The typology of Slavic aspect: a review of the East-West Theory of Slavic aspect

Until recently, differences between the verbal aspectual systems of the individual Slavic languages have not received much attention and data was often interpreted to hold for all Slavic languages. Only in the past decades has this situation changed and to date a number of studies comparing two or more Slavic languages have seen the light of day. Independently from each other, Barentsen and Dickey have devised theories which account for the observed differences between respective Slavic languages. Their approaches are so similar that we think it is justified to speak of a single, comprehensive theory which attributes the differences in the functioning of the systems to differences in the meaning of the perfective and imperfective aspect for the individual languages. This leads to a typology in which there is an Eastern and a Western type of Slavic verbal aspectual system, hence the name ‘East-West Theory’. In this paper, we provide a critical analysis of this theory, focusing on three context types: habitual contexts, narrative contexts and retrospective contexts. Our analysis shows that the theory adequately and convincingly explains most of the data. However, we will also demonstrate that there are still areas in which the theory needs to be developed further, and we provide some suggestions as to how this can be approached.


Introduction
With respect to verbal aspect, Slavic languages occupy a special position because of their morphologically encoded opposition between a perfective and an imperfective aspect, which holds for the entire verbal system and applies to the complete verbal paradigm. In fact, the term 'aspect' itself is borrowed from the Russian vid (cf. Klein 1994, p. 72), and many general linguistic descriptions of the category of aspect across languages devote special attention to aspect in Slavic (see for example Comrie 1976;Dahl 1985;Smith 1997;Croft 2012;Binnick 2012;cf. especially Gvozdanović 2012). Even though it is generally acknowledged that one can speak of a common aspectual system for all Slavic languages (see for example Galton 1976), various studies that have appeared in the last decades have pointed out that there are considerable differences in the use of aspect between the Slavic languages (see for example Galton 1976;Ivić 1983;Mønnesland 1984;Stunová 1993;Dickey 2000;Barentsen 2008;Alvestad 2013). Several linguists have discussed such differences, but surprisingly few have actually tried to explain them in a cross-Slavic fashion. In this paper, we will provide a critical analysis and review of the most comprehensive theory of Slavic aspect to date, which takes these differences into account and tries to explain them. This is the East-West Theory of Slavic aspect (short EWT) 1 as developed by Adriaan Barentsen and Stephen Dickey. 2 It should be noted that there is no such thing as a single fully explicit EWT or paradigm, as for example laid down in a single book or article. In fact, there are two authors who (largely independently from each other) have developed very similar theories, sharing their central hypotheses. From this perspective it is justifiable to speak of a single theory (cf. Dickey and Kresin 2009, p. 125).
The aim of this paper is two-fold. First, by providing a critical analysis of EWT we aim to contribute to its further development and to provide a deeper understanding of Slavic aspect. Second, in doing so we also hope to bring EWT to the attention of a larger group of linguists than is currently the case.
It should be noted that both Barentsen and Dickey have a usage based (functionalist) approach to aspect, and do not work within a formal semantic model. In the same vein our analysis must also be seen as a contribution to the usage based study of Slavic aspect, and its aim is not to present a formalization of aspect.
In reviewing EWT, we will focus on three context types or usage types in which the use of aspect is known to differ among Slavic languages: • habitual contexts, • narrative contexts, • retrospective contexts ('perfect' use of the past tense).
For each context we will discuss data from various Slavic languages, explain how they are analyzed and accounted for within EWT, and provide an evaluation of this analysis. Whenever relevant to the discussion, we will also refer to contexts other than these three. We will conclude with an overall evaluation and a general discussion of how EWT may be further developed.

General outline of EWT of Slavic aspect
In this section we will discuss the central ideas on aspect of the two main proponents of EWT-Barentsen and Dickey. This section serves as a background for the following sections in which the data are discussed in detail.

Barentsen, pf aspect in Russian
Adriaan Barentsen has set out and developed his theory of aspect in various publications, amongst others: Barentsen (1985Barentsen ( , 1995Barentsen ( , 1998Barentsen ( , 2008. Even though most of his publications focus on Russian, he has considered other Slavic languages as well (see, e.g., Barentsen 2008). He also supervised Stunová's (1993) PhD-thesis on the comparison between Russian and Czech aspect. The central idea of his theory is that the pf and the ipf aspect in Russian each have a specific invariant meaning, which explains their use. According to Barentsen, the meaning of the pf is made up of three 'layers', forming a hierarchical structure (Barentsen 1995(Barentsen , p. 4, 1998. This means that the pf is used if: (a) the event expressed by the predicate is terminative (predel'nyj; cf. 'telic'), (b) the event is seen as a totality (celostnost'), meaning that a terminus is reached such that there is a change of situation (smena situacij), and (c) the event expressed by the pf verb is sequentially connected to a following and / or preceding situation (sekventnaja svjaz'). 3 Before we discuss this in more detail, it should be noted that Barentsen uses the term 'terminative' instead of 'telic'. This term is a broader one than what is usually defined by the term telic, because it also accounts for the so-called delimitative and perdurative perfectives in Russian, which contain the prefixes po-and pro-respectively. An example is the sentence On pospal (he po-slept-'He slept for a while'). Such perfectives express the idea of a temporal boundary (terminus), but not the idea of a goal (telos) or 'internal end point'. Furthermore, terminative is also used for semelfactives such as kriknut' 'shout', where there is also no inherent end point (see also Barentsen 1995, p. 5, who refers to predel'nost' v širokom ponimanii 'terminativity in a broad sense' as used by Bondarko and specifičeskaja predel'nost' 'specific terminativity' as used by Maslov for his term terminative). 4 For an extensive discussion of the differences between telicity and terminativity and the way in which various authors use these terms, see Genis (2008, pp. 91-100). In this paper we use Barentsen's term terminative.
Turning back to the three conditions for the pf given by Barentsen, we may illustrate this by having a closer look at a pf verb expressing an event like 'reading a book'. This event could schematically be presented as shown in Fig. 1 (cf. Barentsen 1985Barentsen , p. 61, 1995. 5 Terminativity means that the event can be regarded as a constellation of three situations. In this case situation X is the situation is which the book has not yet been opened, while situation Z represents the situation in which the book has been read. Situation Y stands for the 'event itself', the activity of reading transforming the initial situation into the final situation. The change of situation from X to Y and from Y to Z in Fig. 1 is accomplished by crossing the boundaries 'In' (initium) and 'Tr' (terminus). Barentsen indicates the crossing of these boundaries with 'd' (distance or difference). The presence of these boundaries shows that we are dealing with a terminative predicate, fulfilling the first requirement. This means that we can present the situation as complete, which the circle around situation Y symbolizes. This is the second requirement. However, in Russian these two characteristics of the event are not enough to use the pf aspect. For that, the event also needs to be sequentially connected to another situation, as such fulfilling the third requirement for the pf. Barentsen (1995, p. 16) provides the following definition of the feature of sequential connection, which, according to him, is part of the invariant meaning of the pf in Russian; because of its importance we provide the whole quote and our translation: Etot tretij priznak my uslovno oboznačaem terminom 'sekventnaja svjaz' '. Ego sut' zaključaetsja v podčerkivanii svjazej, kotorye celostnoe predel'noe dejstvie imeet so svoim okruženiem, čerez situacii Z i / ili X.Ėto značit, čtoėti situacii, ili xotja by odna iz nix, sovpadajut s vremennymi otrezkami, opredelennym obrazom uže izvestnymi iz konteksta ili rečevoj situacii. Čerez situaciju X dejstvie možet kak by 'ottalkivat'sja' ot momenta, raspoložennogo do samogo dejstvija. V takix slučajax obyčno podčerkivaetsja potencial'nost' vozniknovenija dejstvija Y. Čerez situaciju Z dejstvie 'privjazyvaetsja' k bolee pozdnemu momentu, i takim obrazomėtot moment xarakterizuetsja suščestvovaniem 'itogovogo sostojanija' dannogo dejstvija (Barentsen 1995, p. 16).
'For this third feature we will use the term 'sequential connection'. Essentially, this feature underlines the links that the total terminative event has with its surroundings through situation Z and / or X. This means that these situations, or at least one of them, coincide with temporal segments [a reference point or an other event, E. F., J. K.] that are, in a specific manner, already known from the context or from the speech situation. Through situation X the event is able to 'push itself away' from a moment preceding the event itself. In such cases, the potentiality of the realization of the event Y is usually emphasized. Through situation Z the event is 'tied' to a later moment in time, and in this way this moment is characterized by the existence of the 'final / resultant phase' of the event. ' The idea of sequential connection will be discussed extensively in Sects. 3-5, but we will provide one example here: 6 (1) My ustali pf . 'We are tired.' (Ru; cf. Barentsen 1998, p. 51) In (1) we find a fully complete terminative event (X, Y, Z), consisting of the transition to a state of tiredness that is sequentially connected through Z, to an externally given other situation, which in an isolated utterance like this can only be the speech situation. In cases like these, with a resultative event, the fact that the feature 'sequential connection' is part of the meaning of the pf aspect, makes that the utterance expresses the actual presence of the resultative state at the moment of speech.

Barentsen, ipf aspect in Russian
It is important to emphasize that the requirement of sequential connection has immediate consequences for the use of the ipf. The ipf is used in case any of the three features expressed by the pf aspect is absent. This means that we have ipf if (a) the situation expressed by the predicate is non-terminative (cf. 'atelic'; in the lexical meaning of the verb no specific boundaries between a preceding and subsequent situation and the 'action itself' are implied and accordingly there is no situational change), (b) if it is terminative but the terminus is not reached; i.e. the event is non-complete, non-total (e.g. in progressive contexts, in which situation Y is focused on), or (c) if it is terminative and complete but not sequentially connected. 7 Regarding the third feature Barentsen (1998, p. 53) notes, "[n]aša model' predpolagaet takuju vozmožnost', kogda NSV otličaetsja ot sootvetstvujuščej formy SV isključitel'no tem, čto otricaetsja priznak 'sekventnaja svjaz' '." 8 We can illustrate this by comparing the so-called general-factual use of the ipf in Russian (see also Sect. 5.2 on retrospective use of the ipf), with a regular pf past tense as given earlier in (1), for example: (2) [T]am možno spirt kupit' i kokain. Sovsem nedorogo.-Čto, pokupal ipf ? ' 'There you can buy alcohol and cocaine. Not at all expensive.' 'What, did you buy some?' ' (Ru; RNC: В. Пелевин. Хрустальный мир. 1991) Even though in both examples a complete terminative event is implied, it is only in (1) that the event is sequentially connected through Z, that is, only in (1) does the speaker express the fact that the realization of the event has an effect on the moment of speech. This is not the case in (2), where the speaker does not relate the result of buying alcohol and / or cocaine to the moment of speech (which would have been the case if the speaker would have used 6 Examples without a source indicated are our own (E. F., J. K.).
7 Even though the semantics of predicates are in some cases more and in others less easily associated with the idea of a terminus, in Russian most events can be presented as terminative because of the presence of prefixes such as delimitative po-which turn typical aterminative predicates into terminative ones (e.g. sidet' ipf 'sit' → posidet' pf 'sit for a while'). Cf. the broader sense of terminativity vs. telicity. 8 'Our model presupposes the possibility that the ipf differs from the corresponding pf form only insofar as the feature 'sequential connection' is negated.' kupil p ), but only focuses on the question of whether a complete buying event ever took place or not, which is a question of general experience. 9 As is argued by Barentsen, the feature of sequential connection is absent in some other Slavic languages such as Czech (Barentsen 1998, p. 55). This means that in Czech the meaning of the pf is: (a) the situation expressed by the predicate is terminative, (b) it is complete. Even though Barentsen does not explicitly discuss the meaning of the ipf in other Slavic languages such as Czech, one could infer from the logic of this theory that the ipf can be used in Western languages if the action is non-terminative or if it is terminative, but the terminus is not reached (i.e. non-completed, non-total). However, as we will see in our discussion below, this definition does not hold for some contexts.

