This paper provides an analysis of the ‘frustrative’ marker séna7 in St’át’imcets (Lillooet Salish), and compares it to similar elements cross-linguistically. Séna7 appears in a range of discourse contexts, including when events have an unexpected outcome, fail to continue, or fail to take place optimally. We argue that séna7 felicitously applies to a proposition p only if there is a salient true proposition q and the speaker did not expect p and q to both be true. Séna7 encodes epistemic modality, refers only to the speaker’s epistemic state (ignoring the common ground), and has no effect on at-issue truth conditions (séna7(p) entails p). We show that séna7 provides a diagnostic for distinguishing between entailments and implicatures in the language, and a clear diagnostic for the distinction between futures and prospective aspects. We compare séna7 with similar elements in Tohono O’odham, Kimaragang and Tagalog. We argue that séna7 and the Kimaragang frustrative can be captured by the same analysis once independent features of their tense/aspect systems are taken into account. Following Kroeger (2017. Frustration, culmination and inertia in Kimaragang grammar. Glossa: A Journal of General Linguistics 2(1). 56. 1–29), but pace Copley and Harley (2014. Eliminating causative entailments with the force-theoretic framework: The case of the Tohono O’odham frustrative cem. In Bridget Copley & Fabienne Martin (eds.), Causation in grammatical structures (Oxford Studies in Theoretical Linguistics 52), 120–151. Oxford: Oxford University Press), we argue that frustratives should not be unified with non-culminating accomplishments, and can be analyzed without appealing to causality or efficacy.
This paper is a contribution to the growing literature on so-called ‘frustrative’ elements, focusing on data from St’át’imcets (Lillooet Salish). Frustratives have been defined by Overall (2017: 479) as in (1), based on his study of 54 Amazonian languages.
Overall notes (2017: 479) that there has been ‘little typological analysis of this category’. Until recently, this was also true of formal analyses; recent formal approaches include Copley (2005), Copley and Harley (2014), Kroeger (2017) and Carol and Salanova (2017).
Introductory examples of the St’át’imcets frustrative séna7 are provided in (2)–(4). Séna7 appears, for example, when events have an unexpected outcome (2), fail to continue (3), or fail to take place in an optimal fashion (4).
This paper provides a unified analysis of séna7 which captures all its effects. We propose that séna7 can be felicitously applied to a proposition p only if the discourse context contains a salient true proposition q and the speaker did not expect p and q to both be true at the same time. The analysis entails the following claims about séna7: (i) it is inherently context-dependent, since it depends on a proposition provided by the context; (ii) it encodes epistemic modality; (iii) it refers only to the speaker’s epistemic state (it does not place any restriction on the common ground); (iv) it has no effect on the at-issue truth conditions: an utterance of séna7(p) asserts p.
We show that séna7 provides a clean diagnostic for distinguishing between entailments and implicatures in the language. This enables us to confirm the difference between predicates which only implicate, as opposed to entail, culmination in the perfective aspect. It furthermore provides a clear diagnostic for a temporal distinction which can otherwise be difficult to tease apart in general, in any language: the distinction between futures (which place the evaluation time before the reference time), and prospective aspects (which place the reference time before the event time).
In the last part of the paper we address the relation between séna7 and other similar elements crosslinguistically including the Tohono O’odham frustrative cem (Copley 2005; Copley and Harley 2014; Hale 1969), the Kimaragang frustrative dara (Kroeger 2017), and the Tagalog ability/involuntary action marker (Alonso-Ovalle and Hsieh 2017a, 2017b, 2018). We argue that in spite of apparent empirical differences between St’át’imcets séna7 and the Kimaragang frustrative, the two elements can be captured by an identical analysis, with the differences deriving from independent features of the tense/aspect systems of the languages. In the debate between Copley and Harley (2014) and Kroeger (2017) over the best analysis of frustratives, we find evidence in support of Kroeger’s view: frustratives should not be (even partially) unified with non-culminating accomplishments, and frustratives can be analyzed using standard modal tools without needing the additional notions of causality, forces, or efficacy.
In the remainder of this section we provide background information on the language, séna7, and our methodology. In Section 2 we show how our proposal works by providing a systematic overview of the interpretations of séna7-clauses with different Aktionsarten. Section 3 presents our formal analysis and explores more detailed predictions: for example, we show that séna7 is speaker-oriented, not at issue, and does not induce a causality effect. Section 4 shows how séna7 interacts with, and distinguishes between, the two grammaticized forms of future time-reference in St’át’imcets, and argues that séna7 acts as a semantic diagnostic for prospective aspect within the class of motion verbs. In Section 5 we compare séna7 to other frustrative elements and their analyses crosslinguistically. Section 6 concludes and points to future research directions.
1.2 Background on St’át’imcets and on séna7
St’át’imcets, also known as Lillooet (ISO 639-3 lil), is a Northern Interior Salish language spoken in the southwest interior of British Columbia, Canada. It is highly endangered, with fewer than 100 first-language speakers at the time of writing (Dunlop et al. 2018). All unattributed examples in the paper come from original fieldwork by the authors. In cases where the data is taken from published narratives, the speaker/storyteller is identified by name.
Séna7 is one of a small closed class of lexical adverbs in St’át’imcets; these adverbs generally occur after the first predicative element of a clause (including any enclitics). Unlike enclitics, séna7 is prosodically independent and may also occur clause-finally or – less frequently – in other post-predicative positions. We do not address its clause-internal distribution in detail here, but we assume that it always takes sentential scope.
Séna7 can appear in both monoclausal and biclausal structures. Biclausal cases were given in (2)–(4) above, and a monoclausal case is shown in (5).
We will argue that semantically, séna7 always relates two propositions, but one of them can be either contextually provided or accommodated.
Séna7 has previously been glossed as ‘though’ (van Eijk 1997), ‘counter-to-expectation’ (Davis 2017), ‘often untranslatable; expresses an unfulfilled condition, a change of mind or some other contradiction or contrast’ (van Eijk 2013), and as ‘against expectations (either the speaker’s, the hearer’s, or somebody else’s); often difficult to translate into English’ (Alexander et al. in prep.). These informal characterizations give something of the flavor of séna7, but do not offer full insight into its semantic or pragmatic contribution. The first attempt at formal analysis of séna7 was by Davis and Matthewson (2016); the current paper builds on and extends the proposals made there.
We will henceforth gloss séna7 as cntr, for ‘contra expectation’.
Several data collection methodologies were employed in this study. Primarily, we conducted targeted elicitation using standard semantic fieldwork methods involving controlled discourse contexts (see, for example, papers in Bochnak and Matthewson 2015; Krifka 2011; Matthewson 2004b; Tonhauser and Matthewson 2016). In addition to the usual methods of eliciting acceptability judgments and translations in context, we utilized two less common techniques as a response to the radical context-dependence of séna7. First, we sometimes provided the consultants with a sentence containing séna7 and asked them to provide a suitable discourse context in which the sentence could be uttered. Second, we conducted a variant of the ‘cloze’ test familiar from language acquisition studies: we provided the speakers with a clause containing séna7, and asked them to provide a felicitous completion (i.e., a follow-up clause). Instances of this elicitation method are marked with ‘…’ between the first and second clauses. Thus, wherever the data includes a ‘…’, the material after the dots was volunteered by the consultant.
Finally, we checked our generalizations against all instances of séna7 in five text collections (Alexander 2016; Edwards et al. 2017; van Eijk and Williams 1981; Matthewson 2005; Mitchell in press), as well as all the example sentences in a forthcoming comprehensive English-Upper St’át’imcets dictionary (Alexander et al. in prep). We also examined the large number of instances of séna7 which have arisen in our elicited data over the years, many of them spontaneously offered in contexts where we were targeting other grammatical phenomena.
2 How séna7 works: case studies of Aktionsarten
Our proposed analysis is given semi-formally in (6).
According to this proposal, séna7 does not affect the truth conditions of its prejacent proposition; instead, it imposes a condition on the relation of the prejacent to another salient proposition (explicit or implicit) within a discourse context.
In order to show how our appoach to séna7 works, in this section we offer a systematic exploration of its effects on lexical aspectual classes (Aktionsarten). We will show that the proposal summarized in (6) successfully unifies all of séna7’s empirical effects. For background on lexical aspectual classes, see Filip (2012, 2020 and references therein.
2.1 Séna7 with states and activities
Atelic predicates in St’át’imcets – states and activities – show the following interpretations with séna7: (i) some expected outcome of the eventuality fails to hold; (ii) the eventuality fails to continue; (iii) the eventuality unexpectedly co-occurs with another one; (iv) the eventuality does not happen ‘well’ or successfully.
