Table of Contents
2. The interpretation of an isolated past tense sentence
3. The progression of the reference time in sentence sequences
4. The Past Tense used for the expression of substitutionary speech
With the presentation of his theoretical model for the semantic analysis of tenses, Hans Reichenbach (1947: 287 - 298) supplied to linguistic research a tool that has since then been made use of in a large number of publications. His model, consisting of the three core elements point of speech (S), point of event (E), and point of reference (R), has been widely accepted to constitute an appropriate theory for the analysis of the English verbal tenses. Nevertheless, since its first publication, several attempts and proposals for modifications on Reichenbach’s model have been presented.
In most of these works, the focus is pointed on research on the past tense. A large number of linguists as well as logicians have proposed modifications concerning different temporal extensions of both reference and event time, interpretations of these two times in terms of definite and indefinite points in time, their relation to each other, and, finally, successive reference points in sentence sequences. Therefore, when Alfred Schopf (1987) presents his proposal, it has a basis of a long tradition of earlier works containing different views and proposing different approaches which can either be followed or rejected.
In this paper, Schopf discusses three aspects of the use of the English past tense and refines Reichenbach’s analysis of this tense. First, Schopf explains the information conveyed by an isolated past tense sentence by interpreting it in terms of a search instruction. Secondly, he contributes to the discussion about the progression of the point of reference in a sequence of sentences. Finally, Schopf presents his proposal for an account of the use of the past tense in substitutionary speech.
The aim of the present paper will be to discuss Schopf’s approaches in comparison to Reichenbach’s model. It will be attempted to demonstrate which elements of Reichenbach’s account have been left unchanged by Schopf and which elements, on the other hand, have been modified by him. Since not all parts of Schopf’s analysis have their origin in Reichen- bach’s theory, it will furthermore be discussed in which way he adds modules of other approaches to Reichenbach’s model and also introduces own proposals. Furthermore, each aspect of Schopf’s approach will be discussed for its ability to fulfil the aims Schopf sets in his paper.
2. The interpretation of an isolated past tense sentence
Among the basic decisions to be made for analysing tense, one of the most important is the characterisation of both the event time and reference time in terms of their extension in time and their relation to each other. In Reichenbach’s theory, the reference time is solely regarded as being punctual while the event time can either constitute a point or an interval in time. The distinction between these two possibilities depends on the aspect of the tensed verb to be analysed. For the simple past, Reichenbach assumes a punctual event time (Figure 1) while he suggests that the English language indicates an extended event time by the use of the past progressive (Figure 2).
Abbildung in dieser Leseprobe nicht enthalten
Although Reichenbach does not explicitly state that the reference time remains punctual for the past progressive, this must be the inevitable conclusion from his discussion since he solely mentions "time extension of the event” (1947: 290) and that "the event covers a certain stretch of time” (Ibid.). However, in Reichenbach’s model, any distinction between a punctual and an extended event time depends not on event notions or any similar linguistic category that would distinguish between different types of verb meaning but on aspect.
This rather limited view on possible distinctions between different types of event times without taking into considerations different types of reference time has been subject to major modifications in later publications. Schopf agrees with Reichenbach in so far as he regards the progressive aspect as indicating an extended event time in which the reference time is included (1987: 190). But beyond this one instance of an extended event time, Schopf does not agree to Reichenbach’s analysis and chooses to integrate the concept of event notions into his approach. He stresses that "the identification of the event type [...] is an indispensable operation because it determines the type of reference time [...] to be looked for” (1987: 188). In Schopf’s model, the reference time can constitute a point in time as well as an inter val and it is the type of event notion the proposition belongs to that not only influences the choice of the type of reference time but determines it.
To demonstrate this determining interaction, Schopf lists the five event notions states, punctual changes, initially determined processes, accomplishments, and achievements, analyses which type of reference time they require and in which relations this reference time stands to the event time. Of these types of event notions, punctual changes, initially determined processes, which are rather an instance of punctual changes, and achievements require a punctual reference time while accomplishments have to be combined with a reference interval. States, on the other hand, can be linked to both a punctual or a extended reference time since the peculiarity of this event notion is not the type of reference time but its relation to the event time, which "must necessarily surround, include or overlap it” (Schopf 1987: 181). By means of this relation (R _ E), states differ considerably from the other four event notions in which the event time is included in or identical with the reference time (R r E).
