392 8. Temporal domains and relative tenses: theoretical foundations tense forms which naturally takes place when an utterance is reported in the form of past represented speech. This analysis hinges on the following claims: (a) A situation time that is ‘originally’ (i. e. in the original direct speech utter- ance) T-related to t 0 (and thus establishes a domain of its own), is now (i. e. in past represented speech) T-related to the central orientation time of a past domain (which, in (3)Ϫ(4) and (7)Ϫ(10), is the situation time of said), which means that the tense which locates the situation time in time no longer establishes a domain but merely expresses a T-relation in one. (b) It follows that a tense form which originally represents its situation time as the central orientation time of a past domain, such as told in (1)Ϫ(2), is now replaced by a tense form expressing a T-relation in a (new) past domain. This is because the situation time is now T-bound by the central orientation time Ϫ which is the situation time of said in (3)Ϫ(4) and (7)Ϫ (10) Ϫ rather than being directly related to t 0 (by an absolute tense form). The relative tense form in question expresses T-anteriority (i. e. it is a past perfect form) because the original past tense form also expresses T-anteri- ority (but then to t 0 rather than to a past orientation time). (c) Out of context, when a complement that-clause in the past tense depending on a head clause in the past tense is homogeneous, it must as a rule be taken to express T-simultaneity. This is indeed the way the that-clause is interpreted in (1) [I told Bill that Jane was angry with Jim]: Jane’s being angry is W-simultaneous with my telling Bill about it. For a good understanding of this point, it is necessary to draw attention to a number of things: Ϫ Firstly, the generalization made in the first sentence of (c) obviously does not apply if the past tense form in the that-clause is the past counterpart of a ‘futurish form’ Ϫ see 2.9 Ϫ as in Mary said that Bill was coming over when school broke up the next week. Ϫ Secondly, the reason why the generalization applies only if the that-clause is homogeneous is that T-simultaneity means coincidence (see 2.17). If the situation time of a durative situation is represented as T-simultaneous with a punctual situation time, it is represented as punctual. As explained in 2.17, this is only possible if the durative situation clause (whose situation time is T-bound and punctual) is a ‘homogeneous’ clause. (If it is hetero- geneous, the situation time can only be the time of the full situation.) Thus, the situation time of the that-clause of Mary said that Bill walked to town cannot be T-simultaneous with the situation time of said, but that changes when the that-clause is made homogeneous by the use of the progressive form: in Mary said that Bill was walking to town the walking can (and out of context is) interpreted as simultaneous with the situation time of said. III. Arguments for distinguishing between the absolute and the relative past tense 393 Ϫ Thirdly, if a that-clause is itself the head clause of an adverbial when-clause, its situation time must be interpreted as W-anterior to the situation time of the matrix clause (see 2.17 and 8.26.1). Thus, in (2) [I told Bill that Jane was furious when Jim refused to take her out] the situation of Jane being furious is interpreted as W-anterior to that referred to by I told Bill. In our terminol- ogy, this means that was is now an absolute past tense form ‘shifting the do- main’. That is, the temporal domain established by told in I told Bill remains unexpanded, because the past tense form was in Jane was furious establishes another (W-anterior) past domain which also comprises the situation time of the when-clause. (The reason why the preterite in a that-clause supporting a when-clause cannot be analysed as a relative preterite expressing T-simulta- neity will be explained in 8.26.1.) (d) In section 9.29 we will see that some types of clauses (such as adverbial when-clauses) under certain conditions allow ‘indirect binding’, i. e. the situation time in question is not T-bound by the situation time of its head clause, but by the situation time of a higher clause. The most typical pattern is that realized in He said that when he had seen the house he had immediately fallen in love with it, which is an alternative to He said that when he saw the house he had immediately fallen in love with it. Whereas saw is an example of ‘normal’ (ϭ ‘direct’) binding, because it represents the situation time of the when-clause as T-simultaneous with the situation time (had fallen) of its head clause, had seen is an example of indirect binding: it represents the situation time of the when-clause as T-anterior to the situation time of said. This is possible because it is pragmatically clear that the situation time of had seen must anyhow be interpreted as W-simultaneous with the situation time of had fallen. (e) Indirect binding is never possible in complement clauses depending on a verb of communication (say, tell, hear …): we cannot interpret Jill admit- ted that she had said that Ian had been angry as equivalent to Jill admitted that she had said that Ian was angry. 