Talk:Marshallese language/Archives/2019/November

Mora value of coda glides

Currently the Phonotactics section states that glides take up one mora. Assuming that short vowels have one mora and long vowels have two, a vowel plus coda glide should always yield a long vowel. But currently the module yields [mˠɑːzʲɛ͡ʌlˠ, mˠɑ.tʲɛ͡ʌlˠ] for {m‌̧ahjel‌̧}, in which the second version, with the short [ɑ], is the enunciated version. So enunciation is deleting a mora. Then again, I think the module has always transcribed word-final vowels, which in the phonemic representation are followed by a coda glide, as short, as in {habah} [ɑbˠɑ], which should be three morae but looks like two (two short vowels). So the mora deletion has always been there in that case. — Eru·tuon 21:48, 9 November 2019 (UTC)

Ahh, so you noticed my new cluster isolating mode, for enunciations. And yes, the final mora of utterances does seem deleted, but I have a (quite OR) theory it's more of a pausa situation: Just like how final obstruents are articulated yet not fully released, so are final glides. For the enunciated forms, imagine each hard syllable break not just as a syllable break, but an actual pause in speaking. As...if...I'm...speak...ing...clear...ly. In truth, it's not possible for a fully qualified Marshallese word to be fewer than two morae, though affixes can be: {ri-}, {-in}, because they attach directly to consonant phonemes of other words. Go figure. - Gilgamesh (talk) 22:16, 9 November 2019 (UTC)

The idea is...if a heavy syllable break neighbors a vowel (where it's an underlying consonant cluster involving one or more glides), it's an extra mora. I'd use an even clearer symbol, but I don't know of anything better than [.]. Can you think of one that fits this purpose and is clear to read? - Gilgamesh (talk) 22:19, 9 November 2019 (UTC)

Ultimately, the isolating mode may be unnecessary for pronunciation guides. Dictionary overkill. Was just trying something. I can easily deactivate its functionality and display only the uninterrupted speech pronunciation mode. - Gilgamesh (talk) 22:21, 9 November 2019 (UTC)

Oh, just realized it's you and not Austronesier. Yes, well... Do you think it's overkill for Wiktionary? I mean, it's largely for the benefit of words like Jālwōj {jalwȩj}, which expands epenthetically to {jalȩwȩj} in uninterrupted speech and phonetically surfaces as [tʲælʲoːtʲ]. I believe it's a kind of sandhi. The name has a fused alternative spelling of Jālooj which is a continuous {jalȩwȩj}. But if you separate the components of Jālwōj at the cluster boundary and pronounce them in isolation, it's [tʲælʲ...wɤtʲ]. - Gilgamesh (talk) 22:26, 9 November 2019 (UTC)

Eh, I often get scatterbrained when working deeply on a project, like this one, and sometimes I have a tendency of making multiple small comments rather than one big organized one. But forget what I said about pausa and unreleased glides—I was trying to make a point, but realized after the fact I wasn't being very clear. So, let me put it a different way. Yes, all Marshallese utterances do begin and end with a consonant phoneme, even if it's an otherwise invisible sound that only colors vowels. So a vowel uttered in isolation is never just a vowel, but a vowel with a glide at its beginning and a glide at its end, and each of these glides count as a mora: a is {hah}, u is {wiw}, etc. In words like nana {nahnah}, one can analyze it as two sequences of {nah} [nʲɑ]. But when you fuse them together, the process of epenthesis adds a prosthetic vowel between them, changing the morphophonemic sequence {nahnah} with its unstable consonant cluster {hn} to a more stable {nahanah} in uninterrupted speech. Therefore, as counterintuitive as it may seem, [nʲɑ] + [nʲɑ] = [nʲɑːnʲɑ]. But if you enunciate it, and separate its components into separate utterances of a whole, it's [nʲɑ...nʲɑ]. - Gilgamesh (talk) 23:43, 9 November 2019 (UTC)