Dickey's typology
In many ways the theory presented by Stephen Dickey (2000), and further elaborated upon in subsequent articles (among others Dickey 2001Dickey , 2005Dickey , 2011Dickey , 2015, 10 can be seen as a verification of and an elaboration on Barentsen's ideas, specifically on the difference between the meaning of the pf aspect in Russian and in Czech. In contrast to Barentsen's theory, which focuses mainly on Russian, Dickey's theory has a stronger typological character, because he discusses various Slavic languages. Another important difference to Barentsen is that Dickey presents his ideas within the framework of cognitive linguistics. This explains why he does not speak of invariant meanings and different uses-terms which are typical for the European structuralist framework-but of prototypical (or central) meanings and derived meanings. Furthermore, Dickey formulates Barentsen's notion of sequential connection in terms of 'temporal definiteness' following Leinonen (1982). 11 Like Barentsen, Dickey argues that the semantics of aspect is not identical in all Slavic languages. Based on a comparison of aspect usage between the various modern Slavic languages in a number of contexts he comes to the conclusion that there are two main groups within Slavic, in which aspect is used in different ways: an Eastern group (Ru, 12 Uk, Br and Bg 13 ) and a Western group (Cz, Sk, Sn and Sorb).
For the pf Dickey proposes one central concept, 'temporal definiteness', for the eastern group, and another, 'totality', for the Western group. For the ipf the central concept in the Eastern group is 'qualitative temporal indefiniteness', while the central concept for the Western group is 'quantitative temporal indefiniteness'. 9 The theory presented by Barentsen should not be interpreted in such a way that the question whether the event is complete (total) can be objectively determined on the basis of the state of affairs in the actual world. As such, the idea of completeness or totality has to do with the way the event is presented (cf. Comrie 1976, p. 18, who argues that it is incorrect to speak about a 'completed' event, and uses the term 'complete' event). 10 In these subsequent articles Dickey does not only refine the theory as proposed in Dickey (2000), but also ties differences in the productivity of particular markers of perfectivity to the east-west division and discusses diachronic developments. 11 Definiteness is reminiscent of the nominal domain. The linking of different domains to each other is typical of the cognitive approach to language, where similar concepts in different domains are often linked to each other, for example because they are seen as cognitively similar, or because a concept from one domain is understood in terms of a more basic concept from another domain. But note that Barentsen (1995, p. 11) also links the pf in Russian to the nominal domain, stressing the relationship between delimitative perfectives and countable nouns. Mehlig (1996) treats such analogies between Russian aspect and nouns more extensively. 12 We use the following abbreviations for language names: BCS-Bosnian, Croatian, Serbian, Br-Belarusian, Bg-Bulgarian, Cr-Croatian, Cz-Czech, Mc-Macedonian Pl-Polish, Ru-Russian Sr-Serbian, Sk-Slovak, Sn-Slovene, Sr-Serbian, Sorb-Sorbian, Uk-Ukrainian. 13 In a recent study (Dickey, to appear), Dickey characterizes Bulgarian as a peripheral member of the eastern group as far as the so-called general-factual use (see Sect. 5.2) is concerned. We will examine the definitions of aspect in the Eastern group first. As mentioned above, in this group the meaning of pf is temporal definiteness. An event 15 is temporally definite if it is uniquely locatable in a context, i.e. if it is viewed as contiguous in time to qualitatively different situations (Dickey 2000, pp. 26-27).
In Fig. 2 (cf. Dickey 2000, p. 27) we see the circle representing event X being flanked by two external events, (Y) and (Z). The letters above situations Y and Z are in parentheses to indicate that only one of them is needed for event X to be seen as temporally definite. This can be compared to Barentsen's definition of the pf in which the event is also linked to an external situation through either situation X or Z. As is pointed out by Dickey and Kresin (2009), the notion of temporal definiteness is similar to the notion of sequential connection provided by Barentsen (1998), since "temporal definiteness has as a practical effect the limitation of pf verbs in the eastern languages to contexts of (explicit or implicit) sequentiality" (Dickey and Kresin 2009, p. 125). In our view the crucial point of overlap between these theories is indeed to be found between the features sequential connection and temporal definiteness. We see the relationship between these concepts as follows: sequential connection is a feature of the pf aspect in the Eastern languages which requires the presence of a contrastive situation (cf. Barentsen 1985 who uses the term 'contrastive sequential connection') prior and / or subsequent to the event described by the pf verb, which makes the event temporally definite. Dickey (2000, p. 109) argues that the ipf aspect in the Eastern group expresses qualitative temporal indefiniteness, that is: the inability of an event to be assigned to a single, unique point in time relative to other states of affairs (see also Dickey and Kresin 2009, p. 126). 16 Dickey (2000, p. 109) represents this graphically (Fig. 3) as negating the schema for temporal definiteness. 14 Dickey uses the letter X to refer to the 'action itself' and Y and Z to refer to situations with which X is contrasted, while Barentsen uses Y to refer to the 'action itself' and X and Z to refer to contrastive situations. However, in the case of Dickey's theory the Y and Z are to be understood as external situations, while for Barentsen X and Z are part of the terminative event itself. Note that Dickey's representation with circles seems to suggest that the external situations are always complete (total). This is, however, probably not an intended feature of the theory. In the case of Barentsen's theory no such suggestion is made. 15 Dickey uses the term 'situation' where we use 'event'. To prevent terminological confusion as much as possible, we will change his term 'situation' to 'event' when it is used in the sense that we have defined for 'event '. 16 In this definition it is not clear whether these 'other states of affairs' in the definition are to be interpreted as preceding and / or subsequent situations, or that it also includes concurrent situations in processual usage, like the moment of speech in the actual pres, where all Slavic languages use an ipf aspect. However, Dickey and Kresin (2009, note on pp. 126-127) remark that although in those cases some phases of the event are clearly assignable to a single unique point in time, the entire event cannot be uniquely located, so the other states of affairs with which the event contrasts, are either preceding or subsequent situations.
Furthermore, the inability to be assigned to a single unique point in time is not simply a matter of 'objective inability', but it can also be a choice of the speaker to present an event that way in discourse contexts that facilitate or allow such a presentation (Dickey, personal communication). In Fig. 3 'NA' negates the complete representation of the pf aspect as given in Fig. 2. This is in fact reminiscent of Barentsen's 'negative' definition of the ipf aspect in Russian, where the ipf is used in case either one of the three layers of which the pf is made up of is not present.
The negation of the schema for temporal definiteness (see Figs. 2 and 3) can mean two things: 1. Event X is construed as more conceptual points on a timeline (i.e. in the case of a nonterminative event or if the terminative event is not complete, for example in a durative context). 2. Event X is construed as one conceptual point on a timeline, but without external situation Y and / or Z (e.g. in the case of general factual use as in (2)). The (perhaps unintended) implication of this description is that in this case the Russian (Eastern) ipf is conceptually identical to the Western pf aspect. 17 For the Western meaning of the pf aspect, totality, Dickey refers to Comrie (1976), and speaks of a synoptic construal of a situation (event), i.e. as an indivisible whole (Dickey and Kresin 2009, p. 124). This definition of the pf in the West is in fact similar to Barentsen's analysis of the Western pf 18 and can be graphically represented as shown in Fig. 4 (Dickey 2000, p. 26).
The circle represents event X as a single indivisible whole. In contrast to the Eastern pf, there is no relationship between X and its surrounding situations (Y or Z). As we have already 17 Barentsen does not explicitly discuss this point. We might argue that in his theory, which is essentially structuralist, the idea of totality in the case of the eastern ipf is an interpretation rather than a meaning. This means that it is not part of the meaning of the form, in other words: it is not expressed by the form, but it is something that can be inferred from the context in which the form-meaning element is used. 18 Barentsen (1985Barentsen ( , 1995Barentsen ( , 1998 does not speak about the 'synoptic construal' of an event, but stresses that in the case of a pf a terminative event is presented as complete, i.e. with the changes of state in focus. The similarity with Dickey's approach is that both see the pf aspect in the Western languages as expressing one feature less than the pf in the eastern languages. For Barentsen this is the lack of a sequential connection, for Dickey it is the lack of temporal definiteness, hence the absence of external situations Y and Z in Fig. 4. remarked above, the Western pf aspect is conceptually identical to the eastern ipf aspect in the case of a fully completed terminative event as in (2).
The ipf in the Western group revolves around the concept of quantitative temporal indefiniteness meaning the assignability of a situation to several points in time (Dickey 2000, p. 107). Dickey represents this graphically as illustrated in Fig. 5 (ibid., p. 108).
Again we see the timeline, but now event X is represented as occupying several points in time. Typical contexts in which we find this meaning are processual contexts where the focus is on an ongoing (i.e. non-complete) event and habitual contexts where the same type of event is repeated a non-specific number of times, which can be conceptualized as the event occupying several points in time. We will discuss this in more detail in the sections below, but it should already be noted that this meaning of the ipf in the West cannot be seen as the negation of the meaning of the pf. In other words, the ipf cannot be defined as 'non-totality'. This is because the Western ipf, just like the eastern ipf, is not only used in the case of nonterminative situations, or in the case of terminative situations that are not fully complete (e.g. in durative contexts), but also in the case of fully complete terminative situations-i.e. in the case of total situations-as long as these situations have some duration, i.e. if they occupy more points on a timeline for example: 20 (3) Kdo šil ipf ty šaty?
'Who made (lit. sewed) this dress?' (Cz; Filip 1999, p. 186) Besides the two main groups and their aspectual meanings as described above there are transitional zones, which have so-called polycentric networks in which the relative prominence of totality and temporal definiteness varies slightly from context to context. The northern transitional zone is formed by Polish, while the southern transitional zone consists of BCS and Macedonian. 21 As we will explain below, according to Dickey, the meanings postulated by him can explain several typological correlations within Slavic.
We have now discussed the general or central meanings for the pf and the ipf aspect in the East and the West respectively provided within the EWT of aspect. As we have seen, the 19 Dickey draws circles with different colours on both sides of the timeline. The point of this shading is to indicate the vagueness of the 'more than one point' (Dickey, personal communication). 20 The same is true the other way around: the Western pf aspect is not incompatible with situations that occupy more points on a timeline, cf. our example (32). So while pf verbs conceptualize the event as total, the same event can be presented as consisting of more points on a timeline by other means in the utterance. And while ipf verbs conceptualize the event as occupying more points on a timeline, this is not incompatible with a total interpretation. 21 Dickey (2000) does not discuss Macedonian, but Kamphuis (2014) shows that Macedonian can be seen as a transitional zone between the Eastern group and the Southern transitional zone proposed by Dickey, BCS. In Macedonian the pf aspect behaves like the pf aspect in the eastern group. The use of the ipf aspect, however, differs in some respects. notion of sequential connection (or temporal definiteness) of the Eastern pf is a crucial notion within the theory. In the following sections, we will discuss three contexts or usage types (habitual contexts, narrative contexts, retrospective contexts) and provide a critical analysis of how EWT accounts for the difference in aspectual use between the Eastern and the Western group in these contexts.

Introduction of habitual expressions
We define the term 'habitual expression' as an expression in which an event is presented as repeated an indefinite number of times. 22 Habitual expressions are typically accompanied by words (quantifying expressions) expressing concepts such as 'always', 'usually', 'often', 'sometimes', 'never', 'seldom'. Another term for habituality is 'unbounded repetition', which is opposed to 'bounded repetition' (cf. Barentsen et al. 2015) in which case there is a limit to the number of repetitions (e.g. 'twice', 'a few times', 'five times'). Contrary to bounded repetition, unbounded repetition, i.e. habituality, does not need to be expressed by a special form, such as an adverb, in Slavic, as is illustrated by the following example from Russian, where the mere use of the ipf past tense forms already suggests habituality in the given context: The use of the imperfectives daval, ob"jasnjal and govoril in the context in example (4) implies that the events were not realized only once, but occurred over a period of time, on various occasions. In English this can often be expressed by a construction with 'would', like in (4) 'would give'. Habitual expressions across Slavic are known to behave differently with respect to verbal aspect; an overview is given in Table 1.
The differences in aspectual use are generally linked to the two levels on which aspect can work in habitual contexts (Mønnesland 1984, p. 54;Stunová 1993, p. 35). The first level is the micro-level, which is the level of the individual sub-event. In the case of (4) this is each individual instance of the event of giving, explaining or speaking. The second level is the macro-level, the level on which the individual sub-events form a collective macroevent (Timberlake 1982, p. 315). In the case of (4) this is the whole complex of the repeated giving, explaining and speaking events. On the micro-level it is possible to see each repeated situation as a totality, whereas this is impossible by definition on the macro-level because of the presentation of the repetition of the events as unbounded. As we will show, languages 22 'Indefinite' is used here as opposed to 'definite.' So the definition does not imply that habitual events are being presented as repeated a countless number of times, rather that the number of repetitions to which the habitual expression refers is not fixed, as it is in bounded repetition. Habitual expressions can be seen a generalization over a number of occurrences (cf. Carlson 2012, p. 829) and the number of occurrences needed for such generalizations depends on the particular event and the context. Past habitual contexts ipf / pf ipf differ as to which degree they allow habitual events to be conceptualized as total on the micro-level.
In the following sections we will say more about the use of aspect in the different Slavic languages in habitual expressions, and how the EW-model accounts for this use. Since there is a difference between aspect usage in habitual expressions in the pres and in habitual expressions in the past (Dickey 2000, p. 77), we will treat the way the EW-model deals with these contexts separately. Mønnesland (1984, p. 54) shows that there is a division in Slavic between languages that use the pf pres in habitual contexts, and languages that use the ipf pres. In Russian, which is a representative of the Eastern aspectual group, the normal means of expressing a present habitual is the ipf pres, and the pf pres is not acceptable in most habitual contexts (Forsyth 1970, p. 172;Mønnesland 1984, p. 61):

Habitual expressions in the present
(5) Každyj den' ja vypivaju ipf (*vyp'ju pf ) rjumku vodki. 23 'I drink (finish) a glass of vodka every day.' (Ru; cf. Mønnesland 1984, p. 61) A similar situation can be found in the other Eastern aspectual languages like Bulgarian and also the transitional languages Polish and Macedonian that behave like Eastern languages in non-past habitual contexts. The Russian situation differs from the Western aspectual Slavic languages such as Czech, Slovak or Slovene and also the transitional language BCS, which aspectually behaves like a Western language in non-past habitual contexts. In these languages it is possible to use the pf pres in habitual contexts, as is illustrated by the following Czech example: (6) Vypije pf jednu skleničku vodky denně. '(S)he drinks a glass of vodka every day.' (Cz; Dickey 2000, p. 52) However, in Czech, as in the other Western languages, it is also possible to use the ipf pres in habitual contexts: (7) Denně dostávám ipf několik dopisů. 'I receive some letters every day.' (Cz; Petruxina 1978, p. 60) This use of the ipf is associated with the macro-level and can also be triggered by specific adverbs, like stále or pořád in the meaning 'all the time' (see for example Petruxina 1983;Dübbers 2015, pp. 201-204). 24 Note that in Russian, as in Czech, the pf pres is also used as the pf fut tense, that is, to denote a single completed terminative event in the fut, for example: 25 (8) Ja vyp'ju pf rjumku vodki.
(Ru) (9) Vypiju pf skleničku vodky. (Cz) 'I will drink (finish) a glass of vodka.' Such use differs from its use in sentences like (6) in which the pf pres does not have a fut tense meaning.
Within the EW-model, the restriction in Russian (and other Eastern languages) on the pf pres form in habitual contexts is explained by pointing at the fact that the pf expresses sequential connection, or to put it differently, it needs a temporally definite context in which it can be contrasted with either a preceding or a subsequent event. Barentsen (1995, p. 21) argues that in the case of an (isolated) fut tense use of the pf pres as in (8), the pf aspect expresses a sequential connection with the moment of speech, preceding the realization of the pf pres event. The pf pres event Y is 'pushed away', as it were, from the situation at the reference point X, creating a contrast with the reference point, which results in a fut or potential interpretation. However, in a habitual context like in (5), with a non-terminative macro-event and no contiguous, qualitatively different situations relative to the micro-event, the pf aspect cannot occur in Russian, so the only choice is to use the ipf aspect. 26 This differs from the situation in Czech.
The possibility to use the pf in the Western group as in (6) can be explained in terms of the absence of the feature sequential connection. In the Western group, the pf aspect expresses totality. In the case of (6), the pf expresses a total, fully completed terminative event on the micro-level, which is indefinitely 'multiplied', in this case by the adverb denně 'every day'. Unlike Russian, Czech can focus on both the unbounded repetition of the (total) events, which creates a non-terminative event on the macro-level by using the ipf, as in (7) or on the totality of the (terminative) event on the micro-level by using the pf, as in (6). One could be tempted to say that Russian focuses on the macro-level by using the ipf in examples like (5), but the fact that there is no choice of aspect, shows that we are not dealing with a deliberate emphasis of the macro-event, but with a necessary consequence of the meaning of the Russian pf.

