(L&N) Q&A Thread - Quick questions go here

[Salmoneus]

Hard to say. I think David Gil argues that you just don't, at least in some Indonesian varieties.

I'm just wondering how you parse clauses if the same words/phrases could be either nouns or predicates/verbs.

Mostly it will be clear from context.

If it is not clear from context, you can:

- narrow down possible interpretations making certain interpretations require marking (eg direct/inverse, transitivity, etc)

- allow disambiguative paraphrasis

- have non-free word order.

Or a combination of the above.

What sentence were you thinking of in particular that you found ambiguous?

[Omzinesý]

What sentence were you thinking of in particular that you found ambiguous?

I have no specific sentence in my mind. In English you usually have an article marking that what follows is an NP. I started thinking that if you don't have articles how does it work. The key idea of my next lang could be conversion.

I think Chinese heavily utilizes topic-comment structures which tells that what precedes the small pause is an NP. (I don't know how much conversion Cinese even has.)

Analytic languages are still a bit unknown territory to me.

[Salmoneus]

Well, if you want to think how a language would deal with a certain problem, it's good to begin by thinking of some situations where the problem could arise. Then you can consider how much ambiguity there really is, and how it could be prevented.

More to the point, without an example, it's hard to offer any answer beyond the very vague (as I did), because there's nothing concrete to comment on (and it's not always even clear if both people are thinking about the same problem).

I'm not trying to be mean, please do think of a couple of examples and I'm happy to offer some suggestions!

[Nel Fie]

Something that's been on my mind: Are there any known instances of a moraic natlang where particular phonological contexts can force a monomoraic element (e.g. nucleus, vowel, coda consonant, etc...) to "stretch" to one-and-a-half mora? Or inversely reduce to a half-mora? (The other half being either dropped or taken up by another element)

[Creyeditor]

What does half a mora mean?

[Salmoneus]

I thought by definition you couldn't have a half-mora? The mora is atomic, isn't it? If the atom in your language is the "half-mora", isn't that just a mora that's been mislabelled?

[Nel Fie]

I thought by definition you couldn't have a half-mora? The mora is atomic, isn't it? If the atom in your language is the "half-mora", isn't that just a mora that's been mislabelled?

This is hinting at the point I'm trying to elucidate.

What does half a mora mean?

I'm not entirely sure how specific and unified the concept of mora is, since I've seen it defined a number of ways. A rough sketch of the definition I'm thinking of could be: "a feature of phonological structures, dominantly expressed in terms of duration, the attribution of which is specific to a language and whose identification is necessary to explain other phonological specificities and requirements of that language".

A half-mora would therefore be held as an element that behaves, in the context of the language's phonology, as if it were only half of a mora long.

To give a made-up example:

Imagine a language that counts structures like CV and V to both be one mora each. However, it also has diphthongs VV̯, which the language counts to also be one mora, and which it distinguishes from vowel sequences V.V by counting the latter as two mora long.
The language would thus distinguish:

/ai̯/ - dog
/a.i/ - chief

... with the latter being twice as long as the former. One could analyse that the vowels in those diphthongs are each a half-mora long.

To the drive the point home, one could imagine other complexities, like half-moraic roots which undergo various phonological changes when free or bound to form a full mora, unlike their monomoraic or polymoraic counterparts.

But if I were to take the definition provided by Salmoneus in the most literal way, of mora as the smallest atomic structure in a language, would then the more correct analysis be that CV, V and VV̯ are in fact each two morae long, that V.V is in fact four morae, and that in a VV̯ diphthong, each vowel is one mora?

[Creyeditor]

Okay, what is a mora?
If a mora is a unit of duration (as in the idea of mora-timed languages), then you could in principle have one half a mora or 0.4457436 moras because duration is a gradient phonetic dimension. That's true for the first part of your diphthong example.
If a mora is what a heavy syllable has two of and a light syllable has one of (and syllable types are referenced by generalizations on stress, as in Larry Hymans work) then you cannot have half a mora.
If a mora is something more abstract that you use for complex calculations on stress, e.g. stress the vowel in position √(x*2) then you can have half a mora. I don't know if anybody does this. The problem is that I cannot see how it would influence phonology. But maybe that's what your alluding to in the second part of your dipthong example.
The closest thing I can think of are Meso-American and South-American languages that distinguish between consonantal and vocalic moras. Also, if you google "strong mora" and "weak mora" you probably get some relevant examples.
Btw, CV, V and VV̯ are a light syllable, a light syllable and a light or heavy syllable (language dependent), so 1, 1, 1 or 2. Atomic doesn't mean that every segment has to have a mora. Onsets can share their mora with a nucleus; and offglides can also share their moras with the rest of the nucleus.

