3.2. Viewpoint-Aspect Distinctions
Like all Slavic languages, BG employs morphological means to encode the category of viewpoint aspect (see
Section 2.1). Thus, for instance, the simplex verb in (15) is imperfective. Its perfective counterpart with (nearly) the same lexical meaning is the prefixed verb in (16a). Unlike (16a) where the prefix seems to be semantically vacuous,
19 other perfectivizing prefixes obviously contribute meaning. The result is a new lexeme like (16b). Be that as it may, prefixation gives rise to perfective stems as a rule.
(15) | pìšă | pìšeš | (imperfective) |
| write.prs.1sg | write.prs.2sg | |
| ‘write’ |
(16) | a. | na-pìšă | na-pìšeš | (perfective) |
| | on-write.prs.1sg | on-write.prs.2sg | |
| | ‘write [down]’ |
| b. | pre-pìšă | pre-pìšeš | |
| | over-write.prs.1sg | over-write.prs.2sg | |
| | ‘write [over] (i.e. copy)’ |
Irrespective of whether the prefixed stem differs in meaning from the simplex base, it can further be regularly transformed into a so-called
secondary imperfective by adding the productive suffix
-va-.
20 All resulting forms belong to conjugational class III (see
Section 3.1). Examples are given in (17) (for more details, see
Karagjosova, 2024b;
Manova, 2007).
(17) | a. | napìs-va-m | napìs-va-š | (sec. imperfective) |
| | on.write-ipfv-prs.1sg | on.write-ipfv-prs.2sg | |
| | ‘write [down]’ |
| b. | prepìs-va-m | prepìs-va-š | |
| | over.write-ipfv-prs.1sg | over.write-ipfv-prs.2sg | |
| | ‘write [over] (i.e. copy)’ |
Secondary imperfectives with semantically vacuous prefixes are pragmatically marked, in that their range of use is restricted as compared to those with ‘meaningful’ prefixes.
The examples in (15) to (17) illustrate what
Manova (
2002,
2007) calls
aspectual triple opposition.
21,22 In addition, Manova argues that the members of that triple are regularly derived from each other, as shown in (18).
(18) | imperfective | → | perfective | → | secondary imperfective |
| pìšeš | | na-pìšeš | | napìs-va-š |
In template (14), aspectual prefixes occupy the
pref-slot. However, it must be noted that the prefixes themselves are lexical rather than grammatical elements (which applies as well to seemingly vacuous prefixes; see note 19). Still, by turning atelic into telic predicates, the grammatical impact is that the resulting telic verbs become perfective in the first place. To capture this,
Biskup (
2017) makes the following claim:
“I take the position that prefixes […] contribute telicity, in addition to perfectivity. I analyze the connection between telicity and perfectivity in the way that the prefix (an incorporated preposition) introduces a causal relation between the verbal part and the prepositional part of the prefixed predicate and in addition has a perfective property […]. The incorporated preposition bears a Tense-feature with the value [perfective] and values the Tense-feature of the aspectual head […]”.
Omitting details, this means that prefixes cause and reflect perfectivity without being aspect markers sensu stricto. By contrast, the imperfectivizing suffix
-va- is such a marker and should therefore occupy the
asp-slot. Thus, I analyze the secondary imperfective
prepìsvame as in (19). Note that, according to (18), the
-va-suffix must somehow overwrite/neutralize the perfectivity initially associated with the prefix. I take it that this neutralization affects only the formal
pfv-feature, which is responsible for the perfective interpretation of prefixed forms. When this feature becomes annulled by
-va-, there will also be no perfectivity. Instead, the
ipfv-feature introduced by
-va- causes the imperfective interpretation.