Exceptional cases in the Eastern group
Even though the use of the pf pres form in Russian is not possible in habitual contexts like (5), there are specific habitual contexts in which the use of a pf pres is possible. This is, for example, the case if the repeated situations are linked to each other, either as a pair, or in a chain, the so-called 'habitual-correlative constructions' (Bondarko 1971, pp. 197-208). In (10) the habitual interpretation is triggered by the use of the habitual adverb vsegda 'always', which occurs at the beginning of the sentence as a 'multiplier', having scope over the pair of events, and which, in this case, is orthographically separated from the rest of the sentence: (10) On vsegda / vyp'et pf kofe i pojdet pf na rabotu// 'That's what he always does-drinks his coffee and goes to work.' (Ru; Zemskaja 1983, p. 125) The EW-model can explain the occurrence of the pf aspect in this construction because there is another situation on the micro-level that provides the temporally definite environment needed for the pf to occur in. Other languages from the Eastern aspectual group, such as Bulgarian, behave like Russian by only allowing pf forms in pres habitual contexts when sequentiality is in play. 27 Another use can be found in sentences with a 'when' clause, in which situations in the main and the dependent clause are contrasted. In (11), the realization (full completion) of the first terminative event zaxočetsja (get a desire) is linked to the occurrence of another event smogu posmotret' (be(come) able to see him): (11) Vsegda, kogda mne zaxočetsja pf , ja smogu pf posmotret' na nego [. . .].
'Always, when I feel like it, I can look at him.' (Ru; RNC: O. Čexova. Moi časy idut inače. 1973) However, the event in the main clause can also be expressed by an ipf pres (cf. Bondarko 1971, p. 198, who also provides an example without habitual expression): One might argue that in such sentences there is a sequential connection with the situation given in the main clause. In that case we have to conclude that an event expressed by an ipf verb can be an instance of what Dickey calls a 'qualitatively different situation'. However, this view seems at odds with Barentsen's remark that in the case of the ipf, the feature or sequential connection is either absent or negated (Barentsen 1995, p. 18). We will come back to this issue in Sect. 3.3. In Russian, there are also cases where a pf pres form occurs in a habitual context without it being part of a pair or chain, for example: 27 Also, in the transitional language Macedonian, the ipf pres tense is normally used in the case of habitual contexts, and the pf pres (in a construction with the modal markerḱe) is only used when there is a sequential connection with another event: BCS, the immediate neighbor of Macedonian in the transitional zone, allows for the use of perfectives as they are used in Czech with no special definite environment. Polish, the transitional language in the north, allows the use of perfectives to a greater extent than Russian, but their use is still more limited in this regard than Czech and BCS (Dickey 2000, pp. 68-71 Forsyth (1970, p. 173) speaks of 'singularization' in such cases (cf. the traditional Russian term nagljadno-primernoe značenie 'visual exemplary meaning') that is, one occasion or one complete performance is selected or quoted, and held up as a sample of, or to exemplify a recurrent phenomenon. Because of the absence of an explicit temporal definite context, sentences such as these are more difficult to explain using EWT, as is also acknowledged by Dickey (2000, p. 57). Dickey argues, however, that in cases like (14) there is a contrast between the 'sudden, exceptional event' expressed by the pf pres verb and the background against which the event occurs. This contrast creates a conceptualization of temporal definiteness, connecting the unexpected event expressed by a pf pres, to the preceding situation. 28 However, such an analysis is difficult to apply to instances like (15) where the idea of unexpectedness is not clearly present:  Forsyth (1970, p. 174) argues that the pf pres expresses the action as a contingency in such cases-an action that is expected to occur from time to time when the appropriate circumstances for it occur. Following this line of reasoning, a sentence like (15) can be paraphrased as 'In a German school, whenever the appropriate circumstances arise, it will always (certainly) be the case that . . .', in which we again have two situations that are contrasted. Such sentences therefore have a potential character, which is absent in comparable ipf sentences, cf. (16): (16) -A vospitatel'nicy vsegda govorjat ipf , čto sadit'sja na zemlju nel'zja, možno prostudit'sja i ispačkat'sja. 'And educators always say that you cannot sit on the floor, otherwise you could get cold and dirty.' (Ru; RNC: E. S. Ginzburg. Krutoj maršrut: Čast' 2. 1975-1977) A similar analysis is also given by Barentsen (1995). Barentsen discusses the use of instances of singularization by pointing out a relationship with sentences like the following, where the situations are not localized at a specific moment in time: (17) Esli, naprimer, brosit' metalličeskij šar v jaščik s peskom, on šlepnetsja pf i ostanovitsja pf . 'If you, for example, throw a metal ball into a box of sand, it will plop down and come to a stop.' (Ru; Barentsen 1995, p. 21) Since such conditional cases describe a general rule ('every time that X, Y'), the difference between an interpretation of the event as occurring once, and an interpretation of the event as occurring repeatedly is blurred. Whereas pf pres forms usually connect with a situation X, for example the speech moment or another reference point, in the case of exemplary use there is a connection with a non-concrete or non-unique point, an arbitrary point of a certain type. Barentsen (1995, p. 21) therefore argues that because of this, it is not surprising that in Russian such temporal-aspectual forms are very wide-spread in different versions of the 'vivid-exemplary meaning.' 29 As such, Barentsen suggests that in Russian there must be a sequential connection with another situation, and that this situation can be either explicitly mentioned, like in a narrative, or implicitly like the moment of speech in retrospective languages (see also our discussion of the retrospective context) or a reference point that is not uniquely localized in real time but that is, as it were, 'created' by the restriction sequential connection puts on the situation in which pf verbs occur (cf. Zel'dovič 2002, p. 31).
Notwithstanding the specific explanation of these more complicated examples, all these habitual contexts in which the pf pres is used have the fact in common that a representative example of the repeated micro-events is singled out, making it possible to connect the single completed terminative event to a reference point, similar to the case of fut reference in (8).

Habitual expressions in the past
The difference in aspect between the Eastern group and the Western group is even more prominent in the case of past tense habitual contexts. In the past tense, habitual expressions allow both the pf and ipf aspect in the Western group, whereas in the Eastern group they allow for the ipf only. This can be illustrated by the following English fragment from Winniethe-Pooh taken from the Amsterdam Slavic Parallel Aligned Corpus (ASPAC), 30 which is translated with pf past-tense forms in Slovene and ipf past tenses in Russian: (18) Every morning he went out with his umbrella and put a stick in the place where the water came up to [. . .].
(Winnie-the-Pooh) 'Každoe utro on vyxodil ipf s zontikom iz doma i paločkoj otmečal ipf mesto, do kotorogo podnimalas' ipf voda.' (Ru translation) 'Vsako jutro je z dežnikom odšel pf ven in zataknil pf palico tja, do kamor se je vzdignila pf voda [. . .].' (Sn translation) In Russian, the ipf is obligatory in past-tense habitual expressions, whereas in Slovene both the pf and the ipf are possible. In (18), Slovene uses pf verbs, but as (19) shows, in some habitual-like contexts Slovene also uses the ipf, for example in the following sentence in which there is a constant repetition of the same action at a particular moment in time: (19) He was taking the balloon out, and putting it back again.
(Winnie-the-Pooh) 'On opuskal ipf šarik v goršok i vynimal ipf ego snova i snova.' (Ru translation) 'Ves srečen je jemal ipf balon iz lonca in ga spet deval ipf vanj.' (Sn translation) The difference between Russian and Slovene can be explained in the same way as in the case of pres tense habitual expressions. In Slovene, it is possible to focus on the unbounded repetition of the event on the macro-level by using the ipf, by which the situation is presented as 'non-total', i.e. occupying more than one conceptual point on a timeline. Because of the meaning of the pf in Slovene, totality, it is also possible to focus on each individual completed repeated situation (micro-situation) and use the pf in contexts like (18). Russian, on the other hand, does not allow for the possibility to focus on each individual situation (micro-situation) by using the pf, because the pf does not only express totality, but totality and something more-viz. sequential connection / temporal definiteness, which is generally incompatible with habitual contexts, as we have shown in the previous paragraph. In these cases, one has to use the ipf in Russian, because there is no unique situation with which a sequential connection can be made (see e.g. Stunová 1993 and Barentsen 2008 for a more elaborate discussion). The difference with habitual expressions in the pres tense is that there are almost no exceptions to the restriction on the pf in past tense habituals in Russian, like in case of the pres tense (see also Mønnesland 1984). Sentences like the following, which is the past tense equivalent of (15) given earlier, are not acceptable, even though one could imagine that the pf in (20) would create a context of singularization just like in the pres tense: (20) V nemeckoj škole vsegda *skazali pf , na kakoj stranice kakoj abzac čitat' i kakie imenno frazy nužno zapisyvat'.
(Ru) (Intended meaning: In a German school they would always tell you, on what page you had to read what and exactly which phrases you had to write down.) In habitual-correlative constructions the use of pf past is also very limited (cf. Dickey 2000, pp. 74-75). Examples like (21), which is the past tense the equivalent of (10), in which there is a pause between the quantifier and the habitual events, are only possible in spoken language, according to Zemskaja (1983, p. 125) (cf. Bondarko 1971: (21) On vsegda vypil pf kofe i pošel pf na rabotu//.
(Ru) (That's what he always did-he would drink his coffee and go to work.) The way in which habitual expressions differ in the pres tense is, however, not very easy to explain. It might have to do with the fact that in the case of a past tense event, both the microevents and the macro-event are already realized, which makes it difficult to single out one micro-event as exemplary for the complete macro-event. For past tense situations the vantage point lies after the realization of the habitual event, which also means after the realization of all the micro-events, which makes it impossible to single out one event and present it as a representative instance of the habitual macro-event. When the pres tense is used, the temporal orientation lies before the realization of the event, which makes it possible to single out one instance of an event as being representative for the habitual event. This is probably because the idea of a general rule or law is in accordance with an 'if X then Y'-structure, which has an inherent potential meaning (cf. our discussion of (17)). In that case the verb refers at least on the macro-level to a non-actual event that is not yet realized at the moment of speech. Hence, the pf pres, as in (15), refers to a non-actual event that is not yet realized at the moment of speech. Because of this, it is probably easier to construe a context in which the speaker pushes himself away from an imaginary (and therefore also potentially repeatable) point than in the case of the past tense, as such singling out one individual repeatable event. 31

