3.1 Grammatical aspect: Temporal definitions and classifications

The two main (i.e. conceptually most general and typologically widespread) categories of grammatical aspect are perfectivity and imperfectivity. A whole set of related, yet not identical temporal definitions of these categories are to be found in the literature. The perfective/imperfective opposition is characterized by Comrie (Reference Comrie1976: 16) as follows: ‘perfectivity indicates the view of a situation as a single whole, without distinction of the various separate phases that make up that situation; while the imperfective pays essential attention to the internal structure of the situation’. Similar definitions, based on a situation’s (lack of) completeness, can be found in e.g. Dahl (Reference Dahl1985: 78) and Smith (Reference Smith1997: 3, 66), who point out that a perfective viewpoint involves a total view of a situation, while an imperfective viewpoint implies a partial view. This viewpoint opposition can be understood as an opposition between an ‘external’ (perfective) and an ‘internal’ (imperfective) perspective (Michaelis Reference Michaelis1998). Closely related to these are definitions in terms of boundedness (perfective) and unboundedness (imperfective), proposed by, for instance, Chung & Timberlake (Reference Chung, Timberlake and Shopen1985) and hinted at by Smith (Reference Smith1997: 301–302). These two types of theory (i.e. completeness and boundedness) are subsumed by Dickey (Reference Dickey2000) under the term ‘synoptic theories’. Formalizations of these synoptic theories have been suggested in, for instance, Klein (Reference Klein1994) and other neo-Reichenbachean approaches to aspect, where perfective aspect is analyzed as indicating the full inclusion of the event time (ET) in the reference or topic time (TT), while imperfective aspect involves the full inclusion of TT within ET.

An important body of work has been devoted to going beyond these purely temporal, synoptic theories, by concentrating on the discourse functions that aspect markers can take on. Central accounts in this respect are Hopper (Reference Hopper and Givón1979, Reference Hopper and Hopper1982) and analyses within the formal framework of Discourse Representation Theory, such as Partee (Reference Partee1984) and Kamp & Reyle (Reference Kamp and Reyle1993). The use of (tense and) aspect markers (like the perfect) to index participants (see e.g. Lakoff Reference Lakoff1970) also belongs in this group. Perfective markers are typically said to foreground situations and to move the narrative time forward by referring to events as part of a temporal sequence, while imperfective markers have more of a backgrounding function, in that they set the scene against which other events take place. Not unrelated to these discursive accounts is the analysis in Dickey (Reference Dickey2000) of eastern Slavic aspect in terms of temporal (in)definiteness. According to this analysis, perfective aspect in Russian, for example, indicates temporal definiteness, i.e. the unique temporal location of a situation, for instance as a consequence of it being part of a sequence. Conversely, imperfective aspect is said to signal temporal indefiniteness (see Dickey in this volume for more details). While these discursive and temporal-definiteness accounts undeniably constitute important advancements in a more inclusive conception of aspectual meaning, we should point out that (most of) these accounts still center around the relation between one situation and other situations in time, i.e. they presuppose a temporal basis for the semantics of aspect.

