I had always vaguely wondered about the place name Bikini (after which the famous swimwear was named); now, thanks to an exhaustive investigation by piloklok (Bob Kennedy’s linguistics blog, which has been promoted to the blogroll for this service to etymology), I know that the Marshallese form is Pikinni and that this “is composed of pik ‘surface’ and ni ‘coconut’.” Now I have only two questions: 1) Is the stress, as my Webster’s Geographical Dictionary and the Wikipedia article say, on the first syllable? and 2) What does “piloklok” mean?

(Via Literal-Minded.)

Update. See now Jory Dayne’s extremely informative comment in this Wordorigins thread:

Pik and Ni are glossed as ‘plane surface’ and ‘coconut’ in The Marshallese-English Dictionary—and according to Abo, Bender, Cappele & DeBrum, Pik,Ni is the origin of the place name Pikinni. I guess the real mystery, however, is why the Marshallese opted to single that particular islet out for that specific feature, when nearly all the other islets in the whole of the group share almost identical features: namely, a flat surface where coconuts are growing.

There is another gloss for Pik, and that is to fly, as in the flight of birds, or flapping. Given the tendency to name places for an apparently arbitrary, isolated event, this seems like it could also be a possibility—perhaps in a storm or what have you; that’s pure speculation on on my part, however.

As for stress, I would offer that PIK(ih)NI was probably the original pronunciation—for a couple reasons.

1. I’m willing to bet that the second ‘i’ in Pikinni/bikini is actually just an excrescent vowel… For instance, the Marshallese word for ‘doctor’ is taktõ (dahkduh)—borrowed from English. It is pronounced, however, as DAHK(ih)Duh. A non-loan word, jerbal follows the same pattern. It is pronounced JEHR(ih)bahl.

So you’re probably looking at the name actually being Pikni, with the excrescent vowel inserted between the two parts to help it conform to custom. Other place names also follow this pattern…

2. The ‘N’ in Ni, is a heavier ‘n’—the doubling in the current spelling (Pikinni) is probably to reflect that (although they have recently switched to using a cedilla beneath the heavy consonants to indicate this). So while the stress would be placed on Pik, the weight of the ‘n’ would give that syllable a stress of its own.

It’s worth mentioning, though, that the current pronunciation of Pikinni seems like it has changed to match the one common to English speakers (biKIni). I offer that with the caveat that I have only been in contact with Marshallese who have relocated to the U.S. (although most very, very recently)—so it may just be that group, while native Marshallese are “keepin it real.”


  1. I have an off-topic Bikini anecdote.
    I think that the “bikini” name was part of a marketing campaign — sort of like “what a dynamite swimsuit” (except kilotons bigger). There was a combined link to the bomb and to the sexy South Pacific, IIRC; I think that there was a Polynesian photoshoot. (Technically, Bikini is Micronesian, but PR don’t care).
    In the same way, the ball-point pen was called the “atomic pen” (yuanzi bi) in Chinese. This was back at the beginning, when the cost $10 rather than 3 cents. (Because of their difficult but beautiful writing system, I think that Chinese tend to be connoisseurs of writing implements.)
    Anyway, I used to know some Micronesians (mostly from Truk). They said that in Truk toplessness was no big deal, but that showing the legs above the knee was as provocative as toplessness is to us. (That’s what those “grass skirts” are all about).
    So once when the Micronesian guys caught some Micronesian girls out in their bikinis, the girls got into the water and wouldn’t come out until the guys went away.

  2. Not having heard Marshallese yet, I’ll have to answer the first question with more bookwork. There are actually conflicting accounts, unfortunately. Rehg (‘Proto-Micronesian Prosody’, in Edmonson & Gregerson (Eds.), Tonality in Austronesian Languages, 1993) cites a manuscript by Bender placing Marshallese primary stress on the final syllable of polysyllables. Secondary stress appears on alternating preceding light syllables.
    I’ve also come across a 1977 dissertation by one F.X.N. Zewen, ‘The Marshallese language: a study of its phonology, morphology, and syntax’, in which primary stress is claimed to be initial.
    I think when it comes to Marshallese phonology, what Bender says, goes. It’s possible that the generalizations made by Zewen (which may be the source of Webster’s and Wikipedia’s) are based on pitch rather than prominence. These often do not occur on the same syllable in Micronesian languages. (See Rehg’s article for details).
    As for piloklok, I made the word up in an attempt to get a clever blog name. Not sure if it worked though.

