III. Arguments for distinguishing between the absolute and the relative past tense 399 coinciding) with t 0 . This means that the situation time receives its temporal specification from the location at t 0 , which is the strongest possible kind of temporal specification, since t 0 is by definition ‘given’: in sentences with an unanchored time-specifying adverbial, the situation time derives its temporal specification exclusively from the adverbial, because the Adv-time ‘contains’ the situation time (in terms of coincidence or inclusion Ϫ see 2.23.1). A situa- tion time cannot derive its temporal specification from two different mecha- nisms at once, even if the times they refer to coincide with one another. In sum, a sentence whose situation time is represented as T-simultaneous (ϭ coinciding) with t 0 cannot contain an unanchored punctual time-specifying adverbial specifying a time with which the situation time coincides. Something similar happens when a past tense is used. Consider: (11) I was still thirsty at five o’clock. (12) At five o’clock Jim whispered that he was still thirsty. (13) *At five o’clock Jim whispered that he was still thirsty at five o’clock. (The two time-specifying adverbials at five o’clock are meant to be read as specifying the same time.) (14) (At five o’clock) Jim whispered that he was still thirsty at three o’clock. In (11), the absolute past tense form was locates its situation time in the past time-sphere. The precise location of this situation time in the past is specified by at five o’clock. In (12), was [thirsty] is a relative past tense form. It repre- sents its situation time as T-simultaneous (ϭ coinciding) with the situation time of the head clause and in so doing derives its temporal specification from the latter, which itself derives its temporal specification from at five o’clock. As shown by the ungrammaticality of (13), we cannot add at five o’clock to the that-clause of (12). If we add this, the that-clause can no longer derive its temporal specification from the T-relation with the situation time of the head clause. As noted above, a situation time cannot derive its temporal specification from two different times at once, even if these times happen to coincide with one another. This explains the unacceptability of (13). Things are different in (14), where at three o’clock refers to a different time from the time of the whispering. In our terminology, this means that was [thirsty] must now be read as an absolute tense form establishing a domain of its own. (The different adverbials make it clear that the central orientation time of the domain in question is interpreted as W-anterior to the situation time of the head clause, but this relation is not expressed by the tense forms: the absolute preterites whispered and was [thirsty] T-relate their situation times directly to t 0 .) The difference between (11) and (14) on the one hand and (12) and (13) on the other shows that an absolute past tense form is compatible with an unanchored 400 8. Temporal domains and relative tenses: theoretical foundations punctual time-specifying adverbial, whereas a relative past tense is not. 17 This is a cogent syntactic argument in favour of the claim that a distinction should be made between T-absolute and T-relative past tense forms in English. It should be noted that the restriction observed in (13) is quite similar to that observed in *John is here at five o’clock. What we claim in connection with the relative past tense is therefore based on a more general principle which also governs the use of unanchored time-specifying adverbials in present tense sentences referring to a single situation. It might be argued that the above argument fails because the problem with (13) [*At five o’clock Jim whispered that he was still thirsty at five o’clock]is simply redundancy. However, redundancy may lead to reduced acceptability but not to ungrammaticality, unless the redundancy is a question of reduplicat- ing a constituent within a single syntactic slot (function). Compare: *On his head he was wearing a bowler hat on his head. (This is ungrammatical because the locative adverbial on his head is reduplicated, so that there are two identical constituents fulfilling the same syntactic function. An alternative syntactic analysis, e. g. that the second on his head forms part of the noun phrase a bowler hat on his head, is not available: in He was wearing a bowler hat on his head the constituent on his head can only be analysed as a locative adverbial, not as a post- modifier to the noun head bowler hat. This is clear from the fact that What he was wearing on his head was a bowler hat is a quite normal thing to say, while the following is not: *What he was wearing was a bowler hat on his head.) * ? John told the girl standing next to him that he liked the girl standing next to him. (This is far less acceptable than John told the girl standing next to him that he liked her. The repetition of the girl standing next to him makes for redundancy, which drastically reduces the acceptability of the sentence. However, there is no way one can say that the sentence is syntactically or/and semantically ungrammatical.) *At five o’clock Jim whispered that he was still thirsty at five o’clock. (If the two time-specifying adverbials at five o’clock are meant to be read as specifying the same time, this sentence is simply ungrammatical, not just relatively unacceptable because of redundancy. Note also that it differs from the first of these three examples in that at five o’clock is not reduplicated: the two time-specifying adverbials have a syntactic function in different clauses.) 