Page 12 of 13

Re: Random phonology/phonemic inventory thread

Posted: 19 Nov 2020 20:02
by this_is_an_account
I haven't posted here in a bit, here's a phono.

/m mʲ n nʲ ŋ/
/p pʲ t tʲ d dʲ ɖ k kʲ ɡ ɡʲ ʔ ʔʲ/
/ɓ ɓʲ tʼ tʲʼ kʼ kʲʼ/
/v vʲ s sʲ z zʲ ʂ h hʲ/
/tʂ dʐ/
/tsʼ tsʲʼ tʂʼ/
/r rʲ l lʲ j w/
/i e o a/

The syllable structure is (C)V(C), where the initial consonant cant be /ŋ/ and the final consonant is one of /m n ŋ/. Palatalized consonants only appear before /i e/. In some areas, /v vʲ/ are [ɦ ɦʲ].

Re: Random phonology/phonemic inventory thread

Posted: 22 Nov 2020 03:21
by eldin raigmore
Phoneme inventory
All 13 consonants are pulmonic egressive.
By manner of articulation:
Plosives: /p b t k g/
Nasals: /m n N/ (/N/ is the voiced velar nasal. IPA notation would be engma, but it’s hard for me to type that.)
Fricatives: /s h/
Approximants: /w j/ (/w/ is the voiced labial-velar approximant.)
Lateral approximant: /l/

By place of articulation:
Bilabial: /p b m w/ (/w/ is co-articulated labial-velar approximant)
Alveolar: /t n s l/
Palatal: /j/
Velar: /k g N/ (I’m writing the voiced velar nasal as /N/)
Glottal: /h/

Close to open:
Close: /i u/
Mid: /e o/
Open: /a/

Front to back:
Front unrounded: /i e/
Central unrounded: /a/
Back rounded: /u o/

I use C to denote any consonant, M to denote any nasal, K to denote any stop or /s/, J to denote any glide or liquid, V to denote any vowel.
The nasals /m n N/, and only the nasals, can appear in codas; and they can’t appear anywhere else.
The other ten consonants can all appear in onsets, but not codas. They can all be one-consonant onsets.
The stops and one of the fricatives /p b t k g s/ can be the first consonant of an onset-cluster. No other consonant can be the first consonant of an onset cluster, and none of those six can appear anywhere else in an onset-cluster.
The glides and the liquid /j w l/ can be the last consonant of an onset cluster. No other consonant can be the last consonant in an onset cluster, and none of those three can occur elsewhere in an onset cluster.
The glottal fricative /h/ can only be a one-consonant onset. It cannot occur in a cluster nor a coda.

Phonotactics and syllable-structure
CV(M) or KJV(M)
Mandatory onsets can be one or two consonants
Optional codas can be one consonant
Nuclei can be one vowel
10 consonants K+J+H can be one-consonant onsets but can never be in codas
6 consonants K can be first consonants of onset-clusters, or only consonants of simple onsets, but can never be in codas and can never be last consonants of clusters
3 consonants J can be last consonants of onset clusters, or only consonants of simple onsets, but can never be in codas and can never be first consonants of clusters.
1 consonant /h/ can only be a single-consonant onset
3 consonants M can only be single-consonant codas

6*3+10 = 18+10 = 28 onsets
5*3+5 = 15+5 = 20 rimes
28*20 = 560 syllables
10*5 = 50 CV syllables
6*3*5 = 90 CCV syllables
10*5*3 = 150 CVC syllables
6*3*5*3 = 18*15 = 270 CCVC syllables

Phonological shapes of morphemes.
No morpheme can contain more than one copy of any single phoneme.
I’m going to concentrate on morphemes with two or fewer syllables. I leave aside 3-syllable morphemes for now.
Affixes and particles will be four phonemes or fewer.

All morphemes with four or fewer phonemes will be affixes and particles.
They’ll include all one-syllable morphemes, of which there could be up to 560.
They’ll also include all CV.CV two-syllable morphemes, of which there could be up to
10*5 * 9*4 = 50*36 = 1800.
So 1800+560 = 2360 affixes and particles.
English reportedly has around 800 affixes. I don’t know how many particles English has.

The most common, most frequently-used roots, will be 5-phoneme bisyllables.
They’ll consist of
6*3*5 * 8*4 = 90*32 = 2880 CCV.CV at a max
8*5 * 6*3*4 = 40*72 = 2880 CV.CCV at a max
10*5*3 * 9*4 = 150*36 = 5400 CVC.CV at a max
10*5 * 9*4*3 = 50*108 = 5400 CV.CVC at a max
5-phoneme 2-syllable morphemes, for a total of up to 16,560 roots.
That’s more than three times as many as needed for most everyday conversations.