[Salmoneus]

Imagine a language that counts structures like CV and V to both be one mora each. However, it also has diphthongs VV̯, which the language counts to also be one mora, and which it distinguishes from vowel sequences V.V by counting the latter as two mora long.
The language would thus distinguish:

/ai̯/ - dog
/a.i/ - chief

... with the latter being twice as long as the former. One could analyse that the vowels in those diphthongs are each a half-mora long.

What vowels? If the diphthong is a phoneme, then it does not contain other phonemes, and so those other phonemes cannot have characteristics such as length, because they don't exist, and non-existing things have no properties. If the diphthong is not a phoneme, but only a sequence of two phonemes that happen to be in sequence together, then each phoneme presumably has (/is/represents) one mora, and the vowels in the second example are simply long vowels of two morae each.

To the drive the point home, one could imagine other complexities, like half-moraic roots which undergo various phonological changes when free or bound to form a full mora, unlike their monomoraic or polymoraic counterparts.

Do those roots actually occur in those unbound forms? Then why say they are "half" anything? If they have the most basic unit of duration possible, then their duration is one mora, and calling one mora half a mora is just playing with words. If the roots do not actually occur in those unbound forms, then those hypothetical unbound forms do not have any duration at all, because they don't exist, and non-existing things have no duration. The actual words they are bound into would have duration, of course, which could be one or more morae long.

But if I were to take the definition provided by Salmoneus in the most literal way, of mora as the smallest atomic structure in a language, would then the more correct analysis be that CV, V and VV̯ are in fact each two morae long, that V.V is in fact four morae, and that in a VV̯ diphthong, each vowel is one mora?

Only if a language treats them that way, by treating V, CV and CVC syllables as being different lengths or weights. [and my meaning was that they were atoms of significant duration, not atoms of language per se].

You seem to be treating "mora" as a term on the phonetic level. But it's not - you're confusing it with "second". It's a term on the "phonemic" level. The fact that a consonant physically has a duration is irrelevant to mora counts unless a language regards that duration as in some way significant.


[yes, I'm aware I'm probably using 'mora' as synonymous with 'chroneme' here, and probably I should be technically talking about a mora as the bearer of the atom of significant duration, but who cares]

[Nel Fie]

So, I was about to respond but I came across an idea that might explain what I'm misunderstanding about mora. Please bear with me and consider the statement:

When spoken about in terms of duration, mora-timed languages, etc... Is it that morae are not features of phonological structures themselves, but rather that morae are more like independent and suprasegmental units of timing, and that phonological content merely happens to occupy them and can shift between them, without changing the morae themselves?

Or by analogy, that morae are essentially like bottles that are all the same size, and that phonological elements are merely like liquids that fill the inner volume of said bottles, and while different liquids can be mixed in the same bottle, or a liquid may only fill them up halfway, the bottles themselves do not change.

In that sense, one could not have a "half-mora" for the same reason that a bottle which is half-full is still itself an entire bottle, not a "half bottle" - it is only the contents which occupy half of its volume.

Is that a mora accurate way to think about morae?

[Creyeditor]

I don't know enough about the (either non-falsifiable or unproven) theory of Isochrony in order to answer this question.

[Nel Fie]

Isochrony wasn't really on my mind here - I was trying to figure out what I was getting wrong, since both you and Salmoneus seemed rather adamant that I hadn't quite understood what morae are.

But thank you for bringing it up, I had forgotten about isochrony and it'll probably be a good thing to investigate.

Well, I'll assume that we probably aren't going to make much more progress here, so thank you both for taking the time to engage with my original question. For what it's worth, I found something that approaches what I was looking for in an analysis of Kiribati (a.k.a. Gilbertese), especially if my analogy of "mora-as-bottle" is accurate.

In broad brush strokes and according to the document, it has a tri-moraic stress foot, which is also the constraint on its minimal prosodic word size. And within that contexts, there are instances where segments shift or undergo phonological changes, or epenthetic vowels are produced to satisfy the constraint. Four examples from the document:

a. Final moraic nasal + nonhigh vowel

E bon ataia.
3s verily know-3s
[e.pwon.na.tai.a]
'S/he really knows.'

b. Final moraic nasal + high vowel

Bon iran atuuna.
verily hair-of head-3s
[pwo.nii.ran.na.suu.na]
'It was his [head] hair.'

c. Final moraic nasal not licensed in rime (4vii)

Bon ngngai
verily Is
[pwo.ni3.g3ai]
'It was me.'

d. Final moraic nasal licensed in rime

Bon te kamea.
verily ART dog
[pwon.te.ka.me.a]
'It was a dog.'

For anyone interested, the source is "Trimoraic Feet in Gilbertese" by Juliette Blevins and Sheldon P. Harrison, released in Oceanic Linguistics, Vol. 38, No. 2 (Dec., 1999).

