1. Introduction: the essence of temporality

The human experience of time relies on many contributing factors. To name a few, it is shaped by the structure and operations of the human brain upon which our conceptualization of reality is conditional; by the socio-cultural environment that makes us partition time in a certain way and adopt one or another perspective on the passing of time; by the awareness of the finiteness of the human life; and, arguably, it is also reliant, at least on the level of communicated thoughts, upon the language in which we obtain and convey information. There is an endless list of pertinent questions we can pose concerning the nature of the human concept of time, as well as the nature of time itself. In what follows I will focus on a small subset of these, centred round what I want to call here the essence, or the core, of the human concept of time as it is represented in natural-language discourse. In particular, I will explore the relation between objectivity and perspective-taking, in that humans, in conveying information, while striving for objectivity, cannot avoid adopting a perspective. The latter is what indexicality in language allows us to do; in ascribing temporal reference, we adopt a perspective. The consequence of this is that the human concept of time can no more be dissociated from the perspectival thought than the concept of the self can. The corollary of this observation is that perspectival temporality can no more be excluded from the semantic representation than the notion of the self can: neither can be reduced to the bare referent for the purpose of semantic representation if the latter is to retain cognitive plausibility. In what follows, I will present such a semantic qua conceptual approach to temporal reference developed within my theory of Default Semantics (Jaszczolt Reference Jaszczolt2005, Reference Jaszczolt, Heine and Narrog2010, Reference Jaszczolt2016a) – a radical contextualist approach that allows for an unconstrained independence of the semantic representation from the logical form of the uttered sentence whenever the goal of representing the primary meaning intended by the speaker and recovered by the addressee calls for it. In doing so, I build upon my theory of time as epistemic modality developed in Jaszczolt (Reference Jaszczolt2009a, Reference Jaszczolt, Jaszczolt and de Saussure2013, Reference Jaszczolt and Lewandowska-Tomaszczyk2016b), according to which, on the level of conceptual qua semantic building blocks, temporality disappears: instead, there are degrees of detachment from the certainty of the here and the now into which the concept of time is reducible. In other words, temporality is a complex concept that can be broken down into simple, primitive, atomic concepts, and these are modal in nature. They are also scalar, pertaining to degrees of epistemic modality. I also address the question of temporal asymmetry between the past and the future that my scalar account throws up, and move on to the next seminal question of this paper, namely the relation between metaphysical time (called here time_M), psychological time (time_E, where ‘E’ marks the domain of epistemological enquiry), and time in natural language (time_L), concluding that the perspective-infused time_E and time_L are compatible with time_M of mathematical models of spacetime: all are definable through possibility and perspectivity.

The structure of my investigation is as follows. First, in the remainder of this section, I introduce two vintage approaches to metaphysical time, the so-called A- and B-theories which, although both somewhat flawed, have given rise to the vast majority of the current vivid debates in the philosophy of time, and which have informed semantic approaches to temporal reference. Having adopted a complex stance that uses essentially B-theoretic devices appropriate for time_M but yields conceptual representations where A-theoretic time emerges by appealing to the perspective of the speaker (time_E), I then move in Section 2 to some examples of cross-linguistic differences in how temporal reference is expressed in natural languages, pointing out that these differences can easily be accounted for by what I call lexicon/grammar/pragmatics trade-offs. I take on board the significance of tenseless languages, as well as the phenomenon of time–tense mismatches and the associated problems with accounts based on grammatical tense.

Throughout this paper, ‘tense’ is used in three ways: in the sense of a predicate pertaining to grammatical categories (linguistic tense, or ‘tense_L’); in the sense in which it is used in metaphysics when philosophers talk about ‘tensed reality’ (metaphysical tense, or ‘tense_M’; see, e.g., Ludlow, Reference Ludlow1999, Reference Ludlow, Jaszczolt and de Saussure2013, and McTaggart, Reference McTaggart1908, on A-theory); and, finally for the concept of time as the past, the present, and the future that does not necessarily correspond to the linguistic tense and that I call epistemic tense, or tense_E. Tense_E is introduced here in order to capture the multifarious provenance of information about temporal reference of eventualities that is then represented in conceptual qua semantic, but not necessarily qua structural-linguistic, representations. ^{Footnote 2} The concept tense will only be relevant when I consider the question as to whether reality, thought, and language are themselves tensed; that is, affected by the distinctions between the past, the present, and the future (leaving aside the question of their reality). When I discuss time on the level of metaphysical time, psychological time, and time in language (in its lexicon, grammar, as well as pragmatics), I use the analogous distinction between time_M, time_E, and time_L. The indices will also sometimes be applied to the labels ‘past’, ‘present’, and ‘future’, in an analogous way.