8.26.2 There is another observation supporting the claim that an absolute past tense form is compatible with an unanchored punctual time-specifying adverbial, whereas a relative past tense form is not. It concerns unanchored punctual time-specifying adverbials that are added to the description of a dura- 17. The only exception is when the adverbial in the subclause is a pro-form for the adverbial in the head clause, as in At five o’clock John said that he was thirsty {then / at that time}. Such sentences, which anyhow sound awkward because the addition of the pro- form is completely redundant, are disregarded here. They do not form any counterevi- dence to the argument that has been offered. III. Arguments for distinguishing between the absolute and the relative past tense 401 tive bounded situation. As noted in 2.30.2, this is only possible if there is a suitable inceptive (inchoative) or terminative reading. Compare: At seven o’clock he was eating a three-course dinner. (This sentence, in which was eating is an absolute tense form, is fine on the interpretation that the situation time coincided with the punctual Adv-time, because the progressive form represents the telic situation as nonbounded and homogeneous, which means that the situation time can be a punctual portion of the time of the full situation and can therefore be represented as coinciding with the punctual Adv-time denoted by at seven o’clock.) At seven o’clock he ate a three-course dinner. (This sentence is ungrammatical if it is to express that the situation time coincided with the punctual Adv-time, because the nonprogressive form of a telic verb phrase represents the durative situation as bounded (and therefore as heterogeneous), which means that the situation time is also the time of the full situation, which is durative and therefore cannot coincide with a punctual Adv-time. However, when such a sentence is uttered by a coopera- tive speaker, a cooperative hearer will look for a suitable reading and assume a partly inchoative interpretation: ‘He ate a three-course dinner and started doing so at seven o’clock’.) John came to the station at four. (The most suitable interpretation here is a partly terminative one: ‘John came to the station and arrived there at four’.) In order to see whether we can get the same inchoative or terminative reading in clauses using a relative tense it suffices to consider the past represented speech versions of the previous two examples: John said that at seven o’clock he ate a three-course dinner. (The only possible interpretation of the that-clause is a partly inchoative reading. However, the begin- ning of the eating cannot be interpreted as T-simultaneous with the situation time of said. The sentence only allows the reading in which the eating precedes the saying, i. e. the reading that is to be expected if ate is an absolute past tense form.) John said he came to the station at four. (The most suitable interpretation here is a partly terminative one: ‘John came to the station and arrived there at four’. How- ever, John’s arrival at the station cannot be interpreted as T-simultaneous with the situation time of said. The sentence only allows the reading in which came is an absolute tense form establishing a past domain which is interpreted as W-anterior to the past domain established by said.) In sum, the combination of an unanchored punctual time-specifying adverbial and the description of a durative bounded situation can trigger a partly incho- ative or terminative reading in complement clauses, but only if the past tense that they use is the absolute past tense. There is thus a constraint on combining a punctual unanchored time-specifying adverbial with the relative past tense, while there is no such constraint where the absolute past tense is concerned. This further corroborates our claim that there are two types of past tense in English. 402 8. Temporal domains and relative tenses: theoretical foundations 8.27 Argument 5: a relative past tense form cannot be replaced by used to In a sentence referring to a past habit, an absolute past tense form can be replaced by used to (which expresses T-anteriority to some orientation time), whereas a relative past tense form cannot. 8.27.1 A scrutiny of examples with the auxiliary used to reveals that its tem- poral meaning is: ‘The situation time is represented as T-anterior to some orien- tation time’. The orientation time in question may be t 0 ,asinTerry used to be a nuisance, or a past orientation time, as in Terry said that as a child he used to be a nuisance, or a post-present orientation time which is treated as if it were t 0 when another situation time is T-related to it, as in the following ex- ample: [Teach your child to cook healthy food. When he grows up and is cooking for himself,] he will remember the things he used to make in his childhood [and will feed himself well]. (www) In this example, used to locates the situation time of make in a ‘pseudo-past’ period, i. e. in a period which is past with respect to the time of the