Seic languages

[gach]

Here's one more conlanging project that I've been thinking for a while but have still mostly been developing ideas for. For a long time it wasn't much more than an idea that I should try to do something polysynthetic some day and a place where I could mentally put grammatical ideas I had got from reading about such languages as Chocktaw and Yimas. I still haven't got much concrete to show but at least I have some sketches to write about. The focal point language in this project is called Janti but it has a whole family behind it which I'm calling the Seic languages. The name refers to a common root that's found in the form of sei or something similar across the family and denotes a type of a long seaworthy boat that the speakers of the coastal branches of this family use widely. It's a bit of a misnomer for Janti, since it belongs to an inland branch, but it still reflects the names given to many of the speakers of the Seic languages by their neighbours.

You've already seen the phonology I have for the proto language of the family. This was the first one I posted last week in the random phonology thread:

Code: Select all

p    t    ts   tʃ        k    ʔ
     t'   ts'       tɬ'  k'
          s    ʃ    ɬ    x
m    n         j    l    w

Code: Select all

i  ɨ           ɨ:
e     o
   a           a:

I'm intending a time depth of at least 3000 years between this stage and Janti and for the most part I'm still thinking how to fill all this time with sound changes and mix up the correspondences. What I do know is that the phoneme inventory in Janti is going to be

Code: Select all

p    t         k    ʔ
     tt        kk
     s              h
m    n
          j    w

Code: Select all

i  ɨ  u
e  ɵ  o
   a

with an additional /l/ in some dialects, the loss of /h/ in others, and a good deal of variation in the vowels between dialects. The double consonants <tt> anf <kk> in the consonant table aren't geminates but refer to strong stops that don't ever undergo voicing. The regular stops all have voiced allophones between inherently voiced phonemes (vowels, glides, nasals).

Some changes that I'm considering are:

• The raising of /o/ word finally into /u/ and fronting it otherwise into /ɵ/
• The collapse of long vowels into short ones plus the labialisation of the old long /a:/ into /o/
• Some chain shift on the vowels
• The dropping of a lot of intervocalic sonorants and merging adjacent vowels to mix the vowel history
• The loss of laterals, /tɬ'/ > /ts'/, /ɬ/ > /s n j/, /l/ > /n j/
• The debuccalisation of the postalveolar and velar fricatives, /ʃ x / > /h/
• The loss of affricates into fricatives, /ts ts'/ > /s/, /tʃ/ > /s h/ but possibly also into the stops /t tt/ in some environments
• The lenition of /p/ > /w/ everywhere except at the start of the word
• The generation of new labial nasals from /nw/ > /m(w)/

I will need the language to have decently complex clusters throughout its development to have the room for more changes and to still retain some phonological complexity in the end product.

A characteristic feature of Janti is reflecting the Seic plain vs. ejective stop series as weak vs. strong stops. In Janti the weak stops undergo voicing lenition when surrounded by voiced phonemes (/p/ > [b ] or /w/, /t/ > [d ɾ] or /j/, /k/ > [g ɣ ʝ] or /j/) while the strong ones stay invariably voiceless (/tt/ = [t], /kk/ = [k]). The glottal stop naturally can't be voiced but I might still add a conditional dropping rule for it. I was thinking for a while if I should derive the strong stops from original plain or ejective ones but in the end settled on the ejectives. The history will be so that through the most of the development of Janti's ancestors the language had plain vs. ejective stops and affricates. The plain series could experience more lenition as time went on but this was always blocked on the ejectives. The ejectives lost their glottalisation only a little before the final splitting of Proto Janti into the dialect continuum that forms modern Janti and haven't yet had time to evolve any further from their current state of being plain voiceless stops.

On the side of grammatical ideas, two of the most important characteristics of Janti will be complex polypersonal agreement and classification both by gender and verbal classifiers. There are three genders: animate, neuter 1, and neuter 2. The animate gender continues an original animate and the two neuter genders derive from old inanimate count (1) and mass (2) nouns. This gender system is shared with or without the mass noun distinction across the Seic family but history has mixed the gender membership in the languages to a varying degree. In Janti the neuter 2 gender has lost its association with mass nouns to a large degree and you can find animals, for example, in all three of the genders. Gender membership affects agreement on verbs and adjectives and the forms that the deictics take.

The verbs in Janti have three sets of agreement pronominals for animate participants, including the sets for agent (A), patient (O), and dative (D) functions, and a good deal of split alignment similar to in Chochtaw. The alignment patterns are lexically determined and all combinations of different agreement sets are attested in the lexicon. Neuter participants don't have proper A- and D-agreement but they do have O-agreement plus special intransitive S-agreement that corresponds with the A-agreement of animate subjects on the same verbs. Bypasses for the holes do exist and for example neuter nouns in the transitive A-role are coded using an indefinite subject affix.

