So, looking at Navajo, Koyukon, and similarly complex polysythetic langs, a lot of the affixes combine together in crazy ways.
You might have any number of changes, from metathesis to suppletion. It looks like you sort of have to define some morphophonological rules (like metathesis), but in other places, you just say when -'a- meets -di-, they become -ke-.
Is that a fair description?
In my lang Táálen, there are 5 slots before the root, and 4 after, and I know there's a lot of the transformation similar to the previously named langs, but I'm not sure whether I'm describing how they work correctly. I feel like I'm missing something in how to put these parts together realistically.
Affix combination in polysynthetic langs
Re: Affix combination in polysynthetic langs
A lot of it is historical stuff. Maybe do something that might be plausible historically?
Yo jo moy garsmichte pa
Re: Affix combination in polysynthetic langs
Kashaya Pomo is a (fairly extreme) example that I'm fairly familiar with. It's traditionally analyzed as having 2 position classes for prefixes and 15 for suffixes, but the reality is more complex. The morphological rules are dazzlingly elaborate, but here's a quick and dirty summary of some of them:
*Several suffixes trigger a process referred to as the "decrement", which consists of the loss of a glottal segment, either by deleting an /h/ or /ʔ/ or by causing an aspirated or glottalized consonant to become it's plain counterpart. Other suffixes trigger debuccalization of certain consonants in subsequent suffixes, and a few trigger both shifts.
*Verbs may be marked to indicate that they are performed by more than one agent by shifting all instances of /ˀn/ in suffixes to /tʃ/. This process also affects the final consonant in the verb root.
*Position class 3 consists of the "terrestrial" suffix /-tʃ/ and the "super-terrestrial" suffix /-m/, which are optionally used to indicate that the action is taking place on or off the ground, respectively. These suffixes may also cause the final stem vowel to lengthen, but not all stems are affected.
*Position class 4 contains a number of suffixes, including the "semelfactive" /-tʃ/, which can indicate a number of things depending on the nature of the verb, and the "inceptive" suffix /-jitʃ/, which indicates that the action of the verb is beginning. The former takes the form /-htʃi/ after certain verb roots, while the latter may only be affixed after stems ending in a vowel. There are other ways of forming the inceptive after other kind of stem. If the semelfactive is directly followed by the reflexive suffix /-itʃ'/, the two affixes merge to form the suffix /-tʃtʃitʃ'/, which is most frequently used to indicate that the action moves backward with a quick motion.
*The "plural act" suffix, also belonging to position class 4, which serves to indicate that multiple objects are undergoing the action described in the verb or that it is being performed multiple times with regard to a single object, has 13 different allomorphs. Although these are conditioned by a variety of phonological factors, the environments in which they may occur significantly overlap, and the choice of allomorph is idiosyncratic to the particular verb root. Some of these are actually infixed before certain root-final consonants, and all but three of them trigger the decrement. With some roots ending in a plain dental or alveolar stop, the plural act marker consists solely of the decrement.
*The final class 4 suffix is "plural movement," used with verbs of motion to indicate that multiple persons or objects are moving, which takes the form of the infix <h> before stem-final consonants and the suffix /-ht/ after stem-final vowels. The later form is usually reduced to /-t/ before additional suffixes beginning with /m/.
*Position class 5 consists of various directional suffixes, though in practice these may occur at one (or more) of several places in the inflected verb. Directionals ending in a vowel trigger the decrement, and an epenthetic /w/ is inserted before vowel-initial directionals that follow another suffix ending in a vowel.
*The reflexive, which belongs to position class 6, is /-jitʃ'/ after most vowel-final roots and /-itʃ'/ after consonant-final roots and any suffixes of preceding position classes.
*The causative (position class 7) is relatively straightforward, taking the form /-hqa/ after a vowel or /-qa/ after a consonant, usually triggering aspiration.
*Position class 8 is the "locomotory" suffix, which may imply constant or intermittent movement or a gradual change of state, depending on the nature of the verb root. It takes the form /-aˀnaˀn/ following syllables with a short vowel and /-aˀnaˀnaˀn/ after a syllable with a long vowel. It also sometimes takes the form /-waˀnaˀn/ after vowels, and before a directional or distributive suffix the final /-aˀn/ is omitted.
*Position class 9 is the "durative" suffix, denoting that the action takes place over "a little while" (an indefinite but bounded period of time), which has 10 phonologically-conditioned allomorphs.