remember- ing Ϫ see 9.18.1. (In other words, used to represents the making as T-anterior to the situation time of will remember, which is the central orientation time of the post-present domain and is therefore treated as if it were t 0 when another situation time is T-related to it Ϫ see 9.17.) It follows from the temporal meaning of used to (viz. ‘The situation time is represented as T-anterior to an orientation time’) that, if the reference is to a habit lying in the (real) past, ‘used to ϩ verb’ can alternate either with an absolute past tense (expressing T-anteriority to t 0 ) or with a past perfect (ex- pressing T-anteriority to an orientation time in a past domain): [When I was a child] I {was / used to be} afraid of spiders. Chris said that she {had played / used to play} tennis until three years ago. In cases where the past perfect can alternate with an absolute past tense estab- lishing a W-anterior domain (e. g. He said he {left / had left} early that morn- ing), used to can alternate with the past tense as well as with the past perfect: [He said that when he was a child] he {had been / was / used to be} afraid of spiders. (Had been expresses T-anteriority in the past domain established by said; was and used to be create a past domain of their own, which is interpreted as W- anterior to the domain established by said.) III. Arguments for distinguishing between the absolute and the relative past tense 403 On the other hand, it follows from its meaning that used to cannot alternate with a tense form representing its situation time as T-simultaneous (ϭ coincid- ing) with t 0 or with a past, pre-present or post-present orientation time. In the following sentences it is each time unacceptable, at least on a simultaneity read- ing: Now he {is / *used to be} afraid of frogs. (reference to a present habit) I saw that he {was / *used to be} afraid of frogs. (reference to a past habit that is represented as T-simultaneous with the past situation time of saw) I’ve often heard him say that he {was / *used to be} afraid of frogs. (reference to a habit that is represented as T-simultaneous with the pre-present situation time of have heard) [If you teach him to cook as a child] he will tell other people later how he {cooks / *used to cook}. (reference to a post-present habit that is represented as T-simulta- neous with the post-present situation time of will tell) In conclusion, in a sentence referring to a habit, used to can alternate with a tense form expressing T-anteriority to t 0 (ϭ an absolute past tense) or T-anteri- ority to another orientation time (e. g. a past perfect), but not with a tense form representing its situation time as T-simultaneous (ϭ coinciding) with t 0 (ϭ a present tense) or as T-simultaneous with another orientation time (e. g. a relative past tense). 8.27.2 We interpret the data offered in the previous subsection as meaning that used to cannot express T-simultaneity. One might object to this, saying that all that has been shown is that used to cannot express W-simultaneity, or that a W-simultaneity interpretation blocks the use of used to. This objection would involve the claim that we do not need the notion of T-simultaneity to explain the data, and that all past tense forms are absolute tense forms. It is up to us to refute this analysis. It is not difficult to show that a W-simultaneity interpretation does not block the use of used to. In a subclause, used to can alternate with a past tense form establishing a domain of its own which is to be interpreted as W-simultaneous with the head clause domain, provided the verb forms of the two clauses ex- press T-anteriority to the same orientation time. Compare: Sheila visited her grandmother regularly and she also gave her flowers. (Both past tenses represent their situation time as T-anterior to t 0 . In doing so they create two separate past domains, which can be interpreted as W-simultaneous with each other. Since both clauses can refer to a habit, the clauses can be interpreted as referring to W-simultaneous past habits.) Sheila used to visit her grandmother regularly and she also used to give her flowers. (The two clauses again establish separate past domains, which can be pragmatically interpreted as W-simultaneous with each other. Both clauses express T-anteriority to 404 8. Temporal domains and relative tenses: theoretical foundations t 0 . Another interpretation Ϫ for example, the habit expressed in the second clause is interpreted as anterior to the habit expressed in the first clause Ϫ is completely ruled out.) Similarly, used to can alternate with a past perfect form effecting ‘indirect binding’ (see 9.28.1), provided the situation time in question is interpreted as anterior to the same orientation time as also binds the situation time of the head clause: [I admitted that] when I had been in front of the camera or microphone I had been very nervous. (Both past perfect forms express T-anteriority to the situation time of admitted. Their situation times are interpreted as W-simultaneous with each other, which means that the when-clause is bound indirectly.) [I admitted that] when I used to be in front of the camera or microphone I used to be very nervous. (Both used to-forms express T-anteriority to the situation time of admitted. Their situation times are interpreted as W-simultaneous with each other.) In