Beside perfective and imperfective aspect, other more specific subtypes of grammatical aspect can be distinguished. According to Comrie (Reference Comrie1976: 28), the category imperfective comprises two subtypes: habitual and progressive aspect. As we have already observed for perfective and imperfective aspect, definitions of progressive aspect comprise many related temporal concepts (sometimes specifically proposed for the meaning of the English progressive), including continuity at a particular reference point (Comrie Reference Comrie1976, Huddleston Reference Huddleston2002), ongoing activity (Dahl Reference Dahl1985), dynamicity (Rydén Reference Rydén, Nevalainen and Kahlas-Tarkka1997), incompleteness (Leech Reference Leech2004), time framing (Jespersen Reference Jespersen1931), and (limited) duration (Leech Reference Leech2004). There seems to be a general agreement nowadays, at least among non-formalist accounts, that the progressive is used to zoom in (i.e. create an imperfective perspective) on a dynamic event. By zooming in on an event, very similarly to what happens when one continually closes up on a visual object (like a cow), that segment of the situation that is focused upon becomes unbounded and homogeneous, just like with the conception of a state. According to Michaelis (Reference Michaelis2004), the progressive in fact selects state phases in the temporal configuration of events, i.e. it selects the state that lies in between an event’s boundaries, where states change quality (Altshuler & Michaelis in this volume provide more details regarding such stativization processes). Habitual aspect is expressed by dedicated grammatical constructions in some languages (e.g. English used to/would for past habits), but it can also be expressed by more general imperfective or progressive constructions. Habitual situations involve a series of repeated tokens of an event type that is considered to be ‘characteristic of an extended period of time, so extended in fact that the situation referred to is viewed not as an incidental property of the moment but, precisely, as a characteristic feature of a whole period’ (Comrie Reference Comrie1976: 27–28). Expressions of habituality, such as (2), are related to generic statements, such as (3), the only difference being that the latter involve non-specific subjects (Bybee, Perkins & Pagliuca Reference Bybee, Perkins and Pagliuca1994: 151–152):

It is sometimes argued that both imperfective and perfective aspect can be split up into more specific temporal subcategories, as is suggested by Bybee et al. (Reference Bybee, Perkins and Pagliuca1994: Chapter 3). However, while Comrie (Reference Comrie1976) proposes that imperfectivity, progressivity, and habituality (and related senses) may be conceptually ordered as a hierarchy – imperfectivity being the more general category – Bybee et al. (Reference Bybee, Perkins and Pagliuca1994) focus on the diachronic semantic links between categories associated with perfectivity and imperfectivity, respectively. Thus, the categories completive, perfect, resultative, perfective (i.e. uses that are not subject to the restrictions characterizing the three previous categories), and, finally, indefinite anterior and definite past-tense uses are analyzed as constituting various stages in the typical diachronic development of perfective markers, reflecting progressive degrees of grammaticalization with increasingly less specific usage conditions. Completive aspect markers can be considered as types of perfective markers that emphasize the final boundary of a situation. Defining the grammatical category of perfect as a subtype of perfective aspect may be more complicated, in view of its tense-like properties (see e.g. Ritz Reference Ritz and Binnick2012). Very generally, the perfect – which can appear in combination with past-, present-, and future-tense constructions – is said to indicate the continuing relevance (Comrie Reference Comrie1976: 52) of a prior situation at reference time. In some languages, such as German, French, and Dutch, the perfect has evolved into a definite past-tense construction (i.e. more general than indefinite anterior uses). But even for languages where a perfect construction still has a clear aspectual profile (e.g. focusing on an event’s state of completion), the strong link with temporal location (e.g. anteriority vis-à-vis a reference time) turns perfect constructions into potential rivals of past-tense markers in the domain of past-time reference. Yet the fact that the perfect can combine with tense marking is a good reason to regard it as an aspectual construction nonetheless. On top of this formal argument, a semantic motivation is given by Michaelis (Reference Michaelis1998, Reference Michaelis2004), who analyzes the perfect, just like the progressive, as a stativizing type-shifting construction, selecting states in the temporal configuration of events, such that a situation involving a dynamic event is given an overall stative profile. The difference between the progressive and the perfect is that the latter involves the selection of a posterior state rather than a medial one.