I changed the syllable break [.] to a space [ ]. I experimented with [..], and even [ | ], but [ ] was less visually disruptive. - Gilgamesh (talk) 00:21, 10 November 2019 (UTC)

Honestly, I'm really confused by the idea of either syllable breaks or spaces having moraic value at all. Usually their only meaning in IPA transcriptions is to separate syllables or words. I would think an extra mora due to a syllable-final glide should show up in the phonetic transcription as vowel length. But I'm getting the impression (partly from listening to the Marshallese Bible recordings) that phonetically there are at least three vowel lengths: a vowel before a non-glide consonant in the next syllable, a vowel before a coda glide, a vowel–glide–vowel sequence (resulting from epenthesis or not). Maybe this corresponds to one, two, and three morae, because the article says that the vowel–glide–vowel sequence is three times as long as a single vowel. But that would imply that /CV-G-V-C/ should be four morae, because of the final /C/, which contradicts the statement that /CV-CV-C/ is three morae. Still, such a distinction could be transcribed with ˑ and ː. This is definitely in OR territory. (Wiktionary doesn't frown on that as severely as Wikipedia.) And I might be completely out to sea.

Fun and possibly useful: you can install wikt:User:Erutuon/scripts/Marshallese.js in your common.js on Wiktionary to add an input box to search the table in wikt:Module:mh-pronunc by the orthography or any of the transliterations, or by notation containing CGV# (non-glide non-vowel, glide, vowel, word boundary). For instance, searching VGVGVGV shows Jāioon and Nuwio‌̧o‌̧k. — Eru·tuon 05:57, 10 November 2019 (UTC)

I'll definitely have to give that Marshallese.js a look soon. But your senses are keen to be skeptical about mora length. Yet Willson (2003) points out in Section 5 (Syllable Structure) that only sequences of CV or solitary C carry units of prosodic weight. That's how CVC is two, CVCVC is three, CVCCVC is four, etc. So when a vowel phoneme comes after a consonant phoneme, the vowel assumes the entire mora for both phonemes. It's clear from Choi (1992) that the glides still take up an amount of time similar to that of vowels, which can explain why long monophthongs are audibly three times as long (rather than two), but glides don't have a weight equaling vowels if a vowel follows one. Considering that Marshallese has an Oceanic ancestor that consisted of mostly only consonant-vowel syllables (CV, CVCV, CVCVCV, etc.), it's logical that the vowels that disappeared (lan̄ {lag} from *laŋi, "sky; heaven") cheshirized their prosodic weight into the solitary consonants left behind (slight OR on my part, but it helps my point). And these unpaired consonants still sometimes spring an epenthetic vowel if they're the first member of an unstable consonant cluster, even across word boundaries. - Gilgamesh (talk) 09:46, 10 November 2019 (UTC)

Adding to this, I noticed something else in Marshallese speech. It's already acknowledged that utterances that end with vowels can (though not always do) terminate them in an off-glide of [j] or [w]. So io̧kwe {yi'yakʷey} in isolation or at the end of an utterance can be [i̯æ͡ɒɡʷɛ] or [i̯æ͡ɒɡʷɛj], in free variation. (OR alert.) Though it's said that no literal [ɰ] ever surfaces, I've noticed that utterances ending in otherwise silent {h} can often terminate with a sharp glottal stop, so enana {yenahnah} is often [ɛnʲɑːnʲɑʔ]. Choi (1992) observes that Marshallese speakers seem intuitively aware of mora count, so this may just be the closest {h} comes to an enunciated off-glide. (End OR.) Come to think of it, perhaps off-glides would be a clearer way of accomplishing what I've been trying to do in isolate mode. - Gilgamesh (talk) 10:11, 10 November 2019 (UTC)

