Jump to content

Talk:Marshallese language/Archives/2019/October

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia


Native name of the language

Back when I first became interested in Marshallese I heard that "Ebon" was the correct native name and thought I'd confirmed it - but can't remember any of my sources. A couple of years later I saw "Kajin M̧ajeļ" - but not in this correct up-to-date Unicode (nice!) - can we try to find out if they are both correct or one might need to go. Marshallese dictionary anyone? (: — Hippietrail 00:03, 16 Sep 2004 (UTC)

Kajin M̧ajeļ is literally "tongue of Marshall" ("M̧ajeļ" being the closest you're going to get to "Marshall".) The Ethnologue lists Ebon as an alternate name, but it doesn't say who uses it—whether it's a native designation, an alternate English designation, or what. Incidentally, the 1911 encyclopedia refers to Ebon as the main name, with Marshall in parentheses (e.g. [1]).
It looks like it was called Ebon after the Ebon atoll in the Marshall Islands—especially so as the two dialect names given are Ratak and Ralik, also geographical names (see Geography of the Marshall Islands)—apparently because Ebon was one of the first islands with European trade (according to History of the Marshall Islands). I'm guessing it got to be called "Marshall(ese)" proper after it was identified with the group as a whole being "The Marshall islands". Yesno? —Muke Tever 04:07, 16 Sep 2004 (UTC)
The most commonly used form is Kajin Majõl or simply Majõl. Occasionally it is even expressed as ri-Majõl (technically ri- is a personal prefix, so ri-Majõl properly refers to a Marshallese person, frequent usage notwithstanding). mssever (Talk | Blog) 01:12, 30 May 2006 (UTC)

Ampersand

I could have sworn I read that Marshallese used the ampersand ("&") as a vowel. Is this correct? Ampersand links here this reason --[jon] [talk] 17:09, 14 Mar 2005 (UTC)
Yes you did read that but it is wrong. There exist some simplistic teaching aids for the Marshallese language which were typed on machines which lacked the exotic Marshallese letters. Some people seem to have assumed that this is the correct orthography. Hardly surprising since information on Marshallese is so scarce. There is a dictionary in existence however, and it contains a brief description of the phonology and orthography. You won't find it in too many bookshops but university libraries should have a copy. — Hippietrail 11:12, 15 Mar 2005 (UTC)
This site (which unfortunately appears to be commercial), quotes from a book, (The Languages of the World, by Kenneth Katzer) that appears to indicate that Marshallese does use the ampersand to indicate "a vowel sound somewhere between e and i." I'm sure that I have seen a version of the quoted book somewhere at a library here. If I can find it I could get a page number. Miss Madeline | Talk to Madeline 22:23, 5 July 2006 (UTC)
Indeed, Katzner says so: The alphabet includes an ampersand (&) which represents a vowel sound somewhere between e and i, as well as seven special consonants:. I am unable to find Unicode equivalents for the seven consonants; they look like m,n,t with a vertical stroke on top (placed like an accent, but vertical instead of diagonal) and g,n,l,r with two of these vertical strokes. Katzner also gives a sample text that shows the ampersand (which seems to be frequent) and some of the special consonants; also, an apostrophe appears after the vowel i in some words. 88.73.5.184 22:52, 9 October 2006 (UTC)
The ampersand is found in Byron Bender’s “Phonetic Spelling” in the book Spoken Marshallese. Katzner took his example text from this book (pg. 61). I believe Katzner mistook the phonetic spelling for the practical orthography. The characters m̍ n̍ t̍ g̎ n̎ l̎ r̎ are all available in Unicode. The single vertical line accent is U+030D and the double vertical line accent is U+030E. Thus to type n̎, U+006E U+030E. Languagegeek 11:12, 1 November 2006 (UTC)

re: orthography

i've heard that there is more than one orthography used to write Marshallese. the one chosen depends on religious affiliation. that's all i know. just noting it here because i wouldnt want to Wikipedia to offend some speakers. peace – ishwar  (speak) 04:45, 25 October 2005 (UTC)

I would if I knew them. Unfortunately, my greatest familiarity is with the modern official standard. I never became fluent in the language, and left the archipelago when I was a teenager, and only started studying it again lately. Resources online are pretty scarce... - Gilgamesh 08:15, 25 October 2005 (UTC)
I haven't heard of differing orthographies by religious affiliation, but my primary association was with Protestant Christians. Catholics are a minority, and other religions are very small. Of course, indigenous religious practices creep in. But I don't know of any linguistic difference by religion.mssever (Talk | Blog) 01:12, 30 May 2006 (UTC)
I might be wrong, but I think I read somewhere that South-African Muslims (or Muslim clericks) write/wrote Afrikaans with the Arabic alphabet. But whether they used it because of their religion, or they didn't have any knowledge about Latin letters, I don't know. Mulder1982 15:35, 11 October 2006 (UTC)

Marshallese Scouting

Can someone render "Be Prepared", the Scout Motto, into Marshallese? Thanks! Chris 04:32, 8 August 2007 (UTC)

Marshallese language template

If you are a native speaker of Marshallese then you can help translate this template into your own language:


mahThis user is a native speaker of Marshallese.

Edit


--Amazonien (talk) 22:55, 21 January 2009 (UTC)


Although I am not a native speaker of the marshallese language i did spend two years in the islands teaching for the Lds church. Be prepared could be translated to

Kwon pojak... You should be prepared...Be prepared

Hope this helps —Preceding unsigned comment added by Noniep (talkcontribs) 07:32, 17 July 2010 (UTC)

Any publication activity?

Are there any periodicals or any publishers of books in the language? If so, it might be noted what spelling those use, and if not, the fact might be stated. -- 14:29, 13 December 2010 (UTC) —Preceding unsigned comment added by 77.7.148.95 (talk)

Syllable structure

"All syllables begin and end with a consonant..."

I can immediately think of common words without CVC structure, which contradict this statement, such as: pronouns ña (me), kwe (you); articles ņe (there-by-you), eo (the); "nouns" āt (name), ņo (wave); and "verbs" lo (find), bu (shoot).

What is the source of information for the section "syllable and phonotactics"? Does such a source have explanations for these other syllable structures?

Seraphimek (talk) 05:33, 16 February 2011 (UTC)

According to Willson's "...Marshallese Phonology", "Marshallese syllable structure may be CV, CVC or VC" (p7). This source contradicts current section on "syllable and phonotactics" in the current article.

I will post Willson's statement in the current syllable and phonotactics section above the CVC constraint statement, for which I will post a "citation need."

The CVC constraint is dubious and needs to be cited. Entries in the Marshallese-English Dictionary (MED) give CVC syllable structures for all of the examples of "contradictions" in the above post: < ña > is / ŋah / ... basically the MED implies there are consonants in the phonology that do not appear in the orthography. So perhaps the apparent counter-examples of CVC structure indeed have that structure.

However, there is not a stated CVC constraint in the MED, and more importantly, velar approximate / h / is not given as a phoneme either in the MED or in Willson's "Introduction to Marshallese Phonology" (p2).

Seraphimek (talk) 22:04, 16 February 2011 (UTC)

IPA?

