TLFKAT: phonology for a polysynthetic conlang

[DesEsseintes]

0. Welcome to the Esseintial TLFKAT Scratchpad!

I've decided to start a scratchpad to present a work-in-progress nicknamed TLFKAT for now.

So What is the Language Called?

Unfortunately, I'm having great difficulty deciding on a name for the language, especially as I keep changing the phonology.

The original name was Tlííeno. That name soon became incompatible with the phonology and kept changing (Tłííʼoengwo, Tłííyego, etc...) so I decided to use TLFKAT (The Language Formerly Known As Tlííeno) until I settle on a final version of the phonology. In my mind I tend to think of it as Toffeecat.

Design Objectives

TLFKAT is a polysynthetic language inspired by Wakashan, Navajo, Caddo, and - to a lesser degree - Kalaallisut. The phonology especially is meant to look Native North American, and my choices regarding the orthography are meant to reflect this.

I aim to make TLFKAT a verb-heavy language with a complex verb structure that is primarily templatic, but to some extent hierarchical. Noun incorporation is present, but is mainly restricted to body parts and elements of a locative, instrumental or adverbial nature.

Although I've been thinking about TLFKAT for a few months now, many aspects of the language remain undeveloped. I foresee very slow progress, but hopefully having this scratchpad will provide some impetus.

I will number my posts for ease of reference, but that does not imply that any rigid order will be followed in presenting material on the language. This is a scratchpad, after all.

Everything presented herein is provisional, and subject to change, but that probably goes without saying.

[DesEsseintes]

1. Phonology

The analysis of TLFKAT phonology presented here is perhaps somewhat questionable. Comments, suggestions, and criticism are welcome.

Consonants

In this analysis of the phonology, TLFKAT has 34 consonant phonemes as laid out in the following table. A further 5 marginal phonemes are listed in parentheses.

Code: Select all

p    m        (pʷ   mʷ)
t              tʷ
t͡s   s         t͡sʷ  sʷ   
t͡n̥   n         t͡n̥ʷ  nʷ
t͡ɬ   ɬ    l    t͡ɬʷ  ɬʷ   lʷ  
t͡ʃ   ʃ 
t͡ɻ̝̊   ɻ         t͡ɻ̝̊ʷ  ɻʷ
          j
k    x
                    w
              (k͡ŋ̊ʷ) ŋʷ
              (k͡ʟ̝̊ʷ) ʟ̝̊ʷ   ʟʷ
q   (ʁ)
ʔ              ʔʷ

Disregarding labialisation and arranging the phonemes according to MoA with orthography, gives us the following:

/p t k q ʔ/ p t k q ʼ*
/t͡s t͡n̥ t͡ɬ t͡ʃ t͡ɻ̝̊ k͡ŋ̊ k͡ʟ̝̊/ ts tn tł tš tr kg kƛ
/s ɬ ʃ x ʟ̝̊ ʁ/ s ł š x ƛ r̊
/l ɻ j w ʟ/ l r y ų λ
/m n ŋ/ m n g

*the glottal stop is normally not written word-initially

Labialisation is represented orthographically by the following vowel.

Notes on the Consonant Inventory

• the uvular fricative /ʁ/ is rare and mostly occurs as an allophone of the uvular stop
• the velar nasals and laterals /k͡ŋ̊ʷ ŋʷ k͡ʟ̝̊ʷ ʟ̝̊ʷ ʟʷ/ have no unlabialised counterparts and are mostly in complementary distribution with their alveolar counterparts /t͡n̥ʷ nʷ t͡ɬʷ ɬʷ lʷ/ except before the vowel [ɔ] where the difference is phonemic (see vowel section below)

Nasal "Affricates"

The nature of the phonemes /t͡n̥ k͡ŋ̊ʷ/ tn kg is perhaps more accurately described as pre-stopped nasals. However, I choose to group them with the affricates as they behave in exactly the same way as the more orthodox affricates at the morphophonological level.

Vowels

TLFKAT has a simple five-vowel system. However, each plain vowel has an "aspirated" counterpart. I've used the breathy voice diacritic to indicate this.

/a e i ɤ ɯ/
/a̤ e̤ i̤ ɤ̤ ɯ̤/

After a labialised consonant /ɤ ɯ/ are rounded while /a e i/ are realised with a labial onglide.

However, the actual realisations of the vowels tend to be significantly lower than the cardinal values:

Plain: [ä ɛ e̝~ɪ ʌ ɤ̝] a e ı ę ı̨
Labialised: [o̯ä o̯ɛ ʏ̯ɪ ɔ o̝] oa oe uı o u

The aspirated counterparts are marked with an acute accent.