Special tense forms in Macedonian and Bulgarian
There is more evidence that the vantage point plays an important role. Note that while Macedonian and Bulgarian behave similarly to Russian in the case of pres tense habituals, the situation is different in the case of the past tense. In Bulgarian and Macedonian, habitual past events can be expressed by a special tense form. In Macedonian, it concerns a pf or ipf 31 Dickey (2000, pp. 77-80) also provides an analysis, which is to some extent similar to our analysis. imperf in combination with certain particles or conjunctions (e.g.ḱe, da, ako, koga). The typical construction is an imperf with the particleḱe, which is aptly called 'future-in-the-past' (Koneski 2004, pp. 491-496), and is also used to shift back the vantage point to a moment in the past before the realization of an event, for example in the irreal or conditional mood. 32 When such a form is used, an additional vantage point is created: the action which is referred to is prior to the first vantage point, due to the imperf tense, but a second vantage point lies before the realization of an event, for example in the irreal or conditional mood, which is due to the pf aspect. The Bulgarian pf imperf is used in the same function (Lindstedt 1985, p. 241). The availability of the pf imperf gives these South Slavic languages more freedom to use pf verbs in habitual constructions than in Russian. The ipf imperf (and not the pf aor) is used in past habitual contexts in Bulgarian (22) and Macedonian (cf. Kamphuis 2014, pp. 134-135) where no explicit situation is present that can function as a context for sequential connection: Subordinate clauses seem to facilitate the use of the pf in habitual contexts. We already discussed this with respect to the pf pres as in (11)-(13), and we find similar cases with non-32 Remember the 'if X then Y'-function of the Russian pf pres in singularizing contexts, which is similar to the conditional or irreal mood. Kalsbeek (2012, p. 347) argues that it is exactly this overlap in function that leads to the use of the conditional in BCS to denote repeated actions in the past. 33 In BCS, the conditional is used in the same fashion, although, unlike the pf imperf in Macedonian and Bulgarian, it can be used without a contrasting situation being mentioned in the context (Dickey 2000, pp. 72-73;Kalsbeek 2012). 34 http://www.bermama.ru/blogs/chuzhih-detei-ne-byvaet/gostevoi-dlja-detei-sirot-naskolko-on-vazhen.html. finite forms such as the inf, as in sentences with pered tem, kak 'just before' where the use of the pf inf is very common: 35 (26) Očen' často, pered tem kak skazat' pf pravdu, ja naprimer, dumaju, a nado li?
'Very often before telling the truth, I think, for example, is it really necessary?' Even though the subordinate clause facilitates the use of the pf in (24), for the pf past tense to be used more factors are needed. In (24) this is the meaning of the conjunction posle togo, kak 'after', which is strongly associated with a retrospective (perf) meaning. Similarly in (25) the conjunction do togo, kak 'until' focuses on an event which blocks the full completion of the event (raskrylis' 'open'). In other contexts with other conjunctions the use of the pf tense is not always possible. In (27) with kogda 'when' the pf past tense is not acceptable, and instead the ipf past tense is chosen. This is due to the presence of vsegda 'always', which can be illustrated with (28), where the pf is acceptable: (27) Vsegda kogda on *prišel pf / prixodil ipf domoj, on srazu ložilsja spat'. 'Always when he came home, he immediately went to bed.' (28) Kogda on prišel pf domoj, on srazu ložilsja spat'. 'When he came home, he immediately went to bed.' Also note that the pf is possible in the semantically similar construction with a pf ger, which again underscores that an analysis of this phenomenon has to take into account various factors such as the meaning of the verb form (past tense versus ger in this case) and the meaning of the conjunction, cf. (29): (29) Vsegda, pridja pf domoj, on srazu ložilsja spat'. 'Always, after he had come home, he immediately went to bed.' The reason why subordinate clauses facilitate the use of the pf in habitual contexts, in some cases even with the past tense, is not fully clear to us. One possible factor is that the subordinate clause provides an additional piece of information about the main clause. This semantic and syntactic separate status probably causes habitual adverbs such as vsegda 'always' to not be directly applied to the event in the subordinate clause. We may further illustrate this with English, where one can say Always, after I get home from work, I spend the rest of the night working on a piece, or After I get home from work, I always spend the rest of the night working on a piece, but not *After I always get home from work, I spend the rest of the night working on a piece. It could be argued, therefore, that even though a habitual context normally speaking triggers an ipf, this is not necessarily the case in subordinate clauses, even if the subordinate clause contains a past tense. If the past tense event in the subordinate clause is strongly associated with the idea of full completion (result), this can overrule the tendency to use the ipf. This is possible because the subordinate clause can form a kind of 'island' for the habitual form. 36 How such cases are accounted for within EWT is not fully clear to us, since we find an ipf in the main clause in the examples (24), (25), (26) and (29) provided by us. One could hypothesize that the pf event in the subordinate clause is linked to the microevent in the main clause. Cases such as these require further study. 37 Such an analysis should 35 In the RNC the query 'vsegda pered tem kak' yielded 8 examples, all of which contained a pf inf. 36 Also see the discussion in Fortuin (2008, pp. 215-219), which shows that pf inf can occur in habitual contexts if they are dependent on an ipf predicate. 37 In some cases, we find a pf both in the subordinate clause and the main clause. An example of this is (11) given earlier in the text which has two pf pres forms. Another example is given below with the habitual expression po utram 'in the mornings', the subordinator kak tol'ko 'as soon as', and an imp in the main clause: pay attention to the different verb forms (pf pres, pf past tense, pf inf, pf ger, pf imp), type of subordinators and type of habitual expressions.

Habitual sequences of events in the Western languages
Finally, it should be noted here that in Western languages such as Slovene and Czech, the presence of a sequence of events often triggers the use of the pf aspect, similar to what happens in habitual expressions in the pres tense in the Eastern group (see Dübbers 2015, for Czech). For instance, while in Czech sequential connection is not an inherent feature of the pf, perfectivity and sequentiality show correlation there as well. In the following example, the pf is the preferred choice of aspect, because of the fact that the habitual events occur in a sequence: (30) V Žižkově ulici měl tehdy lahůdkářský obchod pan Brůžek. Tam velmi často muž zašel pf a koupil pf bud' dva pomeranče, nebo banány a mě a hocha podělil pf . 'In the Žižka street, Mr.
Brůžek had a delicatessen shop in those days. The man very often went there and bought either two oranges or bananas and gave one to me and one to the boy.' (Cz; Dübbers 2015, p. 200)

Evaluation of the explanation of habitual contexts within the EW-model
In habitual expressions, both in the non-past and the past, EWT is able to account for the observed variation across Slavic by postulating the presence of the feature sequential connection for the pf in the Eastern aspectual group. The restrictions this feature places on the contexts in which pf verbs can occur, namely the presence of a preceding or subsequent contrasting situation, is not in accordance with habitual contexts in which there is no other situation with a unique status to which the event can be connected. Since in Czech and other Western languages 38 sequentiality is not an obligatory feature of pf aspect, in those languages there is a choice to either express unbounded repetition on the macro-level by using an ipf verb, or focusing on a representative instance on the micro-level by using a pf verb. The feature of sequential connection can also account for some exceptional cases in the Eastern group, which allow for the use of the pf pres in habitual contexts. All these exceptional contexts can be explained in terms of the possibility to create a sequential connection. In some instances there is a sequence of (micro-)events, where one event is linked to the other, whereas in other instances there is a context of singularization, where a connection is created with another situation that is presupposed by the usage of a pf form. All these cases also have a specific meaning that seems to be absent in the case of the Western pf aspect in pres habitual contexts, which further corroborates that it is the feature of sequential connection that facilitates these uses in the East. 39 We have also shown that past tense contexts do not (i) Po utram, kak tol'ko privedeš' pf sebja v porjadok, navedi pf porjadok i na svoej planete.
'In the mornings, as soon as you have attended to yourself (i.e. get ready), attend to your planet.' (Ru; translation of Le Petit Prince, Saint-Exupéry; Fortuin and Pluimgraaff 2015, p. 224) In such cases the habitual form po utram has scope over the entire sentence, in which the situation in the subordinate -clause is sequentially connected to the situation in the main clause. We have not found such instances with the past tense. More empirical research is necessary before any conclusions can be drawn. 38 Including BCS in the transitional zone. Polish mainly falls within the Eastern group. 39 It should be noted that there are instances where Czech seems to behave more or less in the way that Russian does. An example is the potential reading of the pf pres (see Sonnenhauser 2008, for an overview). Within the EW-model, the modal character of this use can probably only be explained in terms of sequential connection similar to the exemplary use of the pf pres in the Eastern languages. The same usage type also seems to occur in Czech, where the pf aspect does not have the feature sequential connection.
allow for such exceptions in the majority of cases. This is harder to explain within the theory, even though we have argued that the past tense probably makes it difficult to single out one instance of a repeated situation. A better understanding of the concept of 'reference point' is necessary, including a discussion of exceptional cases such as (24). Something which merits special attention is the fact that a sequence of events in Czech often triggers the use of the pf, as in (30), even though it is exactly these two notions that in other cases are used to explain the difference between the Eastern and the Western group. Dickey (2000, p. 264) states that the notions of totality and temporal definiteness are conceptually proximate and that sequentiality entails totality. Following this line of thought one could perhaps argue that a context of sequences of events is the most natural context for the meaning of totality, which is the meaning of the pf in the West, and part of the meaning of the pf in the East. In the East, the pf also signals sequential connection in addition to totality. This explains why in the East there is an almost one-to-one correspondence between contexts of sequences of events and the pf, whereas in the West there are particular exceptions, as we will show in Sect. 4.

Introduction to sequences of event in narration
Two major differences between the Eastern group and the Western group have been described with respect to sequences of events in narration. The first difference is that in the Eastern group (e.g. Russian) in past tense narrations a pf verb is required in most contexts in which sequences of events are presented, whereas in the Western group (e.g. Czech) ipf verbs are much more common. Secondly, in a pres tense narration (e.g. historical pres) the Eastern group requires an ipf pres tense, whereas in the Western group, both the ipf pres tense and the pf pres are possible. So, in both instances the Western group displays more variety in aspectual choice, similar to the situation with respect to habitual expressions. The two general rules for the Eastern group seem contradictory at first sight, but we will see that the theory can explain this difference (Table 2). We will discuss these contexts, and how they can be accounted for within EWT by starting out with past tense narration in Sect. 4.2 and turning to pres tense narration in Sect. 4.3.

Past tense narration
In Russian, ipf verbs are generally not allowed in sequences of events in the past tense; the use of the pf aspect is obligatory in most of those contexts. This can be illustrated by the following Russian sentence taken from Barentsen (1998, p. 52): 40 (31) Vot mi i svernuli pf (*svertyvali ipf ) nalevo i koe-kak, posle mnogix xlopot, dobralis' pf (*dobiralis' ipf ) do skudnogo prijuta. 'So we turned left and somehow, after much trouble, we managed to get to the meager shelter.' (Ru; M. Ju. Lermontov; cited in Barentsen 1998, p. 52; imperfectives added by E. F., J. K.) In the Western group, the use of the pf is also favored in similar past tense narrative contexts, even if the event itself has a durative character like Czech example (32); in this case the past tense narrative, with different events that occur one after the other, favors the meaning of totality over the meaning of 'several points in time' (the same is true for Slovene as well): (32) Pak se najednou zastavil pf , udělal pf vlevo v bok a pomalu přešel pf ulici na druhej chodník. 41 (Cz) 'Then he suddenly stopped, turned left, and slowly crossed the street to the pavement on the other side.' Note, however, that in the Western languages, the use of ipf verbs is sometimes possible in sequences of events, unlike in the Eastern languages. As is shown by Stunová (1993), the use of past tense imperfectives in narrative contexts in Czech is often found with verbs of a certain lexical class like verba sentiendi, verba dicendi and determined verbs of motion (Stunová 1993, p. 128), for example: (33) Když me viděl ipf , poroučel ipf se té paničce a šel ipf ke mně.
'When he saw me, he said goodbye to the little lady, and walked up to me.' (Cz; Čapek; cited in Stunová 1993, p. 123) According to Stunová (1993, p. 112, p. 122) in cases like (33), the use of three pf verbs is also possible. The use of the ipf suggests that the actions do not follow each other in a strict successive order, as would have been the case when pf verbs were used. However, the three ipf verbs also do not express three parallel processes, but rather a succession of events that could be interpreted as partly overlapping (Stunová 1993, p. 112). Something similar can be said to be the case in (34), where we have two pf and one ipf verb: below. In (i) and (ii) the meaning of the verb, or the fact that there is no natural pf (cf. Janda 2007) to these ipf verbs, seems to play a role in the use of the ipf, and in (iii) the use of the ipf portrays the event as having duration: It is not fully clear to us how EWT accounts for such instances. To give an example, how are sxvatili and rasstreljali in (iii) sequentially connected to the surrounding context? This again raises the question of whether it is possible to make a sequential connection to a situation that is expressed by an ipf verb. Alternatively, one might argue that in these cases, there is a mix of a narrative sequence of events and a retrospective style of narration, which facilitates the use of the ipf. In addition, lexical factors may play a role, such as the meaning of modal verbs (see for example Barentsen 2002, for the semantics of (s)moč' 'can'  Ivančev 1961, p. 83) Here, the use of the pf forms kousl and zkřivil implies strict sequentiality, while the ipf vracel creates an effect of smooth transition or continuity of the action (Galton 1976, p. 70;Dickey 2000, p. 217). Such effects can be attributed to the fact that the ipf does not express totality, so there is no clear boundary to the action. 42 Overlap or smooth transition are, however, not the only factors that trigger the ipf aspect. In many instances in Czech, the ipf is used to express events that clearly follow each other in a sequence, often with an ingressive meaning like in (35) In (35), the ipf verb clearly refers to an action that occupies more than one point on a timeline. However, at first sight, cases like vracel in (34) seem difficult to account for when several points in time is taken to be the basic meaning of the ipf aspect. Berger (2013) discusses these kinds of examples. He shows that the ipf can always be replaced by a perfective partner, and comes to the conclusion that the use of the ipf aspect in these cases can be seen as a way of 'slowing down the action' (ibid., pp. 40-41). This, of course, stretches the definition of several points in time to a point where it has little to do with actual duration anymore. Nevertheless, one can imagine that slowing down time makes it possible to view an event that normally is seen as momentary, as having a substantial duration, to create a dramatic effect, much like slow motion effects in movies. Stunová (1993, pp. 116-117) shows that in cases where Czech uses an ipf verb, Russian often has an ingressive pf verb (typically with za-, po-, or u-, or a construction with the phase verbs stat' and načat'), and in some cases a delimitative verb (with po-). In Russian, the ipf of example (35) would be rendered with a pf verb, in this case expressing ingressivity, cf. (36): (36) Raz kak-to vzjal pf klarnet i zaigral pf . . . (Ru;Dickey 2000, p. 225) In his overview of several Slavic languages, Ivančev (1961, pp. 102-103) shows that there is indeed a negative correlation between the use of za-as an ingressive prefix and the use of the ipf past tense in narrative contexts (cf. Dickey 1999). 43 Within EWT, the difference between Russian and Czech aspect usage is explained by pointing out that the ipf in the Eastern languages signals the absence, or denies the presence, of a sequential connection (temporal definiteness), unlike the pf (Barentsen 1995, p. 18). Since events in past tense narration are typically presented as being connected, this context is not compatible with the ipf aspect. In the Western languages, on the other hand, ipf aspect does not deny the unique location of an event in a sequence (Dickey 2000, p. 232), or, in other words, the existence of contrasting situations, and is therefore not incompatible with past sequences of events. At first sight it is not entirely clear how one can explain the obligatory use of delimitative (po-) or inchoative (po-, za-) perfectives in Russian past tense narratives 42 Our Czech informant does not think that this example to expresses an overlap, but rather as a concentration on the action of 'giving back' itself, leaving the possible consequences of the action out of focus. This creates a kind of dramatic moment in the sequence of events. 43 Ivančev (1961, p. 48) considers Slovene in this respect to be a transitional zone between Russian and Bulgarian on the one hand and Czech on the other. Note that this differs from most of the other contexts that we are discussing, where Slovene seems to occupy the most extreme position within the Western languages.
instead of the unprefixed base verbs (e.g. posidet' 'sit for a while', postojat' 'stand for a while', pojti, 'leave', zapet' 'start singing' etc.), solely based on the basic meanings of pf and ipf aspect. Why would a speaker, for example, not be able to choose to present a chain of events in an isolating manner by using ipf forms like in the historical pres (cf. Sect. 4.3 on the historical pres)? Nevertheless, the past tense narrative context itself seems to require the use of these prefixes, at least in many contexts (cf. the restriction on the use of pf verbs in habitual past contexts in the Eastern languages). Dickey and Hutcheson (2003, pp. 25-26) connect the relative absence of ipf verbs in past sequences of events in modern Russian to the rise of delimitative verbs in Russian. From a diachronic perspective, one could say that sequentiality and perfectivity became more closely intertwined in the Russian system from the sixteenth or seventeenth century onwards. One of the effects of this development was the increasing use of pf delimitative verbs in sequences, where Old Russian uses ipf verbs, just like modern Czech. Therefore, even if the situation itself is aterminative, a specially derived pf verb is required in contexts of sequentiality in the past tense. This is in line with Barentsen (1998, p. 52) who argues that a past tense narrative context with completed terminative events is the most basic instance of a sequential connection. As Dickey (2000, p. 77) puts it, the past tense itself "bears a certain resemblance to the concept of temporal definiteness [. . .] just as the past tense contrasts one situation (that denoted by the predicate) to another (the present), temporal definiteness construes a situation as unique relative to other situations (and often to the present)".

