From aspect to information structure: Non-aspectual functions of change of state markers in Austronesian and beyond

1 Introduction

In this paper, I discuss a phenomenon that has received little attention in the existing literature: the presence of polyfunctional markers, which combine an aspectual meaning of change of state with information and discourse structuring functions, in several languages from different branches of the Austronesian family. An example of such a marker comes from Lakurumau, an under-described Western Oceanic language spoken on New Ireland, Papua New Guinea, where the enclitic particle asang is found. As an aspect marker, asang encodes change of state, while its non-aspectual uses include the expression of contrast and emphasis. Examples (1a) and (1b) illustrate the polyfunctionality of asang. In (1a), asang modifies the predicate masai ‘(be) ripe’ with an aspectual function: it asserts that the predicated state of affairs holds at reference time and that it has come about as a result of a transition from a previous state of affairs of opposite polarity. I label this aspectual function as change of state (see the discussion in Section 2.1). In (1b), asang attaches to the directional axo(m) ‘far away; above’ and has no aspectual meaning. Instead, it strengthens and emphasizes the contrastive interpretation of the second clause (e.g. ‘The garden is not nearby. It is up there on the mountainside’).

Markers with similar aspectual semantics, morphosyntactic characteristics (enclitics with few or no restrictions in the choice of their lexical host), and non-aspectual functions to the Lakurumau asang are well documented in descriptions of other Austronesian languages. One example is the enclitic =te ‘completive; contrastive focus’ in Bantik (Sangiric; Utsumi 2020) and the enclitic na’a ‘perfect; marker of contrast’ in Longgu (Southeast Solomonic; Hill 1992); these two languages, despite their common family affiliation, are quite distant from each other genetically, geographically and typologically. Although descriptions of polyfunctional aspect markers in individual Austronesian languages do exist, no comparative study has been conducted.^[1] As a result, several aspects of this phenomenon remain unclear, such as the distribution of these markers within the Austronesian family and the relationship between their change of state semantics and their non-aspectual uses.

In this paper, I address these unanswered questions by analyzing a sample of 52 languages belonging to all major branches of Austronesian (Table 1).^[2] The paper is organized as follows: in Section 2, I discuss the semantics of change of state, which I define as a grammatical aspectual operation based on two notions: current relevance and transition to a new situation. In Section 3, I present the non-aspectual functions of the analyzed markers in the sample languages. In Section 4, I provide examples of similar polyfunctional markers in languages outside the Austronesian family. In Section 5, I argue that the development of information and discourse structuring functions in change of state markers derives from the application of notions of current relevance and transition to a new situation not to eventualities, but to elements of propositions. I also suggest that this process may be universally valid and could explain the similar development of change of state markers into information and discourse structuring devices attested in language families other than Austronesian. In Section 6, I draw conclusions and present further research directions and needs.

2 Defining the field of inquiry: change of state markers

All the markers I analyze in this paper have the same aspectual function: they assert the establishment of a new state of affairs, which holds at reference time and is the result of a recent transition from a previous state of affairs of opposite polarity. I have labeled this function change of state (COS) and, henceforth, refer collectively to the morphemes analyzed in this paper as COS markers (see Table 2).

Admittedly, the term “change of state” is ambiguous, as it has been used to denote different things in the literature. For example, both the lexical semantics of verbal roots such as the English words redden and ripen (Koontz-Gaborden 2007) and the grammatical function of markers such as the Yoruba tí and Mandarin Chinese le (Bisang and Sonaiya 1997; Li et al. 1982) have been referred to as change of state. Here, I understand change of state as a grammatical aspectual operation defined by two semantic notions: current relevance and (recent) transition to a new situation. Change of state markers assert a state of affairs that holds at reference time and is pictured as the result of a recent transition from a state of opposite polarity (see examples (2a) to (2d)). As example (2d) shows, COS markers give rise to an interpretation incorporating “no longer” when combined with negation.

The definition of change of state I offer in this paper only considers the behavior of COS markers in combination with stative predicates, as shown in examples (2a) to (2d). This does not mean COS markers do not also interact with dynamic predicates. In example (3) from Lakurumau, the particle asang modifies the dynamic predicate waan ‘go’, resulting in an immediate future interpretation.