16 (However, the rule is not so strict 16. This impossibility of indirect binding in past represented speech accords with the impos- sibility of using the future tense in similar examples referring to the post-present: (i) (direct reported speech) Next time I meet him I will say to him: “Some day I will ask you if you are hungry.” (ii) (indirect reported speech) Next time I meet him I will tell him that some day I will ask him if he {is / *will be} hungry. The report in (ii) must do justice to the original temporal relations. This means that the absolute form will ask from (i) becomes the relative tense form will ask (expressing T- posteriority to the future situation time of will talk, which is treated as a ‘pseudo-t 0 ’Ϫ see 9.17.1 and 9.22.1) and that the relative present tense form are from (i) remains the relative present tense form are, because the T-relation it expresses, viz. T-simultaneity, remains unchanged. Using will be instead would be an instance of indirect binding, which is ungrammatical in complement clauses. 394 8. Temporal domains and relative tenses: theoretical foundations if the ‘reporting verb’ is a cognition verb (e. g. realize, be aware, feel, know… Ϫ see 9.23.3.) The above principles explain the data provided by (3)Ϫ(10): (a) Sentence (1) [I told Bill that Jane was angry with Jim] can be reported as (3) [Jill said that she had told Bill that Jane was angry with Jim] because it is normal for an absolute preterite (told) to be ‘backshifted’ in past represented speech (because what is originally anterior to t 0 is now anterior to the situation time of said) and because it is equally normal for a relative preterite (was) to be retained (because the relation of T-simultaneity is not affected by the reporting and because indirect binding is not allowed in complement clauses depending on said). Sentence (1) can also be reported as (4) [Jill said that she told Bill that Jane was angry with Jim] because the absolute preterite told can anyhow only be interpreted as establishing a W-anterior domain (and hence as interpretationally equivalent to had told) because, as pointed out in 8.25.3, the situation time of a durative bounded (and hence nonhomogeneous) subclause (told) cannot be repre- sented as T-simultaneous with the virtually punctual situation time of its head clause (said). (b) Sentence (1) [I told Bill that Jane was angry with Jim] cannot be (faithfully) reported as (5) [Jill said that she told Bill that Jane had been angry with Jim] or (6) [Jill said that she had told Bill that Jane had been angry with Jim] because the substitution of had been for was changes the T-simultane- ity relation expressed by was in (1) into a T-anteriority relation. That is, (5) and (6) are both reports of I told Bill that Jane had been angry with Jim rather than reports of (1). (The use of had been in (6) cannot be analysed as an instance of indirect binding, because indirect binding is not grammatical in a that-clause depending on a head clause with said as verb Ϫ see principle (e) above.) (c) Sentence (2) [I told Bill that Jane was furious when Jim refused to take her out] can be reported as (7) [Jill said that she (had) told Bill that Jane had been furious when Jim refused to take her out] because this report does justice to the temporal relations expressed in (or inferrable from) the reported clause. Had told represents its situation time as T-anterior to the situation time of said, in the same way as told in (2) represents its situation time as anterior to t 0 . Told is also fine, because it can be interpreted as establishing a past domain that is W-anterior to that of said. The situation time of had been furious is represented as T-anterior to that of (had) told, which is in keeping with the fact that in (2) the situation time of was furious is interpreted as W-anterior to the situation time of told. The past tense refused from (2) is retained in (7) because the T-relation to be ex- pressed is the same in the two sentences, viz. T-simultaneity. III. Arguments for distinguishing between the absolute and the relative past tense 395 (d) Sentence (2) [I told Bill that Jane was furious when Jim refused to take her out] can also be reported as (8) [Jill said that she (had) told Bill that Jane was furious when Jim refused to take her out] because this version too does justice to all the temporal relations expressed in (or inferrable from) (2). The situation time of (had) told is interpreted as W-anterior to that of said, which is in keeping with the fact that told in (2) represents its situation time as anterior to t 0 . Because of the presence of the when- clause (which is to be interpreted as modifying Jane was furious), the situa- tion time of was furious cannot be interpreted as T-simultaneous with that of (had) told. This means that was in (8) must be an absolute past tense creating a new domain, which (for reasons explained in 8.25.3) can only be interpreted as W-anterior to the domain established by said. Finally, the situation time of refused is represented as T-simultaneous with that of was furious, just as it is in (2). (e) Sentence (2) [I told Bill that Jane was furious when Jim refused to take her out] can also be reported as (9) [Jill said that she (had) told Bill that Jane had been furious when Jim had refused to take her out] because a when-clause that is interpreted in terms of W-simultaneity can be