6-phoneme 2-syllable root morphemes could have the following shapes.
CCVC.CV 6*3*5*3 * 8*4 = 270*32 = 8,640
CV.CCVC 8*5 * 6*3*4*3 = 40*216 = 8,640
CCV.CVC 6*3*5 * 8*4*3 = 90*96 = 8,640
CVC.CCV 8*5*3 * 6*3*4 = 120*72 = 8,640
CVC.CVC 10*5*3 * 9*4*2 = 150*72 = 10,800
CCV.CCV 6*3*5 * 5*2*4 = 90*40 = 3,600
for a total of up to 48,960 roots.

So far that’s 65,520 possible roots. That’s a few more than needed for most expert conversations.

7-phoneme 2-syllable roots could have the following shapes.
CCVC.CCV 6*3*5*3 * 5*2*4 = 270*40 = 10,800
CCV.CCVC 6*3*5 * 5*2*4*3 = 90*120 = 10,800
CCVC.CVC 6*3*5*3 * 8*4*2 = 270*64 = 17,280
CVC.CCVC 8*5*3 * 6*3*4*2 = 120*144 = 17,280
for a possible total of 56,160 roots.

So far that’s 121,680 roots.

8-phoneme 2-syllable roots, if there are any, have only one possible shape:
CCVC.CCVC 6*3*5*3 * 5*2*4*2 = 18*15 * 10*8 = 270*80 = 21,600 possible roots of this shape.

That’s now a possible 143,280 roots; still a few myriads fewer than the OED says English has.


Time to consider 3-syllable roots?

3-syllable roots, if there are any, could have 6 to 12 phonemes.
I’m going to arbitrarily forbid morphemes longer than 11 phonemes.

Let’s start with 6-phoneme trisyllables.
They’ll all have one shape.
CV.CV.CV 10*5 * 9*4 * 8*3 = 50*36*24 = 43,200 possible roots.
That gives us 65,520+43,200 = 108,720 possible roots with 6 or fewer phonemes in 3 or fewer syllables.

So far we have:
16,560 possible 5-phoneme 2-syllable roots
48,960 possible 6-phoneme 2-syllable roots
43,200 possible 6-phoneme 3-syllable roots
56,160 possible 7-phoneme 2-syllable roots
164,880 possible roots so far.

The OED says English has about 177,000 roots.
If I want to beat that I may need to go further.


If I want to go further I have a choice. I can use the 8-phoneme 2-syllable root morphemes to get 164,880 + 21,600 = 186,480 to exceed English’s (about) 177,000 root-inventory;

Or I can instead consider 7-phoneme 3-syllable roots.

7-phoneme 3-syllable roots, if there are any, have the following six possible shapes:
CVC.CV.CV 10*5*3 * 9*4 * 8*3 = 150*36*24 = 129,600
CV.CVC.CV 10*5 * 9*4*3 * 8*3 = 50*108*24 = 129,600
CV.CV.CVC 10*5 * 9*4 * 8*3*3 = 50*36*72 = 129,600
CCV.CV.CV 6*3*5 * 8*4 * 7*3 = 90*32*21 = 60,480
CV.CCV.CV 8*5 * 6*3*4 * 7*3 = 40*72*21 = 60,480
CV.CV.CCV 8*5 * 7*4 * 6*3*3 = 40*28*54 = 60,480
for a total of as many as 570,240 possible roots.
164,880 + 570,240 is 735,120 possible roots of 7 or fewer phonemes and 3 or fewer syllables. More than 4 times as many roots as English has.

If I include both the 5-to-8-phoneme 2-syllable possibilities and the 6-or-7-phoneme 3-syllable possibilities I have
767,720 possible roots; about 4-and-1/3 as many roots as English actually has.

I don’t believe I need to consider any 8-or-more-phoneme 3-syllable root possibilities!
I’m approaching a million possible phonetic combinations, and I think most grammatical utterances with 8 or more phonemes and 3 or more syllables are likelier to be full words than morphemes; at least in this conlang. Admittedly that’s just a feeling!


The phoneme inventory might be unnaturalistic.
Given the phoneme inventory, the phonotactics is naturalistic enough.
The requirement that no phoneme occur twice in the same morpheme is naturalistic enough, though nothing like universal.
The idea that every morpheme consist of an entire syllable or entire syllables, may well be unnaturalistic and maybe even unrealistic.

Anyway; between any two repetitions of the same phoneme must come a syllable boundary (which is naturalistic and pretty common) and a morpheme boundary (also naturalistic among languages satisfying the equivalent requirement that no phoneme occur twice in the same morpheme).
Also, every morpheme boundary is also a syllable boundary (since syllables never span morpheme-boundaries).

I haven’t included a way to detect word-boundaries.