[Creyeditor]

Isochrony wasn't really on my mind here - I was trying to figure out what I was getting wrong, since both you and Salmoneus seemed rather adamant that I hadn't quite understood what morae are.

How does mora-timed language make sense without Isochrony?

---

As for Gilbertese. Yes, segments can be moraic or non-moraic. And moraic and non-moraic segments can alternate. And maybe one could say (reversing your metaphora) that segments are bottles that contain moras or not. Or syllables are bottles that contain moras or not. But (a) this is the phonological concept of a mora that is not directly related to the phonetic dimension of duration and (b) I don't understand how you need half a mora.

[Nel Fie]

How does mora-timed language make sense without Isochrony?

---

As for Gilbertese. Yes, segments can be moraic or non-moraic. And moraic and non-moraic segments can alternate. And maybe one could say (reversing your metaphora) that segments are bottles that contain moras or not. Or syllables are bottles that contain moras or not. But (a) this is the phonological concept of a mora that is not directly related to the phonetic dimension of duration and (b) I don't understand how you need half a mora.

I wouldn't know, since I only just remembered Isochrony as a thing, and since I obviously don't understand what morae are. For the same reason, the question of "half-mora" is moot at this point, since it was based on this misunderstanding.

So, let's take it from the top. Based on what you have written:

Mora are an abstract property possessed by certain phonological structures. The exact shape of said structures varies from language to language, from single segments to specific patterns of natural classes. Said property can be expressed as a binary, in so far that a structure either:
(1) Has a mora
(2) Does not have a mora

A mora can be shared, but it is retained as the whole and sole property of one the elements sharing it, which can be identified as such.

Furthermore, this property can express itself in a wide range of ways, but said expression is not inherent to the concept of a mora itself, but rather as separate processes which merely use morae as "reference points" or "units". Said processes can include:

1) Stress assignment based on syllable weight, measured in mora
2) Timing of individual mora-bearing structures or larger moraic feet
3) Synchronic phonological changes
4) Prosody and prosodic constraints

... as well as other things, or in case of many languages, no processes at all.

Is that about correct?

[Salmoneus]

I'm ignorant of linguistics, so I don't have a true theoretical definition, let alone any ontological specificity, only a practical notion that seems to cover how languages I've seen seem to work.

That being said, I'd suggest a simpler model:

- two words, or two syllables, can be systematically regarded by a language as taking more or less time to say. In this case, the word or syllable that takes longer (according to the rules of the language) to say contains more morae than the other

Furthermore:

- all phonemes in an utterance are associated with a mora
- morae are non-discontinuous (if two phonemes are associated with the same mora, the intervening phonemes must also be)
- morae are non-overlapping
- morae are determined by linguistic rules, being neither arbitrary nor interlinguistically objective


I think that covers everything. Have I missed something out?

A segment is described as moraic if adding or removing it alters the number of morae in the word, and as non-moraic if it does not.

[Creyeditor]

@Nel Fie:
I think number 2) is true (because of "processes can include") but somehow less so than the other ones. A closed heavy syllable can have the same phonetic duration as an open syllable even if they differ in mora count for the purposes of stress assignment. A closed heavy syllable does not need to be as long as a long vowel in its phonetic duration. Talking in terms of Isochrony, a stress-timed or a syllable-timed language can have a mora--based stress system, making moras differ drastically in their phonetic duration. Arabic is an example of a stress-timed language where you roughly stress the syllable that bears the third-last mora.

---

I think Sals Furthermores 2,3,4 are really crucial for the concept of a mora in any kind of Prosodic Phonology or Autosegmental Phonology framework of moras.

[Nel Fie]

Alright. Although I don't feel like I fully understand what morae are yet, maybe that's the point. At least, it comes across as if being hard to define is part of the point. Either way, I think I'll call it day. Maybe I'll take the time to read up on them in a more targeted way once I've managed to digest what we've talked about.

Thank you both very much for taking the time to provide your own insights on the matter and being patient with me. I appreciate it a lot.

[Creyeditor]

It is a abstract phonological unit but I think you got the gist. If was asked to provide a tl;dr it would be the following.

• A mora is a unit of phonological weight/length not a unit of phonetic duration.
• Isochrony moras and weight moras might not be the same.

If you want to do some further reading, I would suggest really anything by Larry Hyman on moras. I always found his writing style relatively easy to understand even without a background in theoretical phonology. His book, a theory of phonological weight, is probably most comprehensive, but really anything will do.

[Nel Fie]

Thank you. I'll keep the recommendation in mind once I find the time to read up.

[Creyeditor]

I really like Wiktionary-style entries that show the same (orthographic) word in several languages and lists their meanings. This made me wonder: What (orthographic) word is a word in the highest number of languages (excluding trvivial cases like letter names (e.g. be) and international loans (e.g. internet)?