bound indirectly (had refused) rather than directly (refused). (f) Sentence (2) [I told Bill that Jane was furious when Jim refused to take her out] cannot be reported as (10) [Jill said that she had told Bill that Jane was furious when Jim had refused to take her out] because had re- fused cannot be an instance of indirect binding (accompanied by a W- simultaneity interpretation) if its head clause is not in the past perfect too. (This restriction is explained in 13.10.1Ϫ2.) The form had refused in (10) can therefore only be interpreted as expressing T-anteriority. As a result, (10) does not accurately represent all the temporal relations expressed in the original utterance, since refused expresses T-simultaneity in (2). In other words, (10) is not a report of (2) but rather of I told Bill that Jane was furious when Jim had refused to take her out. In sum, the data provided by (3)Ϫ(10) are all naturally accounted for in our theory. 8.25.4 As noted before, at least some of these data cannot be accounted for by a theory that holds that ‘sequence of tenses’ is a mechanical (automatic, purely formal) rule in past represented speech. Consider the following, in which the (b) sentences are meant to be the indirect reported versions of the (a) sen- tences: (a) I was in Greece at the time. (absolute past tense) (b) Jane said she had been in Greece at the time. (a) Millie complained that she wasn’t feeling well. (relative past tense) (b) John told us that Millie had complained that she {wasn’t /*hadn’t been} feel- ing well. 396 8. Temporal domains and relative tenses: theoretical foundations These examples show that in past represented speech an absolute preterite can be backshifted whereas a relative one cannot. It is only a theory that distin- guishes between absolute and relative preterites that can account for this fact. 8.25.5 This also refutes any analysis which claims that in past represented speech both the head clause and the that-clause simply locate their situations in the past and are therefore both absolute past tenses (in our terminology). Like our analysis, such an analysis holds that the report must do justice to the original utterance, hence that two situations that are W-simultaneous in the original utterance must be reported in such a way that they can be interpreted as W-simultaneous. However, unlike our analysis, such an analysis assumes that the way to do that is to use the past tense in both the clauses of the report. According to this analysis, the two past tenses in Bill once told me his right foot had only four toes simply locate their situations in the past and the two situations are automatically interpreted as W-simultaneous because My right foot has only four toes represents the situation referred to as simultaneous with t 0 and a faithful report has to do justice to this temporal relation. In this analysis both the past tense forms in Bill once told me his right foot had only four toes are absolute tense forms. This analysis cannot explain all the data provided by (3)Ϫ(10): (a) The analysis does not explain why the unmarked report of My right foot has only four toes is Bill once told me that his right foot had only four toes rather than Bill once told me that his right foot has only four toes. If My right foot has only four toes is true at some time before t 0 , then (if Bill is still alive) it must also be true at t 0 . We would therefore expect the present tense to be more natural than the past tense. That this is not the case follows from the fact that a shift of domain from the past to the present is subject to restrictions, whereas it is always possible to expand a past domain created by an intensional verb like said or thought. This be- comes even clearer if My right foot has only four toes is an erroneous utterance or thought: we can say Bill {imagined / dreamed} that his right foot had only four toes, but hardly ?? Bill {imagined / dreamed} that his right foot has only four toes. (b) A heterogeneous (ϭ nonhomogeneous Ϫ see 1.36) situation cannot be lo- cated at t 0 , because its situation time coincides with the time of the full situation and one cannot locate a durative situation at (ϭ as coinciding with) a point of time. This requirement, that the situation time located at t 0 must be punctual, follows from the semantics of the present tense: it represents the situation time as T-simultaneous with t 0 , not as W-simulta- neous with t 0 . (W-simultaneity is a question of interpreting the W-relation between t 0 and the time of the full situation.) In past represented speech we notice exactly the same requirement of T-simultaneity between the that- III. Arguments for distinguishing between the absolute and the relative past tense 397 clause situation time and the situation time of said. A heterogeneous situa- tion cannot be located at the situation time of said, because you cannot locate a durative situation at (ϭ as coinciding with) a time which is consid- ered as punctual, or at any rate as shorter than the time of the situation which is located at it. Our analysis therefore predicts the unacceptability of He said he wrote a long letter on the simultaneity reading, while the other analysis does not. If both past tenses are absolute tenses, it should be possible for the two situations (domains) to be interpreted