Re: Random phonology/phonemic inventory thread

Posted: 22 Nov 2020 19:01
by Vlürch
DesEsseintes wrote: 18 Nov 2020 17:20This is almost exactly the kind of vowel system I like to use at the moment.
That kind of vowel inventories are nice, but in practice I personally can't handle speaking anything that distinguishes /ɛ ɔ/ from /e o/ since I can't pronounce either pair consistently except in isolation (and not even then always) and they just come out as [e̞ o̞], or /ɛ/ as [æ] and /e o/ as [i u​] if I try to pronounce them as distinctly as possible haha. [>_<]

Something meant to be kinda similar to the languages of India, but more "minimalistic":

/m n̪ ɳ [ŋ]/ <m n ṇ (n)>
/p t̪ ʈ k/ <p t ṭ k>
/ʂ/ <ṣ>
/ʋ θ ɻ j x/ <v s ṛ y h>
/ɾ/ <r>
/l̪ ɭ/ <l ḷ>

/a e i o u [ə]/ <a e i o u (Ø)>
/aː eː iː oː uː/ <ā ē ī ō ū>

/p t̪ ʈ k/ are voiced in phonemic intervocalical and post-nasal positions and aspirated in open syllables, including when voiced. In clusters after liquids, they're voiced but not aspirated.

[ə] occurs word-initially when a word begins with /ɳ ʈ ʂ ɻ ɾ ɭ/ (and isn't preceded by a word that ends in a vowel or one of the phonetically allowed cluster consonants), between consonants to break up most clusters, and word-finally after /p t̪ ʈ k ɾ/; these effects occur across word boundaries. Since its presence or absence is entirely predictable, it's not written. Its exact realisation is [æ~ɐ~ə~ɘ~ɨ~ɯ~ɤ~ʌ] in free variation.

[ŋ] only occurs as an allophone of /n̪/ before /k x/ and word-finally when the following word begins with a vowel, or utterance-finally.

Nasal consonants cause nasalisation on the preceding vowel in closed syllables.

rūk tari ṣopnā heṛṭa ompīkin
/ɾuːk t̪aɾi ʂopɳaː xeɻʈa ompiːkin/
[əɾuːkə t̪ʰa̠ɾi ʂopəɳa̠ː xe̞ɻɖa̠ õ̞mbʱiːgĩŋ]

Re: Random phonology/phonemic inventory thread

Posted: 29 Nov 2020 01:23
by Shemtov
Started as Nahuatl meets Indo-Aryan:
/p t̪ t̪ʰ t̪͡s̪ t̪͡s̪ʰ t̪͡ɬ̪ t̪͡ɬ̪ʰ ʈ ʈʰ ʈ͡ʂ ʈ͡ʂʰ ʈ͡ꞎ ʈ͡ꞎʰ t͡ʃ t͡ʃʰ k kʰ kʷ kʷʰ ʔ~h/ <p t tʿ tz tzʿ tl tlʿ ṭ ṭʿ ts tsʿ ṭl ṭlʿch chʿ c cʿ cu~uc cʿu~ucʿ h>
/f s ʂ ʃ/ <f z s x>
/m n/ <m n>
/l ɭ/ <l ḷ>
/j w/ <y hu~uh>

/ i o a/ <i o a>
/i: o: a:/ <ī ō ā>
/ɛi ai oi aʊ/ <e ai oi ao>
/a:ɪ o:ɪ/ <āi ōi>
/ɐi: ɔi:/ <aī oī>


The glottal stop is realized as [h] before another stop.

Re: Random phonology/phonemic inventory thread

Posted: 29 Nov 2020 20:44
by Shemtov
The above, with some changes:
/p t̪ t̪ʰ t̪͡s̪ t̪͡s̪ʰ t̪͡ɬ̪ t̪͡ɬ̪ʰ ʈ ʈʰ ʈ͡ʂ ʈ͡ʂʰ ʈ͡ꞎ ʈ͡ꞎʰ t͡ʃ t͡ʃʰ k kʰ kʷ kʷʰ ʔ~h/ <p t tʿ tz tzʿ tl tlʿ ṭ ṭʿ ṭẓ ṭẓʿ ṭl ṭlʿch chʿ c cʿ cu~uc cʿu~ucʿ h>
/f s ʂ ʃ x xʷ/ <f z ẓ x j ju~uj>
/m n/ <m n>
/l ɭ/ <l ḷ>
/j w/ <y hu~uh>

/ i o ɛ a/ <i o e a>
/i: o: e: a:/ <ī ō ē ā>
/ɛi ai oi ao ɛo/ <ei ai oi ao eo>
/a:ɪ o:ɪ e:ɪ a:o e:o/ <āi ōi ēi āo ēo>
/ɐi: ɔi: ɛi: ɐo: ɛo: / <aī oī eī aō eō>

The glottal stop is realized as [h] before another stop.

