Origin and Distribution of Proto-Micronesian p_w m_w

[chris_notts]

Hi all, I've been reading "Historical Linguistics: Theory and Method" because it contains a section on the sound changes that produced Marshallese, and I have a couple of questions about proto-Micronesian.

Firstly, I was wondering if anyone could tell me the origin of the proto-micronesian phonemes /p_w/ and /m_w/? Having rounded/velarised labials without any other consonants with secondary articulations seems a little odd.

I am also interested in their distribution. Looking at the various examples in the book, I am guessing that these plain labials and the velarised series only contrasted before front vowels, but this isn't stated anywhere so I can't be sure whether it's true or just a coincidence because of the small sample size.

Finally, can anyone tell me the nature of the two dental/alveolar stops and fricatives? The book writes them t T s S, but it doesn't say anywhere what the nature of the series written with capital letters is. The different isn't actually reflected in Marshallese, since it seems that t, T -> t_j and s, S -> t_G, but I would be interested to know what is was.

I ask because I am interested in languages which have experienced transfer of qualities from vowels to consonants followed by reductions in their vowel systems. One interesting thing about Marshallese is that it seems that the consonant splits were actually somewhat limited. The following consonants have splits:

(i) The velars k and ŋ gain labialised versions before back vowels
(ii) r,l,and n show a three-way split into palatalised, velarised and labialised depending on the following vowel

The following proto-phonemes don't split, but always show the same reflex regardless of the following vowel: p p_w f m m_w x j w t T s S ʒ c ɲ. Some of these reflexes are either velarised or palatal(ised) consonants, but they are not conditioned by neighbouring vowel segments.

So, as you can see, the splits that occurred between Proto-Micronesian and Marshallese don't look like they would preserve the front-back distinction in the majority of cases. This is why I am interested in, for example, where there was a p vs p_w contrast before back vowels in Proto-Micronesian, because if there wasn't then the distribution of phonemes at that stage already partially correlated with the backness of the following vowel.

[chris_notts]

I'm interested in this because I want to do something similar for a conlang and I'm reading up on different languages which have had these kinds of changes. Examples include:

Arrernte and other Arandic languages, which have partially or completely neutralised the contrast between i and u but gained labialised consonants in the process

A number of Mongolian languages, which have gained a palatalised series of consonants when a high vowel followed, and then neutralised all unstressed short vowels to @. Note that since most Mongolian languages have vowel harmony and the initial vowel is typically preserved, the main vowel contrast that is lost in this case in a height contrast.

This change is claimed for a number of Caucasian languages, but I can't find any details about their histories.

It has been claimed that short vowels in Irish are only distinguished by height, and get their front/back specification from neighbouring consonants. I can believe it, since Irish seems to have been fairly thorough in palatalising its entire phoneme inventory before front vowels historically, but I don't know enough about Irish to be sure either.

[Thomas Winwood]

The distinction between *p/*m and *pʷ/*mʷ would appear to be inherited from Proto-Oceanic, which has separate *p/*ᵐb/*m and *pʷ/*ᵐbʷ/*mʷ. (Citation: Ross, Malcolm D.; Andrew Pawley; Meredith Osmond, eds. (1998). The lexicon of Proto-Oceanic: Volume 1, Material Culture. Canberra: Australian National University. ISBN 9780858835078.)

As for *t/*T, *s/*S and *c/*Z, read page 68, footnote 3. The precise phonetic distinction is undecided.

[chris_notts]

The distinction between *p/*m and *pʷ/*mʷ would appear to be inherited from Proto-Oceanic, which has separate *p/*ᵐb/*m and *pʷ/*ᵐbʷ/*mʷ. (Citation: Ross, Malcolm D.; Andrew Pawley; Meredith Osmond, eds. (1998). The lexicon of Proto-Oceanic: Volume 1, Material Culture. Canberra: Australian National University. ISBN 9780858835078.)

As for *t/*T, *s/*S and *c/*Z, read page 68, footnote 3. The precise phonetic distinction is undecided.

Do you know if the contrast between *p/*m and *pʷ/*mʷ was neutralised before back vowels?

[Salmoneus]

Not in proto-Oceanic, no.

EDIT: however, it BECAME neutralised in many later oceanic languages - and some even still have free variation between pwi and pu (etc). [It seems unclear whether there were three or two labiovelars in POc - earlier reconstructions had only two, but I think they go with three these days]. However, some languages apparently do maintain the distinction, so it hadn't been lost in POc.

You probably want to read "The Proto-Oceanic Labiovelars: Some New Considerations", by John Lynch. I would tell you what it said, but not having JSTOR (etc) access, I can't read it myself...

EDIT 2: in case it's not clear, it's apparently thought that the "labiovelar" series originated in POc, rather than being inherited. Looking at the abstract of the Lynch, it seems he blames borrowing, phonological conditioning, subsequent dissimilation of rounded vowels next to labiovelars (does this disagree with what I said before? Maybe it's only in some circumstances?), use of the distinction to disambiguate, and "increased functional load" of labiovelars in newly created words.

[chris_notts]

Not in proto-Oceanic, no.

EDIT: however, it BECAME neutralised in many later oceanic languages - and some even still have free variation between pwi and pu (etc). [It seems unclear whether there were three or two labiovelars in POc - earlier reconstructions had only two, but I think they go with three these days]. However, some languages apparently do maintain the distinction, so it hadn't been lost in POc.

Thanks - it sounds as if there might have been a correlation then, even if not a neutralisation.

You probably want to read "The Proto-Oceanic Labiovelars: Some New Considerations", by John Lynch. I would tell you what it said, but not having JSTOR (etc) access, I can't read it myself...

I would really love to have access to sources like JSTOR, but unfortunately I don't have it either. IIRC, last time I looked, it seemed pretty expensive for individual access.

EDIT 2: in case it's not clear, it's apparently thought that the "labiovelar" series originated in POc, rather than being inherited. Looking at the abstract of the Lynch, it seems he blames borrowing, phonological conditioning, subsequent dissimilation of rounded vowels next to labiovelars (does this disagree with what I said before? Maybe it's only in some circumstances?), use of the distinction to disambiguate, and "increased functional load" of labiovelars in newly created words.

Some of those sources are pretty interesting. That's part of what I find hard about coming up with histories for conlang - the bits that aren't perfectly regular or have a complex mixture of origins.

Incidentally, I have read that Basque m spread for similar reasons. All the evidence is that proto-Basque's only nasal was n, but m was acquired and spread through the lexicon in a number of different ways including borrowings from IE languages, simplification of sequences like *nb, expressive uses in compounding, conversion of existing b to m in other contexts (it's not clear to me if the process was regular or not) etc. I guess that it's quite common for languages which acquire a new phoneme to want to put it to use.

EDIT: It looks like in Basque, b > m is sporadic, but it occurs more often word initially and if another nasal follows in the word

[Soap]

So youre saying pʷ bʷ mʷ are actually labiovelars, not simply rounded bilabials?
That would make sense, since I believe they turn up as plain velars in some descendants, but I thought it was generally reconstructed as a rounded bilabial with no other coarticulation.

[Salmoneus]

They're called "labiovelar", because they're labial and velar. Obviously we don't know what they were in either proto-Micronesian or proto-Oceanic. There's a lot of variation in how they've ended up in daughter languages: labials, velars, velarised labials, rounded labials, velarised and rounded labials, rounded labials, co-articulated labials and velars, labials with off-glides, and combinations of the above.

It's true they generally seem more closely associated with labials than with velars; but then, in many language 'labiovelars' are very closely associated with velars.