In the preceding chapter, it was argued that perfect constructions have a uniform compositional semantics. It is well-known, however, that present perfect construct- ions can exhibit a wide range of readings or meaning effects. In this chapter, it will be argued that the variability of readings does not contradict the assumption of a uniform semantics. In fact, it will be shown that the particular semantics of the construction, taken together with some additional but independently motivated pragmatic principles, predicts that the construction displays the specific meaning effects it does.
From a theoretical point of view, the multiple meaning effects of the present perfect are especially important because some of them appear to show that the construction is a tense with the same meaning as the past tense. Other effects, however, appear to show that it is an aspect which can express - among other things - the completedness of the situation expressed by the VP. In some approaches, the effects of the present perfect are treated as different meanings of two homophonous constructions; other approaches treat them as different readings that are due to a genuine ambiguity, e.g. in terms of a scope ambiguity (cf. Klein (1997)). Moreover, there are some approaches which focus only on some of the effects and disregard others. It is clear that neither of these views is compatible with the semantics motivated in chapter II.
Hence, I will propose a quite different account of the problem in this chapter:
both the semantics and the syntax of perfect constructions is uniform, but certain pragmatic principles, which can be shown to play an important role in natural language independently, operate on the semantics. The semantics contains vagueness at certain points - e.g., where the contextual restrictions of quantifiers are concerned. These vaguenesses can be assigned content, and the various possibilities for assigning content lead to the wide range of effects mentioned above. In such an approach, there is no need to assume ambiguities, nor is it necesssary to add any stipulations with regard to the composition of the construction. Quite to the contrary:
the occurrence of the effects follows automatically from the semantics of the present perfect construction and the pragmatic principles that operate on the semantics.
1* For discussions and comments, I want to thank Thanks to Wolfgang Klein and Arnim von Stechow, Karin Donhauser and the students in our joint class "Tempustheorien und die Entwicklung des deutschen Tempussystems" (Wintersemester 1997/98) as well as the audience of my talk "Das Perfekt im Deutschen" at the DGfS-Jahrestagung in Düsseldorf, 26.- 28.2.1997, where I first presented the ideas and observations that are presented in this chapter.
An earlier version of this chapter is published as "Die Lesarten des Perfekts", Literaturwissenschaft und Linguistik, 1999.
Section 2 introduces the meaning effects of the construction, which constitute a measure to what an account of the present perfect must be able to explain. Section 3 discusses some previous accounts of the effects and shows their shortcomings.
Section 4 is concerned with pragmatic principles that are relevant in a theory of temporal interpretation. Sections 5-11 show how the semantics that was introduced in chapter II taken together with the pragmatic principles discussed in section 4 predict the effects of the construction. Finally, section 12 summarizes the results of the chapter.
For reasons of space, this chapter is only concerned with present perfect clauses whose tense time is an environment of the time of utterance, i.e. with present perfect clauses that compete with past tense clauses insofar as both of them locate the situation time of their VP before the time of utterance. Taking some additonal principles into account that are crucial to futurate interpretations of the present tense in present perfect constructions (e.g. the behavior of present and future adverbials that was discussed in chapter II), the principles discussed in this chapter apply analogously in these cases. Moreover, note that where this appears desirable for presentational reasons, comparisons to the behavior of past perfect constructions in German or to the behavior of English present perfect constructions will be used.
2. The effects of the present perfect
This section serves to introduce the most important meaning effects of present perfect constructions that will be discussed in the remainder of this chapter. Most of the effects are well-known and repeatedly discussed in the literature; others, however, have to my knowledge not been noted before.
PRESENT PERFECT AND PAST TENSE CAN OFTEN BE SUBSTITUTED BY EACH OTHER. The simple past tense and the present perfect in German often can be used for describing a situation in the past without triggering any change of acceptability or meaning. (2-1) lists some illustrating examples.
(2-1) a. Gestern rannte Lola.
yesterday ran Lola ('Yesterday, Lola ran.') b. Gestern ist Lola gerannt.
yesterday is Lola run ('Yesterday, Lola ran.')
c. Martin verabredete sich gestern zum Tischtennisspielen.
Martin made-an-appointment himself yesterday for-the tabletennis d. Martin hat sich gestern zum Tischtennisspielen verabredet.
Martin has himself yesterday for-the tabletennis-playing made-an- appointment
e. Heute morgen rief Ralf an. Er hatte starkes Kopfweh.
today morning called Ralf at. he had strong headache
f. Heute morgen hat Ralf angerufen. Er hat starkes Kopfweh gehabt.
today morning called Ralf at. he has strong headache had
COMPLETEDNESS EFFECTS. The past tense sentence (2-2a) and the present perfect sentence (c) suggest that at a particular time in the past, Hans was already busy
eating or was working in the garden. However, the present perfect sentences (b) and (d) seem to imply that Hans finished eating and working in the garden, respectively.2 (2-2) a. Hans aß schon.
Hans ate already
b. Hans hat schon gegessen.
Hans has already eaten
c. Hans hat im Garten gearbeitet und das Telefon nicht gehört.
Hans has in-the garden worked and the telephone not heard d. Hans hat im Garten gearbeitet und ist jetzt müde.
Hans has in-the garden worked and is now tired
UNSPECIFIC PRESENT RELEVANCE. The data in (2-3) - (2-6) show effects that are often described as 'present relevance' effects. However, the present relevance of the examples can appear in quite different flavors. Contrasting with the past tense clause in (2-3a), for example, the perfect clause in (2-3b) suggests unspecifically that the emptying of the houses is relevant now.3
(2-3) a. Letzte Woche Mittwoch wurden die besetzten Häuser geräumt last week Wednesday became the occupied houses emptied
b. Letzte Woche Mittwoch sind die besetzten Häuser geräumt worden.
last week Wednesday are the occupied houses emptied become
PRESENT RELEVANCE: CAUSE AND STILL EXISTING OR EFFECTIVE RESULT ('perfect of result'). In (2-4b), the present perfect suggests more specifically that Kleve now consists of brickstone houses to a large extent. I.e. the result of the VP-situation is still given. The past tense in (a), however, is easily compatible with a situation where Kleve's development in the direction of a town consisting of brickstone houses ended at some point after which other types of houses were built.4
(2-4) a. Nach dem zweiten Weltkrieg wurde Kleve immer mehr zu einer after the second worldwar became Kleve always more to a Klinkerstadt. Von der weißen Stadt blieb so gut wie gar
brick-stone-town. from the white town remained as good as PARTICLE nichts mehr übrig.