(19) | pref | – | base | – | asp | – | tns | – | agr |
| pre | | pìs | | va | | | | me |
| pfv | | ‘write’ | | ipf | | (prs) | |
1pl |
Note furthermore that round brackets around a grammatical feature denote that this feature lacks formal expression, so as to be interpreted by default. Thus, in (19), the absence of a dedicated tense suffix gives rise to the default value
prs. Moreover, I claim that if there is an empty default slot as just described, the next suffix to its right adopts the function of realizing the relevant default feature in addition to its own feature(s). For (19), this means that the agreement suffix
-me assumes the additional function of a tense marker; see (20).
(20) | pref | – | base | – | asp | – | tns | – | agr |
| pre | | pìs | | va | | | | me |
| pfv | | ‘write’ | | ipf | | | | prs.1pl |
Having thus determined the place of aspect markers in BG, I address individual tense form(ation)s in the subsequent sections.
3.3. Present Tense
A comparison of these verb forms allows us to isolate the present-tense inflections in
Table 6.
23This shows that 2sg, 3sg, and 2pl share the same inflectional suffixes across all conjugations, while class III has individual variants in 1sg, 1pl, and 3pl.
Table 5 shows that verbs from classes I and II employ the thematic vowels
-e- and
-i-, respectively, to build their present-tense forms. In my analysis, thematic vowels occupy the
asp-slot, so it is likely that they encode some aspectual meaning. My claim is that they reflect the openness of the assertion-time (AT) interval, as proposed in
Section 2.2.
In
Section 3.1, I decided to follow
Scatton’s (
1984) approach, saying that BG verbs have one single underlying stem.
Table 7 shows what this stem looks like for each of the illustrative verbs in
Table 5.
24To show how these theoretical ingredients interact to yield the surface forms in
Table 5, I analyze present-tense forms from classes I and II in (21) and (22), respectively.
25 The feature
open reads ‘open AT-interval’.
(21) | a. | base | – | asp | – | tns | – | agr | (class I.1) |
| | /čet/ | | /e/ | | | | /š/ | |
| | čet | | è | | | | š | |
| | ‘read’ | | open | | | | prs.2sg | |
| b. | base | – | asp | – | tns | – | agr | (class I.2) |
| | /pìjna/ | | /e/ | | | | /ă/ | |
| | pìjn | | | | | | ă | |
| | ‘sip’ | | (open) | | | | prs.1sg | |
| c. | base | – | asp | – | tns | – | agr | (class I.3) |
| | /igrà/ | | /e/ | | | | | |
| | igrà | | e | | | | | |
| | ‘play’ | | open | | | |
(prs.3sg)
| |
| d. | base | – | asp | – | tns | – | agr | (class I.3) |
| | /làja/ | | /e/ | | | | /ăt/ | |
| | làj | | | | | | ăt | |
| | ‘bark’ | | (open) | | | | prs.3pl | |
(22) | a. | base | – | asp | – | tns | – | agr | (class II.1) |
| | /vis/ | | /i/ | | | | /š/ | |
| | vis | | ì | | | | š | |
| | ‘hang’ | | open | | | | prs.2sg | |
| b. | base | – | asp | – | tns | – | agr | (class II.2) |
| | /stroì/ | | /i/ | | | | /ă/ | |
| | stroj | | | | | | ă | |
| | ‘build’ | | (open) | | | | prs.1sg | |
| c. | base | – | asp | – | tns | – | agr | (class II.3) |
| | /mìsli/ | | /i/ | | | | | |
| | mìsl | | i | | | | | |
| | ‘think’ | | open | | | |
(prs.3sg)
| |
While examples (21a), (21c), and (22c) are straightforward, the remaining examples involve the application of phonological operations to adjust illicit sound combinations. Thus, in (21b), the thematic vowel /e/ causes the deletion of the stem-final vowel /a/,
26 while the vocalic agreement suffix /ă/ causes the same in turn with /e/. These subsequent truncations yield the surface form
pìjnă. In the same vein, the theme /e/ in (21d) deletes the stem-final /a/ and is itself deleted by the agreement suffix /ăt/, which gives the surface form
làjăt. Similarly, the stem-final /
/ and /ì/ in (22a) and (22b), respectively, fall victim to the thematic vowel /i/, which inherits their lexical stress.
