Field/Lexical Affixes in Your Polysynthetic Conlangs?

[Omzinesý]

How is the system of field/lexical affixes in your polysynthetic conlangs?
What meanings are coded in the same slot?

Mattissen's "A structural typology of polysynthesis"
https://www.tandfonline.com/doi/abs/10. ... 4.11432546 poses that they are the main feature making a lang polysynthetic. They are affixes whose meanings "a normal lang" expresses with adverbs.

Lately I've been thinking about lexical suffixes. What to expresses in the verb.

[Omzinesý]

My Dlor, which is not polysynthetic in the full sense of the concept but has some features, has prefixes expressing body parts, 'to do with hands', 'to do with mouth/teeth' ... If the word, accent falls on the prefix its function is instrument, if on the root, its function is object.

[VaptuantaDoi]

Ai is not really polysynthetic either, but it is fairly synthetic. It's got five directional prefixes (up, down, upstream, downstream, and towards/into) and some fairly lexical "aspect" prefixes which include some adverb-ish senses like "until" and a transformative. Sadly none of my other polysynthetic languages have really made it off the ground (e.g. Wåpwåpwohisi).

[Omzinesý]

[...] and some fairly lexical "aspect" prefixes which include some adverb-ish senses like "until" and a transformative.

Have you described them somewhere?

[VaptuantaDoi]

[...] and some fairly lexical "aspect" prefixes which include some adverb-ish senses like "until" and a transformative.

Have you described them somewhere?

No, not yet. I may expand this category before I write it up too.

[Pabappa]

My Lava Bed languages (the clade that enwraps Play) might qualify as polysynthetic, but for a long time I've focused on building the language for its own sake rather than seeking labels. I don't think Play has any fixed slots that serve as field or lexical affixes, so maybe it's not polysynthetic, but it has the unusual trait (rare even in Lava Beds) of having the 1>2 verb marker be a zero morph, so the sentences I've posted in a few recent threads come quickly to mind when speaking Play, not just for violent actions but also for affectionate ones (papim! "let me kiss you!") and those with complex semantics.

Poswa has what might be called field and lexical affixes (I admit I didnt really understand the PDF), but the slots aren't rigidly fixed like in some languages. For example there is an affix -i- that indicates an action cost money ("take" vs "buy" and many other verbs), but it can be affixed to any verb, not just to verbs that have a certain number of slots filled between the root and that affix.

Pabappa indicates the noun class of the patient on every verb, using a fusional infix that for all practical purposes means that the learners have to memorize every form for every verb. I'm making it difficult on purpose. And it also marks "purpose" on many of its inanimate nouns, essentially answering the question "what is this for?" Most of those are passive but some are active (e.g. "animal that scares me").

[Knox Adjacent]

Yeah I got them and slowly ever more so. Instrumentals, elevation, direction, time of day, a slot for incorporated adverbs/adverbial prefixes, I guess some aspect forms, such much. Honestly polarity and speaker evaluation might be one of the few things without an affixal option.

Bummer about Ainu. (Spot the Ainu-isms anyway in Knox Adjacent)

[Glenn]

I have read Mattissen's article before; thank you for giving me the opportunity to read it again!

I have a few scattered notes for a polysynthetic conlang that I first started years ago after learning about polysynthetic languages on the ZBB, with the placeholder name of "Northlander." (The speakers are, unsurprisingly, referred to as Northlanders; at one point, I began creating an endonym with the elaborate translation of "Those Who Dwell in the Shade of the Cedars," but it has been entirely lost, apart from the fact that it began with /tɬ/ and probably had at least two or three other laterals as well; the language was envisaged as lateral-heavy].)

In Mattissen’s terms, Northlander is intended to be “mixed II,” or possibly “compositional” (I see it as having incorporation of noun roots into the verb, but not necessarily verb roots, although I remember thinking that verb roots might be incorporated if they were nominalized first), agglutinative and “templatic” as opposed to “scope-oriented” (having clearly distinguished slots in a fixed order, partly because that was the first model I was introduced to, partly because I figured that it would be easier to keep track of.

