs.

Examples of "improvements" in the 1969 textbook orthography:

• 'Hello'
  Old: Yokwe yok
  New: Yi'yaqey y&q


• 'I'm going to Ailinglaplap / Jaluit'
  Old: Ij etal ñan Ailinglaplap / Jaluit
  New: Yij yetal gan Hay&l&gļapļap / Jalw&j

Since Marshallese makes too many distinctions for the standard keyboard, a linguistically optimal solution to facilitate literacy in Marshallese could go in either of two directions. The first direction seems by and large to prevail.

• Write more vowels than strictly necessary in order to keep them less abstract and because vowel diacritics are easier to keyboard, while relying on the neighboring vowels to show some of the consonant distinctions. This allows people to write with lower levels of linguistic or computer literacy.


• Write only the minimal (four) vowel distinctions, and add diacritics to distinguish all the consonants in order to show the full beauty of the underlying phonological system. This requires higher standards of linguistic and computer literacy before people can write their own language.

I would suggest that a socially optimal orthography might get by with even fewer alphabetic distinctions. People could write fewer vowels and consonants than would be optimal in isolation, while relying instead on sentential, semantic, or social context to reduce ambiguity. But this approach would make linguists feel rather less useful.

A revised Marshallese Bible was published in 2002. I'm not sure which of the several previous orthographic practices it relies on. Marshallese editions of (portions of) the Bible go back to the the 1860s, after the first missionaries had arrived, some of them from Hawai‘i.

Sample PDFs of Marshallese materials in a vowel-rich, consonant-poor orthography are accessible from the Pacific Area Language Materials website.

SOURCES: Heather Willson, A Brief Introduction to Marshallese Phonology (PDF, UCLA); Byron Bender, Spoken Marshallese (Hawai‘i, 1969); R.W.P. Brasington, Epenthesis and deletion in loan phonology (PDF, U. Reading, 1981).

UPDATE: David at Rishon-Rishon examines the question of "social optimality" at greater length, with evidence from Russian and Hebrew, noting that Russian writes consonantal palatalization on the vowels, while Hebrew writes velarization on the consonants.