Some data that are more difficult to account for
Even though at first sight the data can largely be accounted for within EWT, there are also some linguistic data, which are more difficult to account for. The data from Bulgarian, an Eastern language, like Russian, are largely compatible with the theory, but also show that aspect is certainly not the sole factor that plays a role. The general preterit in Bulgarian is the aor, which, like in Russian, is usually pf in narrative contexts with sequenced events. However, there are also examples of ipf aor in narrative sequences, often used to refer to a durative situation in a sequence of events like in (37): (37) Tja spa ipf.aor i započna pf.aor da raboti.
'She slept and began working.' (Bg; Lindstedt 1985, p. 181) In those cases, the aor is used to express a temporal boundary, which is a typical function of the aor. With verbs that are not terminative this use is very reminiscent of the use of the delimitative verbs in Russian. This characteristic of the aor is the reason Barentsen (1998, p. 52) calls the function of the perfective past to present past sequences of events in Russian, the 'aorist meaning' (following Bondarko 1971). The Bulgarian use of the ipf aor shows that the possible choice in tense forms plays an important role in the choice of aspect as well. It also shows that the ipf is not necessarily incompatible with presenting past events in a sequence in Bulgarian. 44 Another interesting example from Bulgarian is the use of ipf imperf 44 Such examples of events in a sequence with ipf aor should not necessarily be interpreted as presenting the events as sequentially connected, or temporally definite (cf. our discussion of the way events are presented in the historical pres, Sect. 4.3). The aor is used to refer to a total event, while the ipf aspect is used to indicate that the event is not terminative, has no inherent limit. The Bulgarian ipf aor is thus the ultimate example of a total event presented by an ipf verb. In other languages, like Russian, the ipf is not incompatible with a total interpretation (cf. the ipf general factual usage, Sect. 5.2), but it does not express totality, which is exactly what the ipf aor in Bulgarian does. Bulgarian (and to a lesser extent Macedonian) is thus the only Slavic language in which totality and temporal definiteness are also formally discernable.
as a kind of reportive past tense (the past counterpart of the reportive pres), cf. Lindstedt (1985, p. 141) Stankov 1969, p. 97;cited in Lindstedt 1985, p. 141) In such a narrative context, the ipf imperf can be used to present events in a more detached way (cf. the discussion of the historical pres in the next paragraph). As such, this usage does not have to be seen as a counterargument against EWT, but again shows that there is an intricate interaction between certain tense forms and aspect, since apparently the imperf tense form makes this use in Bulgarian possible. Furthermore, it raises the question of why Russian would preclude a more detached way of presenting past tense sequences of events, while Bulgarian does allow it. Clearly, past tense sequences of events are not by nature incompatible with the ipf aspect in the Eastern languages, as is also shown by a number of counterexamples from Russian that Dickey presents (Dickey 2000, p. 215), like (39): (39) Potom ona vdrug obratilas' pf k knjazju i, grozno naxmuriv pf brovi, pristal'no ego razgljadyvala ipf . 'Then she suddenly turned to the prince and, having frowned threateningly, examined him closely.' (Ru; Dostoevskij. Idiot; cited in Ivančev 1961, p. 43)

Present tense narration
In pres tense narration, events that happen one after the other are presented with a pres tense form. A typical example of pres tense narration is the historical pres, where events are presented as having occurred in the past, but are narrated using pres tense forms. Again we see a division here between the Eastern and the Western aspectual group. In Russian, the historical present requires the ipf aspect, in contrast to the past tense narration, which requires the pf. The difference between past tense narration and pres tense narration can be illustrated using the following Russian sentences from Barentsen (1998, p. 54): (40) Petr prišel pf , našel pf ključ i sprjatal pf ego v karman. (Ru) 'Peter arrived, found a key and hid it in his pocket.' (41) Petr prixodit ipf , naxodit ipf ključ i prjačet ipf ego v karman.
(Ru) 'Peter arrives, finds the key and hides it in his pocket.' Note that the pres tense is used even with predicates that usually do not express a durative situation (achievements) as in (41). This stresses the fact that the historical pres cannot be seen as an instance of a processual use of the pres tense (cf. ' ? Peter is coming, he is finding the key and. . .'). 45 Barentsen (1998, p. 54) discusses pres tense narration in Russian and suggests that the historical pres as such is characterized by a presentation of events in an isolating manner (Barentsen 1985, p. 223). The ipf is suitable for this, whereas the pf, which expresses the feature of sequential connection, is not (cf. Dickey 2000, p. 154). This is in accordance with the idea expressed by Dickey that in the East, the ipf expresses qualitative temporal indefiniteness, that is, the non-assignment of a situation to a single unique point in time.
Russian differs from languages of the Western Slavic group such as Czech, where both the ipf pres and the pf pres can be used as a historical pres and in similar contexts, like stage directions. This can be explained within EWT in the following way: the Western pf aspect does not express sequential connection (temporal definiteness) but totality. The feature totality, unlike the feature sequential connection, is compatible with an isolating manner of presenting the elements of a narrative chain of events, which is typical for the historical pres style of narration (cf. Stunová 1993, p. 175).
Even though Czech and other Western aspectual languages allow for the use of the pf pres in the case of the historical pres, Stunová (1993, p. 179) shows that the ipf is predominant in Czech as well in these contexts. According to Stunová (1993, p. 178, p. 190) the ipf is more typical for situations with a longer duration, whereas the pf is more typical for situations with a shorter duration or to indicate a sudden entry. This is in accordance with Dickey's point of view that the ipf in the West expresses quantitative temporal indefiniteness or the assignability of a situation to several points in time. This meaning excludes the use of the ipf with single (as opposed to iterated) situations without a clear durative phase. Esvan (2015, p. 214), however, shows that duration, or processuality, is not the only factor that triggers the ipf in Czech. According to Esvan, the use of ipf verbs irrespective of their duration may create a specific effect, which is absent in the case of a pf verb; cf. (42) and (43) According to Esvan (2015, p. 214) the use of the ipf in (43) has a different character than its perfective counterpart in (42). He calls this use of the pres tense the 'tabular present', which is a narrative style that is typically used to describe a moment in a person's life. In Esvan's view, this ipf tabular pres gives rise to an 'expressive effect of fragmentation'. In other words: the events are presented in an isolating manner, perhaps similar to the historical pres in Russian, which, according to Barentsen (1985, p. 223) presents the events in an isolating manner. The use of the pf forms in Czech, as in (42), creates an effect of smoother transition between the events than the use of ipf verbs. It should be noted, however, that not all informants agree on the interpretation of the ipf aspect as creating an effect of fragmentation; the ipf is sometimes also interpreted as expressing fluidly overlapping events, with no clear boundaries between the events (in medias res). Perhaps both interpretations can be brought back to the following essence: the events are not presented as they normally would be, namely as a connected chain of events; the use of the imperf shifts the focus away from the chain to the action itself and this can create both a fragmented effect, and an in medias res effect. No matter what the specific effect, this usage seems to need extra explanation within the EW-model. Note also that according to the native speakers of Slovene we asked, the ipf is ruled out in contexts like (43), where the verb refers to non-repeated single terminative events that are not clearly durative. This shows that there are differences within the Western aspectual group (as we will also see later in our discussion).

Evaluation of the explanation of narrative contexts within the EW-model
Aspectual behavior within narrative contexts can largely be explained within EWT, but the data are less straightforward and therefore harder to analyze and explain than in the case of habitual contexts. Generally speaking, one can conclude that in past tense narratives, there is no big difference between the Eastern and the Western group. Both in Russian and Czech the pf is preferred in such contexts. In Bulgarian, where more verb forms are available, the default choice for the pf is more easily overruled than in Russian. The main difference between Russian and Western languages such as Czech is that, whereas a language like Czech can sometimes use imperfectives, Russian mostly uses specialized ingressive pf prefixes (za-, po-) or pf ingressive verbs (stat') in similar instances. Dickey and Hutcheson (2003) and Dickey (2011Dickey ( , 2015 show that the very development of these specialized (po-) forms from the seventeenth century on points to a closer association between contexts of sequential connection and the pf in Russian and as such can be considered one of the most important developments in the Russian aspectual system, setting it apart from, among others, the Czech aspectual system. This is in accordance with the observation made by Ružička (1962, p. 316) that in older versions of Russian, the use of the ipf was more common in sequences of past events (see also Dickey 2007Dickey , 2011. 46 The data in the case of pres tense narratives are to some extent clearer than in past narrative contexts. Whereas Russian does not allow for the pf pres in such contexts, in the Western languages such as Czech it occurs relatively often. 47 This can be explained by pointing out the presence of the feature of sequential connection in the East, which is incompatible with the isolating presentation of the elements of the narrative chain of events in the historical pres. Note that the explanation of EWT is also in accordance with the fact that in older stages of Russian, the pf pres was also possible in the case of the historical pres (see Manning 1939). This, together with the observations by Ružička (1962, p. 316) on the use of ipf verbs in pres tense narration in older stages of Russian, accords well with the assumption that the specific meaning of sequential connection is a Russian innovation. However, the data strongly suggest that in some contexts the ipf in the West is acceptable with (single) non-durative events, and it is not immediately obvious that such cases have a specific slow-motion effect. This is not 46 This can also be illustrated using examples from the Bible. Note the following example from Matthew: (i) Togda d'javol ostavil pf Ego, a k Iisusu pristupili pf angely i služili ipf Emu.
'Then the devil left Him; and behold, angels came and began to minister to Him.' (Ru; Matthew 4:11, Slovo Zhizny, https://www.biblegateway.com/passage/?search=matthew+4%3A11&version=SZ) This Slovo Zhizny translation uses an ipf verb in a sequence of events, probably influenced by older versions of the Slavic Bible translation (NB: all Old Church Slavonic Gospel codices have the verb služiti here). It is interesting to see that the English translation uses an ingressive construction here and that the Russian Easy-to-Read Version (https://www.biblegateway.com/passage/?search=matthew+4%3A11&version= ERV-RU), which generally uses a more colloquial style, translates this verse with an ingressive construction as well (cf. also Dickey 2000, pp. 219-222): (ii) I togda d'javol ostavil pf Ego, i prišli pf k nemu Angely i stali pf Emu služit'. in full accordance with the definition of the Western ipf given by Dickey as 'assignability to more than one point in time'. We have also seen that the pf in the Western languages is easily triggered by sequences of events. Clearly, sequences provide an ultimate environment for the expression of total events. One can perhaps explain the use of the ipf with non-durative predicates as a way of preventing this sequencing effect, similar to the use of the reportative past in Bulgarian. This is in accordance with the effect of fragmentation one achieves by using the ipf in pres tense narration. In the same vein, the use of the pf in the Western languages is very similar to the use of the pf in the Eastern languages. In both groups the use of the pf has the typical aor function, used to narrate a chain of events, which can be attributed to the sequencing effect that the pf carries in these languages. This suggests that it is the feature of totality which explains the similarities between the different languages.
As a general remark regarding narrative contexts, we would like to point out that unlike pres tense habitual contexts, in which the choice of aspect is clearly dependent on the environment, EWT implies a categorical incompatibility of the context with one of the aspects in narratives in the Eastern group (ipf in past tense and pf in pres tense) (cf. the strict restriction on the ipf in past tense habitual contexts). However, it is not entirely clear how this incompatibility is connected to the proposed meanings of the pf and ipf aspect. There is a certain circularity in the argument that a historical pres style of narration is an inherently isolated manner of narration, and therefore not compatible with the Russian pf pres, or that the past tense sequences are incompatible with the isolating effect of the Russian ipf. However, we have shown that these connections between certain contexts and the pf or ipf aspect occur throughout the Slavic languages, in both the Eastern and the Western group. For example, Czech also prefers the pf in past tense narration and in habitual sequences. This can function as independent evidence of the influence of certain contexts on the choice of aspect. The differences in meaning of the aspects between the languages and groups can then explain why this attraction is stronger or weaker in some cases.