3.2 Lexical aspect: Definitions and classification

Since issues of lexical aspect are less central to most of the analyses presented in this special issue, we will touch upon this topic only briefly, even though, of course, the distinctions between states and events and telic and atelic situations constitute pivotal building blocks for any general aspectual account. One of the most commonly employed actional classifications is that of Vendler (Reference Vendler1957/Reference Vendler1967), who divides English verbal predicates into four classes – states, activities, accomplishments, and achievements – on the basis of the features [$\pm$ stativity], [$\pm$ telicity], and [$\pm$ duration]. Other classifications, related more or less closely to Vendler’s, include Verkuyl (Reference Verkuyl1993), Bache (Reference Bache1995), Dik (Reference Dik1997), and Smith (Reference Smith1997) (see also Croft Reference Croft2012: 45–52 for an overview). It is generally agreed upon that the most basic actional distinction is that between states and dynamic events (although there is less agreement on how to operationalize this distinction), echoing the general grammatical categories of imperfective vs. perfective aspect. Events can be further subdivided on the basis of additional distinctions such as inceptive/non-inceptive or homogeneous/heterogeneous. Stative verbs, a relatively less heterogeneous collection, are also sometimes classified into different subtypes along temporal parameters, for instance on the basis of their temporal extension (Croft Reference Croft2012: 58). One of the most fine-grained actional classifications to date can be found in Croft (Reference Croft2012), who incidentally also takes into account qualitative differences between situation types, rather than only temporal parameters.

In contrast with the study of grammatical aspect marking, many if not most available classifications of lexical aspect historically show ‘but little typological awareness’ (Tatevosov Reference Tatevosov2002: 322). That is, classes of actionality are frequently defined on the basis of a set of universal semantic features, as is done by Vendler and his followers, and it is tacitly assumed that this classification (usually established on the basis of English data) is not subject to cross-linguistic variation: all languages will fundamentally have recourse to (a subset of) the predicted possible classes. However, substantial cross-linguistic variation among classes of lexical aspect is not uncommon (Tatevosov Reference Tatevosov2002; Croft Reference Croft2012: 50; Bar-El Reference Bar-El, Bochnak and Matthewson2015), and it remains a particularly thorny issue to define actional classes in a typologically relevant fashion, i.e. such that allowance is made for peculiar, language-specific (possibly unique) ways of organizing the temporal profiles of situation types (see Crane, Nichols & Persohn Reference Crane, Nichols and Persohn2019).

An equally challenging, and ultimately highly related issue is whether it is possible to classify a verb as by definition belonging to only one given actional class, or whether instead the lexical aspect of a verb (phrase) always needs to be defined in context (implying that it is inherently variable (see the discussion on coercion in Section 4.2 for more details).

4.1 Non-temporal meanings of aspectual constructions

Non-temporal uses of aspectual constructions have been especially prominent in analyses of general imperfective and progressive constructions. Typological and language-specific data from, among others, Fleischman (Reference Fleischman, Bybee and Fleischman1995), Iatridou (Reference Iatridou2000), Brisard (Reference Brisard2010), and Patard (Reference Patard, Patard and Brisard2011, Reference Patard2014) show that (specifically past) imperfective constructions are often recruited to evoke hypothetical or irreal situations rather than establishing actual past-time reference. This is clearly illustrated in pretense-play contexts, such as (4), in which children use a past imperfective (the imparfait in French) to describe what is happening in the game they are playing at the time of speaking:

Other uses in which the imparfait or analogous tenses in other languages are used to refer to situations that are not (presented as) real or known at the time of speaking are found in, among others, protases of conditional clauses, wishes, and politeness contexts (see Brisard Reference Brisard2010 for an overview). Since these uses pertain to the (construed) reality status of the reported situation, they can be called modal (epistemic) uses of the constructions under consideration. A central point of discussion is whether these modal uses constitute pragmatic extensions of the temporal semantics of these imperfective constructions in interaction with (non-temporal) features of the context, as is argued by, for instance, James (Reference James1982), Fleischman (Reference Fleischman1989), and Patard (Reference Patard2014). In Cognitive Grammar, the question pertains to whether these uses reflect grounding properties of the aspectual constructions at issue: grounding predications like tense basically involve epistemic concerns, which makes modal uses of such predications anything but ‘pragmatic’. The issue, as noted with respect to the hybrid tense/aspect status of certain constructions, is then whether aspect markers (themselves grounded by tense) may also qualify as grounding predications in some respects, or in certain contexts (see Brisard forthcoming).