Unexpected outcomes of stative eventualities are illustrated in (7)–(11). Notice that the contextually salient proposition q may be provided by the second clause of the utterance itself (as in (7)–(9)), or not (as in (10)–(11)). In cases where q is not obvious from the utterance itself, we indicate it below the data.
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.
Like states, activity predicates also appear with séna7 when some expected outcome of the event fails to happen. Examples are given in (12)–(16).
Sometimes, the expected outcome of a state or activity is simply that it continues, so séna7 flags the fact that the eventuality no longer holds. This is shown in (17)–(18) for states, and in (19)–(21) (and (3) above) for activities.
Séna7 also appears on states and activities when the issue is not a failed outcome, but simply an unexpected co-occurrence with another eventuality. In (21), singing a sad song does not cause one to be unhappy, and in (22), having a bath does not cause one to wash one’s hair. It is simply that these two pairs of eventualities usually co-occur, so the co-occurrence of the opposite is unexpected.
Finally, states and activities allow séna7 in contexts where the eventuality does not happen successfully or very well. This is illustrated in (23)–(27). The English translation often includes ‘try’, but this is not literal; it is an attempt by the speakers to render the ‘not very well’ effect.
Because the prejacent can contrast in various ways with another true proposition, it is easy to find minimal sets with identical séna7-clauses, but different qs. This confirms the context-dependence of séna7. One such minimal pair is (28)–(29): in (28), the speaker contrasts their earlier hunger with the failure of the hunger to continue, while in (29), the hunger contrasts with the failure to eat.
Another pair is (30a) and (30b): an earlier state of wanting is contrasted either with the failure of the wanting to continue, or with the failure of the wanted event to be realized.
An activity pair is given in (31a), and (31b). The interpretations are respectively ‘event in vain’ and ‘not very well’.
So far we have only given positive data – environments where séna7 is felicitous – which do not yet prove that séna7 itself is contributing the relevant interpretation. Negative data are given in (32)–(37). These show that séna7 is unacceptable with states and activities if there is no salient proposition q which the speaker does not expect to be true at the same time as the prejacent.
In this section we have shown that séna7 appears with states and activities when there is a failure of an expected outcome (including a failure of the eventuality to continue), or more generally when something unexpected happens during or after the eventuality, including cases where the activity is not performed successfully. If none of these conditions obtain (or can be reasonably accommodated), séna7 is infelicitous.
2.2 Séna7 with achievements and accomplishments
Achievement and accomplishment predicates behave similarly to each other in many respects when co-occurring with séna7, but there is one important difference relating to whether event culmination is entailed. We will show that this difference provides support for our proposal that séna7 cannot alter the truth conditions of its prejacent.
First, some background on these aspectual classes in St’át’imcets. Achievements are intransitive and unaccusative; they completely lack an external argument. Accomplishments are transitive and have agentive subjects. Crucially, achievements entail culmination in the perfective aspect, but accomplishments with control transitivizers do not: they merely implicate culmination (Bar-el et al. 2005; Matthewson 2004a). The basic facts are illustrated in (38)–(39). The same root, mays ‘get fixed’, entails culmination when it surfaces without transitivizing morphology (38), but only has a cancellable implicature of culmination when it appears with the directive (‘control’) transitivizer (39).
When séna7 is added to achievements and accomplishments, the former allow a subset of the interpretations allowed for the latter. The available interpretations are predictable given the difference between the two aspectual classes with respect to culmination entailments.
We begin with the ways in which achievements and accomplishments behave similarly, and then turn to the differences in the following sub-section.
2.2.1 Similarities between achievements and accomplishments
Séna7 can appear on achievements and accomplishments when an expected outcome of the event or of its result state fails to materialize. (40)–(43) show achievements, and (44)–(46) accomplishments.
The second environment where séna7 appears with achievements and accomplishments is when the expected result state of the event doesn’t hold. This is shown in (47)–(49) for achievements and in (50) for accomplishments; the expected result states are him being there, the fish being all gone, and ‘it’ being in a fixed state.
With both achievements (51)–(53) and accomplishments (54), séna7 also allows an interpretation that the event didn’t happen well or successfully. (Notice that (51)–(52) contain the same predicate mays ‘get fixed’ as in (49), with a different interpretation.)
2.2.2 Differences between achievements and accomplishments
Control accomplishments with séna7 allow an interpretation which achievements do not allow: that the expected culmination of the event did not take place. This is illustrated in (55)–(58).
Crucially, achievements cannot fail to culminate with séna7. The consultant for (59) corrected the predicate to the accomplishment verb máysen, and in (60) the predicate was corrected to zúqwalmen ‘almost die’.
The fact that achievements can never fail to culminate in the perfective aspect with séna7 is an important finding: it shows that while séna7 encodes an unexpected outcome or occurrence, it cannot take away entailments. Séna7 does not alter the truth conditions of the proposition to which it attaches. This means that séna7 cannot be captured by the analysis proposed for the Kimaragang frustrative by Kroeger (2017); we return to this point in Section 5.
The proposal that séna7 cannot cancel the truth conditions of its prejacent correctly predicts that even with accomplishments, some part of the event, specifically its initial process part, still has to happen. Thus, séna7 does not license an interpretation in which the event fails to start at all. This is shown in the minimal triplet in (62)–(64): two different unexpected occurrences are possible in (62)–(63), but it is not possible for no cooking at all to happen, as in (64).
We also correctly predict that a set of transitive predicates which truth-conditionally do entail culmination in the perfective aspect (a subset of predicates formed from achievement roots and containing the causative transitivizer) still must culminate with séna7; this is shown in (65).
Séna7 with these culminating predicates gives rise to the usual unexpected/unsuccessful interpretations, as for example in (66)–(67).
2.3 Summary of empirical landscape
Table 1 summarizes the interpretations we have discovered with séna7 for each Aktionsart. The result state and culmination tests are not applicable to states or activities, since these do not involve changes into result states.
|Unexpected outcome/co-occurring event||Unsuccessful event||Failure of result state||Failure of culmination|
We propose that all the attested semantic effects can be unified under a single generalization: séna7 marks the unexpected co-occurrence of two true propositions. The columns in the table are thus not separate readings, but simply common ways in which the conditions on séna7 can be met. In many cases, the proposition q which contrasts with the prejacent is provided by a generalized implicature deriving from the lexical semantics of the prejacent’s predicate: the implicatures that accomplishments will culminate, that achievements and accomplishments have persistent result states, and that eventualities will happen successfully.
The explanation for the lack of a ‘failure of culmination’ interpretation with achievements is, as already discussed, that séna7 does not have the power to defeat entailments of the proposition to which it applies.
In the next section we formalize our analysis further, and go through some finer predictions it makes.
3 Analysis and detailed predictions
Our proposal – that séna7 (p) conveys that speaker did not expect p to be true as well as another contextually salient proposition q – is stated more formally in (68).
In this formula, EPISsp(c) is an epistemic modal base for a speaker in a context c. ∩EPIS sp(c)(w) is the set of worlds which are epistemically accessible to the speaker of c in w (worlds which are compatible with the speaker’s beliefs). STEREO is an ordering source; BESTSTEREO(w) orders a set of worlds according to stereotypicality relative to w, and selects the most stereotypical ones.
Putting all this together, the speaker of séna7(p) asserts p, and conveys at a not-at-issue level that there is a true proposition q, and there is no world w’ among the most stereotypical worlds epistemically accessible to the speaker such that p and q are both true in w’.
This is a modal analysis, using standard Kratzerian conversational backgrounds (an epistemic modal base, a stereotypical ordering source). There is thus a similarity with familiar epistemic modals, as in (69).
A standard analysis of (69) is that it means ‘In all worlds which are compatible with the speaker’s beliefs/evidence (epistemic modal base), and in which things proceed in a maximally normal manner (stereotypical ordering source), Michl is the murderer.’ In our analysis, séna7 quantifies over the same set of worlds as an epistemic modal, but conveys that there are no maximally stereotypical epistemically accessible worlds in which p and q are both true. Another important difference between ordinary epistemic modals like must and séna7 is that séna7 conveys modality in the not-at-issue realm: it doesn’t directly assert its modal contribution.
Our analysis makes some further detailed predictions. These are laid out in (70) and tested in the following sub-sections.