Re: Random phonology/phonemic inventory thread

Posted: 02 Dec 2020 21:14
by Vlürch
Just something random with potential to get tongue-twistery:

/m n ŋ/ <m n ng>
/p t k/ <b d g>
/pʰ tʰ kʰ/ <p t k>
/t͡s t͡ʃ/ <dz j>
/t͡sʰ t͡ʃʰ/ <ts c>
/s ʃ/ <s x>
/ʋ j h/ <v y h>
/r l/ <r l>

/a~r̩~ɬ̩~s̩/ <a>
/i~r̩ʲ~ɬ̩ʲ~ɕ̩/ <i>
/u~r̩ʷ~ɬ̩ʷ~s̩ʷ/ <u>

<a i u> are [a̠ i u] after /pʰ tʰ kʰ t͡sʰ t͡ʃʰ r l/ and word-finally
<a i u> are [s̩ ɕ̩ s̩ʷ] after /t͡s s/ before voiceless consonants
<a i u> are [r̩ r̩ʲ r̩ʷ] after other consonants before voiced consonants
<a i u> are [ɬ̩ ɬ̩ʲ ɬ̩ʷ] after other consonants before voiceless consonants

/r/ is [ɾ] intervocalically when preceded and followed by <a i u> as [a̠ i u]
/p t k t͡s t͡ʃ s ʃ/ are voiced intervocalically when followed by <a i u> as [a̠ i u] or [r̩ r̩ʲ r̩ʷ]
/ʋ/ is devoiced to [f] before <a i u> as [ɬ̩ ɬ̩ʲ ɬ̩ʷ]
All consonants are palatalised before /i/; /t͡s t͡ʃ t͡sʰ t͡ʃʰ s ʃ/ merge as [t͡ɕ t͡ɕ t͡ɕʰ t͡ɕʰ ɕ ɕ]

pagura adixung vasbul sigxa
[pʰa̠guɾa̠ ɬ̩tʲɬ̩ʲʒr̩ʷŋ fɬ̩spr̩ʷl ɕɕ̩kʃa̠]

Re: Random phonology/phonemic inventory thread

Posted: 06 Dec 2020 05:18
by Porphyrogenitos
Cross between Rotokas and Maxakali?

/m n ɲ ŋ/
/p t c k/

/a e i o u/

Maximal syllable structure is CVC. Stress is phonemic. (One source of phonemic stress is coalescence of auxiliaries to produce new verbal inflections, like the Spanish future and conditional affixes.)

The palatals historically originated from [Cj] clusters, which in turn originated from /iV/ and /eV/ in hiatus following a consonant (as in Latin → Romance). Some dialects may retain labial+j clusters (or palatalized labials, if you wish to analyze them that way).

Phonological rules:
  • Nasals assimilate to the place of the following consonant.
  • Stops are voiced after nasals. Thus: /Np Nt Nc Nk/ → [mb nd ɲɟ ŋg]
  • Postvocalic stops are lenited, varying between true fricatives and approximants: [β~β̞ z~ɹ ʝ~j ɣ~ɰ]
  • But what of word-final stops? I am not sure if they will be lenited and voiceless [ɸ s ç x] or actual stops, perhaps unreleased and tending towards glottal stop. This could be a source of dialectal variation.
  • Also, I know that two adjacent identical stops will be an unvoiced geminate.
  • But what of historic clusters of non-homorganic stops, e.g. /pk/ and /tp/? I am not sure if stops will assimilate to a following stop, forming a geminate, or if the first stop will be lenited. I kind of want to do a thing like Latin debitum → Spanish deuda. Perhaps that can be saved for a later stage of the language, after some syncope takes place.
  • Another thing I am considering is some form of consonant gradation, where a singleton stop or nasal becomes geminate under Finnic-type conditions.
  • Lastly, as for stop-nasal clusters...hmm, probably they will just become geminate nasals. Maybe some dialects will preserve them as non-homorganic nasal clusters, cf. PIE *swepnos → Latin somnus.
Nonsense words:


Re: Random phonology/phonemic inventory thread

Posted: 06 Dec 2020 21:47
by eldin raigmore
Phoneme Inventory
12 phonemes.

9 consonants
/m k j p w n t l s/

3 vowels
/i u a/

These are all the phonemes that are contained in either more than two-thirds of the inventories in UPSID’s sample,
or in more than two-thirds of the inventories in PHOIBLE’s sample.

So for any three of them, there is at least one language in one sample or the other that contains all three.

(Actually, on the average, it’s likely that about 29.63% of the inventories in one sample or the other contain all three of any three of these phonemes.)

Syllable Shape
The maximal syllable shape is (C)V(C).