nothing more remaining
b. Nach dem zweiten Weltkrieg ist Kleve immer mehr zu einer Klinkerstadt after the second worldwar is Kleve always more to a brick-stone-town geworden. Von der weißen Stadt ist so gut wie gar nichts mehr
become. from the white town is as good as PARTICLE nothing more übrig geblieben.
remaining remained
PRESENT RELEVANCE: EINSTEIN AND PRINCETON.(2-5) illustrates a phenomenon which the German present perfect, in contrast to the English present perfect, does not show. We will see later, however, that the contrast between the German and theEnglish present perfect in this respect is quite interesting. The English example in
2 The examples are taken from Klein (1994) and Klein (1997), respectively.
3 Cf. Herweg (1990:203).
4 Cf. Ehrich (1992:90f).
(a) strongly and wrongly suggests that Einstein is still alive; since we know that Einstein is dead, (a) appears infelicitous. However, the passivized sentence (b) does not show this effect; (b) just suggests that Princeton is still in existence, which is right. But the corresponding German data in (c) and (d) do not show this contrast, or at most very weakly.
(2-5) a. #Einstein has visited Princeton.
b. Princeton has been visited by Einstein.5 c. (#)Einstein hat Princeton besucht.
Einstein has Princeton visited
d. Princeton wurde von Einstein besucht.
Princeton became by Einstein visited
PRESENT RELEVANCE AND SWITCHES FROM STAGE-LEVEL TO INDIVIDUAL- LEVEL PREDICATES: COUGHING AND OTHER DISEASES. The examples in (2-6) are important to us in more than one respect. (2-6a) is an assertion about Hans, an individual which we will take to be still alive now. (2-6b-d), however, are assertions about Mozart and Goethe, two individuals which have been dead for quite a while now. (2-6a, b) are present perfect clauses with the verb husten ('cough'); they differ only with regard to the subject. Despite this little difference they behave quite differently. For the moment, let's focus on two of the effects of (2-6): contrasting with (2-6b), (2-6a) exhibits a presence relevance effect. Ehrich's (1992) intuition about this is that the present relevance effect arise because of the present perfect and the semelfactive Aktionsart of the verb. This effect cannot arise in (2-6b), however, because a coughing by Mozart can hardly be relevant now. Because of this, Ehrich finds the sentence marked. But when a temporal adverb is added as in (2-6c), the present relevance effect vanishes. Moreover, the effect does not arise with individual-level predicates or longterm properties at all.6
(2-6) a. Hans hat gehustet.
Hans has coughed b. #Mozart hat gehustet.
Mozart has coughed
c. Mozart hat an seinem 24. Geburtstag um 11 gehustet.
Mozart has at his 24th birthday at 11coughed d. Goethe hat in Weimar gelebt.
Goethe has is Weimar lived
But the markedness of (2-6b) does not necessarily occur: (2-6b) is fully acceptable when the stage-level predicate (vd. Kratzer (1989)) husten, which usually refers to a temporary property of individuals, is reinterpreted as an individual-level predicate - i.e. when Mozart's coughing is understood as a longterm property of the composer.
Under this reading of (2-6b), husten seems to refer to a generic-habitual property of individuals. this effect does not arise in (2-6a). In other words: while (2-6a) can very well refer to a single coughing of Hans, (2-6b) strongly suggests that Mozart had some chronic lung disease.
5 These examples are briefly discussed in Declerck (1991:344f), for example. The observation seems to go back to Curme (1931) and Jespersen (1931).
6 For (b) cf. Ehrich (1992), for (c) and (d) cf. Ehrich and Vater (1989).
SWITCH FROM INDIVIDUAL-LEVEL TO STAGE-LEVEL PREDICATES (in past perfect clauses). Interestingly, in past perfect sentences like (2-7), we discover the opposite effect - namely, that a typical individual-level predicate is reinterpreted as a stage- level predicate.
(2-7) a. Hans war intelligent gewesen.
Hans was/has intelligent been b. Eduard war Schweizer gewesen.
Eduard was/has Schweizer been
INDEFINITE READINGS ('experiential perfect' cf. Comrie (1976:58f), 'existential perfect' cf. von Stechow (1999)). In contrast to past tense clauses, present perfect clauses can have indefinite readings.
(2-8) a. Martin aß Regenwürmer.
Martin ate earthworms
≠ There is (at least) one time in the past (which is not specified) at which Martin ate earthworms.
b. Martin hat Regenwürmer gegessen.
Martin has earthworms eaten
= There is (at least) one time in the past (which is not specified) at which Martin ate earthworms.
While the present perfect in some of the sentences listed above - i.e. in the ones in (2-1) - behaves exactly like the past tense, it shows aspectual effects in others. This is the case with the completedness effects in (2-2) and perhaps (according to a wider and more vague view of aspect) with the present relevance effects in (2-3) - (2-6). Moreover, it often shows effects that are neither of a tenselike nor of an aspectual nature; this applies to the life-time effect in (2-5), to the stage- level to individual-level switch in (2-6), to the individual-level to stage-level switch in (2-7), and to the indefinite readings in (2-8). In the remainder of this chapter, we will be concerned with the question how the special effects of the present perfect can be accounted for. Section 3 provides a survey of some previous attempts to explain the occurrence and distribution of the effects and points out why these accounts fail.
Section 4 sketches the basic ideas and the architecture of my own account. In the remainder of the chapter, these will then be applied to the special effects of the construction.
3. Previous accounts
Not surprisingly in view of the observations mentioned above, there is an extended discussion in the literature on whether the German present perfect is a tense or an aspect semantically, or ambiguous between a temporal and an aspectual reading, or consisting of a temporal and an aspectual component. This section aims at presenting various accounts of the present perfect under the perspective of how they account for the meaning effects of the construction - or how they could account for them.
The accounts are grouped according to their basic assumptions concerning the semantics of the present perfect. Since there are so many accounts on the market, it is neither possible nor necessary to discuss each of them in detail. Hence, for each type of approach, I will focus on one or two typical recent analyses. The interested reader
is referred to the literature that presents discussions of other or more pieces of literature on the construction, especially Thieroff (1992:159ff). As we will see, all of the accounts have serious shortcomings, so that from section 4, we will have to look for an alternative approach.
3.1. The present perfect as a tense
This subsection sketches accounts that view the present perfect as a tense category, either one expressing a past tense meaning (Comrie (1976)) or one expressing an extended-now meaning (Fabricius-Hansen (1986), von Stechow (1999)). These accounts do not attribute any aspectual semantic component to the construction.