*The "distributive" suffix (position class 10) indicates that "the action takes place in different places or moves in a haphazard manner", and may only come after the locomotory or durative suffix or one of a few specific directionals. It may be realized as /-waˀn/ or /-iwaˀn/.
*The negative suffix (position class 12) is generally /-tʰ/ before a vowel and /-tʰi/ before a consonant, but takes the form /-tʰu/ before an imperative.
*Although other affixes are generally optional, every verb must take an affix from position class 14, which contains 7 evidentials, 5 irrealis mood markers, and 4 different kinds of imperative. However, none of these markers regularly experiences any significant allomorphy (save that triggered by previous markers), although two evidentials that begin with vowels take an epenthetic initial /w/ after suffixes ending in a vowel. Class 14 also contains the absolutive marker, which is /-w/ after a vowel, /-i/ after /n/, and /-ʔ/ after any other consonant, triggering glottalization. There are also 6 switch-reference markers in this class, but these are kind of complicated to explain and don't undergo much allomorphy.
There are other processes as well that are phonologically conditioned and not associated with particular position classes, but that's a long story for another time. The examples above seem like they can, for the most part, be fairly easily explained in terms of historical sound changes, particularly phonetic reductions related to the grammaticalization of what may once have been independent coverbs and particles. Hopefully they can also provide you with some inspiration if you're stuck for ideas regarding allomorphic rules.
*Several suffixes trigger a process referred to as the "decrement", which consists of the loss of a glottal segment, either by deleting an /h/ or /ʔ/ or by causing an aspirated or glottalized consonant to become it's plain counterpart. Other suffixes trigger debuccalization of certain consonants in subsequent suffixes, and a few trigger both shifts.
*Verbs may be marked to indicate that they are performed by more than one agent by shifting all instances of /ˀn/ in suffixes to /tʃ/. This process also affects the final consonant in the verb root.
*Position class 3 consists of the "terrestrial" suffix /-tʃ/ and the "super-terrestrial" suffix /-m/, which are optionally used to indicate that the action is taking place on or off the ground, respectively. These suffixes may also cause the final stem vowel to lengthen, but not all stems are affected.
*Position class 4 contains a number of suffixes, including the "semelfactive" /-tʃ/, which can indicate a number of things depending on the nature of the verb, and the "inceptive" suffix /-jitʃ/, which indicates that the action of the verb is beginning. The former takes the form /-htʃi/ after certain verb roots, while the latter may only be affixed after stems ending in a vowel. There are other ways of forming the inceptive after other kind of stem. If the semelfactive is directly followed by the reflexive suffix /-itʃ'/, the two affixes merge to form the suffix /-tʃtʃitʃ'/, which is most frequently used to indicate that the action moves backward with a quick motion.
*The "plural act" suffix, also belonging to position class 4, which serves to indicate that multiple objects are undergoing the action described in the verb or that it is being performed multiple times with regard to a single object, has 13 different allomorphs. Although these are conditioned by a variety of phonological factors, the environments in which they may occur significantly overlap, and the choice of allomorph is idiosyncratic to the particular verb root. Some of these are actually infixed before certain root-final consonants, and all but three of them trigger the decrement. With some roots ending in a plain dental or alveolar stop, the plural act marker consists solely of the decrement.
*The final class 4 suffix is "plural movement," used with verbs of motion to indicate that multiple persons or objects are moving, which takes the form of the infix <h> before stem-final consonants and the suffix /-ht/ after stem-final vowels. The later form is usually reduced to /-t/ before additional suffixes beginning with /m/.
*Position class 5 consists of various directional suffixes, though in practice these may occur at one (or more) of several places in the inflected verb. Directionals ending in a vowel trigger the decrement, and an epenthetic /w/ is inserted before vowel-initial directionals that follow another suffix ending in a vowel.
*The reflexive, which belongs to position class 6, is /-jitʃ'/ after most vowel-final roots and /-itʃ'/ after consonant-final roots and any suffixes of preceding position classes.
*The causative (position class 7) is relatively straightforward, taking the form /-hqa/ after a vowel or /-qa/ after a consonant, usually triggering aspiration.
*Position class 8 is the "locomotory" suffix, which may imply constant or intermittent movement or a gradual change of state, depending on the nature of the verb root. It takes the form /-aˀnaˀn/ following syllables with a short vowel and /-aˀnaˀnaˀn/ after a syllable with a long vowel. It also sometimes takes the form /-waˀnaˀn/ after vowels, and before a directional or distributive suffix the final /-aˀn/ is omitted.