Class-III verbs differ as they are athematic. I claim that this corresponds to the absence of the
asp-slot.
27 As a result, the consonantal agreement suffixes can be added without adjustments. On the other hand, class-III verbs cannot encode aspectual semantics the way class-I and class-II do. I illustrate this analysis of class-III verbs in (23).
(23) | a. | base | – | tns | – | agr | (class III: simplex) |
| | /ìska/ | | | | /š/ | |
| | vis | | | | š | |
| | ‘want’ | | | | prs.2sg | |
| b. | base | – | tns | – | agr | (class III: secondary imperfective) |
| | /prepìsva/ | | | | /š/ | |
| | vis | | | | m | |
| | ‘copy’ | | | | prs.1sg | |
Note that (23b) is the analysis of the secondary imperfective prepìsvame, which I already analyzed in (19). The difference between the two analyses is only apparent: Focusing on their derivation—which I did in (19)—secondary imperfectives involve the addition of -va-, which undoubtedly occupies the asp-slot. On the other hand, the addition of -va- determines that the resulting verb belongs to conjugation III and therefore lacks the asp-slot—which is what analysis (23b) highlights. In other words, the asp-suffix -va- derives secondary imperfectives, at the same time excluding the addition of any further aspect suffix. This explains, among other things, why secondary imperfectives cannot repeatedly be imperfectivized by adding a second -va- (*prepìsvavam), and also why there can be no (underlying) thematic vowel in class-III verb forms.
There are two crucial implications following from the above analyses.
First, the thematic vowels in classes I and II reflect the openness of the AT-interval (
Section 2.2). However,
Nicolova (
2017, p. 375) argues that this feature is non-distinctive since, in the present tense, there is no ‘closed’ (non-continuative) counterpart. As will be shown in
Section 3.4 and
Section 3.5, the situation is different in the past tense as well as in
-l-periphrases, where the ‘open’ imperfect contrasts with the ‘closed’ aorist, which makes
open a distinctive feature in these forms.
28Second, class-III verbs are generally incapable of encoding the openness of AT. This holds true not only for the present tense but also for the imperfect. Therefore, class-III verbs are largely ambiguous between aorist and imperfect in the past tense (
Section 3.4) and in perfect periphrases (
Section 3.5).
3.4. Past Tense
Regarding the past tense, BG stands out from all other Slavic languages in two respects: First, it retains and productively employs the synthetic past-tense forms aorist and imperfect. Second, it productively derives both the aorist and imperfect forms from either viewpoint aspect.
29 In other words, BG has imperfective and perfective imperfects as well as imperfective and perfective aorists. This is illustrated in (24) and (25).
(24) | a. | Kupìx | si | dnès | mnògo | knìgi, |
| | buy.pfv.aor.1sg | refl | today | many | books |
| | ‘I bought a lot of books today,’ |
| b. | kupùvax | gi | cjàl | čàs. | |
| | buy.ipfv.aor.1sg | them | whole | hour | |
| | ’I was buying them for a whole hour.’ (Andrejčin, 1978, p. 178) |
(25) | Štom | napìšeše | pismò, | tòj | ti | otgovàrjaše. |
| as.soon.as | write.pfv.ipf.3sg | letter | he | you.dat | answer.ipfv.ipf.3sg |
| ‘As soon as you wrote the letter, he would answer you.’ |
| (Bertinetto & Delfitto, 2000, p. 215) |
For the sake of comparability, I repeat
Table 4 from
Section 3.1 as
Table 8 to supply the reader with full aorist past-tense paradigms.
30 Table 9 gives the imperfect paradigms.
Table 8 shows that to form the aorist past tense the vowels that underlie the division into conjugations (
-e-,
-i-, and
-a-) are replaced by other vowels or no vowel at all in classes I.2, I.3, I.4, and II.1. By contrast, they are left unchanged in classes I.1, II.2, II.3, and III. In case of the imperfect past tense, the latter holds for classes I.2, I.3, I.4, and III. On the other hand, the vowels are replaced with
-a- or
-e- in class II and to some extent also in class I.1.