The usage of IPA is very confusing. /ɦʲɜtʲ ɦʲɘtˠ ɦˠamˠ mʲɘɦˠʷɨrˠ/ or [ɛ̯ɛt͡ɕ e̯e͡ɤt̴ ʕɑ͡æm me͡owu͡ɯr̴]

  1. There is no explanation why these two pronunciations. Are they different dialects?
  2. You should use slashes for broad transcriptions and square brackets for narrow transcription, not for variant pronunciations.--90.179.235.249 (talk) 18:35, 24 June 2012 (UTC)
They are broad and narrow transcriptions, not separate dialects. Marshallese phonology is very underspecified in ways that make IPA transcriptions a challenge. Vowel phonemes are undefined by backness or roundedness, and approximants are undefined by height. However, both have a large selection of allophones whose clear pronunciation is essential for telling both the vowel and consonant phonemes apart when they directly touch each other. It's explained better in the Marshallese phonology article. However, to clear up confusion, I changed the Template:IPAc-mh text to verbally specify that one pronunciation is broad and the other is narrow. - Gilgamesh (talk) 11:11, 25 June 2012 (UTC)

Phonology issues

I see the article(s) have been greatly modified in my absence, and I'm trying to review them.

Vowel allophones

It seems that A Brief Introduction to Marshallese Phonology and the Marshallese-English Online Dictionary: The Sounds of Marshallese disagree on the quality of the vowel allophones, in ways that overlap into different phonemes. Brief uses:

i ɯ u
ɪ ɤ ʊ
e ʌ o
ɛ a ɔ

But MOD uses:

i ɯ u
e ɤ o
ɛ ʌ ɔ
æ ɑ ɒ

I've been using the latter in articles and IPA templates because I've been using the MOD heavily — I believe Brief is otherwise correct. The question is, which system are we to use in the articles? It is actually possible that both pronunciations are in active use as dialectal differences, but MOD's system is more intuitive both to Marshallese orthography and to vowel height.

Oh, so that's what "MOD" means: Marshallese-English Online Dictionary. It's used in the article with no explanation. I'll add one at the first use. --Thnidu (talk) 05:36, 2 January 2014 (UTC)

Approximants

It seems ambiguous to use /j ɰ w/ for the approximants, since the actual phonemes are unspecified for vowel height, while the symbols [j ɰ w] are specified by the IPA as strictly equivalent to close semivowels [i̯ ɯ̯ u̯]. The approximants assume the height of the vowels they neighbor. And since some vowel syllable components are desyllabified in speech (like the i in io̧kwe analyzed in the MOD as {yi'yaqey}, i.e. /jɨ̯jakʷɜj/ or [jæ͡ɒɡʷɔ͡ɛɛ̯]), the articulation [j] is actually an allophone of the phoneme /j/ only where it neighbors /ɨ/, or in the desyllabified sequence /jɨ̯j/. So while there is precedent in using /j ɰ w/ as phoneme symbols for their simplicity, I propose they not be used where they clash with allophonic articulations. Another potential option, is to retain /j ɰ w/ as phonemic symbols only, and only use semivowels for allophonic articulations [i̯ ɛ̯ æ̯   ɯ̯ ɤ̯ ʌ̯ ɑ̯   ɔ̯ ɒ̯]. (Before, we had been using [ʕ] in some allophonic transcriptions, but this was strictly a graphical simplification of [ɑ̯], the same way [j] is a graphical simplification of [i̯].)

Signature

Thoughts? - Gilgamesh (talk) 04:39, 28 December 2012 (UTC)

Here are my thoughts:
Vowels: The two different systems for notating Marshallese vowels aren't really contradictory. The IPA symbols are useful, but one can hear finer divisions in the vowel space than the IPA can notate. In fact, the same speaker will pronounce a vowel slightly differently each time he pronounces it, depending on slight differences in how he positions his articulators and the context that the vowel is in. A vowel like /æ/ in English doesn't just have one height and backness value -- it's actually a "cloud" of possible values, some more likely than others. Detailed phonological experiments can figure out what these clouds look like.
Now, a Marshallese vowel like /ɪ/ will sometimes be pronounced more high and sometimes more low. The IPA doesn't give an exact cutoff between /e/ and /ɪ/, and for most intermediate values one could use either symbol. Since Marshallese has four vowel heights, and the IPA has five possible heights, it seems to be mostly a matter of notation whether you'd rather use /i ɪ e ɛ/ or /i e ɛ æ/. Therefore, I think it's probably better to stick to the former, since it's in a published paper rather than on a website.
Approximants: I'm not comfortable using new notation which is not used in any of the sources, since it might be considered original research. However, if you can find a published paper or book which proposes this, then it's definitely worth mentioning in the article. Mo-Al (talk)
The cloud paradigm made sense, but the two systems still pose an ambiguity — the contradictory overlaps of [e ɛ o ɔ] between them. The broad vowels are not in dispute, but the narrow vowels are necessarily going to vary, just like the highly variable narrow vowel systems of different accents of English. The MOD (Marshallese-English Online Dictionary) is not just a website — it is also a hypertext adaptation of the MED (Marshallese-English Dictionary), first printed in 1976. Brief is a useful scientific paper, but MED remains the most authoritative dictionary of the Marshallese language. When I started adding more Marshallese entries on Wiktionary, I used its system of vowels just as I used the online dictionary itself for linked references.
As for approximants, I changed the IPA templating system to use plain approximant symbols for the broad transcription, and pure semivowels for the narrow transcription — theoretically, even most of the semivowels could be omitted, but they're useful for establishing syllable boundaries at consonant-approximant, approximant-consonant and approximant-approximant margins. - Gilgamesh (talk) 05:51, 29 December 2012 (UTC)
I'll concede on the status of the MED, though it would be best to have a reference to the print version with a url=... attribute. Maybe I'll get working on that. (Update: done!)
I will say that the mid vowels are still not contradictory. The IPA vowels /e/ and /ɛ/ actually do overlap, as they may both be used to indicate mid vowels. It actually turns out that some languages with both vowels /e ɛ/ have /e/ mid and /ɛ/ mid-open, while some have /e/ mid-closed and /ɛ/ mid. Therefore I've changed to wording in the article to say that the MED and Willson *notate* the vowels differently, rather than saying that they give different phonemic realizations (which they may not -- at least, not neccessarily).
Regarding the approximants, I'd really rather not use anything which isn't sourced. Writing "breadfruit" as [mææ̯] isn't supported by the sources that the article uses. Since we seem to disagree about this, I think it might be good to not use the IPAc1-mh template until we can come to an agreement about this issue. Mo-Al (talk) 02:54, 30 December 2012 (UTC)
No, not necessary...a few small easily-revertable changes will omit the coda approximants. I'll do that now. - Gilgamesh (talk) 05:41, 30 December 2012 (UTC)
Great! Thanks. Mo-Al (talk) 05:56, 30 December 2012 (UTC)
Just to let you know — the phoneme gruntwork is done in Template:c1-mh for broad transcriptions, and Template:c2-mh for narrow transcriptions. Template:IPAc-mh common doc documents the usage and syntax. - Gilgamesh (talk) 06:01, 30 December 2012 (UTC)
One more issue: When using this template, the article contains the following: "[mæ] 'breadfruit', [mæ͡ɑ] 'but', and [mæ͡ɒ] 'taboo'" This seems problematic. The source implies that these words should be a (seeming) minimal pair for the vowels [æ ɑ ɒ] in the context [m___]. It doesn't give any hint of there being a preceding [æ] glide. While we do have a source for the existence of Marshallese diphthongs, it only mentions the case where there are *explicit* consonants with different secondary articulations -- not when there are *underlying* /w j G/ which don't surface. Therefore I'm not sure that this transcription is supported by the source. Mo-Al (talk) 06:09, 30 December 2012 (UTC)
Oh, I looked them up in the dictionary. Breadfruit and taboo both use the palatalized /mʲ/ phoneme, with their MED phonemic notation being {may} and {maw} respectively ({m} is /mʲ/, {ṃ} is /mˠ/). I couldn't find the corresponding entry for "but", but perhaps I'm just not looking right. So yes, the [æ] diphthong onset necessarily has to be there, because only front vowels can immediately neighbor /mʲ/ within a syllable. - Gilgamesh (talk) 07:35, 30 December 2012 (UTC)
What source says that "only front vowels can immediately neighbor /mʲ/ within a syllable"? (In fact, Willson (2003:7) seems to explicitly claim the opposite...) Mo-Al (talk) 07:40, 30 December 2012 (UTC)