A similar controversy surrounds non-aspectual uses of the (specifically present) progressive, which have been described for French, Dutch, and English in, respectively, De Wit, Patard & Brisard (Reference De Wit, Patard and Brisard2013), Anthonissen, De Wit & Mortelmans (Reference Anthonissen, De Wit and Mortelmans2019), and De Wit & Brisard (Reference De Wit and Brisard2014). Such non-aspectual uses are for instance attested in certain performative utterances in English (De Wit, Brisard & Meeuwis Reference De Wit, Brisard and Meeuwis2018):

The fact that the (more canonical) simple present and the present progressive can be used interchangeably in this context illustrates that the progressive is not used for aspectual reasons but rather has some expressive/qualifying function here. While these non-temporal uses are typically considered subsidiary (if they are considered at all in studies of performativity), they are taken to reflect a fundamental non-aspectual meaning in, among others, analyses within the French enunciativist tradition (e.g. Adamczewski Reference Adamczewski1978), as well as by Williams (Reference Williams2002) and De Wit & Brisard (Reference De Wit and Brisard2014). De Wit et al. (Reference De Wit, Brisard and Meeuwis2018) also show more generally that the choice of aspectual marking for performative statements in a given language crucially depends on other than temporal considerations that could broadly be called epistemic. Overall, many aspect-related phenomena that have been observed in the literature (including in formal-semantic accounts), such as the imperfective paradox (Dowty Reference Dowty1979, Portner Reference Portner1998) or the sub-interval property of stative predicates, pertain to issues of (mis)alignment between the observation/conception of a situation and its simultaneous (complete) identification and report, which is clearly a matter of qualifying a state of affairs and thus of construing a modal assessment.

Moving beyond the realm of modality, it has been observed that certain aspectual constructions, most notoriously the perfect, can develop evidential and mirative meanings – notions that are of course not unrelated to modality (see e.g. DeLancey Reference DeLancey2001, De Wit Reference De Wit2017a). That is, perfect constructions can be recruited to indicate the source of information or a feeling of surprise on the part of the speaker. A particularly interesting construction in this respect is the Turkish perfect suffix -miş, which – in contrast to its unmarked counterpart in the past-time paradigm -di – expresses a sense of indirectness (Johanson Reference Johanson and Aikhenvald2018). More specifically, as the following illustration by Slobin & Aksu (Reference Slobin, Aksu and Hopper1982: 187) shows, -miş can convey inference, hearsay, and/or surprise:

In order to unify the analysis of various uses of the -miş perfect, Slobin & Aksu (Reference Slobin, Aksu and Hopper1982) argue that we need to look beyond its temporal import and instead, again, come up with a knowledge-based analysis: as they put it, ‘the essence of all uses of -miş is to encode situations for which the speaker is not somehow prepared – situations on the fringe of consciousness, learned of indirectly, or not immediately assimilable to the mental sets of the moment’ (Slobin & Aksu Reference Slobin, Aksu and Hopper1982: 195). More generally, there appears to be a straightforward semantic connection between evidentiality and the perfect, in the sense that both convey a sense of indirectness in terms of acquiring knowledge of a situation (DeLancey Reference DeLancey2001: 378). The perfect involves past situations of which the conceptualizer witnesses some current consequences (e.g. in the form of a result). A sense of indirectness is equally central to evidential statements, which ‘are indirect in the sense that the narrated event is not stated directly, but in an indirect way, by reference to its reception by a conscious subject, a recipient’ (Johanson Reference Johanson and Aikhenvald2018: 511). Hence, it makes sense that languages should recruit the same construction to convey these two meanings. By extension, mirative interpretations follow naturally from the evidential meanings: as DeLancey (Reference DeLancey2001: 378) observes, ‘a fact which one knows only when ones sees secondary evidence for it is necessarily unexpected to some degree’. This evidential/mirative/aspectual polyfunctionality of perfect constructions makes one wonder to what extent we can still speak of constructions that are purely or primarily aspectual.

