2.1 Examples of the most common uses

In addition to the core semantic function of the frustrative (“non-realization of some expected outcome”), Overall also identifies a number of “extended” (i.e., secondary) functions that are frequently expressed by frustrative markers. He shows that within his sample of Amazonian languages, different sets of functions are attested in different languages. (This is what we would expect if the core semantics of the frustrative marker varies somewhat from language to language.) A number of these secondary functions are also associated with the Kimaragang frustrative, as illustrated in this section.

The Kimaragang examples in (3a,c) illustrate the classic frustrative pattern: actions which are performed but do not have the desired or expected effect, i.e., actions performed in vain. Example (3b) is very similar except that the dara clause describes a state rather than an event. As Overall observes, frustratives often occur in biclausal structures in which the unmarked clause expresses either an explicit statement of non-realization, as in (3a), or the cause of the non-realization, as in (3b,c). (In the following examples I will use init to represent the proposition asserted in the dara clause, which corresponds to the initiating circumstance, and exp to represent the expected outcome.)

An important secondary function of the frustrative is the discontinuous past, or cessative, which indicates that a situation was true in the past but does not persist into topic time (Plungian & van der Auwera 2006). In Kimaragang, this use of dara occurs primarily with stative predicates, as illustrated in (4).

Overall combines into a single functional category the “Incompletive” use, marking an event which is begun but not completed, and the Avertive (or “Action Narrowly Averted”; Kuteva 1998; 2000), which marks an action that was on the verge of happening but didn’t. He points out that these two uses are sometimes difficult to distinguish, e.g. ‘about to fall’ vs. ‘started to fall’. Notice that the avertive reading involves interpreting the non-past verb form as prospective aspect (‘about to X’). (There is no morphological marker for prospective aspect in Kimaragang.) As discussed in section 3, I take this to be a kind of coercion effect.

Overall treats statements of frustrated intention as falling within this same category. Like the avertive reading, frustrated intention derives from prospective aspect: when a volitional action is described using prospective aspect (e.g., ‘about to eat’), it is often interpreted as a statement of intention. The use of dara to express frustrated intention is quite common in Kimaragang.

Overall lists seven languages in his Amazonian sample set (out of a total of 54 languages) that use the frustrative to mark counterfactual conditionals. Kimaragang dara shares this function as well.⁴ As these examples illustrate, dara in counterfactual conditionals can occur either in the antecedent (conditional clause), as in (8b), or in the consequent (main clause), as in (8a).⁵ This appears to be the only use of dara in which the particle can potentially occur in either clause without changing the meaning of the sentence. Some other discourse-type particles in Kimaragang are similar to dara in this respect: they make the same contribution to the meaning of the sentence whether they occur within the main clause or within a subordinate clause. In light of this fact, it seems plausible to assume that dara takes scope over the whole conditional sentence, as suggested by an anonymous reviewer.

It is not immediately obvious how to derive the counterfactual function from the definition of dara proposed in (2). It may ultimately be necessary to treat this function as a polysemous sense of dara, but for the purposes of this paper, I will tentatively assume that when dara occurs in a conditional construction, the relevant outcome is (by convention?) identified with the consequent (apodosis) of the conditional. Under this assumption, a sentence of the form “If A then B dara” asserts that the conditional statement If A then B is true, and (in most contexts) implicates that B is false. By the rule of modus tollens, this would imply that A is also false, giving rise to the counterfactual interpretation.

In contexts where the implicature that B is false does not arise, we would predict a hypothetical reading: the truth of B cannot be asserted at the time of speaking. Consistent with this prediction, it turns out that dara can sometimes occur within a hypothetical conditional, although such examples are rare. Kroeger (2017) indicates that the hypothetical reading for (9) is more prominent, but the counterfactual reading is also possible, depending on context.

