The Absconditive revealed : Attention alignment in the grammar of Coastal Marind

Speakers of Coastal Marind, a Papuan language of the Anim family, use a special inflectional form of the verb to signal that the state of affairs that the verb describes is outside the addressee’s current focus of attention. A central claim of the paper is that speakers use this verb form, which I call the Absconditive, to signal that the addressee should realign their attention to achieve shared access to the state of affairs. The paper describes the function of this attentional-epistemic grammatical category and provides examples of its use, mostly drawn from video recordings of face-to-face interaction. I also contrast the Absconditive with constructions with related functions, such as the use of morphology expressing information-structural notions (narrow focus), and a verb form that appears to express shared access to a referent.


Introduction
The grammar of Coastal Marind, a Papuan language of Southern New Guinea, provides speakers with an array of inflectional options for marking a simple statement such as 'Mother is coming'. Three of the inflectional possibilities are shown in (1). In this paper I will argue that a key to understanding the uses of these alternatives is that they reflect differences in the attentional and/or epistemic states of the speaker and addressee-the types of grammatical meanings for which Evans et al. (2018a,b) propose the label engagement. Although the propositional content is identical in the three versions, which only differ in the strings of prefixed material before the verb stems, each of them would be appropriately used under a different set of discourse circumstances. (1a) is a focus construction, and would make a good reply to a content question ('Who is coming?') or a partial correction of a previous, incorrect statement ('Father is coming-No, Mother is coming'). In contrast to (1a), (1b) would be fine as an out-of-the-blue utterance, but also adds an attentional element. The Presentative prefix hat-, typically in combination with a gender proclitic, is mainly used in situations where the state of affairs is accessible to both speaker and addressee, e.g. because they are looking in the same direction and can see Mother approaching. (1c) likewise has an attentional component, and shows the structure that will be the main concern of this paper: a verb prefixed with the Absconditive prefix Vp-, here with various others) under the heading "locus", but other than some vague references to locational deixis (he calls them plaatsaanduidende praefixen), Geurtjens provides no explanation of their use. The grammar by Drabbe (1955) has the major advantage of a more consistent morphemic analysis, but offers few clues as to the function of the forms: Drabbe identifies the prefix series as markers of present tense (1955: 37-39) but notes that he can not find any difference in meaning between these prefixes and the other forms that he labels present tense (1955: 38).

Formal aspects 2.1 The place of the Absconditive within Coastal Marind inflection
Coastal Marind syntax can be classified as mildly verb-final, with the syntactic placement of non-verb constituents being relatively free. The immediately preverbal position hosts a focussed constituent, e.g. the interrogative pronoun or phrase in a content question. Nominal morphology is very limited: a few common nouns have overt marking of gender and number, and some kinship terms have possessor prefixes. There is no case marking of core argument roles. The main locus of complexity is the verb, which consists of a prefixal complex hosting a large number of inflectional categories, and a verb stem. The categories realised in the prefixal complex include participant indexing, markers of TAM and valency change, and a special set of prefixes that indicate the presence and role of a preverbal focussed constituent-these will be seen in Section 4.2 below. The verb stem follows the prefixal complex. Most verb stems can exhibit formal alternations realising a variety of categories (indexing of a patient-like participant, pluractionality, applicative formation). A position after the verb stem often hosts a suffix (marking e.g. aspect).
An important feature of the language is that phonological criteria split the morphological contents of the verb into two units: the prefixal complex is one phonological word, and the verb stem (including any following suffix) is another. Syntactically, however, the prefixal complex and the verb stem form an inseparable unit. In the morphemic segmentation used in interlinear glossing, I indicate the phonological independence of the two parts of the verb by means of a blank space, while simultaneously adding a hyphen after the final prefix of the prefixal complex to indicate that the string of prefixes forms a morphosyntactic unit with the ensuing verb stem. Although not licensed by the Leipzig Glossing Rules, I believe that this idiosyncratic solution is useful for visualising the bipartite structure of the verb, while at the same time allowing for easier matching of morphs with their category labels on the second line.
In Olsson (2017) I adopted a templatic model for the description of the prefixal complex, in which position classes are set up according to the paradigmatic restrictions of the prefixes (which prefixes are mutually exclusive) and their syntagmatic ordering (which prefix comes after which). Two prefixes are said to belong to the same position class if they (i) are mutually exclusive and (ii) always occur in the same position relative to prefixes belonging to other position classes. Applying these diagnostics, I arrived at a template with 17 prefixal position classes preceding the verb stem (numbered from −17 to −1; see Olsson, 2017: 205). The position class template can be thought of as a blueprint for composing a well-formed inflected verb: the template is scanned from left to right, and the relevant prefixes are lined up before the stem in the order that corresponds to their class membership. There is no obligation to add a prefix from any specific position class, but every inflected verb must contain a prefix indexing person/number of the agent-like participant (with default 3sg prefixing if the verb is avalent or only takes a patient-like participant); however, these 'Actor' prefixes belong to different position classes depending on their person specification, so 1st and 2nd person are in one class, and the 3rd person prefixes in another. All of the Actor prefixes still stand in paradigmatic opposition to each other, which shows that prefixes may be mutually exclusive even if they belong to separate position classes (another example is the Past Durative prefix, which is mutually exclusive with the Future prefixes, despite differing position class membership).