At the same time there's classification for syntactic objects and intransitive subjects by verbal classifiers which originate from classificatory incorporation. These are different from the gender system in that their selection is always based on either transparent semantics or in conventionalised extension of a semantic trait from a reference noun. There are also a lot more classifiers than there are genders, and despite there being strong tendencies for which classifier to use with which noun, you can commonly change a conventional classifier to another for intentional semantic effect. Also unlike genders, not all nouns are classified. Human participants only receive a classifier in order to add special non-standard traits to them. Typical classifiers describe generic-specific relations (like "tree" for trees or "animal" for larger animals) or shape or consistency (like "long", "round", or "liquid") but there are also rarer more specific classifiers.

That's it for now. As you can see, I still have most of the concrete stuff undone. I need to make a note for myself to clear the terminology that I'm going to use for describing the alignment and the syntactic roles since I need to talk separately about semantic and syntactic roles as well as the labels of the functional agreement sets.

[Frislander]

Spoiler:

Here's one more conlanging project that I've been thinking for a while but have still mostly been developing ideas for. For a long time it wasn't much more than an idea that I should try to do something polysynthetic some day and a place where I could mentally put grammatical ideas I had got from reading about such languages as Chocktaw and Yimas. I still haven't got much concrete to show but at least I have some sketches to write about. The focal point language in this project is called Janti but it has a whole family behind it which I'm calling the Seic languages. The name refers to a common root that's found in the form of sei or something similar across the family and denotes a type of a long seaworthy boat that the speakers of the coastal branches of this family use widely. It's a bit of a misnomer for Janti, since it belongs to an inland branch, but it still reflects the names given to many of the speakers of the Seic languages by their neighbours.

You've already seen the phonology I have for the proto language of the family. This was the first one I posted last week in the random phonology thread:

Code: Select all

p    t    ts   tʃ        k    ʔ
     t'   ts'       tɬ'  k'
          s    ʃ    ɬ    x
m    n         j    l    w

Code: Select all

i  ɨ           ɨ:
e     o
   a           a:

I'm intending a time depth of at least 3000 years between this stage and Janti and for the most part I'm still thinking how to fill all this time with sound changes and mix up the correspondences. What I do know is that the phoneme inventory in Janti is going to be

Code: Select all

p    t         k    ʔ
     tt        kk
     s              h
m    n
          j    w

Code: Select all

i  ɨ  u
e  ɵ  o
   a

with an additional /l/ in some dialects, the loss of /h/ in others, and a good deal of variation in the vowels between dialects. The double consonants <tt> anf <kk> in the consonant table aren't geminates but refer to strong stops that don't ever undergo voicing. The regular stops all have voiced allophones between inherently voiced phonemes (vowels, glides, nasals).

Some changes that I'm considering are:

• The raising of /o/ word finally into /u/ and fronting it otherwise into /ɵ/
• The collapse of long vowels into short ones plus the labialisation of the old long /a:/ into /o/
• Some chain shift on the vowels
• The dropping of a lot of intervocalic sonorants and merging adjacent vowels to mix the vowel history
• The loss of laterals, /tɬ'/ > /ts'/, /ɬ/ > /s n j/, /l/ > /n j/
• The debuccalisation of the postalveolar and velar fricatives, /ʃ x / > /h/
• The loss of affricates into fricatives, /ts ts'/ > /s/, /tʃ/ > /s h/ but possibly also into the stops /t tt/ in some environments
• The lenition of /p/ > /w/ everywhere except at the start of the word
• The generation of new labial nasals from /nw/ > /m(w)/

I will need the language to have decently complex clusters throughout its development to have the room for more changes and to still retain some phonological complexity in the end product.

A characteristic feature of Janti is reflecting the Seic plain vs. ejective stop series as weak vs. strong stops. In Janti the weak stops undergo voicing lenition when surrounded by voiced phonemes (/p/ > or /w/, /t/ > [d ɾ] or /j/, /k/ > [g ɣ ʝ] or /j/) while the strong ones stay invariably voiceless (/tt/ = [t], /kk/ = [k]). The glottal stop naturally can't be voiced but I might still add a conditional dropping rule for it. I was thinking for a while if I should derive the strong stops from original plain or ejective ones but in the end settled on the ejectives. The history will be so that through the most of the development of Janti's ancestors the language had plain vs. ejective stops and affricates. The plain series could experience more lenition as time went on but this was always blocked on the ejectives. The ejectives lost their glottalisation only a little before the final splitting of Proto Janti into the dialect continuum that forms modern Janti and haven't yet had time to evolve any further from their current state of being plain voiceless stops.