The morphological analysis of aorist and imperfect forms is controversial. However, most authors agree that imperfect forms have the marker /ě/
31 added to the root or in place of the present-tense vowel (see, e.g.,
Tilkov et al., 1983, pp. 325–327;
Scatton, 1984, p. 185;
Krăstev, 2005, pp. 105–106;
Hauge 1999, p. 98;
Nicolova, 2017, p. 397). The resulting form is then usually called the
imperfect stem.
But opinions diverge as concerns the remaining inflectional material, and most crucially the overall analysis of aorist forms. Many descriptions present the inflections -x, -xme, -xte, and -xa as markers of both aor and ipf, with the exception of the 2/3sg -še and -⌀, which are peculiar for ipf and aor, respectively. This means that the burden of discriminating aorist from imperfect is primarily laid on the vowels that precede the inflections.
Scatton (
1984) is more explicit as he isolates the suffix
-x-, which he claims marks the past tense. Concerning /ě/, Scatton analyzes it as a dedicated imperfect suffix (his “*A”). Moreover, his compositional analysis implies that verb stems including /ě/ are specified as imperfect
before they become specified as past tense—which will be helpful when analyzing imperfect participles in
Section 3.5. In (26), I apply Scatton’s analysis to the imperfect past tense
pìjnexme, ‘(we) sipped’ (class I.2), placing the imperfect suffix in the
asp-slot.
(26) | base | – | asp | – | tns | – | agr |
| /pìjna/ | | /ě/ | | /x/ | | /me/ |
| pìjn | | e | | x | | me |
| ‘sip’ | | ipf | | pst | |
1pl |
Note that despite the fact that the surface imperfect stem
pìjne- is identical to the present stem (see
Table 3), the two stems differ in their underlying nature: The present stem involves the vowel /e/, whereas the imperfect stem contains /ě/. The same holds for classes I.3 and I.4. Thus, unlike stated above, the addition of the imperfect /ě/ takes place in all conjugations except class III. This receives a straightforward explanation under the assumption that these verbs are athematic and lack the
asp-slot (see
Section 3.3). In other words, there is no room for the imperfect suffix /ě/ in class-III verbs. As a consequence, they are ambiguous between imperfect and aorist, with the exception of the 2
/3sg due to their specific inflections.
Unlike imperfects, aorists do not exhibit a uniform marker. As
Table 8 shows, the slot where imperfects have /ě/ is occupied by one of the vowels
-o- (
-e-),
-(j)a-, or
-i-. Analyzing these vowels as dedicated aorist suffixes, one will hardly be able to find any systematic pattern to determine their choice. Moreover, it seems improbable that the same grammatical notion should be encoded by three vowels that can by no means be treated as allomorphs.
The one-stem approach (see
Section 3.3) avoids this issue. The vowel that occurs in aorist forms is just the final vowel of the single underlying stem. This can be shown by comparing the ‘aorist stems’, as illustrated in
Table 8, with the underlying stems listed in
Table 7—see
Table 10.
There are only two differences that call for an explanation:
(i) The ‘aorist stem’ in class I.1 shows the vowel
-o- (or
-e-), while the underlying stem does not. A straightforward explanation is that the underlying stems of verbs from class I.1 are consonantal (another example is /pek/ ‘bake’). Since the aorist inflections are consonantal, too, their addition requires the insertion of a fill-in vowel.
32 Consequently, the ‘aorist stem’ of the relevant verbs is actually identical with their underlying stem, and there is no exception whatsoever. The fill-in vowel is inserted to ensure pronounceability; see analysis (27) for the form
čètoxme, ‘(we) read’.