However, the study of the interaction between COS markers and dynamic predicates would require the analysis of several detailed factors, such as the actionality class of specific verbs and their temporal specifications, and is therefore far too extensive to be undertaken for this paper. The majority of the grammatical descriptions consulted for this study do not reach this level of detail when discussing the semantics of aspectual markers, making a true comparison impossible. For this reason, in this paper I restrict myself to examining the interaction between change of state and stative predicates.

The markers I analyze as expressing change of state receive different glosses in the source materials, such as perfect, perfective, already, now, completive, change of state, inchoative, and iamitive (see Table 2).^[3] This implies that these markers may be instances of different grammatical categories. However, this is not a problem for the present analysis: I do not claim that all the markers discussed here are instances of the same grammatical category (e.g. the perfect). I only claim that they all have at least one aspectual function in common: they express change of state when interacting with stative predicates. This does not imply the creation of a putative change of state grammatical category, to which all the sample markers belong and that is distinct from the perfect, the completive, or “already.” Indeed, the hypothesis that all the sample markers belong to the same grammatical category is an intriguing one. However, it is a hypothesis I cannot prove now: to do so, I would need to analyze their interaction not only with stative predicates but with dynamic predicates of different actionality classes, too (see Krajinović 2019 for an analysis of Oceanic perfect markers). Therefore, I take the definitions in the descriptive materials I have consulted at face value and allow for the possibility that the sample markers may be instances of different grammatical categories (i.e., the completive, the perfect, and “already”): what is relevant for my analysis is only the marker’s ability to express change of state when modifying stative predicates.

3 Non-aspectual functions of COS markers in the sample languages

Most of the COS markers in my sample only have an aspectual function. Polyfunctional markers are found in only twelve languages (Table 3).

Despite their relative rarity, polyfunctional COS markers are distributed throughout the entire family, crosscutting the major typological and geographical divide between Western and Eastern Austronesia. They tend to occur in clusters of closely related or geographically contiguous languages, with three main clusters found: languages of the Philippines and North-East Borneo (Malaysia); languages of Sulawesi (Indonesia); and languages of Guadalcanal, Malaita, and Makira – three islands in Solomon Islands. Formal cognacy is not present at the family level but at the cluster level: COS markers have the phonological shape =n[V] in the Philippines/Borneo languages; =mo in the Sulawesi languages, except for Bantik; and =na[‘V] in the Guadalcanal/ Malaita/Makira languages.

Morphologically, all polyfunctional COS markers in the sample are enclitics and can attach to any type of lexical host. Some of them, such as Ilocano =(e)n and Tajio =mo, have restrictions regarding their position in the clause: they always occur in second position, directly following the leftmost element. Crucially, the morphological flexibility of COS markers is not limited to elements in predicate position. This specification is important because the Austronesian languages have a notoriously labile distinction between nouns and verbs. Virtually any nominal root can be used as a predicate and can take person indexing as well as tense, aspect, and mood marking: see examples (8a) and (8b) from Lakurumau, where the lexical item skul ‘school’ is used as a noun in (8a) following a preposition and as a verb in (8b), with person indexing.

One might think that COS makers only attach to non-verbal roots when these are used predicatively and are therefore functionally verbs. However, this is not the case, as example (9) from Kimaragang shows. In (9), the noun phrase I=Jim, Jim is clearly non-predicative – it is the subject of the verbal form minanakaw ‘steal’. Additionally, the marker no performs a non-aspectual function, which Kroeger (2021a) labels as ‘focus’.^[5]

As well as having non-aspectual functions in the modification of non-predicative elements, COS markers can be used with non-aspectual functions on predicates. For instance, in example (10) from Tondano, the COS enclitic =mow does not encode change of state, rather it “encodes a sense of certainty” (Brickell 2014: 426).

In the remainder of this paper, however, I will only focus on the interaction of COS markers with non-verbal elements, as this is the environment where their non-aspectual function is most evident.

4 Beyond Austronesian

In the previous section, I limited my investigation to the Austronesian languages. However, markers that combine an aspectual meaning of change of state with information- and discourse-structuring functions are also attested in other language families. This section briefly presents some examples from Quechua (Quechuan), Latin (Indo-European), Tulil (Baining), and Kuot (isolate).