Consequently, they can explain the tense-like behavior of the construction, but it is not obvious how they can deal with the observation that the present perfect often exhibits additional aspectual effects.
3.1.1. Past tense accounts
Comrie (1976) suggests that the present perfect in German, especially in southern German dialects, has the same meaning as the past tense. This immediately captures the fact that the present perfect and the past tense often can be substituted by each other, of course.
The approach cannot at all account for the fact that the present perfect quite often shows additional aspectual effects, however. Hence, it can be dismissed without further discussion. However, there are some more subtle arguments against past tense accounts. As von Stechow (1999: section 5) points out, the past tense and the present perfect in German behave differently when embedded under a past tense, for instance. Thus, while a past tense embedded under a past tense allows for simultaneous as well as shifted readings, a present perfect embedded under a past tense only allows for shifted readings but not for simltaneous readings.
(3-1) a. Fritz dachte, daß Maria krank war.
Fritz thought that Maria sick was
= SIMULTANEOUS READING: Fritz thought at t that Maria was sick before t.
= SHIFTED READING: Fritz thought at t that Maria was sick at t.
b. Fritz dachte, daß Maria krank gewesen ist.
Fritz thought that Maria sick been is
≠ SIMULTANEOUS READING: Fritz thought at t that Maria was sick before t.
= SHIFTED READING: Fritz thought at t that Maria was sick at t.
This shows immediately that the past tense and the present perfect cannot be synonymous.
3.1.2. Extended-now accounts
McCoard (1978) suggests an extended-now analysis of the present perfect construction in English. Fabricius-Hansen (1986) and von Stechow (1999) try to apply this idea to the German present perfect.7 Here is a summary of von Stechow's proposal.8
Von Stechow (1999) pursues a pronominal account of tenses. Thus, tenses are not taken as relations between points or intervals of times but rather denote time intervals that are determined by the context or correpond to time variables.
According to von Stechow, in perfect constructions the past participle as such does not carry any anteriority meaning. Rather, the auxiliary in the perfect construction introduces a past interval; this interval, however, is not an absolute past interval but a relative past interval. It starts at some contextually or lexically limited time in the
"past" (i.e. in the past calculated from an evaluation time) and is limited on the right side by the right edge of the reference time R of the clause. Since in the canonical case of a present perfect construction it includes thus the time of utterance, the interval is called an "extended now" interval, abbreviated here "XN."9
The XN-interval is linked to the nonfinite part of the clause by an adverb of quantification. This adverb of quantification takes the XN-interval as a restrictor and the participial VP as its nucleus. It can quantify either over events or over times.
The adverb of quantification determines whether the construction receives a universal reading (3-2a) or an existential reading (3-2b); a perfect construction which does not contain an overt adverb of quantification like always or for two hours does never receive a universal perfect reading. In the case of an existential reading, the reading triggers an anteriority effect and sometimes a completion effect, too. If I understand von Stechow right, the completion effect arises when the adverb of quantification quantifies over events, but not when it quantifies over times.
Quantificational adverbs like einmal ('once') or zweimal ('twice') always requires an event reading.
(3-2) a. UNIVERSAL PERFECT reading:
Ich bin immer krank gewesen.
I am always sick been
"For every time interval included in XN, I am sick."
b. EXISTENTIAL PERFECT reading with mere anteriority:
Ich bin krank gewesen.
I am sick been
"There is a time of my being sick contained within XN."
c. EXISTENTIAL PERFECT reading with completion:
Ich bin (einmal) krank gewesen.
I am (once) sick been
"There is a situation of my being sick contained within XN."
7 The account of Grewendorf (1995) is related to this type of approach insofar as it also assumes a kind of extended now. Since it contains an aspectual component, however, it will be dealt with as a complexity account.
8 Von Stechow applies this approach to the future tense, too.
9 In Fabricius-Hansen (1986), it is called "unechter Vergangenheitsbereich" ('improper past interval').
It is not clear to me how this approach extends to the analysis of past perfect constructions and of perfect infinitives. The main problem is that if the perfect auxiliary denotes an extended-now interval, it should do so in general. But if this is so, then it is not clear how the meaning of the past perfect and of perfect infinitives can be derived without further stipulations.
In any case, the approach does not have much to say about many of the effects introduced in section 2, e.g. the switches between stage-level and individual- level readings. Another problem is that the account cannot explain several phenomen that were accounted for in chapter II - i.e. the stative character of the construction, the behavior of temporal adverbials, and the distribution and effects of focus accent.
Moreover, the device of distinguishing between mere anteriority readings and completion readings, i.e. the distinction between quantification over times and quantification over events seems unnecessary. I.e. as will become clear in the course of this chapter, the three readings illustrated in (3-2) can be derived in my account without further stipulations.
3.2. The present perfect as an aspect
Aspectual accounts were suggested by Brinkmann (1962) and Glinz (1970). They assume that the present perfect has an aspectual reading, meaning either completion or present relevance. They cannot explain the lack of the aspectual effects in some examples - i.e. Hans hat geschlafen or Goethe hat in Weimar gelebt.
3.3. The present perfect as tense/aspect ambiguous
Tense/aspect ambiguity accounts assume that the present perfect is ambiguous between a past tense and an aspectual completion reading. Such accounts were proposed by Wunderlich (1970) and Bäuerle (1977, 1979). In this type of account, it remains completely unclear which factors determine the choice of one reading or the other.
3.4. The present perfect as a tense/aspect complex
Complexity accounts of the present perfect assume that the construction consists of both a tense component and an aspect component. The tense component is provided by the morphosyntactic present tense, while the aspectual component is characterized as "completion." Accounts of this type were proposed, for instance, by Ballweg (1988a, 1988b), Janssen (1988), Grewendorf (1995), Ehrich and Vater (1989), Ehrich (1992), and Zeller (1994). Their main problem is that obviously, the aspectual completion component is not always at work (cf. the data presented above); if the semantics of the present perfects always provides such an aspectual component, it remains unexplained how and why this component sometimes shows and sometimes hides. Moreover, not all of the meaning effects of the construction can be captured as triggered by the components a complexity semantics provides.
Let us take a more detailed look at the account of Grewendorf (1995).
Grewendorf analyses the present perfect as a combination of a present tense meaning, an aspectual component expressing completion, and anteriority ('Vorzeitig- keitsbezug').