Willson page 2.

"Marshallese has four vowel phonemes. These phonemes are specified for [height] and [ATR] but are not for [back] and [round]. ...... Marshallese vowels are assigned the qualities [back] and [round] by the surrounding consonants. Vowels become [-back, -round] in the environment of palatalized consonants, [+back, - round] in the environment of velarized consonants and [+back, +round] in the environment of labialized consonants."

And page 3.

"When a vowel appears between consonants that have different secondary places of articulation, the vowel retains its height specification but consists of a smooth transition from one target to another (Bender (1968), Choi (1992)). For example if a vowel is found in the environment CʲVCʷ, there is steady movement of the tongue throughout the vowel from a [-back, -round] to a [+back, +round] position, such as in jok ‘shy’ where there is a transition between [e] and [o]: /tʲeokʷ/. In other words, [eo] is a dipthong. There are therefore twenty four Marshallese diphthongs, six at each specification of height and ATR."

Since Marshallese is a vertical vowel system, the vowel's backness and roundedness is intimately linked to the consonant it neighbors, which is why it diphthongizes when it is surrounded by two consonants of differing secondary articulations. So though the twelve articulations and 24 diphthongs are technically allophones, at the same time they are not — they directly reflect the qualities of the consonants they neighbor. - Gilgamesh (talk) 08:21, 30 December 2012 (UTC)

Also, I remembered another reason why I used semivowels to mark the approximant articulations. In many environments they seem to disappear, but they don't completely, since they mark a moraic coda, and where they cluster with other consonants, they mark an abrupt repositioning of the tongue's secondary articulation. For example, Jaluit has two name versions: Jālooj has the MED phonemes {jalẹwẹj} /tʲalʲɘwɘtʲ/ [t͡ɕæle͡oo̯o͡et͡ɕ], while Jālwōj is {jalwẹj} /tʲalʲwɘtʲ/ [t͡ɕæl.o̯o͡et͡ɕ]. I used semivowels not as original research, but as an inference of approximants and semivowels actually being the same thing — in the strictest specification, [j] is interchangeable with [i̯], [w] is interchangeable with [u̯], [ʕ̞] is interchangeable with [ɑ̯], etc. And since the approximants assume the height of the vowels they neighbor, the approximants may be effectively invisible in many places, but they still have the metric length of a consonant wherever they occur — MED has plenty of words where geminated /jj/, /ɰɰ/ and /ww/ occur, and an approximant consonant has to necessarily exist metrically for it be geminable. In Ij Yokwe Lok (one of the most iconic songs of my younger years), one of the lyric words is indeeo, which MED analyzes as {yindẹyyew}, in other words /jɨnʲrʲɘjjɜw/ [i̯inr̟ee̯ɛ̯ɛ͡ɔɔ̯] - Gilgamesh (talk) 08:21, 30 December 2012 (UTC)