The modifications that Schopf applied to Reichenbach’s model become apparent by analysing a past tense sentence that expresses a state and in which a time adverbial supplies a definite reference time. While in Reichenbach’s theory the example sentence would be interpreted like every other simple past sentence and hence would be regarded as consisting of a punctual event time and a punctual reference time (Figure 3a), it would be accounted for by Schopf’s approach as having a punctual reference time included in an event time that lasted on anterior to reference time and lasts on posterior to it (Figure 3b).
Abbildung in dieser Leseprobe nicht enthalten
Surely the analysis following Schopf’s account describes more accurately the situation conveyed by the example sentence than the analysis following Reichenbach’s model does since in Figure 3a, the event time anterior and posterior to the reference time is totally neglected. But this difference between the interpretations by the two accounts resulting out of a deficiency in Reichenbach’s theory can easily be explained by the simple fact that the idea of event notions, as originally introduced by Zeno Vendler (cf. 1968: 106 for the time schemata of states and other event types), was presented after Reichenbach had published his theory and hence was not known to him. But by embedding Vendler’s concept of event notions into Reichenbach’s tense model and the definition of the relations between the event times and reference times of the various event notions, Schopf modifies Reichenbach’s theory in a way that enables it to yield, at least for the analysis of states, more appropriate results.
Another basic decision to be made in an approach towards the interpretation of past tense sentences is whether the event time and the reference time respectively to be looked for are to be characterised as in terms of definite or indefinite points in time. When such an approach is based on a definite interpretation, it will try to localise one particular point in time while it would, when based on an indefinite interpretation, consider the proposition expressed by the sentence to be true at some time within an interval or would employ paraphrases such as "There is at least one time in the past at which...” (cf. Schopf: 179f.). In Reichenbach’s theory, this characterisation is not made and since it is, with regard to linguistic research since then, hardly possible to introduce a model for the interpretation of past tense sentences without deciding whether it is based on definite or indefinite event times and reference times, Schopf has to rely on theories developed by other linguists following the Reichenbachian tradition.
For this decision, Schopf points out (1987: 180) that his approach is influenced by several proposals on the interpretation of past tense sentences made by Barbara Partee (1973, 1978, 1984). He follows her suggestion to regard a past tense sentence as referring to a "definite interval whose identity is generally clear from the extra-linguistic context” (Partee 1973: 603). Partee regards, first of all, the reference time and event time as constituting intervals instead of points in time but, what is more important for Schopf’s model, she also considers the proposition in a past tense sentence as being true at, and referring to, definite times. This view is contrary to other approaches from tense logicians (cf. Schopf 1987: 179f.) in which the past tense is regarded as referring to all times posterior to speech time and, thus, to an indefinite time interval.
Partee’s example sentence I didn’t turn off the stove can be, strictly logical, interpreted as referring to all time from the day of birth of the person uttering this sentence on to the point of speech since the only temporal information given is the past tense of the verb. If the sentence is regarded in isolation, there is, therefore, nothing to object against such an analysis. Partee, on the other hand, refers to the extra-linguistic context of the sentence "when uttered, for instance, halfway down the turnpike” (1973: 602) and thus chooses to pay attention also to the communicative and pragmatic circumstances of the utterance. The interpretation of I didn’t turn off the stove as referring to an indefinite time is only possible if the situation in which it is uttered is neglected and it would also miss to express the information the speaker wishes to convey.
The problem in Partee’s interpretation and also in Schopf’s approach is to receive the temporal information needed in order to determine the definite time to which a past tense sentence refers. One possible solution for at least some sentences is the phenomenon which Partee labels as temporal anaphora and which she demonstrates as in (4).
(4a) Sam is married. He has three children.
(4b) Sheila had a party last Friday and Sam got drunk.