3.1 Séna7 takes only one syntactic argument
Séna7 applies to a prejacent proposition p; however, the contrasting proposition q is not an argument of séna7, but is existentially quantified over. This predicts that q is not syntactically required to be present. A strong piece of evidence for this is that mono-clausal sentences containing séna7 are possible (and frequently volunteered). We have seen several examples of this above, and (71)–(73) are more cases where the contrasting proposition q is not overtly given.
Even when there are two overt clauses, the contrasting proposition q is not necessarily represented by one of them. In (74), for example, it is not unexpected that a potential place to stay would be both good and expensive, so the contrast is not between the two overt clauses. Rather, the fact that the place seems good (p) contrasts with the implicitly conveyed proposition q ‘We won’t stay here’.
In (75), séna7 encodes the unexpectedness of my not having another drink, even though I have money. Crucially, q is not the second overt clause, ‘I’ve already had enough to drink’. Instead, q is an implicature of the second overt clause.
Similarly in (76), q is provided by conversational implicature. Here, séna7 is contrasting going out with not having fun, which is implicated by not having any money.
3.2 The unexpectedness requirement is symmetrical between the two propositions
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. This is supported by the fact that in a linear sequence of two clauses, séna7 is not restricted to appearing in the first one. Although it appears on the first clause in most of our data, there are second-clause examples, as shown in (77)–(80). Example (78) is a minimal pair with (21) and is interpreted identically, showing that the clause séna7 is placed in has no effect on the meaning; Examples (94)–(97) below provide two more minimal pairs with séna7 in opposite clauses, with no effect on meaning.
3.3 Séna7’s contribution cannot scope under other operators
Our claim that séna7 contributes its semantic content in the not-at-issue dimension predicts that its unexpectedness contribution cannot take scope under operators such as negation. This is correct. In (81), the negation targets the at-issue truth conditions of the first clause (they did not allow us to run and play); it crucially does not negate the unexpectedness (i.e., the sentence does not mean ‘It is not unexpected that we ran and played in spite of them not allowing us to’).
Similarly in (82), séna7’s contribution is not targeted by the negation. The sentence asserts that the speaker didn’t want to go to the school, and séna7 contrasts the lack of wanting to go with the fact that he had to go anyway.
3.4 The unexpectedness requirement holds only for the speaker
The requirement that p and q are not expected to both be true is placed only on the speaker. This predicts that the addressee need not share the speaker’s assumptions about what counts as unexpected. We test this in (83)–(84). Here, the contexts do not provide a contrasting q for the addressee, yet the sentences are fine.
3.5 Séna7’s prejacent clause need not be inherently unexpected
According to our analysis, the speaker of a séna7-clause does not expect the prejacent to be true at the same time as some other salient proposition q. The speaker crucially does not have to believe that the prejacent proposition itself is inherently unexpected. We see this in (85)–(86); in these cases, the prejacent of séna7 when considered in isolation is stereotypically often true.
3.6 The requirement is about expectations, not intentions
The unexpected co-occurrence of p and q includes, but is not limited to, situations where some agent had an intention which failed. In (12) above, séna7 accompanies a report of a failed plan (to kill deer), but in (87), there was no plan that ‘they’ (riders in a ‘suicide race’) would get hurt. The speaker simply did not expect them to escape unscathed from this dangerous situation.
Further cases where there is no failed intentional plan are given in (88)–(89).
These data show that the contribution of séna7 cannot be unified in terms of involving frustrated intention. However, the data can all be unified in terms of unexpectedness. Our proposal that St’át’imcets séna7 does not semantically convey frustrated intention is in accord with Overall’s (2017: 485) observation that in the Amazonian languages he discusses, “The sense of unfulfilled intention or desire … seems in most cases to be epiphenomenal.”
3.7 The requirement is about expectations, not causality
We gave examples above where the unexpectedness of p and q both being true did not derive from a failed causal relation ((21)–(22)); further examples are given here. In (90), the issue is not that their teaching us to cook (p) is expected to cause them to know how to cook (the negation of q). Rather, it is simply unexpected for p and q to both be true. Similarly in (91), the chicken being cooked (p) would not cause the potatoes to be cooked (the negation of q), in (92), the fence getting fixed would not cause the gate to be fixed, and in (93) getting to the meeting would not cause the car not to break down, yet séna7 is fine in all three examples.
In (94), séna7 is licensed by the common expectation that of 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. As further evidence that causality is not involved here, we elicited this sentence also with séna7 in the opposite clause, as shown in (95).
Examples (96) and (97) are identical except that séna7 appears in the first versus second clause. A causal analysis would have to conclude that the examples have quite different meanings, an idea for which there is no evidence. Moreover, each of the potential causal claims are somewhat implausible: either that liking to eat cake causes one to not be able to each much cake (96), or that not being able to eat much cake causes one to like eating cake (97).
A final piece of evidence that the two propositions séna7 relates need not stand in a causation relation is given in (98). In this context, the event of reaching Mount Currie is part of a larger event of traveling to Lillooet from Vancouver. Yet a causing and a caused event must be fully distinct and cannot stand in a part-whole relation (Menzies and Beebee 2020).
This will be important below where we argue that Copley and Harley’s (2014) cause-based analysis of frustratives cannot be applied to St’át’imcets séna7.
Having shown that the fine-grained predictions of our analysis are upheld, we turn now to further implications. We show that besides the distinction between achievements and accomplishments discussed in Section 2.2, there are other subtle semantic distinctions in the language which séna7 allows us to delineate clearly, further highlighting its use as a diagnostic tool for illuminating contrasts beween entailments and implicatures. In Section 4.1 we undertake an examination of future time reference, showing that séna7 allows us to distinguish between a future modal and a prospective auxiliary, and in Section 4.2 we extend the diagnostic to show that some motion verbs must contain prospective semantics.
4 Séna7 and future time reference
4.1 Future versus prospective aspect
When séna7 co-occurs with markers of future time reference, the results are as predicted by the analysis. Furthermore, séna7 distinguishes semantically between the two grammaticized forms of future time reference in St’át’imcets.
The two grammatical markers of future time reference in St’át’imcets are the modal clitic =kelh and the aspectual auxiliary cuz’. Both appear in (99). As a rough approximation, =kelh corresponds to English will or future-oriented might, while cuz’ corresponds to is going to. See van Eijk (1997), Matthewson (2006b), Rullmann et al. (2008) and Davis (2017) for discussion.
We assume a neo-Reichenbachian approach to tense and viewpoint aspect that involves reference to (at least) three time intervals: a reference time (the time about which the sentence makes a claim), an event time (the time at which the event takes place), and an evaluation time (with respect to which tenses are evaluated). The evaluation time is by default the utterance time in matrix contexts. Thus, for example, a past tense places the reference time before the utterance time in a matrix clause. See Klein (1994) for this type of approach.
Glougie (2008) argues that the St’át’imcets clitic =kelh places the reference time after the evaluation time, while cuz’ is a prospective aspect which places the event time after the reference time. In simple cases, the meanings are difficult to tease apart, but Glougie shows that the elements diverge in cases where an event is already planned at the utterance time. Here, only cuz’ is acceptable, not =kelh, as shown in (100). Glougie notes that (100b) would be appropriate if the speaker was only considering going away for the weekend and had not yet purchased a bus ticket.
With both =kelh and cuz’, the evaluation or reference time need not be the utterance time, but can be a past time. This is parallel to the situation in English, where will has a past-shifted form would, and is going to has a past-shifted form was going to. Past-shifted examples of =kelh and cuz’ are given in (101) and (102) respectively.
When séna7 co-occurs with these markers of future time reference, it gives rise to two quite distinct readings. With =kelh, séna7 imparts that the event described by p will happen, in spite of some other proposition q, while with cuz’, séna7 conveys that the prejacent event was going to happen, but the event described by q happened instead.
Data with =kelh are given in (103)–(104). Here, the speaker is making a prediction about a future event, and in addition there is some contextually recoverable true proposition q, and the speaker finds it unexpected that q is true as well as p.
These data are as predicted by Glougie’s analysis of =kelh and ours of séna7. The future modal =kelh places the reference time after the evaluation time, which in these examples is the utterance time. Séna7’s prejacent, which contains =kelh, asserts that an event will take place at that future reference time in all relevant possible worlds. Séna7 contributes that the speaker doesn’t expect that the future proposition p and some contextually available proposition q are both true. In other words, the speaker asserts that an event will happen in the future, and conveys that something unexpected will also happen. This gives an ‘in spite of’ or ‘anyway’ reading.