(27) | base | – | asp | – | tns | – | agr |
| /čet/ | | | | /x/ | | /me/ |
| čèt+o | | | | x | | me |
| ‘read’ | | (aor) | | pst | |
1pl |
(ii) The above comparison shows that aorists can have their stress on the pre-inflectional vowel
-a-, which serves to disambiguate them from imperfects primarily in cases where the two forms would otherwise be homonymous. Except for the 2
/3sg, homonymy regularly occurs in class III due to the impossibility of adding the imperfect suffix (see
Section 3.3 and above). Stress on the pre-inflectional
-a- is strongly marked since the basic (lexical) stress position of class-III verbs is always on the syllable preceding it. I take it that the relevant stress shift is a post-morphological operation. (28) illustrates my analysis for the aorist
iskàxme, ‘(we) wanted’. I add the
aor-feature below the stem since it is the marked stress on
-a- which reflects the aorist. Without stress shift (
ìskaxme), there would be no encoding whatsoever of this category.
(28) | base | – | tns | – | agr |
| /ìska/ | | /x/ | | /me/ |
| iskà | | x | | me |
| ‘want’ | | pst | |
1pl |
| aor | | | | |
To summarize, BG aorists are aspectually unspecified except for the—anyway ubiquitous—viewpoint-aspect category. In contrast, imperfects involve an additional aspectual specification (marked by /ě/) which does
not assert that the past relevant situation ceased before the time of evaluation, as AT is an open interval (see
Section 2.2).
All things considered, it is a reasonable conclusion that aorists encode that AT is a closed interval, i.e., that the situation denoted in the stem ceased before the time of evaluation (no current relevance). As aorists are morphologically unmarked, it is likely that the ‘closed perspective’ is the trivial default when speaking about past situations—as
Klein (
1994, p. 117) puts it: “It is just the nature of time. Things in the past are settled”. In other words, AT-intervals are closed by default when they are located in the past.
33 The following observations provide further evidence for the claim (advocated by, e.g.,
Aronson, 1967;
Maslov, 1959;
Scatton, 1984;
Sonnenhauser, 2006a) that the aorist is the unmarked member of the opposition.
While the aorist is inherited from Late Proto-Indo-European, the imperfect is a Proto-Slavic innovation formed with the suffix
-ěa-. Crucially, the suffix
-x- was adopted from the aorist by analogy (see
Bielfeldt, 1961, p. 281;
Trunte, 2018, p. 43). In other words, the aorist was the basic past tense already in Proto-Slavic.
While the imperfect past is an anaphoric tense requiring a referential anchor to hook up to in discourse, the aorist introduces such anchors (see, e.g.,
Becker, 2010;
Becker & Egetenmeyer, 2018). Therefore, the aorist is the prototypical narrative tense in BG.
But aorist and imperfect stems are used not only to build past-tense forms, but also to derive participles to be used in the diverse perfect(oid) periphrases of BG.
3.5. Perfect(oid) Periphrases
BG has a wide range of periphrases based on participles formed with the suffix /l/ (“
-l-participles”). But in addition to the ‘ordinary’
-l-participle available in all Slavic languages, BG has, as a rather recent innovation (see, e.g.,
Scatton, 1993, p. 215;
Nicolova, 2017, p. 63), a second
-l-participle based on the imperfect stem, which I refer to as the
imperfect participle. By analogy with the opposition in the past tense, I refer to the ‘older’
-l-participle as the
aorist participle.
34 Table 11 give examples of the two.
The morphological formation of aorist participles parallels that of aorist past-tense forms, with the participial suffix /l/ taking the place of the past-tense marker /x/; see (29).
(29) | base | – | asp | – | tns | – | agr |
| /pìjna/ | | | | /l/ | | /i/ |
| pìjna | | | | l | | i |
| ‘sip’ | | (aor) | | perf | | pl |
Imperfect participles are formed in parallel to imperfect past-tense forms as they involve the same aspectual suffix /ě/; see (30).