as W-simulta- neous with (ϭ overlapping) one another, unless the theory contains a spe- cial interpretive principle blocking this interpretation. Our theory does not contain such a special (ad hoc?) interpretive principle and does not need one. In other words, the kind of simultaneity that is relevant to the use and under- standing of past represented speech sentences is coincidence (T-simultaneity), not overlap (W-simultaneity). This accords with the claim that the past tense in the that-clause is a relative past tense, but not with the claim that it is an absolute past tense whose situation is interpreted as W-simultaneous with the situation of the head clause. (c) As we will see in 9.9Ϫ12, the tenses used to express T-relations in a past domain can under certain conditions also be used to express T-relations in a pre-present domain, even if the present perfect is given a ‘co-extensive’ T-interpretation (i. e. the situation time is co-extensive with the pre-present and therefore leads up to t 0 Ϫ see 5.2.2). The following examples of pre- present represented speech (or thought) come from section 9.12: [I watched Oprah today and George Michael was on there and I enjoyed his in- terview.] I have known for some time that he was gay [but you know he is still sexy]. (www) (T-simultaneity) [“You don’t think it’s on the small side?” Ϫ “It looks fine to me.”] Ϫ “I’ve been thinking lately it was rather small.” (CP) (T-simultaneity) I have known for some time that she was in great pain, [but unfortunately the only thing I could and can do is pray that God reveals his plan for her and be ] (www) (T-simultaneity) I have known for some time that she would not last forever, [but her death still comes as a shock]. (www) (T-posteriority) [In fact, terrorism experts say that] for years their worst fear has been that a suicide bomber would hit inside US borders. (www) (T-posteriority) It is clear that the italicized past tenses in the first three examples cannot be analysed as absolute past tenses, because they do not represent the situation they refer to as T-bygone, and certainly not as temporally disconnected from the present. On the other hand, analysing them as relative pasts accords not 398 8. Temporal domains and relative tenses: theoretical foundations only with the simultaneity interpretation which they receive but also with the observation that when a pre-present domain is expanded, it is always expanded as if it were a past domain. (Abundant evidence for the existence of this type of shift of temporal perspective will be adduced in 9.9Ϫ12.) 8.26 Argument 4: the incompatibility of a punctual adverbial with a relative preterite The claim that there are two types of past tense in English, an absolute and a relative one, is supported by the fact that an absolute past tense form is compatible with an unanchored punctual time-specifying adverbial (i. e. an adverbial of the type at 5 o’clock), whereas a relative past tense form is not. 8.26.1 Observe the following: (said at 5 o’clock) John is here {now / at present / today / at this very moment / *at five o’clock}. (The acceptability judgement on ‘at five o’clock’ assumes that the present tense form has its basic meaning and refers to a single situation rather than being a historic present or referring to a present habit or characteristic. Note also that the relevant difference between at this very moment and at five o’clock is that the former expresses an Adv-time which is anchored to t 0 while the latter does not.) This example shows that when a situation time is represented as T-simulta- neous (ϭ coinciding) with t 0 , its temporal location cannot also be specified by a punctual unanchored time-specifying adverbial like at five o’clock. The only seeming exception is when the time-specifying adverbial does not specify a punctual Adv-time containing the situation time in terms of coincidence but belongs to the proposition describing the situation whose situation time is lo- cated at t 0 , as in the following: (said at 5 o’clock) [You always claim that John is never here at five o’clock. But now you can see for yourself:] John is here at five o’clock. Examples like these, in which at five o’clock does not function as a time- specifying adverbial but forms part of the untensed proposition, will be disre- garded in what follows because they are irrelevant to the argument to be pre- sented. So will sentences in which the present tense is a historic present and sentences that express a present characteristic or habit Ϫ see the comment added to the first example. The reason why, at five o’clock, we cannot say *I am here at five o’clock is that the present tense represents its situation time as T-simultaneous (i. e. . shorter than the time of the situation which is located at it. Our analysis therefore predicts the unacceptability of He said he wrote a long letter on the simultaneity reading, while the other analysis. the existence of this type of shift of temporal perspective will be adduced in 9.9Ϫ12.) 8.26 Argument 4: the incompatibility of a punctual adverbial with a relative preterite The claim that there. with the situation time (had fallen) of its head clause, had seen is an example of indirect binding: it represents the situation time of the when-clause as T-anterior to the situation time of said.