Next, in Section 3, I move to the subjectivity imposed by the speaker’s perspective and to the role of context in providing the perspective that, in radical cases, can lead to the conclusion that even the past can change. In Section 4, I briefly present the theory of temporality as epistemic modality developed within Default Semantics (Section 4.1), also attending to the question of the asymmetry of the past and the future that is seemingly irreconcilable with a degree-based modal account (Section 4.2), followed by a representation of an example of a tense–time mismatch in Default Semantics (Section 4.3). In Section 5, I move to addressing the question of how deeply rooted the modality of time is and what it suggests about the compatibility between the L, E, and M levels. In other words, I consider the meaning of modality vis-à-vis the time_L/time_E/time_M distinction. Section 6 concludes with a general discussion of the inalienable essence of time on the dimensions of experience and representation. I conclude that basic epistemic modal concepts together with the de se perspective allow us to represent reality as if it was tensed_M, but only so because time_M of spacetime models is itself intrinsically metaphysically modal.

Most current debates concerning the metaphysics of time stem from the argument for unreality of time proposed over a century ago by McTaggart (Reference McTaggart1908). He put forward two views on time, found them both flawed, and concluded as a result that time is unreal. According to the first view, called by him the ‘A-theory’, time passing is real: reality is tensed_M, there is change in the world. In metaphysics, the stance that time passing is real, called tensism, is an umbrella term for four views, differentiated by what parts of the temporal spectrum are considered real: (i) the past, the present, and the future; (ii) only the past and the present; (iii) only the present and the future, in the sense that only these two have the property of existence; and finally, probably the most popular stance in this camp, (iv) only the present exists – the view called presentism. According to the alternative view, ‘B-theory’, reality is tenseless_M, and time_M does not flow. Events are real and are ordered according to the earlier than / later than relation, but there is no past_M, present_M, or future_M as their real attributes. Now, using the A-series we are able to represent change, but the series itself is flawed in that temporality in it is irreducible, which leads to circularity: on the one hand, an event cannot be simultaneously past, present, and future, and on the other, when one removes the qualifier ‘simultaneously’, one presupposes the existence of time and introduces circularity. Next, using the B-series we are able to represent the ordering of events, but the series is flawed in that in order to make use of the earlier-than / later-than relations in the first place, we have to presuppose the existence of time. Without time we would have just a timeless ordering on a meaningless scale which McTaggart calls the ‘C-series’. McTaggart therefore concludes that the past, the present, and the future are properties not of reality but rather of human experience. Humans remember events, perceive them, or anticipate them, which produces the experience of time and of the passage of time. But this also leads him to conclude that time itself is unreal.

Nevertheless, there is a clear sense that time is real: events that took place cannot be ‘undone’; they belong to collective memory and to historical accounts irrespective of the experience of one particular individual; memories may differ somewhat or may fade completely, but there is some ‘reality’ beyond them that is composed of what (with some approximation) various groups, societies, and nations choose to call historical facts. In a different way, time also appears to be real for scientists working with spacetime models and for those laymen for whom spacetime with the associated corollaries predicted by mathematical models constitutes a concept actively employed in reasoning. Basic understanding of the ideas underlying Einstein’s theory of relativity results in the belief – or knowledge, depending on the degree of understanding – that time is not absolute and that it varies with the velocity of the moving object. Here B-theory comes up trumps. But there is a conceptually opposing force in human folk thought, founded on an accepted conviction that events can be placed on a pretty much fixed timeline, and characterized as past, present, or future, as they are in A-theory. The latter reflects the classical Aristotelian conception of time, adopted in Isaac Newton’s Principia Mathematica: time can be represented as a horizontal line extending infinitely in both directions, into the past and into the future. But just as the simplicity of the picture is appealing, so the evidence, collected even at that time, was disturbing. It was observed that the time of the eclipses of the moons of Jupiter appeared, when observed from the Earth, to be related to the distance of the moon. Henri Poincaré’s and subsequently Albert Einstein’s view that time is not absolute follow as a natural consequence. What we have now is the special and the general theory of relativity that can still feel at odds with intuitions. ^{Footnote 3}

Versions of A-theory fare better with intuitions but encounter problems with truth-makers, especially for the temporal location that is denied existence: presentism, for example, has a problem with truth-makers for the past and the future and has to resolve to theoretical tricks which have dubious metaphysical support. ^{Footnote 4} They also appear to be incompatible with Einstein’s special relativity. For example, special relativity extends to relativity of simultaneity, while presentism is founded on the ontology of times with unique properties arranged on a scale (see Einstein, 1920/2010, and, e.g., White, Reference White2015). ^{Footnote 5} On the other hand, the human experience of time as tense_E can be dissociated from the properties of real time when we accept that the flow of time is an illusion caused by the experience of living in time where events can be experienced as happening now and can be linked by causation. In other words, time can still be real but not flow (Mellor, Reference Mellor1998). This is the view compatible with the multidimensional view on spacetime in mathematical models of contemporary physics. In the terminology adopted here, we would have tense_E but no tense_M: reality has time_M as some of its multiple dimensions but does not have pastness, the now, or the future. It is the de se perspective that allows us to think of time_M as if it was tense_M. In Section 3 we will link this stance with perspective-taking and with expressing time as an aspect of expressing the self, using evidence from a domain that is not normally used for this purpose: the lexicon/grammar/pragmatics trade-offs in discourse as they are reflected in a variety of natural languages.