*Position class 9 is the "durative" suffix, denoting that the action takes place over "a little while" (an indefinite but bounded period of time), which has 10 phonologically-conditioned allomorphs.
*The "distributive" suffix (position class 10) indicates that "the action takes place in different places or moves in a haphazard manner", and may only come after the locomotory or durative suffix or one of a few specific directionals. It may be realized as /-waˀn/ or /-iwaˀn/.
*The negative suffix (position class 12) is generally /-tʰ/ before a vowel and /-tʰi/ before a consonant, but takes the form /-tʰu/ before an imperative.
*Although other affixes are generally optional, every verb must take an affix from position class 14, which contains 7 evidentials, 5 irrealis mood markers, and 4 different kinds of imperative. However, none of these markers regularly experiences any significant allomorphy (save that triggered by previous markers), although two evidentials that begin with vowels take an epenthetic initial /w/ after suffixes ending in a vowel. Class 14 also contains the absolutive marker, which is /-w/ after a vowel, /-i/ after /n/, and /-ʔ/ after any other consonant, triggering glottalization. There are also 6 switch-reference markers in this class, but these are kind of complicated to explain and don't undergo much allomorphy.
There are other processes as well that are phonologically conditioned and not associated with particular position classes, but that's a long story for another time. The examples above seem like they can, for the most part, be fairly easily explained in terms of historical sound changes, particularly phonetic reductions related to the grammaticalization of what may once have been independent coverbs and particles. Hopefully they can also provide you with some inspiration if you're stuck for ideas regarding allomorphic rules.
Re: Affix combination in polysynthetic langs
Thanks CatDoom, that exactly what I needed. It was feeling like I was making some sort of kitchensink-newb-lang because of the craziness of the rules, but that really is the only way to explain these changes, and they really are that non-intuitive.
Re: Affix combination in polysynthetic langs
Glad I could help! In case you're interested, here are the specific allomorphs of the most variable affixes, listed according to the conditions under which they may occur:
Plural Act (position class 4):
Before any root-final consonant: <t> (triggers decrement)
Before root-final /k'/: <h> (triggers decrement)
Before root-final /m, q, qʷ, tʃ/: <tu> (triggers decrement; note that /qʷ/ is more often analyzed as a cluster)
After /j/: -ʔta (triggers decrement)
With some roots ending in /t̪ ,t̺/: -0 (triggers decrement)
After /l, n, nʰ, ˀn, tʃ'/: -ta (triggers decrement)
After /l, n/ and before the durative suffix: -at (no decrement)
After a long vowel: -m (triggers decrement)
After any vowel or consonant: -m (no decrement)
After /l/: -aq, -ataq (both trigger decrement)
After a vowel and before the durative suffix: -w (no decrement)
After a long vowel and before the durative suffix: -w (triggers decrement)
Durative (position class 9):
Directly following monosyllabic roots or after disyllabic stems ending in a vowel: -tʃiˀn
After other vowel-final stems: -meˀn
After a suffix (but not root) ending in /ˀn/: -itʃeˀn
After /tʃ/ if [+plural agent]: -iˀn
After /tʃ'/ if [+plural agent]: -iwaˀn
(I'm not certain if the plural agent process turns the above two affixes into -itʃ and -iwatʃ; I'm guessing not, since that would seem to be the only reason for distinguishing the post /tʃ/ allomorphs in +/- plural agent environments)
After a single syllable ending in /tʃ, tʃ'/ or after the directional suffix /-matʃ/ when the final stem vowel is short [-plural agent]: -iˀn
Otherwise, after /tʃ, tʃ'/ [-plural agent]: -eˀn
After any other suffix ending in a consonant other than /ˀn tʃ tʃ'/: -aˀn
After the causative suffix, the allomorph -waˀn may be used, but this seems to be in free variation with -tʃiˀn and -meˀn (depending on the structure of the stem)
The durative may also be repeated to indicate that the action takes place over a relatively longer period of time, similar to English constructions like "CatDoom went on and on about Kashaya."
It's also worth noting that /ˀn/ is realized as [d] when it winds up in syllable onset position. The Pomoan languages are sometimes analyzed as having the phonemes /b/ and /d/, but more recent scholarship usually regards these as allophones of /ˀm/ and /ˀn/, respectively.