Some prefixes are mutually exclusive with all prefixes from several adjacent position classes, in which case I consider the prefix to be a "multi-class prefix" spanning more than one position class. The Absconditive is one of these multi-class prefixes and is mutually exclusive with a range of other prefixes occurring at the left edge of the position class template. The categories with which the Absconditive is incompatible on such morphotactic grounds include: some tense markers (e.g. the Perfect prefix), a prefix marking the verb as subordinate, the 'Orientation' prefix series indicating focus on the preverbal constituent, and various prefixes that mark verbs in non-declarative contexts (questions, imperatives, prohibitives). For many of the prefixes that compete with the Absconditive for the same prefixal slot, one could argue that they express conflicting semantic categories and therefore stand in paradigmatic opposition. For example, I will argue in Section 4.2 that clauses with the Absconditive lack the ability to express constituent focus, which is an important function of the Orientation prefixes. This difference would explain their mutual incompatibility, whereas for other prefixes I am not aware of any non-ad hoc explanation of the cooccurrence restrictions.² The Absconditive is also incompatible with some affixes in other parts of the verb, because of conflicting semantics, e.g. the Past Durative prefix (the Absconditive is restricted to present time reference). The set of prefixes with meanings that are the most closely related to the function of the Absconditive, which I labelled the 'Speaker attitude prefixes' in Olsson (2017), do not belong to the same class as the Absconditive-as could have been expected, were the template more neatly organised according to semantics-but appear in a separate position class closer to the verb stem. Except for the Presentative prefix hat-, which was glimpsed in (1b) and will be further discussed below, the combinatorics of the Absconditive and the Speaker attitude prefixes are not particularly well understood and will not be discussed further here.³ The corollary of this complicated templatic architecture is that it is not possible to fit the Absconditive into any clearly defined, paradigmatically contrasting set of markers with opposing engagement values. In this respect the engagement categories in Coastal Marind differ from the tight-knit engagement systems described for Andoke (Landaburu 2007) and Kogi (Bergqvist 2016), which both have an affixal position that is filled with one out of a set of mutually exclusive engagement markers. The system of oppositions that the Coastal Marind Absconditive enters into involves choices between tenses, clause types, and morphological marking of information structure, which has the consequence that the range of contexts that permit the engagement use of the Absconditive is limited to declarative main clauses about present states of affairs, that are presented as 'all-new' information. These language-internal mechanisms conspire to make the functional range of the Coastal Marind Absconditive comparatively limited, especially in comparison with more 'canonical' engagement markers in languages such as Andoke and Kogi, where the choice between different engagement appears to be more or less obligatory.

The Absconditive and its relationship to the demonstratives
The shape of the Absconditive prefix series is easiest to grasp by comparing them to the independent demonstratives. Coastal Marind has three demonstrative series, which can be represented as a variable initial vowel V-, realising agreement in gender and number, followed by a stem expressing one of three distance contrasts. The three demonstrative series are Proximal Vhe, Distal Vpe and Remote Vhan. The meanings of the first two series are straightforwardly glossed as expressing 'here' (the Proximal) vs. 'there', or perhaps rather, 'not here' (the Distal). The semantics of the Remote series are more complicated but need not concern us here, because only the Proximal and Distal demonstratives correspond to variants of the Absconditive (see Olsson, 2017: 84-86 for the Remote Vhan).
The independent demonstratives agree in gender and number whenever they are used in contexts that require agreement, e.g. when used as attributive modifiers within an NP ('this N', 'that N'), where they agree in gender/number with the head nominal. The details of the gender system are not relevant here, suffice  to say that gender I contains male humans (patul e-pe 'that boy'), gender II contains female humans and all other animates (nggat u-pe 'that dog'), whereas inanimates are divided between genders III (yahun e-pe 'that canoe, those canoes') and IV (katal i-pe 'that money'). Number is only distinguished for animates, and is marked on the demonstrative, but not on the noun itself (patul i-pe 'those boys', nggat i-pe 'those dogs').
The gender-number forms of the demonstratives are tabulated in Figure 1, with the gender marking vowel separated from the stem by means of a hyphen. In non-agreeing contexts, e.g. when used as demonstrative adverbials, the demonstratives appear in their gender III form per default, e.g. ehe 'here' and epe 'there'. The Absconditive prefixes most likely derive from preposed demonstratives that have been incorporated into the prefixal complex. As seen in Figure 2, the members of the Absconditive series are identical to the corresponding forms of the demonstratives, except that the final /-e/ of the demonstratives is missing. I stress that this similarity should be understood in diachronic terms: although there are various constructions that involve preposing a deictic element (sometimes realised as a proclitic) to the verb, it is clear that the Absconditive series must be described as prefixes, because they trigger allomorphic alternations in adjacent material in the same way as other prefixes do, and obey the kind of morphotactic combinatorics displayed by the other categories in the position class template. Crucially, the Absconditive series also differs in meaning from the independent demonstratives. In most uses the Absconditive lacks the concrete deictic function of pointing to an entity, whereas this meaning is central to the demonstratives. This means that the contrast between Proximal and Distal is largely neutralised for the Absconditive: the forms based on the Distal (Vp-etc.) can be considered the unmarked form of the Absconditive, and the contrast between Proximal Vhand Distal Vpis only relevant in the few contexts where the Absconditive retains the function of pointing to an entity, as discussed in Section 3.3. The difference in form between the free demonstratives and the Absconditive prefixes, as well as the accompanying meaning differences, indicate that the development of the prefixes from the demonstratives should be understood in terms of grammaticalisation, although I will have little to say about the exact process here (but see Section 5 for a suggestion; see also Evans et al., 2018b: 159ff. and Schapper and San Roque 2011 for the grammaticalisation of demonstratives into stance-marking particles with clausal scope).

