Page 1 of 1

Sound Changes - Monosyllabisation

Posted: Sat Oct 01, 2011 5:17 pm
by chris_notts
Hi all, sound changes and diachronics haven't historically been my strong point. But since I spent quite a lot of time writing my own sound change applier I thought I'd better use it for the project that's been an idea in my head for quite a while. I'm going to outline my sound changes below and I'd appreciate comments.

I'll outline the changes first, then add posts about natlangs where similar changes have happened where I know of them and some example words.

Some of these changes could be reordered without changing the results, and in others there are several possible pathways to the same result.

The goal I set myself was to start off from something vaguely similar (although not identical) to proto-Austronesian and then generate something more typical of South-East Asian languages, i.e. development of a large number of monosyllabic morphemes and an increase in the number of possible syllables.

The proto-language

The proto-language was deliberately chosen to have a relatively simple structure and little that's particularly odd or remarkable.

The proto-consonants are:

p t c k ʔ P T C K s h m n ŋ r l w j

where /P/ /T/ /C/ /K/ are a fortis series of stops, which probably evolved from geminates at some point historically. The fortis stops only occur intervocally, and therefore don't occur either word initially or word finally. /ʔ/, /r/ and /l/ also don't occur word initially. The palatals may have been pure palatals or post-alveolar affricates.

This means that the set of consonants which can occur word initially are:

p t c k s h m n ŋ w j

The proto-vowels are the typical:

i e a o u

Syllable structure is simple. All syllables must be open except for the final syllable, which can be closed by any consonant except for the fortis stops. All syllables must have onsets apart from the initial syllable of the word, which can be onset-less.

Stress was on the second syllable, and on every other syllable after that. Content words had a minimum weight constraint: mono-syllabic words had to have a closed syllable.

Early developments

(1) Change of stop contrast to voiced vs voiceless

Lenis stops become voiced except for word finally, and fortis stops become voiceless. This means that the only stops that can begin a word are voiced, which sounds odd but is actually the case in native Basque vocabulary. Basque also historically had a stop contrast of some other kind that developed in this way into a voicing distinction under the influence of neighbouring languages.

(2) Merger of e with other vowel phonemes

E becomes a copy of the following vowel, and is raised to i word finally. This probably happened via centralisation, i.e. e developed to something like ə, which was then influenced by following vowels. The actual changes are:

e -> V1 / _ C V1
e -> i / _ #

After this, the remaining vowels shift their pronunciation into a more box-like formation.

i a o u -> i æ ɒ u

Crucially, /æ/ and /ɒ/ were both low vowels.

(3) Shift of the palatals

The palatals /c/ and /ɟ/ fricativised and then shifted forwards as part of a chain shift. The exact change was:

c -> s -> θ
ɟ -> ʃ

Development of secondary articulations and further vowel changes

(1) Development of secondary articulations

Consonants gain secondary articulations based on the preceding vowel. Consonants follow i are palatalised, and consonants following either u or ɒ are labialised. No change occurs after æ.

C -> Cʲ / i _
C -> Cʷ / [vowel,rounded] _

(2) Stressed vowels undergo a split depending on the following vowel height

The 4 vowel system had 2 high vowels and 2 low vowels. In stressed syllables, these vowels underwent a split based on the height of the following vowel.

[vowel,high,stressed] -> [high-mid] / _ C [vowel,low]
[vowel,low,stressed] -> [low-mid] / _ C [vowel,high]

Everywhere except word finally,

æ, ɒ -> a

Word finally, where this split didn't occur because there was no following vowel, there was a shift in vowel quality:

æ -> ɛ
ɒ -> a

(3) Reduction of unstressed vowels

All unstressed vowels reduce to ə.

