Conclusion: Describing Koasati’s Semantic Alignment System Part 2

Continued from Part I, where I give an overview of the different types of agreement and their uses.

Type I and Type II

Although most active verbs take Type I agreement stative verbs take Type II agreement, this is clearly not the factor which determines which verbs take which agreement.  As I have shown in (2) and (5) in Part I, there are in fact a number of stative verbs which take Type I agreement and active verbs which take Type II agreement respectively.  The only semantic distinction that accounts for most of the variation between Type I and Type II agreement is control.  Verbs like “to eat,” “to buy,” “to be prepared,” and “to sing” are things that the subject has clear control over, so they take Type I agreement.  However, verbs like “to fall,” “to bleed,” “to be short,” and “to be clean” are all things that are mostly out of the direct control of the subject; thus they take Type II agreement.

We find even more evidence for agreement selection based on control when we examine verbs that can take either Type I or Type II agreement.  Many verbs fall into this category; most, if not all, of these are either stative verbs or intransitive active verbs:

(10) Verbs taking Type I and Type II agreement

Afolohkalit or Achafolohkat                           “I yawned”

Hifoskal or Chahifosk                                    “I’m breathing”

Nochilit or Chanochit                                     “I slept”

Afaakal or Achafaak                                       “I’m laughing”

Chahakbisk or Hakbiskal                               “I’m sneezing”

Chahikohk or Hikohkal                                  “I’m hiccupping”

Yahkalit or Chayahkat                                    “I cried”

Salatlilit or Chasalatlit                         “I slid”

Akostiniichilis or Achakostiniichis                 “I thought”

Limitkal or Chalimitk                                     “I’m swallowing”

Chalkafihlis or Ilkafihlilis                               “I was strong”

Imathakilis or Chaimathakis                           “I was different”

Chapalkahosis or Palkahosilis                        “I was fast”

Chakappalosit or Kappalosilit            “I was quiet”

Chahimathaahosis or Himathaahosilis            “I was loud”

The semantic difference between the cha- and –l forms of these verbs is very subtle.  When I asked speakers about what the difference is between the cha- and –l forms of many verbs, they even said that there was no difference in meaning, only that they’re used in different contexts.  So to try to get at the precise semantic difference, I elicited the verbs in different contexts, some implying a non-controlling subject and others implying a controlling subject, and observed which agreement type the speaker used.

For example, the verbal suffix /–honk/, meaning “to do something accidentally” strongly implies that the subject was not in control of the action that they performed.  Thus, verbs that can take Type II agreement always did when elicited with this suffix:

Chalimitkahonk                                  “I swallowed it on accident”

Chatammihonk                                   “I fell on accident”

Chanochitahonk                                 “I fell asleep on accident”

Similarly, the suffix /–aapis/, meaning “to almost do something,” also implies that the subject was not in control of the action that they performed in many contexts.  Thus it can also be used to elicit the cha- forms of verbs:

Chanochaapiis                                    “I almost fell asleep”

Chatatammaapiis                                “I almost fell”

Chalimitkaapiis                                   “I almost swallowed it”

Stachombatikaapiis                             “I almost got stuck”

To imply a controlling subject, on the other hand, I asked speakers to imagine a context in which they are actors in a play, and the director tells them to do something, and then they do that action.  In this scenario, even actions such as yawning, sneezing, and laughing, which are usually done without the control of the subject, are controlled.  And, as expected, this scenario always led to Type I agreement:

Afaakallahon ammahkan afaakalit                               “He told me to laugh, so I laughed”

Ammankat hakbiskallahon akamis hakbiskalit            “He told me to sneeze, so I sneezed”

Stimaatik ammahkan hafolohkalis                               “The director told me to, so I yawned”


Type II and Type III

The semantic difference between Type II and Type III agreement is an even more subtle one than that between Type I and Type II agreement, if there is any difference at all.  Type II agreement is much more common than Type III agreement, so it’s possible that the words requiring Type III agreement are more or less exceptions that just need to be memorized.  However, there are two verbs that can take either Type II or Type III agreement and have different meanings:

Chahoopat                              “I was sick”

Ahoopat                                  “I was hurt,” “I was injured”

Chakaan                                 “I am well,” “I am good”

Ankaan                                   “I have an orgasm” (Kimball 1991, p. 254)

Kimball posits that the distinction is based on time: Type II agreement for states that are more permanent, and Type III agreement for states that are mainly temporary.  D. Hardy (1988), when describing a similar system in the related language Creek, claims that the distinction between Creek’s Type II and Type III agreement is more based on affectedness, or as he describes it “envelopment.”  This means more or less that Type II agreement indicates that the subject is more fully affected by the state described in the verb.  To me, this classification seems better fit to describe Koasati than Kimball’s description; both “being sick” and “being hurt” are temporary states of existence, but “sick” implies being more fully affected than just “hurt.”



Like all grammatical phenomena, semantic alignment is somewhat messy, with many subtleties and exceptions.  The semantic distinctions that languages can make are practically limitless, and as a result pinning down precisely what distinction is being made where is challenging.  However, after an in-depth analysis of semantic alignment in Koasati, I am able to confidently claim that the choice between Type I and Type II or III agreement is primarily based on control.  The precise distinction between Type II and Type III agreement is less clear because there are so few examples of contrast between those two sets of agreement markers, but I can at least say that it seems to be based on affectedness.

Kimball, G. (1991). Koasati grammar. Lincoln, Nebraska: University of Nebraska Press.

Hardy, D. E. (1988). The semantics of creek morphosyntax (Doctoral).