In his analysis, the present tense is a neutral tense that localizes the situation time of the main predicate of its clause as intersecting with a contextually given reference time ('Betrachtzeit'). The reference time is obligatorily definite and determined only by the context; i.e. it is in no way temporally constrained by the present tense.10 The past participle expresses the completion of the situation expressed by the verb. The anteriority comes about as a cosequence of the completion aspect: the assertion of completion implies the assertion of anteriority of the situation denoted by the verb stem.
The present tense auxiliary is associated with a reference time R1. The construction as a whole asserts that the completion of the verb situation, which is expressed by the participial verb, is given in R1.
However, when did the situation expressed by the verb stem take place?
Grewendorf claims that the verb stem is associated with a reference time, too.11 Let us call this second reference time R2. Like R1, R2 can be provided by the context or by temporal adverbials, too. In principle, R2 can start at any point in the past, i.e.
roughly speaking, anywhere on the left side of a time arrow. It is limited, however, on its right side; Grewendorf proposes that its right side is limited by the right edge of the interval given by R1. Hence, the analysis implies that the completion of the verb situation must be reached at the end of the reference time R1. This is illustrated in (3- 1). Note that in the case illustrated, the two reference times R1 and R2 overlap.
According to Grewendorf, this corresponds to the characteristic reading of the present perfect. They have the option not to overlap, however, triggering a past tense reading in this case. This is indicated in (3-3) by the dotted line on the right side of R2.
(3-3)
TU| R1
limited by the context or R2 by a temporal adverbial
limited by the right edge of the present-tense-Rl
Grewendorf's account is constructed so as to deal with two typical readings of the present perfect - with past tense readings on the one hand, and with so-called universal readings, where the situation time of the verb reaches up to the time of utterance, on the other hand.
However, the account suffers at least from the following shortcomings. First, as was illustrated above, the present perfect does not assert completion of the verb
10 When the clause occurs out-of-the-blue, the reference time includes the time of utterance, because this time is always especially prominent in the context. Future readings of the present tense are restricted by conversational implicatures, i.e. they are not acceptable when the speaker is not maximally informative. Finally, historical present tense readings can occur when the context or explicit temporal adverbials provide a definite time interval to which the tense can express simultaneity.
11 In fact, Grewendorf says that the participle is associated with the second reference time. But Grewendorf clearly means to say that this second reference time is responsible for the location of the situation expressed by the verb.
situation at all. In principle, the verb situation can reach beyond the time of utterance and/or beyond the reference time (or, in the present terminology, tense time) infinitely into the future. This comes out expecially clearly with predicates like be dead.
Second, although in this respect the account could in principle perhaps be augmented by pragmatic principles, in the present version, Grewendorf's account does not have much to say about the switches from stage-level to individual-level readings and from individual-level to stage-level readings of the predicates in perfect constructions that were introduced in section 2.
3.5. The present perfect as scope ambiguous
Klein (1997, 1998a) develops a scope proposal in order to deal with the past tense reading and the completion reading of the present perfect in German. As we have already seen above, these two readings can be brought about by continuations of the clauses as in Klein's examples in (3-4).
(3-4) a. Peter hat gearbeitet und ist müde.
Peter has worked and is tired
b. Peter hat gearbeitet und wollte nicht gestört werden.
Peter has worked and wanted not disturbed become/to-be
Klein describes the reading of the two examples as follows. In (3-4a), the speaker describes a present state as the result of an earlier situation of working. However, in (3-4b), he does not describe what Peter is like right now as the result of the earlier situation of working; rather, he talks about this situation in the past itself. As Klein points out, "(u)nder both readings, there must be some situation in the past at which the person designated by the subject worked... What is different, is the way in which this situation relates to the present" (p2). The contrast can also be brought about by the use of temporal adverbials. When the construction co-occurs with a present time adverbial, the first reading is brought about; when it co-occurs with a past time adverbial, the second reading is brought about.
Interestingly, the contrast does not arise with the present perfect in English;
English shows only a reading corresponding to (3-4a), but not one corresponding to (3-4b). Moreover, the lack of the contrast seems to be related to some other well- known observations concerning the present perfect in English. Consider the famous contrast in (3-5).
(3-5) a. Einstein hat Princeton besucht.
Einstein has Princeton visited b. #Einstein has visited Princeton.
Klein argues that the two readings in German arise because of a scope ambiguity that does not occur in English. This accounts both for the occurrence of the ambiguity in (3-4) and for the acceptability contrast between English and German in (3-5). Simplifying somewhat, the past participle morpheme and the auxiliary together denote a function called "POST" that assigns posttimes to its argument. The function can apply either to the VP or to the VP plus the subject. I.e.
(3-5a) results in two readings that can be represented as in (3-6). The resulting posttimes can be identified with the topic time TT of the clause (which can, for the present purpose, be identified with out notion of tense time and common notions of
reference time; with the present tense PRES, TT is located at or after the time of utterance).
(3-6) a. POST has scope over the VP plus its subject:
[[POST [Einstein [Princeton besuch-]]] PRES ]
= TT is a posttime of a situation where Einstein visits Princeton.
b. POST has scope over the VP only:
[[Einstein [POST [VP Princeton besuch-]]] PRES ]
= Einstein has the posttime property of visiting Princeton at TT.
The reading represented in (3-6b) is only felicitous when still Einstein exists at TT.
Since Einstein is not alive anymore, this reading is infelicitous. Klein argues that English does not have the option of exhibiting a logical form corresponding to (3- 6b). This captures the fact that (3-5b), the English equivalent to (3-5a) is infelicitous.12
Although the scope account seems interesting as well as desirable because it draws on independent evidence and captures some apparently unrelated aspects of the behavior of perfect constructions in English and German, we have to ask how much the proposal really helps us in explaining the behavior of the present perfect and its numerous special effects. Unfortunately, like the types of approaches discussed above, it explains only a relatively small range of the readings and uses of the present perfect. Thus, it does not help to understand how the stage- level/individual-level switches come about.
Of course, from the fact that the assumption of a scope ambiguity explains only part of the phenomena related to the present perfect we cannot cnclude that the scope account is fundamentally inadequate, i.e. that there does not exist a scope ambiguity of the sort described by Klein. We will come back to an evaluation of the approach in chapter IV.
3.6. Conclusion and outlook
As was shown in the preceding subsections, each of the previous accounts of the present perfect has serious shortcomings. Many accounts have difficulties dealing with the occurrence of past tense readings and completion readings. Moreover, none of the aforementioned accounts seems to be able to provide satisfactory explanations for the full range of effects that perfect constructions in general and present perfect construction in particular can have - especially for the switches between stage-level and individual-level readings.