I agree that Marshallese has a vertical vowel systems, and that the sources say that the vowels become diphthongs around *surface* consonants with secondary articulations. The thing is, I haven't seen any source that says explicitly that the vowels become diphthongs around *underlying* consonants that don't surface, like /j w G/. Also, Bender (1968:21–22) really does give the phonetic forms of the three words as [mæ] [mɑ] [mɒ], *not* as [mæ] [mæ͡ɑ] [mæ͡ɒ]. (Incidentally, Willson (2003:7) implies that this diphthonization may not happen next to long vowels, and she goes as far as to suggest that long vowels may need to be specified for backness!)
As for the approximant issue -- this is an interesting observation, but anything which is a non-obvious "inference" probably qualifies as OR. (Also, it's worth noting that [ɑ̯] and [ʕ̞] are actually quite different sounds, and [w] and [u̯] may also differ phonetically from each other.)
Also, the article should probably use [mʲ] throughout rather than [m̟], since this is what is used in the sources... Mo-Al (talk) 17:03, 30 December 2012 (UTC)
Check my talk page. Someone is challenging the two-tiered narrow/broad transcriptions, adding yet another layer to my headache...if I had to choose between the two, I'd choose the narrow transcription because that's what you actually hear — but I don't really have a good idea how to simplify it to make it look as reader-friendly as Spanish IPA. ...okay, I read the page 7 reference. I think I actually understand what it's talking about, because I do recall the jouj in kōn jouj sounding more like [tɕou̯tɕ] or perhaps [tɕe̯ou̯i̯tɕ] (a common historic alternate spelling is joij). But I do know that [mæ] [mæ͡ɑ] [mæ͡ɒ] (or perhaps [mæ] [mæ̯ɑ] [mæ̯ɒ]) really do occur, every time I heard phrases like enana, which is {yenahnah} /jɜnʲaɰnʲaɰ/ [ɛ.næ͡ɑ.næ͡ɑ], heard here. - Gilgamesh (talk) 02:24, 31 December 2012 (UTC)
Okay, I have another way of illustrating this. Spelling (even within the same orthography) is highly variable, right? But the pronunciations of the spelling variants remain the same. otemjej/wōtōmjej, Kuwajleen/Kuajleen, Ujae/Wūjae, io̧kwe/iakwe, Jijej/Jisōs, Kūraij/Kraist, ikkure/iukkure, ekkwaļ/eokkwaļ, jouj/joij, etc. The MED even has a page that discusses common patterns of spelling variation for people trying to look up words in the dictionary. Considering the sounds of Marshallese being what they are, this is because even the official orthography provides more than one way to spell the language's highly abundant diphthongs.
As for the long vowels...to be honest I've never been 100% sure how to transcribe those, other than transcribe the two short vowels twice and leaving it to underlying mechanics. I don't have the clearest idea how to distinctly transcribe MED sequences like {kak} vs. {kahk} vs. {kahak} vs. {kahhak} in semi-narrow IPA without the use of approximant or semivowel symbols. The MED system is easy — they provide it, but it's not IPA. Broad transcription is almost as easy, since it translates well from MED's system.
But relatively narrow ear-friendly transcription has never been easy to do without some risk of accusation of original research. We hear the words, and intuitively understand how they're pronounced, but I've been accused of making them too broad under the analogy that Spanish IPA has simple vowels and simple consonants and no diacritics, etc. Marshallese is not Spanish...and phonologically eccentric languages like Marshallese stretch the eurocentric IPA to its limits; I imagine linguists have had a similarly difficult time with languages like Ubykh. - Gilgamesh (talk) 06:25, 31 December 2012 (UTC)
I agree that Marshallese is a complex situation, and the data that you're bringing up are indeed quite interesting. However, the sources *do* give IPA, and I don't think we have the mandate to modify what they give. Ultimately, we are writing a summary of the published research on Marshallese, so we have to use what is written in the sources. For example, our source gives "[mæ] [mɑ] [mɒ]" -- I don't think we have the ability to modify this without another source that *explicitly* states that we should. Mo-Al (talk) 07:07, 31 December 2012 (UTC)
Also, by the way, I think templates like IPAc1-mh and IPAc2-mh are not normally used in linguistics articles. I think they may be normally used just for clarifying the pronunciation of article titles. For example, such a template is used in the first sentence of the article Brașov, but not in the Romanian phonology article. Correct me if I'm wrong, but I'm not 100% sure that this is the right place for this template. Mo-Al (talk) 07:13, 31 December 2012 (UTC)
Update: I've found a rather detailed paper (Choi 1992) which has some nice acoustic analyses of Marshallese vowels. I'll be adding the information from this article, which hopefully will clear some of this up. (He mentions, for example, that vowel-initial words are sometimes, but not always, pronounced with an onglide.) Mo-Al (talk) 07:26, 31 December 2012 (UTC)
We still know that [mæ] [mɑ] [mɒ] are oversimplifications, and [mɑ] [mɒ] especially are (on their face) ambiguous about which of the two [m] phonemes are being used (though the dictionary shows the /mʲ/ is being used). That particular source illustrated a specific point, but is problematic in other ways. And if you listen to a lot of Marshallese, you hear the diphthongs at work. I am not at all comfortable not using the diphthongs, especially when the same words can be spelled multiple different ways (with different vowels) without at all changing the pronunciation. Em̧m̧an {yeṃṃan} /jɜmˠmˠanʲ/ is not [ɛmˠmˠɑnʲ]; it's actually [ɛ͡ʌmˠmˠɑ͡ænʲ] (I know at least four YouTube clips that corroborate this pronunciation[2][3][4][5]), whether it's spelt em̧m̧an or eōm̧m̧wān or anywhere inbetween. It's neither one vowel nor the other — it's both at once, in a diphthong, because vowel backness and roundedness are simply not phonemic in Marshallese and are completely at the mercy of the consonants they neighbor. Marshallese orthography has just never been able to indicate this with one-grapheme-to-one-phoneme exactitude, because the Indo-European-speaking missionaries found it easier to think in eurocentric phonetic paradigms that have no one-to-one correspondence with Marshallese phonology. - Gilgamesh (talk) 07:40, 31 December 2012 (UTC)
(Edit conflict.) Show me the paper — I'd like to review it too. - Gilgamesh (talk) 07:40, 31 December 2012 (UTC)
I've added a link to the paper in the bibliography section. Mo-Al (talk) 07:43, 31 December 2012 (UTC)

Okay, been reading the paper. So far, it supports the existence of diphthongs in short vowels, including in words that begin or end with approximants. It also says (here on page 24) that initial and final approximants are sometimes realized as an off-glide, but that this does not consistently occur, and that it has the same height as the vowel it neighbors (if that's what "with narrowing at one of the three secondary restriction sites" is meant to mean — please tell me if you think this is incorrect). I've been skimming much of the paper, looking for relevant references, and am currently continuing from page 24. - Gilgamesh (talk) 08:13, 31 December 2012 (UTC)

Technically, this "narrowing" probably means that the glide is higher than the surrounding vowels... Mo-Al (talk) 08:18, 31 December 2012 (UTC)
Oh... Well, I'm still reading. And I noticed in page 67 that a long vowel with /tʲaɰatʲ/ seems to have a U-bow-shaped curve between the three secondary points of articulation, rather than a V-shaped line. Also, the way it describes long diphthongs being really just two short vowels...sounds more like the mechanics of mora-timing. - Gilgamesh (talk) 08:27, 31 December 2012 (UTC)

So, since I've been challenged to simplify the "too narrow" transcription, I have an idea.

  • For the sake of transcription, assume that /ɜ/ and /ɘ/ really are distinct vowels (even if they may not actually be), and go by the MED phonetic transcriptions where it uses {e} (lower) or {ẹ} (higher). (As a side note, we should probably put Bender and Choi's vowel inventory analysis in the article as a third opinion.)
  • Do not mark approximants at all. Do not mark secondary articulation of consonants, except for /tʲ/ being [tɕ ~ ʑ]. Let the vowel allophones be all the secondary articulation they need to be. Mark consonants are voiceless at the beginnings, ends and when geminated, and voiced elsewhere.
  • For the short vowels, use the 12 monophthong allophones and 24 diphthong allophones as normal. Probably no need for tie markers, as long as we're marking ambiguous syllable boundaries with [.].
  • For the long vowels and long diphthongs, first take the two pairs of diphthong vowels; let's call them V1A V1B V2A V2B. As usual, if C1 and G have the same secondary articulation, omit V1A, and if G and C2 have the same secondary articulation, omit V2B. If V1 and V2 have the same vowel height, mark them as a long vowel with [ː]. Otherwise, just string the whole vowel symbols together as a syllable.
  • For {CG} clusters (where G is an approximant), use [.] syllable break in place of the approximant.
  • For {GC} clusters, use [ˑ] to mark the slightly metrically longer vowel+approximant.
  • Mark epenthetic vowels only as [ᵊ]. Its actual articulation is no more phonemic than its existence.
  • Examples:
    • aelōn̄ {hayẹlẹg} /ɰajɘlʲɘŋ/ = [ɑæeleɤŋ].
    • Bokaak {bekʷhak} /pˠɜkʷɰak/ = [pʌɔɡ.ɑk].
    • em̧m̧an {yeṃṃan} /jɜmˠmˠanʲ/ = [ɛʌmːɑæn].
    • enana {yenahnah} /jɜnʲaɰnʲaɰ/ = [ɛnæɑˑnæɑ].
    • Epjā {yepjay} /jɜpʲtʲaj/ = [ɛbᵊʑæ].
    • Ibae {yibahyey} /jɨpˠaɰjɜj/ = [iɯbaˑɛ].
    • io̧kwe / iakwe / yokwe {yi'yakʷey} /jɨ̯jakʷɜj/ = [iæɒɡɔɛ].
    • Jālooj {jalẹwẹj} /tʲalʲɘwɘtʲ/ = [tɕæleoːetɕ].
    • jouj / joij {jẹwij} /tʲɘwɨtʲ/ = [tɕeouitɕ].
    • Kuwajleen / Kuajleen {kʷiwajleyen} /kʷɨwatʲlʲɜjɜnʲ/ = [kuɒæʑᵊleːn].
    • M̧ajeļ / M̧ajōļ {ṃahjeḷ} /mˠaɰtʲɜlˠ/ = [mɑˑʑɛʌl].
    • Mājro {majrẹw} /mʲatʲrˠɘw/ = [mæʑᵊrɤo].
    • Pikinni {pikinniy} /pʲɨkɨnʲnʲɨj/ = [piɯɡɯinːi].
    • tata {tahtah} /tˠaɰtˠaɰ/ = [tɑˑdɑ]. (Note that *{tatah} */tˠatˠaɰ/ would be *[tɑdɑ] without [ˑ], and that *{tahatah} */tˠaɰatˠaɰ/ would be *[tɑːdɑ] with [ː].)
    • Wōjjā {wejjay} /wɜtʲtʲaj/ = [ɔɛtɕːæ].
    • wōtōmjej / otemjej {wẹtẹmjẹj} /wɘtˠɘmʲtʲɘtʲ/ = [oɤdɤemᵊʑetɕ].