2. Temporal reference: from time_L to time_E

The ways languages express temporal reference point clearly in the direction of some form of tense_E but away from universal tense_L and also away from tense_M. They also reveal the existence of tense–time mismatches. In other words,

1. (i) languages have no difficulty with expressing the human concepts of the past, the present, and the future (tense_E);

2. (ii) not all natural languages have grammatical tense (tense_L); when they do, tense_L need not correlate with tense_E.

We will add to this the following:

1. (iii) (iii.a) the perspectival and egocentric representation of temporal distinctions, as well as (iii.b) the epistemic-modal building blocks of temporal concepts that (iii.c) in themselves exhibit perspectivity strongly suggest that the property ‘tensed’ emerges on the level of human conceptualization rather than being traceable to ontology (tense_M).

In this section I will attend to (i) and (ii), leaving (iii.a) for Section 3, (iii.b) for Section 4, and (iii.c) for Section 5, which will complete the argument.

Although grammatical tense (tense_L) by its very raison d’être externalizes the concept of temporal reference, the distinctions within this category can be more, or less, finely grained. Matses, a Panoan language of the Amazon region, has three past tenses: recent, distant, and remote. When the sentence concerns a past eventuality reported on the basis of inferential evidence, the source of information has to be overtly specified using its evidential system. It has to be stated overtly, by means of inflection called double tense (Fleck, Reference Fleck2007), (i) how long ago the eventuality took place and (ii) how long ago the speaker obtained this evidence, which results in categories such as ‘distant past inferential’ plus ‘recent past experiential’, or ‘distant past experiential’ plus ‘recent past inferential’, adding up to nine markers for the past-time reference (see also Aikhenvald, Reference Aikhenvald2012, p. 181). Central Pomo, a native American language spoken in California, distinguishes between two kinds of future: realis and irrealis, used in accordance with the speaker’s judgement about the degree of probability of the eventuality. Next, while we may think that realis/irrealis is a distinction subsidiary to the temporal one, Caddo, another native American language, expresses the future as realis (de Haan, Reference de Haan and Frawley2006, pp. 41–42). St’àt’imcets (Lillooet Salish) spoken in British Columbia has only the future/non-future distinction. However, since the marker of futurity kehl can also be used for the future in the past, akin to the English will/would, arguably the language is not inherently tenseless: the marker can be theoretically analyzed as combining with a phonologically empty, or unarticulated, tense morpheme (Matthewson, Reference Matthewson2006). Present- and past-time reference are expressed by potentially ambiguous temporally unmarked constructions where the aspectual class triggers the default reading: present for states and past for accomplishments and activities. Arguably, some overtly tenseless languages yield better to a tensed semantics with a null TENSE morpheme, while others, like Paraguayan Guaraní (Tupí-Guaraní), are ‘inherently’ tenseless: analyzing them using TENSE would not work (see Tonhauser, Reference Tonhauser2011). In Tupí-Guaraní, future-time reference is expressed by using aspectual, modal, and mood markers such as desiderative/hortative, prospective, or possibility. Generally, in Tupí-Guaraní and Arawak languages, aspect, modality, and mood are expressed (optionally) through inflections on the verb, while temporal distinctions are expressed through the lexicon (see Aikhenvald, Reference Aikhenvald2012, p. 180). To compare, Yucatec Maya makes use neither of tenses nor of temporal adverbials: temporality has to be gleaned from the aspectual marking through a default pragmatic interpretation (see Bohnemeyer, Reference Bohnemeyer2002). Languages with optional tense and aspect, such as Thai (see Srioutai, Reference Srioutai2006), often make use of the same strategy, with no loss of informativeness or clarity.

Considered from an extra-theoretic perspective, this means that, although a language may not have grammatical tenses, the way temporality is realized can still be closer to, or further from, the meaning of the devices present in the system of the language itself: sometimes the lexicon provides the necessary adverbials of temporal location, which means that we can still glean regularities from the system itself, while at other times resorting to context used in pragmatic inference or in conversational defaults ^{Footnote 6} is the only way of analyzing temporal reference. While we cannot yet at this point address the question of universality or relativity of the human concept of time (time_E) and the concepts of temporal reference (tense_E), we are able to use this extensive evidence to argue in favour of the equivalence between grammatical, lexical, and pragmatic means of expressing these concepts, as well as their interaction – a process that I call elsewhere the grammar/lexicon/pragmatics trade-offs (Jaszczolt, Reference Jaszczolt, Filipović and Jaszczolt2012). In this vein, Bittner (Reference Bittner2005), for example, analyzes West Greenlandic (Kalaallisut), an Eskimo-Aleut tenseless language, emphasizing the precision with which temporality can be expressed in it. Tense–time mismatches in English, where, for example, the continuous or simple present can be used for future eventualities to signal their planned or predetermined status (‘futurate progressive’ and ‘tenseless future’, respectively), or for the past to enhance the interest value of narration (‘past of narration’), demonstrate how the trade-offs can work: a temporal adverbial and the context, sometimes in conspiracy with the aspectual class, combine with information that is associated with the grammatical category to yield the exact required effect. ^{Footnote 7,Footnote 8}