TLFKAT also possesses at least the following diphthongs

/ai au ei ou ɯi ia ɯa ua/ [aɪ̯ aʊ̯ ɛɪ̯ ɔʊ̯ ɯ̯̽ɪ ɪ̯a ɯ̯̽a ʊ̯a] aı au eı ou ı̨ı ıa ı̨a ua

All monophthongs can occur short or long. Long vowels are written double.

Aspiration

The vowels listed as breathy voiced in the inventory affect adjacent consonants.

Preceding stops and affricates are aspirated.

p t t͡s t͡n̥ t͡ɬ t͡ʃ t͡ɻ̝̊ k k͡ŋ̊ʷ k͡ʟ̝̊ʷ q ʔ
→ pʰ tʰ t͡sʰ t͡n̥ʰ t͡ɬʰ t͡ʃʰ t͡ɻ̝̊ʰ~t͡ʂʰ kʰ~k͡xʰ k͡ŋ̊ʷʰ k͡ʟ̝̊ʷʰ qʰ~q͡χʰ ʔʰ

Preceding fricatives and nasals are devoiced (if voiced) and lightly aspirated. Approximants fricativise.

m s n ɬ ʃ ɻ x ŋʷ ʟ̝̊ʷ (ʁ)
→ m̥ʰ sʰ n̥ʰ ɬʰ ʃʰ ɻ̝̊ʰ ŋ̊ʷʰ ʟ̝̊ʷʰ (ʁʰ)

Labialised consonants follow exactly the same pattern so I skipped them to save space.

Notes

• /l j w ʟʷ/ never precede aspirated vowels
• /t͡ɻ̝̊ʰ ɻ̝̊ʰ t͡ɻ̝̊ʷʰ ɻ̝̊ʷʰ/ are [t͡ʂʰ ʂʰ t͡ʂʷʰ ʂʷʰ] for some speakers
• /kʰ qʰ/ are frequently realised as [k͡xʰ q͡χʰ]
• Note that there is no phonemic /h/ although there is an aspirated glottal stop
• /ʁʰ/ is as of yet unattested

Restrictions on CV sequences

Although the labial vs plain consonant distinction is mostly straightforward for the alveolar and glottal consonants, there are several irregularities in the system:

Uvular consonants can only be followed by one of a e o. Note that the back vowel is rounded even though the uvular consonants are not.

The velar nasals and laterals (kg- and kƛ-series) can only be followed by one of oa oe o.

The retroflex consonants (tr-series) can only be followed by back and/or rounded vowels a ę ı̨ oa oe o u uı. That is to say, they can not be followed by e ı.

The tš-series and y behave as if labialised before back vowels and can therefore only be followed by one of a e ı o u.

The labialised tł- and tn-series can only be followed by o u uı but not oa oe. Therefore, we can see that these are in complementary distribution with the kƛ- and kg-series except before o.

Bilabial consonants (p-series) can occur with any of a e ę ı ı̨ o u. This is the remnant of a labialisation distinction in velars that evolved into bilabials (kʷ → pʷ), but the labialisation distinction was lost before central and front vowels. I'm not sure about the details.

I probably should make some ginormous table to illustrate all this but I don't know how.

Syllable Structure

CV(C)(C)

All syllables must have a consonant onset.

The nucleus may be long or short, or a diphthong.

Permissible codas are the following:

ʼ q ƛ λ g k x š r ł l n s t p m

Word-finally the set of permissible codas is different:

ʼ q k x š r tł ł l ts s tn n t p

Furthermore, the following clusters are permitted word-finally:

{s n l ł r} + {p t ts tn tł k q}

That is to say, any alveolar resonant followed by any of the permissible word-final stops and affricates.

Such clusters are probably a result of word-final vowel deletion.

Labialisation is lost in coda position, although it can resurface before a suffix.

Sample Words

Here are a few simple words just so readers can get a feel for the sound of the language.

amottsíłpıtł [ʔämɔtt͡sʰe̝ɬpe̝t͡ɬ] n small branch, twig
eeyıtłut [ʔɛːje̝t͡ɬo̝t] n whale
sı̨spáí [sɤ̝spʰaɪ̯] n village
tsóáárxe [t͡sʰo̯äːɻxe] n birch


The next post will contain more details on allophony and introduce some morphophonological processes.

[DesEsseintes]

2. Allophony and Morphophonological Processes - Part I

Here I will outline some of the more common morphophonological processes in TLFKAT.

Consonant Sandhi I

Here are some of the processes that apply when two consonants come into contact.