- Gilgamesh (talk) 11:52, 31 December 2012 (UTC)

I think this is much improved -- thank you for taking the time to go over this! I do have a couple more comments:
  • It might make more sense to write the phonetic realization of /tʲ/ as [tʲ], since according to Choi this is one of its allophones (in free variation with the others).
  • Correct me if I'm wrong, but my understanding was that the secondary articulations of consonants really are present on the phonetic level, and are *not* just abstractions to explain the vowel system (unlike, for example, some of the glides). If this is so, then shouldn't we keep the secondary articulations in the phonetic transcriptions?
  • Is there really any acoustic difference between things like [pʌɔɡ.ɑk] and [pʌɔɡɑk]? If not, the syllable divider shouldn't be present next to consonants in the phonetic transcriptions.
  • Could there really be a surface contrast between things like [tɑdɑ], [tɑˑdɑ], and [tɑːdɑ]? This would be pretty remarkable, since three-way length contrasts are quite rare, and would probably require a source. (And if so, it should definitely be mentioned in this article!)
Also, would it be okay to remove the IPAc1-mh and IPAc2-mh templates from this article? I think the style convention on Wikipedia is to only use these types of templates in the first sentences of articles to give the pronunciation of names (like in the article Kofi Annan). Mo-Al (talk) 01:37, 1 January 2013 (UTC)
Okay. It's New Year's Eve and doing some things, but I think I can comment briefly on this for now.
Yes, we can use the secondary articulations. I just thought omitting them (and letting the vowels they neighbor tell the story) would simplify the narrow transcription. I suppose using both secondary articulations and vowel articulations just keeps things clear.
  • Okay, [tʲ] works. But still using [ʑ] between vowels.
  • MED treats the approximant boundaries as distinctive, so all the phonemes in {bekʷhak} are distinctive. [pʌɔɡɑk] is an ambiguous transcription, because it could be {bekʷhak} or {bewkak}. Same with {tatah}, {tahtah}, {tahatah} distinction — if MED recognizes such distinctions as meaningful, we should do the same in transcription. With [ˑ] in {GC} and [.] in {CG}, we can do this minimally without making too many additional assumptions. Especially considering that Bokaak is also spelt Bok-ak (connecting the two morphemes bok and ak with a dash instead of as a compound), and this is a somewhat special spelling rule where the "a" is doubled to mark the morpheme boundary.
  • Yes, let's replace IPAc1-mh etc. in the article. Just please remember to use the {{IPA|}} template for IPA and {{lang|mh|}} template for Marshallese snippets — it assists style sheets (including custom user style sheets) who display the class="IPA" and lang="mh" directives in a special way. - Gilgamesh (talk) 04:51, 1 January 2013 (UTC)
  • Would it be possible to use [dʲ] between vowels? I'm not clear on why the manner of articulation would change...
  • I agree that the MED distinguishes between {beqhak} and {bewkak}, and between {tatah}, {tahtah}, and {tahatah}. The question is whether these are all distinguished *acoustically*. Do you think that Marshallese has distinct short, semi-long, and long vowels in pronunciation? This is quite rare in the world's languages, and if this were true I find it surprising that the sources don't seem to comment on it. For that matter, what does a syllable divider even mean pronunciation-wise? Mo-Al (talk) 05:45, 1 January 2013 (UTC)
  • [dʲ] is a possible realization, and I have heard it before. It may be that [ʑ] is a distinction more common in specific dialects or among younger generations, or something like that. [dʲ] could be conservatively used, I guess.
  • Acoustic difference? Possibly, or possibly not — Marshallese has never been particularly easy for non-Marshallese to study in this regard, even when you hear it spoken it every day. Perhaps [ˑ] may not be the wisest or least-OR option for {GC} boundaries. But I see unacceptable oversimplification issues here in arbitrarily disregarding MED distinctions. So how do we distinguish it? My earlier approach was to use semivowels, like with [pˠʌɔɡʷɑ̯ɑk] and [tˠɑɑ̯dˠɑ], to mark the boundary. Spellings like Bokaak for {beqhak} seem to reflect a consciousness of this. Maybe we could do this, though only where there would otherwise be a phonemic ambiguity. - Gilgamesh (talk) 06:38, 1 January 2013 (UTC)

Unfortunately, I don't think using semivowels like [ɑ̯] is so feasible. The thing is, a sound sequence like [ɑ̯ɑ] should be phonetically equivalent to [ɑ] or [ɑˑ]. When you chose to write this as [ɑ̯ɑ] rather than [ɑɑ̯] or [ɑ] or [ɑˑ], you're adding some lower-level phonemic or morphophonemic structure which isn't present on the phonetic level. The question should just be: How is this word physically pronounced?

I don't know much about Marshallese grammar, but I have to wonder whether the MED is using morphological criteria to determine the form of the word. Perhaps beqhak and beqak or beqahak really are pronounced exactly the same, but because beqhak is composed of two morphemes beq+hak the MED assumes that there is no intervening vowel... Mo-Al (talk) 06:50, 1 January 2013 (UTC)