sum, the data in connection with used to referred to in 8.27.1 cannot be explained by the assumption that a W-simultaneity interpretation blocks the use of used to. 8.27.3 In the previous sections the following principle has been illustrated: Take a complex sentence in which both the head clause and the subclause express a past habit and use an absolute preterite form (so that two different past domains are established). Suppose further that the subclause domain is to be interpreted as W-simultaneous with the head clause domain, as in In those days Bill walked to work because his wife needed the car. In that case we can replace the the past tense form of the subclause (as well as that of the head clause) by a form with used to without changing the temporal relations: in In those days Bill {walked / used to walk} to work because his wife used to need the car, the two situations are again T-anterior to t 0 and are again interpreted as W-simultaneous with each other. How- ever, if the subclause uses a relative past tense form, we cannot replace that form by a form with used to because this would change the T-simultaneity relation expressed by the relative past tense form into a T-anteriority relation. It follows that if sentence (15) below established two independent past domains which were interpreted as W-simultaneous with each other, it should be pos- sible for the sentence to alternate with (16): (15) My father often said that he was afraid of mice. (The intended reading is that on which the two situations, which are both T-anterior to t 0 , are interpreted as simultaneous.) (16) My father used to say that he used to be afraid of mice. Actually, (16) cannot be interpreted as having the same meaning as (15). It can only be interpreted as equivalent to My father always said that he had been afraid of mice. This strongly supports our claim that (on the intended inter- pretation) was afraid in (15) is not an absolute preterite form establishing a W- III. Arguments for distinguishing between the absolute and the relative past tense 405 simultaneous domain but is a relative preterite form expressing T-simultaneity. (Evidently, a tense form expressing T-simultaneity cannot alternate with (and be equivalent to) a used to-form expressing T-anteriority.) The crux of this argument is that the impossibility of replacing (15) with (16) proves that it is not W-simultaneity that blocks the use of used to but T-simultaneity. The use of used to is not blocked in the examples in 8.27.2, which illustrate W-simulta- neity, but it is blocked in (16), because was in (15) expresses T-simultaneity. 8.27.4 In sum, in a that-clause referring to a past habit, an absolute preterite can be replaced by a used to-form, whereas a relative preterite cannot. The following examples further illustrate this: Millie believed that her son {didn’t smoke / didn’t use to smoke} when he was younger. (Because of the presence of the when-clause, the situation of not smoking must be interpreted as anterior to Millie’s belief Ϫ see 8.26.1. This means that didn’t smoke must be an absolute preterite establishing a domain that is interpreted as W- anterior to the domain established by believed. Such an absolute preterite, expressing T-anteriority to t 0 , can be replaced by a used to-form on a habitual interpretation.) Millie believed that her son {didn’t smoke / #didn’t use to smoke}. (For lack of a context triggering another interpretation, the habitual situation of not smoking is interpreted as simultaneous with Millie’s belief. If, on this interpretation, didn’t smoke were an absolute preterite establishing a domain that was interpreted as W- simultaneous with the domain established by believed, it would be interchangeable with didn’t use to smoke, because both forms would express T-anteriority to t 0 .The fact that didn’t use to smoke (which expresses T-anteriority to t 0 ) cannot replace didn’t smoke (on the intended reading) 18 can only mean that didn’t smoke does not express T-anteriority to t 0 . In other words, didn’t smoke cannot be an absolute preterite. Our analysis, which treats it as a relative preterite (expressing no other relation than T-simultaneity) is in keeping with this conclusion and also provides a natural explanation for the data.) 8.28 Argument 6: the use of the preterite rather than the conditional tense The existence of a relative past tense (expressing T-simultaneity with an orientation time in a past domain) accounts for the obligatory use of the past tense (rather than the conditional tense) in the adverbial time clause of a sentence like John decided that 18. It is this fact Ϫ that didn’t smoke cannot be replaced with didn’t use to smoke Ϫ that is indicated by the symbol # in the example. . unanchored punctual time-specifying adverbials that are added to the description of a dura- 17. The only exception is when the adverbial in the subclause is a pro-form for the adverbial in the head clause, as. establishing a domain of its own. (The different adverbials make it clear that the central orientation time of the domain in question is interpreted as W-anterior to the situation time of the head clause, but. begin- ning of the eating cannot be interpreted as T-simultaneous with the situation time of said. The sentence only allows the reading in which the eating precedes the saying, i. e. the reading