  3. Atomic pen! That’s great. I know a bar in Ueno that sells “Electric Brandy” (電気ブランデー), which is so named solely because electricity happened to be the hip new thing when they first started making it.

  4. Not sure if it worked though.
    Works for me!
    Electric Brandy: When you want to get lit…

  5. John Emerson: Linking ‘atomic’ to ‘pen’ was done in marketing the first ballpoints in the US – in 1945, Milton Reynolds advertised his pen at Gimbels as a “fantastic, atomic era, miraculous pen.” It went for $12.50 made Reynolds millions of dollars, but the thing didn’t work and two years later the price had dropped to 69 cents.
    It looks like either ‘atomic era pen’ was borrowed into Chinese as ‘atomic pen’, or some enterprising marketer had the same idea.

  6. Marshallese phonology (and orthography) is almost certainly the most confusing I have ever come across.

  7. Andrew Dunbar says

    If anyone here need it, I’ve taken some notes on Marshallese orthography and phonology. I wasn’t able to find it on the Internet a couple of years ago so if it’s still not around I can Unicodise what I have and post it to Wikipedia, Everything2, or right here.

  8. I posted on Marshallese spelling reforms a while back and included links to a few online resources.

  9. Great post, Joel. This is something I’ve never understood:
    Linguistic experts were overzealously committed to the “one phoneme, one symbol” principle of orthography design.
    Yeah, that principle has appeal in a sort of childishly literal way, but seriously, what can possibly be wrong or unscientific about using two symbols to represent a phoneme, especially when it can so greatly reduce the cost and difficulty of a writing system? I would have thought the possibility of using a standard keyboard would trump almost any other considerations, but no, generations of linguists and others competed to devise bizarre combinations of symbols for people who had not been literate. “I know, we’ll use an e with a vertical line over it for the first vowel, and an ampersand with an umlaut for the second!” “No, no, let’s use a modified Cherokee syllabary, with elements of Tibetan!” Grr.
    Andrew: I’d love to read what you’ve got if you post it somewhere.

  10. Hat,
    Just check out the Wikipedia entry on Marshallese and you’ll see plenty of Andrew’s Hippietrail. Good stuff.

  11. Andrew Dunbar says

    For those wishing to hear Marshallese, I just found this phrasebook site:
    Pronounciations courtesy of Mr. Alfred Capelle. The name “Capelle” comes up very frequently in relation to Marshallese or the Marshallese language. Sadly, I’m at work where none of the office computers have speakers so I’m unable to listen to it myself!
    As for my notes, they’re more sparse than I thought but I’m 99% sure they came from this dictionary:

  12. Great! Here‘s the phrasebook link with audio files, and here‘s the Marshallese-English Dictionary.

  13. Here’s what I could make out from my notes. I don’t have a photocopier card so I just took some hurried notes of a table which maps the orthographic symbol to IPA.
    The dictionary may be using some non-standard IPA symbols. I’m using a superscript “j” here which is the palatization modifier. My notes use two different superscript “w”s. One is the usual angular one. The other is curved and may be a superscript “ɯ” or “ɰ”. I’ll check again next time I’m at the university library.
    If you see some squares, try copying and pasting into MS Word and then changing the font to something like Arial Unicode MS or Code2000.
    a → ɑ
    ā → æ
    b → bʷ
    d → unclear
    e → e, ɛ
    I → I
    j → tʲ, c
    k → k, kʷ
    l → lʲ
    ļ → lʷ, lʷ
    m → mʲ
    m̧ → mʷ
    n → nʲ
    ņ → nʷ, nʷ
    n̄ → ŋ, ŋʷ
    o → o, ɔ
    o̧ → unclear
    ō → ə, ʌ
    p → pʲ
    r → rʷ, rʷ
    t → tʷ
    u → u
    ū → ɯ
    w → w
    y → y

  14. Just commenting so this thread will show up in Recent Comments and people will have a chance to notice the update with Jory Dayne’s information.