Data with cuz’ are given in (105)–(108). Here we get a quite different interpretation.
Again, the results fall out from the analysis. Cuz’ places the event time after the reference time, which in these examples is a past time. Séna7’s prejacent thus makes a claim about a pre-state of an event (the state of something being about to happen). Séna7 conveys that there is some other proposition q that is unexpected given the prospective p (the claim that there was a pre-state of an eventuality). The most natural case is that q entails that the expected eventuality did not take place. The cuz’ data are similar to cases where séna7’s prejacent is a lexical stative, as discussed in Section 2.1. Just as séna7 when applied to a proposition about wanting something frequently conveys that the expected outcome of that desire (getting the thing) remains unfulfilled, séna7 on a cuz’-proposition conveys that the expected outcome of the pre-state of an eventuality happening (the eventuality actually happening) remains unfulfilled.
The reader may have noticed that the =kelh + séna7 data involve present evaluation times (‘will’, not ‘would’-readings), while the cuz’ + séna7 data involve past evaluation times (‘was going to’, not ‘is going to’ readings). Our analysis predicts that =kelh cases can also, in a rich enough context, allow past evaluation times, with readings such as ‘the event described in p was predicted to happen, in spite of q.’ This is correct, as shown in (109).
We have shown in this section that séna7 gives rise to different interpretations with the two markers of futurity, =kelh versus cuz’. With =kelh, the truth conditions are that the prejacent event will happen, and séna7 conveys that something else will happen which is not expected to simultaneously be true (‘p will/would happen, in spite of q’). With cuz’, the truth conditions are that the prejacent event was planned to happen, and séna7 conveys that counter to expectations, it didn’t happen after all (‘p was going to happen, but q happened instead’). These are exactly the readings predicted by Glougie’s (2008) analysis of =kelh and cuz’ as a future-oriented modal and a prospective aspect, respectively. This provides a very clean diagnostic for the distinction between futures (which place the evaluation time before the reference time), and prospective aspects (which place the reference time before the event time).
The ability of séna7 to diagnose the semantics of prospective aspect leads to a further result: it enables us to identify a subset of motion verbs in St’át’imcets which must be analyzed as containing prospective semantics. We discuss this in the next sub-section.
4.2 Séna7 as a diagnostic for prospective aspect: extension to motion verbs
|motion towards speaker||t’iq ‘get here’||ts7as ‘come (here)’|
|motion away from speaker||tsicw ‘get there’||nas ‘go (there)’|
|motion ‘along’||t’ak ‘go along’|
Examples of each motion verb are given in (110)–(115), from Davis (2017, Ch. 16). (There are two examples for t’ak ‘go along’, as it does not specify the direction towards or away from the speaker.) As discussed by Davis, the different tenses used to translate t’iq ‘get.here’ and tsicw ‘get there’ (past) versus ts7as ‘come’ and nas ‘go’ (present) do not reflect a real tense effect. They are the default interpretations when combining telic versus atelic predicates with the null non-future tense (Matthewson 2006b).
When we add séna7 to sentences containing telic motion verbs, nothing unexpected happens. Like the other achievement predicates discussed in Section 2.2, t’iq ‘arrive’ and tsicw ‘get there’ retain their culmination. Séna7 indicates some unexpected happening, such as the failure of the result state to hold or the failure to meet the person one was intending to visit.
The non-cancelability of the culmination with t’iq/tsicw and séna7 is illustrated in (120)–(121).
Ts7as ‘come’ and nas ‘go’ show a different pattern. As they are atelic, they allow an interpretation where the subject fails to reach her destination, as in (122)–(124) (which contrast minimally with (120)–(121)). However, they also allow an interpretation which is not available for ordinary activity predicates: that no motion took place. This is illustrated in (125)–(126), and it suggests that ts7as and nas contain prospective semantics. Notice that (118) and (126) form a minimal pair with different interpretations.
As expected, consultants freely accept minimal pairs involving different contextually provided contrasting propositions, either involving noncompletion of the motion event (127a), failure of the result state to hold (128a), or a complete failure to move ((127b), (128b)).
The behavior of ts7as ‘come’ and nas ‘go’ matches that of the prospective aspect cuz’ as discussed above: unlike other predicates, they allow an interpretation with séna7 where the prejacent event does not take place. We conclude that they have a reading as prospective aspects.
The fifth motion verb, t’ak ‘go along’, is partially similar to ts7as ‘come’ and nas ‘go’, and partially similar to t’iq ‘arrive’ and tsicw ‘get there’: it is atelic, but non-prospective. This shows that the two features – (a)telicity and (non-)prospectivity – are separable. Example (129) shows that t’ak is atelic (the motion does not have to reach a final destination), and (130) shows that t’ak is non-prospective (the motion cannot fail to start at all).
A minimal triplet contrasting the three motion verbs which allow motion away from the speaker is given in (131). Telic, non-prospective tsicw ‘get there’ entails that the motion was completed; atelic, prospective nas ‘go’ allows no motion at all, and atelic, non-prospective t’ak ‘go along’ entails that some motion took place but does not require that the destination is reached.
In summary, not only are the predictions of our analysis of séna7 confirmed, séna7 interacts with future time-reference in a predictable way and provides a clear diagnostic for the presence of prospective semantics in a subset of the motion verbs of the language.
5 Comparison with other frustratives
In this section we compare séna7 to similar elements crosslinguistically, and explain why previous analyses are not applicable to séna7. We also propose a potential re-analysis of another frustrative marker, Kimaragang dara, to make it parallel to our analysis of St’át’imcets séna7. 
5.1 Tohono O’odham cem
A well-known frustrative marker is Tohono O’odham cem (Copley 2005; Copley and Harley 2014; Hale 1969; see also Devens 1979 on the cognate in closely related Pima/Akimel O’odham). Copley and Harley (2014: 123) remark that ‘Descriptively speaking, sentences with frustratives can express the fact that the subject intended to do something that is not realized; that [the] subject does something in vain; that a situation is unsatisfactory or does not develop as expected, or that a state does not continue.’ Examples are given in (132)–(133).
Copley (2005) argues that cem(p) sentences (a) assert that all inertia worlds for the topic situation s are worlds in which p(s), and (b) presuppose that the actual world is not an inertia world for s. Copley and Harley (2014) replace the inertia worlds analysis with an approach involving forces (see also Copley and Harley 2015). Forces are inputs of energy which act on situations. An efficacious situation is one whose normal expected result (given the forces in the situation) obtains (see Copley and Harley’s paper for the full formal definition). They argue that cem(p) sentences presuppose that the topic situation s is not efficacious (i.e., its normal result does not obtain). Their denotation for cem is given in (134).
According to this denotation, cem(p) is truth-conditionally identical to p (just as we have proposed for séna7). This feature of Copley and Harley’s analysis might initially seem to clash with the characterization given above that cem is licensed by contexts in which the subject intended to do something that is not realized, or with example (132) in which Juan did not manage to open the door. However, the role of aspectual morphology is crucial: when the clause is perfective, cem’s prejacent is actualized, so sentences like (132) which allow non-realization are necessarily in the imperfective. A more literal translation of (132) would presumably be ‘Juan was opening the door,’ which is truth-conditionally compatible with him not opening it.
Apart from the technical tools used (forces and efficaciousness as opposed to quantification over possible worlds), the other difference between Copley and Harley’s analysis of cem and ours of séna7 is that the former relies on the normal progression of situations and their expected results or outcomes. We have argued on the basis of examples like (90)–(98) above (e.g., ‘The potatoes got cooked but the chicken didn’t’) that séna7 conveys unexpectedness but does not always rely on causes and effects. As Copley and Harley do not discuss data like (90)–(98), it is difficult to be sure whether (failed) causation is a crucial requirement of cem.
A related issue is the effect of cem on imperfective accomplishments. Copley and Harley write that ‘In the case of the imperfective, a sentence with cem as in [(132)] conveys that Juan does something to open the door, but the door does not open’ (2014: 148). This is in line with the idea that cem signals the failure of the normal, expected outcome of pulling on a door: that it opens. However, séna7 in a parallel case can not only have the non-culmination interpretation, but can also convey a non-causally-related unexpected event. This is shown in (135).
Copley and Harley do not provide data like (135) involving imperfective accomplishments which eventually culminate, but involve other unexpected eventualities.
Beyond the analysis of cem, Copley and Harley have the larger goal of partially unifying frustratives with non-culminating accomplishments; the latter were discussed in Section 2.2 and are illustrated again in (136) for St’át’imcets.