Next, in addition to the differences in the granularity of the distinctions within the past, present, and future categories, ranging from null to several temporal subclasses, there are orthogonal concepts: consecutive tense in Swahili, for example, is orthogonal to the past/present/future distinction; there is a grammatical marker signalling the ‘and then’ relation (Givón, Reference Givón2005, p. 154). There is also the issue of the orientation of the speaker vis-à-vis the arrow of time that is amply discussed in the cognitive semantic literature. Maori, for example, points to the conceptualization where the speaker faces the past: the past is known and as such is spread in front of the speaker’s eyes (Thornton, Reference Thornton1987). To repeat, we are not yet in a position at this stage in our argument to draw conclusions concerning the relative or universal status of temporal concepts: this question will have to wait until Section 4. However, what we can conclude is that, since it is fairly incontrovertible to assume a correlation between concepts and linguistic expressions, different aspects of temporality are brought to salience in different natural languages, and it would take much more than our lexicon/grammar/pragmatics trade-offs to explain them because languages differ not only with respect to which of the three kinds of devices they employ for expressing a concept, but also with respect to what their speakers choose to conceptualize. There is much more to time_L than tense_L, but when time_L is construed in this contextualist way it maps onto time_E. It remains to be seen what this construal tells us about the question of the relevant conceptual universals. The route we will follow here goes through the question of the interaction between expressing temporality and the de se perspective, followed by the question of the complexity of temporal concepts as reflected in natural language lexicons and grammars. The methodology of this move is as follows. The assumed trade-offs allow us to go beyond tense_L; it is an investigation of time_L that will allow us to enquire about the complexity of time_E and the associated tense_E (in the sense of past_E, present_E, and future_E). But instead of assuming that time_E gives us an insight into time_M, as it is done in the so-called descriptive metaphysics, ^{Footnote 9} we will reverse the order of explanation, ask whether time_M plays a part in the construal of time_E, and enquire about the foundations for their compatibility.

3. Time, perspective, and expressing the self

Theories of psychological time invariably draw on experience and on the understanding of experience. For Kant, Husserl, or Heidegger, time is founded on retaining images. Husserl (Reference Husserl and Brough1928) distinguishes the apparent time pertaining to external objects and the internal time pertaining to mental acts. Cognition is permeated with the properties of mental states that at best (but not always, depending on the version of intentionality adopted in a particular phenomenological approach) are directed at an external object, so it is the internal time that we ought to focus on: time comes into being as part of consciousness. It is imposed on our experiences and only through them indirectly on real eventualities. External time only matters insofar as it is reflected in experiences (for the present), memories (for the past), and anticipations (for the future). The present, the past, and the future of the external time are reanalyzed in this way as properties of consciousness: primal impression (from perception of the present moment), retention (from memory of the past), and protention (from the expectation of the future). ^{Footnote 10}

Now, since internal time is constituted by consciousness, it follows that understanding time is part of understanding the ego. Just as human life is bounded by the events of one’s birth and death, so is the human concept of time (see Heidegger, Reference Heidegger and Stambaugh1953); just as human life is composed of experiences perceived, remembered, and anticipated, so is the human concept of time composed of the present, the past, and the future. As was exemplified in the previous section, some of these concepts in this three-way distinction make it into the grammar and/or the lexicon of natural languages – others do not, but as conceptual correlates of experiences they are always salient.

In this picture, the three-way distinction in the domain of temporal reference into past_E, present_E, and future_E acquires the status of a conceptual universal. By theoretical assumption, for us, it will also acquire the status of a semantic universal, irrespective of what the syntax and the lexicon of the particular language choose to externalize, in that on our contextualist construal semantic representations capture the meanings conveyed in discourse irrespective of their relation to the logical form of the sentence, in agreement with the lexicon/grammar/pragmatics trade-offs. ^{Footnote 11}

At this level of generality, the three-way distinction making up tense_E exhibits obvious parallelism with that of tense_M: we need not assume a flow of time, or the present, the past, or the future on the level of tense_M in order to uphold it. In metaphysical spacetime, in the second law of thermodynamics the ‘arrow of time’ points in one direction where the entropy (disorder) of a closed system always increases in the world and what has happened, happened irreversibly. Likewise, in cosmological explanations of time, the universe is progressively expanding (see, e.g., Hawking, Reference Hawking1988, p. 144), and analogously, in psychological time, the future always becomes the present and the present always becomes the past.