  15. John Cowan says

    Doing the same, plus explaining that what’s at stake in Marshallese is the four vowel phonemes (high, mid-high, mid-low, and low) with at least three allophones each. Shall they be given distinct symbols (thus violating “one phoneme, one symbol” from the other side) as was historically done? The linguists say no!

    Or shall the allophony be blamed on the consonants? This is the modern phonemic analysis, but makes for an inconvenient number and distribution of consonant letters. The Marshallese consonant system is stop vs. nasal, labial vs. coronal vs. velar, and palatalized vs. velarized vs. lip-rounded (which front, back, and round the neighboring vowels). Then there are three laterals with the same three coarticulations, three rhotics ditto, and plain /j/ and /w/, which do not merge with the coarticulated consonants.

    There’s a gap at the rounded velarized coronal stop, which means only 25 instead of 26 graphs needed for the consonants and four for the vowels. This is a system that does not fit the Latin alphabet framework at all, with its three vowel heights and useless front vs. back vowels, and only voiced vs. voiceless vs. nasal distinctions in the consonants (and not even those in the liquids and nasals). It can be done, it has been done, but only at the expense of unlovely artifice.

    Note that both Russian and Irish blame consonant phonemes (not even allophones any more) on the vowels in their writing systems (which is why Irish should use Cyrillic, dammit, even apart from its great advantage of Flummoxing the English).

    Update: A nearby post is entitled “The Head Heeb on the Fur of Cairo”, which I thought was a pronouncement by the Ashkenazi Chief Rabbi of Israel (there are two) about the quality or kosher-ness of Egyptian fur for shtreimls, but turns out to be a reposting from a blog called The Head Heeb (defunct) about the Fur (people from Darfur) refugees in Cairo.

  16. Thanks for bumping the thread; I’ve taken the opportunity to add the italics I lazily left out of the Dayne quote from Wordorigins a dozen years ago.

  17. Also, I’m sorry to see Kennedy stopped posting in 2008, glad to see the blog is still up, and sorry to see he’s let it get infested by spammers (why don’t people close off comments when giving up on posting?).

  18. Proto-Malayo-Polynesian *niuR coconut.

  19. There’s now a free revised and expanded version of the Marshallese dictionary online, with an orthography slightly revised to make it easier to type, I think.

  20. Nice, thanks!

  21. Lars Mathiesen says

    I don’t have a photocopier card — that was 2005. Now you just take a picture with your phone. I’d keep mine for that even if I uninstalled all the other apps.

  22. January First-of-May says

    I’d keep mine for that even if I uninstalled all the other apps.

    I used to keep a phone that had a really good camera (by cheap phone standards) until, sometime around January this year, its second battery started working so badly that even on charger it could barely take a flash photo before shutting down. (That was after we replaced its first battery, which prior to replacement also ran to similar levels of non-workingness.)

    I’m hoping that some day we’ll find a third battery for it somewhere and it will start doing photos again, but it’s such an old model (Samsung Galaxy 4 or thereabouts) that I doubt we could find something that exotic outside places like eBay (and even there it would probably come used, i.e. with lessened quality).

  23. David L. Gold says

    “This is something I’ve never understood:
    Linguistic experts were overzealously committed to the “one phoneme, one symbol” principle of orthography design.
    Yeah, that principle has appeal in a sort of childishly literal way, but seriously, what can possibly be wrong or unscientific about using two symbols to represent a phoneme, especially when it can so greatly reduce the cost and difficulty of a writing system?”

    Rather than “one phoneme, one symbol,” the desideratum is that each phoneme represent just one grapheme and that each grapheme be represented by just one phoneme (= orthographic bi-uniqueness, to use the term coined in the Prague Linguistic Circle, probably by Josef Vachek).

    A grapheme may be a monograph (such as s representing the Hungarian phoneme /š/), a digraph (such as sh representing the English phoneme /š/), a trigraph (such as sch representing the German phoneme /š/), and so on.