The Absconditive as an engagement marker
This section describes the most important aspects of the use of the Absconditive category. I discuss its restriction to clauses with present time reference (Section 3.1), its use to realign the attentional focus of the addressee (Section 3.2), the lack of deictic contrast in most uses of the Absconditive, which clearly distinguishes it from the independent demonstratives (Section 3.3), and a use that seems to emphasise that the information contradicts the beliefs of the addressee (Section 3.4).

The present tense restriction
One of the primary conditions that restrict the distribution of the Absconditive is its limitation to present tense contexts. Tense prefixes such as the Past Durative dor Future meare incompatible with the Absconditive, which only appears in verb forms that are unmarked for tense, and can be interpreted as having present time reference.
(2) a. *upa-d- The restriction to present time contexts also implies that the Absconditive only combines with verbs that are aspectually durative, such as ɣet 'be moving'. Punctual verbs may not appear in present time contexts, unless they have been marked with durativising morphology, e.g. the 'Extended' suffix -la (-a after a consonant), which derives stative-resultative verbs from the corresponding change-of-state verbs. Compare punctual tapeb 'fly away, take off flying' with durative tapeb-a 'be flying, in flight': The present tense restriction has the consequence that the Absconditive is largely absent from the text type that dominates my corpus: past tense narratives. Most occurrences of the Absconditive in narratives are found in reported speech. In Coastal Marind narratives, such material is always rendered as 'direct speech', i.e. the speech is quoted verbatim with pronouns, tense markers and other deictic elements remaining intact. Therefore it is not surprising that the Absconditive is common in narrated dialogue, just like it is common in faceto-face conversation. In this paper, I do not use reported speech data taken from narratives. My claim is that a function of the Absconditive prefixes is to manage the addressee's attention, which in the case of reported speech would mean that the character to whom the speech is attributed is using the Absconditive to manipulate the attention of the character that s/he is addressing within the story. Using reported speech in narratives to support my description of the Absconditive would depend on the interpretation of such imagined speech situations. It is better to use video recordings of actual face-to-face interaction to study the use of the Absconditive, because the ability to observe the physical orientation of the interlocutors and the referent, along with pointing gestures and gaze direction, makes it possible to avoid some of the subjectivity associated with the exegesis of narrated dialogue.⁴ It is likely that there are other restrictions on the clause types in which the Absconditive can be used. The Absconditive is so far unattested in negated clauses. As mentioned in Section 2.1, the prefix is not compatible with the non-declarative morphology used to form questions and imperatives, although I have observed the Absconditive in yes/no-questions formed with the tag-like sentence-final particle ay. Because of the lack of data, I will leave the use of the Absconditive in such clause types for future studies.

Realignment of the addressee's attention
The most straightforward, and probably most common, use of the Absconditive prefixes in face-to-face conversation is found in situations where the addressee is engaged in some activity and the speaker points out something that is taking place outside the addressee's field of vision or is judged to be outside the addressee's current focus. During the early stages of my fieldwork, when I was trying to understand how the Dutch colonial descriptions of Marind related to the language spoken by villagers in Wambi, it struck me that I heard the Absconditive prefixes (the "present tense" according to Drabbe's grammar) mostly when speakers with whom I was walking along the beach were pointing out something happening in the distance ("Mother is sitting over there") or when trekking through the bush with kids searching for birds to shoot ("A cockatoo is flying there"). These observations correspond to the most concrete sense in which the Absconditive prefixes mark a state of affairs as being outside the addressee's focus of attention.
I will provide some initial illustrations of similar situations of non-alignment using data from a video recording of some teenagers cleaning the well outside my house in Wambi. The recording of this task-which involved emptying most of the water in the well and removing algae from its bottom and walls-turns out to be one of the most high-yielding sources of Absconditive tokens in my corpus, with a total of 7 occurrences in the 15 minutes that I transcribed during my fieldwork. It is easy to see why: most of the action and the accompanying conversation is focussed on the main project of emptying the well, but this sphere of shared attention is occasionally punctuated by remarks (either by the main participants or by passers-by) about other events unfolding in the proximity. Most of these statements are implicitly requesting the addressees to redirect their focus of attention, and are therefore marked with the Absconditive prefixes.
