In this section, I will discuss temporal reference of temporal adverbial clauses as illustrated by (62)-(64).
(62) a. Ta lai de-shihou, wo hui gaosu ta he come when I will tell him ‘When he comes, I will tell him.’
b. Wo zhu zai meiguo de-shihou, chi-guo longxia I live in America when eat-Asp lobster
‘When I lived in America, I ate lobsters (I had the experience of eating lobsters).’
c. Ta lai de-shihou, wo (zheng) zai zhu fan he come when I right Prog cook rice ‘When he came, I was cooking.’
(63) a. Wo qu zhiqian, (wo) hui xian da dianhua gei ni I go before I will first make phone-call to you
‘Before I go, I will call you first.’
b. Wo qu zhiqian, da-guo yi-tong dianhua gei ta I go before make-Asp one-Cl phone-call to him ‘Before I went, I made a phone call to him.’
c. Ta lai meiguo zhiqian, shi ge yanyuan he come America before be Cl actor
‘Before he came to America, he was an actor.’
(64) a. Wo kaoshang yanjiusuo yihou, hui mai yi-liang xin che I admitted graduate-school after will buy one-Cl new car
‘After I am admitted to a graduate school, I will buy a new car.’
b. Wo kaoshang yanjiusuo yihou, mai-le yi liang xin che I admitted graduate-school after buy-Asp one-Cl new car ‘After I was admitted to a graduate school, I bought a new car.’
c. Kao-wan shi yihou, ta-de xinqing hen qingsongxvi examine-finish test after his mood very relaxed ‘After he finished the test, he was very relaxed.’
The subordinate clauses in (62)-(64) contain neither an aspectual marker nor a temporal adverbial, but they all have a fixed temporal reference just as the matrix clauses do. To put the (c) examples aside for the moment, the (a) and (b) examples seem to indicate that temporal reference of a temporal adverbial clause can be determined by that of the matrix clause. For example, the matrix clauses in the (a) examples have a future interpretation because of the use of the modal auxiliary hui
‘will’ and so do the adverbial clauses. Similarly, both the matrix and embedded clauses of the (b) examples receive the same past interpretation because of the use of le or guo in the matrix clauses. If we look at these two sets of examples alone, it seems very tempting again to suggest that a control theory--for example, something like the TCH mentioned in the last section, may account for temporal reference of Chinese temporal adverbial clauses. That is, one first determines the temporal reference of the matrix clause as if the adverbial clause did not exist and then assigns
the same temporal value to the adverbial clause. Plausible as the above control theory might sound, it is not general enough to cover the (c) examples in (62)-(64). An important difference between the (a), (b) examples and the (c) examples in (62)-(64) is that the matrix clauses of the (a), (b) examples describe telic events, whereas the matrix clauses in the (c) examples describe atelic states. As discussed, an atelic imperfective sentence without any temporal adverbial or tense-aspectual marker must be assigned a present reading. Thus, according to the control hypothesis, the matrix clauses in the (c) examples should have a present interpretation, if we pretend that the temporal adverbial clause is not there. As a consequence, the control hypothesis predicts that the temporal adverbial clauses in the (c) examples in (62)-(64) have a present interpretation. However, this prediction is wrong, because the matrix and embedded clauses in (62)-(64) are asserted to be true of a past interval only. They do not assert that the matrix clauses are true of the speech time at all. This clearly indicates that temporal reference of Chinese temporal adverbial clauses must be determined by something other than the simple but incorrect control hypothesis such as the TCH. In what follows, I will pursue a different approach to account for the data in (62)-(64).
From (62)-(64), we have learned that temporal reference of a Chinese temporal adverbial clause varies with the context in which it appears. For the sake of argumentation, let us assume that any temporal interpretation can be assigned to a temporal adverbial clause but the assignment is subject to certain semantics or pragmatics constraints to be discussed later. Moreover, let us also assume that just as the tense node— if it exists--must agree with the interval denoted by zuotian
‘yesterday’ or 1996 nian ‘the year of 1996’ as noted earlier, the tense node of the matrix clause in (62)-(64) must agree with the topic time introduced by the temporal adverbial clause. In addition, each temporal connective specifies a temporal relation
of precedence or overlap between the adverbial clause and the matrix clause. I will argue that the above assumptions, together with some other semantics or pragmatics constraints to be discussed later, will enable us to account for the temporal interpretation of Chinese temporal adverbial clauses.