Phonotactics and Morpho-phonemics
The consonant /s/ cannot occur in an onset; only in a coda.
So there are 27 possible syllable bodies; /i u a/ and /m k j p w n t l/+/i u a/.

The consonant /l/ cannot occur in a coda; only in an onset.
So there are 27 possible syllable rimes; /i u a/ and /i u a/+/m k j p w n t s/.

Every morpheme begins with a consonant.
The syllable-body /ta/ cannot be the first two phonemes of a morpheme.
So 23 syllable-bodies can be the first two phonemes of a morpheme.

The syllable-rime /un/ cannot be the last two phonemes of a morpheme.
So 26 syllable-rimes can end a morpheme; a single vowel, or a vowel plus a consonant other than /l/, except not /un/.

Bisyllabic Morphemes
I’m very interested in the bisyllabic morphemes.
A lot of what I’ve said so far, and a lot of what I’ll say next, applies to monosyllabic morphemes and trisyllabic morphemes as well as to bisyllabic morphemes; but I’ll consider the effects on bisyllabic morphemes first.

There can’t be two consecutive vowels in a morpheme; so, no hiatus between an open syllable and a following vowel-initial syllable.

So bisyllabic morphemes can be shaped like

The Maximum Onset Principle applies to syllabification.
That is, if a consonant could be interpreted as either (part or) all of the coda of an earlier syllable,
or (part or) all of the onset of a later syllable,
then it will be interpreted as (part or all of) the onset of the later syllable.

So the only CVC.V(C) morphemes are CVs.V(C) morphemes.
If that central intervocalic consonant is not an /s/, the morpheme is shaped like CV.CV(C).

There are no “geminated” consonants internal to any morpheme.
If a morpheme contains a closed syllable followed by a syllable with an onset, the coda of the earlier syllable cannot be the same as the onset of the later syllable.

Furthermore the consonant-cluster /pw/ is prohibited morpheme-internally.
So the intervocalic syllable-boundary-straddling two-consonant cluster in a CVC.CV(C) morpheme can be one of
1*1 + 1*7 + 7*1 + (7*6-1) = 1+7+7+(42-1) = 15+41 = 56 pairs.
Adding in the 9 single consonants, the number of consonants or two-consonant clusters that can occur between the vowels of two consecutive syllables in a morpheme, is 65 = 56+9.

So the number of possible two-syllable morphemes is
23 CV morpheme-initial syllable bodies
multiplied by
65 C or CC intervocalic consonantal sequences
multiplied by
26 V(C) morpheme-final syllable-rimes.

So 23*65*26 = 38,870 bisyllables might be morphemes.

One- and Three-Syllable Morphemes

There are 8*8=64 CiC syllables, 8*7=56 CuC syllables, 7*8=56 CaC syllables, 8 each Ci and Cu syllables and 7 Ca syllables.
So 64+56+56+8+8+7 = 72+64+63 = 199 single syllables can be morphemes.
There are also 8 each iC and aC syllables and 7 uC syllables; as well as 3 V syllables; but none of those 26 syllables can be morphemes.

So 39,069 one-or-two-syllable sequences might be morphemes.
That’s right in the middle of the 30,000-to-50,000 range of lexicon sizes for carrying on expert-level conversations.

23*65*3*65*26 = 7,579,650 three-syllable sequences have so far not been ruled out as possible morphemes. I don’t know that any of them will be necessary. Their number could be drastically reduced by prohibiting the same phoneme from being used more than once in any three-syllable morpheme. Perhaps there still might be 3.4 to 3.5 times as many of them as two-or-fewer-syllable morphemes. At any rate there’d be at least 70,560 possible CVC.CVC.CVC morphemes even with that restriction.

All roots are free morphemes.
All free morphemes are either roots or particles.
Particles can’t take inflection nor derivation; they are mandatorily one-morpheme words.
All derivation and inflection is by means of affixes.
All affixes are either prefixes or suffixes.

The morphology is a position-class morphology.
There are at most about a hundred parts-of-speech, of which at most about twenty-five are “major”, large, open word-classes.

There are also at most about a hundred classes of affixes, of which at most twenty-five are ‘major”, large, open classes, that can be applied to word-bases of more than one part-of-speech.

For any given part-of-speech there is a “template” —— a string of morpheme-classes —— laying out which classes of morphemes can appear in such a word and in which order.
One of them — the root — is mandatory.
The rest are optional.
In each position, at most one (or, in the case of the root, exactly one) morpheme of that class can be included.


That’s all so far.
What do you think?

In certain cases it will be unambiguously obvious, to speakers of the language, where the morpheme-boundaries go.
Every morpheme-boundary will also be a syllable-boundary, but not vice-versa.