Consider first the scene depicted in the video still of Figure 3. The young man (Pau) on the left is drawing water from the well (his blurred, right arm is pulling up a bucket) and has just instructed two children (the boy on the right and the small girl in the background) to take off their clothes so that he can bathe them using water from the well. The small girl responds enthusiastically and starts to remove her shirt, encouraged by Pau who addresses her while continuing to draw water, whereas the small boy, who has just said that he does not want to bathe at the moment, looks down into the well and announces that a fish is swimming in the well. His utterance is given here as example (4). Pau, who is intended as the main addressee of this utterance, is not attending to the bottom of the well, so the use of the Absconditive prefix is a signal to the addressee to redirect his attention so that he can perceive the (in this case probably imagined) fish in the well. Consider also the configuration in Figure 4, which represents the next occurrence of the Absconditive prefix in this recording, a few minutes later. In the image, Pau (in the middle) and his sister Yakoba (to the right) are seen drawing water, while the speaker, Dula, is standing on the far left. Dula is watching a boy approaching from further away, behind Yakoba's back (not visible in the video still). The boy had been asked earlier to bring buckets. Dula wants Pau and Yakoba to shift their attention away from the well to the approaching boy, and signals this by using a verb prefixed with the Absconditive, as given in (5). Note that only the first clause in his utterance employs the Absconditive; the second clause is a focus construction (with a focussed preverbal object argument, so the 'Object Orientation' prefix mis used) and can be seen as answering an implicit question 'How many is he bringing?'. Focus constructions do not allow the use of the Absconditive prefixes, as mentioned in Section 2.1.

The deictic contrast
An important aspect of the claim that the Absconditive prefix series has grammaticalised from free demonstratives into bound engagement markers is the fact that the deictic contrast present in the demonstratives (ehe 'this', epe 'that') has largely been lost in their prefixed descendants. The Distal demonstrative series (based on the stem Vpe) correspond to the unmarked Absconditive series Vp-, which is used in contexts of diverging attention regardless of the location of the state of affairs described by the clause. This is a crucial detail, because it would not be particularly noteworthy if the Absconditive series retained the full meaning of the free demonstratives, and the only feature that distinguished them were that the Absconditive prefixes are bound to the verb. Instead, the Absconditive series has lost the most salient feature of the demonstratives (the deictic distance contrast) and acquired a meaning that the demonstratives lack, viz. the attention-realignment function (see e.g. Traugott (2003) for the development of subjective and intersubjective functions in grammaticalisation). In this section I discuss the lack of a distance specification in the main uses of Absconditive, and also show that it is still retained in copula clauses. The examples of the Absconditive in use that have been presented so far are compatible with a description of them as being ordinary demonstratives attached to the verb, retaining their deictic meaning. According to this description, the meaning of the Absconditive prefix upin example (4) would be the same as the meaning of the independent demonstrative upe 'that', i.e. pointing to a fish that is 'not here'. Similarly, in (5) the prefix epwould be equivalent to the demonstrative epe 'that', pointing to the boy in the distance. This description is wrong, in my view: verbs prefixed with the Absconditive assert states of affairs, and do not point to referents, because they lack the function of spatial deixis associated with demonstratives. The clearest evidence for the lack of a distance specification in the Absconditive is the use of the unmarked Vpseries (derived from the Distal demonstratives) in contexts where the asserted state of affairs is 'here', e.g. is taking place within the space that the speaker could not refer to by using the Distal demonstrative epe 'there'. I will illustrate this with an example from the corpus.
The utterance given in (6) occurs later in the same recording as the previous examples. The participants had been looking for an additional rope for tying to a second bucket, which would enable two people to draw water from the well simultaneously. As seen in Figure 5, the space around the well was getting crowded as people from the surrounding houses came to see what was going on and what I was filming. The speaker in this example-the young boy wearing a Jamaican-themed sleeveless shirt and grasping a stick with his left hand-had been discussing where to find a bamboo pole to use for climbing down into the well when two small children squeezed past on each side of him and peered down the well. He interrupts the discussion about the bamboo and makes the jocular comment in (6),⁵ which in turn triggers a response from Yakoba who screams at the children to go away. The utterance is (6) is intended for the participants drawing water from the well, and refers to the two small boys peeking out on both sides of the speaker. If a demonstrative had been used to point to the small boys in this spatial configuration, the only possible choice would have been the Proximal Vhe-series, i.e. ihe isahih 'these children', because the referents are in the close proximity of the speaker and the addressees are further away. In contrast to the demonstratives, the Absconditive prefix in (6) is insensitive to the distance of the referents and appears in its unmarked Vp-form. Further support for the claim that the distance contrast is neutralised in the Absconditive comes from the fact that the Vp-forms may combine with 1st person verb forms. No such forms occur in my corpus, but speakers told me that this combination is natural and commonly used.⁶ The example in (7) was originally composed by me and sent in a text message to a Coastal Marind speaker who lives in the provincial capital Merauke. I was late for an appointment, and my friend texted me asking whether I was coming or not. I guessed that my response might provide an appropriate context for an Absconditive 1st person form, but when I arrived at the destination the recipient informed me that my wording had been wrong. He suggested the following context for the appropriate use of this expression: a group of people are walking on a path through the bush, and a person walking further ahead asks 'Where is Bruno?'. Then I can reply by shouting the form in (7).⁷ (7) epa-no- Again, this form uses the unmarked Absconditive Vp-series, derived from the Distal Vpe-demonstratives, but the Distal forms of the independent demonstratives would not be possible in a context where I refer to my own location while addressing someone further away, which would require the Proximal Vhe-demonstratives. This distance contrast is not relevant for the Absconditive prefixes, in which the spatial deixis component has been bleached, presumably as it went through the grammaticalisation process from free demonstrative to a bound inflectional prefix.