Introduction to the retrospective (perfect) use of past tense forms
We use the term 'retrospective use' for what could also be called the perf use of the past tense. 48 In most Slavic languages, except Bulgarian and Macedonian, 49 the Late Common Slavic compound perf tense form (a pres tense form of byti 'to be' and a past part), has become the general past tense. Since this form is now also used in narrative contexts, it is often no longer referred to as perfect, because, as Lindstedt (2000, p. 371) remarks: "When a perfect can be used as a narrative tense [. . .] it has ceased to be a perfect". Nevertheless, it is still possible to speak of the perf use of the past tense, just as it is possible to speak of the narrative use of the past tense. Moreover, it is necessary to make this distinction, since, as we will see, aspect usage also interacts with these different uses.
It is not easy to give a general description or even definition of the perf, or perf use, of the past tense in Slavic, since the forms referred to as perf differ in function between languages, 48 Note that we use the term 'retrospective' for perf use in a broad sense, including uses of the ipf aspect. This is different from perfektnoe značenie in Russian literature, which is generally used only for the typical resultative meaning of pf verbs. This terminological issue was the reason Barentsen (1992, p. 14) coined the term 'retrospective', which covers both pf and ipf perf usage as opposed to narrative usage. 49 And to a lesser extent (dialects of) BCS. but as a general starting point we may take Lindstedt's (2000, pp. 259-260) translation of Maslov's (1990) definition of the perf: [A]n aspecto-temporal form of the verb, expressing a present state as a result of a preceding action or change, and / or expressing a past action, event or state that is somehow important to the present and is considered from the present point of view, detached from other past facts. 50 Maslov's definition is interesting because it actually defines two types of perf, namely one where the result of a past action is present at the moment of speech and one where a past action is somehow relevant to the present, even if no specific result of that action is present at the moment of speech. Both these sides of the perf are present in Slavic past tense forms, and are strongly correlated with aspect use. As Lindstedt (1995, p. 99) puts it: " [i]n dialogues, the distinction between the Slavonic ipf and pf often parallels the distinction between the so-called experiential (or existential) and resultative meaning of the perf tense in such languages" [i.e In (44), we find an ipf past tense, even though the situation the predicate refers to ('receive the highest ordination') can be viewed as being fully completed (cf. example (2) given earlier). This use of the ipf to refer to total events in the past is named the 'general-factual use' of the ipf in the Russian linguistic tradition (see e.g. Padučeva 1996). 51 As we will discuss in Sect. 5.2, this use of the ipf aspect differs between the Eastern and the Western group. The main difference between the groups is found in terminative events with no process phase (achievements); here the Eastern group allows for ipf verbs, while this use is strongly restricted in the Western group.
In (45), the pf past refers to the resultative state of a completed event, which holds at a certain reference point, normally the moment of speech. This use has been described for Russian (e.g. Barentsen 1995Barentsen , p. 19, 1998, and Barentsen connects this usage to the defining characteristic of the pf aspect in Russian: sequential connection. It might therefore be expected that this use of the pf past tense to refer to the resultative state of a completed event is typical for the Eastern group as a whole, and less typical for the Western group, but as we will show in Sect. 5.3, the data indicate that this is not entirely the case. In Sects. 5.2 and 5.3, we will discuss both the pf and ipf aspect in retrospective contexts, and discuss and evaluate how they are analyzed within the EW-model. We start out by discussing the general factual use in Sect. 5.2 and then discuss the concrete factual use in Sect. 5.3 (see Table 3). 50 In the way we discuss the perfect instead of 'aspecto-temporal form of the verb' one should read 'aspecttemporal usage of the verb,' given the fact that e.g. Russian and Czech do not have a special perf form. 51 The term itself seems to have been introduced by Maslov (1959). Pf past tense for resultative state Yes (but often a construction with a past pass part is used) Yes in Ru (but in Bg often a construction with a past pass part is used) 52

General factual use and related uses
Before we discuss the general factual use of the ipf past tense, it is useful to point out that both in the Eastern and in the Western group, the ipf past tense is used for terminative situations if the situation is not completed, but ongoing, for example in typical processual contexts (both narrative and retrospective), cf. (46), (47): ( C z ) 'Snow was falling / I was falling (e.g. from a great height).' Similarly, in both groups, the ipf past tense can also be used to express a conscious attempt to complete an action with a number of verbs (see for example the overview of verbs given by Forsyth 1970, pp. 71-73). In this respect, Russian is similar to Czech and Slovene (see e.g. Galton 1976, p. 65). Some Czech and Slovene examples of this so-called conative use of the ipf are given below: (48) Dajal ipf sem mu denar, pa ga ni hotel vzeti.
(Cz) 54 'When the aircraft tried to return (lit. 'was returning') to the airport for the third time, it was flying low and losing altitude.' (50) "Tudi če bi jo v to prepričeval ipf , verjamem, da je ne bi prepričal pf ", je dodal njen brat Ivica. (Sn) 55 ' "Even if I had tried to persuade her, I wouldn't have persuaded her", added her brother Ivica.' In these contexts, the ipf is used to express the fact that the natural end point (telos) of the situation has not been reached. Note that the conative use cannot be equated with a processual use or with a presentation of the situation as being stretched out. The verb dajati 'give' in (48) does not imply that there was a relatively long attempt to give (as in 'I offered him something 52 In Russian, the pass part construction is also used to express a resultative state (see for example Knjazev 2007, e.g. Sect. 5.2). 53 http://bos.zrc-sazu.si/c/PL/neva.exe?name=pl&expression=kr=vb.%20impf.&hs=801 (February 2014). 54 http://www.uzpln.cz/pdf/incident_aRhUgv8d.pdf (February 2014). 55 http://www.24ur.com/sport/ostalo/ante-kostelic-janica-se-ne-bo-vrnila.html (February 2014). a couple of times'; 'I was trying to give him something'), but may apply to a context in which the giving occurred only once, without emphasizing any specific duration. 56

Fully completed terminative situations in the past
In retrospective contexts, the ipf past tense can also be used for fully completed terminative situations. Two examples from Russian are given below, see (51) and (52) Rassudova 1969, p. 37) In sentences like (51) the speaker uses the ipf because (s)he is mainly interested in whether the addressee has (ever) read War and Peace, and the question whether it was really finished or not is irrelevant (cf. Forsyth 1970, pp. 82-84). This use of the ipf is often called the 'existential general factual use' of the ipf (see e.g. Padučeva 1996, pp. 48-52). In (52), the speaker uses the ipf because (s)he wants to focus on the question of who made the dress, and the fact that the action is fully completed is taken as a given, or as irrelevant. Padučeva (1996, pp. 48-52) calls this use, where the focus is on particular circumstances of the action, such as the question of who performed it, or where the situation took place, the 'actional general factual use' of the ipf. The explanation of this usage type in Russian given by Barentsen (1998, pp. 53-55) is that the ipf signals the absence of a sequential connection with the resultative situation (Z in his model) and thus focuses on the event Y itself (Barentsen 1998, p. 54) (cf. also the explanation given by Dickey 2000, pp. 108-109, who focuses on the isolation from the context). In the words of Lindstedt (2000, p. 369), this type of usage is "a way of referring to a past situation without referring to a particular occasion, that is to say, it is characterized by nonspecific past time reference". The fact that the use of the ipf in Russian signals the absence of a sequential connection makes it possible to use the ipf past tense to refer to completed situations since they are presented as being fully detached from the surrounding context as well as the moment of speech. Note that since the ipf is the mirror image of the pf in the eastern languages, Dickey and Barentsen can argue that the occurrence of this use is correlated with the feature of sequential connection which is part of the meaning of the pf aspect. A complicating factor, however, is that this usage of the ipf is not confined to the eastern group. In Czech, for example, one would use the ipf in the same contexts as well (see also Kopečný 1962, pp. 53-54): (53) Četl jste ipf Vojnu a mír? Četl ipf .
'Who made this dress?' (Cz; Filip 1999, p. 186) In Slovene, a sentence like (53) can be translated with an ipf as well: 56 This seems to contradict the definition Dickey gives for the ipf aspect in the Western group, namely the assignability of a situation to several points in time. One could argue, however, that by emphasizing the noncompletion of the event, the preparatory phase is 'stretched out,' and as such presented as having some duration. Such an analysis implies, however, that the duration of a situation expressed by a predicate can only be measured relative to particular linguistic contexts, and cannot be determined objectively. We will say more about this issue later.
(55) Ste brali ipf Vojno in mir? 'Have you read 'War and Peace'?' (Sn; Derganc 2010, p. 76) Instances of the actional type like (54) are, however, more restricted. In Slovene, a sentence like Kdo je šival ipf to obleko; 'who made (lit. sewed) that dress', cannot refer to a completed single event and necessarily implies various people working on the dress at various occasions. 57 Dickey (2000, pp. 97-110) argues that there is a crucial difference between the Eastern and the Western group. When a terminative event without a process phase (or 'achievement' in Dickey's terms) is referred to, the Eastern group can still use an ipf verb or pf verb (with some difference in meaning), while languages in the Western group must use pf verbs in such contexts. Dickey illustrates this with the following examples, in which the general factual use of the ipf is possible with the achievement verb 'fall' in Russian, but not in Czech (or other Western languages): (56) Ja pomnju, v detstve odnaždy ja upal pf / padal ipf sėtogo dereva.
(Ru; Dickey 2000, p. 99) (57) Jako dítě jsem jednoho dne spadl pf / *padal ipf z toho stromu. 'I remember, as I a child I once fell from this tree.' (Cz; Dickey 2000, p. 101) 58 57 There is an intriguing factor for which we have no full explanation. Dickey (2000, pp. 117-118) claims that in general experience, in questions in the Western languages, an indication like 'from the beginning till the end' or 'fully' is only compatible with the pf. He provides the following example from Czech: (i) Přečětl pf / *Četl ipf jsi někdy vůbec tu celou knihu? 'Have you ever read this whole book?' This could be seen as a corroboration of the meaning of totality of the Western pf: when the event is explicitly presented as fully complete, the pf is the only possible option in Czech. This differs from Russian, which prefers the ipf in contexts like (i). Note, however, that this restriction also seems to be due to the specific construction with někdy (vůbec) 'ever', since the ipf is acceptable in the case of concrete experience questions: (ii) Četl ipf jsi to celé? 'Did you fully read that?' (http://www.motorkari.cz/forum-detail/?ft=166023&fid=63 (September 2014)) Interestingly, the situation in Slovene is different. According to the native speakers we asked, in Slovene the ipf is possible with the verb 'read' in general experience questions and in concrete experience questions, even if the completeness of the realization is made linguistically explicit, for example: (iii) Ste že kdaj brali ipf Božjo besedo od začetka do konca? 'Have you ever read the word of God from the beginning till the end?' (http://plus.iskreni.net/postna-akcija/320-prvi-postni-vikend-spodbuda-k-branju (September 2014)) (iv) A je sploh kdo to klikal in bral ipf do konca?
'Has anyone clicked on it and read it till the end?' (http://www.joker.si/mn3njalnik_oldy//index.php?s=ea44009130f3e22f6d668f8706217bc5 &showtopic=71115&st=0&p=1063290480&#entry1063290480 (September 2014)) In (iii) the ipf can be chosen to emphasize the importance of the book that was read and the reading of it. It is not clear to us why sentences like these are not acceptable in Czech, especially because in general, Czech more readily allows for the ipf in general factual type contexts than Slovene. More empirical research is necessary before any conclusions can be drawn. 58 In the original Czech translation in Dickey (2000, p. 99) the word jednou 'one time' is used as an equivalent for the Russian odnaždy. We are thankful to Hana Filip who pointed out that jednou is more like the Russian odin raz 'one time', which in Russian would probably trigger the pf aspect as well. However, the aspectual difference between Russian and Czech as described by Dickey is still present when the Czech adverb jednoho dne 'one day' is used as an equivalent for odnaždy.
Dickey explains the difference between the Russian and Czech aspect usage, or the difference between the Eastern and the Western group, by pointing at the fact that the definition of the Czech ipf is 'assignability of a situation to more than one point in time' (Dickey 2000, pp. 124-125). This is incompatible with predicates without a process phase (i.e. terminative events that have an momentaneous character). Put differently, in (53)-(55) the ipf is used in accordance with the above given definition, because of the fact that the event has a process phase, which outweighs the fact that the events can be understood as fully completed. The choice of aspect in such cases depends on the question which phase of the event is in focus: the result or the process phase (Dickey 2000, p. 107). In the Eastern group, the definition of the ipf aspect is 'non-assignment of a situation to a unique location in time relative to other states of affairs', which does not rule out the use of an ipf verb to refer to events without a process phase; the use of the ipf aspect in such predicates presents the event as a single point in time detached from other past events. It should be noted that even though it is correct that in the West the ipf past tense easily triggers a processual (or habitual) interpretation, in some contexts in Czech it is in fact possible to use the ipf past tense even if it has no clear process phase. As we will show, this is especially the case for the actional type of the general factual use.