Plural Act (position class 4):
Before any root-final consonant: <t> (triggers decrement)
Before root-final /k'/: <h> (triggers decrement)
Before root-final /m, q, qʷ, tʃ/: <tu> (triggers decrement; note that /qʷ/ is more often analyzed as a cluster)
After /j/: -ʔta (triggers decrement)
With some roots ending in /t̪ ,t̺/: -0 (triggers decrement)
After /l, n, nʰ, ˀn, tʃ'/: -ta (triggers decrement)
After /l, n/ and before the durative suffix: -at (no decrement)
After a long vowel: -m (triggers decrement)
After any vowel or consonant: -m (no decrement)
After /l/: -aq, -ataq (both trigger decrement)
After a vowel and before the durative suffix: -w (no decrement)
After a long vowel and before the durative suffix: -w (triggers decrement)
Durative (position class 9):
Directly following monosyllabic roots or after disyllabic stems ending in a vowel: -tʃiˀn
After other vowel-final stems: -meˀn
After a suffix (but not root) ending in /ˀn/: -itʃeˀn
After /tʃ/ if [+plural agent]: -iˀn
After /tʃ'/ if [+plural agent]: -iwaˀn
(I'm not certain if the plural agent process turns the above two affixes into -itʃ and -iwatʃ; I'm guessing not, since that would seem to be the only reason for distinguishing the post /tʃ/ allomorphs in +/- plural agent environments)
After a single syllable ending in /tʃ, tʃ'/ or after the directional suffix /-matʃ/ when the final stem vowel is short [-plural agent]: -iˀn
Otherwise, after /tʃ, tʃ'/ [-plural agent]: -eˀn
After any other suffix ending in a consonant other than /ˀn tʃ tʃ'/: -aˀn
After the causative suffix, the allomorph -waˀn may be used, but this seems to be in free variation with -tʃiˀn and -meˀn (depending on the structure of the stem)
The durative may also be repeated to indicate that the action takes place over a relatively longer period of time, similar to English constructions like "CatDoom went on and on about Kashaya."
It's also worth noting that /ˀn/ is realized as [d] when it winds up in syllable onset position. The Pomoan languages are sometimes analyzed as having the phonemes /b/ and /d/, but more recent scholarship usually regards these as allophones of /ˀm/ and /ˀn/, respectively.
Re: Affix combination in polysynthetic langs
Wow, those allomorphs can be even more non-intuitive than I thought. Mithun's "Languages of Native North America" either doesn't highlight how crazy they get, or I missed the examples.
Re: Affix combination in polysynthetic langs
My source is Theoretical Aspects of Kashaya Phonology and Morphology by Eugene Buckley.
The durative allomorphs seem relatively straightforward, at least in that they all end in a non-back vowel + /ˀn/.
With regard to the Plural Act marker, it's possible that the original morpheme was something like <tmiq>. As I recall, it's believed that in Pre-Kashaya or Pre-Proto-Pomo, /q/ was lost next to vowels other than /a/, while /m/ synchronically causes a following /i/ to lower to /a/. Maybe the cluster was resolved through the deletion of /m/ under most conditions, and /q/ was deleted following /i/, resulting in the (V)t(V) form that appears under many conditions. Later, some other process might have been triggered by certain stems that wound up ending in /-l/, which got rid of /tm/ but left the lowered vowel /a/, allowing /q/ to survive. Under yet different conditions, only /m/ survived, turning into /w/ intervocalically.
That's pretty much all speculation on my part, however.
The durative allomorphs seem relatively straightforward, at least in that they all end in a non-back vowel + /ˀn/.
With regard to the Plural Act marker, it's possible that the original morpheme was something like <tmiq>. As I recall, it's believed that in Pre-Kashaya or Pre-Proto-Pomo, /q/ was lost next to vowels other than /a/, while /m/ synchronically causes a following /i/ to lower to /a/. Maybe the cluster was resolved through the deletion of /m/ under most conditions, and /q/ was deleted following /i/, resulting in the (V)t(V) form that appears under many conditions. Later, some other process might have been triggered by certain stems that wound up ending in /-l/, which got rid of /tm/ but left the lowered vowel /a/, allowing /q/ to survive. Under yet different conditions, only /m/ survived, turning into /w/ intervocalically.
That's pretty much all speculation on my part, however.