The point of this experimental engelang is to prove that a 9-consonant 3-vowel phoneme-inventory with (C)V(C) syllables can have a pretty big lexicon without enormously lengthy morphemes, even with somewhat restricted phonotactics.
If the word-templates aren’t too long there can be lots of words, too, even without lengthy words.

Re: Random phonology/phonemic inventory thread

Posted: 06 Dec 2020 22:12
by LinguoFranco
/m n ŋ/
/p t (ʈ) k (kʷ) q/
/s (ʂ) ʃ~ɬ h~ɦ x/
/l r j w~ʋ/

/ɪ~i iː ʊ~u u:/
/a aː/

The most complex syllables permitted are CVC and CVV. There is a simple pitch accent system.

The phonemes in parentheses are not yet a part of the inventory, but I'm thinking about adding them anyway.

/x/ is supposed to be the voiceless uvular fricative, but it keeps showing up as the voiceless velar fricative, instead.

The only major alteration I might make to this inventory is either removing /r/ or just expanding the rhotic phonemes by adding either a retroflex or a uvular rhotic.

Re: Random phonology/phonemic inventory thread

Posted: 10 Dec 2020 00:15
by this_is_an_account
/m n ŋ/ m n g
/b t d dʑ ɖ k ʔ/ p~b t~c d j ṭ~ḍ k q
/f s ʂ h/ f s~x ṣ h
/ʋ l/ v l
/i u e o a/ i u e o a
/ju jo/ iu io

A syllable can begin with any consonant, or any consonant + /ʋ l/ (including clusters of /ʋl lʋ/). It can also begin with a cluster of any two plosives of the set /b t ɖ k/ (except for two of the same consonant, or /tɖ ɖt/), which can also be followed by /ʋ l/. A syllable can end in any consonant of the set /m n ŋ s h l/.
The syllabe structure is then this:
Where C is any consonant, P is one of /b t ɖ k/, and N is one of /m n ŋ s h l/.

The initial plosive clusters are pretty interesting in that they are always voiceless, so /bt tb/ are pronounced [pt tp]. Even when both components are phonemically voiced, they become voiceless, so /bɖ ɖb/ [pʈ ʈp]. These voiceless allophones of /b ɖ/ are written <p ṭ>

/dʑ/ is an interesting phoneme. /t s/ become [tɕ ɕ] before /i ju jo/ (when they are written <c x>), and /dʑ/ only ever appears before those sounds, so it might seem like its part of the same set of allophones, but /d/ also appears before those vowels.

Re: Random phonology/phonemic inventory thread

Posted: 12 Dec 2020 16:58
by eldin raigmore
A thought-experiment.
Imagine a language fitting the following description.
To keep it simple I’ll say it’s analytic and isolating.
I want to find out how many CVCVCVC trisyllabic morphemes/words it could have with the following restrictions.
(These are phonotactic rules and morphophonemic rules.)

No phoneme may appear twice in such a string of phonemes.
To talk about where the phonemes go I’ll write
C1 V1 C2 V2 C3 V3 C4
Let’s assume the following.
There are 12 consonants and 6 vowels.

To make it easy the 6 vowels will be PHOIBLE’s 6 commonest vowels: /i u a e o ɛ/
Front close unrounded /i/, front close-mid unrounded /e/, front open-mid unrounded /ɛ/;
Central open unrounded /a/;
Back close rounded /u/, back close-mid rounded /o/.
Or to list them in a different order;
Close front unrounded /i/, close back rounded /u/;
Close-mid front unrounded /e/, close-mid back rounded /o/;
Open-mid front unrounded /ɛ/;
Open central unrounded /a/.
The vowels are divided into three classes as follows.
A): Two /i ɛ/ can be V1 or V2 but not V3;
E): Two /u o/ can be V1 or V3 but not V2;
I): Two /a e/ can be V2 or V3 but not V1.
So 4 vowels (classes A and E) can be V1; 4 vowels (classes A and I) can be V2; and 4 vowels (classes E and I) can be V3.

To make it easy the 12 consonants will just be the 12 commonest consonants in PHOIBLE; /m k j p w n t l s b ŋ g/
perhaps modified if some pair of them never occur together in any natlang’s inventory.
By manner of articulation:
Plosives: /p b t k g/
Nasals: /m n ŋ/
Fricative: /s/
Approximants: /w j/
Lateral approximant: /l/
By place of articulation:
Bilabials: /p b m w/
Alveolars: /t n s l/
Palatal: /j/
Velars: /k g ŋ/
The consonants are divided into six classes as follows.
B): Three /k j b/ can be C1 or C2 or C3 but not C4: and
C): three /m g ŋ/ can be C2 or C3 or C4 but not C1.

D): Two /s p/ can be C1 or C4 but not C2 nor C3: and
F): two /w l/ can be C2 or C3 but not C1 nor C2.

