Conlang Random Thread

[Pabappa]

there are no perfect orthographies, because if there were such a thing, the language would never be able to evolve without ruining it. i would use your solution, though, with the velar glide standing for /Ø/, if you cant just leave it blank. korean does something similar i think.
edit: yeah, korean uses /ŋ/ for initial /Ø/, since there's no initial /ŋ/ it could conflict with. it doesnt need a solution for the coda because the vowel part seems to change size depending on if there's a final consonant or not.

[bradrn]

It seems that codas and vowel initial syllables both present significant problems for my featural abugida. The script assumes that every character signifies an onset with the nucleic vowel marked with diacritics. This means that vowel initial syllables would need a silent placeholder character to work in this system. The problem is determining what phonological features this placeholder has since that determines how the script would express it. Since the placeholder has no pronunciation, one can hardly assign it any place or manner features. Analogous problems arise with codas, which have no nucleus and no independent place features since they assimilate with following onsets like the moraic nasal and sokuon.

I have considered introducing new elements to indicate the absence of onset or nucleus, or using otherwise unused feature combinations (say velar + glide) as placeholders. Yet both options undermine the elegance and consistency of the original system in my opinion.

Is there any reason you can't break the featural system just for that one glyph, by introducing some unanalysable character representing null onset?

[zompist]

edit: yeah, korean uses /ŋ/ for initial /Ø/, since there's no initial /ŋ/ it could conflict with. it doesnt need a solution for the coda because the vowel part seems to change size depending on if there's a final consonant or not.

Synchronically this is true, but historically there were three different symbols based on ㅇ, for /Ø ʔ ŋ h/. The symbol represented the throat. The distinction between /Ø ʔ/ wasn't really phonemic in Korean, and neither contrasted with /ŋ/, so the symbols all merged, leaving ㅎ /h/ as the only other member of the family.

But anyway, yes, Korean definitely shows how you could treat syllable onset as if it were a consonant with particular features.

[Richard W]

Analogous problems arise with codas, which have no nucleus and no independent place features since they assimilate with following onsets like the moraic nasal and sokuon.

Can't you just add features of nasal and stop? Admittedly that may be awkward with the implicit vowel.

Your script could mark phonetic distinctions rather than phonemic distinctions.

[malloc]

But anyway, yes, Korean definitely shows how you could treat syllable onset as if it were a consonant with particular features.

It occurs to me that the phonological rules for initial vowels actually do imply an inaudible onset that one could fairly consider dorsal and vocalic. Which solves that problem at least.

[quinterbeck]

I'm still screwing around with phonologies, trying to create something fun and weird in the spirit of Danish and themed around west Africa, so weirdass allophony/patterninɡ and questionable boundaries of what a phoneme is in a smallish consonant inventory.

This is what I have so far:

More: show

/i e u o a ɨ/ + lenɡth
/n/
/p t k/
/ts tʃ/
/f s ʃ x/
/b d ɡ/
/w l j/

(C)V(C) syllables, maximum possible cluster is VCCV. Possible clusters heavily restricted.

I think the best way to describe this clusterfuck would be to go from phoneme to phoneme and note all the different allophones/replacements.

/n/ - The only nasal that's always a nasal, everywhere. Also can form /Cn/ clusters with /b d j l w/, which then become [mn ɾn jn ln mn] in surface realisation. All other /Cn/ clusters forbidden. Becomes palatal /ɲ/ before /i/. When it ends in a cluster in front of absolutely any C whatsoever, the post-nasal fortition happens, see below.

/p t k/ - Unaspirated and voiceless by default. Partially to fully voice to /b d ɡ/ after lonɡ vowels. Lenite to /f s x/ before another plosive, if that plosive is voiced, the lenited form is then either voiced to [v z] for /p t/ or elided with compensatory lenɡtheninɡ of the precedinɡ vowel for /k/. /t/ palatalises to /tsʲ/ before /i/.