But when we enquire into the linguistic representations of past-, present-, and future-time reference actually performed by interlocutors in discourse, even in one particular language, we can see considerable variety of conceptualization and a considerable degree of choice. Combining temporal, modal (including evidential), ^{Footnote 12} and aspectual resources allows the speakers to represent eventualities with a different degree of inherent detail and with a different degree of commitment. For example, past events can be reported in English using regular past tense, as in (1):

1. (1) He went to London and met the Mayor;

past of narration, as in (2):

1. (2) This is what happened yesterday: He goes to London and meets the Mayor;

epistemic necessity past, as in (3):

1. (3) He is not here; he must have gone to London to meet the Mayor;

or epistemic possibility past, as in (4):

1. (4) I wonder why he is not here; he may have gone to London to meet the Mayor.

The present and the future also allow for considerable variation – free variation with respect to temporal reference but often more constrained, principled variation from the point of view of epistemic commitment and considerations to do with informativeness, social and cultural appropriateness, and so forth. For example, focusing on the internal structure of the event, one could represent it as protracted; focusing on evidence, one could use the evidential conditional. These are well-known facts of the grammar and pragmatics of English and we will not rehearse them here. What is of interest to us is the fact that this availability of a relatively free choice suggests that reporting on events is permeated with the de se perspective: just as in tense_E time is only of interest where it translates into the properties of human experiences, so in tense_L time is intertwined with perspective, including commitment, and the level of detail with which the eventuality is presented. Further variation can be ascribed to the perception of causal links between events and to what is conceptualized as a unique event in the first place. The literature on the individuation of events is vast and philosophical discussions are ample. Suffice to say here that the question of the granularity of events applies not merely to units that correspond to propositional thought but at a higher level as well: William James’ specious present, for example, is a concept constructed in order to capture the intuition that we are in direct contact not only with what is happening now, but also with what immediately preceded and will immediately follow the now; in other words, the intuition is that we directly perceive not only the present, but also chunks of the past and the future, and they all form one experiential unit.

Bringing together perspective, the self, and the properties of temporal reference appears indispensable when we attend to questions pursued in the philosophical literature concerning the role of change as a seminal definitional characteristic of time. Barlassina and Del Prete (Reference Barlassina and Del Prete2015) discuss the following actual situation. Lance Armstrong won the Tour de France in the year 2000. Hence, the proposition expressed in (5) is true:

1. (5) Lance Armstrong won the Tour de France in 2000.

(from Barlassina & Del Prete, Reference Barlassina and Del Prete2015, p. 61). But in 2012 it was discovered that Lance Armstrong had used illegal substances and his title of winner was withdrawn. Barlassina and Del Prete assume that, after this event, (5) becomes false, even when it is uttered by the same speaker who, in addition, may not be aware of the withdrawal of the title. They say, “… in moving from Context A to Context B, the past (of the actual world) has changed” (p. 61). When we think of events as properties of time (Montague, Reference Montague and Thomason1960), then the year 2000 ceases to have the property of winning the Tour de France by Armstrong. ^{Footnote 13} In order to solve the problem they try to employ contextualism, on which winning the Tour de France is a context-sensitive predicate, or relationism, on which ‘winning the Tour de France’ is a relational predicate that relates a person and a predication in a context, concluding that neither will work because (6) would acquire a wrong analysis:

1. (6) It is no longer the case that Lance Armstrong won the Tour de France in 2000.

(from Barlassina & Del Prete, Reference Barlassina and Del Prete2015, p. 65). They take it for granted that (6) is intuitively true when uttered in 2012 and that contextualism and relationalism both require that the meaning of (5) as embedded in (6) is tied, respectively in one way or the other, to the context relevant at the time of uttering (6).

Now, if we follow them in accepting their first assumption, namely that (6) is intuitively true, then there is indeed a problem with implementing a contextualist or a relativist analysis, at least on their version of the accounts. We would either have to resort to minimalism and the conclusion that the past has changed, or tweak the contextualist or relativist accounts in such a way that they are made to conform to the authors’ intuitions about (6). Another way out consists of denying the intuitive plausibility of (6), for example by claiming that intuitively it is indeed the case that Lance Armstrong did win the Tour de France in 2000 but, say, that the event of winning and the property of being a winner have different characteristics: he won in 2000, tout court, but he was a winner in 2000 and is no longer a winner in 2012. We can easily tweak the linguistic descriptions and avoid a puzzle. If (6) is false, we have the desired result.

But let us go along with the authors on the plausibility of (6) and see how we can tweak a contextualist account to accommodate it. In order to accept the intuition that (6) is true, we would have to attend to its problematic structure. And since it is precisely the structure and the associated scoping that leads the authors to the formulation of the alleged puzzle, unravelling the structure is likely to annihilate the puzzle itself. In order to agree with the speaker who truthfully utters (6) we would have to make it mean something to the effect of (7), where p stands for (5).

1. (7) It is no longer the case that what was true in 2000, namely p, is true in 2012.

(7) is equivalent to (8):

1. (8) It is no longer true that what was true in 2000, namely p, is true in 2012.