(Partee 1984: 245)
The sentences in (4a) show what is commonly known as nominal anaphora insofar as the pronoun he refers backwards to the noun Sam which has been introduced before. In (4b), a very similar anaphora can be found for the interpretation of the sentence’s last event Sam got drunk since it is not accompanied by a time adverbial that would provide some information on the time at which this event occurred. This information is provided, instead, by the adverbial last Friday, which accompanies the preceding event Sheila had a party, but which also provides the time frame for the second event. For past tense sentences making use of temporal anaphora, Partee comes to the conclusion that "the past tense can be viewed as an anaphoric element inasmuch as it is not understood as meaning ‘at some time in the past’, but as referring to some relatively definite past time, the specification of which is provided by a non-linguistic or linguistic antecedent” (1984: 245). Temporal anaphora clearly is a linguistic antecedent and therefore relatively easy to define, compared to non-linguistic antecedents, which would have to account for the understanding of a sentence like I didn’t turn off the stove when uttered in a car after having left from home. This would be no case of temporal anaphora but nevertheless, the hearer would understand that this utterance means that the stove is still, at the very moment, switched on. The explanation for this intuitive interpretation in terms of a definite time is much harder to give and treated rather vaguely by Partee with her reference to a non-linguistic antecedent.
In Schopf’s model, the first step in the complex task of the interpretation of a past tense sentence is the identification of the type of event notion the sentence’s verb belongs to. But according to Schopf, such a sentence "would remain uninterpretable if a specific reference time were not supplied either co-textually or contextually” (1987: 186). The latter distinction stresses that the reference time of an isolated past tense sentence that consists only of a subject and a verb cannot be determined if no additional information is given. Normally, this additional temporal information is transmitted by a time adverbial the existence of which in a sentence containing a tensed verb turns this sentence into a "fully tensed sentenced, i.e. a sentence containing tense and time adverbials” (Schopf 1987: 188). For the interpretation of a fully tensed sentence, Schopfs regards it as the "ultimate goal to be attained” (Ibid.) to "locate the event reported by it in time” (Ibid.) or, put differently, to determine the event time, which can only be achieved by a former determination of the reference time as well as the relation between these two times, the latter being defined in Schopf’s account of the temporal relations for the various event notions.
Schopf proposes to regard the processing of the interpretation of a past tense sentence as following a search instruction by which the interpreter would "look out for a co- textually or contextually given specific time in the past” (1987: 186f.). By stressing that this given time has to be specific, he expresses his decision to have his approach, like Partee’s, based on the concept of a definite reference time and event time and to reject any approaches that advocate the idea of indefinite time intervals. The task of a search instruction would be to narrow down the possible time span in which the proposition is true as much as possible rather than to consider it to be true within a large interval that could contain, in its largest extent, all time anterior to speech time. But Schopf is aware of the fact that not all past tense sentence allow an interpretation in terms of a definite reference time that can be located on the time axis as the first example discussed by him shows.
A past tense sentence containing a time adverbial as in (5a) would be represented as in (5b) to demonstrate the order in which it is interpreted by the search instruction.
(5a) John lost his watch yesterday.
(5b) Yesterday (Past (John lose his watch))
(Schopf 1987: 187)
The interpretation starts with the innermost bracket and the identification of the type of event notion the untensed proposition belongs to. In this example, it would be identified as a punctual change requiring a punctual reference time to be looked for (cf. Schopf 1987: 182). After having completed this first task, but not yet having found an appropriate reference point, the next information Past instructs the hearer to look for it in the past. This very vague time span is specified by the next temporal parameter yesterday, which limits the time interval in which the reference time is to be located. But since this time span cannot be filled by a reference time that is defined as being punctual, the hearer would have to narrow down also this interval. But this final task cannot be accomplished since the hearer lacks any additional information about a definite point in time contained in the interval yesterday. Therefore, the reference time has to remain indefinite since the information conveyed by this sentence is that the watch was lost at some unspecified point of time included in the interval yesterday.
Due to a different type of even notion, the example in (6a) as represented in (6b) requires a slightly modified interpretation.
(6a) At 8 o’clock it was raining hard.