/ts tʃ/ - in complementary distribution before /i/ and /ɨ/, no /ts/ before /i/ and no /tʃ/ before /ɨ/. Deaffricate to /s ʃ/ before plosives, with subsequent voicinɡ to /z ʒ/ before voiced plosives. Also deaffricate after lonɡ vowels. /sts stʃ ʃtʃ ʃts/ clusters simplify to /s ʃ ʃ s/, respectively.

/f/ - voices to /v/ after lonɡ vowels, lenites to /w/ in clusters.

/s ʃ/ - in complementary distribution before /i/ and /ɨ/, just like their affricate counterparts. /s/ becomes /ʃ/ before and after stops. Both sibilants elide after affricates. /ʃ/ is the only non-liquid phoneme that can occur after /p t k/ and keep them [p t k] on the surface. /ʃ/ voices to /ʒ/ before /b d ɡ/.

/x/ - elides in most clusters with compensatory lenɡtheninɡ if the adjacent vowel was short, otherwise tracelessly. Only remains in clusters with /w l j/ and after /b d ɡ/.

/b d ɡ/ - the wierdest ones. [m ɾ ŋ] when after a lonɡ vowel or precedinɡ any consonant whatsover. [m ŋ] then triɡɡer post-nasal fortition, as described below. Intervocalic /d/ is always [ʒ] before /i/, irrespestive of the lenɡth of the precedinɡ vowel. /ɡ/ becomes /n/ in coda, reɡardless of anythinɡ else.

/w l j/ - /l/ is [ɫ] when next to /u o ɨ/ unless there's an /i/ next to it as well. After lonɡ vowels, [ɫ] vocalises to /w~u̯/. Otherwise, these ɡuys can be in any cluster whatsoever, only underɡoinɡ post-nasal fortition.

Post-nasal fortition - a very productive phonoloɡical process that, basically, fortitions everythinɡ after a phonemic or allophonic nasals into a voiced plosive. Labials become [mb], coronals except [tsʲi] become [nd], palatals and [tsʲi] become [ɲɟ], velars become [ŋɡ]. This process affects even liquids and ensures that nothinɡ in the lanɡuaɡe can stand after a nasal aside from the aforemenrioned voiced plosives.

What do you think?

I'd love to see a sample, comparing the phonemic elements of words with the surface forms.

[dɮ the phoneme]

Right now I'm playing around with some ideas for a language with serial verb constructions as well as V > T raising, and I can't really find anything on how these features would intereact. In particular, since both verbs in an SVC are main verbs, which one raises to T? Both?

[Xwtek]

Is it realistic to have a diphthong like this:

/æɪ ɑʊ ɪæ ʊɑ/

[dhok]

isn't that just Old English

[Xwtek]

isn't that just Old English

No, the Old English diphthong is: iu̯ eo̯ and æɑ̯

[WeepingElf]

Is it realistic to have a diphthong like this:

/æɪ ɑʊ ɪæ ʊɑ/

I think Saami languages have a similar inventory of diphthongs (/ei ou ie uo/), so I guess this is realistic.

EDIT: No, I was wrong. But I think I have seen that inventory somewhere, though I no longer remember where. I think it was in one of Peter Schrijver's writings on the phonological history of various northwestern European languages, but I can't tell. Anyway, I think Akangka's diphthong inventory is realistic.

[Xwtek]

Is it realistic for a language to lack Obstruent+Liquid cluster? In my language, the only liquid is /m n l r j w/. However, /r/ comes from /d/ in certain positions, and it would devoice instead of turning into /r/ after voiceless obstruent, (the contrast of /r/ and /d/ is only present in /rd/ vs /dd/, and only word medially, as both cluster are banned word initially, and only /rd/ allowed word finally). Also /l/ fortify into /tɬ/ after obstruent. /j w/ is disallowed because it would form diphthong with vowel instead.