At this point, it is obvious that any adequate account of the present perfect must provide a considerable degree of flexibility: the construction exhibits numerous meaning effects that come and go depending on a large number of semantic and/or pragmatic factors. Some such effects appear to be of a more or less vaguely aspectual nature, some others, however, are not.
Of course, the choice of the basic account of the present perfect plays a role for how the meaning effects of the present perfect can be dealt with. Accounts that do not provide a basis for the aspectual effects within the semantics of the
12 Interestingly, the literature on English and German syntax and the syntax-semantics interface perhaps provides independent evidence for the assumption that the scope ambiguity arises in German but not in English.
construction, for instance, have to attribute the effects to pragmatics in some way.
Obviously, this also holds for the account I proposed in chapter II. Given that pragmatics often has a wastebasket-function, this prospect at first sight may not appear desirable. However, as was already pointed out above, if the effects were due only to the semantics of the construction, then it would remain unexplained that the effects are sometimes lacking. This consideration certainly helps to sympathize with a pragmatic explanation in general. In the remainder of this chapter, I will propose a pragmatic account that is able to derive all the meaning effects introduced in section 2 quite naturally.
4. How pragmatic principles operate on the semantics
In this section, I will show that the semantic characteristics of the perfect construction invites the application of certain pragmatic principles. Specifically, I will argue that the special effects associated with present perfect constructions are in each case the result of the interaction of the following components: 1. the semantics of the perfect constructions, 2. the semantics of the verb contained in it, 3. a pragmatic principle that functions on the basis of the Gricean maxim of informativity which I will call the "Principle of Informative Contrast" (PIC), 4. the free assignment of topical status to times, 5. the possibilities how time variables can get values depending on the context in which they occur, and 6. some general principles concerning the interpretation of tense times and other temporal elements.
(4-1)
Readings of perfect- constructions 1. Semantics of
perfect-
constructions 2. Verb semantics
6. Principles of temporal interpretation
3. Principle of Informative Contrast between TS and TT
5. Assignment of time values to context- ually given variable that restrict TTs or TSs
4. Free assignment of topical status to TTs or TSs
It is important to note that the principles that play a role here can also be observed independently of perfect constructions. Hence, they do not have to be stipulated in order to analyse the behavior of perfect constructions descriptively adequate; rather, they are independently motivated. Moreover, note that the lack of the relevant effects with the past tense is due to the fact that the semantic structure of the past tense does not provide the same conditions for the operation of the pragmatic principles. Let us start by having a closer look at the ways in which we expect pragmatic principles to function on the semantic input.
4.1. Situation times of VP and resultant-state: informative contrasts
Recall from chapter II that perfect constructions refer to two crucial time intervals, the situation time of the complex constituted by [VP+PART+AUX], i.e. of the resultant- state, and the situation time of the VP. If one refers to two distinct time intervals, however, then the two time intervals should provide different content or information.
The idea behind this consideration is a combination of the Gricean maxims of relevance and informativeness, of course: if reference to two distinct time intervals is being made, then each of them should be relevant by providing information different from the one provided by the other time interval. If one of them does not provide any crucial information, then why refer to it?
Hence, it is to be expected that the use of perfect constructions must be licensed by the satisfaction of the PRINCIPLE OF INFORMATIVE CONTRAST (henceforth, PIC), which is stated in (4-2).13 It requires the two time intervals crucial in the interpretation of the perfect construction - i.e. the situation time of the VP and the situation time of the resultant-state - to differ in some relevant way.
(4-2) Principle of Informative Contrast (PIC)
In a perfect construction, the situation time of the VP and the situation time of the resultant-state differ in some relevant way.
Because of the PIC, by using a perfect construction, a speaker always implicates that there is some contrast between the situation time of the VP and the situation time of the resultant-state. Its being a pragmatic principle accounts for some of its properties - i.e. that its force can be strengthened or weakened by contextual factors and that for pragmatic reasons it can attain different appearances. Its possible appearances and what their nature depends on will be illustrated in more detail below. Note that it is to be expected that what counts as a relevant and informative contrast will presumably correlate with other semantic or pragmatic factors that have to do with relevance and informativeness - e.g. topic-focus structure. Some of the possible appearances of the PIC will be illustrated below.
4.2. Values of times
Above, we identified the tense time of a clause and the situation time of its VP as two times that are relevant for temporal interpretation. According to the formal semantic account presented in chapter I, the tense time is represented as the restrictor of an adverb of quantification "∃T", while (the relevant part of) the situation time of the VP is introduced by an adverb of quantification "∃A". As was shown in chapter II, the situation time that is introduced by "∃T" in perfect constructions is in fact the situation time of the resultant-state, i.e. of the complex [VP+PART+AUX]. Finally, an adverb of quantification "∃P" introduces a relevant part of the situation time of the underlying, i.e. non-participial, VP. Each of these adverbs of quantification may in principle be phonetically empty - in which case it receives an existential default- interpretation - or realized by an overt adverb of quantification. Moreover, each of
13 I am not quite sure about its relationship to Klein's (1994:207) PRINCIPLE OF REASONABLE
CONTRAST: "The assertion, which extends over the focused part of the lexical content, must allow for a reasonable contrast."
the adverbs of quantification can, as any quantifier, be restricted by various types of material. Thus, "∃T" is usually restricted by the tense of the clause, "∃A" by an aspectual relation, and "∃P" by the anteriority relation of the past participle. Each of these quantifiers may, however, also be restricted by time values given by temporal adverbials or by implicit time values that are provided by events mentioned in the context, by contextually salient times, or even by times suggested by the denotation of noun phrases.
In this subsection, we will illustrate a number of possibilities for providing such restricting time values. For presentational reasons, we will start with a consideration of the possibilities for providing time values as restrictors of "∃T" only, but later we will also consider the options for providing time values that serve as restrictors of
"∃P".
4.2.1. Direct time values
Let us see in more detail now how the quantifier "∃T", which introduces the tense time by the material in its restrictive clause, can be restricted. The tense of a clause provides a very large initial value for the tense time of a clause. Thus, a past tense designates all the time before the time of utterance as the initial tense time of its clause (= TT<TU). It must be observed, however, that the past tense requires there to be at least some minimal distance between the tense time and the time of utterance.