To begin with, let us consider (62a). The temporal connective de-shihou ‘when’
dictates that the event time of the matrix clause overlaps or begins right after the event time of the adverbial clause (Partee (1984)). Moreover, the tense of the matrix clause— if it exists, must agree with the topic time introduced by ta lai de-shihou
‘when he come’, i.e., the interval at which he is here holds. Since the matrix clause in (62a) contains the modal auxiliary hui ‘will’, which can be assumed to occupy the tense node, the topic time should refer to a future time. Consequently, the adverbial clause must refer to a future time. The overlapping or right-after requirement introduced by de-shihou then forces the adverbial clause to refer to a future time just as the matrix clause. This explains why the temporal adverbial clause in (62a) has a future interpretation.
Similar remarks apply to (62b). Due to the use of the experiential marker guo, the matrix clause has a past interpretation. So the topic time is a past interval. It follows from the overlapping or right-after requirement that the temporal adverbial clause is about the past, too.
(62c) is more complicated. The matrix clause in this example does not contain any aspectual marker or modal auxiliary. This means that no direct evidence can tell us what temporal interpretation the matrix clause has. Notice that even though the tense, of the matrix clause---if it exists, must agree with the topic time introduced by the temporal adverbial clause, the temporal adverbial clause does not provide us with sufficient information to tell what its temporal interpretation is. It is thus mysterious how the temporal locations of the matrix and temporal adverbial clauses in (62c) is
determined. For the sake of argument, let us pretend that the temporal location of the temporal adverbial clause can be about the present or the future and see what violation the temporal specification might encounter. Consider present interpretation first. This temporal interpretation is not allowed for two reasons. The first reason has to do with Kamp and Reyle’s (1993) observation about temporal prepositional phrases such as in April or on Sunday. They point out that these phrases cannot be used to refer to periods containing the utterance time. Thus, if today is Sunday and you know this, then you cannot utter the following sentence.
(65) Mary wrote the letter on Sunday.
They propose that this constraint be analyzed as a presupposition on the interpretation of such phrases. With this in mind, now let us consider the following Chinese sentences.
(66) Zhongqiujie de-shihou, wo jian-guo ta moon-festival when I see-Asp him
‘I saw him on the day of Chinese moon festival.’
The phrase zhongqiujie de-shihou ‘the day of Chinese moon festival’ exhibits properties similar to those of temporal expressions such as in April or on Sunday. If today is moon festival and you know this, you cannot felicitously utter (66). This fact suggests that de-shihou ‘when’ is subject to a constraint similar to in April and on Sunday. If this is correct, then the temporal location of the event denoted by the adverbial clause in (62c) cannot be about the present. The second reason is related to Maxim of Quantity, a conversation principle formulated by Grice (1975), which
recommends speakers to say as much as he can. According to this maxim, if a speaker is cooking at the speech time, he should say it as such. In other words, if the speaker wants to use a time adverb to indicate the present moment, he should use expressions such as xianzai ‘now’ rather than a temporal adverbial clause that has no fixed temporal reference.
Next let us consider the examples in (63), involving the temporal connective zhiqian ‘before’. This temporal connective requires that the event denoted by the matrix clause, indicated by e1, precede the event denoted by the temporal adverbial clause, indicated by e2. Moreover, the whole temporal adverbial clause introduces an interval, i.e., the period of time before the event described by the zhiqian-clause
‘before-clause’, to serve as the topic time of the matrix clause. I use Ttime to stand for it. The time schemata for sentences containing a zhiqian-clause look like the following:
(67) Matrix clause telic (perfective) Matrix clause atelic (imperfective) Ttime Ttime
---> --->
e1 e2 e1 e2
Now let us consider (63a). Due to the use of the modal auxiliary hui ‘will’, the matrix clause refers to a future event. In other words, e1 in (67) must follow the speech time. Since the adverbial clause event e2 follows the matrix clause event e1, e2
must follow the speech time, too. This explains why the adverbial clause in (63a) has a future interpretation as the diagram in (68) indicates.