    Esperanto spelling is fully bi-unique and Estonian spelling almost is (“Although the Estonian orthography is generally guided by phonemic principles, with each grapheme corresponding to one phoneme” + details []).

    Another desideratum, which, like orthographic bi-uniqueness, may not be fully attainable, is employment of just the symbols available on whatever keyboard is most used or most available in the speech community in question.

  24. David Eddyshaw says

    using just the symbols available on whatever keyboard is widespread in the speech community in question

    I’ve never quite fathomed the rationale of the pre-2016 Kusaal orthography in this regard (although I’ve developed a reluctant respect for it over the years, specifically because of how its various somewhat odd conventions interlock to result in a system with remarkably little actual ambiguity, given that the language has 20 distinct consonant phonemes, nine distinct short vowels along with contrastive length (where there is a three-way distinction), nasalisation and glottalisation in most possible combinations of all those features, and lots of diphthongs in a weird asymmetrical distribution caused by relatively low-level, probably recent, sound changes.)

    Typeability on a keyboard set up for English seems pretty clearly to have been a factor, and the system succeeds in this perfectly with sole odd exception that it also uses ŋ (/ŋ/ is not at all a rare phoneme in the language, either); the more mysterious as a small tweak to the existing spelling conventions would have made it perfectly easy to use ng instead.

    The 2016 revision seems to have capitalised on the fact that people typing on phones can happily summon up keyboards at will to suit the language, which seems fair enough (though they should have just introduced tildes to mark nasalisation while they were at it.)

  25. There was, broadly speaking, a back-and-forth in orthography fashions in the past few decades: ASCII, because computers can only do ASCII; then anything Unicode, because computers can do anything Unicode; then ASCII again, because texting can only do ASCII; and now back to Anything Unicode Goes.

  26. John Cowan says

    Up to a point, Minister. Granted, I have written a keyboard driver (for Windows only, alas) that allows using a standard US or UK/Irish keyboard to type almost 1000 symbols while usurping only AltGr (right Alt). But to do so and still maintain some sort of mnemonicity requires ungraceful and sometimes painful hand positions that severely slow down touch-typing. (It helps, I find, to have learned the piano first.) So the keyboard is the true limitation on rich orthographies, and this is unlikely to change soon or ever.

  27. @John Cowan: “Touch typing,” as it was originally conceived, largely no longer exists, because it was a skill that was specific to a certain kind of technological and business practice. As I recently commented on a Language Log post on the word keyboarding:

    … However, although the physical apparatus used for typing was a computer, the class (and the computerized typing instruction programs that the teachers recommended) still focused on teaching students how to type up handwritten documents. This meant, for example, that students were forbidden to look at the keyboard, on the assumption that they would need to keep their eye on the document they were keying in. It took a few more years before it dawned on the teachers and programmers teaching keyboarding that henceforth, people would be composing their own documents at the computer, so there was nothing wrong—for example—with looking at the keyboard for some guidance.

    I can touch type, in the sense that I no longer need to look at the keyboard at all in order to type, although it certainly goes more quickly if I glance down occasionally.

  28. With me it’s the opposite — if I look at the keyboard, I type slower and worse. No idea why.

  29. That passage is incorrect. My 9-year old touch types. On a tablet no less. Her friends are all learning that way. Watching the keyboard would be like learning to watch your own mouth while you talk.

  30. David Marjanović says

    Typewriters set up for French are why the Albanian spelling system uses ë and ç but sh, zh, xh.

    (…X is the rather rare /d͡z/.)

    The more di- and trigraphs there are, the harder it generally is to make sure they remain unambiguous. In German, you need to understand the morphological composition of a word to know whether word-internal sch followed by a vowel is /ʃ/* or /sx/. Or consider Gotham City in English, which seems to be universally pronounced with /θ/ but is etymologically goat + home and has been spelled Gottam before.

    * /ʃː/ where applicable

    I touch-type on the German keyboard layout. Anything that’s not on that layout I copy & paste from the character map; that includes ë, ï and ç.

  31. @Ryan: So, like those high-school keyboarding teachers a quarter century ago, I am now the one thinking a generation of technology behind.