Copley and Harley claim that frustratives and non-culminating accomplishments both involve non-efficacy, but differ in whether this is enforced or only allowed. Frustratives presuppose that the topic situation is non-efficacious (as in (134)). Culminating accomplishments presuppose that the topic situation is efficacious; it is therefore entailed that the end result of the net force (the culmination) actually occurs. Non-culminating accomplishments fail to presuppose this, and therefore allow the absence of culmination.
Using data from the Austronesian language Kimaragang, Kroeger (2017) argues against Copley and Harley’s efficacy-based partial unification of frustratives and non-culminating accomplishments. One empirical argument advanced by Kroeger is that in Kimaragang, the frustrative marker (which according to Copley and Harley would presuppose non-efficacy) can co-occur with non-volitive marking, which enforces culmination on accomplishments and which therefore according to Copley and Harley would presuppose efficacy.
The same is true in St’át’imcets. There is a small class of transitive predicates in St’át’imcets which test as achievements, due to the fact that their instantaneous or near-instantaneous running time prevents them from being initiated without also culminating (see Footnote 12). As shown in (137)–(138), members of this class can co-occur with séna7 (see also (66)–(67) in Section 2.2.2). This would result in a fatal clash of presuppositions in Copley and Harley’s account.
A second argument against unifying frustratives with non-culminating accomplishments, again originally due to Kroeger (2017), is that the available interpretations for the two phenomena are quite different. In St’át’imcets, accomplishments without séna7 only allow an event to be ‘non-normal’ in the sense that its culmination need not take place. Séna7, in contrast, allows a broader range of interpretations, as we showed in Section 2: not only failure to culminate, but also other unexpected outcomes, failure of the result state to hold, or that the event did not happen well.
It is not easy to demonstrate via negative evidence that accomplishments without séna7 only license unexpected interpretations which involve non-culmination, because any simple predication not containing a frustrative can be followed by a clause saying that the event had an unexpected outcome or was not successful. However, indirect evidence comes from consultants’ responses to monoclausal, out-of-the-blue sentences containing only accomplishment verbs, as opposed to their responses to séna7-sentences. We illustrate this in (139)–(140).
In (139), an accomplishment predicate, either with or without séna7, can be followed by a query about whether the culmination was reached. No special context is needed in order for the possibility of non-culmination to seem natural. The presence of séna7 merely makes this more probable, as indicated by the consultant’s comment on the séna7 version.
In (140), on the other hand, the responder asks more generally “What seems wrong?”, rather than specifically asking about non-culmination. As in (139), the consultant comments on the role of séna7 in conveying that something went wrong, but unlike in (139), the consultant remarks that the non-séna7 version needs a visual cue that something went wrong. This difference between non-culmination (as in (139)) and general unexpectedness or failure (as in (140)) supports the idea that the two phenomena of non-culminating accomplishments and frustrativity should not be unified in terms of a single notion of ‘nonefficacy’.
A final argument against the unification of séna7 and non-culminating accomplishments, not discussed by Kroeger (2017), relies on morphological evidence. Recall that according to Copley and Harley, non-culminating accomplishments presuppose nothing about efficacy, and it is culminating accomplishments which bear a presupposition (that the topic situation is efficacious). Thus, ‘[t]he absence of a culmination is the basic case’ (2014: 135). In support of this, Copley and Harley claim that ‘non-culminating accomplishments do not require any special construction or morphology to indicate the failure of a normal or expected outcome to occur,’ and that this ‘allows us to treat cases of defeasible causation straightforwardly, instead of first generating and subsequently undoing a causative entailment’ (2014: 134).
This analysis makes the wrong predictions for Salish languages. In St’át’imcets and other languages in the family, non-culminating accomplishments do need special morphology (directive transitivizers). The bare root is always a telic achievement (as shown in Section 2.2), so the morphological evidence suggests that we do need to first generate a culmination and then undo it. This in turn suggests that we need to assign semantic content both to séna7 and to the directive transitivizer, and since these two elements are morphologically distinct, there is no morphological argument that they should be partially unified semantically.
5.2 Kimaragang dara
In this section we show that the Kimaragang frustrative particle dara, discussed by Kroeger (2017), is very similar to St’át’imcets séna7. We further argue that the two may be even more similar than Kroeger’s own analysis of dara suggests, and propose a potential re-analysis along the lines we have proposed for séna7.
Dara is found in a range of contexts, including cases of unfulfilled desires or intentions, failed attempts, former states that no longer obtain, states that do not lead to expected results, things done in vain, and counterfactual conditionals (Kroeger 2017: 2). Two of these uses are shown in (141)–(142).
Kroeger unifies all the uses of dara as expressing ‘frustrated expectation or intention’ (2017: 1). He proposes that dara asserts that some salient proposition is true in all optimal (i.e., highest-ranked) accessible worlds, and presupposes that the actual world is nonoptimal in the relevant respects (thus, that the salient proposition is false). The unrealized proposition can be dara’s prejacent, or if this is not possible, then it ‘may be inferred from context, and typically describes a successor event or result state of the situation described in the dara clause’ (Kroeger 2017: 15). For example, (141) asserts that in all the most optimal worlds, the speaker isn’t bitten, and presupposes that the actual world is non-optimal, so the speaker did get bitten.
In Kroeger’s analysis, dara-clauses make either epistemic or bouletic modal claims. (141) and (142) are epistemic: they have ‘frustrated expectation’ readings. They assert that in all stereotypical worlds compatible with the speaker’s knowledge, the dog leaves the speaker alone/the speaker still has money, and at the same time they presuppose that these optimal propositions are false: the dog did not leave the speaker alone, and the speaker no longer has money.
A bouletic (‘frustrated intention’) case is shown in (143). Here, the optimal but false salient proposition is the prejacent itself. According to Kroeger’s analysis, the sentence asserts that in all worlds that are compatible with the relevant circumstances and in which Mother’s desires or intentions are fulfilled, she binds the fish trap; it also presupposes that she does not bind it.
Kroeger’s analysis and ours are similar in that they both invoke modality and use no extra theoretical tools beyond the standard ones of restricted quantification over possible worlds. However, there are two important differences between the analyses.
First, Kroeger’s analysis employs modality in the at-issue truth conditions: a dara-clause makes a modal assertion. This allows dara’s prejacent to be false in the actual world. In contrast, we have argued that séna7’s contribution is not-at-issue and has no effect on truth conditions; thus, séna7’s prejacent is entailed to be true in the actual world. Second, Kroeger allows intention readings (with a circumstantial modal base and a bouletic ordering source), while our analysis is purely epistemic: the only factor for séna7 is speaker expectation.
Nothing theoretically would rule out frustratives varying in these ways cross-linguistically. It is already known that languages encode a range of different fine-grained modal distinctions. Moreover, what is conveyed in the at-issue realm by one element can be conveyed in the not-at-issue realm by another element in the same or another language. For example, DeVeaugh-Geiss (2014) and Zimmermann (2018) argue that the German particles wohl and schon contribute not-at-issue modal semantics corresponding respectively to the at-issue modal elements werden and eher in the same language.
However, we suspect that dara may in fact be fully compatible with our analysis of séna7, which would be an interesting result, as St’át’imcets and Kimaragang are unrelated languages. We propose that a unified analysis can be given for séna7 and dara while still capturing the apparent empirical differences between the two frustratives.
With respect to whether two different modal flavors (epistemic vs. bouletic) are required, we observe that in both languages, the facts are the same, namely that both expectation-related and intention-related interpretations are available. However, we propose that a unified epistemic analysis can capture the facts. By adopting one extra assumption – that the expected outcome of an intention is that the intention is fulfilled – we can reduce the failure of intention cases to failure of expectation cases. In fact, we already made this assumption in Section 2.1, following Copley and Harley’s (2014) 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.
An apparently more substantial obstacle to the unification of séna7 and dara is that séna7(p) entails p (as we argued extensively above), while dara(p) does not (as for example in (143)). This is what leads Kroeger to adopt an at-issue modal semantics for dara. However, there are some significant exceptions to Kroeger’s claim, where p is in fact entailed by dara(p). These include cases where the prejacent clause is marked for past tense, as well as predicates which describe states in the past or present. In these cases, dara entails its prejacent, just like séna7 does. An example is given in (144): we see that the past tense-marked version of the sentence does not allow a non-realized interpretation with dara.