Overall describes an evaluative function, in which the frustrative marker is used to indicate that the described situation is unsatisfying, disappointing, or annoying to the speaker, even in contexts where there is no sense of unrealized expectation. This usage seems to be fairly widespread among the languages in his sample, but (so far) is not attested in Kimaragang.⁶

On the other hand, the Kimaragang frustrative is used in some ways that are not mentioned by Overall. The most productive of these is the optative/desiderative function. In most examples of this type the frustrative particle dara reinforces the meaning of a lexical predicate or derivational morpheme which expresses the optative or desiderative meaning, as illustrated in (10a-b). However, it is also possible for dara to function as the only marker of optative or desiderative meaning, as seen in (10c). The presence of dara in these examples marks them as being hypothetical. Sentence (10a) is judged to be simply an expression of the speaker’s opinion (“Wouldn’t it be nice if…”), whereas the same sentence without dara is heard as a definite suggestion that something be done. Sentence (10c) is heard as an expression of the speaker’s desire or hope, which may or may not be fulfilled, whereas the same sentence without dara is heard as a statement of intention (‘I will borrow his video machine’), as if the speaker has a definite plan to borrow the device. The desiderative prefix ti- cannot be used to express a definite plan, and as a result sentence (10b) sounds unnatural if dara is omitted.

The use of dara to mark desires is discussed in more detail below. The key idea is that when a volitional agent expresses a desire to do something within his power, under normal circumstances this will create an expectation that the action will in fact be carried out.

As illustrated in examples (11a-b), dara also functions as a politeness marker which makes a request less coercive. This use may well have developed as a conventionalized statement of frustrated desire, e.g. in example (11a): ‘I would like to get you to bring me some tobacco dara, (but I am too shy to ask)’.

Examples (11a-b) with dara are interpreted as polite requests, but if dara is omitted they are interpreted as orders or demands: the addressee is not free to say “No”. My consultant used the Malay phrase tidak boleh tidak ‘cannot not (do it)’ to express the force of these utterances without dara. Hale (1969: 206) notes a very similar “polite desiderative” use of the Tohono O’odham frustrative particle cɨm. Sharon Estioca and Bill Hall (p.c.) inform me that the Subanon frustrative marker bosia also has this function.

Finally, dara occasionally seems to function as a marker of non-universality, indicating that the proposition asserted is not in fact true for all the members of a relevant topic set. This function is illustrated in (12a-b):

2.2 The contribution of dara

As mentioned in section 1, I follow Davis & Matthewson in arguing that the “frustrated expectation” meaning associated with dara is “not at issue”; it does not affect the truth conditional meaning of the clause in which it occurs. Indeed, for some examples, such as (3a) and (3c), leaving dara out appears to make only a subtle difference to the meaning of the sentence, without reducing its naturalness or acceptability. This is not true in all cases, however.

My consultant states that example (3b) (‘We love each other dara but her father doesn’t like me’) clearly implies that the couple does not plan to marry (or at best that the marriage plans are on hold). If dara is omitted, however, it suggests that the couple does plan to marry in spite of the father’s opposition. My consultant says that this second version is fully grammatical, but it strikes him as something no one in that culture would ever say. Notice that this difference in meaning is not truth-conditional. The at-issue propositional content is the same in both versions: ‘We love each other and her father doesn’t like me.’

For discontinuous past examples like those in (4), dara is strongly expected. Omitting dara does not typically seem to affect the meaning of the sentence but does degrade naturalness. For example, my consultant says that (4c) (‘There was a lot of fruit on Soyinsin’s mango tree dara, but they all fell off before they got ripe.’) is perfect; but if dara is omitted, the sentence would be only marginally acceptable. Similar judgments are reported for the “Action Narrowly Averted” usage illustrated in (6).