Then, (8) reanalyzes as (9):

1. (9) p was true in 2000 and p is no longer true in 2012.

In virtue of the content of (5), (9) reduces to (10):

1. (10) p and it is no longer true in 2012 that p.

It is apparent that the problem is strictly intra-theoretic and is triggered only on the account on which events are taken to be properties of time. But this is a rather peculiar way of posing the problem. It is true that Montague (Reference Montague and Thomason1960) and following him Partee (Reference Partee1973) adopted the construct of events as event types and as such properties of time, but several decades later one would have to have a good independent reason to adopt this construal of events. As such one can present nothing more than an intra-theoretic puzzle constructed merely for its own sake. In the past several decades ample analyses and arguments put forward in formal semantics have demonstrated clear explanatory advantages of construing events as particulars, to wit Davidson (Reference Davidson1970), Higginbotham (Reference Higginbotham1985), or Parsons (Reference Parsons1990), followed by the formal treatment of temporality in dynamic semantics such as Discourse Representation Theory (Kamp & Reyle, Reference Kamp and Reyle1993).

If we step outside Barlassina and Del Prete’s (Reference Barlassina and Del Prete2015) ‘puzzle for the puzzle’s sake’, what we have here is a fairly straightforward and unproblematic scoping: at t ₁(e) and at t ₂ (at t ₁(e)). It is straightforward because we can now, for example, plug in a speech-act-theoretic account and treat ‘at t ₁(e)’ and ‘at t ₂ (at t ₁(e))’ as assertions where truth of an eventuality is relativized to times, resulting in ‘at t₁(at t ₁(e))’ and ‘at t ₂ (at t ₁(e))’, respectively. What was presented as a puzzle about the past now appears to be merely the fact that the structure of natural-language sentences is itself to be analyzed in a contextualist manner since it is the context that allows pragmatic, syntactic, and lexical means of expressing meaning to mix and to work together in producing utterance meaning.

Now, having rejected the intra-theoretic problem with contextualism with respect to the logical form of (6) by pointing out the overwhelming preference for a construal of events as particulars in formal semantics, we can move full steam to the benefits of a contextualist outlook. There is something intuitively appealing in saying that the past has changed: after all, (7)–(10) are all true, and if so, then the past has changed, although not in the sense in which Barlassina and Del Prete (Reference Barlassina and Del Prete2015) use this term. The past has changed because the contents of the speech acts captured in ‘at t ₁ (at t ₁(e))’ and ‘at t ₂ (at t ₁(e))’ are different. These contents are relativized to the perspective: either the perspectives of different speakers, or, as was suggested above, even the same speaker, ignorant of the revoking of the title, on two different occasions. What appeared to be a puzzle now appears to be the standard state of affairs calling for an equally standard contextualist analysis: speakers alter their view and alter their judgement, or the subject of their judgement may change if someone else fills a given role. In the latter case the puzzle becomes just another version of the pseudo-puzzle associated with roles and reference exemplified in (11) and (12a)–(12b).

1. (11) The President of the United States changes every four years.

2. (12) a. The President was assassinated in Dallas in 1963.

   vs.

   b. Barack Obama was assassinated in Dallas in 1963.

These are pseudo-puzzles indeed, and in addition well-mulled-over ones.

The intuitive folk conclusion is then this: the past has changed because the speaker’s perspective, attitude, knowledge of the past, have changed, and these are essential for the representation of the past. On the account developed here, past-time reference is built out of attitudes to events that are in the epistemic sense remote from the self, and it is precisely the variation in this remoteness that changes. ^{Footnote 14} It seems that in languages such as Matses discussed in Section 2, where double tense would be used for the purpose of (5) and (6), combining, say, distant past inferential and recent past experiential, the problem would not arise: the combinations of tenses and evidential markers distinguish between different takes the speakers adopt on p, and result in appropriate semantic representations.

All in all, time_E and time_L are permeated with perspective. First, they are perspectival in the sense that the eventuality is portrayed from the vantage point of the speaker, and as such it is presented as tensed experience, with different degree of attention to different aspects of the situation and, what will have particular importance for the next section, with different degrees of epistemic commitment. Next, they are inherently perspectival in the sense of temporal anaphora, namely that changing vantage point can alter the future to, say, future in the past, or the present into the future. This is well known from the need for distinguishing between event time, speaker time, and reference time proposed by Reichenbach (Reference Reichenbach1948). It has to be remembered that the vantage point can change as discourse progresses, as is exemplified in (13):

1. (13) On Sunday I thought I would spend the day reading War and Peace. I was going to have coffee in bed and read. I started reading but then I remembered I had to go to a wedding and got up speedily.

