III. Arguments for distinguishing between the absolute and the relative past tense 385 This test is corroborated by the fact that it makes the correct predictions in cases in which the relevant T-relation is T-anteriority rather than T-simultane- ity: John expects that he will feel lonely when Mary {has left /*will have left}. John expected that he would feel lonely when Mary {had left /*would have left}. Compare also: [The thought of his case never left him now. He had often considered whether it would not be better to draw up a written defence and hand it in to the Court.] In this defence he would give a short account of his life, and when he came to an event of any importance he would explain for what reasons he had acted as he did, inti- mate whether he approved or condemned his way of action in retrospect, and adduce grounds for the condemnation or approval. (TTR) (The past tense forms came, did, approved and condemned are all relative tense forms expressing T-simultaneity with the situation time of their head clause. Note that initimate is short for would inti- mate.) (…) In this defence he will give a short account of his life, and when he {comes / *will come} to an event of any importance he will explain for what reasons he has acted as he has, intimate whether he approves or condemns his way of action in retrospect, and adduce grounds for the condemnation or approval. (The forms comes, has acted, has, approves and condemns all express a T-relation in a post-present domain. The present tense forms all express T-simultaneity with the situation time of their head clause.) 8.23.2 This test rests on the assumption that the possibilities and restrictions in connection with the use of absolute and relative tense forms in particular types of clauses are the same in a past domain as they are in a post-present one. Though the above examples all corroborate this assumption, we have not yet adduced any really conclusive evidence that the assumption is correct. The test must therefore provisionally be treated as tentative. 8.24 Argument 2: past tenses whose situation time is W-posterior to t 0 Unlike the first argument, this argument is a cogent one. In a sentence like Mary told me on the phone last night that at the dinner party tomor- row she would suddenly say that she felt sick, the past tense form felt refers to a situation which lies in the post-present (i. e. which is W-posterior to t 0 ). This observa- tion demonstrates the inadequacy of the assumption that there is only one past tense 386 8. Temporal domains and relative tenses: theoretical foundations in English, which is used to locate a situation time prior to t 0 . However, the observation is naturally explained on the assumption that felt is a relative past tense form expressing T-simultaneity. This constitutes very strong evidence in favour of a theory which as- sumes the existence of a relative past tense (next to an absolute one), which expresses T-simultaneity in a past domain, irrespective of the W-location of the binding time relative to t 0 . 8.24.1 We have argued that the tense structure (ϭ semantics) of the two kinds of past tense are quite different from each other: the semantics of the absolute preterite is ‘The situation time establishes a past domain’, while that of the relative preterite is ‘The situation time is represented as T-simultaneous with a binding orientation time belonging to a past domain or subdomain (or to a ‘pseudo-past subdomain’ Ϫ see 9.9.1 and 9.18.1)’. Since it is not the existence of the absolute preterite that is subject to debate, but the existence of the relative preterite, what we need to show is that there are preterite forms that unequivocally express the tense structure typical of the relative preterite. In our opinion, the italicized tense forms of the following examples fulfil this requirement Ϫ see Figure 8.6: [A few weeks ago he told us about what he planned to do.] When he was fifty-five he would retire. [He would move to some place in the south and …]. [Yesterday the Prime Minister repeated that] if the party lost next week’s elections, this would not induce it to change its views on nuclear energy. Until a moment ago I thought that at tomorrow’s meeting of the board he would argue that it was necessary to sell the company. [But he’s just assured me that he isn’t going to do that.] Mary told me on the phone last night that at the dinner party tomorrow she would suddenly say that she felt sick. Figure 8.6. The tense structure of Mary told me on the phone last night that at the dinner party tomorrow she would suddenly say that she felt sick. III. Arguments for distinguishing between the absolute and the relative past tense 387 In these examples, the situations described by was [fifty-five], lost, was [neces- sary] and felt [sick] are interpreted as W-posterior to t 0 (because of the tempo- ral indications in the relevant clauses), but this