The presence of dara does affect the interpretation of the sentence in example (7a), which literally says: ‘The cat licked the plate that I ate from, when I still eat dara.’ As discussed in more detail in section 3.2, the atemporal verb form in the first clause is used here in its “narrative tense” function (Kroeger 1991), which establishes a topic time in the past for the sentence as a whole. In this context the non-past verb in the dara clause normally gets a prospective aspect reading (‘when I was still about to eat more’ or ‘when I still intended to eat some more’). The sentence is interpreted as describing a situation where the speaker got up and left his plate for some reason, planning to return and continue eating. The cat licked the plate while he was gone. When dara is absent, however, prospective aspect is disfavored, and the most natural interpretation of the non-past verb form mangakan ‘eat’ is ‘action in progress’: the cat licked the plate while the speaker was still eating. This interpretation seems to be facilitated by the fact that in this example the verb mangakan ‘eat’ occurs within a circumstantial adverbial clause, where the non-past verb form can have a relative present tense interpretation to indicate simultaneity with a past event.

In other examples involving the frustrated intention usage, e.g. (7b), omitting dara results in an unnatural or unacceptable utterance. Without dara, the first clause in (7b) is most naturally interpreted as a simple statement of present intention (‘I will not take the money he gave me’), but the past tense verb form in the second clause seems to require a topic time in the past for the sentence as a whole, and renders the sentence incoherent. We will discuss the role of tense and aspect in the interpretation of dara in more detail in section 3.2.

Dara can sometimes occur in monoclausal sentences as well. The most common utterances of this type involve the optative/desiderative use (ex. 10; cf. sec. 3.1) or polite requests (ex. 11), but this is also possible with other uses of dara. In some such examples, like those in (13–14), the presence of dara does not change the truth-conditional meaning of the sentence but implies that there is more to be said.

In other cases, like those in (15–16), the presence of dara does correlate with a change in the truth-conditional meaning of the sentence. When dara co-occurs with a non-past verb form in a context where there is no indication of a frustrating circumstance, the most natural interpretation would be the optative: something the speaker hopes for. In sentences like (15b) and (16b), however, the optative reading (e.g., ‘I hope to be bitten by the dog.’) would be highly implausible. As discussed in section 3.2, such utterances may trigger a coerced shift in the time reference of the verb, from non-past to past tense with prospective aspect. Under this interpretation, (15b) literally says ‘I was about to be bitten by the dog dara’, and (16b) literally says ‘My car was about to fall down the hill dara’.

In summary, there are many cases where omitting dara merely renders the sentence unnatural, but in other cases this omission appears to trigger a change in the meaning of the sentence. Such changes are often due to coercion effects, as discussed in the following section.

3.1 Extending the analysis to Kimaragang

Davis & Matthewson suggest that essentially the same analysis could work for Kimaragang dara, as well. In doing so, they note that examples like (19) pose a problem for their hypothesis, because the “prejacent” or base proposition p (literally ‘I will kill that cobra’) is apparently not asserted to be true:

Kroeger (2017) proposed an analysis for such examples which assumed a shift in the meaning of Kimaragang tense morphology: past vs. non-past in most contexts, realis vs. irrealis in clauses marked by dara.¹⁰ Davis & Matthewson note that in analogous St’át’imcets examples, the verb would be overtly marked for prospective aspect. They suggest that a better analysis of the Kimaragang examples can be achieved by assuming that in frustrative clauses, the Kimaragang non-past verb form gets interpreted as prospective aspect, as suggested by the free translation of (19). Under this hypothesis, the asserted base proposition in (19) is ‘I was about to kill’, while the expected outcome that remains unrealized is the actual killing, and dara contributes its normal frustrative interpretation. This same reasoning applies to the avertive (“action narrowly averted”) examples in (6) above.