Loss of unstressed syllables part 1 - development of non-final unstressed syllables

(1) Schwa is lost before a syllable containing a stressed vowel, producing a large number of onset clusters

(2) If the second element of the cluster if a semi-vowel, that semi-vowel overrides any secondary articulation the consonant already has

Cj -> Cʲ
Cw -> Cʷ

If the first element is a semi-vowel and the second is a glottal, then the same occurs:

j [glottal] -> [palatalised]
w [glottal] -> [rounded]

(3) h is lost before or after other consonants

(4) Secondary articulations are lost from a consonant preceding another consonant

(5) If the second element of the cluster is r or l and the first is a stop, fricative or a nasal, then the cluster is retained. l -> r after a consonant. Any secondary articulation is lost on r.

(6) Stops are devoiced before r

(7) Nasals become voiced stops before r

(8) t,d,s,θ,ʃ -> s / _ C

(9) nasals assimilate to a following obstruent

(10) s, nasals, ʔ lengthen a following obstruent and delete. ʔ also lengthens a preceding obstruent and deletes.

(11) bilabial stops and velar stops are lost without any reflex before another obstruent.

Re: Sound Changes - Monosyllabisation

Posted: Sat Oct 01, 2011 5:18 pm
by chris_notts
Loss of secondary articulations

(1) palatalised velars become palatal

(2) labialised alveolars and post alveolars become retroflex. The retroflex rhotic becomes an approximant.

(3) clusters of (post-)alveolar + r become retroflex

(4) ɭ -> ɻ

(5) Palatalisation and rounded separate from glottals before a stressed vowel. e.g. ʔʷ -> ʔw

(6) Palatalised separates from bilabials before a stressed vowel, e.g. pʲ -> pj

(7) Other secondary articulations are lost

Development of geminates

(1) Shift in manner for stops

Initially, the geminate and singleton stops are distributed like this in terms of glottis spreading:

|----- t tː ----- d dː -----|

A number of these geminates were word initial and hard to perceive. The differences were accentuated by shift along this dimension, so that long unvoiced stops gained breathiness and long voiced stops became creaky, with the creak probably continuing onto the following vowel. This produced a system as follows:

| tʰ ----- t ----- d ----- d̰ |

The short voiced stops may have been slightly breathy (d̤ instead of d).

This was then followed by:

(i) Lenition of the plain voiced stops to fricatives or approximants

d -> ð
b -> β
ɖ -> ɻ
g -> ɣ
ɟ -> ʝ

(ii) Loss of voicing of (very) creaky voiced stops, yielding ejectives

(iii) Voicing of the unvoiced unaspirated stops

This gave a three way aspirated vs voiced vs ejective system of stops.

(2) ʃː,c -> cʰ

(3) Long fricatives become affricates

(4) Long nasals are devoiced

(5) Length was lost otherwise

Further developments in consonants

(1) ʈʼ -> cʼ

(2) ʝ -> z

Development of finals and post finals (second consonant in a cluster)

(1) Assimilation of final schwa to following consonants:

əw -> u
əj -> i
əh lost

(2) Loss of remaining schwas, which may produce a cluster of two final consonants

(3) Devoicing of obstruents word finally

(4) ð,ɣ -> j / word finally or before an obstruent, β -> w in the same environment

(5) l,r,ɻ -> h / _ word finally or before a consonant

Development of post final obstruents

(1) stops and affricates shift to glottal stop when preceding a nasal or following another obstruent word finally

(2) fricatives and approximants shift to h when preceding a nasal or following an obstruent word finally

(3) other consonants (ie nasals) are lost after another consonant word finally

(4) first in a cluster of two consonants finally is generally protected

Development of singleton

(1) alveolar and retroflex stops become glottal stop word finally

(2) fricatives become h word finally

Development of glottal final consonants - incipient tone

(1) Words with a glottal stop in the final (cluster) develop high tone on the final syllable and lose it

(2) Words with a h in the final (cluster) develop low tone on the final syllable and lose it

(3) Words with neither have a middling tone

Further changes

(1) Semi-vowels lost after similar vowels, e.g. j lost after i

(2) Front vowels centralised and merged after retroflex consonants.