before-I-go (Ttime)
(68) --- --->
s* call(e1) go(e2)
Next, let us consider (63b). In (63b), the experiential marker guo appears in the matrix clause. Therefore, the matrix clause has a past reading and the topic time must be about a past interval. To satisfy these requirements, however, there are three possibilities, as given in (69). In these diagrams, the topic time and the calling event are both before the speech time, but the location of the going event differs. In (69a), the time of going is before the speech time; in (69b), the time of going is in the future;
in (69c), the going event overlaps the speech time.
before-I-go (Ttime)
s*
(69) a. --->
call(e1) go(e2)
before-I-go (Ttime)
s*
b. --->
call(t1) go(t2)
s*
c. --->
call(e1) go(e2)
According to (69a), (63b) should have a reading according to which both the calling event and the going event happened before the speech time. Indeed, (63b) has this reading. In contrast, (69b) and (69c) require that the calling event happen before the speech time but the going event will take place in the future or is on-going at the speech time. Unfortunately, these are readings that (63b) lacks. The problem is why
(63b) does not have these readings. In what follows, I will argue that the representation of (69b) and (69c) is ruled out by an independent principle proposed by Kamp and Reyle (1993).
Kamp and Reyle (1993) have made a very interesting observation about the combinations of tenses in main and subordinate clauses in English. They point out that English sentences like the following are deviant.
(70) Bill will leave before Mary arrived.
(71) Bill left before Mary will arrive.
According to Kamp and Reyle (1993, 652), the oddity of (70) can be easily accounted for in terms of inconsistency because the word before requires that the event of the matrix clause precede the event of the subordinate clause but the tenses express the reverse. However, the same inconsistency account cannot be extended to (71), because a past event is certain to precede a future event. Despite this, they point out that what (70) and (71) have in common is that “their locating adverbs fail to produce a genuine constraint on the set of times compatible with the interpretation of tense”.
Thus, they propose a “non-triviality constraint” to capture this. This constraint essentially says that “temporal adverbs must impose a genuine restriction on the location time”. This requirement is not satisfied in (71) because a past time is always before a future time. So there is no genuine constraint on the tenses and the meaning of before.
Although Chinese does not have morphological tenses, the situation in (69b) is completely parallel to that in (70) and (71). Thus, I suggest that Kamp and Reyle’s Non-triviality Constraint not be taken as a constraint on morphological tenses but a universal principle on permissible temporal interpretations for temporal connectives.
This then excludes the possibility of representing (63b) as (69b).
(69c) has a similar problem as (69b). As an event happened before the speech time is certain to precede an event that is taking place at the speech time, a diagram like (69c) cannot constitute a case where the temporal relation between the matrix and embedded clauses is genuinely constrained by the temporal connective zhiqian
‘before.’
The last example in (63) is (63c). (63c) is much like (62c) in that the matrix clauses in both sentences do not contain any aspectual marker or modal auxiliary. (72) is the diagram that indicates all possible locations of the speech time. In this diagram, five possible locations for the speech time are indicated by an arrow on the time axis.
before-he-come-to-America (Ttime)
(72) --->
be-an-actor(e1) come-to-America(e2)
If the speech time is the leftmost arrow, then both the event of coming to America and the state of being an actor should hold at a future time. In other words, both the matrix and adverbial clause have a future interpretation. This is not permitted, however. As noted earlier, if a matrix clause is to express futurity in Chinese, an overt modal auxiliary or temporal adverbial indicating a future time is obligatory. However, (63c) does not contain a modal auxiliary such as hui ‘will’ or any time adverb such as weilai
‘in the future’ clearly indicating a future time. Though the adverbial clause ta lai meiguo zhiqian ‘before he come to America’ is a temporal adverbial, it is not that kind of temporal adverbial that inherently refers to a future time. Next, let us consider the second possibility, where the speech time is included within e1. This possibility means that the state of being an actor is asserted to be true at the speech time. But if this is
the assertion that the speaker wants to make, he should use a more direct expression such as xianzai ‘now’ to indicate this rather than using the temporal adverbial clause ta lai meiguo zhiqian ‘before he come to America’, which as noted does not have a fixed temporal reference. In other words, if the speech time falls within e1, the use of the construction in (63c) violates Grice’s (1975) Maxium of Quantity. As for the third possibility, it requires that the event of going to America take place in the future but the state of being an actor be true of a past interval. This possibility should be equivalent to the English sentence He was an actor before he will come to America.