The above-noted use of (past) imperfective constructions for the expression of politeness or, more generally, mitigation (e.g. of illocutionary force) as well as the attested mirative uses of the perfect point to the exploitation of various aspectual constructions for interactional/interpersonal purposes. Other interpersonal meanings that have been associated with aspectual constructions are authority, necessity, and obligation. In her analysis of the Russian perfective/imperfective contrast with verbs of communication, Israeli (Reference Israeli2001) notes that one motivation for selecting a perfective rather than an imperfective marker is that the former can convey a sense of authority that is absent in the case of the latter. Interpersonal functions can also motivate the use of verbless statives in Modern Standard Arabic (Mansouri Reference Mansouri2016: Chapter 4). In (7), for instance, the stative predication is interpreted as having a sense of obligation or necessity:

Still other non-temporal, pragmatically motivated extensions of the use of aspect markers are connected with information structure. This has been observed by Güldemann (Reference Güldemann2003) for present progressive constructions in Bantu, which have evolved out of focus constructions and can still express this focal meaning (see De Wit et al. in this volume). Aspectual constructions can furthermore be used as clause-chaining devices – indicating different types of inter-clausal dependency. This is elaborately discussed by Robert (Reference Robert2010) for various aspectual constructions in the Niger-Congo language Wolof. Similarly, François (Reference François2010) notes that the so-called Background perfect can be used to mark subordination in its own right (i.e. without needing other subordination markers such as conjunctions) in the Oceanic languages Hiw and Lo-Toga (see Bril Reference Bril2010 for additional illustrations of aspectual means for indicating clause hierarchy).

In sum, aspectual constructions can function as markers of modality, evidentiality, mirativity, authority, necessity, obligation, focus, and clausal structure. Integrating these non-temporal functions within a general semantics of aspect requires us to move away from an exclusively temporal analysis and identify the basic features in the conceptual schemes evoked that are responsible for the many extended uses that are attested. For that, it seems plausible to turn to the relation between types of temporal profiles (e.g. stative vs. dynamic) and implications in terms of how knowledge of them can be construed (as is done in the first three contributions to this special issue).

4.2 The use of aspectual encoding in non-canonical contexts

The crucial role of context cannot be overstated when analyzing aspectual semantics. One notion that immediately comes to mind in this respect is that of coercion. Coercion effects come about when there is a semantic mismatch between a higher-order construction and a (lexical) element embedded in that construction (see e.g. de Swart Reference de Swart1998; Michaelis Reference Michaelis2004, Reference Michaelis2011). Aspectual coercion can occur when a canonically stative verb combines with the progressive in English, as in I’m loving it. In cases such as this, the verb is exceptionally construed as being dynamic, i.e. in this context, it gets a different, non-canonical meaning. Instances of coercion indicate the need for a context-based approach to actional classification, in which we recognize the potential of verbs to shift lexical class depending on the context in which they appear and abandon the idea of a ‘natural’ aspectual class (at least for certain verb types). But context also plays a role in the analysis of grammatical aspectual constructions, which can, as we have seen in Section 4.2, be remarkably polysemous. A notable case in this respect is the Japanese -te iru construction, which may, depending on the context (e.g. the lexical aspect of the verb), take on stative, progressive, and perfect meanings (see Ebert Reference Ebert, Bertinetto, Bianchi, Dahl and Squartini1995 for other examples of such contextually triggered progressive–perfect ambiguities). Contextual effects can also consist of one aspectual construction affecting another one within the same verb phrase. This phenomenon, which is known as aspectual stacking (see e.g. Altshuler Reference Altshuler2016: 155–158), often seems to seems to go hand in hand with a meaning shift of (at least) one of the two constructions (De Wit Reference De Wit2018: 228–229). For example, in Russian, the perfective prefix pere- can be added to the imperfective verb stem pisat’ ‘write’ to form the perfective perepisat’ ‘rewrite’. Yet by adding the suffix -yvaj to perepisat’, such that we get perepisyvat’, the derived perfective value of the construction is undone, so to speak, and the meaning becomes imperfective again (‘being in the course of rewriting’). Still within the context of the verb phrase, aspectual meaning is furthermore determined by tense. For instance, the English progressive might have a prototypical meaning of ongoingness in present- or past-time contexts, but when it is used in combination with will (as in Your train will be stopping in Manchester), it typically gets a non-aspectual, matter-of-course meaning (Celle & Smith Reference Celle and Smith2010). More generally, interactions of certain aspectual constructions with present- (but less with other) tense marking often leads to the kinds of phenomena noted above with respect to epistemic status (the use of progressives to mark incongruity) or tense-like behavior (the use of perfects as preterites).

Aspectual constructions may also exhibit special behavior in certain non-canonical contexts, as with non-finite verb forms (infinitives and participles), imperatives, conditionals and other types of subclauses, performatives, and specific types of text, such as narratives or instructions. For instance, in their analysis of wine reviews, Hommerberg & Paradis (Reference Hommerberg, Paradis, Glynn and Sjölin2014: 218) demonstrate that aspect choice contributes to ‘the construal of the tasting event as a joint writer–reader enterprise’, a matter of stance. Similarly, both De Wit et al. (Reference De Wit, Brisard and Meeuwis2018) and Fortuin (Reference Fortuin2019) argue that, in many languages, performative contexts involve distinct forms of aspectual behavior, reflecting the alternative, non-temporal function aspectual constructions fulfill in such contexts (see example (5) above).

Once more, these observations indicate that we need to reflect on whether or not we can come up with a uniform semantic analysis of a given aspectual construction, incorporating this wide range of contextually induced variation.

4.3 Non-canonical aspect marking: Beyond the verb

As indicated in Section 2 above, a central tenet of unidimensional approaches to aspect is that aspectual meaning can be expressed at various levels of the utterance (see the seven aspectual tiers distinguished by Sasse Reference Sasse2002: 263). Encoding strategies that rely on linguistic tools beyond the (finite) verb are typically considered more marginal, and they are often neglected in overviews of aspectual meaning and marking. Yet such alternative types of grammatical marking force us to explore the limits of aspectual semantics (as a category associated with (finite) verbs) and to investigate possible links between certain types of aspectual meaning and certain expression forms.

For instance, light verb constructions are said to fulfill an aspectual function, in that they impose a telic construal (Wierzbicka Reference Wierzbicka1982, Bonial & Pollard in this volume). This is illustrated in the following opposition:

While (8a) involves an unbounded viewpoint, the use of the light verb have invokes an inherent endpoint to the denoted event. Similarly, reduplication constitutes yet another tool to express aspectual meaning. In Mandarin Chinese, for instance, it serves ‘a delimitative function, i.e. it expresses that a situation continued for some (relatively brief) period of time’ (Dickey Reference Dickey and Riemer2016: 345). This is illustrated in example (9):

Beyond the verb proper, it has been noted that tense, aspect, and modality can also be marked on nominals and modifiers within an NP (Nordlinger & Sadler Reference Nordlinger and Sadler2004; Bertinetto, published online 3 March 2020). Although Nordlinger & Sadler (Reference Nordlinger and Sadler2004) primarily focus on tense and mood distinctions, they also cite examples showing that aspect marking can be done within the confines of the noun phrase, possibly in combination with verbal aspect marking, as is the case in the following example from Sirionó (Tupí-Guaraní, Bolivia):

Recent work by Bertinetto (published online 3 March 2020: Section 3) provides more robust evidence for the existence of nominal aspect. Note, finally, that adverbials (e.g. in the form of prepositional phrases) can carry aspectual meaning as well (Sasse’s fourth tier). This is amply demonstrated for the adverb already by Michaelis (Reference Michaelis1996) and temporal adverbs such as now by Altshuler (forthcoming), as well as in the joint contribution of these two authors to the current volume.