i,e -> ɨ / [retroflex] _
ɛ,a -> ɑ / [retroflex] _

(3) Simple vowel shifts

ɨw -> u
iw -> ju

Results

Consonants in onsets as a result of the changes include:

pʼ tʼ cʼ kʼ
pʰ tʰ ʈʰ cʰ kʰ ʔ
b d ɖ g
m̥ n̥ ɳ̥ ɲ̥ ŋ̥
m n ɳ ɲ ŋ
t̪θ ts ʈʂ
θ s ʂ ʃ h
β ð z ɣ
l r ɻ
j w

This represents approximately a doubling of the number of onsets.

I may do some mergers of the nasals, especially the voiceless nasals since I think place is harder to perceive for them. Perhaps a collapse of the retroflex nasals with something else to start with, either alveolar or velar, and a further merger of two POAs in the voiceless series.

Onset clusters with r, w and j occur, although these are more frequent with some POAs than others.

There are now seven phonemic vowels, with some of the contrasts neutralised after retroflex consonants:

i e ɛ a ɔ o u

The codas are more limited, and amount to:

w j k p c m n ɳ ɲ ŋ

(Again, I probably need some nasal mergers)

There is an incipient tone system, with high and low tones occurring on final vowels. In my 400 word experiment, approximately 20% - 30% of the words have a high or low final tone, with middle tone being by far the most common. I probably should assign default tones to some of the middle tone final syllables based on onset consonants. There is no tone contrast on non-final syllables of multi-syllable words.

There has been a massive reduction in word size: words with up to three syllables are monosyllabic as a result of the changes. Words with four syllables originally retain two.

I've formulated this all in HaSC, although it's a bit messy because it evolved rather than being completely planned from the beginning.

Re: Sound Changes - Monosyllabisation

Posted: Sat Oct 01, 2011 5:31 pm
by chris_notts
Some example changes:

sepe -> pʼi
mena -> n̥ɛ
saru -> θru
woKisi -> gì
tumaTo -> m̥á
sitoha -> tʼà
miTeha -> tʰà
ciPeCe -> pʰì
joPe -> bi
koʔesi -> kʼì
siPa -> pʰjɛ
kepas -> βà
alam -> lam
teŋu -> ŋ̥u
narehe -> ɻì
ŋipaKu -> pʼjɛk
seral -> θrà
oka -> ɣɛ
heTeme -> dim
cehato -> ʃaj
jatal -> ðà
ŋukuso -> kʼò

Of course, serious compounding or rebuilding of words will have to occur alongside this demolition. For example, I'm planning to grammaticalise a few of these as noun class prefixes, and restore distinctiveness via a class system.

Re: Sound Changes - Monosyllabisation

Posted: Sat Oct 01, 2011 5:53 pm
by chris_notts
Examples of natlangs which have done similar things. Mostly off the top of my head (based on a few months of reading up about these kinds of changes):

Reduction and loss of unstressed initial syllables with compensatory changes in stressed syllable

This has happened repeatedly in East Asian languages. Many of them were historically 'sesqui-syllabic', with many words having two syllables with iambic stress, but have since developed primarily monosyllabic forms. Some of these languages mostly just dropped the unstressed initial syllable, whereas others integrated some of the phonological material into the stressed syllable in various ways. I'm not aware of any East Asian language developing word initial geminates as compensation for the loss, but I don't see why it can't happen.

In addition, quite a few Australian languages historically deleted the whole initial syllable. This is despite the fact that those languages had, at least at some point in their history, word initial stress. Dixon explains this by claiming that the stress occurred late in the vowel, which made initial onsets less prominent than might be expected and led to their loss. After the loss of initials, stress shifted and the initial vowel was often then lost.

Many of these Australian languages did partially preserve the quality of the lost vowel as labialisation / palatalisation as I have done.