I never learned enough conversational Marshallese either, when I had the chance; I regret that today; I also always only lived in the American expatriate community and never in the Marshallese community, which raises a cultural barrier that was never easy to thin after I moved away from the islands. I care a great deal about the Marshallese language, though I cannot speak conversationally and it is not easily learnt online or frequently used online. Anyway, what I understand is that the tongue/lip positions are closely linked to the secondary articulation, both in the vowels and in the consonants. {beqak} would have a slope in {qak}, but the mouth would have already changed positions in {beqhak} during the {qh} cluster before {a} was reached. So {beq} has a slope, and {qak} has a slope, but {hak} has no slope from beginning to end. Therefore, the position changes already between {q} and {h}, and {h} has no labialization. Now, thinking about this, I suppose that a transcription of [pˠʌɔɡʷɑk] could suffice without any additional need for disambiguation. But there's still an issue with {CG} and {GC} clusters that have the same secondary articulation, such as in {tahtah}, because [tˠɑdˠɑ] is superficially ambiguous that way. Could this be one of those situations where we do use [ɰ] as a disambiguator? I understand the complications, but as someone who myself is not fluent, I am very uncomfortable deviating from the MED's specifications. I wish there were more educated fluent Marshallese speakers on the internet to assist us — but the Marshall Islands is poor country, and most people are not online. Come to think of it, I have before emailed the MOD webmaster with questions, when I wanted to know Marshallese forms of the names of various Kwajalein atoll islets I remembered from my youth; they conveyed the answers back after a time, but it involved passing the questions along to people actually living in the islands, getting the answer, and then passing it back to me — that's how I obtained some of the additional islet names in the Kwajalein Atoll article. I wonder if they can also be asked questions about the acoustic distinctiveness of the MED system. I have no way of knowing just how well the webmasters of the online MOD understand the fully meaning of the original MED they adapted. - Gilgamesh (talk) 07:25, 1 January 2013 (UTC)
Interesting. Well, I don't think you need to worry about deviating from the "MED's specifications" -- the MED doesn't give a *phonetic* transcription at all! I would say that it's best to stick with what we know. We know the consonant quality and vowel quality in words like [tˠɑdˠɑ]. The only thing we don't know is whether there is some extra length. [tˠɑdˠɑ] isn't technically incorrect even if there is some semi-long vowel (its length would just be unspecified in the transcription), so [tˠɑdˠɑ] should be sufficient. Mo-Al (talk) 07:53, 1 January 2013 (UTC)
Well, I think I'll keep the two-tiered system for Wiktionary Marshallese articles, but it seems clear now that Wikipedia only needs one pronunciation per Marshallese name in articles. Simplifying the template system will surely take a redesign and overhaul. Also, I've been asked that Wikipedia:IPA for Marshallese be rewritten to be easier to understand. That won't be easy as long as Marshallese orthography itself does not perfectly reflect its phonology, but just a rough eurocentric impression of it. I'm not sure how to go about rewriting it. - Gilgamesh (talk) 11:45, 1 January 2013 (UTC)
Makes sense. Well, anything involving Marshallese phonology is guaranteed to be challenging... I may try making some edits to the IPA page (per the things that we agree upon), though my main interest is this article. Mo-Al (talk) 01:34, 2 January 2013 (UTC)
Okay, thanks for your help. - Gilgamesh (talk) 04:12, 2 January 2013 (UTC)
My interest, as elaborated in my post on Gilgamesh's talk page, is to simplify the presentation of Marshallese transcriptions at Wikipedia articles for the sake of users, most of which is accomplished by having just one transcription.
I don't think 12 monophthongs and 24 diphthongs are too much (that's only slightly more than what English has).
I hesitate to accept [ᵊ] for an epenthetic schwa if it's just a matter of it being phonemic or not; if there is enough variation in whether it's pronounced, this is more justified, though a parenthetical schwa may instead be in order, given that superscript letters often indicate a modified pronunciation in the IPA.
I'm fine with whichever representation of consonants as long as it's clear. Do we want to utilize the velarization marker? It seems kind of redundant to me. Russian has a similar contrast between palatalized and velarized sounds and only palatalization is marked, though Irish marks both palatalization and velarization. — Ƶ§œš¹ [ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ] 14:40, 4 January 2013 (UTC)
The epenthetic vowel is not necessarily a schwa, but its actual height articulation is not distinctive, which is why I suggested the schwa (though the vowel still undergoes secondary articulation coloring, and its vowel height is predictably intermediate to the two nearest vowels). But in the circumstances where the epenthetic vowel exists, it is always pronounced. This expectation even caused hypercorrective vowel deletion in loanwords like Kūrjin ("Christian"). Kūrjin and *Kūrijin are actually homophonous.
Marshallese is actually more similar to Irish in that palatalization, velarization and labialization are equally critically distinct. And Marshallese orthography saw fit to give diacritics to "heavy" velarized and labialized consonants (rather like Arabic emphatic consonants), and not to "light" palatalized consonants. One of my previous approaches (which could easily be revisited), was to keep palatalized consonants unmarked (except for /tʲ, rʲ/ being [tɕ~ʑ, r̟]), and marking velarized and labialized (actually labiovelarized) consonants with the infixed tilde diacritics ([p̴, t̴], etc.), and including either [ʷ] or the diacritic [k̹] for consonant rounding. But really, none of the consonant phonemes can be considered "clear" — they are all secondarily articulated in one of the three ways, and one secondary articulation is no weaker than another. - Gilgamesh (talk) 21:18, 4 January 2013 (UTC)
Let me throw in my two cents:
It's misleading to use [ə] for the epenthetic vowel if it can really be high or low. In a phonetic transcription, [ə] refers to a vowel which is pronounced as a central vowel. Since this is a phonetic transcription, you must write what is actually produced. I think we should find a published source to add to this article on epenthetic vowel quality, and then reflect what it says in the transcriptions.
I don't see a reason to treat the secondary articulations in Marshallese any differently than in Irish.
Use [ʷ ʲ ˠ], not things like [k̹], [r̟], etc. There are good reasons for this which I think are pretty self-evident. Mo-Al (talk) 01:33, 6 January 2013 (UTC)
I agree on all counts. — Ƶ§œš¹ [ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ] 04:30, 6 January 2013 (UTC)
I nixed the schwa; back to articulated epenthesis, since the height and contour are predictable even though they aren't phonemic then. And ʲ ˠ] is how things are being done now, so...seems like things are alright, now? At least, pending a better understanding of whether (and how) there's an accoustic contrast to MED sequences like {ata}, {ahta} and {atha}; currently all three are [ɑdˠɑ]. I hope someday we can find a clarification on that. - Gilgamesh (talk) 04:53, 6 January 2013 (UTC)

I just realized an IPA ambiguity issue which means we may need to reinstitute vowel tie bars. Right now, both *{wey} and *{wewyey} are transcribed *[ɔɛ]. But whereas *{wey} implies a smooth transitioning *[ɔ͡ɛ] sequence, *{wewyey} is *[ɔ-ɛ] with no actual diphthonging. And *[ɔ-ɛ] is not intrinsically two-syllable *[ɔ.ɛ] either — I know from growing up hearing Marshallese people speak among themselves in Marshallese, that they say Ibae {yibahyey} with the acoustic appearance of two syllables instead of three. If this phonemic information is correct, it means Marshallese can form syllables of multiple vowels with or without diphthong contours; so, in [i͡ɯbˠɑɛ], [i͡ɯ] has a smooth contour from [i] to [ɯ], while [ɑɛ] has an abrupt transition between [ɑ‿ɛ] with no contour.
In my experience, such a difference is noticeable even in American English, because eye as [a‿i] (still one syllable) tends to sound like a non-native accent, because native speakers tend to contour the diphthong as [a͡i]. This difference is not phonemic in English (which is why English IPA tends not to require tie bars), but apparently it is phonemic in Marshallese.
How do you think we should handle this? Is it alright if restore tie bars where the smooth transitions occur? And how to handle long vowels like in naaj? Is [nʲæ͡ɑː͡ætʲ] okay? And if so, should the long diphthong word jouj be [tʲe͡o‿u͡itʲ] using the IPA undertie diacritic, or just plain [tʲe͡ou͡itʲ] to keep it simple? - Gilgamesh (talk) 11:19, 9 January 2013 (UTC)