As mentioned above, there is one environment in which the Absconditive series maintain the proximal/distal distance contrast, so that the Proximal Vh-series are used when the referent is (presented as being) close to the speaker, and the Distal Vpwhen it is (presented as being) outside the vicinity of the speaker. This happens when the Absconditive is prefixed to the copula, which in non-past contexts (recall that the Absconditive is restricted to present tense clauses) consists of a zero stem (given here as 'Ø') optionally followed by the Imperfective suffix -e. The use of the Absconditive in copula clauses differs from its use with standard verbs, because such copula clauses are used to talk about the existence of some entity. Intuitively, such clauses place emphasis on this entity, whereas in other contexts it is the entire proposition that the speaker tries to bring to the attention of the addressee. Therefore it makes sense to talk about referents when discussing copula clauses, whereas I use the term state of affairs when talking about the crucial piece of information that standard sentences with the Absconditive convey.
In copula clauses, the distance contrast between Vhand Vpis straightforward and can be illustrated by the two observed utterances in (8) and (9) below. In the first example, an elderly, blind man is groping around among his belongings looking for his tobacco. His son is sitting across the room with the father's tobacco in front of himself, and informs the father of this by saying (8). His utterance uses the Proximal version of the Absconditive, because the tobacco is distant from the addressee, but close to the speaker, who is the deictic centre. In this context, demonstrative reference to the location of the tobacco would likewise be made using the Proximal demonstrative ehe 'here'. In the second example, a mother is sitting with her daughter on her lap, with the daughter facing the same way as herself (so that the daughter's face is not visible to the mother). Across from her sits her mother-in-law, who notices mucus coming from the child's nose, and informs the daughter-in-law of this by means of (9). Since the referent (the mucus) is distant from the speaker, she uses 6 See also the observed utterance in (10) below. Absconditive-marked verbs with 1st and 2nd person subjects are frequently attested in relative clauses, but in this specific construction the Absconditive lacks the engagement semantics and simply marks the verb as subordinate, as mentioned in Section 5. Independent verbs with 2nd person subjects marked with the Absconditive are completely absent from my corpus-but see footnote 1. 7 My interpretation of the infelicity of (7) in this SMS conversation is that the context does not involve a realignment of attention. The question 'Are you coming or not?' is appropriately answered by giving the correct alternative, e.g. no-manem 'I'm coming', without the Absconditive, as the question already presupposes the possibility of me coming. In the context suggested to me by the Coastal Marind speaker, the question 'Where is Bruno?' suggests that (the speaker believes that) Bruno is not here, and thus invites a correction of this false belief in the form of a request to realign the focus of attention.
the Distal version of the Absconditive, just like she could have used the Distal demonstrative epe 'there' to refer to the location of the mucus. Note that throughout this paper, the Distal member of the Proximal/Distal Absconditive pair is treated as the unmarked option, and glossed ' ', whereas the Proximal is glossed ' : ' in the morphemic analysis.  (8) and (9), the addressee's attention must be realigned with that of the speaker before the intended referents can be perceived, which is what motivates the use of the Absconditive. Although I have no clear explanation for why the distance contrast in the Absconditive is present with the copula but not with standard verbs, it seems likely this has to do with the fact that this type of predication puts more emphasis on the location of an entity than the type of activity in which it is involved. Such copula clauses are primarily intended to make the addressee attend to the location of some entity, rather than a state of affairs, which is reflected in the choice between the Proximal and Distal/unmarked versions of the Absconditive, according to the relative proximity of the referent in question.⁸

Updating common ground: inaccessible and contradictory information
The data given above exemplified the use of the Absconditive in contexts where the addressee's attention needed to be redirected in order to achieve mutual access to the state of affairs. In this section I will claimalthough so far there is very little data bearing on the issue-that the Absconditive has a more abstract attention-realignment use, because it can be used to assert a present state of affairs that the speaker judges to be inaccessible to the addressee, not because of a lack of visual attention, but because the assertion of the state of affairs contradicts some belief held by the addressee, or because it concerns something that the addressee cannot have knowledge about (it is outside the addressee's 'epistemic realm'), such as the mental activities of the speaker. This further supports the idea that the function of the Absconditive (at least when used with verbs other than the copula) is not to point out entities, but rather states of affairs. As in the more concrete uses in contexts of diverging perceptual attention, the use of the Absconditive should probably be seen as a request that the addressee accept this attentional or epistemic update so that it becomes part of the common ground. I base these claims mainly on observed-not recorded-uses, so my evidence is weak and the claims must be reevaluated once more spontaneous language use has been documented. The idea that the Absconditive can signal that the statement contradicts a belief of the addressee is based on uses such as the one in (10). During a feast in the village, I had been assigned to a chair next to an elderly woman with whom I had not interacted previously. The woman spoke to me in Coastal Marind as we watched the festivities, but during a hiatus in the conversation the villager who had assigned me to the chair happened to walk past and must have assumed that we had been sitting in silence, because he told the woman that she should talk to me, so that I learn the language. The woman retorted as in (10). Based on the discourse context and later discussion of this example with speakers, I interpret this use of the Absconditive as signalling that it asserts something that contradicts the assumptions of the addressee-in this case, that the woman had not been talking to me. Note that the most common forms of such corrective statements involve replacive focus, as in the minidialogue 'Father is coming-No, Mother is coming' mentioned in connection with (1a) above. Sentences with constituent focus do not allow the use of the Absconditive. The statement in (10) does not involve constituent focus, rather it expresses what is known as polarity focus (Halliday 1967) or verum focus (Lohnstein 2016), emphasising the truth of the statement, as opposed to the belief of the interlocutor. I have observed a possibly related use with statements that report private experiences that are not directly observable to others. For example, discussing video recordings and words written in my notebooks I have heard speakers use the Absconditive with the verb 'to see', in the sense 'Yes, I can see it' (epano-idihe). It seems reasonable that it is the inaccessibility to others of the private experience of being able to see something that motivates the use of the Absconditive. Support for this interpretation also comes from the textual example in (11), which is an adapted⁹ version of a sentence from one of the texts collected by Petrus Drabbe. This is one of the rare occurrences of the Absconditive (Drabbe: present tense) in the corpus of narratives collected by Drabbe. It occurs with the verb hwetok 'think, remember', so one could speculate that the use of the Absconditive reflects the fact that only the speaker has the 'epistemic authority' (see e.g.