This is a good suggestion, which provides a good analysis for the avertive and frustrated intention uses of dara, and I adopt this proposal going forward. However, more research is needed concerning the availability of prospective aspect in Kimaragang. Davis & Matthewson (p. 1387) state: “If Kimaragang unmarked eventive predicates allow future time reference, then dara-clauses with these unmarked predicates could in effect be parallel to St’át’imcets séna7-clauses with prospective aspect cuz’.” In other words, they suggest that in general, Kimaragang eventive predicates allow either future or prospective readings in the absence of past tense marking. In fact, while the non-past verb form does normally allow a future tense reading, it appears that the prospective aspect reading is available only under certain circumstances, which I do not yet fully understand. In the following section I offer some brief comments on tense and aspect in Kimaragang, and how these categories affect the interpretation of dara.

3.2 Tense, aspect, and the interpretation of dara

Examples (15–16) and (19) illustrate how the presence of dara can license not only the prospective aspect interpretation but also a shift in time reference, from non-past to past. The use of the non-past (i.e., unmarked) verb form in these examples to refer to a past time is somewhat surprising. Example (20) demonstrates that in a simple main clause, the non-past verb form is compatible with present or future time reference, but not with past time reference.

How then can we explain the “prospective aspect in the past” interpretation for the non-past verb in examples like (19)? In most cases, the critical factors seem to be (a) the tense marking of other clauses in the sentence, and (b) contextual indicators that the expected outcome was not realized.

According to the first clause of Davis & Matthewson’s proposed definition, p dara asserts that p is true. Now the basic, literal meaning of the initial clause in example (19) is ‘I will kill that cobra’, but the continuation of the sentence makes it clear that this proposition is not true. The sentence can only be made non-contradictory by assuming that this literal meaning is not the base proposition p which dara operates on. The past tense form in the continuation also indicates a topic time in the past for the sentence as a whole. These contextual constraints force a shift in the temporal interpretation of the first clause, resulting in: ‘I was about to kill the cobra’. This is the proposition p that is asserted to be true. The salient expected outcome that fails to be realized is that the cobra will be killed.

This kind of context-driven shift in aspectual interpretation is often referred to as coercion (Moens & Steedman, 1988). De Swart (1998) describes such shifts as follows:

Typically, coercion is triggered if there is a conflict between the aspectual character [i.e., Aktionsart–PK] of the eventuality description and the aspectual constraints of some other element in the context. The felicity of an aspectual reinterpretation is strongly dependent on linguistic context and knowledge of the world.

Since there is no grammatical marker for prospective aspect in Kimaragang, this reading arises primarily through coercion. To help clarify the role of coercion in frustrative constructions, let us briefly consider some coercion effects which can be triggered by a different particle, the “completive” aspect marker no. Kroeger (2021 ms.) describes the aspectual functions of no as being very similar to those of the sentence-final particle le in Mandarin, which Olsson (2013) refers to as a iamitive.¹¹ Some of the most common uses of the completive particle are illustrated in the examples below.

Soh & Gao (2006) and Soh (2009) propose that Mandarin sentence-final le marks “Change of situation”. More specifically, they propose that a sentence of the form p le (i) asserts that p is true at topic time, and (ii) presupposes that p was not true prior to topic time. Let us assume that this is also the core meaning of Kimaragang no. When no combines with a stative predicate, as in (21b), the proposition (‘we are out of water’) is asserted to be true now, and presupposed not to have been true earlier. The result is an inchoative (change of state) reading, ‘We have run out of water’.

When no combines with a non-past verb describing a process, as in (22b), at least two readings are possible: inceptive ‘It has started to rain,’ or (by coercion) prospective aspect ‘It is about to rain.’ The first reading takes the rain event itself as being the new situation: ‘it is raining now, but was not raining before’. The second reading takes the immediate “pre-time” of the rain event as being the new situation: ‘it is about to rain now, but was not about to rain before’. The choice between these two readings will depend on context. Additional examples involving prospective aspect are seen in (15a) and (23).