Kroeger’s account of this ‘realis’ effect with dara on past-inflected eventives is that ‘we see a kind of shift in the function of the tense morphology: it marks a contrast between past versus non-past time reference in main clauses and similar contexts, but realis versus irrealis in dara clauses (2017: 18).’ However, simply adding realis marking to a clause containing an at-issue modal does not actually achieve the effect of requiring the prejacent proposition to be true in the actual world (this is true whether the realis semantically scopes over or under the modal). To have the intended effect, the realis contribution of the past tense marker would have to actively cancel dara’s at-issue modal contribution, something which would be compositionally problematic. In addition, postulating a semantic ambiguity in the contribution of the past/realis marker is less desirable conceptually than having a unified analysis. Finally, the proposed realis reading of the past marker does not account for the realis effect with non-future statives. For these, Kroeger writes that he ‘do[es] not have a good explanation’ (2017: 20).
Our proposed alternative analysis of dara, which leads to further predictions about Kimaragang which await testing, is that it has an identical semantics to séna7. Under this analysis, the empirical difference between the languages – the fact that dara appears to allow false prejacents and séna7 does not – derives not from a difference between the two frustrative markers, but from independent differences in the tense/aspect systems of the languages.
The temporal systems of the two languages are in fact different: St’át’imcets has a future/non-future tense system (with non-future being phonologically null; Matthewson 2006b), while Kimaragang has a past/non-past tense system for eventives (with non-past being phonologically null), while statives are not normally marked for tense (Kroeger 2017). Thus, temporally unmarked predicates in St’át’imcets can only be interpreted as having past or present time reference, while temporally unmarked eventive predicates in Kimaragang allow present or future time reference. 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’. The apparent ‘unrealized’ status of dara’s prejacent would then derive not from dara itself, as in Kroeger’s analysis, but from the inherent futurity/unreality of the prejacent, as in our analysis of St’át’imcets séna7-clauses with prospective aspect.
The behavior of stative predicates in Kimaragang could potentially also fall out from this reanalysis, since stative predicates interpreted in the present or past require realis interpretations with dara (Kroeger 2017: 20). This follows if (a) dara(p) entails p, as in our reanalysis, and (b) statives, unlike eventives, do not allow prospective or future interpretations without overt temporal marking. Kroeger does not give examples of stative predicates with future interpretations, so further research is required to determine whether this prediction is upheld.
This proposed reanalysis of the Kimaragang facts has an additional advantage: it does away with the presupposition that the optimal proposition is unrealized. This is a welcome result because in many cases (including (141) and (142)), the postulated presupposition is overtly introduced by, or at least implicated by, a follow-up clause. Presuppositions by definition are assumed to already be in the common ground and therefore are not usually overtly stated (cf. #The King of France is bald, and there is a unique King of France).
Before leaving Kimaragang, we need to reiterate that we are fully in agreement with what we take to be the main point of Kroeger’s paper: that the meaning of frustratives like dara is not unifiable with the meaning of non-culminating accomplishments, pace claims by Copley and Harley (2014); see discussion in the previous sub-section.
5.3 Tagalog AIA (ability/involuntary action)
The final related phenomenon we discuss is ability/involuntary action morphology on Tagalog verbs (Alonso-Ovalle and Hsieh 2017a, 2017b, 2018). The effect of AIA morphology is illustrated in (145)–(146). In (145), the verb has neutral morphology; the sentence simply asserts that Lisa opened the door. In (146), extra meaning is conveyed by the AIA marking.
Alonso-Ovalle and Hsieh (2018) argue that a sentence containing AIA morphology asserts the core proposition p (e.g., that Lisa opened the door), and presupposes that, given the facts that p is assumed to causally depend on, ¬p was expected. This analysis shares with our analysis of séna7 the fact that the core proposition is asserted, and that modality is introduced in the not-at-issue dimension. The modality for AIA morphology relies on a set of worlds defined by a set of causally relevant facts (cf. Kaufmann 2013), plus a stereotypical ordering source.
Although the not-at-issue status of the modality is parallel between Tagalog AIA morphology and St’át’imcets séna7, there are also differences which seem to speak against a unified analysis. One major one is that AIA morphology is said to enforce the inherent unexpectedness of the prejacent p itself. Thus, AIA morphology is inappropriate in (147).
We saw in (85) (‘The sun came up, but it was cloudy, so we couldn’t see it’) that St’át’imcets séna7 is compatible with inherently expected prejacent events. Séna7 is also possible even in an out-of-the-blue, monoclausal statement about the sun parallel to (147), as long as it is possible to accommodate some salient unexpected second proposition. This is shown in (148).
We have argued that this result follows because séna7 does not force its prejacent to be inherently expected, but merely marks the unexpected co-occurrence of the prejacent with some other proposition. While the contrast between (147) and (148) is certainly suggestive, it would be interesting to see whether AIA morphology could become acceptable in ‘sun’ cases by making the contrasting proposition salient in the context, for example in a biclausal case like (85).
6 Summary, open issues and implications
The St’át’imcets frustrative séna7 poses a prima facie analytical challenge, due to the apparently wide range of readings it gives rise to: failure of expected outcome, noncontinuation of an eventuality, unexpected co-occurrence of two eventualities, an eventuality not happening very well, the failure of a result state to hold, the failure of culmination, and even that an event didn’t happen at all (only with prospective aspect).
We have argued that the meaning of séna7 is best captured by the analysis in (149). According to this, séna7 takes one semantic and syntactic argument: its prejacent clause. It has no effect on the at-issue truth-conditions of this clause, so an utterance of séna7(p) asserts p. In the not-at-issue dimension, séna7 conveys that the discourse context contains a separate salient true proposition q, and the speaker does not expect p and q to both be true. The contrasting proposition q can be provided by a subsequent clause, an implicature of asserted material, real world knowledge, or other means; as such, the interpretation of séna7-clauses is highly context-dependent.
We have further shown that séna7 can be used as a diagnostic to tease apart entailments from implicatures, using telicity as a case study: séna7 helps to distinguish achievements, which have a culmination entailment, from control accomplishments, which only have culmination implicatures. Séna7 also distinguishes between two ways of expressing future time reference: with the modal clitic =kelh, séna7 asserts that an event will happen in the future, and conveys that something unexpected will also happen (‘p will happen, in spite of q’), whereas with the prospective auxiliary cuz’, it is the pre-state of an eventuality which contrasts with a second proposition q; the most common interpretation is ‘p was going to happen, but q happened instead’. This provides a diagnostic for teasing apart futures (which place the reference time after the evaluation time) and prospective aspects (which place the event time after the reference time). Finally, we showed that séna7 distinguishes motion verbs along both parameters: telic versus atelic (requiring vs. not requiring the reaching of an endpoint) and prospective versus non-prospective (allowing vs. not allowing no motion at all to take place).
Crosslinguistically, we showed that séna7 yields similar interpretations to other frustratives, including Tohono O’odham cem, Kimaragang dara and Tagalog AIA morphology. We argued that the differences between séna7 and dara may reduce to independent differences in the temporal systems of the languages; this paves the way for a unified analysis, but requires empirical confirmation in future research. We have in addition argued (following Kroeger 2017, but pace Copley and Harley 2014) that non-culminating accomplishments are fundamentally different from frustratives.
Methodologically, this paper contributes to semantic and pragmatic fieldwork along two dimensions. First, we have shown that rich contextual specification, partially co-created with our language consultants, can yield precise formulations of subtle not-at-issue phenomena such as the meaning of frustratives. Second, we have shown that once their precise contribution is understood, frustratives such as séna7 can themselves be employed as diagnostic tools to tease apart implicatures and entailments, as we demonstrated in our analysis of non-culminating accomplishments, prospective aspect, and motion verbs in St’át’imcets.
6.2 Open issues and implications
An interesting topic for future research is the effect of séna7 inside questions, as in (150).
One recent analysis of not-at-issue content inside questions is that of Davis and McCready (2016). They argue that when an expressive element appears in a question, it can operate on whatever is the true answer to the question. Applying this idea to séna7, we would predict that in (150), the speaker is (a) asking whether the addressee is hungry, and (b) conveying that whatever the true answer is to the question, it is unexpected. This makes sense, since either answer would be unexpected in such a context. If the addressee isn’t hungry, it’s unexpected that they are still eating (as in the consultant’s volunteered context for the utterance). If the addressee is hungry, that is unexpected given that they just ate a lot. However, further research is required here.
Similarly, future research should extend the analysis of séna7 to capture its contribution inside imperatives. One example is given in (151).