This isolate mode increasingly seems more problematic than it's worth. I just hoped to provide something more phonetically readable than either of the phonemic transcriptions that also shows how individual components can be pronounced in isolation. But if it appears no less convoluted, it's failed its purpose. I've already removed it from the wikt:Template:mh-ipa-rows. I'm considering removing it from the module code entirely. - Gilgamesh (talk) 11:01, 10 November 2019 (UTC)

Right, /VGV/ should have two morae, analogous to how in Latin and Ancient Greek syllable-initial consonants don't count towards syllable weight in poetry. But I was talking about syllable-final glides. Both mā /mʲæj/ and māj /mʲætʲ/ have two morae, but the /j/ doesn't surface, so the vowel in the realization of /mʲæj/ should be longer than the one in /mʲætʲ/ in order to explain the extra mora. To explain another way, the current transcription [mʲæ] doesn't seem logical: because [mʲætʲ] has two morae and the syllable-final [tʲ] counts as a mora, [mʲætʲ] minus [tʲ] should have one mora; but the word transcribed [mʲæ] has two morae. The extra mora comes out of nowhere.

So I was suggesting that the extra mora (and extra length) could be indicated with ˑ, because ː is already used. So māj /mʲætʲ/ → [mʲætʲ], /mʲæj/ → [mʲæˑ], /mʲæjælʲ/ (māāl) → [mʲæːlʲ]. With the same notation and the application of vowel epenthesis, nana /nʲæɰnʲæɰ/ would be [nʲɑːnʲɑˑ].

But if the glide is realized as [j] or [ʔ], that suffices to mark the extra mora: [mʲæj]. I think I did hear the glottal stop a few times in pre-pausal positions in the Bible recordings. — Eru·tuon 00:05, 11 November 2019 (UTC)

Yes, basically. An off-glide would suffice, but in free variation with its absence. Since it's impossible for a vowel to end a word without the extra mora being there, it is simpler (and reflected in most mainstream phonetic-level Marshallese transcriptions) to simply not mark it, any more than one would phonetically transcribe a French word in isolation to include the silent final consonant that reappears during liaison (French). A major difference is that in French liaison, that reappearing consonant is only mandatory in certain word sequences depending on grammatical construct and is normally omitted elsewhere, and it has to be learnt for each word. At least in Marshallese, depending on which vowel ends a word, the identity of the final glide is always predictable. And since the final vowel itself is colored for backness and roundedness, one can't say that the glide responsible is ever completely silent.

The purpose of this request for peer review is to improve the state of the way these wikis cover Marshallese in a way that not only pores over available reliable references in ways I haven't done so by myself (since most of the time I was the only editor working on this article), but also to reduce the use of original research. Austronesier, in offering to help, pointed out that sometimes a certain amount of synthesis in expanding the details is inevitable in situations like these, but it should be reasoned on strong references and agreed upon through editor consensus. For years this article already covered the basic topic of prosodic weight, as well as epenthetic vowels that occur between non-glide consonants, both as topics in the limited context described by Willson (2003). But the article did not cover the effects of epenthetic vowels neighboring glides at all until their behavior was confirmed and described by Bender (1968), a crucial reference I previously did not access in all these years the article has been worked on. Before reading Bender (1968), I never fully understood the nature of glides as they pertained to consonant clusters. And Austronesier has never encountered an Austronesian language with a phonology more complicated than that of Marshallese; it can be difficult to describe and explain even in far more ideal circumstances, since a lot of the concepts involved (like vertical vowel systems) are not exactly novice-friendly linguistics topics.

How much simpler it would be over all if we had ready access to editors fluent in Marshallese, but they are very rare because certain geographic and economic factors make internet access prohibitively expensive to communities in the Marshall Islands. Even the Marshallese-English Online Dictionary is not being designed principally for online access through modern digital devices, but with the intent of eventually being distributed in the Marshalls as an offline website hosted on removable media for use on whatever computers are locally available. - Gilgamesh (talk) 03:42, 11 November 2019 (UTC)