Arguably, every instance of a use of grammatical tense reflects a perspective when the perspective is understood as a creation of ‘deictic consciousness’ (see de Saussure, Reference de Saussure, Jaszczolt and de Saussure2013, and his discussion of origo) pertaining to (i.a) the speaker, or (i.b) shifted to somebody else, or sometimes (i.c) generalized to a greater or lesser extent (as in the case of, say, analytic truth on the one hand and physical laws on the other), and analogously to (ii.a) the speech time, or (ii.b) shifted away to the past or the future, or (ii.c) generalized over time.

This phenomenon is not unlike the phenomenon of the reference shift in the case of some indexicals such as I, you, here, or now discussed in the philosophical literature. According to Kaplan (Reference Kaplan, Almog, Perry and Wettstein1989), the semantic value of an indexical is fixed in the context of the current speech act, except for its use in direct quotation. ‘Monster contexts’ are contexts that would violate this rule. But in the past few years, following Schlenker’s (Reference Schlenker2003) observation that Amharic allows for the displacement if I, as in (14), linguists have engaged in a hunt for languages that allow for such ‘monsters’. ^{Footnote 15} In (14), ‘me’ refers to the speaker and ‘I’ refers to the speaker’s brother, and the shift of reference cannot be attributed to quotation: the Amharic sentence is not quotative. Instead, what we have is a narration from two different perspectives, where the choice depends on which perspective is chosen as more relevant and salient.

1. (14) wändәmme käne gar albälamm alä

   my-brother “with-me I-will-not-eat”, he said

   ‘My brother refused to eat with me.’

(from Leslau, Reference Leslau1995, p. 778). While this phenomenon has normally been analyzed as a case of ‘context shift’, it seems more appropriate to describe it as the case of focusing on the semantically prominent perspective. In the spirit of Lewis’s (Reference Lewis1979) centered worlds, Roberts (Reference Roberts2014) uses for this purpose the concept of ‘doxastic centers’, and Mount (Reference Mount2015, p. 20) ‘mutually-accepted perspectives of interlocutors’. It appears that such mutually accepted perspectives also provide a theoretical solution to temporal reference: the vantage point, the degree of epistemic commitment, the aspect of the eventuality that the speaker chooses to bring to focus all testify to the fact that perspective and time_E are inherently interconnected. This will become important in the following section when we descend to the conceptual building blocks of the concept of time. We are now in a position to move to the properties of time_E itself and to address the question of the complexity and universality of the concept.

4. Time_E as degrees of acceptability

5. L, E, M, and detachment

In Section 4.2 we made a methodological decision to reject descriptive metaphysics and instead treat (i) expressing temporal reference in natural language (time_L) and real time as understood in mathematical models of spacetime (time_M) as sources of independent evidence for the concept of time (time_E). The rationale behind this assumption is that there is substantial evidence that the way we speak about an aspect of reality is strongly correlated with the concepts that speakers of a language form – to wit extensive research on the surface relativity (i.e., relativity in the neo-Whorfian sense adopted in Section 3) in the conceptualization of space. ^{Footnote 23} We have seen that the degrees of epistemic detachment can be read off linguistic expressions as they are externalized there through the lexicon, grammar, or context-driven pragmatic inference and defaults. But what is the ultimate cause of this modality? Is it merely the egocentric perspective on temporality imposed by the observer? Or is there an equivalent in time_M with which time can be correlated? A simple but negative answer would be to follow the terms of the A-/B-theoretic construal and maintain, as we did in Section 2, that reality itself is B-theoretic while the concept of time is tensed, tout court. But this is not the whole answer.

In order to attempt a fuller answer, we ought to ask first whether time_M is perspectival and as such indexical in any relevant sense of these labels. Can the perspectivity and indexicality of time_L and time_E be reconciled with the characteristics of time_M?

First, we have to remember that perspectivity affects not merely complex temporal concepts but their modal building blocks as well. Just as time is egocentric and indexical, so is the degree of commitment to the truth of a given proposition about an eventuality. The fact that the property of perspectivity is preserved all the way down allows us to shift the discussion of the correlation between the E and the M domains to the level of modal building blocks because the relevant properties are not lost: modal supervenience of time guarantees by definition that both detachment and egocentric perspective belong to the definitional characteristics of time_E.

Now we are in a position to ask about the correlates of graded detachment and perspectivity within the domain of time_M. In the most general terms, the correlate of epistemic modality is metaphysical modality: propositions are known or believed just as eventualities must or can happen. The correlate of the gradation of epistemic possibility would then be the gradation of metaphysical possibility. The latter obeys certain physical laws such as the law of increasing entropy predicted by the second law of thermodynamics. It is also restricted by mathematical models of spacetime where time cannot be explained by a fixed ordering of its units: simultaneity, for example, is on this model relative; events that appear simultaneous from one perspective may appear not synchronized when we vary the movement of the observer. ^{Footnote 24} In this general sense, time_M is modal, and it is also perspectival in the sense of fitting within the constraints of a mathematical model of spacetime.