Ironically, field/lexical affixes of the kind described above and in Mattisen’s article were almost completely undeveloped in Northlander, in part because I was initially focused on noun incorporation (initially without a real understanding of how noun incorporation actually works in practice) and polypersonal agreement, and in part because while I was aware of the existence of affixes having an adverbial function, I didn’t really know how to incorporate them, and I was not fully aware of their range of potential functions or what an important feature they can be in polysynthetic languages (as set forth in the article). If I ever resume work on Northlander, I will certainly need to bear this in mind.

[DV82LECM]

I will read this paper but, for now, I wanna explain Skasti, a bit. It is intended as polysynthetic but I'm not entirely sure what the difference between synthetic and polysynthetic really is. It allows three prefixal slots and six suffixal in its verbal complex. There is a tendency for heavy collapse of morpheme suffixes when slots are not utilized, rendering very structurally-ordered alterations that result in somewhat dense fusionality. Much meaning can be represented in what is typically 2-syllable morphemes. Because of historical sound changes, many negative forms show somewhat-radical phonetic divergence.

/-heʃtadwã-/ ("be fed")
/-heʃtaglõpã-/ ("not be fed")

/-heʃtastẽpã-/ ("have been fed")
/-heʃtaksudwã-/ ("not have been fed")

[Omzinesý]

and in part because while I was aware of the existence of affixes having an adverbial function, I didn’t really know how to incorporate them, and I was not fully aware of their range of potential functions or what an important feature they can be in polysynthetic languages (as set forth in the article). If I ever resume work on Northlander, I will certainly need to bear this in mind.

That is the hard part. I'm neither very well aware of what they could express and even less how to make them a harmonic system. The body part affexes are very handy but there could be much more.

The ZBB introduction to polysynthesis lacks them altogether.

[Imralu]

Thank you for the link to that article. It was an interesting read.

I'm currently working on a polysynthetic language which is tentatively called Miwosh or (Míwoš). It's a revamp of another language of mine, Balog, and the main idea is that I wanted to spoonerise almost all of the particles with the root so that they become -VC- infixes. I had previously decided that a dialect of Balog did this, but it's becoming its own language now and I'm putting it in a different universe. Although I'm carrying across most of the vocabulary, I'm changing it enough that I'm going quite slowly and thinking very carefully so that I make aesthetic decisions that I'm more happy with. Miwoš is a compositional polysynthetic language with hierarchical, recursive infixing, scope-ordered (non-templatic) organisation and incorporation of lexical roots (i.e. compounding).

How is the system of field/lexical affixes in your polysynthetic conlangs?
What meanings are coded in the same slot?

I can't really answer that yet as (a) I don't have all the morphemes yet and, more importantly, (b), there are no slots as it's not templatic but has hierarchichal, scope-ordered organisation with the capacity for recursivity within words.

E.g.

Lizobíyouhāž.
l<iz><ob><iy>ouh-ā-ž
sleep<PST><want><INCEP>-SBJ.R2-be.1S
I wanted to fall asleep.

Liziyóbouhāž.
l<iz><iy><ob>ouh-ā-ž
sleep<PST><INCEP><want>-SBJ.R2-be.1S
I got tired. / I started to want to sleep.

Lizobiyóbouhāž.
l<iz><ob><iy><ob>ouh-ā-ž
sleep<PST><want><INCEP><want>-SBJ.R2-be.1S
I wanted to get tired. / I wanted to start to want to sleep.

(The stress would ordinarily be on the "ou", but both líyouh 'fall sleep' and lóbouh 'be tired' have lexicalised.)

Here is how Miwoš falls according to Johanna Mattissen's classification in table 3 (with the final parameter taken from table 2):

Code: Select all

Non-root bound morphemes:       +
Scope ordered:                  +
Templatic:                      -
Noun incorporation*             +
Verb serialisation*             +
Roots per verb form:            ≥1

According to Mattissen's table, that puts it with Pano, which I'm assuming refers to the Panoan languages as a whole (or possibly specifically Pánobo) rather than the extinct Pano language.