However, I have shown that the semantics of such sentences violates Kamp and Reyle’s (1993) Non-triviality Constraint. As for the fourth, it is excluded because the speech time can never be included within a telic (perfective) situation. The remaining possibility is the last arrow. In this possibility, both the state of being an actor and the event of coming to America precede the speech time. Since both events happened before the speech time, their temporal relation can be genuinely constrained by the semantics of the subordinator zhiqian ‘before’. Indeed, this is the only interpretation that does not violate any temporal constraint. Therefore, (63c) can be uttered in a situation where the last arrow is the utterance time. It follows from this that both the matrix clause and the adverbial clause have a past interpretation.
Finally, let us consider temporal reference of yihou-clauses ‘after-clauses’ in (64).
The temporal connective yihou ‘after’ requires that the event described by the adverbial clause precede the event described by the matrix clause. So, the time schemata of sentences containing a yihou-clause should look something like those in (73).
(73) Matrix clause telic (perfective) Matrix clause atelic (imperfective) Ttime Ttime
---> --->
e2 e1 e2 e1
Applying the above schemata to (64a), we obtain the following diagram:
after-admission (Ttime) (74) s*
--->
s* admitted(e2) s* buy(e1)
Because the matrix clause in (64a) contains the future modal auxiliary hui ‘will’, the event denoted by it must take place in the future. To satisfy this requirement, there are three possibilities: the speech time may follow the event of admission but precede the event of buying, or the speech time precedes both the event of admission and the event of buying, or the speech time overlaps the time of admission. However, the first option violates Kamp and Reyle’s Non-triviality Constraint, because a time before the speech time always precedes a time after the speech time. Therefore, there is no genuine constraint imposed by the time connective yihou ‘after’ in the first option.
The third possibility is excluded because it is impossible for the speech time to be included within the run time of an achievement. An achievement situation must culminate before one is able to talk about it, unless it is about a future event. However, if the speech time precedes both events, i.e., the third arrow, no violation of Non-triviality Constraint will arise. This explains why the adverbial clause in (64a) has a future interpretation just like the matrix clause.
Next, consider (64b), which differs from (64a) only in the use of a different aspectual marker in the matrix clause. The use of le in (64b) indicates that the matrix clause is about a past event. Therefore, the event denoted by the matrix clause should
take place before the speech time. Since the semantics of yihou ‘after’ dictates that the event denoted by the yihou-clause precedes the event denoted by the matrix clause, it follows that the former also precedes the speech time. This is shown by the diagram in (75).
after-admission (Ttime) (75)
--->
admitted(e2) buy(e1) s*
The last case is (64c). The matrix clause of (64c) is unbounded, so it should have a diagram like (76).
after-admission (Ttime) s* s* s* s*
(76) --->
finish-taking- relaxed(e1) the-exam(e2)
In this diagram, we have four possible speech times to consider. If the speech time is the leftmost arrow, this means that both the matrix clause and the adverbial clause have a future interpretation. This possibility is excluded, because future interpretation in Chinese requires an overt expression indicating a future time but no such expression is available in the structure. The temporal adverbial clause does not inherently refer to a future time, so it does not count as an expression indicating a future time. The second possibility has a similar problem, because it requires that the matrix clause be interpreted as a future clause but there is no overt expression inherently referring to a future time. In addition, achievement events denoted by sentences with a resultative verb such as kao-wan ‘finish examing’ can never include
the utterance time. The remaining possibilities are the third and the fourth arrows. The third possibility is excluded by Kamp and Reyle’s Non-triviality Constraint, because an on-going situation always precedes a past situation. Finally, the fourth possibility claims that both the event of exam taking and the state of being relaxed happened in the past. Since two past situations can be genuinely constrained by a temporal connective, the fourth possibility does not violate Kamp and Reyle’s Nontriviality
the utterance time. The remaining possibilities are the third and the fourth arrows. The third possibility is excluded by Kamp and Reyle’s Non-triviality Constraint, because an on-going situation always precedes a past situation. Finally, the fourth possibility claims that both the event of exam taking and the state of being relaxed happened in the past. Since two past situations can be genuinely constrained by a temporal connective, the fourth possibility does not violate Kamp and Reyle’s Nontriviality