The Absconditive contrasted
It is instructive to contrast the use of the Absconditive prefixes in contexts where the speaker tries to redirect the attention of the addressee with uses of other prefixes in contexts where the speaker is not attempting to redirect the attention of the addressee. Speakers often talk about presently occurring states of affairs without recourse to the Absconditive series, for example when mutual access has already been achieved, or in contexts where shared visual attention is not important or relevant for the discourse goals. In this section I discuss two grammatical options used in such contexts, the Presentative (Section 4.1) and the focus construction, especially as it is used in locative predication (Section 4.2).

Shared attention and the Presentative
A common option in contexts of shared attention is the Presentative prefix hatwhich was introduced briefly in connection with example (1b) above. This prefix typically occurs in the so-called Presentational construction, which involves either a gender proclitic or a demonstrative preceding the verb, which is inflected with the 'Neutral Orientation' prefix k-, and combines mainly with intransitive verbs describing the positional state or dynamic activity associated with the subject argument.
In the exchange given in (12) (see also the video still in Figure 6), Pau (A) is standing right behind Buang (B), and asks where he can find a certain bamboo pole. They are both oriented so that they have approximately the same view of the surroundings. Buang replies as in line 2 and points in a direction that is more or less directly in front of them. Unlike the contexts in which the Absconditive is used, the addressee does not need to reorient himself to gain access to the referent, as it is more or less right in front of him, and this is emphasised by the use of the Presentative in Buang's response. I have often observed the use of the Presentative in utterances that follow an earlier turn using the same verb marked with the Absconditive prefix. The turn featuring the Absconditive is used before mutual attention has been achieved, after which the second turn, marked with the Presentative, confirms that the mutual lockin has been established. Example (13) shows one such exchange from the corpus. I was recording a narrative monologue, and a dog was lying on the ground behind the person who was talking into the camera. Another villager (A) was sitting behind the camera and watched the action on the monitor of the digital video camera as I started the recording. She found it amusing that the sleeping dog was visible on the monitor, and alerted a second villager (B) sitting nearby about this. The utterance given in line 1 employs a verb marked with the Absconditive because the addressee's attention is not yet on the video camera, and A wants B to lean over and inspect the monitor so that she too can perceive the sleeping dog. B does so, and confirms that she sees the dog using the exclamation in line 2. A then repeats the verb from the first turn, now prefixed with the Presentative, which perhaps could be glossed as 'There it is lying, as both you and I can see'. Another context in which the Presentative is highly frequent is when a speaker hands over an item to somebody and says 'here you go', which in Coastal Marind is rendered with the copula inflected with a 2nd person Possessor prefix plus the Presentative, thus affirming shared attention on the gift that is being presented (cf. the discussion of the copula in Section 3.3).
In certain ways the Presentative seems to provide a mirror image of the Absconditive prefix, since it is used in situations where the speaker and addressee share perceptual access to a referent or state of affairs described by the verb. But a closer look at the more than 100 attestations of the Presentative in the corpus suggests that its function should perhaps not be described as signalling shared access: rather it seems to convey that the state of affairs is in 'plain sight', or easily perceivable, i.e. potentially accessible for anyone to attend to if they wish to do so. (Note that it is not restricted to visual perception. The Presentative is often used when the occurrence of a state of affairs is evident through sound only, e.g. a motorcycle approaching in the distance.) Evans et al. (2018a,b) emphasise that engagement markers make reference to the attentional or epistemic states of speaker and addressee, forming "grammaticalised systems for monitoring and adjusting intersubjective settings"-engagement is "grammaticalised intersubjectivity" (Evans et al., 2018a: 113). It is not clear that the Presentative has an intersubjective meaning, and in fact I believe that this makes the Presentative a much less canonical engagement category than the Absconditive, whose tracking of the addressee's non-attention makes it more clearly intersubjective. My main reason for suggesting that the Presentative makes little or no reference to the attentional state of the addressee is that it is common to hear the Presentative in discourse contexts where the addressee does not have access to the relevant state of affairs, which would be surprising on the assumption that the Presentative signals that the addressee already has access to the state of affairs (or referent). Instead the Presentative seems to have the more general meaning that no effort is required to gain access to the referent, either because it is right in front of the addressee or because it is easily perceptible more generally.