G): One /t/ can be C1 but not C2 nor C3 nor C4: and
H): one /n/ can be C4 but not C1 nor C2 nor C3.

So 6 consonants (classes B D and G) can be C1; 6 consonants (classes C D and H) can be C4; and 8 consonants (classes B C and F) can be C2 or C3.


How many trisyllabic strings of four distinct consonants and three distinct vowels, with no clusters (neither consonant-clusters nor vowel-clusters), could be formed satisfying all these phonotactic rules?


The calculation is too long to list here.
But it turns out there are 1418 sequences of four consonants allowed by those rules; each of which can be interpolated by any of the 40 sequences of three vowels allowed by those rules; to yield 56,720 CVCVCVC sequences allowed by those rules.

If these are morphemes or words, there will frequently be cues about where the boundaries are.

Re: Random phonology/phonemic inventory thread

Posted: 22 Dec 2020 23:25
by this_is_an_account
/m n/ m n
/b t d k q ʔ/ b t d k q '
/θ s z x~χ ħ h/ c s z j x h
/r l j w/ r l y w
/i iː u uː e ə o õ õː ã ãː r̩/ i ii u uu e a o ǫ ǫǫ ą ąą r
The syllable structure is CV(C). The only allowed coda consonants are /t k q ʔ r/ and the only vowels that can have a coda are /e ə/.

Re: Random phonology/phonemic inventory thread

Posted: 19 Jan 2021 18:15
by Frislander
/p t k/
/f s x/
/m n/
/w r j/

/i u/
/e o/

The coronal consonants /t s n/ are palatalised to /t͡ɕ ɕ ɲ/ before the high front vowel /i/ and in coda position.

Syllable structure is (C)V(C), where vowel-initial syllables are restricted to word-initial position. Any consonant may occur in the coda phonetically, however the voiceless fricatives /f [ɕ] x/ and the sonorants /w j r/ are in complementary distribution in this position, with the former occurring before voiceless consonants and the latter occurring elsewhere. For example, /akaruj/ "fish own" plus /-kin/ "animate plural" gives [akaɾuɕkiɲ] "fish owls", while /kemer/ "to weep" plus /-fa/ "First person singular perfective" gives [kemexfa] "I wept".

Re: Random phonology/phonemic inventory thread

Posted: 07 Feb 2021 09:24
by Omzinesý
p t k ʔ
f θ s x h
m n ŋ

i i᷈ u u᷈
e e᷈ o o᷈
a a᷈

There could also be some kind of a glottalized/creaky voice vowel quality.

CV syllables
V can appear word-initially but /ɾ/ cannot.

Makes 130 syllables.

Re: Random phonology/phonemic inventory thread

Posted: 07 Feb 2021 14:55
by Davush
I recently feel like making a maximally a-priori "SAE" language (phonology and grammar) which, ironically, is outside of my comfort zone.

I am playing with this inventory:

/p t k/
/b~v d~ð ɡ~ɣ/
/m n/
/ts tʃ/
/f v s z ʃ ʒ x/
/r r: l l:/

/a ɛ e i ɔ o u/ (possibly also with length)

I think this seems quite "SAE" in having a lot of fricatives with voicing distinction, affricates (I was considering also having /dz dZ/ but probably will have them as allophones of /z Z/) , and nothing further back than velar.
Voiced stops lenite intervocalically, which reminds me more specifically of Iberian, so it may or may not remain...

Re: Random phonology/phonemic inventory thread

Posted: 16 Feb 2021 14:20
by Omzinesý
p t k ʔ <p t k '>
f s θ x h <f s ŧ ꝁ h>
l j w <l j w>
m n <m n>
ɾ <r>

/l j w/ are devoiced to [ɬ ç ʍ] in coda.

(C)(C)V(C) Onset clusters with increasing sonority

i u <i u>
e o <e o>
ɛ ɔ <ê ô>
ä <a>, <â> word-finally

i, e => ɪ <i>
u, o => ʊ <u>
ɔ, ɛ, ä => ə <a>

Word-finally only open syllable with [ə] can be unstressed.

Tri/biconsonantal morphology

Re: Random phonology/phonemic inventory thread

Posted: 17 Feb 2021 07:13
by Nortaneous
Davush wrote: 07 Feb 2021 14:55 I recently feel like making a maximally a-priori "SAE" language (phonology and grammar) which, ironically, is outside of my comfort zone.
That got me thinking about a maximally SAE non-SAE phonology. If I've done everything correctly, which I almost certainly haven't, this should score 100 on that old SAE phonology test:

Code: Select all

     t̪ ts  
b bʷ   d    
f fʷ   s   x   h
v vʷ     ʑ    
m mʷ n̪ 
  ʙʷ   ɺ

i     u
iɛ yœ uɔ
e  ø  o
 ɛ a    ɑʴ(= ɹ̩)
Max syllable: FFCVCPF. There are probably some restrictions on which F - maybe the only FF cluster is xsCVCCx or something. (If so, there's voice assimilation.)
/mʷ/ can only occur in coda position, presumably for historical reasons. (or intervocalically, where you can call it coda)
The contrast between high-mid diphthongs and mid vowels is neutralized in unstressed syllables.
There are probably distributional limitations on the two labial series, or something. ʙʷ can only occur before high or high-mid rounded vowels, but is fully contrastive.