First let's classify our consonants

Stops: p t k q ʼ
Resonants: m* s n l* ł r ṣ g λ* ƛ x* r̊*
Affricates: ts tn tł tr tṣ kg kƛ

*these resonants don't have affricate counterparts

Below I'm going to use the following notation:
V = vowel
C = stop
R = resonant
S = syllabic resonant
CR = affricate

And here are the sandhi rules:

1. Resonant + resonant
RR → RCR
The second resonant undergoes fortition to the corresponding affricate
en + nı → entnı son
páál + ług → pááltług heart (organ)

2. Resonant + affricate
RCR → RCCR
An excrescent stop corresponding in PoA to the resonant is inserted before the affricate.
rug + tłán → rugktłan - satchel
ųeeyın + tníín → ųeeyınttníín - to come into sight, appear

3. Short vowel + affricate + resonant
CVCRR → CVVRCR
The affricate weakens to the corresponding resonant but the preceding vowel undergoes compensatory lengthening. The resonant undergoes fortition to affricate.
pétn + łanaq → pééntłanaq - soothsayer, shaman
ayetł + sos → ayeełtsos - to pout

4. Short vowel + affricate + stop/affricate
CVCRC(R) → CVVRCC(R)
The affricate weakens to the corresponding resonant but the preceding vowel undergoes compensatory lengthening, and an excrescent stop corresponding in PoA to the original affricate is inserted after the resonant.
pétn + kgo → pééntkgo - spell, curse

5. Long vowel + affricate + resonant
CVVCRR → CVVCSCR
The release of the affricate syllabifies, and the resonant fortifies to the corresponding affricate.
yootł + ná → yootłtná [jɔːtɬ̩t͡n̥ʰaʰ] - everywhere

6. Long vowel + affricate + stop/affricate
CVVCRC(R) → CVVCSCC(R)/CVVCSSC(R)
The release of the affricate syllabifies, and the following stop or affricate onset geminates. In some cases, the gemination may instead occur on the syllabic resonant. I'm not sure yet what conditions determine this, but it may well have something to do with stress.
mı̨ı̨tš + pé → mı̨ı̨tšppé [mɤ̝ːtʃ̩ppʰɛ] → mı̨ı̨tššpé [mɤ̝ːtʃ̩ːpʰɛ]
Note: I'm actually not entirely decided about this.

As we can see in rules 5 and 6, a preceding long vowel prevents an affricate from leniting to the corresponding resonant. We will see that the same principle applies to stops preceded by a long vowel.

Following a short vowel, the alveolar stop /t/ assimilates to a following consonant and geminates it. If the following consonant is an affricate the onset is geminated.

7. Short vowel + t + consonant
CVtC → CVCC
Examples:
tsı̨t + nı̨naq → tsı̨nnı̨naq - mouse
óyat + gotṣ → óyaggotṣ - to mix up, confuse
alı̨t + kı̨t → alı̨kkı̨t - salmon

If the stop is /p k/ things become a bit more complicated. If the following consonant is a resonant, the stop assimilates to it but a glide /w j/ is inserted before the resonant. This causes the preceding vowel to change, often resulting in a diphthong.

8. Short vowel + p + resonant
CVpR → CVwRR
Using s as an example of a following resonant, the actual realisations are as follows:
{ap op oap oep} + s → ouss
{ep ıp up uıp} + s → uss
{ęp ı̨p} + s → ı̨ss
Examples:
kƛoap + neek → kƛounneek - west
yıp + súyutł → yussúyutł - honour, excellence

9. Short vowel + k + resonant
CVkR → CVjRR
Using ł as an example of a following resonant, the actual realisations are as follows:
{ak ek} + ł → eıłł
ık + ł → ıłł
{ok oak oek uk uık} + ł → uıłł
{ęk ı̨k} + ł → ı̨ıłł
Examples:
tnósok + nı → tnósuınnı - (meaning unknown)
kı̨́k + mat → kı̨́ímmat - spring (the season)

The sequences ap ak become au aı before a stop or affricate

10. Cap Cak → Cau Caı / _C
Examples:
sak + tá → saıtá - now
áp + tran → áútran - permission to enter a house

The stops /p t k/ do not assimilate:
- after a long vowel
- before a stop/affricate after a short vowel other than /a/

In those cases /k/ lenites to [x] regardless of whether it is the first or second consonant in the resulting cluster.

Examples:
aap + tan → aaptan
ek + tsáł → extsáł
tół + kıt → tółxıt - owl

As outlined in the phonology, there is only one uvular phoneme: /q/. However, it is rather a common one, especially as many nouns end in -q. Such nouns actually have their own special rule.

11. Short vowel + q + consonant
CVqC → CVaC
When a noun stem ending in aq ıq ı̨q uq is followed by a consonant, the q elides and an epenthetic a is inserted in its place.
Examples:
artłáq + tí → artłáátí - seal
sı̨nı̨q + pıtł → sı̨nı̨apıtł - small beaver
ęruq + sółot → ęruasółot - birch forest

This rule may apply in other cases as well, but I'm not sure yet.