Another fairly common coercion effect that can arise when non-past eventive verbs are followed by the completive aspect particle is “newly formed intention”: the agent did not intend to perform the action before, but now he does.¹² This is illustrated in example (24), taken from an origin myth about the rock formations on the top of Mt. Kinabalu.¹³ A man climbs to the top of the mountain to bring back the soul of his dead wife. He threatens the guardian spirits and begins to chop holes in the mountain top until they relent and agree to send her back. Here the meaning of the clause without the completive particle would be ‘we will cause your wife to return’. When combined with completive aspect, the clause is interpreted as expressing a change of heart: the guardian spirits are now willing to release the wife and send her home, which was previously not the case.

Notice that in these examples where prospective aspect is triggered by completive no, there is no shift in time reference: for all of these sentences the topic time remains in the present. Only when dara is also present, as in (15b) and (25), do we find topic time in the past.

However, I suggest that dara itself does not require this coerced shift in time reference. No such shift is found in statements of wishes or desires, like those presented in (10) and in section 3.3, even though they are marked by dara. One important factor in triggering this shift appears to be tense marking in non-dara clauses. The dara clause in avertive and frustrated intention examples generally contains a non-past verb form, which allows the prospective interpretation. But if the sentence contains another clause expressing the frustrating circumstance, that clause is generally expressed in the past tense, as in exx. (6, 7b, 19), or with the atemporal form of the verb used in its “narrative tense” function, as in exx. (7a) and (25). Both of these forms indicate topic time in the past.

As de Swart pointed out, coercion effects are determined not just by the presence of a particular grammatical morpheme, but also by contextual factors and knowledge of the world. Thus even in monoclausal sentences like (15b) and (16b), where the dara clause contains an eventive non-past verb, if something in the context makes it clear that the relevant expectation was not realized, then the non-past verb is quite likely to be interpreted with past time reference. The prospective aspect reading is clearly more accessible when dara is present, as seen by the unacceptability of (26b) and (27b). However, occasionally this reading is possible even without dara, as illustrated in (28), although such examples seem to be uncommon.

Prospective aspect provides a good analysis for the avertive use of dara, but it may seem less plausible as an analysis for examples like (29). Because the circumstances (specifically, in this case, the lack of rattan) make the described event impossible, it seems a bit odd to say that the event is “about to happen”.¹⁴ The relevant interpretation here has to do with the agent’s intentions. Notice that in this example, as in the other examples of frustrated intention presented in (7), there is no overt word or morpheme that expresses desire or intention; the intention is inferred by coercion.

As discussed in Kroeger (2017), the basic (volitive or “neutral”) past tense verb form entails that the action was initiated, or at least attempted. For this reason, dara clauses that contain past tense eventive verbs typically express either the core frustrative function (action done in vain) or the incompletive function. Dara clauses that contain non-past eventive verbs typically express the avertive and frustrated intention functions.

3.3 A problem for the analysis

The frustrative particle can be used to mark frustrated desires, not only in Kimaragang but also in St’át’imcets. Some Kimaragang examples are provided in (30). No coercion effect is needed to explain these examples, because the component of desire is generally made explicit with either a lexical predicate (as in 30a) or a derivational affix (e.g. desiderative ti-, as in (30b), or petitive poki- as in (30c).¹⁵

In Kimaragang however, unlike St’át’imcets, the frustrative particle is also used to mark desires which are not known to be frustrated. These sentences describe desires which are unfulfilled as of the time of speaking, but could well be fulfilled at some future time. In such examples dara can function as the only indicator of desiderative force in the clause, as in (31a), but more often it co-occurs with either a lexical predicate (31b,c) or a derivational affix expressing desire.

Such examples suggest that Kimaragang dara does not always have a strictly frustrative meaning. We address this issue in the next section.