Development of word initial geminates

This has occurred in at least limited circumstances in some languages. Some dialects of Trique apparently gained monosyllabic forms with geminate initials by losing unstressed word initial syllables with certain shapes. Some Austronesian languages developed word initial geminates by deleting unstressed vowels between identical consonants. In the Austronesian case, what probably aided retention was that a common environment for the sound change was reduplication of the first syllable of verbs, which was morphologically significant. I've therefore tried to aim for a high functional load of initial geminates in my changes to motivate their retention instead of loss.

Development of retroflexes from labialised alveolars

I can't give a straightforward example of a language where this occurred. However, in many languages there is a correlation between retroflex consonants and neighbouring back / rounded vowels, including in Australian languages. In many languages, vowels next to retroflex consonants that are front are centred or lowered as well, which is what I've done.

Loss of all unstressed vowels

Loss of unstressed (alternating) vowels is a common change, although typically it is conditioned more than in my case. One example of a language with a fairly simple structure (like my proto-language) which deleted all almost unstressed vowels, including vowels in the first syllable, is Dorig, which is an Austronesian language spoken in the Vanuatu islands.

Voiced geminates to ejective

I have no direct examples of this, but something suggestive did happen in Southern Sotho:

*mp -> pʰ
*mb -> pʼ

and similarly for other POAs. The intervening steps aren't described in my source, but something like gemination followed by changes in laryngeal setting seems plausible.

Re: Sound Changes - Monosyllabisation

Posted: Sat Oct 01, 2011 6:04 pm
by chris_notts
Some of the Chamic family of Austronesian languages actually went from being polysyllabic to mostly monosyllabic due to influence from neighbouring languages. I've found at least one book about the development of the Chamic languages that I'd love to get my hands on:

From Ancient Cham to Modern Dialects: Two Thousand Years of Language Contact and Change (Oceanic Linguistics Special Publications)
http://www.amazon.com/Ancient-Cham-Mode ... 0824821319

Re: Sound Changes - Monosyllabisation

Posted: Thu Oct 06, 2011 2:33 am
by chris_notts
I guess this post must have been a little boring. I'll post some more here soon about the grammar that will hopefully be more interesting.

Re: Sound Changes - Monosyllabisation

Posted: Thu Oct 06, 2011 5:16 am
by äreo
I'm surprised no one else has replied. This looks pretty awesome. And even helpful - I often have trouble with these kind of extensive sound changes that leave a language unrecognisable.
/t̪θ/ ftw

Re: Sound Changes - Monosyllabisation

Posted: Wed Oct 26, 2011 6:22 pm
by Zaris
I agree with the gent above me. Not only is this post extensive and enjoyable, it's quite entertaining because it raises so many interesting ideas for my own conlangs. But, I did have one question.
(ii) Loss of voicing of (very) creaky voiced stops, yielding ejectives
Being a lover of ejectives, I have to ask. Might you have any real world examples of this change? I would be interested in a language that produced this change

Re: Sound Changes - Monosyllabisation

Posted: Thu Oct 27, 2011 2:01 am
by chris_notts
Zaris wrote:I agree with the gent above me. Not only is this post extensive and enjoyable, it's quite entertaining because it raises so many interesting ideas for my own conlangs. But, I did have one question.
(ii) Loss of voicing of (very) creaky voiced stops, yielding ejectives
Being a lover of ejectives, I have to ask. Might you have any real world examples of this change? I would be interested in a language that produced this change
Well, see here:

chrisdb.dyndns-at-home.com/uploaded/meditsai.pdf

or the other thread, where there's a little discussion on this:

http://zbb.spinnwebe.com/viewtopic.php?f=4&t=39158

The grammar I've been writing up contains natlang examples of interesting changes where I know of them. I don't know of a clear example of this particular change, but it appears that geminate voiced stops have yielded implosives in languages like Sindhi. And if you want an unclear example, it looks like Southern Sotho did changes like this:

*mp -> pʰ
*mb -> pʼ

which is at least suggestive of the kind of changes I've done.