(30) | base | – | asp | – | tns | – | agr |
| /pìjna/ | | /ě/ | | /l/ | | /i/ |
| pìjna | | e | | l | | i |
| ‘sip’ | | ipf | | perf | | pl |
While the respective past-tense forms are finite in the sense that they are specified for mood and tense (and person) and refer to real-world situations on their own, aorist and imperfect participles require the presence of the auxiliary
săm, ‘be’.
35 For instance, prescriptive grammars (starting at least as early as
Andrejčin, 1944) state that the
(present) perfect is built with an aorist participle and a present-tense form of
săm, as shown in (31).
36(31) | Postpil | săm | v | universitèta | na | 1.9.2000 g. |
| enter.pfv.ptcp.sg.m | be.prs.1sg | in | university.def | on | 1.9.2000 |
| ‘I entered the university on 01.09.2000.’ (Nicolova, 2007, p. 419) |
However, perfect periphrases may also contain imperfect participles. An example is (32), dubbed the “perfect progressive” by
Hauge (
1999, p. 113). Hauge argues that speakers use this variant of the perfect in place of the imperfect past tense when they do not (wish to) vouch for the information conveyed.
37(32) | Tuk | v | manastìra | e | ìmalo | učìlište, | v | koèto | vìnagi |
| here | in | monastery.def | be;prs.3sg | have.ipfv.ptcp.sg.n | school | in | which | always |
| se | | e govòrel | | blgarski ezìk. | | | | |
| refl | be.3sg speak.ipfv.ipf.ptcp.sg.m | Bulgarian language | | | | |
| ‘Here in the monastery there has been a school where Bulgarian has always been spoken’. |
| (Hauge, 1999, p. 113) |
Much like the perfect, the pluperfect is formed with the aorist (or imperfect) participle plus an imperfect past-tense form of
săm; see (33).
38(33) | [À]rmijata | stoèše | bezpòmoštna, | sjàkaš | bè | pàdnala | v |
| army.def | stand.ipfv.ipf.3sg | helplessly | as if | be.ipf.3sg | fall.pfv.ptcp.sg.f | in |
| plen. |
| captivity |
| ‘The army was standing still helpless as if it had been captured.’ |
| (Jordan Radičkov; cited from Nicolova, 2007, p. 430) |
Considering the fact that the morphological makeups of the participles have the same stems as the past-tense forms, it is reasonable to conclude that their semantics is the same, too. Therefore, I claim that aorist participles present the situation denoted in their verb stem as finished, whereas imperfect participles do not.
In (29) and (30), I have located the participial suffix /l/ in the
tns-slot. Indeed, the contribution of /l/ is temporal, but unlike the past-tense suffix /x/, which encodes an
absolute past tense (i.e., the assertion time lies before the time of evaluation), the /l/ denotes a
relative tense. More precisely, it produces the
perfect time span (PTS), as defined in (7) in
Section 2.3. In a nutshell, this means that /l/ locates the original assertion-time interval (AT-1) in the pre-time of a newly introduced second assertion time (AT-2); see (34). The semantics of /l/ is thus not only temporal but also aspectual.
(34) | - - - - - - [AT-1 - - - ][AT-2 - - - ] - - - - - - | (PTS) |
To arrive at an absolute tense interpretation, an auxiliary must be used to determine the relation between AT-2 and the time of evaluation. Thus, for instance, a present-tense auxiliary will give rise to a present perfect, while a past-tense auxiliary yields a pluperfect.
Now, recall that the /l/-suffix is added to the stem
after the latter has been specified as either imperfect or aorist. This means that the semantic contribution of aorist and imperfect affects AT-1, whereupon /l/ introduces AT-2. In
Section 2.2, I argued that imperfect /ě/ specifies AT as an open interval, while its absence (=aorist) means that it is a closed interval. When applied to the PTS, this means that an aorist participle denotes a situation finished within AT-1. The latter interval lies in the pre-time of AT-2, from where the speaker views it in a kind of retrospection. Due to the fact that AT-1 and AT-2 form the overall PTS interval, the finished past situation is somehow still relevant in the state that obtains in AT-2.