On the side of grammatical ideas, two of the most important characteristics of Janti will be complex polypersonal agreement and classification both by gender and verbal classifiers. There are three genders: animate, neuter 1, and neuter 2. The animate gender continues an original animate and the two neuter genders derive from old inanimate count (1) and mass (2) nouns. This gender system is shared with or without the mass noun distinction across the Seic family but history has mixed the gender membership in the languages to a varying degree. In Janti the neuter 2 gender has lost its association with mass nouns to a large degree and you can find animals, for example, in all three of the genders. Gender membership affects agreement on verbs and adjectives and the forms that the deictics take.

The verbs in Janti have three sets of agreement pronominals for animate participants, including the sets for agent (A), patient (O), and dative (D) functions, and a good deal of split alignment similar to in Chochtaw. The alignment patterns are lexically determined and all combinations of different agreement sets are attested in the lexicon. Neuter participants don't have proper A- and D-agreement but they do have O-agreement plus special intransitive S-agreement that corresponds with the A-agreement of animate subjects on the same verbs. Bypasses for the holes do exist and for example neuter nouns in the transitive A-role are coded using an indefinite subject affix.

At the same time there's classification for syntactic objects and intransitive subjects by verbal classifiers which originate from classificatory incorporation. These are different from the gender system in that their selection is always based on either transparent semantics or in conventionalised extension of a semantic trait from a reference noun. There are also a lot more classifiers than there are genders, and despite there being strong tendencies for which classifier to use with which noun, you can commonly change a conventional classifier to another for intentional semantic effect. Also unlike genders, not all nouns are classified. Human participants only receive a classifier in order to add special non-standard traits to them. Typical classifiers describe generic-specific relations (like "tree" for trees or "animal" for larger animals) or shape or consistency (like "long", "round", or "liquid") but there are also rarer more specific classifiers.

That's it for now. As you can see, I still have most of the concrete stuff undone. I need to make a note for myself to clear the terminology that I'm going to use for describing the alignment and the syntactic roles since I need to talk separately about semantic and syntactic roles as well as the labels of the functional agreement sets.

Love it. I think there should be more polysynthesis on this board, and this will be a welcome addition. I love the phoneme inventories, particularly the ejective lateral affricative without a corresponding plain: very Salishan.

One small quibble: does Choctaw really count as polysynthetic? It may have pretty robust verb morphology including polypersonal marking, but noun incorporation shows only traces and there don't seem to be vast quantity of derivational affixes. (this is not true of all Muskogean languages: Koasati's verbs look positively monstrous).

[gach]

Love it. I think there should be more polysynthesis on this board, and this will be a welcome addition. I love the phoneme inventories, particularly the ejective lateral affricative without a corresponding plain: very Salishan.

Thanks. I had a look a few years ago for which languages have an ejective lateral affricate but not a pulmonic one since I had a feeling of seeing that repeatedly. I might not have found the best statistics but it is a hole that does occur to varying degrees in a bunch of languages around the world. I do like this particular hole and doing that bit of research gave me ultimately the motivation for including it for this family.

One small quibble: does Choctaw really count as polysynthetic? It may have pretty robust verb morphology including polypersonal marking, but noun incorporation shows only traces and there don't seem to be vast quantity of derivational affixes.

Not that much no. Chocktaw's primary influence for me has anyway been for including a system of complex split alignment. But I should perhaps also say that I'm not that concerned about the exact definition of polysynthesis myself, largely because I don't usually know that well what other people mean by it when they use the term. For Janti the polysynthetic label basically means higher morphological complexity on the verb, polypersonal agreement, and incorporation in the form of the verbal classifiers and a group of verbs that demand an incorporated object. If you wish, you might also try an analysis where you say that the classifiers are classifiers and the incorporating verbs are in fact denominal derivations, and by doing so describe the language completely without invoking incorporation. It's a bit of a muddy business.

[Frislander]

Not that much no. Chocktaw's primary influence for me has anyway been for including a system of complex split alignment. But I should perhaps also say that I'm not that concerned about the exact definition of polysynthesis myself, largely because I don't usually know that well what other people mean by it when they use the term. For Janti the polysynthetic label basically means higher morphological complexity on the verb, polypersonal agreement, and incorporation in the form of the verbal classifiers and a group of verbs that demand an incorporated object. If you wish, you might also try an analysis where you say that the classifiers are classifiers and the incorporating verbs are in fact denominal derivations, and by doing so describe the language completely without invoking incorporation. It's a bit of a muddy business.

That's fine: mucking about with the definitions of the actual terms is the linguist's job. For me polysynthetic languages select from a range of features which are characteristic but none of which are obligatory. Such features include polypersonal marking (surprisingly enough not universal - Nuu-Chah-Nulth only marks one argument and Salish can be even more sparse); extensive bound morphology of an often lexical nature; noun incorporation; verb incorporation.