2.2 Seminal Work
2.2.1 Tense
According toComrie (1985), tense is “the grammaticalized expression of location
2.2 Seminal Work
rily and that it is morphologically bound. That is, tense is typically conveyed by inflectional morphology rather than by separate words or phrases that are optional in a sentence (such as adverbs or adverbial phrases), even though the latter can also be employed to describe time. Many languages, such as English or Portuguese, use verbal morphology to encode tense.
The description of the meaning of tense owes much to the work of Reichenbach
(1947). He observes that three points in time are necessary to account for past
perfect forms, such as the sentence in (7).
(7) John had left on Monday.
These three points are the point of speech S (when the sentence is uttered), the point of the event E (when the event being described occurred) and the point of reference R (a third point that can sometimes be described by modifiers such as the
phrase on Monday in (7)). A sentence with a past perfect verb can be ambiguous.
For instance, the sentence in (7) has two readings. In one reading, the event of John
leaving occurs on Monday. In the other reading, this event has already happened on Monday, i.e. it precedes Monday. In both readings, E is when the event of John leaving takes place. In the first reading, the phrase on Monday is used to describe the temporal location of E, but in the second reading it describes R.
Reichenbach’s contribution consists in (i) defining these three times, (ii) using this three-point system for all tenses, (iii) defining the meaning of the different tenses through temporal relations between S and R on the one hand and R and E on the other—the temporal relation between S and E is not represented directly—and (iv) resorting to only two temporal relations: simultaneity and precedence.
Table2.1presents Reichenbach’s analysis of the English tense system. Each cell
in that table corresponds to a combination of two temporal relations that results in a specific semantic value of tense (simultaneous present, simultaneous future, etc.), which in turn is associated with different grammatical tenses (such as the English simple present, simple future, etc.). Temporal simultaneity is represented with a comma (,) and temporal precedence with a dash (−). It must be noted that English makes no grammatical distinction between posterior present (e.g. Now I shall go, where now identifies R) and simultaneous future (e.g. I shall go tomorrow, where
S, R S − R R − S simultaneous present simultaneous future simultaneous past
E, R simple present: simple future: simple past:
I see John I’ll see John I saw John
posterior present posterior future posterior past
R − E simple future: (none) conditional:
I shall go I would see John
anterior present anterior future anterior past E − R present perfect: future perfect: past perfect: I have seen John I’ll have seen John I had seen John Table 2.1: Reichenbach’s representation of English tenses
tomorrow is R), lacks a posterior future, and the conditional can be used in many different ways besides denoting a posterior past.
Reichenbach’s analysis offers a natural account of the difference between the En-
glish simple past and present perfect. The contrast in (8) and (9) shows that while
the simple past in (8) can combine with time expressions denoting a past time, such
as last month and 1957, this is not the case of the present perfect, as the ungram-
matical examples in (9) illustrate.1 This is because these temporal expressions (last
month and 1957) arguably refer to R.
(8) a. I visited the Parthenon last month. b. I visited the Parthenon in 1957.
(9) a. * I have visited the Parthenon last month.
b. * I have visited the Parthenon in 1957.
Reichenbach’s theory has received some criticism, one of the reasons being that it allows more tenses than the ones usually found in natural languages.
Nevertheless, this sort of decomposition of tenses as involving more points in time than just the speech time and the event time has been enormously influential in the
subsequent literature on tense and aspect. For instance,Comrie(1985) distinguishes
1
We follow the common practice in the linguistics literature of presenting ungrammatical ex- amples preceded by a star (*).
2.2 Seminal Work
absolute tense (simple present, simple past and simple future) and relative tense (present perfect, past perfect and future perfect), and argues that only the latter
need to be represented with a reference point. Prior(1967) suggests the use of two
reference points R1 and R2 in order to account for examples like I shall have been
going to see John, with the tense structure S − R2− E − R1. Hornstein(1990) uses Reichenbach’s system to explain the possible combinations of tenses with temporal adverbs in English. An adverb like now is represented with the semantics of the simultaneous present, yesterday is considered similar to the simultaneous past, and tomorrow is akin to the simultaneous future. He focuses on why combinations like John leaves tomorrow are possible (where the simple present tense form leaves gets a future interpretation), whereas combinations like * John has left yesterday are not.
Another line of research is concerned with the temporal flow of discourse. Las-
carides & Asher (1993) present a formal account of how to determine discourse relations between propositions introduced in a text, and the relations between the events they describe. They seek to explain the different temporal orderings in nar- ratives such as:
(10) a. Max stood up. John greeted him.
b. Max fell. John pushed him.
In (10a) the first sentence describes a situation that temporally precedes the
situation described in the second sentence. In (10b) the temporal ordering is the
opposite. The authors explain this difference through defeasible constraints (φ > ψ “φ normally entails ψ”) and non-monotonic inference:
• Defeasible Modus Ponens φ > ψ, φ |= ψ
E.g., birds normally fly, Tweety is a bird |= Tweety flies • Penguin Principle
φ → ψ, φ > ¬χ, ψ > χ, φ |= ¬χ
E.g., penguins are birds, penguins normally don’t fly, birds normally fly, Tweety is a penguin |= Tweety doesn’t fly
The sentence in (10a) illustrates the default interpretation of discourse according to which situations happen in the temporal order in which they are described in
a discourse. This is defeasible, however, as (10b) exemplifies. Here, the reverse
temporal ordering is imposed by a causation relation between push and follow. The piece of knowledge that causes precede effects is not defeasible.
Precisely determining the temporal relation holding between situations men- tioned in consecutive sentences can, however, depend on more factors. For instance,
tense seems to have anaphoric properties (Webber, 1988): just like pronouns pick
up entities previously introduced in a discourse, verb tense can also refer to times
previously mentioned. In (11), the second sentence picks up the event time of the
first sentence (the two playing events happen at the same time).
(11) a. John played the piano.
b. Mary played the kazoo.