Some unexpected cases of ipf general factual in the Western languages
We will first discuss the use of the general factual in the West in more detail by starting with the so-called existential use. In constructions where the focus is on the question whether the action has (ever) been realized (i.e. the existential type), the use of the ipf in the Western languages is as outlined above: it is possible with terminative events with a process phase (accomplishments) as in (53), but not with terminative events without a process phase (achievements). As already mentioned, Dickey (2000, p. 107) argues that this is because the presence of a process-phase makes it possible to use an ipf verb that shifts the focus away from the result of the action and focuses on the action in development. This differs from instances with achievements in which there is no development to focus on. It should be noted, however, that even though most of the predicates that allow for the existential ipf can have an accomplishment reading, there is no indication in these examples that the development of the event or its durative character plays a crucial role in the choice of aspect. This can be illustrated by the following sentences in which the ipf has been chosen (even if the pf is not ruled out either), cf. Filip (1999, p. 186), who also provides an example with platit 'pay': (58) Už jste někdy platili ipf pokutu. Jaká byla nejvyšší?
(Cz) 'Have you every paid (lit. given) a bribe?' Even though platit pokuta / dávat úplatek 'paying a fine / bribe' or kupovat po internetu 'buying something on the Internet' can in fact be seen as something that takes time (cf. 'I am paying the fine right now'), in these sentences there are no clear additional clues that this process (i.e. the duration) is an important element, or that the durative character triggers the use of the ipf instead of the pf, which is something the definition of the Western ipf as 'assignability of the situation to several points in time' seems to implicate. Note that according to the native speakers we asked, such cases also do not have a conative meaning (e.g. 'try to buy', 'try to pay', 'try to give') and refer to complete single events. Instead, the ipf is triggered in such cases because it is not the result that is relevant here, but rather the action itself. However, platit pokuta or kupovat po internetu, or dávat úplatek can be seen as presupposing a particular development, which requires different stages, no matter how minimal. It could be argued that this is enough for the ipf to be used. Also note that in the case of (60), which expresses a relatively momentaneous event, data from the Internet suggest that the pf is much more common than the ipf. In this construction, however, it is not acceptable at all to use the ipf for even clearer instantaneous events without clear internal structure: (61) Už jste někdy dal pf (*dával ipf ) gól?
(Cz) 'Have you ever scored (lit. given) a goal?' However, when the focus is on the circumstances of the realization of the situation, i.e. the actional general factual type, the requirement of duration to use an ipf verb is less strict. This can be illustrated by comparing (61) to (62) with the ipf dávat 'give', which does not have a clear durative character or process phase to emphasize: (62) Kdo dával ipf ten gól?
In (62), we find the verb dávat 'give' in the construction 'kdo + V past.ipf + X'. In this construction, the predicate expresses a fully completed event that has no clear process phase (or to put it differently, scoring a goal can be seen as an achievement). Note, however, that in the same construction we cannot use a verb like padat 'fall', see (63): (63) Kdo spadl pf / *padal ipf z toho stromu?
(Cz) 'Who has fallen from that tree?' The difference in use between these verbs cannot be attributed to the actual duration of the event in reality, but rather to the different meanings of the verbs. 61 A verb like dávat gól 'score a goal' more clearly presupposes various stages of development (transfer of possession from one participant to another). This differs from a predicate like 'fall from a tree', which does not clearly presuppose various stages of development, but instead expresses a simple movement from one place to another (both of which may or may not be linguistically expressed). Because of its more complex internal structure, it is possible to focus on this internal structure even if the event is fully completed and no process phase is present. Note, however, that the restriction can be overruled, as is shown in the following example (even though this sentence contains a number of typos, the use of the ipf padal is fully acceptable according to the native speakers we asked): 62 61 According to the test advocated by Vendler (1957) for English, both 'fall from a tree' and 'score a goal' must be seen as achievements, since they can occur with an 'at + time expression'-phrase, and do not combine with For how long?, at least in English: He fell from the tree at 15:00; He scored at 15:00 sharp; ?For how long did he fall from the tree; ?For how long did he score a goal. Also note that both verbs cannot be compared to predicates like reach the top, which have a preparatory phase and an instantaneous culmination point, presupposing the transition of one state to the other (cf. Moens and Steedman 1988). 62 Dickey (2000, p. 100) remarks that "in the Western half Slavic territory, impv [ipf, E. F., J. K.] achievement verbs are much less acceptable [italics our] in general factual contexts-the pv [pf, E. F., J. K.] is used instead".
In the examples he provides, however, the pf is marked as ungrammatical (*), and not as (much) less acceptable.
(Cz) 63 'Then there was one incident in which I voluntarily fell on the ground, in order not to take a child that had decided to roller skate from one side of the road to the other without looking around, down with me. The kid crashed into the side of my front wheel and ended up on the ground, but the wheel stopped turning and I fell to the ground. Nothing happened to the child, nor to me; I was just cursed at by the women who came for the child and had not seen what had happened (a guy who had seen it did not say anything). Their main concern was whether the child had fallen on its head, while it was me who had fallen from a height onto the asphalt.' One could of course argue that such contexts can only be interpreted as durative, that is, as expressing 'more points in time.' For some of the predicates in this example this is indeed the case. In (64), our native speaker informant does indeed interpret the falling in a durative, slow motion like way, creating a dramatic effect for the first use of padat ('I found myself falling to the ground'). This is, however, not clearly the case for the second instance. This use of padat is triggered by the contrastive context ('. . .while I was the one who had fallen'), which shifts the focus away from the possible result of the event to the actor, making it a typical example of the actional general factual use of the ipf. 64 In our view, arguing that the situation is conceptualized as durative here would, without additional explanation, stretch the concept of durativity in an undesirable way. It would be more correct to say that in Czech, the actional use is always possible with terminative events that have a process phase and with some terminative events without process phases, in some cases only if particular conditions are met. 65 Note, furthermore, that with some ipf predicates the use of the actional general factual is ruled out, even in a contrastive context. For example, we have not found any examples with the verb ztrácet ipf 'lose', referring to a non-intentional or non-controllable event, whereas the corresponding Russian verb terjat' ipf 'lose' can freely be used in actional general Note that these examples are all instances of the existential use of the general factual. In Dickey (2015), the issue is treated from a somewhat different angle, and Dickey argues that 'Czech allows the non-resultative IGF [ipf general factual, E. F., J. K.], the actional impf and the existential IGF with accomplishments; however, in contrast to Russian it does not allow the existential IGF with achievement verbs (. . .). The concrete IGF is also very uncommon in Czech, as is the two-way IGF (. . .)'. Even though Dickey does not explicitly say that the actional use is possible both with accomplisment verbs and achievement verbs in Czech, one could draw this conclusion a contrario because he does mention the restriction on ipf achievement verbs for the existential general factual. On the other hand, the example he gives of the actional general factual is A kde jsi kupoval ipf ty roury?, 'And where did you buy the pipes?' With respect to this example (and other examples) he remarks that 'Western languages primarily allow the IGF in cases in which a process component can be identified'. This suggests that in Dickey's view, the actional type is only possible with accomplishment predicates only and not with achievements. It should be emphasized here that the differentiation between accomplishments and achievements based on the criteria used by Dickey (2000, pp. 13-14) is not always easy, as our discussion shows. 63 http://www.bike-forum.cz/forum/pes-nepritel-bikeru/995934/forum.html (February 2014). factual sentences like Kto terjal ključi? 'Who lost his keys?' In the following Czech example, (65), the ipf is therefore excluded, but not in the Russian example (66) In contrast to padal 'fell', in (64) a contrastive context does not facilitate the use of the ipf with this predicate, as is illustrated in (67): (67) Myslel jsem si, že to bude on, kdo bude tak blbej a ztratí pěněženku, ale byl jsem to samozřejmě já, kdo ztratil pf (*ztrácel ipf ) jeho peněženku.
(Cz) 'I thought that he would be so stupid to lose his wallet, but of course I was the one who lost his wallet.' The difference between padat na asfalt 'falling on the asphalt' and ztratit klíče / peněženku 'losing one's key / wallet' cannot easily be explained in terms of the duration of the event in reality. Instead, we think the difference may have to do with the fact that in the case of padat the subject is conscious of the situation, which makes it possible to focus on the situation itself, rather than on the result of the situation, even if the speaker does not want to emphasize the duration of the situation. This differs from ztratit peněženku, which is both a non-intentional and unconscious event. Note that Dickey (2000, pp. 101-102) considers volitionality not to be of primary relevance for aspect choice. In his opinion, there is an indirect correspondence, mediated by the fact that many achievements are non-volitional.
To conclude, the definition of the Western ipf in EWT ('several points in time') is not enough to predict or even fully explain the use of the ipf general factual in Czech. This is mainly because it is not entirely clear what counts as several points in time and what does not. There does not seem to be an independent test to determine whether or not verbs have enough of a process phase to refer to a fully completed terminative event by means of an ipf verb. Furthermore, to what extent a verb can be used in such contexts depends not only on the specific verbal semantics, but also on the construction in which it is used (for example existential general factual or actional general factual), and on other contextual clues, which facilitate a shift in focus from the resultative phase to the situation itself, or to participants associated with this situation (for example contrastive contexts with some ipf verbs in the case of the actional general factual). We will say more about this issue in Sect. 5.4.

Annulment, or 'Two-way action'
Another use which is related to the general factual use of the ipf, is the so-called annulment use (see Rassudova 1968Rassudova , 1969Rassudova , 1984Forsyth 1970). In Russian, it is possible to use an ipf past tense in situations that have been fully completed if the result has been annulled, cancelled or reversed at a specific reference point, for example the moment of speech. Another Unlike Czech, where a contrastive context sometimes facilitates the use of the imperfective, like in (64), in Slovene a contrastive context does never facilitate the use of the ipf (ipf padati, instead of pf pasti). As such, the data suggest also that Czech more easily allows for the ipf in the case of fully complete terminative events than Slovene. This is in accordance with the conclusions provided by Fortuin and Pluimgraaff (2015) with respect to the use of aspect in the imp.
term for this use is 'two-way action ipf' (see e.g. Forsyth 1970, p. 78), because the use of the ipf refers to the fact that the result of a previous action is reversed or annulled: Both Barentsen (1995Barentsen ( , p. 20, 1998 and Dickey (2000, pp. 110-116) argue that the ipf is used in these situations because the result of the event is no longer present at the moment of utterance. The ipf aspect creates a gap, as it were, between the moment of speech (which is the natural contrasting situation in retrospective contexts) and the situation, which is presented by the ipf verb. In Barentsen's version of the theory, the ipf in Russian can be used if the terminative situation is presented as complete, but not sequentially connected to another situation through the resulting situation Z; in cases in which the reversed action situation Z no longer exists, which makes a connection to the moment of speech via Z impossible. However, in the Western group we can find a limited set of verbs that are used in a similar fashion, again seemingly breaking the rule that ipf verbs always assign the situation to more than one point in time and the pf aspect always presents an event in its totality. These ipf forms behave just like a Russian ipf verb in a terminative completed situation without a preceding or subsequent contrasting situation, at least in questions: (71) Otvíral ipf jsi okno? (Cz; Dickey 2000, p. 112) Dickey refers to Galton (1976) and Bareš (1956), who point out that Czech allows for both pf and ipf aspect in a context where the window has been closed again. In Russian, however, such contexts require the ipf. Dickey (2000, p. 115) argues that the fact that the Western pf does not revolve around the concept of temporal definiteness, can explain the use of the pf in such contexts. At the same time the fact that the default interpretation (but not the meaning) of the pf in the Western languages is the presence of a result, is responsible for the use of the imperf to deny the presence of the result, in verbs that denote clearly reversible actions, like 'open'. It should be noted that the annulment use is indeed much more wide-spread in Russian than in Czech or Slovene. Dickey (2015) reexamines example (71) and comes to the conclusion that, even though informants accept the usage of otvírat in this context, real examples are almost impossible to find and it never occurs with verbs of motion such as 'come' or 'return' (see Bareš 1956, p. 577), which is the typical context for such usage in the Eastern group. A Czech sentence like the following is therefore always interpreted as indicating a process: (72) Letoun se vracel ipf .
( C z ) 'The plane was returning.' (Cannot mean: The plane returned (but has now left again).) The two-way use also seems absent, or at least rather restricted, with verbs like 'give' or 'take' in the Western languages.
'He is very tired, he hardly moves his feet.' (Ru; Barentsen 1998, p. 51) For a better insight into the typical perf function of the pf past, one could translate this sentence as follows: 'He has become very tired. . .'. Hence, the verb expresses both a limit that has been reached in the past and the result of that event at the moment of speech. An example in which the point of reference is not the point of speech comes from Lermontov's Geroj našego vremeni, in which it becomes clear from the context that the reference point has to be a past moment: (78) Lošadi izmučilis' pf , my prodrogli pf . 'The horses were exhausted; we were chilled.' (Ru; M. Ju. Lermontov; cited after Barentsen 1998, p. 51) The interesting difference with the narrative use of pf verbs is that in this case the events expressed by the verbs are not interpreted as linked to each other, but both resultative states are linked to one and the same reference point and even the events themselves could be viewed as occurring simultaneously. Contrary to what ipf verbs express when used retrospectively, namely the denial of a direct link from the event to the moment of speech, pf verbs in this context express that there is such a link: the result of the past event is present at the reference point. This emphasis on the resultative state is so strong in retrospective contexts in Russian that a pf past tense verb can be used to express simultaneity of the resultative state with another state of event present at the reference point, as in example (77), where the resultative state of ustal is simultaneous to the action that is referred to as occurring at the moment of speech with a pres tense form peredvigaet (Barentsen 1998, p. 51).
Although at first sight the use of pf past forms in the perfect function that we saw in the Russian examples (77) and (78) might not stand out as the most compelling evidence for the existence of sequential connection as a unique feature of the pf aspect in the Eastern languages, it is striking that in other Slavic languages, except for the North Slavic languages, the standard way of expressing that one is tired, is with a construction of a pass part and a copula. Consider for example the following translations from English, where only Russian and Ukrainian employ a pf past tense, while all other languages (including Bulgarian!) use a part construction: 67 (79) "I'm sorry," Langdon said, "but I'm very The difference between the part construction and the pf past tense seems to lie in the fact that the pf past tense clearly links the action itself to the moment of speech and as such contrasts these two situations, in other words: the event itself is inextricably bound to the resultative situation. The part construction does not contrast the moment of speech with a prior event, but only refers to the state of the person at the moment of speech. It should be noted, however, that similar constructions with the pf past tense do occur in other Slavic languages as well.
In Macedonian and Bulgarian, for example in (90), a pf aor could easily be used to translate the Russian examples (1) (repeated in (45)): (90) Jako se izmoriv pf.aor . 68 (Mc) 'I am (lit. have become) very tired.' When translated like this, the interpretation is more like the Russian examples in which the event is linked to the speech moment. Furthermore, although in Czech the standard translation of (79) is construed with a past part construction, it is also possible to use a pf past tense: (91) Unavil pf jsem se.
(Cz) 69 'I got tired.' This means that, while in the examples above the part constructions seems to be the default choice in most Slavic languages that are not part of the Eastern group, and even in Bulgarian as a (peripheral) member of the Eastern group, the usage of the pf past in retrospective language is certainly not absent in those languages and leads to interpretations similar to interpretations of the pf past in the Eastern group. A more comprehensive comparison between the various modern Slavic languages in this particular context is necessary to be able to draw more definite conclusions.