The next question to ask is whether the degree-based epistemology is compatible with a degree-based metaphysics. There is no tense_M and no passage of time_M (Mellor, Reference Mellor1998; Le Poidevin, Reference Le Poidevin2007). In order to answer this question one would have to stipulate what the M correlate of such degrees ought to mean. To follow this up would require a different kind of enquiry and very different expertise, but suffice it to say that the cosmological ‘arrow of time’ is a quantifiable concept: the universe expands and expansion can be measured and explained in mathematical terms.

This construal leaves us with two kinds of reductionism. First, in DS, we have performed conceptual reductionism where we reduced semantics to epistemology in awarding semantic status to mental representations. ^{Footnote 25} Now we have the option of saying that the concept of time and the reality of time are interconnected in such a way that time_E, with its egocentric, perspectival degrees, maps onto time_M with its degrees assigned by the physics of spacetime. Perspectivity can be written into the L and the M levels without disturbing the correlation: on the E level it comes from the egocentric construal of time, and on the M level from the models of spacetime. And where we have the correlation, we can look for further reduction to the level of properties of reality. But while we can be fairly certain that the correlation between concepts and their linguistic expression (time_E and time_L) has the same source, to complete the reductionism we would have to say that the source of this conceptualization is the reality itself: the human concept of time is what it is because real time is what it is. By rejecting descriptive metaphysics we opened up this issue for discussion. But by focusing on egocentric perspectivity we pointed out that the issue is not so simple: memories, experiences, anticipations, and the strength of commitment are likely to have their sources both in reality and in the temporally finite human ego. Phenomenologists emphasized the latter but less has been done in philosophy of time by way of a search for correlations between the egocentric perspective and the cosmological perspective. The latter is worthy of more attention because finiteness of the human life and human mental architecture are, after all, part of a bigger picture. The theory of temporality as modality presented here may have its roots at a deeper level than the well-researched time_L/time_E interface would suggest. But, while flagging the possibility of reductionism, we are far from having a theory of this kind of supervenience.

6. Conclusion: in search for the inalienable essence of time

In human life, time passes; humans have a vivid, useful, and productive concept of passing time. I put forward a view that this concept of time is not primitive but rather is composed of simple, atomic concepts that reflect knowledge and beliefs – degrees of epistemic modality. I have also argued that this degree-based view is compatible with the metaphysics of time on which there is no time passage. If we pursue neo-Whorfianism about time further, we have to admit that the modal building blocks of the temporal concepts are either salient or can be brought to salience. Does it mean that the inalienable essence of the concept of time is the detachment from certainty? In the domain of the representation of time, the degree-based modal concepts do indeed constitute the inalienable essence. However, by virtue of the optional surfacing in consciousness, they cannot be assigned this status in the domain of the experience of time. But they are fairly robust there, as we can glean from the clines of commitment, lexicalized or grammaticalized in language, with which we talk about not only future but also past and present events. They are also robust when we consider that the experience of duration is subjective and varies with emotions and evaluation of events among other factors; that we have the aptitude to juggle different timelines associated with different roles we play in life; and add to it the synthetic rather than analytic view that humans have of their lives: we are not able to partition life into objectively delineated events, nor are we able to say where the present moment begins or ends – to wit the important role that has to be assigned to the concept of the specious present. Next, it also appears that some form of modality permeates all three relevant levels: L, E, as well as M. The overall tentative conclusion then has to be that basic epistemic modal concepts together with the de se perspective allow us to represent reality as if it was tensed_M, but only so because time_M of spacetime models is itself intrinsically metaphysically modal.

I will conclude with a quotation from a great historical novel, War and Peace. In his analysis of causes and effects of the Napoleonic wars, Leo Tolstoy observes that one method of analyzing historical change

This leads historians to investigate smaller and smaller units in striving for truth. But, he continues, varying the size of the event comes with varying the perspective from which it is observed; the degree to which perspective permeates the account is inversely proportional to the generalization over individuals, until we reach the entirely egocentric perspective on events. The fact that we can opt for more finely grained or more coarsely grained perspectives means that the perspectives and the fineness of grain are, as Tolstoy emphasizes, arbitrary. It also means that what we see as a causal link can in fact only be superimposed when this arbitrary partitioning into events is already decided upon. In reality, we are “in the middle of a shifting series of events” where “[i]mperceptibly, moment by moment, an event is carved into its meaning” (p. 825).

Leaving aside debates over the adequacy of different visions of history, what matters for us here is the importance of the perspective from which temporality is predicated: this perspective appears to be invariably dynamic and arbitrarily varying with respect to the unit of time, and (as it appears) a fortiori varying on the scale from the individual to the collective. In short, time and perspective, where the latter manifests itself in the construction of units and the assignment of properties and attitudes to them as captured by our ACC_Δ ⁿ ├ Σ, are intrinsically linked on the level of linguistic and conceptual structure – and so, arguably, on the DS-theoretic account adopted here, on the level of semantic qua conceptual structure, all without any adverse effect on what we would have to take metaphysical time itself to be. Epistemic modal building blocks appear to be an adequate and justified way to understand both this complex concept and what it stands for.