(6b) At 8 o’clock (Past (in rain fast [sic]))
(Schopf 1987: 187)
The event notion of the proposition in the innermost bracket would be identified as a state, which can be related to an either punctual or extended reference time (cf. Schopf 1987: 181f.). Again, the hearer is instructed by the parameter Past to look for this yet unspecified reference time in the past and receives more detailed information from the time adverbial At 8 o’clock, by which the reference time is defined as being punctual. As Schopf pre-defines (Ibid.), this reference point is included in the surrounding event time as an, in this case punctual, sub-interval of the larger interval constituted by the event time.
A third variant form of temporal relations between reference time and event time can be observed in (7a), represented by (7b).
(7a) From ten to eleven I wrote a letter to my father.
(7b) From ten to eleven (Past (I write a letter to...))
(Schopf 1987: 188)
The result of the identification of the type of event notion the proposition in the innermost bracket belongs to is that is an accomplishment. This type of event notion is defined by Schopf as consisting "of a momentary initiation phase, a cumulative process phase and a momentary termination phase” (1987: 182). He furthermore points out that these three phases constitute an indivisible whole which is related to a reference interval with the same extension as the three phases of the event time. Applied to this example, both the event time and the reference time, which is identical with it, extend over a time span defined by the time adverbial From ten to eleven. The temporal information Past provides the information that the hearer has to look for this time span in the time anterior to speech time. But Schopf comes to the conclusion that, because of insufficient information about the position of this time span on the time axis, it is not possible to localise the reference time and, therefore, “additional information we need would have to be supplied by the speech situation” (1987: 188).
Search instructions can also be summarised in a formula. For the sentence (5a), Schopf (1987: 189) develops the formula in (8):
(8) 3t (t = R . t < S A t _ I <yesterday> A t q E <John lose his watch>)
According to this formula, the aim of a search instruction it to locate a time (t) in time. The information given by the past tense of the example sentence suggests that both (t) and (R) are anterior to (S). This is, then, specified by the time adverbial which lets (t) be included in an interval defined as yesterday. Finally, the event is, by definition of the temporal relation for its event notion, included in (t).
Due to the integration of event notions into his concept, Schopf’s search instructions are a more elaborated approach towards the determination of the reference time than Rei- chenbach’s original model is. For the interpretation of sentences containing a time adverbial, Reichenbach postulates the principle of the positional use of the reference point (1947: 294). In this principle, he states that a time adverbial refers "not to the event, but to the reference point of the sentence” (Ibid.). The event time receives its position in time from its relation to the reference time, but before this relation can be defined, the reference time has to be located with the help of the information given by the time adverbial. Since Schopf’s search instruction follow the same principle, this is a element taken from Reichenbach’s theory. But the advantage of a search instruction becomes obvious by comparing Schopf’s analysis of the example sentence John lost his watch yesterday as in (5) to Reichenbach’s discussion of the sentence I met him yesterday (1947: 249) with respect to the time adverbial yesterday constituting an interval in time. Reichenbach is not aware of the impossibility of a punctual reference time filling a time interval and, therefore, allows yesterday to function as a reference point. He also does not mention whether he considers the reference point to be definite or indefinite. This weak point in Reichenbach’s model is improved by Schopf’s embedding of event notions into his search instructions and the definition of different types of temporal relations between reference time and event time.
Schopf also relies on Partee’s suggestions on the interpretation of past tense sentences and her concept of temporal anaphora and modifies them by having the type of reference time to be looked for depend on the type of event notion the sentence’s verb belongs to. By maintaining Partee’s proposal of a definite reference time and of linguistic and non- linguistic antecedents that determine the localisation of the reference time and the event time, he adopts two major elements of her approach. Moreover, they constitute an integral part of Schopf’s model which he transforms into his more practical idea of a search instruction.
But like Partee, Schopf does not succeed in accounting for the human ability of localising an event on the time axis the position of which is not or not sufficiently determined by a time adverbial. Search instructions as developed by Schopf can, in many cases, only be used to narrow down the interval in which the reference time and the event time are located but they cannot account for Partee’s example I didn’t turn of the stove, which is interpreted by almost every hearer in the way that the stove is understood as still being switched on during speech time.