The following two examples are taken from a recording of a multi-participant conversation in my host family's kitchen in Wambi, and will be used to support the point that the Presentative does not imply any intention to redirect the addressee's focus of attention. The setting of the first example is seen in the video still in Figure 7. The two young men (Mili and Pau) leaning against poles in the left half of the picture are attending to a scene in the opposite house, where a boy is searching an elderly man for lice (visible through the door). In the background is Karel, who is trying to light a cigarette using ember from the stove, and has not noticed the couple in the neighbouring house. Mili (on the far left) is clearly amused by the activity taking place across from the kitchen, presumably because it is odd to see the couple (who live several houses away) engage in social grooming in somebody else's house, and in the middle of the day. Mili turns around and addresses Karel (lighting the cigarette), saying: In (14), Mili prefixes the Presentative to the verb despite the fact that the state of affairs is not visible to the addressee, and it is also clear from their interaction that he does not expect Karel to realign his attention so that he can perceive the event (which would require getting up and approaching the door) because after a short acknowledgement from Karel the conversation shifts to an unrelated topic.
Another similar use of the Presentative occurs later in the recording, when Yustina (the woman visible on the far right in Figure 7) leaves the kitchen and walks into the adjacent room, which is separated from the kitchen by a thin wall, against which I am leaning as I am recording the interaction in the kitchen. There she notices a bunch of bananas that had been given to me by a neighbour, and that are hanging from a crossbeam. The bananas are visible to anyone who enters the room, but they are not visible to me where I am sitting with the video camera, nor to anyone else in the kitchen. Remaining in the adjacent room, Yustina addresses the utterance in (15) to me. It is clear from the context that she intends this as a reminder for me to take care of the bananas, perhaps by eating them, distributing them, or hiding them away. I interpret the use of the Presentative in her utterance as emphasising the fact that the bananas are out in the open, plainly visible, including to people who might steal them unless I stash them away. In summary, my understanding is that the Absconditive and the Presentative prefixes are far from complementary in their meaning, although their distribution (the Absconditive in contexts of diverging attention, the Presentative in contexts of shared access) might at first glance suggest that they encode opposite values of some engagement parameter. But a speaker using the Presentative does not seem to be making any claim about the attentional or epistemic state of the addressee. Instead, the focus is on the objective property of the referent as being publicly visible or accessible, not on the (inter-)subjective state of the addressee as also attending to or being aware of the referent (or state of affairs) that the speaker is pointing out.
This description is compatible both with the prototypical presentational uses of the Presentative (as when the speaker presents an item by handing it over to the addressee), and with the uses that describe states of affairs that the addressee is not able to perceive, as in examples (14) and (15) above. In the latter examples the use of the Presentative seems to highlight that the states of affairs are in public view, which in (14) probably adds a comic nuance to the utterance, while in (15) it probably highlights the need to take care of the bananas before they get stolen.¹⁰

Focus constructions and locative predication
A recurring context that involves a clear case of epistemic asymmetry, but in which the Absconditive is almost never found, is furnished by questions and their answers. This was hinted at in connection with example (1a), which provided an example of the specialised focus structure that is used in such question-answer pairs in Coastal Marind. Content questions containing interrogative elements expressing who, where, etc. always place this interrogative phrase in the immediately preverbal position, while the verb is prefixed by one of the five so-called Orientation prefixes, marking the role of the interrogative in the clause (e.g. the Neutral k(a)for a preverbal subject, the Object m(a)for an object, and Locational nd(a)for a preverbal constituent expressing a location; see further Olsson, 2017: 293ff). Like focus constructions more generally, these structures can be seen as specifying a presupposed part (everything except the preverbal constituent) and an open variable (the preverbal constituent) which provides information that relates to the presupposed part. Clauses with the Absconditive, on the other hand, are essentially 'thetic', focus-less utterances (Sasse 1987), and present the entire state of affairs as outside the addressee's focus of attention. Although the Absconditive is common in statements about present events taking place in some specific location, it is virtually never used in answers to content questions. Answers to questions of the type 'Where is X?' typically receive a response consisting of a preverbal focussed constituent expressing the location, followed by the copula (which has a zero stem in non-past contexts) or a positional verb inflected by means of the Locational prefix nd-, as in Ø-e -'S/he's inside the house. ' [0061.19052015.2.dmh] A rare counterexample to the generalisation that the Absconditive never occurs in answers to content questions can be found in the recording of the youth cleaning the well-excerpts of which were used in examples (4), (5) and (12) above-but I would suggest that this attestation supports the claim that the Absconditive should be seen as an appeal to the addressee to align their attention with that of the speaker, because it is found in an exchange in which the inquiring speaker fails to locate the referent on the basis of the answer given as initial response.¹¹ The exchange is given in interlinearised form in (17), and represented by the video still in Figure 8. It involves a passer-by (A, standing behind the camera and not visible in the still) inquiring about the whereabouts of Meka, Yakoba (B), who is standing to the right of the well drawing water, and Yakobus (C), who is standing to the left of the well. A's yes/no question in line 1 indirectly asks 'Where is Meka?', and B answers using a standard focus structure (line 2), while gesturing quickly with her right hand towards a nearby house (her right hand is hidden behind her left arm, which is busy holding a rope attached to a bucket) and 'lip-pointing' (which here involves directing her face towards the referent, while furrowing her brows and making her lower lip protrude; she maintains this deictic gesture throughout the following turns) in the 10 There are some additional features of the Presentative-or rather the Presentational construction in which it occurs-that further distinguish it from the Absconditive, most importantly that it is used in situations where there is more emphasis on a prominent participant of the state of affairs rather than on the state of affairs itself. This is reflected in the fact that it almost only occurs with intransitive verbs (such as verbs of motion or position) and transitive verbs used in intransitive frames (such as reciprocals, and verbs with omitted objects; see also Olsson, 2017: 533-535) and that the deictic element preceding the verb shows gender agreement with the highlighted sole participant. The Absconditive, on the other hand, occurs with verbs of all transitivity types, and can agree in gender with any argument of the clause, because it usually does not single out any specific referent, but rather brings attention to the state of affairs as a whole. 11 I do not consider the sentence in (7) to be a counterexample, since it is not a preferred response to the content question 'Where is Bruno?', rather it denies the presupposition on which the question was based. same direction. B's deictic efforts are unsuccessful, because A cannot locate the referent (perhaps expecting Meka to be visible) and asks emphatically enda? 'Where?' (line 3).¹² At this point C makes his contribution to the exchange by extending his arm in a full pointing gesture towards the house and uttering a verb form marked with the Absconditive (line 4).  The use of the Absconditive in line 4 of example (17) runs counter to my observation that focus expressions (such as the one in line 2) constitute the preferred turn type in answers to content questions. I would suggest that C's utterance is strictly speaking not an answer to A's question about Meka, but rather an instruction explaining how A can make sense of B's seemingly incorrect answer in line 2 (Meka is not visible in the direction B is indicating). It is clear from C's answer that Meka is sitting inside the house (the Inessive prefix kais obligatory with positional verbs when they express position indoors)¹³ so C's answer is a request that A take this state of affairs into account, thereby shifting his attention away from the visible space in front of the house and aligning it with the shared focus that B was trying to establish in line 2.
I will end this section with another example of a focus structure with a demonstrative in the preverbal position, given in (18). This utterance provides an exact parallel to the data in (13) above, but shows the use of a focussed demonstrative, instead of the Absconditive, because the attentional focus of the addressee(s) is not at issue in the context. The speaker in (18) is a passer-by who asked to inspect the monitor on my video camera as I was filming the interaction around the well. He looked at the monitor, confirmed that the participants were visible on the small screen,¹⁴ and left without commenting further. In contrast to the speaker in (13), who wanted to draw the attention of her interlocutor to the dog that was visible on the monitor, it does not seem that the speaker in (18) expects any kind of attentional realignment (or reaction) from the surrounding hearers, and therefore he does not employ the Absconditive.

Concluding remarks
Based on the description given above, I claim that the basic face-to-face uses of the Absconditive can be understood and described as signalling that the speaker is urging the addressees to redirect their attention to some state of affairs that is currently outside their 'epistemic realm', for example because the addressee is looking in a different direction, or because the state of affairs is the exclusive knowledge of the speaker. It seems useful to think of the Absconditive as tracking the "epistemic and attentional states of interlocutors", which Evans et al. (2018a) consider to be the core targets of engagement categories. As evidence for my claims I used close readings of a few video-recorded exchanges, along with discussion of additional corpus examples and overheard utterances. Future work, and more documentation of language in use, will undoubtedly refine our understanding of these prefix series, but for now I hope to have shown that looking at the Absconditive through the lens provided by the notion of engagement provides a much better descriptive framework than the uninformative "present tense" label found in Drabbe's-in most respects brilliant-pioneering work. To emphasise that this should not be the last word on the Coastal Marind Absconditive, I end this essay, in lieu of conclusion, with a note on a different facet of this grammatical category. One frequent use of the Absconditive prefixes falls outside the engagement domain: its use in present tense relative clauses. There are two strategies that are used for forming dependent clauses that can be employed to narrow down reference to an entity or location. Either the verb is prefixed with the 'Dependent' prefix ahto form a relative clause with non-present time reference (i.e. past or future, as specified by additional tense inflection), or it is prefixed with the Absconditive to form a clause with present time reference. Clauses of both types typically occur adjoined to the periphery of the main clause, e.g. as a left-detached topic, and are often marked by a final demonstrative (which is also common with topics). A corpus example is in (19). I will not go into any details about the syntax and semantics of this structure, only note that the Absconditive here lacks the function of attention realignment that it has in independent clauses. A speaker who wishes to use a present tense clause to convey nominal or locative reference has no other choice but to attach the Absconditive to the verb of the dependent clause. Such uses are not a problem for the description of the Absconditive as having engagement semantics, because there is no reason to expect grammatical categories to be monofunctional. The present tense relative clause constitutes a delimited context in which the Absconditive lacks what seems to be its main function in independent clauses. It is of course not unusual for grammatical elements to have one use in main clauses and a quite different use in subordinate contexts-think of e.g. subjunctive verb forms or 'pleonastic' negation. The role of the Absconditive prefix in the formation of subordinate clauses is interesting from a diachronic perspective, since many linguists believe that constructions that start their career as subordinate or nominalised may acquire independent uses, often with modal-or stance-like meanings (e.g. Evans 2007, Schapper andSan Roque 2011). Little is known about the diachrony of Coastal Marind inflection, but it may be worth investigating more closely the relationship between the independent demonstratives and the uses of the Absconditive in subordinate and main clauses, once more descriptions are available of other languages in the fascinating Anim family. Acknowledgement: I am indebted to Alan Rumsey, Bethwyn Evans, two reviewers, the editors of this special issue, and the audience of the SLE 2017 workshop Emerging engagement, for useful comments. Most of all, I am indebted to the Coastal Marind community for letting me study their language.