With a present tense, all the time around the time of utterance - i.e. all the time - or the time after the time of utterance (= TT≤TU) can serve as the initial tense time. Finally, according to common assumptions, with a future tense, all the time after the time of utterance can serve as the initial tense time (= TT<TU).14
These very large maximal tense time values can then be further restricted or specified by other factors (cf. Klein (1994)). If they are not further specified, their restriction is minimal; thus, MINIMAL RESTRICTIONS of a time lead to MAXIMALLY LARGE TIME INTERVALS.
The factors that can further specify frames for tense times are numerous. In principle, all contextually prominent elements that can be used to construe time values can be useed for this purpose.15 Thus, the following list of factors and illustrations is not necessarily complete. In addition to tense (4-3a), also previously mentioned events (4-3b), temporal adverbials (4-3c), the time of utterance itself and its close environment, which is always contextually salient (4-3d), and the time just before the time of utterance, which is presumably also always contextually prominent (4-3e), can specify tense times.
(4-3) a. Männer sind klüger als Frauen, und die Erde ist eine Scheibe.
men are smarter than women, and the earth is flat b. (Sie fanden Barschel in der Badewanne.) Er war tot.
they found Barschel in the bathtub. he was dead c. Am Freitagabend trank Maria drei Caipirinhas.
on-the Friday-evening drank Maria three Caipirinhas
14 This may not be the most adequate analysis, though. There is evidence that the present tense, which is clearly morphosyntactically present in future tense constructions, has to be taken seriously on the semantic level, too. For some details, see Musan (1999) on werden.
15 The idea that topic times - or, in her account, "Betrachtzeiten" - can be provided by a variety of factors has also been exploited by Fabricius-Hansen (1986).
d. Hans ist (jetzt) vollkommen verwirrt.
Hans is (now) totally confused
e. Hans hat (jetzt) seine Kreditkarte wiedergefunden.
Hans has (a second ago/now) his creditcard again-found
In principle, all contextually relevant factors can be used in order to construct time values.
It is important to see that tense time specifications cut out pieces from the initial tense time that is provided by the tense of the clause. They can never give us an interval that is not contained in the time designated by the tense. Technically, this is a consequence of the conjunctive, i.e. intersective, interpretation of the material that is contained in the restrictive clause of a quantifier. Thus, when a past tense has given us the time before the time of utterance as the initial tense time value, then a second TT-specifying factor can only cut out a piece from the time before the time of utterance. This is shown in (4-4b).
(4-4) a. Gestern machte Hans einen langen Spaziergang.
yesterday made Hans a long walk
b. Heute machte Hans einen langen Spaziergang.
today made Hans a long walk
The tense time of (4-4b) can only be a past interval that is contained in today. It cannot reach beyond the time of utterance into the future, although today may very well not be over yet at the time of utterance.
The examples we have looked at so far were concerned with the values of tense times. As said above, values for the restrictor of "∃P" in perfect constructions are analogously provide and will be dealt with in more detail below. Note that in the examples mentioned so far, elements in the linguistic or nonlinguistic context provided time values directly. Also noun phrases, however, - especially when topical - can provide time values. This will be shown in the next subsection.
4.2.2. Time values provided by noun phrases
The purpose of this subsection is twofold. First, I want to show that in some cases, tense times can be restricted by noun phrases. Second, I want to introduce the phenomenon of life-time effects, which tenses sometimes seem to impose on their arguments.16 Both phenomena - life-time effects of various kinds as well as tense times that are provided by noun phrases - will help us later to understand better what is going on in perfect constructions semantically and pragmatically.
Let us first consider a past tense clause that contains an individual-level predicate and appears in a temporally specified context (4-5a).
16 A more detailed account of both phenomena can be found in Musan (1995: Ch.II) and in Musan (1997). The idea that topic times - or, in their terms, "Betrachtzeiten" - can be provided by the time of existence of individuals denoted by noun phrases in the clause has also briefly been sketched by Ehrich and Vater (1989).
(4-5) a. (An jenem Tag wurde ich Hans vorgestellt.) Hans war Amerikaner.
(on that day became/was I Hans introduced) Hans was American
b. (4-5a) asserts about the tense time TT - i.e. about the time of that day when I was introduced to Hans - that Hans was American at TT.
How is the second sentence in (4-5a) interpreted? Under its most likely reading, the first sentence provides the tense time for the second clause - namely, the time at which I was introduced to Hans. The second sentence is an assertion about what was the case at that time - namely, that Hans was American. If understood in this way, then the sentence does not say anything about any other time but the time when I was introduced to Hans.
But what happens when such a sentence appears in a temporally unspecified context? In this case, there is no genuine time interval directly available that can provide a specification of the tense time. This is a situation where noun phrases are used for the construction of tense time values. This can be achieved by taking the time of the individual denoted by a noun phrase - i.e. the time when the individual exists or lives - as a value for the tense time frame. Of course, such a construction of a time value from an individual denotation is an extra interpretational effort as compared to the use of direct time values. Consequently, we expect that the construction of time values by noun phrases probably only takes place when no direct time values are available.
Thus, when the noun phrase Hans is present in the clause, the time of existence of Hans may be taken as a value for the tense time. Recall, however, that tense time specifications must be compatible with the tense time provided by the tense; additional specifications can only cut out a part of the tense time that was provided by the tense. Hence, for a stage-level clause (4-6a) we get (4-6b).
(4-6) a. Hans war krank.
Hans was sick
b. (4-6a) asserts about the tense time TT - i.e. about Hans' time of existence in the past - that Hans was sick at TT.
The interpretation of an individual-level clause under similar circumstances, i.e. in temporally unspecified contexts, should happen analogously. However, past tense individual-level clauses in temporally unspecified contexts show an additional effect:
they suggest that the individual denoted by the subject noun phrase is dead now.
This LIFE-TIME EFFECT (marked "†")is illustrated in (4-7).
(4-7) a. †Hans war Amerikaner.
Hans was American
b. (4-7a) asserts about tense time TT- i.e. about Hans' time of existence in the past - that Hans was American at TT.
I will now argue that the life-time effect of (4-7a) arises because in a situation where Hans is still alive, while both the present tense clause (4-8a) and the past tense clause (4-8b) are true, the present tense clause (4-8a) is more informative than the past tense clause (4-8b) (= (4-7a)).
(4-8) a. Hans ist Amerikaner.
Hans is American
b. †Hans war Amerikaner.
Hans was American
The idea is that in contrast to (4-8b), (4-8a) uses the maximally possible degree of informativeness with respect to the temporal duration of Hans' being American.