Another interesting area for future systematic investigation is the interaction of séna7 with the felicity conditions of prior speech acts. The preliminary data we have are compatible with our analysis, under the assumption that the contextually salient proposition q can be provided by specific felicity conditions in the discourse context. Examples are given in (152)–(154) for séna7-utterances following a command, a question, and an assertion. In each case, q is a felicity condition of the preceding speech act.
There is precedent in the literature for the idea that discourse-sensitive elements like séna7 can respond to felicity conditions; for example, Egg (2010) and Egg and Zimmermann (2012) propose that German discourse particles can respond not only to propositional content, but to the felicity conditions of speech acts.
Eventually, frustratives like séna7 should be compared with a broader cross-linguistic set of markers encoding a sense of contrast, including for example conjunctions such as English but or (even) though, German discourse particles like doch, zwar or schon, and Russian correction and adversative markers (Jasinskaja 2012; Jasinskaja and Zeevat 2008). Two clear differences between séna7 and but or (even) though are the fact that séna7 takes only one syntactic argument, with the other clause provided by the context, and that séna7 is semantically symmetrical (it can appear on either of the two propositions it relates, with no difference in meaning, as we showed in Section 3.2; this is not the case for but or even though; see Umbach 2005; Toosarvandani 2014, among many others).
When it comes to German discourse particles which encode contrast like doch, an interesting point of comparison with séna7 is that – at least in many people’s analyses – particles like doch presuppose that certain information is in the common ground (for discussion, see Karagjosova 2009; Egg 2010; Grosz 2011, 2020; Zimmermann 2011, among many others). We showed in Section 3.4 that the unexpectedness requirement of séna7 need only hold for the speaker; it is not a traditional presupposition (although it is in the not-at-issue dimension). This is in line with research showing that St’át’imcets in general lacks presuppositions which refer to the common ground (Matthewson 1998, 2006a, 2009).
A final observation about frustratives crosslinguistically is that they belong to a range of grammatical categories and appear in different syntactic positions. Séna7 is a sentence-level adverb, cem is a pre-verbal particle, dara is a second-position clitic and AIA morphology is a paradigm of verbal inflection. This speaks against a possible cartographic approach in which there would be a dedicated position in the syntactic spine where frustrative semantics is located. Instead, it suggests a ‘semantic building blocks’ approach (cf. von Fintel and Matthewson 2008; Hale 1986), whereby small pieces of meaning recur crosslinguistically, sometimes combining with other semantic building blocks inside single morphemes, and they are distributed across different parts of the syntactic architecture.
We have proposed an analysis of St’át’imcets séna7 which involves only the standard tools used to analyze modal elements, but in the not-at-issue dimension. To capture séna7’s contribution, it is not necessary to rely on concepts such as forces or efficacy, or even causation. Séna7 can be modeled using simply quantification over stereotypical, epistemically accessible worlds. A strong hypothesis would be that all frustratives can be dealt with in this fashion. As Copley (2005) originally pointed out with respect to Tohono O’odham cem: ‘As exotic as it initially may look to English speakers, cem turns out to be only minimally different from other, more familiar modals.’
Funding source: SSHRC Insight Grants
Award Identifier / Grant number: 435-2015-1694
Award Identifier / Grant number: 435-2016-0381
Please see the first footnote for our acknowledgments to the St’át’imcets speakers who made this work possible. We would also like to thank audiences at the 51st International Conference on Salish and Neighboring Languages at Tla’amin, British Columbia, the Leibniz-Zentrum Allgemeine Sprachwissenschaft, Semantics of Under-Represented Languages in the Americas (SULA) 9, the Workshop on Formal, Probabilistic and Typological Approaches to Discourse Particles and Modal Adverbs at the 28th ESSLLI, and the University of British Columbia, as well as two anonymous reviewers.
Research funding: This work was supported by SSHRC Insight Grant 435-2015-1694 to the first author and SSHRC Insight Grant #435-2016-0381 to the second author.
Adaskina, Yulia V. 2005. Tipologicheskii analiz konstruktsii s frustrativnym znacheniem [Typological analysis of constructions with frustrative value]. Moscow State University MA thesis.Search in Google Scholar
Alexander, Carl. 2016. In Sqwéqwel’ Múta7 Sptakwlh: St’át’imcets Narratives by Qwa7yan’ak (Carl Alexander), [transcribed, translated and edited by Elliot Callahan, Henry Davis, John Lyon & Lisa Matthewson]. Vancouver & Lillooet: UBCOPL & USLCES.Search in Google Scholar
Alexander, Carl, Henry Davis, Beverley Frank, Gertrude Ned, Desmond Peters Sr., Linda Redan, Clara Shields, Ceda Scotchman & Rose Whitley. in prep. Nqwal’uttenlhkálha: An English to Upper St’át’imcets dictionary. Vancouver & Lillooet: UBCOPL & USLCES.Search in Google Scholar
Alonso-Ovalle, Luis & Henrison Hsieh. 2017a. Causes and expectations: On the interpretation of the Tagalog ability/involuntary action form. In Dan, Burgdorf, Jacob, Collard, Sireemas, Maspong, & Brynhildur, Stefánsdóttir (eds.), Proceedings of SALT 27. 75–94.Search in Google Scholar
Alonso-Ovalle, Luis & Henrison Hsieh. 2017b. Overcoming the unexpected: The Tagalog ability/involuntary action verbal inflection. Paper presented at the Austronesian Formal Linguistics Association 24. Seattle, WA: University of Washington.Search in Google Scholar
Alonso-Ovalle, Luis & Henrison Hsieh. 2018. Overcoming the unexpected: The Tagalog ability/involuntary action form. In William Bennett, Lindsay Hracs & Denis Storoshenko (eds.), West Coast Conference on Formal Linguistics (WCCFL 35), 59–68. Somerville, MA: Cascadilla Proceedings Project.Search in Google Scholar
Bar-el, Leora. 2005. Aspectual distinctions in Skwxwú7mesh. Vancouver: University of British Columbia dissertation.Search in Google Scholar
Bar-el, Leora, Henry Davis & Lisa Matthewson 2005. On non-culminating accomplishments. In Proceedings of NELS, vol. 35, 87–102.Search in Google Scholar
Cable, Seth. 2017. The implicatures of optional past tense in Tlingit and the implications for ‘discontinuous past’. Natural Language and Linguistic Theory 35(3). 635–681.10.1007/s11049-016-9355-7Search in Google Scholar
Carol, Javier & Andrés Pablo Salanova. 2017. Frustratives and aspect. Handout from Conference on Indigenous Languages of Latin America 8.Search in Google Scholar
Copley, Bridget. 2002. The semantics of the future. Cambridge, MA: MIT dissertation.Search in Google Scholar
Copley, Bridget. 2005. When the actual world isn’t inertial: Tohono O’odham cem. In Michael Becker & Andrew McKenzie (eds.), Proceedings of the 3rd Conference on the Semantics of Underrepresented Languages in the Americas (University of Massachusetts Occasional Papers 33), 1–18. Amherst, MA: GLSA.Search in Google Scholar
Copley, Bridget & Heidi Harley. 2014. Eliminating causative entailments with the force-theoretic framework: The case of the Tohono O’odham frustrative cem. In Bridget Copley & Fabienne Martin (eds.), Causation in grammatical structures (Oxford Studies in Theoretical Linguistics 52), 120–151. Oxford: Oxford University Press.10.1093/acprof:oso/9780199672073.003.0006Search in Google Scholar
Davis, Henry. 2017. A teacher’s grammar of Upper St’át’imcets. University of British Columbia Ms.Search in Google Scholar
Davis, Henry & Lisa, Matthewson. 2016. Against all expectations: The meaning of St’át’imcets séna7. In Marianne Huijsmans, Thomas J. Heins, Oksana Tkachman & Natalie Weber (eds.), Papers for the International Conference on Salish and Neighbouring Languages 51 (UBC Working Papers in Linguistics 42), 37–67. Vancouver, BC: University of British Columbia.Search in Google Scholar