5.1 Epistemically inaccessible outcomes

Example (35) poses a problem for the hypothesis that the interpretation of the Kimaragang frustrative involves purely epistemic expectations. The frustrated desire expressed by the speaker in this case is the natural desire not to die. But since we all die, there is no epistemically accessible world in which this desire is attained. For this example the frustrated expectation must be defined in terms of circumstantial (specifically bouletic) modality: the possible worlds which must be considered are those in which relevant desires are fulfilled. In other words, given the base proposition ‘I don’t want to die’, the relevant outcome is: ‘I will not die’. This outcome will be realized in all possible worlds/situations in which the speaker’s desires are fulfilled, and dara is used here to indicate that the real world is not one of those worlds. Epistemic modality will not work here, because the relevant outcome will not be realized in any possible world which is consistent with what the speaker believes about the real world.

Davis & Matthewson (p. 1343, ex. 11) cite a St’át’imcets example in which séna7 is used to mark an unfulfilled desire: ‘I really want to go and lay out in the sun for a while’ (Context: the weather is beautiful but the speaker has to write a paper). Davis & Matthewson argue that such examples can be accounted for, under their hypothesis that the frustrative meaning in St’át’imcets involves purely epistemic expectations, by invoking the principle that an agent’s desires can create the expectation that an action will be carried out:

For (11) and other similar cases, we assume that the expected outcome of a mental attitude of desire is that the desired situation obtains. Copley and Harley (2014) achieve a similar effect through their Law of Rational Action, which states that a volitional agent with a desire will act as a force which ceteris paribus will result in the desired situation coming about. [Davis & Matthewson, p. 1344]

I believe that this explanation works for some kinds of situations, e.g. intentions to perform a volitional action, but not others. When a volitional agent intends to perform a certain action, the expected outcome of this intention is that (ceteris paribus) he will perform the action. The word intend seems to presuppose that the agent believes it is possible for him to perform the action. However, this presupposition does not arise when we speak of desires, what the agent wants to do. In this case we need to judge, based on knowledge of the world etc., how likely it is that the agent (a) believes the action to be possible, and (b) actually intends to carry it out. When someone says “I really want to go and lay out in the sun for a while,” it is reasonable to expect that, in the absence of some frustrating circumstance, the speaker will carry out the desired action. But this expectation does not arise when someone says, while watching an annoying public figure on television, “I really want to smack him in the face.”

Similarly, a desire for a certain state or non-volitional event to occur does not necessarily create an expectation that the desire will be fulfilled. The desire expressed in example (35) (not to die) is not the intention of a volitional agent, and is not attainable in any epistemically accessible world. Therefore it cannot trigger an epistemic expectation of fulfillment.

Counterfactual uses of dara pose a similar problem for a strictly epistemic approach. Counterfactual statements are by definition not epistemically accessible: they are known or assumed by the speaker to be false. The optative examples in (10a) and (33) express open-ended, hypothetical optatives (e.g., “If only I could find that key!”), i.e., wishes of the speaker that are still hoped for at the time of speaking. But the particle dara can also be used for counterfactual optatives (e.g., “If only I could have found that key!”), i.e., wishes that the speaker knows will not be fulfilled:

In order for the purely epistemic analysis proposed by Davis & Matthewson to account for such examples, we would need to assume that the speaker’s desire for a certain outcome (e.g., for the addressee to be rich) creates an expectation that the desire will be fulfilled (the addressee will become rich), in stereotypical worlds consistent with what is known about the real world. For reasons discussed immediately above, I do not believe that this assumption is valid. The expectations are circumstantial, specifically bouletic, in nature: the relevant outcome can be expected in any world in which the speaker’s desires are achieved. Dara indicates that the real world is not such a world.

In this section I have argued that the expectations dara cancels are not always epistemic in nature: there are at least some uses in which a circumstantial interpretation is necessary. In the following section we will examine some uses of séna7 that should be possible only under an epistemic interpretation. None of these uses seems to be possible for dara. If this assessment is correct, it would lead us to conclude that dara is never epistemic but always purely circumstantial in nature.