Non-root bound morphemes are mostly infixes that are inserted after the first consonant in the base. There are also a few suffixes such as directionals. Stress falls on the first vowel of however far left the base has been lexicalised, although some infixes contrast in stress, for example the causative infix <úw> must be stressed to distinguish it from the always unstressed perfect/retrospective infix <uw>.

* Because there is no lexical noun-verb distinction, as with virtually all of my conlangs, saying that it has both noun-incorporation and verb-serialisation is a bit misleading as they're the same thing. Bases can be modified by other bases, introduced by one of three modifier interfixes:

Code: Select all

        | interfix      | gloss         | meaning                                       |
        |———————————————|———————————————|———————————————————————————————————————————————|
        | -o-           | LK1           | Modifier and head refer to the same entity    |
        |———————————————|———————————————|———————————————————————————————————————————————|
        | -e-           | LK2           | Modifier and head refer to different entities |
        |———————————————|———————————————|———————————————————————————————————————————————|
        | -wē-          | ATTR          | Head also [modifiers]s                        |
        |———————————————|———————————————|———————————————————————————————————————————————|

Examples:

Code: Select all

         qúsum -o-  múkan                                  tsál-o-  lóuh
         be.oil-LK1-be.coconut                             try -LK1-sleep
        "be coconut oil"                                  "try to sleep" 
         (i.e. "kind of oil which is coconut")             (the tryer is/will hopefully be the sleeper)

         qúsum -e-  wáwak                                  váŋ -e-  ším
         be.oil-LK2-be.baby                                hunt-LK2-be.fish
        "be baby oil"                                     "fish"/ "be a fisherman" 
         (i.e. "kind of oil associated with babies")       (the hunter is not the fish)

         qúsum-wē-xed (→ qúsumwexxéd)                      vaŋ-wē-ŋaž (→ váŋweŋŋáž)
         be.oil-ATTR-be.hot                                hunt-ATTR-be.tall
        "be hot oil"                                      "be a tall hunter"
         (i.e. "oil that is hot", not a different kind of oil) 

The attributive interfix is not used all that often as adjectival kind of meanings are usually just expressed with an extra clause, the rank of the subjectifiers (e.g. here R5) unambiguously indicating which referent is intended. Transitivity is also usually expressed across two clauses.

Code: Select all

         Múlāž témiŋŋúm hómī.
         múl      -ā-     ž     | tém      -ī-     ŋúm       | hóm-ī.
         use.mouth-SBJ.R2-be.1S | get.eaten-SBJ.R5-be.cattle | be.tasty-SBJ.R5
         I use my mouth         | beef gets eaten            | it is tasty
        "I eat tasty beef."

It is, however, possible to put it into one clause.

Code: Select all

         Túwemeŋúmwehhómāž.
         t<úw>ém        -e-  ŋúm      -wē-  hóm     -ā-     ž
         get.eaten<CAUS>-LK2-be.cattle-ATTR-be.tasty-SBJ.R2-be.1S
         "I tasty-beef-eat."

As in the Sora language (Munda family), subject marking is not limited to pronominal markers (indicating only person, number, gender etc.), but a full lexical subject can also be incorporated.

Code: Select all

         Túwemeŋúmwehhómavváŋeším ŋážā.
         t<úw>ém        -e-  ŋúm      -wē-  hóm     -ā-     váŋ -e-  ším     | ŋáž    -ā
         get.eaten<CAUS>-LK2-be.cattle-ATTR-be.tasty-SBJ.R2-hunt-LK2-be.fish | be.tall-SBJ.R2 
         The fisherman tasty-beef-eats                                       | he is tall 
         "The tall fisherman eats tasty beef."

In this case, for the time being, the information that the fisherman is tall could not be added to the first clause because it would have to refer back to the fish rather than the hunter, although I will develop a "skip" morpheme that allows modifiers to relate to the entire base before it.

[lsd]

I guess 3SDL is a polysynthetic language
(in addition to being oligosynthetic)
since it's a one-word language...

probably compositional according to Matissen,
there are serial verbs,
adverbs, especially quantitative ones, reduced to bound morphemes,
everything else is a concatenation of affixed free lexical morphemes...
probably Wichita kind...

[Pabappa]

* Because there is no lexical noun-verb distinction, as with virtually all of my conlangs, saying that it has both noun-incorporation and verb-serialisation is a bit misleading as they're the same thing. Bases can be modified by other bases, introduced by one of three modifier interfixes:

Code: Select all

        | interfix      | gloss         | meaning                                       |
        |———————————————|———————————————|———————————————————————————————————————————————|
        | -o-           | LK1           | Modifier and head refer to the same entity    |
        |———————————————|———————————————|———————————————————————————————————————————————|
        | -e-           | LK2           | Modifier and head refer to different entities |
        |———————————————|———————————————|———————————————————————————————————————————————|
        | -wē-          | ATTR          | Head also [modifiers]s                        |
        |———————————————|———————————————|———————————————————————————————————————————————|

Examples:

Code: Select all

         qúsum -o-  múkan                                  tsál-o-  lóuh
         be.oil-LK1-be.coconut                             try -LK1-sleep
        "be coconut oil"                                  "try to sleep" 
         (i.e. "kind of oil which is coconut")             (the tryer is/will hopefully be the sleeper)

         qúsum -e-  wáwak                                  váŋ -e-  ším
         be.oil-LK2-be.baby                                hunt-LK2-be.fish
        "be baby oil"                                     "fish"/ "be a fisherman" 
         (i.e. "kind of oil associated with babies")       (the hunter is not the fish)

         qúsum-wē-xed (→ qúsumwexxéd)                      vaŋ-wē-ŋaž (→ váŋweŋŋáž)
         be.oil-ATTR-be.hot                                hunt-ATTR-be.tall
        "be hot oil"                                      "be a tall hunter"
         (i.e. "oil that is hot", not a different kind of oil) 

This part looks similar to some of my languages .... i have equative compounds and attributive compounds .... so it's good to know I've hit on an idea that works and that others use.

Code: Select all

         Túwemeŋúmwehhómavváŋeším ŋážā.
         t<úw>ém        -e-  ŋúm      -wē-  hóm     -ā-     váŋ -e-  ším     | ŋáž    -ā
         get.eaten<CAUS>-LK2-be.cattle-ATTR-be.tasty-SBJ.R2-hunt-LK2-be.fish | be.tall-SBJ.R2 
         The fisherman tasty-beef-eats                                       | he is tall 
         "The tall fisherman eats tasty beef."

In this case, for the time being, the information that the fisherman is tall could not be added to the first clause because it would have to refer back to the fish rather than the hunter, although I will develop a "skip" morpheme that allows modifiers to relate to the entire base before it.

I'd like to see this. In Play the solution for a similar problem is a series of morphemes ending in -(v)e that reorient the arguments of the following word. I'd like to keep the system clean in the sense that each of these morphemes only affects the immediately following word, and after that the arguments reset.

In Dreamlandic, a hatelang, a similar problem arises, which I will deliberately leave unsolved. Dreamers simply have to repeat every word every time it is used as an argument of any verb, or at least use a word that describes the object (e.g. "fish" instead of "tunafish" would work).

Very nice work. I actually like the term hatelang myself, as it's short and sweet, but every hatelang I work either turns into a lovelang (e.g. Dreamlandic) or becomes such a chore that I refuse to work on it (Thaoa, Tarise, etc).

For what it's worth, the Dreamlandic I was speaking of then and the one I'm working on now are probably not the same language, but I cant be sure. In both cases Im using the name of a family metonymically for a language within it, and the family broke apart ~4000 years before the focus time, so the languages differ quite a lot.