In other cases, the /q/ lenites to /ʁ/ r̊ instead.

12. Short vowel + q + consonant
CVqC → CVr̊C
Note that unlike in most other clusters starting in a resonant, no excrescent stop is inserted.
tłáq + łı̨ → tłar̊łı̨ - to hang down
qéq + nıt → qér̊nıt - yesterday (usually found as qér̊nınná)

As with other stops, /q/ does not lenite after a long vowel.

The Glottal Stop

I have agonised much over whether to include the glottal stop in TLFKAT phonology, and so far it only features in a handful of words, but let's say it is in for the time being.

Like the uvular stop, the glottal stop can occur in a cluster with a following consonant without the insertion of an excrescent stop.

If the preceding vowel is short, the construction of the glottis moves partly on to the vowel resulting in creaky voice. For some speakers, the glottal stop may even become [ʕ]. For example:

teʼ + na → teʼna [tḛʔna~tḛʕna]

After a long vowel, or as the second consonant in a cluster, the glottal stop behaves normally.

Quick Synopsis of Stop Allophony

As we have seen, stops are most susceptible to change after a short vowel before another consonant. In the case of the more front stops this results mainly in elision and assimilation to the following consonant, whereas the more back consonants fricativise and in the case of the uvular and glottal stops become voiced.

///////

In addition to the processes described above, I'm considering vowel reduction processes that may reduce sequences of long vowels and possibly elide short vowels. I don't have anything concrete yet though.

That only leaves the sandhi rules for complex consonant clusters, which I will cover in my next post. In the meantime, I would appreciate it if someone knowledgeable would comment on the viability of these processes, especially the last one whereby the "voicing" of the glottal stop causes creaky voice, as phonation is not my forte.

[WeepingElf]

WOW! This is one of the coolest conlang phonologies I have ever seen. I like the way how nasality is treated like an extra POA, affricate-wise, in the alveolars. Haven't seen that before.

In short: ROCK'N'ROLL!!!

[2+3 clusivity]

Agreed. Your presentation of the phonology is very well thought and set out. I will look forward to seeing the grammar.

[DesEsseintes]

Cheers, Weeping Elf and 2+3!

I'm afraid there won't be much on grammar quite yet, but I'll be posting more on sandhi soon.

[DesEsseintes]

3. Allophony and Morphophonological Processes - Part II

Here I'm going to introduce my favourite morphophonological process in TLFKAT so far.

Sandhi in Complex Consonant Clusters

Some morphemes, especially roots, end in consonant clusters of resonant + stop:

{s n l ł r} + {p t k}

When such clusters are followed by a consonant, metathesis occurs and the stop is brought before the resonant where it elides and if /p k/ causes the same vowel mutations as introduced in rules 8 and 9 in post 2.

Let's take the root łayalp - wing; take flight as an example. To form the noun wing from the root, we need to append the nominaliser suffix -tsíł, so let's see what happens:

łayalp + tsíł → łayapl + tsíł → łayaplttsíl → łayoulttsíł

l and p metathesise, the onset of tsíł geminates because of the preceding resonant (rule 2) and p lenites to /w/ resulting in a diphthong (rule eight).

If the following consonant is a stop, it geminates:

yełp + tú → yepł + tú → yepłtú → yułłtú → yułttú - the force of a throw

I added a step here to try and explain what is transpiring. Note that in the next to last step I wrote the resonant double. This is in keeping with rules 7, 8 and 9 which tell us that a stop assimilates to and geminates a following resonant. This didn't matter in the łayalp example above, as two geminate consonants cannot occur in a row in TLFKAT and the onset of the following affricate was already geminate. However, in this second example the gemination of the resonant shifts onto the following stop.

Some examples with /k/ (notice that it surfaces as x as it's part of a cluster):

sołx (<*sołk) + pappıtł → soxł + pappıtł → soxłpappıtł → suıłłpappıtł → suıłtpappıtł - dusk, twilight
kasx (<*kask) + súyutł → kaxs + súyutł → kaxstsúyutł → keıstsúyutł - intelligence

Note that in the last step of the first example, the resonant geminates the following stop, but the resulting łpp dissimilates to łtp as TLFKAT prefers clusters of the type R₁C₁C₂(R₂) if there is a resonant at the start of the cluster. (The numbers 1 and 2 indicate PoAs.)

If the cluster ends in t and is followed by an affricate, the t seems to simply disappear:

rałt + tsáł → ratł + tsáł → ratłttsáł → rałttsáł - (woven) cloth

So far, I haven't decided whether there will also be morpheme-final clusters of the type {s n l ł r} + q.

That is most of what I have as of yet on morphophonology.