Example words:

Re: Random phonology/phonemic inventory thread

Posted: 23 Feb 2021 01:24
by Omzinesý
p t k ʔ
s x
m n ŋ
l ʟ

Only dentals and velars can appear in coda. Velars don't precede dentals or vice versa, though.
Plosives have voiced fricative allophones [β ð z ɣ ɦ] between vowels.
Plosives can be geminated.
I'm still unsure how /f, s/ behave.

Re: Random phonology/phonemic inventory thread

Posted: 28 Feb 2021 23:05
by Creyeditor
I think I will start posting my least developed conlangs here from time to time now. At least someone can have a look at them and they don't rot away in my files.

Here is a diachronic phonology sketch titled "dialect of the evil language". The idea is to take a language that is orc-ish-like and mangle it through sound changes in order to get something more euphonious.

The (unrealistic for a naturalistic conlang) segment inventory of the "evil language" for reference.
/ŋ ɴ/
/pf ts tʃ tʂ kx qχ/
/p t ʈ c k q ʔ/
/f θ s ʃ ʂ ç x χ h/
/j w/
/ɫ ʟ/

/ʏ ʊ/
/œ ɔ/

Syllable structure is complex, I guess.
Here is a list of the sound changes from the language to the dialect.

1. Simplification of the vowel system by neutralization and dipthongization
1.1 Loss of front rounded vowels
ʏ > i
œ > e
1.2 Loss of low back vowels
ɔ > o
ɑ > au
1.3 Loss of mid central vowel
ə > a
ɜ > eu
2. Simplification of consonant system
2.1 Deletion of glottals
h > ∅
ʔ > ∅
2.2 Place changes in sonorants
ŋ > m
ɴ > n
ʀ > r
ʟ > ʎ
ɫ > l
2.3 Voicing of fricatives I
f > v
θ > z
s > z
ʃ > ʒ
ʂ > ʒ
ç > j
x > ɣ
χ > ɣ
2.4 Deaffrication
pf > f
ts > s
tʃ > ʃ
tʂ > ʃ
kx > x > g
qχ > x > g
2.5 Fortition of glides
j > ʒ
w > b
2.6 Place changes in plosives
ʈ > t
c >tʃ
q > k
3. Simplification of syllable structure
3.1 Consonant Degemination
CC > C
3.2 Prothesis
∅ > e / #_CC
3.3 Consonantal Apocope
C > ∅ / C_#
3.4 Epenthesis
∅ > a / C_C
3.5 Coda deletion
C > ∅ /_$
4. Lenition
{p t k tʃ f s ʃ} > {b d g dʒ v z ʒ} / V_V

This gives us a segment inventory that looks like this.

Dialectal segment inventory
/m n/
/p t tʃ k/
/b (d) (dʒ) g/
/f s ʃ/
/v z ʒ/
/j w/
/l ʎ/

/i u/
/e o/
/eu au/
And here are some example cognates.
Language > dialect
/ɫʏkxəŋ/ > /liga/ 'to borrow'
/pʏʀçəŋ/ > /piraja/ 'house'
/sœɫp/ > /ze/ 'self'
/fœxŋə/ > /veɣana/ 'to sanctify'
/ɴɔqχəŋ/ > /noga/ `to maje'
/tɔkxə/ > /toga/ 'days'
/pɑʀkə/ > /paraka/ 'mountains'
/tsɫœdz/ > /esale/ 'son'
/ʔakxɫ/ > /aka/ 'white'

Re: Random phonology/phonemic inventory thread

Posted: 01 Mar 2021 03:04
by DesEsseintes
Creyeditor wrote: 28 Feb 2021 23:05 Language > dialect
/ɫʏkxəŋ/ > /liga/ 'to borrow'
/pʏʀçəŋ/ > /piraja/ 'house'
/sœɫp/ > /ze/ 'self'
/fœxŋə/ > /veɣana/ 'to sanctify'
/ɴɔqχəŋ/ > /noga/ `to maje'
/tɔkxə/ > /toga/ 'days'
/pɑʀkə/ > /paraka/ 'mountains'
/tsɫœdz/ > /esale/ 'son'
/ʔakxɫ/ > /aka/ 'white'
Interesting how it went from having lots of character to virtually none. [xD]