W-posteriority relation is not expressed by the past tense forms. This means three things. First, the italicized preterites cannot be analysed as absolute tense forms in any tense theory, be- cause the absolute preterite, 11 which is the only kind of (nonmodal) preterite recognized by most tense theories, locates its situation time anterior to t 0 and is therefore semantically incompatible with an interpretation in which the situ- ation in question lies completely in the post-present. Secondly, in the above examples, the italicized preterites not only fail to represent their situation time as T-anterior to t 0 , they fail to represent it as T-anterior to anything. (This means that T-anteriority does not form part of their meaning.) Thirdly, al- though the past tense forms themselves do not express the W-posteriority rela- tion, they are not incompatible with it. This is in keeping with our analysis of these forms as being relative preterite forms: these tense forms merely represent their situation time as T-simultaneous (ϭ coinciding) with a situation time belonging to a past domain. (The conditional tense represents that T-binding situation time as T-posterior to the central orientation time of the past do- main.) Thus, in the third example was [necessary] represents its situation time as T-simultaneous with the situation time of would argue, which is a situation time belonging to the set of situation times constituting the temporal domain established by thought. It should be kept in mind that a tense form representing its situation time as T-posterior to the central orientation time of a past domain does not say anything about the W-relation between the relevant situation and t 0 : as is clear from He said he would do it {yesterday / now / tomorrow}, the situation referred to by the conditional tense (which represents its situation time as T-posterior to an orientation time in a past domain) may be interpreted as W-anterior, W-simultaneous or W-posterior to t 0 . Obviously, the same is true of any situation whose situation time is represented as T-simultaneous (ϭ coinciding) with the situation time of a situation referred to in the conditional tense, as the following example illustrates: He said he would do it when he had time. [So he may already have done it, or he may be doing it right now, or he may be going to do it tomorrow. I don’t know which.] In this example, had is another instance of a past tense form which cannot possibly be analysed as an absolute preterite, nor as a past tense representing its situation time as T-anterior to an orientation time other than t 0 , but which fully satisfies our definition of ‘relative past tense form’ (expressing T-simulta- 11. As noted in 1.14, we use ‘preterite’ to refer to forms like walked, said, went, etc. Other forms involving the past tense morpheme (such as had walked, would walk, would have walked) are not preterite forms. 388 8. Temporal domains and relative tenses: theoretical foundations neity in a past domain). More generally, as is shown by Figure 8.6, a past domain can contain one or more subdomains which are W-interpreted as lying in the post-present. A past tense expressing T-simultaneity in such a subdomain cannot be analysed as an absolute past tense. The following are some further examples in which the situation time of a clause in the relative past tense is interpreted as W-posterior to t 0 : [He told me yesterday that exactly 17 days from now he would tell Mary that] he was inviting her to lunch because it was his birthday. [“What was it he told you yesterday?” Ϫ “He said that, inevitably, the time would come when] he was fed up with working in a factory [and would look out for a better job.”] [He said he would read the article tomorrow while] he was in the dentist’s waiting-room. Here again, the italicized past tense forms not only do not mean ‘past relative to t 0 ’ (as they would do if they were absolute preterites), but do not mean ‘past relative to (any kind of) orientation time’. The temporal relation they express is that of T-simultaneity with a situation time belonging to a past temporal domain (but interpreted as W-posterior to t 0 ). 8.24.2 The above data, which are naturally explained if one assumes the exis- tence of a relative past tense in English (and hence the existence of temporal domains), cannot possibly be accounted for by an analysis that assumes that there is only one past tense, expressing that the situation time is past relative to t 0 . This argument in favour of the existence of a relative preterite is so cogent that it should be sufficient on its own to discard any analysis denying the existence of a relative past tense in English. 