[bbbosborne]

can anyone make <s> with a hacek diacritic and a dot diacritic below the <s>?

[akam chinjir]

Two versions:

ṣ̌ (S + hacek + subdot)
ṣ̌ (S-subdot + hacek)

There's no one unicode codepoint for an S with caron and dot below, but there's actually one with a caron and a dot above: ṧ.

[Moose-tache]

Is it realistic for a language to lack Obstruent+Liquid cluster? In my language, the only liquid is /m n l r j w/. However, /r/ comes from /d/ in certain positions, and it would devoice instead of turning into /r/ after voiceless obstruent, (the contrast of /r/ and /d/ is only present in /rd/ vs /dd/, and only word medially, as both cluster are banned word initially, and only /rd/ allowed word finally). Also /l/ fortify into /tɬ/ after obstruent. /j w/ is disallowed because it would form diphthong with vowel instead.

Lots of languages lack obstruent-liquid clusters. It seems you're including semivowels as well, and I guess that's slightly less common, but it still happens. Japanese would have this feature if it weren't for all the loans from Chinese. And if we classify palatization as not a cluster there are probably several more examples.

[Xwtek]

Is it realistic for a language to lack Obstruent+Liquid cluster? In my language, the only liquid is /m n l r j w/. However, /r/ comes from /d/ in certain positions, and it would devoice instead of turning into /r/ after voiceless obstruent, (the contrast of /r/ and /d/ is only present in /rd/ vs /dd/, and only word medially, as both cluster are banned word initially, and only /rd/ allowed word finally). Also /l/ fortify into /tɬ/ after obstruent. /j w/ is disallowed because it would form diphthong with vowel instead.

Lots of languages lack obstruent-liquid clusters. It seems you're including semivowels as well, and I guess that's slightly less common, but it still happens. Japanese would have this feature if it weren't for all the loans from Chinese. And if we classify palatization as not a cluster there are probably several more examples.

The problem is my language clearly have Obstruent+Obstruent cluster. Also Liquid+Obstruent cluster (Although the cluster is restricted to nasal+voiced stop/affricate in this case)

[Moose-tache]

The problem is my language clearly have Obstruent+Obstruent cluster. Also Liquid+Obstruent cluster (Although the cluster is restricted to nasal+voiced stop/affricate in this case)

Shouldn't be a problem. English allows mb but not *bm (except in compound words), and Korean allows gw but not *wg. If there's a plausible sound change to explain it, you can rule out whatever clusters you like. If it's a weird combination, then maybe it will be unstable over centuries and eventually change. But that doesn't mean that a language, in this moment, can't have those cluster rules.

[Xwtek]

The problem is my language clearly have Obstruent+Obstruent cluster. Also Liquid+Obstruent cluster (Although the cluster is restricted to nasal+voiced stop/affricate in this case)

Shouldn't be a problem. English allows mb but not *bm (except in compound words), and Korean allows gw but not *wg. If there's a plausible sound change to explain it, you can rule out whatever clusters you like. If it's a weird combination, then maybe it will be unstable over centuries and eventually change. But that doesn't mean that a language, in this moment, can't have those cluster rules.

But English certainly has Obstruent+Liquid cluster.

[Richard W]

Is it realistic for a language to lack Obstruent+Liquid cluster?

Yes. Pali comes close; what obstruent+liquid clusters survive are reasonably suspected of being due to the influence of other dialects. Lack of syllable-initial obstruent+liquid clusters can be seen in Persian.

[Xwtek]

Is it realistic for a language to lack Obstruent+Liquid cluster?

Yes. Pali comes close; what obstruent+liquid clusters survive are reasonably suspected of being due to the influence of other dialects. Lack of syllable-initial obstruent+liquid clusters can be seen in Persian.

I mean, I don't have Obstruent+Liquid initial cluster, but have Obstruent+Obstruent and Nasal+Obstruent iniital cluster