With (4-8b) Hans war Amerikaner the largest time interval for which one can say that Hans was American is the time interval that starts at Hans's birth and ends right before now; it is labelled "ta" in the illustrating picture (4-9). With Hans ist Amerikaner the largest time interval for which one can say that Hans is American is the time interval that starts at Hans's birth, includes now, and ends at the time of his death; this time interval is labelled "tb" in (4-9).
(4-9)
TU
....
Hans' birth
ta
tb Hans' death
Thus, provided that both the present tense and the past tense version are true, the present tense sentence (4-8a) provides more information about the duration of Hans's being American than the past tense sentence (4-8b).17
Given that (4-8a) is more informative about the duration of Hans's nationality, Grice's (1975:45) Maxim of Quantity - Be as informative as is required, not more and not less - explains why (4-8b) appears inappropriate: there is a more informative way of making a statement about the duration of Hans's nationality. Hence, (4-10) holds.
(4-10) If for some A and some B both "A is B" and "A was B" are true, then use the utterance "A is B" instead of "A was B".
Strictly speaking this does not yet account for why (4-8b) triggers a life-time effect.
The fact that (4-8a) is more appropriate than (4-8b) in a situation where Hans is alive might, according to (4-8), justify the intuition that (4-8b) is less good than (4-8a), but it does not explain the life-time effect of (4-8b). The difference with respect to informativeness, however, is the basis for an implicature which leads to the conclusion an utterance conveys that Hans is dead. What one implicates directly is that the property of being American is "over". The further implication that Hans is dead follows from the fact that Amerikaner sein is an individual-level predicate. In (4-11), a justification schema for the implicature and its consequences is spelled out for a situation where a speaker has uttered Hans war Amerikaner.18
17 For a more detailed discussion of the issue of informativeness as applied to present tense and past tense clauses, see Musan (1997).
18 The schema presented here differs somewhat from the one presented in Grice (1975, 50).
For instance, some steps of derivation concerning the transmission of information between speaker and hearer are left out, and I am adding steps that concern the consideration of alternative expressions and informativity.
(4-11) Working out of the implicature of Hans war Amerikaner:
(a) The speaker has expressed the proposition that Hans was American at Hans time of existence in the past.
(b) Thus, the hearer assumes that the speaker is maximally informative about Hans's being American at Hans' time of existence19 - in particular about the duration of Hans's being American at Hans' time of existence.
(c) If the speaker thought that Hans's being American at Hans' time of existence is not over, he would have expressed the proposition that Hans is American at Hans' time of existence, since that would have been a more informative alternative utterance about the duration of Hans's being American at Hans' time of existence.
(d) Thus, the speaker couldn't have been maximally informative about Hans' being American at Hans' time of existence unless he thought that Hans' being American at Hans' time of existence is over.
(e) Thus, the speaker has implicated that Hans' being American at Hans' time of existence is over.
(f) Since being American is a property that, if it holds of an individual at all, holds of that individual during all its life-time, the hearer concludes further that, since the speaker has implicated that Hans' being American at Hans' time of existence is over, the speaker has implicated that Hans is dead.
In this way, the utterance of Hans war Amerikaner triggers a life-time effect.
There is more evidence for the specific account I just sketched. Interestingly, there are data that show even more clearly that life-time effects are due to the ability of noun phrases to provide tense time values, and specifically, for the tendency of choosing topical noun phrases for this purpose. Consider the sentences in (4-12), which contain predicates with more than one argument. The property of resemblance is here meant to be understood as a fundamental similarity that amounts to a life-time property and hence, to an individual-level predicate. The examples differ insofar as in the more unmarked (4-12a), the subject noun phrase is topicalized, whereas in (4- 12b), the dative object is topicalized. The sentences also differ with respect to which noun phrase carries a life-time effect.
(4-12) a. †Tante Theresa ähnelte meiner Großmutter.
[SUBJ Aunt Theresa] resembled [OBJ my grandmother]
('Aunt Theresa resembled my grandmother.') b. (†)Meiner Großmutter ähnelte (†)Tante Theresa.
[OBJ my grandmother] resembled [SUBJ Aunt Theresa]
('Aunt Theresa resembled my grandmother.')
The interesting observation about (4-12b) is that compared to (4-12a), the life-time effect has shifted. With (4-12a), there is a life-time effect concerning Aunt Theresa; in order for the sentence to be appropriate when uttered out of the blue, Aunt Theresa has to be dead. The grandmother being dead with Aunt Theresa being alive does not save the sentence. But with (4-12b) it seems that the being dead of either one of Aunt Theresa and the grandmother is sufficient to make the sentence fully acceptable.
19 The addition "in the past" can be dropped here, because in the competing present tense clause, the restriction to the past would, of course, be dropped. Hence, it might in fact be confusing to add "in the past" here. That is the reason why I dropped it.
I propose that the association of life-time effects with particular noun phrases is due to their being more or less topical, because topical noun phrases are preferred above nontopical noun phrases for construing time values. The preference for topical elements as value providers is, of course, not very surprising. Topical elements are in general contextually especially salient. Thus, it seems very plausible that in contexts that do not supply genuine temporal values for times, topical elements are the most preferred candidates for supplying them.
If this is so, then - since subject noun phrases are default topics - it follows automatically that in (4-12) the subject noun phrase Tante Theresa is the strongly preferred value provider. This holds at least UNLESS some additional factor or other - like certain syntactic movements, e.g. topicalization or scrambling, or a marked intonation structure - changes the preferred topic-focus structure of the clause. Such a change of the topic-focus structure is given in (4-12b). Let's assume that the topicalization of meiner Großmutter leads to a situation where this noun phrase is more likely to be the topic of the sentence and competes with the subject for topical status.20 Thus, for (4-12b) we get the assertion described in (4-13a), and the shifted life-time effect of (4-12b) is derived as sketched in (4-13b).
(4-13) a. (4-12b) asserts about my grandmother's time of existence that Aunt Theresa resembled my grandmother.
b. Working out of the implicature of (†)Meiner Großmutter ähnelte Tante Theresa:
(a) About my grandmother's time of existence, the speaker has expressed the proposition that Aunt Theresa resembled my grandmother at my grand- mother's time of existence.
(b) Thus, the hearer assumes that the speaker is maximally informative about the duration of Aunt Theresa's resembling my grandmother at my grandmother's time of existence.
(c) If the speaker thought that Aunt Theresa's resembling my grandmother at my grandmother's time of existence is not over, he would have expressed the proposition that Aunt Theresa resembles my grandmother at my grandmother's time of existence, because that would have been a more informative alternative utterance about the duration of Aunt Theresa's resembling my grandmother at my grandmother's time of existence.