A transition from one vowel quality to another in the same syllable is the definition of a diphthong, so I'm not sure what the phonetic distinction is between an "abrupt transition" and a "contour" one. — Ƶ§œš¹ [ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ] 13:27, 9 January 2013 (UTC)
Gilgamesh -- I don't think it's phonetically possible to have a diphthong without a phonetic contour. When your mouth changes from one vowel position to another, it has to take all intermediate values at some point. Remember that we can't use any information unless there is a reliable, published source. Mo-Al (talk) 13:44, 9 January 2013 (UTC)
I'm talking about the shape of the contour. In the allophonic diphthongs, it's a steady ╱-shaped glide from one position to the other. But the MED uses clusters of {CG}, {GC} and {GG} (where {G} is an approximant) where the transition is presumably more abrupt. I know it's not physically possible to have no contour — I just mean that it's acoustically jerkier. Like a _/¯-shaped contour instead of a ╱-shaped contour. Maybe there's something in Choi's paper about this; I mostly skimmed it before, but this is the kind of acoustic curiosity Choi addressed, and the answer may be in there somewhere. It seems hard to believe that there would be no real acoustic distinction between {wey} and {wewyey} when there is at least a length distinction between {a} and {aha}. - Gilgamesh (talk) 19:18, 9 January 2013 (UTC)
Actually, I have another thought — one that is demonstrable from what we know here. We know there's a difference between short diphthongs the length of one vowel, and long diphthongs the length of long vowels the length of two vowels, right? So, completely agnostic of contours, Ibae {yibahyey} is metrically closer to [ĭɯ̆bˠɑɛ]. And naaj is more like [nʲæ̆ɑ̆ɑ̆æ̆tʲ] or [nʲæ̆ɑæ̆tʲ].
These are quite subtle issues. It's an interesting topic, and if you can find an explicit citation of a fact somewhere that's great, but otherwise it's just speculation. Mo-Al (talk)
My OCD side is demanding I put tie diacritics at least on the short vowel allophones, to indicate they are a single phonemic height unit as opposed to two phonemic vowels in sequence. - Gilgamesh (talk) 06:17, 6 February 2013 (UTC)
OCD or not, it seems from the above comments that this is an example of OR, both in representation and in the distinction being offered. I can get behind doing the former, but not the latter. — Ƶ§œš¹ [ãːɱ ˈfɹ̠ˤʷɪ̃ə̃nlɪ] 13:10, 6 February 2013 (UTC)
Actually, that's what I did. I used the tie diacritic only on the one-vowel height allophones, and not anywhere else. These do have a smooth glide, per Choi and Willson. Is that acceptable? Not using the tie between two adjacent vowel phonemes [VV] does not necessarily represent break, but rather non-assumption, between the actual possibilities of syllable break [V.V], no syllable break [V‿V] and actual contour [V͡V]. Assuming either one is OR, while assuming neither just demonstrates what we can back up with reliable published sources. - Gilgamesh (talk) 02:52, 7 February 2013 (UTC)

Bibliography note

I inherited my dearly-departed mother's collection of Micronesia-related books. And when I examined it, I found two Marshallese linguistics texts: Spoken Marshallese by Bender (1969), and the print version of the MED. She had them all these years and I didn't realize it. I added Bender (1969) to the bibliography, and added some new material. - Gilgamesh (talk) 06:00, 12 February 2013 (UTC)

Time duration of phonemes

It was staring me in the face the entire time... Per Choi, each vowel unit and each consonant unit occupies the same duration of time. I looked at this again when I realized that page 65 was saying that long vowels of /CVGVC/ have three times the length factor of short vowels of /CVC/. And sure enough, all the vowel contour diagrams do mark both consonants and vowels as durations of time. So whether or not the approximants are distinctly articulated as consonants that interrupt two vowels, they do contribute to the time duration. This means that a sequence like MED *{hahah} actually is *[ɑ̯ɑɑ̯ɑɑ̯] on a phonemic level (*/ɰaɰaɰ/ on a morphophonemic level), but it sounds like one continuous long vowel *[ɑː]. Choi states in page 71 that although there may not be an articulation of an approximant as a consonant that would audibly separate two vowels, Marshallese speakers can still intuit long vowels as two syllables (or two morae), and the glides probably exist at deeper level of representation. So, since final consonants (including the approximants) occupy time and also have prosodic weight (Willson section 5), I suggest reviving my idea of either:

  1. Using [i̯ ɛ̯ æ̯ ɯ̯ ɤ̯ ʌ̯ ɑ̯ ɔ̯ ɒ̯] semivowels with the understanding that they bleed into their adjacent vowels and contribute to length and are perceptible consonant anchors without necessarily forming glides.
  2. Marking vowels progressively longer based on how many units they occupy, with */CˠaɰC/ being *[CˠɑˑC], */CˠaɰaCˠ/ being *[CˠɑːCˠ], and */Cˠaɰɰaɰ/ being *[Cˠɑˑ.ɑˑ].

I favor the first option for phonemic accuracy. And of course, citing the hard reliable references. - Gilgamesh (talk) 09:08, 1 March 2013 (UTC)

Years later, this topic continues to not sit well with me, for the reason that even though /ɰ/ is a consonant and it occupies a mora in the final position, it still does not actually surface as an approximant—no real [ɰ], [ʕ̞], etc. In consideration of this, phonetic transcriptions like [ɑ̯] always felt like an awkward kludge that functions more as a phonemic and moraic placeholder than anything else. So, without straying too far in the realm of original research and idiosyncratic transcription, how best can we transcribe sounds that distinguish combinations like */CˠaCˠ/, */CˠaɰCˠ/ and */CˠaɰaCˠ/? The reliable sources say that speakers can distinguish their prosodic length. So if */CˠaɰCˠ/ is perceptibly one mora longer than */CˠaCˠ/, what is the best way to phonetically transcribe */CˠaɰCˠ/, and what distinguishes it from */CˠaɰaCˠ/ which have the same prosodic length while seemingly only */CˠaɰaCˠ/ is a long vowel? It would be so much easier to be told that */CˠaɰCˠ/ and */CˠaɰaCˠ/ are actually pronounced identically, but we can't exactly infer that from the available reliable sources, as they concede that the phonemes may exist on a (still relatively unstudied) deeper level. How does one phonetically transcribe both */CˠaCˠ/ and */CˠaɰCˠ/ as [CˠɑCˠ] when they have different prosidic lengths, and how does one phonetically transcribe */CˠaɰCˠ/ and */CˠaɰaCˠ/ separately as [CˠɑCˠ] and [CˠɑːCˠ] respectively when they have the same prosodic length, all without any actual surface presence of an explicit [ɑ̯] to distinguish all three? This is an enigma. - Gilgamesh (talk) 07:38, 24 September 2019 (UTC)

After giving it some thought, there are some things we can do to simplify the phonetic transcriptions and be entirely within the sources. First of all, we can axe all the [i̯ ɛ̯ æ̯ ɯ̯ ɤ̯ ʌ̯ ɑ̯ ɔ̯ ɒ̯] at the beginnings and ends of words, except in situations like where a strongly present [i̯] represents /jɨ̯j/ rather than /j/. Word-internally, we already indicate long vowels or long diphthongs where an approximant occurs between two vowels. The tricky part is where an approximant directly neighbors another consonant or another approximant. The best I can think of at the moment is to insert a syllable break [.] to indicate the presence of a heavy prosodic break, even though technically it would be just as true to separate all syllables with [.]. If only there were a symbol for heavy prosody that didn't infer things we can't safely infer. - Gilgamesh (talk) 23:06, 24 September 2019 (UTC)

Display issues

The section on Display Issues is thorough and interesting, but misses the nature of the problem regarding the cedilla under the L and N.