The Quechuan languages use the aspect suffix -ña ‘discontinuative; inceptive’ to express the notion of transition to a new situation (Cerrón-Palomino 1994: 140; Shimelman 2017: 257; Weber 1989: 376ff). According to Cerrón-Palomino (1996), -ña can also attach to non-verbal expressions with a contrastive value (30a). This use has been adopted in Andean Spanish, where the particle ya ‘already’ can be used with a contrastive meaning (30b).

The Latin particle iam ‘already’ has both temporal and scalar meanings. Temporally, iam indicates a transition from a negative to a positive stage (Huitink 2005; Kroon and Risselada 2002), as example (31a) shows. Kroon and Risselada (2002) successfully demonstrate that iam has an additional pragmatic function in expressing counter-expectational focality. I believe that in some cases iam may actually have a contrastive function rather than counter-expectational. This is shown in example (31b), taken from the prologue of a comedy by the ancient Roman playwright Terence. As Koon and Risselada explain, in the prologue the playwright “defends his use of stock characters, since he has been accused by Roman officials of having taken his characters from the plays of other Roman playwrights. Iam puts a certain emphasis on vos (‘you’), which refers to the audience, as opposed to the officials” (Kroon and Risselada 2002: 72). Here, iam elicits a contrastive interpretation, opposing one referent (the audience) to another (the officials).

In Tulil, a Baining language spoken in New Britain, Papua New Guinea, the proclitic aspect marker bə= ‘iamitive’ expresses change of state with stative predicates (32a). As Meng (2018: 165) states, bə= can also attach to phrasal-level elements, with “an emphatic, foregrounding function”, see (32b).

In Kuot, an isolate Papuan language of New Ireland, Papua New Guinea, the enclitic aspect marker =(a)r(a) has complex semantics that include the expression of change of state with stative predicates (Chung and Chung 1996: 51; Chung 1999: 6; Lindström 2002: 144); see example (33).

In discourse, =(a)r(a) indicates current relevant state (Chung 1999), marking situations and events that are worthy of attention because they represent pivotal changes in the narrative or contradict the hearer’s presuppositions. Chung (1999: 5) states that =(a)r(a) may also function “as a focus marker: It marks the thing that the speaker wants the hearer to remember”, but he does not provide clear examples of this function in the text.

Table 5 summarizes the non-aspectual functions of COS markers in Quechua, Latin, Tulil, and Kuot.^[12]

Due to the limited number of languages outside the Austronesian family examined in this study, it is not possible to draw any reliable conclusions about the cross-linguistic distribution and frequency of polyfunctional COS. However, even such limited evidence clearly demonstrates that this phenomenon is present outside of the Austronesian family and its sphere of influence. Tulil and Kuot may have replicated the Austronesian model in expanding the functions of their COS markers. Both languages have had intensive contact with Austronesian languages; in particular, Kuot has Lakurumau (see Section 3.5) as one of its contact languages. However, in the case of Quechua and Latin, there is no possibility of a contact-induced explanation.

Therefore, the development of COS markers into information structuring devices is most likely due to universally valid semantic-pragmatic motivations and processes. In the following section, I explore and discuss the possible motivations behind the combination of change of state and information structure.

5 From aspect to information structure

In their discussions of the Bantu system of verbal focus, Hyman and Watters (1984) and Güldemann (2003) expose the close link that exists between some aspectual categories – the perfect, the progressive and the persistive – and information structure. These categories share a component of current relevance, that is, they encode situations that hold at the reference time. Güldemann (2013: 353–354) observed that perfect, progressive and persistive events “demonstrate a pragmatic relevance for the immediate speech situation and thus possess an inherent feature of focality.”

Güldemann’s observation also applies to change of state, as I understand it in this paper: change of state also has current relevance as a fundamental component of its semantic-pragmatic structure. COS markers encode situations relevant to the current state of discourse, both semantically, because of their holding at reference time, and pragmatically, due to the consequences brought for the subsequent discourse. Moreover, COS markers encode inherently contrastive situations: they assert the establishment of a state of affairs that is understood as new in opposition to an old state of affairs. I have labeled this semantic property “transition to a new situation”. I suggest that the emergence of information and discourse structural interpretations of COS markers derives from the application of the notions of current relevance, and transition to a new situation to elements of propositions, rather than eventualities. Here I follow Matić and Wedgewood’s (2012) examination of how information structural interpretations may arise from markers of verbal categories such as aspect, mood, and evidentiality. Matić and Wedgewood discuss the sources of information structural effects in several languages, including Quechua and Somali. In Quechua, the evidential suffix -mi/-n ‘direct evidentiality’ can also express narrow focus, and in Somali, the marker baa has both mood-marking and focus-marking uses. Matić and Wedgewood (2012: 141) observe that “focus in Quechua is plausibly a reading – perhaps partially inferential and partially conventionalized – of the direct evidential, in much the same way that Somali focus may be an effect of applying the realis mood to propositions and elements of propositions.” This same reasoning applies to the Austronesian COS makers discussed in this paper. In their information-structuring function, COS markers apply the notion of current relevance to the elements they modify, thereby drawing attention to them. The reason a speaker decides to mark a particular element as relevant varies depending on the context: information may be made prominent because it is new, unexpected, or counter-expectational. Additionally, as seen in Sections 3 and 4, COS markers are often used to indicate contrast. In this case, both the notion of current relevance and transition to a new situation are applied, that is, rather than contrasting two states of affairs, COS markers are used to contrast two pieces of information.

The process of deriving such non-aspectual functions from change of state through the application of current relevance and transition to a new situation to elements of propositions does not explain only the Austronesian case, but the emergence of polyfunctional COS markers in Quechua, Latin, Tulil and Kuot, too.

6 Conclusions and outlook

In the previous sections, I have shown that polyfunctional markers, which combine an aspectual meaning of change of state with information and discourse structuring functions, are present in various Austronesian languages from both the Western and Eastern branches of the family as well as in some languages outside the Austronesian family. Further, I have proposed that this process involves applying the notions of current relevance and transition to a new situation to elements of propositions rather than eventualities. However, my analysis is tentative and further research on this topic is needed. Firstly, a larger and, crucially, usage-based study of Austronesian change of state markers is necessary, which will consider data from text and speech corpora, and not only from grammars. Investigating natural discourse can clarify the range of interpretive effects generated by COS markers and their interplay with other discourse structuring operations, such as voice choices and left-dislocation.

Secondly, it is worth exploring whether the development of Austronesian COS markers into information and discourse markers was facilitated by their role as foregrounding devices. Here, the term “foreground” refers not only to the actual event line, in contrast to a non-eventful background – that is, non-narrative parts of a text such as parentheticals and descriptions – but also as the main or primary event line (see Becker and Egetenmeyer 2018; Shirtz and Payne 2015). That is, “foreground” represents the events that the speaker subjectively perceives and indicates as the most relevant in the narrative. In several languages in my sample, COS markers are used to indicate events considered pivotal in the narrative’s development. This function is found in Muna (van den Berg 1989: 269), Kimaragang (Kroeger 2005: 414), Sumbawan (Shiohara 2013: 183), Tokelauan (Hooper 1998), Nalögo (Alfarano 2021: 350), Lakurumau, and possibly others.^[13] The development of COS markers into information structuring devices may have first occurred in their role as foregrounding devices. In this function, they mark prominent, pivotal events: in some languages, COS markers may have loosened their categorial restrictions and gained the ability to attach to non-verbal elements as well, marking them as prominent and attention-worthy phrasal elements. The form of COS markers as clitics or particles may have facilitated such development into agnostic markers, with no restrictions on the type of host they attach to. Arguably, it would be more difficult for fusional morphology, such as the English ablaut in the past tense, or complex constructions, such as the English perfect (have + V-en/ed), to be re-analyzed as nominal or phrasal markers, even if they are semantically and pragmatically adequate.^[14]

Finally, further study is needed to explore the relationship between non-aspectual functions and aspects other than change of state. As mentioned in Section 3.1, polyfunctional persistive markers (i.e. ‘still’) are found in some languages in the Philippines and Borneo, with different focus-marking functions. As mentioned previously, the persistive aspect is similar to change of state in that it also has a component of current relevance and is therefore inherently focal (Güldemann 2003). The development of persistive markers into information structuring devices could also be explained through the application of current relevance to propositional elements rather than eventualities. It is my hope that this paper may inspire further research on these topics.

Special abbreviations

av

actor voice

cos

change of state

dist

distal

dyn

dynamic verb

iam

iamitive

lnk

linker

nvol

non-volitive

pers.art

personal article

piv

pivot

r

realis

test

test

th.v

theme voice

uv

undergoer voice