8.24.3 All the examples given in 8.24.1, and in fact in all the other arguments that will be presented below, are instances of ‘represented speech’, i. e. indirect reported speech or free indirect speech (see 8.25.1). This is because represented speech forms the only context in which the use of the conditional tense (to express T-posteriority) is natural. Expressing T-posteriority is tantamount to making a prediction or expressing an expectation, and this can only be done in an intensional (opaque) 12 context like represented speech. 13 (A similar 12. An opaque context (or ‘intensional context’) is one in which the reference is not to the real world but to an alternative (nonfactual) world. Such a context is created by (amongst other things) ‘intensional verbs’ (‘verbs of propositional attitude’) like want, expect, believe, think, imagine, etc. Clauses that form part of an intensional context receive an ‘opaque’ (‘de dicto’) interpretation, i. e. their truth is not evaluated in relation to the real world but in relation to the alternative world referred to. Clauses belonging to a nonintensional context receive a ‘transparent’ (‘de re’) inter- pretation, i. e. their truth is evaluated in relation to the real world. It is typical of such an interpretation that the truth value of the clause is not affected when a referring III. Arguments for distinguishing between the absolute and the relative past tense 389 remark can be made in connection with the future tense: reference to the post- present by itself creates opacity, i. e. a post-present domain is by definition an opaque domain. Thus, in [If you do that,] you will regret for ever that you have done it, the ‘doing it’ is not factual in the speaker’s t 0 -world (ϭ the actual world): it is only factual in the post-present intensional domain, i. e. in a not- yet-factual alternative world.) However, the fact that the data are restricted to cases in which T-posteriority is expressed does not in any way affect the strength of our argument. As pointed out at the beginning of 8.24.1, all that is necessary to prove the exis- tence of the relative preterite in English is to adduce examples of preterites that cannot possibly be analysed as absolute preterites, and a number of such exam- ples have been given here. 8.24.4 It should also be noted that the examples given in 8.24.1 all have nonpast counterparts in which the future tense is used instead of the condi- tional tense in the head clause and in which the present tense is used instead of the relative past tense in the subclause: [He has told us about what he plans to do.] When he is fifty-five he will retire. [He will move to some place in the south and …]. I will read the article tomorrow while I am in the dentist’s waiting-room. [He says that, inevitably,] the time will come when he is fed up with working in a factory [and will look out for a better job.] Exactly 17 days from now I will tell Mary that I am inviting her to lunch because it is my birthday. There can be no doubt that the italicized present tense forms in these examples express T-simultaneity in a post-present domain: these forms cannot be inter- preted as locating their situation time at t 0 . Since these sentences with the future tense in the head clause and the present tense in the subclause are the nonpast counterparts of examples involving the conditional tense in the head expression in the clause is replaced by an ‘identical’ expression (i. e. by an expression with the same referent). Thus, since in the real world the capital of France and Paris have the same referent, we can replace the former by the latter in The capital of France has ten million inhabitants without altering the truth value of the sentence. In sentences that receive an opaque interpretation, the replacement of a term by an ‘identical’ expression may affect the truth value: the sentence Bill thinks that Paris is the capital of Spain may be true even if Bill thinks that the capital of France is the capital of Spain is not true. 13. We are disregarding here the special use of would in the sense of was to,asinTen years later Tom {would / was to} be the richest man in town Ϫ see 9.6.7. (This is the only use in which would is interpreted as referring to a past fact rather than as merely making a prediction.) 390 8. Temporal domains and relative tenses: theoretical foundations clause and the past tense in the subclause, it stands to reason that in both cases the tense in the subclause should be analysed as a tense expressing T-simultane- ity (coincidence) with a situation time that is T-posterior (to t 0 or to a past orientation time, respectively). 8.25 Argument 3: tense use in represented speech Distinguishing between an absolute and a relative past tense explains why some past tense forms are ‘backshifted’ in past represented speech while others are not. For example, the sentence Ian addressed the girl who was sitting behind the largest desk can be reported as [Jim said yesterday that] Ian had addressed the girl who was sitting behind the largest desk. This illustrates the (quite logical) fact that an absolute past tense can be ‘back- shifted’ in past represented speech, whereas a relative past tense remains unaffected (be- cause the T-simultaneity relation which it expresses remains unaffected). 