To represent the category of viewpoint aspect in (34), we have to add the situation time (ST)-interval, which is represented in (35) using curly brackets.
39(35) | a. | - - - - - - {ST - [AT-1 - - - ][AT-2 - - - ] - - - - - - | (imperfective aspect) |
| b. | - - - - - - {ST - [AT-1 - - }][AT-2 - - - ] - - - - - - | (perfective aspect) |
The prototypical candidate to combine with a closed AT-1 interval (aorist) is the perfective aspect—see example (31)—as at least the right boundary of ST is included in AT-1, as depicted in (35b). This is why aorist participles, just like aorist past-tense forms, are mostly perfective. However, aorist participles may also be imperfective, as in (36).
(36) | Ivàn | e | stroìl | pjàsǎčnǎ | kùla | i | predì. |
| I. | be.prs.3sg | build.ipfv.ptcp.sg.m | sand | tower | also | before |
| ‘Ivan has built a sandcastle before as well.’ (Pancheva, 2003, p. 297) |
This may seem contradictory at first sight given the assumption that AT-1 must include at least one boundary of ST so as to count as a closed (or half-open) interval, while it must not, at the same time, include the culmination within ST so as to be imperfective. But this only appears to be a contradiction: The culmination not asserted by the imperfective aspect is an internal boundary, whereas AT-1 must contain at least one external (temporal) boundary to count as a closed interval. In other words, an imperfective aorist stem (both in the past tense and the perfect) does not, on the one hand, assert the culmination of the situation, but does, on the other hand, assert at least its right temporal boundary (cessation). Thus, utterance (36) means that Ivan has been engaged in building a sandcastle at some time before the time of utterance, but apparently stopped the process before its completion. The focus is on the fact that the activity of building, as such, took place. Moreover, (36) is a present perfect, so the past situation is somehow relevant for the state that holds at the moment of speaking. We might paraphrase this relevance by saying that Ivan is (by now) experienced in building sandcastles.
To schematize this state of affairs, I add to (35a) the moment of cessation “/”, which lies before the possible culmination (which I do not represent); see (37).
40(37) | - - - - - - {ST - [AT-1 - - - / ][AT-2 - - - ] - - - - - - | (imperfective aorist participle) |
Next, if we alter the viewpoint aspect in (36) to perfective, we obtain (38).
(38) | Ivàn | e | postroìl | pjàsǎčnǎ | kùla. |
| I. | be.prs.3sg | build.pfv.ptcp.sg.m | sand | tower |
| ‘Ivan has built a sandcastle.’ |
The perfective aspect asserts the culmination “•”, while all else remains the same. Thus, the building of the sandcastle is both completed and finished and somehow connected to the time of utterance; see schema (39).
(39) | - - - - - - [AT-1 - {ST - - • / } - ][AT-2 - - - ] - - - - - - | (perfective aorist participle) |
Tilkov et al. (
1983, p. 320) characterize this variant of the perfect “statal” because it is the target state which is focused.
Nicolova (
2017, p. 421) adds that the auxiliary can be dropped, as in (40), when the speaker perceives only the result of the activity (here, the existence of the sandcastle).
(40) | Ivàn | postroìl | pjàsǎčnǎ | kùla. |
| I. | build.pfv.ptcp.sg.m | sand | tower |
| ‘Ivan (has) built a sandcastle [I conclude].’ |
In such a situation, Nicolova claims, the speaker can only conclude that it was Ivan who built the sandcastle, which is why she dubs this variant “perfect of conclusion”. Moreover, Nicolova argues that this variant is the origin of some of the alleged evidential moods (see note 35). If the existence of such moods is accepted, aorist participles are described as referring to (reported) past situations. Note that this has a straightforward explanation considering the time structure in (39): As AT-1 is a closed interval, the situation must have ceased in the past, so can by no means be running at the time of utterance.
The next combination is illustrated by (41): The participle is imperfective and imperfect.