[Imralu]

Code: Select all

         Túwemeŋúmwehhómavváŋeším ŋážā.
         t<úw>ém        -e-  ŋúm      -wē-  hóm     -ā-     váŋ -e-  ším     | ŋáž    -ā
         get.eaten<CAUS>-LK2-be.cattle-ATTR-be.tasty-SBJ.R2-hunt-LK2-be.fish | be.tall-SBJ.R2 
         The fisherman tasty-beef-eats                                       | he is tall 
         "The tall fisherman eats tasty beef."

In this case, for the time being, the information that the fisherman is tall could not be added to the first clause because it would have to refer back to the fish rather than the hunter, although I will develop a "skip" morpheme that allows modifiers to relate to the entire base before it.

I'd like to see this. In Play the solution for a similar problem is a series of morphemes ending in -(v)e that reorient the arguments of the following word. I'd like to keep the system clean in the sense that each of these morphemes only affects the immediately following word, and after that the arguments reset.

In Dreamlandic, a hatelang, a similar problem arises, which I will deliberately leave unsolved. Dreamers simply have to repeat every word every time it is used as an argument of any verb, or at least use a word that describes the object (e.g. "fish" instead of "tunafish" would work).

OK, I've decided on the SKIP morpheme now. It's -a-.

It interacts with the other interfixes like this:

Code: Select all

-a- + -o-  = -au-   -SKIP.LK1-
-a- + -e-  = -ai-   -SKIP.LK2-
-a- + -uē- = -awē-  -SKIP.ATTR-

Pretty predictable, really!

Bracket notation of a base, without and then with the "SKIP" morpheme.

Code: Select all

váŋešímueŋŋáž
[vaŋ  [-e-   [šim     [-uē-   [ŋaž    ]]]]]
[hunt [-LK2- [be.fish [-ATTR- [be.tall]]]]]

= to hunt tall fish 
= to fish for tall fish 
= to be a hunter of tall fish 
= to be a tall-fish ... hunter

Code: Select all

váŋešímaweŋŋáž
[vaŋ  [-e-   [šim    ]] [-a-    [-uē-   [ŋaž    ]]]]
[hunt [-LK2- [be.fish]] [-SKIP- [-ATTR- [be.tall]]]]

= to hunt tall fish and be tall
= to fish and be tall
= to be a tall hunter of fish
= to be a tall fish-hunter

I wasn't quite sure whether to bracket -a- as I have done, as something that lifts the following phrase up higher in the tree, or to put it on the end of the first phrase, basically as something that brings about the closing of some brackets:

Code: Select all

[vaŋ  [-e-   [šim    [-a-]]]     [-uē-   [ŋaž    ]]]

... but the end effect is the same either way.

So, without further ado: Túwemeŋúmuehhómavváŋešímaweŋŋáž!

Code: Select all

         Túwemeŋúmuehhómavváŋešímaweŋŋáž.
         t<úw>ém        -e-  ŋúm      -uē-  hóm     -ā-     váŋ -e-  ším    -a-  -uē-  ŋáž
         get.eaten<CAUS>-LK2-be.cattle-ATTR-be.tasty-SBJ.R2-hunt-LK2-be.fish-SKIP-ATTR-be.tall-SBJ.R2 
         "The tall fisherman eats tasty beef."

That's not a natural sentence though and, if you really had to say all of that information in one sentence, it would probably be said something more like: Múlavváŋešímueŋŋáž témiŋŋúmuehhóm. (Lit.: The tall fisherman uses his mouth. The tasty beef gets eaten.) Or, even more likely: múlavváŋeším témiŋŋúm hómī ŋážā. (Lit.: The fisherman uses his mouth. The beef gets eaten. It is tasty. He is tall.) It really depends what you're trying to say and it's pretty rare that so much gets said in one sentence in any language. At least one of these referents has probably already been introduced and then you could just say something like Múlā témī hómī. (He uses his mouth. It is eaten. It is tasty.) ... or Ŋážattúwemeŋúm. (The one eating the beef is tall), depending on what information is assumed to be known and the focus of the utterance.