8.25.1 This argument concerns the adaptation of tense forms in indirect re- ported speech and in free indirect speech. For ease of reference, we will sub- sume these two possibilities of indirect speech under the heading of repre- sented speech (where speech should be interpreted as meaning ‘speech or thought’). Since we will only be concerned with the kind of represented speech in which the representation is located in the past, the term past represented speech will be used in the present section. 8.25.2 In past represented speech, the tense forms from direct speech are traditionally said to be backshifted: “I am at home.” J I said I was at home. “I will be at home.” J I said I would be at home. “I have been at home.” J I said that I had been at home. However, this rule, known as ‘sequence of tenses’, does not automatically apply when the presumed original utterance is in the past tense. Consider the follow- ing stretches of direct speech: (1) “I told Bill that Jane was angry with Jim.” (2) “I told Bill that Jane was furious when Jim refused to take her out.” Out of context, sentence (1) will normally be taken to mean that the two situa- tions were simultaneous, and this meaning can only be reported as (3) or (4): (3) Jill said that she had told Bill that Jane was angry with Jim. (4) Jill said that she told Bill that Jane was angry with Jim. III. Arguments for distinguishing between the absolute and the relative past tense 391 The following are not ‘correct’ past represented speech versions of (1) if the two situations in (1) are taken to be simultaneous: 14 (5) Jill said that she told Bill that Jane had been angry with Jim. (6) Jill said that she had told Bill that Jane had been angry with Jim. Both (5) and (6) are reports of I told Bill that Jane had been angry with Jim. By contrast, sentence (2) only allows the following reading: 15 ‘I told Bill [at t 2 ] that Jane was furious [at t 1 ] when Jim refused [at t 1 ] to take her out.’ (t 1 is anterior to t 2 ) The following are possible past represented speech versions of (2) Ϫ see 8.25.3 for further comment: (7) Jill said that she (had) told Bill that Jane had been furious when Jim refused to take her out. (8) Jill said that she (had) told Bill that Jane was furious when Jim refused to take her out. (9) Jill said that she (had) told Bill that Jane had been furious when Jim had refused to take her out. However, the following sentence is not a correct past represented speech ver- sion of (2): (10) Jill said that she had told Bill that Jane was furious when Jim had refused to take her out. (Unlike (2), this report represents Jim’s refusal as anterior to Jane’s being furious. It is therefore a report of I told Bill that Jane was furious when Jim had refused to take her out.) 8.25.3 The data provided by (3)Ϫ(10) cannot be accounted for in a systematic way (i. e. without ad hoc stipulations) by the sequence of tenses rule if the latter is seen as a purely formal operation, which applies automatically in past represented speech. However, they are naturally explained in our analysis, which treats the backshifting rule as the semantically-driven adaptation of 14. By ‘correct’ represented speech version we mean a version which does justice to the temporal relation(s) expressed in the original utterance. If the original speaker says that two situations are simultaneous, then a report of this speaker’s utterance cannot be a genuine report if it represents the original speaker as having said that the two situations are sequential. So if one thinks that the original speech is John knew that Jim was abroad, and this is interpreted as meaning that Jim’s being abroad was simultaneous to John’s knowing it, then this simply cannot (except by a liar) be reported by Betty said that John knew that Jim had been abroad, because the temporal relations between the situations are different in this sentence. Either a sentence is a report or it is not, and if it is a report then it obviously has to convey the same situations in the same temporal order. 15. We are ignoring the (unlikely) interpretation on which the when-clause specifies the time of the head clause situation. . t 0 . Since these sentences with the future tense in the head clause and the present tense in the subclause are the nonpast counterparts of examples involving the conditional tense in the head expression. t 0 . 8.24.1 We have argued that the tense structure (ϭ semantics) of the two kinds of past tense are quite different from each other: the semantics of the absolute preterite is The situation time establishes. the situation time of their head clause.) 8.23.2 This test rests on the assumption that the possibilities and restrictions in connection with the use of absolute and relative tense forms in particular types