The Unicode standard defines the combination L + combining-cedilla and N + combining-cedilla as being canonically equivalent to the Latvian forms Ļ and Ņ [1]. (That's what the equivalent sign "≡" means in the document). So the font rendering engine pre-composes these combinations, before any rules in the font are applied. As was pointed out, a zero-width non-joiner character ( &zwnj; in HTML 4 ) betweeen the letter and accent will prevent this pre-composition. This should work on any modern operating system, and will look good with any reasonably complete modern font.

A font could provide rules, which de-compose the pre-composed letters. It could do this specially if the text is marked as Marshallese script. I doubt that many fonts (or, any of the fonts listed in the table) do this.

If you see a cedilla appearing beneath one of these letters in the table, it probably means the font is missing something, probably the combining cedilla character. In that case, the font rendering software pulls the glyph from another font, but doesn't pre-combine the characters. So you might get a cedilla glyph roughly positoned under the letter glyph, but the combination will likely look bad, because the letter and the mark are from different fonts.

Only one of the font families listed on the table was installed on my operating system (Linux). But even if all the fonts were installed, the results can only be misleading, as explained in the above paragraph. So I'm afraid the table of fonts is of little use.

Stevan White (talk) 10:23, 5 June 2016 (UTC)

References

In consideration of your input, I have expanded the display issues table to include the letters composed with zero-width non-joiners. I think this is good as a possible display workaround if available for desktop publishing and display graphics, but it's still effectively a kludge for something that is, for ordinary electronic text, better addressed by designing fonts with better extended language support. The precomposed letters are still specified by the Unicode specification as having cedillas rather than commas. An extended language support can display letters with commas for Latvian, but true cedillas for Marshallese. - Gilgamesh (talk) 00:32, 8 October 2019 (UTC)

Hello fellow Wikipedians,

I have just modified one external link on Marshallese language. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018. After February 2018, "External links modified" talk page sections are no longer generated or monitored by InternetArchiveBot. No special action is required regarding these talk page notices, other than regular verification using the archive tool instructions below. Editors have permission to delete these "External links modified" talk page sections if they want to de-clutter talk pages, but see the RfC before doing mass systematic removals. This message is updated dynamically through the template {{source check}} (last update: 5 June 2024).

  • If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
  • If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 02:23, 19 January 2018 (UTC)

Further simplification of phonetic IPA transcription

I recently took steps to simplify the phonetic IPA transcription of Marshallese on Wikipedia while sticking to accredited sources. But after reading various YouTube comments on videos about learning Marshallese, it has become clear that this transcription, which people look up on Wikipedia to study, is still difficult to read. Some ideas I have:

  • No more voiceless vs. voiced consonants in transcription. While it is a certain common habit to to pronounce consonants voiceless at the beginnings and ends of phrases and when doubled, and voiced within phrases, these are not fixed rules, and there is considerable variation in how these are articulated, often varying by gender and between speech and song. I'm thinking, transcribe them all as voiceless with the understanding that any or all of them could be voiced.
  • No more [tʲ] vs. [zʲ], not just as a distinction of voice, but as a distinction of plosive vs. fricative—they're all obstruents, underspecified for any more specific oral constriction. I'm thinking transcribe all of them as [tʲ].
  • The locations of epenthetical vowels are predictable, but their vowel heights are less predictable. While one source implied that the vowel's height is transitional between the two nearest neighboring vowels, at least one of the epenthesis examples contradict this (waj ne- in Willson (2003), section 4.3, "Excrescence"), appearing higher than either neighboring vowel. This could be a copyediting error, but still... My idea, for all epenthetical vowels, is simply to transcribe them as [(ə)], as it's underspecified for anything other than habitually existing.
  • Choi (1992) in particular seems to imply on pages 71-73 that long vowels typically have a different behavior than short vowels. Short vowels are characterized by smooth transition between the secondary articulations of neighboring consonants, while long vowels and long diphthongs seem to spend a greater amount of time on the shared vowel nucleus and much less time on the transitions. He mentions this in context to the word jaaj appearing to be an exception to this rule. (If I mistook any of these things, please correct me.) So my idea is, instead of using phonetic transcriptions like [mʲe͡ou͡ɯrˠ] for mour (which is phonemically /mʲɘwɨrˠ/), just phonetically transcribe it for the vowel nucleus: [mʲourˠ]. An exception can be made if the consonants on either side of the long vowel or long diphthong are more glides, so aelōn̄, phonemically /ɰajɘlʲɘŋ/, and currently phonetically transcribed [ɑ͡æelʲe͡ɤŋ], could still be transcribed in this fashion.
  • As for my previously suggested approach of using an explicit [.] to mark an extra mora boundary after a vowel and before another consonant (where an underlying glide phoneme causes it to occupy that extra mora), I realized later that there is a very simple way of representing that extra duration—the indication of appropriately extra length. The long vowel symbol [ː] is already used for long monophthongs, so I'm thinking [ˑ] for the extra mora boundary. So Ibae, which is phonemically /jɨpˠaɰjɜj/, and is currently phonetically transcribed [i͡ɯbˠɑ.ɛ], could instead be transcribed [i͡ɯpˠɑˑɛ]. I seem to recall trying this approach before, but was criticized for it looking like original research. If this remains the case, we could stick with the [.] kludge.
  • I am very tempted to follow Choi's (1992) lead and treat modern Marshallese as having only three vowel phonemes rather than the traditional four, and using his phonetic transcription system for vowels in articles. It would also seem to correlate with the modern orthography's use of only three vowel heights in spelling (collapsing the middle two into one), as well as my own personal memories of hearing the spoken language in my youth (as meaningless as this admittedly is as a reference). Problem is, as valuable as Choi's data is, considering there are at least two Marshallese dialects, it may not be a safe assumption to consider the four-vowel model obsolete. I also sense echoes in this topic of a similar debate over Wikipedia's transcription of vowels in the English language, and over subtle distinctions like /ɔːr/ vs. /ɔər/ and /ə/ vs. /ɵ~o/ which I'm generally in favor of keeping distinct.

- Gilgamesh (talk) 03:48, 7 October 2019 (UTC)

Withdrawing my epenthesis suggestion, as I noticed some other inconsistencies in the example that appear to be copyediting errors, casting doubt on that one example as a reference. - Gilgamesh (talk) 00:04, 8 October 2019 (UTC)
After reading threads posted years ago, it occurs to me that some of this section is rehashing things discussed back then. I think I should reconsider all my suggestions, especially with everything I've learnt in the years since then. - Gilgamesh (talk) 06:39, 9 October 2019 (UTC)