Davis, John. 1978. Pronominal paradigms in Sliammon. In Papers for the 13th international conference on Salish and neighbouring languages, 208–236. https://lingpapers.sites.olt.ubc.ca/files/2018/03/1978_Davis1.pdf.Search in Google Scholar
De Veaugh-Geiss, Joseph P. 2014. Wohl and the semantics of assumptions. Potsdam: University of Potsdam MSc thesis.Search in Google Scholar
Dunlop, Britt, Suzanne Gessner, Tracey Herbert & Aliana Parker. 2018. Report on the status of B.C. first nations languages. Brentwood Bay, BC: First Peoples’ Cultural Council.Search in Google Scholar
Edwards, Bill, Martina LaRochelle & Sam Mitchell. 2017. Sqwéqwel’s nelh Skelkékla7lhkálha (Tales of our Elders): Stories by Bill Edwards, Martina LaRochelle and Sam Mitchell, [transcribed, translated and edited by Davis, Henry, John Lyon, Jan van Eijk & Rose Agnes Whitley]. Vancouver and Lillooet, BC: University of British Columbia Occasional Papers in Linguistics and the Upper St’át’imc Language, Culture and Education Society.Search in Google Scholar
Egg, Markus. 2010. A unified account of the semantics of discourse particles. Proceedings of SIGDIAL 2010. 132–138.Search in Google Scholar
Egg, Markus & Malte Zimmermann. 2012. Stressed out! Accented discourse particles: The case of doch. Sinn und Bedeutung 16(1). 225–238.Search in Google Scholar
van Eijk, Jan. 1997. The Lillooet language: Phonology, morphology, syntax. Vancouver: UBC Press.Search in Google Scholar
van Eijk, Jan. 2013. Lillooet-English dictionary, vol. 2. Vancouver, BC: University of British Columbia Occasional Papers in Linguistics.Search in Google Scholar
van Eijk, Jan & Lorna Williams. 1981. Cuystwí malh ucwalmícwts: Lillooet legends and stories. Mount Currie: Ts’zil Publishing House.Search in Google Scholar
Filip, Hana. 2012. Lexical aspect. In Robert Binnick (ed.), The Oxford handbook of tense and aspect, 721–751. Oxford: Oxford University Press.10.1093/oxfordhb/9780195381979.013.0025Search in Google Scholar
Filip, Hana. 2020. Lexical aspect (Aktionsart). In Daniel Gutzmann, Lisa Matthewson, Cécile Meier, Hotze Rullmann & T. Ede Zimmermann (eds.), The Wiley Blackwell companion to semantics. Hoboken, NJ: John Wiley and Sons.10.1002/9781118788516.sem038Search in Google Scholar
Gerdts, Donna B. 2008. Halkomelem limited control constructions. Paper presented at the University of British Columbia.Search in Google Scholar
Glougie, Jennifer. 2008. Future expressions in St’át’imcets. University of British Columbia Ms.Search in Google Scholar
Grosz, Patrick. 2011. German doch: An element that triggers a contrast presupposition. Proceedings of the Chicago Linguistic Society 46. 163–177.Search in Google Scholar
Grosz, Patrick. 2020. Discourse particles. In Daniel Gutzmann, Lisa Matthewson, Cécile Meier, Hotze Rullmann & T. Ede Zimmermann (eds.), The Wiley Blackwell companion to semantics. Hoboken, NJ: John Wiley and Sons.10.1002/9781118788516.sem047Search in Google Scholar
Haiman, John. 1988. Inconsequential clauses in Hua and the typology of clauses. In John Haiman & Sandra A. Thompson (eds.), Clause combining in grammar and discourse, 49–69. Amsterdam & Philadelphia: John Benjamins.10.1075/tsl.18.05haiSearch in Google Scholar
Hale, Kenneth. 1986. Notes of world view and semantic categories: some Warlpiri examples. In Pieter Muysken & Henk van Riemsdijk (eds.), Features and projections, 233–254. Dordrecht: Foris.10.1515/9783110871661-009Search in Google Scholar
Hinterwimmer, Stefan & Cornelia Ebert. 2018. A comparison of the modal particles fei and aber. Proceedings of Sinn und Bedeutung (SuB) 22, 469–486.Search in Google Scholar
Jacobs, Peter. 2011. Control in Skwxwú7mesh. Vancouver, BC: University of British Columbia dissertation.Search in Google Scholar
Jasinskaja, Katja & Henk Zeevat. 2008. Explaining additive, adversative, and contrast marking in Russian and English. Revue de Sémantique et Pragmatique 24. 65–91.Search in Google Scholar
Kiyota, Masaru. 2008. Situation aspect and viewpoint aspect: From Salish to Japanese. Vancouver: University of British Columbia Dissertation.Search in Google Scholar
Kratzer, Angelika. 1991. Modality. In Dieter Wunderlich & Arnim von Stechow (eds.), Semantics: An international handbook of contemporary research, 639–650. Berlin & New York: De Gruyter.10.1515/9783110126969.7.639Search in Google Scholar
Krifka, Manfred. 2011. Varieties of semantic evidence. In Claudia Maienborn, Klaus von Heusinger & Paul Portner (eds.), Handbook of semantics: An international handbook of natural language meaning, vol 1, 242–267. Berlin & Boston: De Gruyter Mouton.Search in Google Scholar
Louie, Meagan. 2014. The temporal semantics of actions and circumstance in Blackfoot. Vancouver & Kelowna: University of British Columbia dissertation.Search in Google Scholar
Matthewson, Lisa. 1998. Determiner systems and quantificational strategies: Evidence from Salish. The Hague: Holland Academic Graphics.Search in Google Scholar
Matthewson, Lisa. 2004a. On the absence of telic accomplishments in St’át’imcets. In Cristine Ravinski & Yunhee Chung (eds.), Proceedings of the Workshop on Structure and Constituency in the Languages of the Americas [WSCLA] 9 (UBCWPL 15), 65–78. Vancouver, BC: University of British Columbia.Search in Google Scholar
Matthewson, Lisa. 2005. When I was small – I wan kwikws: A grammatical analysis of St’át’imcets oral narratives. Vancouver, BC: UBC Press.Search in Google Scholar
Matthewson, Lisa. 2006a. Presuppositions and cross-linguistic variation. Proceedings of the North East Linguistics Society 36. Amherst, MA: GLSA.Search in Google Scholar
Matthewson, Lisa. 2009. Pronouns, presuppositions, and semantic variation. In Proceedings of SALT XVIII. 527–550.Search in Google Scholar
Menzies, Peter & Helen Beebee. 2020. Counterfactual theories of causation. In Edward N. Zalta (ed.), The Stanford encyclopedia of philosophy (summer 2020 edition). Available at: https://plato.stanford.edu/archives/sum2020/entries/causation-counterfactual/.Search in Google Scholar
Mitchell, Sam. in press. Wa7 Sqwéqwel's Sam: St'át'imcets Stories from Sam Mitchell. [transcribed, translated and edited by John Lyon & Henry Davis]. Vancouver & Lillooet: PNWLL Press and the Upper St’át’imcets Language, Culture and Education Society.Search in Google Scholar
Overall, Simon E. 2017. A typology of frustrative marking in Amazonian languages. In Alexandra Aikhenvald & R. M. W. Dixon (eds.), The Cambridge handbook of linguistic typology, 477–512. Cambridge: Cambridge University Press.10.1017/9781316135716.015Search in Google Scholar
Reisinger, D. K. E. & Marianne Huijsmans. 2019. ǰaqaʔ: A generalized exclamation operator in ʔayʔaǰuθəm. In D. K. E. Reisinger & Gloria Mellesmoen (eds.), Papers of the ICSNL 54, 171–196. Vancouver, BC: UBCWPL.Search in Google Scholar
Tonhauser, Judith & Lisa Matthewson. 2016. Empirical evidence in research on meaning. The Ohio State University & University of British Columbia Ms. Available at https://ling.auf.net/lingbuzz/002595.Search in Google Scholar
Turner, Claire. 2011. Representing events in Saanich (Northern Straits Salish): The interaction of aspect and valence. Guildford: University of Surrey dissertation.Search in Google Scholar
van der Voort, Hein. 2000. A grammar of Kwaza: a description of an endangered and unclassified indigenous language of Southern Rondônia, Brazil. Leiden: Universiteit Leiden dissertation.Search in Google Scholar
Watanabe, Honore. 2003. A morphological description of Sliammon, Mainland Comox Salish. Osaka: Endangered Languages of the Pacific Rim Publications.Search in Google Scholar
Zeevat, Henk. 2005. A dynamic approach to discourse particles. In Kerstin Fischer (ed.), Approaches to discourse particles. Amsterdam: Elsevier.Search in Google Scholar
Zimmermann, Malte. 2011. Discourse particles. In Claudia Maienborn, Klaus von Heusinger & Paul Portner (eds.), Semantics: An international handbook of natural language meaning, vol. 2, 2012–2038. Berlin & Boston: De Gruyter Mouton.Search in Google Scholar
© 2022 the author(s), published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.