(41) | Ivàn | e | strojàl | pjàsǎčnǎ | kùla. |
| I. | be.prs.3sg | build.ipfv.ipf.ptcp.sg.m | sand | tower |
| ‘Ivan was/has been building a sandcastle [I infer].’ |
This is what
Hauge (
1999) calls the “perfect progressive” as it denotes relatively long-lasting activities or processes. Indeed, (41) adds the nuance that the building of the sandcastle took quite a while—and is possibly still running at the time of utterance.
Tilkov et al. (
1983, p. 320) and
Nicolova (
2017, p. 423) use the term “actional perfect”, which is to say that the action or process denoted in the participle is in the speaker’s focus, combined with its relevance for the present. Thus, while the imperfect excludes any temporal boundaries from the assertion, the imperfective aspect excludes any possible culmination. Schema (42) is therefore void of any boundaries whatsoever.
(42) | - - - - - - {ST - [AT-1 - - - ][AT-2 - - - ] - - - - - - | (imperfective imperfect participle) |
Like in the case of (38), the auxiliary may be dropped, as shown in (43). Again, the effect is a distance between the speaker and the information conveyed (see note 35).
(43) | Ivàn | strojàl | pjàsǎčnǎ | kùla. |
| I. | build.ipfv.ipf.ptcp.sg.m | sand | tower |
| ‘Ivan (has) built/is building a sandcastle [they say].’ |
As far as grammars classify this as a special evidential mood (‘narrative’), they add that the imperfect participle is ambiguous between two temporal readings: past or present. The time structure proposed here has the potential to explain this ambiguity in temporal reference: (42) shows that nothing excludes the possibility that the building of the sandcastle is still running at the time of utterance and that the speaker refers to this very activity. On the other hand, the speaker may as well focus on the past portion of the PTS, i.e., on an activity running at some past time (in AT-1). Put differently, the exclusion of the right temporal boundary thanks to the imperfect renders this variant temporally ambiguous.
Finally, an imperfect participle may also be perfective, as in (44).
(44) | Kogàto | Ivàn | postrojàl | pjàsǎčnǎ | kùla, | … |
| when | I. | build.pfv.ipf.ptcp.sg.m | sand | tower | |
| ‘Whenever Ivan built a sandcastle, …’ |
The same combination can be found in the past tense; see (45).
41(45) | Štom | napìšeše | pismò, | tòj | ti | otgovàrjaše. |
| as soon as | write.pfv.ipf.3sg | letter | he | you.dat | answer.ipfv.ipf.3sg |
| ‘As soon as you wrote the letter, he would answer you.’ |
| (Bertinetto & Delfitto, 2000, p. 215) |
Many authors observe that this combination occurs almost exclusively in adverbial clauses being part of complex sentences whose main-clause verb is an imperfective imperfect, and that its interpretation is iterative or habitual (e.g.,
Comrie, 1976;
Gvozdanović, 2012;
Maslov, 1959;
Pašov, 1999;
Rivero et al., 2017). In other words, perfective imperfects denote multiple instances of the same situation. In the model proposed here, this can be explained in terms of an interpretive adjustment: The contradictory combination of imperfect (open AT-interval, i.e., no ST-boundary inside AT) and perfective aspect (AT includes culmination and right boundary of ST) can be resolved if the utterance is taken to be about multiple instances of the same eventuality, each culminated (perfective aspect). These instances form a chain, and it is this chain which the utterance is about. Crucially, AT-1 covers only a portion of this chain, so is an open interval (imperfect). I schematize this in (46).
42(46) | - - - {ST - • } - [AT-1 - {ST′ - • } - {ST″ - • } - {ST‴ - • } - ][AT-2 - - - ] - - - | |
| (perfective imperfect participle) |
The same holds for perfective imperfects in the past tense. However, these involve only one AT-interval in the pre-time of the time of evaluation, as depicted in (47).
(47) | - - - - • } - - - • } - - • } - - • } - ] - - • } - - - ⬤ - - - | |
| (perfective imperfect past tense) |