  32. A. Sasportas says

    @DM “Or consider Gotham City in English, which seems to be universally pronounced with /θ/ but is etymologically goat + home and has been spelled Gottam before”

    You are right that Gotham as an informal name of New York City has /θ/. For Gotham as the name of a village in Nottinghamshire Wikipedia gives /t/.

  33. @Brett I may be overstating on the basis of my sample of one class of 3rd graders.

    But it accords with my own impression that touch typing is amazingly valuable in the computer era, more so than previously, when it qualified you to be a typist. By not focusing on the keyboard I can focus on the output, while typing more or less fluently.

    In fact, it irritates me no ebd to key things in without a keyboard, at quarter sped and less accuracy, as right now.

  34. In fact I’m a fluent touch typist for a range of control qnd alt commands that allow rapid rewriting. I don’t know how people survive without that skill. Do you actually think it through before typing so that it comes out beautifully the first time?

  35. One funny fact is that I make typos more often when I am typing in an inconvenient position.*
    You will say it is expected, but it is phonological typos, just 10 or more times more often.

    Vowel-wovel (Russian does not distinguish v/w).

    *I mean, if keyboard is not right before me or if it is balancing or if I am half-lying on my side

  36. “atomic pen”

    Наномойка – трехфазная мойка автомобиля, при которой на второй и третьей фазе используются наношамунь и нановоск.

  37. Jen in Edinburgh says

    Yetholm is (or are) YETT-om. (And possibly the same name as Gotham – I would have guessed the other proposal, gate-village, but it’s not really far from Gefrin/Yeavering, which is another goat name.)

    Possibly my favourite unguessable pronunciation is Dumyat, dum-EYE-at, but that’s kind of the opposite of a digraph – one letter, two completely different possible sounds.

  38. Good lord, that’s amazing. Wikipedia:

    Dumyat or Dunmyat /dəˈmaɪ.ɪt/ (Scottish Gaelic: Dùn Mhèad) is a hill at the western extremity of the Ochil Hills in central Scotland. The name is thought to originate from Dun (hill fort) of the Maeatae.

    How the hell did anyone come up with that spelling? And why is lenited m (mh) rendered as m instead of v?

  39. PlasticPaddy says

    All i can say is that this happens in Irish placenames also, i.e Dún Mháire / Doonmara. Maybe there are other examples that are not placenames. I think one should probably bear in mind that (a) the process of generating English or Gaelic spellings could be haphazard, (b) the recorder or the informant could have spoken a dialect and (c) the rule about Mh is clear when it is construed as the beginning of a separate word, but multiple-word placenames are subject to erosion and reanalysis.

  40. All very true.

  41. David Marjanović says

    touch typing is amazingly valuable in the computer era, more so than previously

    Exactly. That’s why I learned it, on a typewriter, with my mom’s old textbook. After I had gone through the book and skipped over half of the exercises, I got used to the slightly different computer keyboards, and now my longest paper has almost 400 pages.

  42. Jen in Edinburgh says

    My only guess is that it’s by anology with pyat, magpie, pronounced similarly, which turns up in other placenames – but I’m not sure if there are any locally.
    (ETA: Pyat- names, not magpies!)

    Final -n is a bit funny about lenition anyway – in standard Gaelic d and t don’t bother to lenite after it, and it wouldn’t surprise me at all if other letters were like that in other dialects.

  43. John Cowan says

    And why is lenited m (mh) rendered as m instead of v?

    Perhaps because lenited /m/ was originally /ṽ/, distinct from lenited /b/ which was /v/, so in some varieties or cases it may have lost the labiality and kept the nasality. (The same distinction was made in Middle Sindarin, but more so because Sindarin had no /b/, /d/, /g/ except prenasalized /ᵐb, /ⁿd/, /ᵑg/ plus /ld/, /rd/, which were written with separate letters.)

    Final -n is a bit funny about lenition anyway

    I think that’s because of the Old Irish distinctions between n~N, l~L, r~R/, now lost in most varieties as phonemic distinctions (they are usually called lenis/fortis, but we don’t really know) but still affecting mutation patterns.

    But I may be talking nonsense, as Trune says.

Speak Your Mind