5.2 Failed outcome vs. unexpected co-occurrence

In the passage cited at the beginning of sec. 5, Davis & Matthewson (2022: 1346) state that “[s]éna7 also appears… when the issue is not a failed outcome, but simply an unexpected co-occurrence with another eventuality”. In other words, séna7 marks the failure of an expected correlation between two propositions, rather than the failure of an expected chain of events.

The St’át’imcets example shown in (37) illustrates a mirative use of the particle, expressing surprise at the actual outcome (p) in light of the speaker’s evaluation of the friend’s ability (q). Clearly there is no causal relation between these two propositions, and they are not construed as belonging to a single chain of events. The unexpectedness of the correlation is purely epistemic.

The St’át’imcets example shown in (38) illustrates what we might call a concessive use of the particle. Davis & Matthewson write (2022: 1366): “In [38], séna7 is licensed by the common expectation that… the spring salmon run at the same time as the strawberries are ripe. However, there is no causal connection between the salmon running and the berries ripening; it is simply that they ripen at the same time of year.”

When, as in examples (37-38), the two propositions whose correlation is unexpected are not construed as belonging to a single chain of events, it may be possible for the particle séna7 to be placed in either clause without changing the meaning of the sentence. Davis & Matthewson (2022: 1361) describe this fact as follows:

“According to our proposal, the unexpectedness requirement of séna7 targets two propositions (p and q) symmetrically. That is, although séna7 syntactically appears within one clause (the p clause), the not-at-issue relation it expresses does not prioritize one proposition over the other… [T]he clause séna7 is placed in has no effect on the meaning.”

Davis & Matthewson illustrate this principle by pointing out that ex. (39) is exactly the same as ex. (38) except for the position of séna7, and the meaning of the two sentences is identical: “As further evidence that causality is not involved here, we elicited this sentence also with séna7 in the opposite clause, as shown in [39].” Another similar pair is presented in (40a-b).

Because the two propositions whose correlation is unexpected are not construed as belonging to a single chain of events, and because either clause can be marked with séna7, it is possible for the situation described in p to occur later in time than the situation described in q. An example is seen in ex. (41): the teachers’ lack of knowledge about cooking (q) predates the cooking lesson (p).

These strictly epistemic uses of séna7 are not shared by dara. Kimaragang has separate particles for the mirative (bala) and concessive (poma) functions, and dara does not seem to be used for either of these functions.

Kimaragang dara does not appear to be symmetrical in the way that St’át’imcets séna7 is. It does not seem to be possible to shift dara from one clause to the other without causing significant change to the meaning of the sentence, as in (44–45), or rendering the sentence meaningless, as in (46–50).²⁰

The concessive particle poma, in contrast, can (at least in some cases) mark either clause without changing the meaning of the sentence. An example is shown in (51).²² This potential symmetry between the clauses reflects the inherently epistemic nature of the concessive.

I have found no naturally occurring Kimaragang examples with dara that involve the kind of reverse time sequence illustrated in ex. (41), and my attempts to elicit such an example were unsuccessful. In all the examples I have examined, the clause marked by dara describes a situation that is prior in time to the expected outcome (whether explicit or implied).

All of these observations are consistent with the hypothesis that the expectations relevant to interpreting dara are circumstantial in nature: given an initial circumstance p (which may be a desire or intention of some agent), what outcomes are to be expected if the normal course of events is followed as closely as possible? Because a course of events is unidirectional in time, the relation between initiating circumstance and expected outcome is not symmetrical. Dara marks the initiating circumstance, so shifting dara from one clause to the other will result in this second clause being reinterpreted as the initiating circumstance, leading to a different range of expected outcomes. Even when this is grammatically possible, it will create a clear change in the meaning of the sentence. The fact that a course of events is unidirectional in time also explains why is not possible for the proposition marked by dara to describe a situation that is later in time than the relevant outcome.

In light of this evidence, I offer the following definition of the frustrative particle dara, modeled after Davis & Matthewson’s analysis of St’át’imcets séna7:

Proposed analysis for Kimaragang dara: