Резюме. The present study is concerned with what a unified account of temporality could be based on. The proposal is that a theory that accounts for temporality should be based on some universal expression of this semantic phenomenon. The paper opens with the discussion of some problematic issues the traditional morphologically-based approaches raise. Then it is suggested that a solution could be to build the account taking as a basis the semantic types of the events that could be expressed in a clause. Evidence in favour of the soundness of the proposal comes from a comparative analysis of Bulgarian and Amharic data.

Ключови думи: temporality; aspectuality; perfectivity; imperfectivity; anchoring

In a previous work of mine (Laskova 2017), where I discussed the nature of the subjunctive, I drew some brief conclusions on temporality and I would like to present a more in-depth discussion on the issue in this paper.

Since Reichenbach’s (1947) seminal work ‘tense’ is the category describing the relationship between the eventuality expressed by the verb, the utterance time and another moment called 'reference time’. Such relation is claimed to be encoded in the morphological markers. Aspect, on the other hand, closely related to tense, has been considered the category that analyses the internal structure of an event and defines events, basically, as completed or ongoing (Comrie, 1976).

These definitions are widely shared, especially in the traditional approaches as they represent a description of the “perceptible” semantics of the corresponding morphological phenomenon. Thus, through these definitions, we provide accounts for the phenomena of temporality and aspectuality which derive directly from morphology. We do this by assigning to each morpheme a meaning corresponding to the semantic function it is associated with. Thus, since a Bulgarian native speaker associates the verbs carrying the morpheme -se with the expression of past events, we attribute this function directly to the morpheme -se.

Of course, such a system could serve as a basis for a neat and detailed description of linguistic facts, and it has done so for more than half a century now. Beyond the obvious morpho-semantic correspondences, however, there still remain some questions of both empirical and theoretical nature that need to be answered and, in this paper, I would like to discuss some of them.

Discussing the theory of tense and aspect, in any case, involves dealing with morphology, since the morpheme is the direct linguistic marker signaling the presence of a semantic phenomenon, which could possibly be given the status of a grammatical entity. Of course, the conclusions one draws largely depend on the theoretical method adopted. Though the present paper will be more concerned with raising the problem rather than providing an articulated answer or entering into technical details, the issue will be dealt with in the framework of generative grammar, since it offers a theoretical apparatus allowing to carry out an in-depth inquiry into the nature of the linguistic phenomena. I believe, however, that the discussion would be of interest to any syntactician or typologist, independently of the adopted theoretical instrumentation.

I would like to pose the question of whether semantic phenomena like temporality and aspectuality should be accounted for and, even more generally, assigned grammatical status on the basis of morphological occurrences. One of the motivations for raising this issue is related to the conclusion I drew in Laskova (2017) regarding morphology, while discussing the subjunctive. I suggested that sometimes morphology might be misleading as to how certain semantic phenomena should be analysed. It might provide only partial, not very transparent or even misleading cues on how the phenomenon should be dealt with. As is the case with Romance subjunctive, I suggested that its morphology wrongly suggests a relation between temoral agreement and the nature of the Subjunctive phenomenon. This is the reason for which for four decades no unified account based on its most evident feature, the temporal deficiency, has been offered. Also, as reported in Kiparsky (1968), some forms we label ‘subjunctive’ may not at all be carriers of the semantic meaning of the subjunctive, as is the case of the so called N-tense verbal forms found in Maasai and in the Bantu language Tswana. Later analyses correctly suggest that these are instances of Injunctive and not of Subjunctive forms. Furthermore, in Laskova (2017) I presented similar considerations suggesting that some subjunctive forms turn out to be syncretic expressions of more than one category or function.

With these considerations in mind, we can have a look at some more facts related to temporality and coming from different language families.

If temporality is a universal semantic category and its direct expression is a specific morphological marker (the traditional tense marker), then how come that in some languages temporality is expressed by other categories, like aspect (as is the case in tenseless languages, cf. Decaen) or mood (as is the case in Mokawk, cf. Baker and Travis (1995))? Moreover, how come that in Amharic, though tense is morphologically represented, past completed events are obligatorily morphologically underspecified for tense?

If we for a moment turn to the category of Aspect, traditional, morphologybased approaches lead us to associate, on the one hand, imperfectivity with progressivity and habituality, and, on the other, perfectivity with completedness (or totality) of the event. Such relations are the result of the association of aspectual morphemes with a given semantic meaning. However, if we construct a theory of aspectuality, based on such relations only, we fail to capture, for example, another very rigid cross-linguistic correlation, namely the one between imperfectivity and temporal anchoring (the fact that it is always the case that an on-going event needs to be anchored to a temporal moment). The moment to which the event is anchored could be either explicit or implicit but is obligatorily present (see Giorgi 2001 for a formal account on the Imperfect tense). The suggested correlation is closely related to the analysis of the Imperfect tense.

Notice also that in Indo-European languages, where exists a specific form for the expression of past events (the Aorist in traditional terms), there is strong evidence that this form is not a tense but rather an aspect form. It is a form characteristic of the Indo-European languages which expresses an event in its entirety (not necessarily telic and not necessarily a resultative one). As noted by Mountaut (2016), the aoristic forms are more often analysed as aspectual rather than as temporal forms. It is not by chance that such forms exist in the languages in which the grammatical aspect is morphologically represented. In particular, in her work Mountaut (2016) analyzes the simple form in – (y) a in Hindi/Urdu, providing evidence that it is an aoristic form and is tenseless and perfective. Similar status, i.e. tenseless and perfective, Giorgi and Haroutyunian (2014) attribute to the aorist form in Eastern Armenian. As mentioned before, the aorist is the only simple form, exclusively dedicated to the denotation of a total and complete event.

These data seem to suggest that the aorist on the one hand expresses a temporal relation and on the other appears to be morphologically tenseless. This conclusion pauses a problem for the morphologically-based accounts on temporality.

Since temporality is a phenomenon represented in all languages (even in those in which temporal morphemes do not exist) before turning to Bulgarian, I would like to discuss some interesting data coming from other languages.

Having the above considerations in mind, one might conclude that in order to offer a more profound and unified analysis on temporality one should try to go beyond morphological appearances and consider in more detail some typological and semantic facts.

In my view, when building an account for a universal semantic phenomenon like temporality (and I consider aspectuality very closely related), one should take as a starting point some universal semantic expression of this phenomenon. On the one hand, this could help to achieve cross-linguistic validity of the account and, on the other, could help to avoid the otherwise obligatory explanation of the cross-linguistic morphological variety. I will just note in passing that morphology needs to be considered in any case. The difference is that, on a morphologically-based account, one needs to explain the morphological occurrences and diversity while on an account based on another property one would only keep track of such occurrences and use them simply as indicators for the correctness of the analysis.

To begin with, what we can notice is that whatever the morphological structure of a language, a clause can denote a given number of events, which are roughly the following: an ongoing event (anchored to the utterance time or to another moment in the past or in the future), habitual events (of which I suggest that they can also be analysed as anchored to a given moment in time, similarly to the on-going events), completed events in the past (or also in the future), events completed before a given moment in time (the ones expressed through perfect tense forms), and the combination between ongoing and perfect events.

In the next section I analyse data from Amharic in the light of this typology of events.

Amharic is a Semitic language spoken in Ethiopia exhibiting discontinuous morphology. We distinguish between four different roots, namely P-root, I-root, G-root and a fourth type that appears in imperative phrases and in nominalizations. All exhibit different aspectual and temporal meanings. In morphology, temporality is expressed by means of auxiliary verbs which are not part of the root. Aspectual information, on the other hand, is encoded in the root itself.

Of interest for the present analysis are the first two types of roots, namely the P-root and the I-root. Importantly, the I-roots are not autonomous forms in the sense that they must appear in the sentence always accompanied by a temporal auxiliary. For this reason they are considered underspecified for tense, or atemporal forms (cf. Leslau 1995, Demeke and Meyer 2001, Demeke 2003, Yimam 2006, Halpert and Stolen 2014)1). Through the help of the auxiliary, the I-root can express a present, a future or a past on-going or habitual event. Consider the following examples taken from Meyer (2014):

(1) jɨsäbrallu

j-säbr- [u] -At-u

3P-break.IPFV- [Pl] -Aus.3Pl

“They break” / “They will break” / “They (usually) break“

(2) jɨ-Sabr-u näbbär

3P-break.IMPV Aus.Pst

“They were breaking” / “They broke”

Scholars agree that the I-roots are specified for the imperfective aspect.

Unlike the I-roots, the P-roots are used obligatorily without a temporal auxiliary and are independent forms in this sense. An example of this kind of root is the verb säbbär /“to break”. It generally carries the meaning of an accomplished event. In this sense, the event designated by this verb appears in the past. Observe the following example, quoted by Halpert and Stolen (2014).

(3) Marta wämbär-u-n säbbär-äft.

Marta chair-Def-Acc break.P-3P.Fem

“Marta broke the chair.”

Demeke (2003) suggests, however, (and scholars generally agree with this proposal) that we are not dealing with a past root but with an atemporal one, since in other contexts it can also have a future interpretation. Consider the following example (quoted by Halpert and Stolen, 2014) in which the P-stem appears in a subordinate sentence.

(4) Sara [Marta wämbär-u-n ɨndɨ-ti-säbbär] tɨ fälläg-all-ɨtʃ

Sara Marta chair-Def-Acc Comp-3P-break-P 3P-want.I-NPST-3.Fem

“Sara wants Martha to break the chair.”

In this sentence the P-stem has a future interpretation.

Very interesting is also another fact, namely that the P-roots can become part of a progressive form if the prefix ɨjja (a progressive affix) is added. Notice that in that case the auxiliary is obligatory. We can draw and interesting conclusion, namely that the auxiliary is sensitive to the aspectual information and not to the morphological type of root. (It is interesting to note that in order to create the progressive form, we use not the I-root, which is marked for the imperfect aspect, but exactly the P-root, which expresses perfective meaning.)

The data from Amharic very clearly show that a verbal form expressing an ongoing or a habitual event needs to be anchored to a moment in time (introduced by the auxiliary). We could maintain actually that the morphological expression of tense (i.e. the auxiliary) in Amharic is nothing else but an expression of the anchoring condition, the form itself being atemporal, as suggested in the literature. ‘Perfective’ events on the other hand do not need to be anchored and cannot be accompanied by an auxiliary.

At this point a question arises: If we consider the temporality expressed by the auxiliary in Amharic an expression of the anchoring condition, is it comparable to the temporality expressed in the European languages?

I will approach this issue by referring to a study carried out by Giorgi and Pianesi (2001) and Giorgi (2010) on the Italian imperfect tense. Particularly interesting for our discussion is the conclusion (presented in this form in Giorgi 2010) that the imperfect tense is not a past tense, as is generally thought, but (omitting the technical detail) is simply a form anchored to a moment in the past, a Topic time. This analysis very much recalls the situation in Amharic. The difference between Amharic and Italian is that in Italian the relation with the Topic time is encoded in the morphology of the verbal form itself – in the imperfect suffix -va, according to Giorgi’s (2010) analysis. It is this suffix that yields the feature checking against the Topic projection. In Amharic, on the other hand, it is the auxiliary that refers to the Topic time.

The same temporal relation is established in Amharic not only between an I-root and the past but also between an I-root and the utterance time through a ‘present auxiliary’. In Italian this is done by what is traditionally known as ‘present tense’. These considerations seem to provide an answer to the question posed above. It seems that it is possible to establish parallels between the temporality found in Amharic and the temporal properties of the European languages. Giorgi’s analysis shows that the ‘imperfective’ meaning of the Italian imperfect tense and its temporal properties are rendered through anchoring. And this is exactly what happens in Amharic.

Next, I would like to turn to Bulgarian and consider examples containing present tense but do not refer to the present time.

(5) VČera se razhoŽdam po centăra I izvedn㎠krai men profuČava kola s nad sto kilometra v cas.

What distinguishes the above sentence from a sentence expressing a present event is only the adverb vcera. Apart from the temporal characteristics, the above sentence exhibits also particular aspectual properties. Notice that despite the second clause expresses an instantaneous event – describing the passing of a car – the verb form used, is not perfective but imperfective.

These observations lead to the following conclusion: in order to express a past completed event, we do not necessarily need to use past tense morphology and perfective aspect. It is the context that introduces the situation and determines the correct interpretation.

Indeed, as is well-known, the past use of the present tense is contextually determined. At this point, we could say that the difference between the ‘past’ use of the present tense and its ordinary ‘present’ use is only due to the role of the context. It is the adverbs that tell us that the event in the first clause is an on-going event. Importantly, we understand this by the fact that the event in the first clause is anchored to the moment introduced by the adverb izvednaŽ occurring in the second clause. In addition, the past interpretation is facilitated by the adverb vČera. The adverbial reference thus introduces a meaning like: yesterday at one precise moment. Thinking of Giorgi’s proposal about the Italian imperfect, we could say this moment introduces a Topic time to which the on-going event is anchored. In the absence of past morphology referring to the past, therefore, the adverbial reference ‘yesterday at a certain moment’ is the only indication about the temporal location of the eventuality. The second clause where we have an instantaneous event expressed by the imperfective aspect, suggests that the aspectual morphology is also not decisive for determining the event type. Notice that even activities can also be used as past completed events. Consider the following example:

(6) Toj pie kafeto i izliza. (intended as a present in the past)

In this example, it is the sequence of events that suggests that the eventuality of drinking is represented as a completed event. Through rendering the event completed, the role of the context is actually to identify it, representing it as “one completed whole”. An on-going event, on the other hand, cannot be identified since it is not “one whole”. For this reason it needs to be anchored to a moment introduced by a Topic.

What we notice through the Bulgarian examples is that the types of events in this language are also temporally and aspectually analyzable by making use of the notions of anchoring and totality/completedness, exactly as it is in Amharic. And such analysis is completely independent from the Bulgarian temporal and aspectual morphology. The difference between these two languages is that Amharic morphology more clearly exhibits the relations between anchoring and on-going events and between pastness and perfectiveness while the Bulgarian morphology is not so ‘transparent’ in this sense.

What we can notice is that once we base our analysis on the types of events, we can more easily capture the similarities even between two unrelated languages.

In this paper I limit myself to analysing only the two types of events presented above (ongoing and past completed events), very tentatively assuming that, with slight changes, the rest could be derived from them. Following this analysis we could arrive at a theory of temporality to a large extent independent on morphological appearances. Of course a unified analysis, at least in order to justify its validity, should be able to explain what the role of observable tense morphemes is. In any case, I think that, on the basis of the present short analysis, we can say that it appears to be easier to build a unified account of Bulgarian and Amharic temporality by explaining Bulgarian data in terms of the notions of anchoring and totality than by explaining Amharic data through a morphological analysis.

As to the role of morphology, I think that the observable morphological structure of a language is the result of a historical development within “permissible” limits. It definitely carries hits concerning temporality and aspectuality but can hardly be itself the basis for a universal account of these two phenomena.

Conclusions

The aim of this work was to show that the obvious morpho-semantic correspondences traditional accounts of temporality are based on, could be a basis for a descriptive approach but could hardly be a solid basis for a unified theoretical account of temporality. I suggested that basing the analysis on the cross-linguistically common semantic types of the events, we could more easily capture the temporal similarities even between two unrelated languages.

NOTES

1. For this reason many authors suggest that this form is underspecified for tense.

REFERENCES

Baker, M.; Travis, L. (1995). Mood as Verbal Definiteness in a “Tenseless” Language, ms., McGill University.

Bondarco, A. Vl. (1971). Vid i vremja russkogo glagola (značenije i upotreblenije). Moskva.

Comrie, B. (1976). Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press.

Déchaine, R.-M. (1991). Bare sentences, in S. Moore e A.Z. Wyner (a cura di)., Proceedings from Semantics and Linguistic Theory Conference SALT 1, Cornell University Working Papers in Linguistics 10.

Demeke, G. (2003). The clausal syntax of Ethio-Semitic. Doctoral Dissertation, University of Tromso, Tromso, Norway.

Demeke, G.; Meyer, R. (2001). A reexamination of tense in Amharic, Afrikanistische Arbeitspapiere 65, 143 – 155.

Durrleman, S. (2007). Completive aspect in Jamaican Creole: The complete story? Generative Grammar in Geneva 5, 143 – 157.

Giorgi, A. (2010). About the speaker. New York: Oxford University Press.

Giorgi, A.; Haroutyunian, S. (2014). On the Temporal and Aspectual Value of the Modern Eastern Armenian Aorist: a comparative Perspective. Armenian Folia Anglistika 12, 151 – 177.

Giorgi, A.; Pianesi, F. (2001). Imperfect Dreams: The Temporal Dependencies of Fictional Predicates, Probus 13, pp. 31 – 68.

Halpert, C.; Stolen, M. (2014). Fixed aspect in Amharic conditionals, paper presented at LSA 2014 Annual meeting, Minneapolis, January 2 – 5, 2014.

Kiparsky, P. (1968). Tense and mood in Indo-European syntax. Found. Lang. 4, 30 – 57.

Laskova, V. (2017). On the nature of the subjunctive, Lingua, Elsevier, vol. 189, pp. 19 – 45.

Leslau, W. (1995). A reference grammar of Amharic. Wiesbaden: Frans Steiner Verlag.

Meyer, R. (2014). Aspect and tense in Ethiosemitic languages, Paper presented at the workshop: The morpho-syntactic encoding of tense and aspect in Semitic languages Friedrich-Alexander University, ErlangenNürnberg 26 April 2014.

Montaut, A. (2016). The verbal form -(y)a in Hindi/Urdu: An Aorist with „aoristic“ meanings. Guentchéva, Z. (a cura di), Aspectuality and Temporality. Amsterdam: John Benjamins.

Reichenbach, H. (1947). Elements of Symbolic Logic. New York: The Free Press. Paperback issue 1966, same publisher.

Yimam, B. (2006). The interaction of tense, aspect and agreement in Amharic syntax. In J. Mugane (ed.) Selected proceedings of the 35th annual conference of African linguistics, Somerville, MA: Cascadilla Proceedings Project, 193 – 202.