Evaluation of the explanation of retrospective contexts within EWT
The data from retrospective contexts largely corroborate EWT, but also make clear that the theory can still be further refined to account for all the data. It can explain why in Russian the ipf past tense for completed terminative events is widespread in various contexts (general factual, annulment) with different kinds of predicates (with or without process phase), whereas in the West this use is much more restricted, with differences between types of use, different types of predicates and also differences between the Western languages. However, even though it is clear that the Eastern group uses the ipf much more extensively in this context, it is difficult to account for this difference by only pointing out a difference in meaning between the ipf aspect in both groups. Dickey (2000, p. 115) is also aware of the complications that arise with the data, and argues with respect to the occurrence of (71) (use of the ipf in the case of annulment in Czech) that: [o]f course, one could always posit an associated component of qualitative temporal indefiniteness within the network for the western impv. Given the conceptual proximity of the two concepts, this is not implausible. Moreover, if the pv [pf, E. F., J. K.] in the respective languages represents a radial category, then there seems to be no reason for not assuming the impv [ipf, E. F., J. K.] is a radial category as well.
Even though Dickey (2015) rightly points out the marginal status of sentences like (71), we have also pointed out other instances that are not fully in accordance with the definition of the ipf in the West by Dickey, for example (62) or (64). Following Dickey' (2000) line of thought, who advocates a polysemous account of the ipf with a central meaning, it could be argued that such instances are peripheral, rather than marginal, in the sense that they cannot be seen as prototypical instances of the meaning of the ipf ('assignability to more than one point in time'). Instead, it seems, such uses share some features with the prototypical use of the ipf in the West, but not all. Nevertheless, the ipf conceptualization is more optimal in such cases than the perfective conceptualization, which explains why the ipf is chosen. Also note that we pointed at other contexts in which the use of the ipf does not fully seem to fit the definition of the Western ipf as 'assignability to more than one point in time,' namely the use of nondurative imperfectives in sequences of events in narration (e.g. vracel in example (34)). If we follow a polysemous account, more analysis is necessary of exactly what features of the prototypical ipf meaning are 'highlighted' and what features are 'backgrounded' in such cases, and why the chosen ipf conceptualization is more suitable than the perfective conceptualization. Such an analysis should take the interaction between the meaning of constructions into account, and the predicates that are used in that construction. Different constructions provide different construals of a scene or situation, which may favor the use of the ipf over the pf, depending on the meaning of the predicate (cf. the notion of fusion of grammatical constructions with verbs as discussed by Goldberg 1995). As we have shown, constructions may more or less easily trigger the ipf in the case of a fully completed terminative event. This was already illustrated with respect to the difference between the existential and the actional general factual in Czech. As we have shown, ipf verbs referring to terminative events that lack a clear process phase are used much more easily in an actional construction than in an existential construction. The difference between the two constructions could be explained in the following way: the actional general factual refers to a concrete occurrence of an event in the past, while the existential general factual is a more general statement (or often a question) regarding the occurrence of some non-specific event at a non-specific point in time. Because the event is more concrete in the case of the actional use, it is probably easier to focus on the action (event) itself, and regard the totality of the event as a given. 70 An example was the ipf verb dat 'give' in Czech, which is not allowed in the existential general factual as in (61), whereas it is in the actional general factual as in (62). Another illustration of the importance of the way the predicate interacts with the construction is the difference between the actional general factual, which does not facilitate the use of the ipf ztrácet 'lose' to refer to a complete event as in (65). It is, however, possible to use the same ipf verb in Czech (though not Slovene) to refer to a complete terminative event in the pres tense like the following, probably because the pres tense is less clearly associated with the idea of completion: Again, this shows the importance of taking both the construction (in this case the pres tense 'descriptive' construction), and the meaning of the predicate (in this case 'lose his wallet') into account in the analysis and explanation of aspect, especially for the Western languages. Another factor in the use of the ipf past tense in the case of fully complete events appears to be the possibility of employing special verb forms. In case of the general factual the use of perf forms seems to lead to different aspectual behavior in those Slavic languages in which the proto-Slavic perf form has not become a general past tense, like Bulgarian. 71 Again, the data suggest that a more fine-grained theory is necessary to account for these data.
If we look at the use of the pf past in retrospective use in examples with 'be(come) tired', the data seem to indicate that resultative function is more wide-spread in Russian than in the Western languages. This could be explained in terms of the feature of sequential connection, which is part of the Eastern pf. However, languages of the Western group do employ the pf past in the same vein. More generally, both Western languages and Eastern languages other than Russian use alternative part constructions more in this context. It could be argued that the relative lack of the part constructions in Russian (at least in the contexts given here) could be connected to the feature sequential connection. In languages where the pf past tense in retrospective contexts does not signal 'sequentiality', the past part construction is the most suitable expression to signal result, whereas in Russian the idea of result is an inherent part of the meaning of the pf past tense in such contexts.

Conclusion and further remarks
The EWT of Slavic aspect is currently the only theory that gives a typology of Slavic aspect. The basic idea of the theory is that there are two main aspectual groups in Slavic (an Eastern group and a Western group) and that systematic differences in aspectual behavior between these groups are due to a difference in meaning between the pf in the Eastern and the Western group, and, connected to that, a difference in meaning of the ipf between the eastern and western group. We have argued that the theory is able to explain the entirety of the data in an elegant and convincing way. The clearest differences between the East and West that we discussed in this paper and that the theory is able to explain are summarized below: -In the East, but not in the West, only the ipf aspect can be used in the case of past habitual contexts. In the West, both the pf and ipf are possible; -in the East, but not in the West, only the ipf pres tense can be used in the case of the historical pres. In the West, both the ipf and the pf are possible; -in the East, but not in the West, a past tense narrative context requires the use of the pf in sequences of events. In the West, the use of ipf in such past tense narrative contexts is more common; -in the East, but not in the West, the ipf past tense can be used with single complete terminative events without a process phase. In the West, the ipf past tense is only possible with terminative events that have a process phase.
There are a number of issues that we will mention here that we think are important for the further development of the theory.
The role of the surrounding context of the perfective event in the case of sequentiality or temporal definiteness According to EWT, the pf in the Eastern group has a semantic feature, which is absent in the Western group, namely sequential connection (in Barentsen's version) or temporal definiteness (in Dickey's version). The theory shows that the surrounding context to which the event can be sequentially connected can be a linguistically expressed situation in the pf (or sometimes an ipf), but also a referential point such as the moment of speech or another vantage point. Whether or not the event can be sequentially connected is inherently related to the construction in which the verb is used. To give an example, in the Eastern languages such as Russian a habitual context makes it impossible to anchor the event in past tense contexts, with the exception of subordinate clauses. However, sequential connection is possible in the case of some pf pres tense habituals. Further research could focus even more on the question of what the prerequisites are for the surroundings of a terminative event to allow sequential connection (or in Dickey's terms, for the terminative event to be uniquely locatable, and contiguous in time to qualitatively different situations), and how the meaning of aspect interacts with the type of predicates or constructions in which the predicate is used. This could perhaps lead to an even better understanding of the strength of the factors, which either facilitate or block sequential connection. Furthermore, such an analysis could perhaps also explain contexts in which Bulgarian, due to its different tense forms, behaves differently from Russian. In our view, such a development of the theory could possibly integrate the feature of sequential connection proposed by Barentsen, as a feature of the pf, to the feature of temporal definiteness proposed by Dickey, as a feature of the context in which the pf is used.
Further explanation of data within the Western group that are not in full accordance with the theory According to EWT (specifically Dickey's version), it is possible to use an ipf in the West in the case of a fully complete (terminative) event as long as this event can be interpreted as occupying several points in time. This suggests that in order to use the ipf in the case of a fully complete (terminative) event, the event must have a durative character or a process phase, which makes it possible to conceptualize it as occupying several points in time. Even though the data discussed by us largely corroborate this hypothesis, there seem to be exceptions to this rule in Czech, and it is not fully clear whether we should understand all these exceptions in terms of a durative presentation such as a 'slow motion effect'. In the latter case, a predicate that is normally interpreted as having no clear process phase, is 'coerced' into a durative predicate because of the use of the ipf. The possible exceptions that cannot be explained straightforwardly in terms of 'slow motion' are the tabular present (example (43)), the actional use of the general factual with dávat (example (62)), and the actional general factual use in a contrastive context with padat (example (64)). Since 'slow motion' effects or 'stretching out' effects are difficult to measure it cannot be disproven that these phenomena play a part here. Also note that native speakers' answers seem to differ with respect to the question whether such effects are present and whether such examples are acceptable. Importantly, however, there seems to be a logic behind these exceptions, for example the full restriction on the existential general factual of the ipf with achievements, and the weaker restrictions on similar verbs in the case of the actional type of the general factual. This logic could be taken as evidence that even though duration (or habituality) is the most important factor, which enables the use of the ipf to refer to fully complete (terminative) events, there are also other factors, which can shift the focus away from the resultative phase Z. Such factors are in fact reminiscent of the same factors we encounter in Russian, for example the focus on the agent of the action. If this analysis is correct, we have to reformulate the definition of the ipf in the West so that it may allow for these uses. Since the Western pf expresses totality, the ipf can, in our view, be seen as the negation of totality, namely non-totality. There are a number of factors, which can trigger this definition: -Non-complete (terminative or non-terminative) events; -fully complete events where the focus is not on the resultative phase and which are therefore not presented as a totality because: • the event is presented as habitual and the habituality is construed on the macro-level, as such blurring the boundaries between the different repeated events (cf. Dickey's description in terms of 'several points in time'); • the durative character of the event (i.e. the event occupying several points in time) makes it possible to focus on the event itself or the circumstances associated with the event; • other factors make it possible to shift the focus from the resultative phase to the action itself (or circumstances associated with the action) even if the event has no clear process phase (i.e. even if the event is not clearly durative).
Further research could focus on such exceptional contexts, and determine whether and how they can be accounted for within the theory. As we have shown, this implies a very detailed analysis of the interaction between the meanings of constructions, which provide a particular construal of a scene or situation, and rich verb semantics. As we have pointed out in a number of cases, the analysis should concentrate on the different factors, which can trigger the ipf, and the relative strength among these factors. In doing so, it is also important to look at the possible differences between specific Western languages, since the data suggest, for example, that Czech allows more easily for the ipf general factual use than Slovene (cf. Fortuin and Pluimgraaff 2015, who argue that Slovene is more restricted in the use of the ipf imp with single non-repeated terminative events than Czech). The existence of such exceptional uses of the ipf in the West and the differences between the Western languages is to some extent reminiscent of the concept of transitional languages as proposed by Dickey. Since transitional languages sometimes behave like Western languages and sometimes like Eastern languages, one would expect that their behavior is not random, but systematic. Further research should more explicitly discuss this dimension of the theory.

Different meaning, same aspectual use
At some points in our discussion we have pointed out that particular aspectual behavior can be explained by focusing on the differences in the meaning of aspect of a particular aspectual group, whereas we frequently find similarities as well. A case in point are narrative sequences of events which seem to favor the pf both in the East and the West. Other examples include the use of the ipf aspect in actional general factual contexts and the use of the pf past in restrospective contexts. Even though these uses can certainly not be seen as a falsification of the theory, this topic merits more discussion and explanation.