(d) Thus, the speaker could not have been maximally informative about Aunt Theresa's resembling my grandmother at my grandmother's time of existence unless he thought that Aunt Theresa's resembling my grandmother at my grandmother's time of existence is over.
(e) Thus, the speaker has implicated that Aunt Theresa's resembling my grandmother at my grandmother's time of existence is over.
(f) Since resemblance (in the sense useed here) is a property that, if it holds of an individual at all, holds of that individual during all its life-time, the hearer concludes further that, since the speaker has implicated that Aunt Theresa's resembling my grandmother at my grandmother's time of existence is over, the speaker has implicated that the grandmother is dead.
20 Things are in fact not that simple. It seems that topicalization in German can lead either to a preferredly topical or to a preferredly focal nature of the topicalized element. But since the life- time effect can apparently concern either the grandmother or Aunt Theresa, this is no problem for my approach.
Thus, the assumption that topical noun phrases are able to supply time values for tense time frames can explain the data in (4-12) that would remain unexplained otherwise.
The Apollo-13-Principle:
Failure is not an option!
4.2.3. A survey of the options for providing time values
It was shown above that values for tense times can be provided by the linguistic or nonlinguistic context. The same holds for the values of the situation times of VPs in perfect constructions, i.e. for the time values introduced by "∃P". This subsection aims at providing a brief survey of how the variants of time values for these frames that can possibly occur are further restricted under various specific circumstances, e.g. in temporally specified contexts in contrast to temporally unspecified contexts. Also we will take a look at how present perfect sentences and past perfect sentences differ in this respect.
The tables (4-14) - (4-17) below show the options for receiving time values for the situation times of the VPs and for the tense times in present perfect and past perfect clauses. We consider the following options for time value specifications:
• Minimal: this refers to the case of minimal time value specification, i.e. either by the tense only or by the semantics of the past participle only. In both cases, we get maximally large time intervals - i.e. all the time before the time of utterance by the past tense (abbreviated TT<TU), all the time by the present tense (abbreviated TU≤TT), and all the time before the tense time by the past participle semantics (abbreviated TS≤TT).21
• TU: the time of utterance, which is always contextually salient.
• Just before TU: the time right before the time of utterance, which is presumably also always contextually salient.
• Adverbial/event: a time that is provided by a temporal adverbial or by previously mentioned events in the discourse; this type of specification is by definition only relevant in temporally specified contexts.
• (subject) NP: a time provided by the time of existence of a noun phrase denotation; as was said above good candidates for this are topical noun phrases and hence especially subject noun phrases because they are default topics.
Shading of a field in the tables below signals that the respective option of providing a time value is available.
(4-14) Options for time values in temporally unspecified contexts: past perfect providing value values for the situation time of VP values for the tense time
minimal Possible. TS≤TT. Not possible. Suppose TU≤TT
were the value. Then it would not be possible to calculate a time that precedes TT for the TS of VP.
TU Not possible since the time must be before TU.
Not possible since the past tense requires TT to be before TU.
just before TU Not possible since TU is not a possible value for TT.
Not possible since with past tense, TT must be distant to TU.
adverbial/event Not possible since the context is temporally unspecified.
Not possible since the context is temporally unspecified.
21 This is a consequence of the semantics introduced in chapter II.
(subject) NP Possible if the subject is alive before TT.
Possible if the subject is alive at TT.
(4-15) Options for time values in temporally unspecified contexts: present perfect providing value values for the situation time of VP values for the tense time
minimal Possible. TS≤TT. Not possible. Suppose TU≤TT
were the value. Then it would not be possible to calculate a time that precedes TT for the TS of VP.
TU Possible if TT is after TU. Possible.
just before TU Possible since TT is at or after TU. Not possible because of the present tense.
adverbial/event Not possible since the context is temporally unspecified.
Not possible since the context is temporally unspecified.
(subject) NP Possible if the subject is alive before TT.
Possible if the subject is alive at TT.
(4-16) Options for time values in temporally specified contexts: past perfect providing value values for the situation time of VP values for the tense time minimal Possible if the temporal specific-
ations do not concern the TS of VP.
TS≤TT.
Not possible. Suppose TU≤TT were the value. Then it would not be possible to calculate a time that precedes TT for the TS of VP.
TU Not possible since the time must be before TU.
Not possible since the past tense requires TT to be before TU.
just before TU Not possible since TU is not a possible value for TT.
Not possible since with past tense, TT must be distant to TU.
adverbial/event Possible since the context is temporally specified.
Possible since the context is temporally specified.
(subject) NP Possible if the subject is alive before TT. Because of direct temporal specification not salient, however.
Possible if the subject is alive at TT.
Because of direct temporal
specification not salient, however.
(4-17) Options for time values in temporally specified contexts: present perfect providing value values for the situation time of VP values for the tense time
minimal Possible. TS≤TT. Not possible. Suppose TU≤TT
were the value. Then it would not be possible to calculate a time that precedes TT for the TS of VP.
TU Possible if TT is after TU. Possible.
just before TU Possible since TT is at or after TU. Not possible because of the present tense.
adverbial/event Possible since the context is temporally specified.
Possible since the context is temporally specified.
(subject) NP Possible if the subject is alive before TT. Because of direct temporal specification not salient, however.
Possible if the subject is alive at TT.
Because of direct temporal
specification not salient, however.
In the next subsection, I will argue that the times that are introduced during temporal interpretation can be freely assigned topical status.
4.3. The topical status of times
Recall the concept of topic time used in Klein (1992, 1994) that was briefly sketched in chapter I. Topic times were associated with utterances and described as the time about which the speaker of an utterance wants to assert something. In Klein's account, topic times were claimed to be temporally located by the tense of the clause.
After introducing the concept of topic times in chapter I, I suggested to use the term
"tense time" instead. This step was taken for good reasons: as will become clear shortly, the topical or non-topical status of times in perfect constructions should be reconsidered in some respects. Specifically, I will argue that the topical status of times does not correlate with being located by the tense of the clause in all cases.
The concept of topic time implies that the application of temporal semantics to a proposition presupposes a decision about the topical status of times. Speaking in dynamic terms, it implies that first, the speaker must decide about which time she wants to assert something. In a second step, she temporally locates this time - the topic time - by an appropriately chosen tense and formulates the assertion about the topic time. Note that this is somewhat implausible insofar as it amounts to saying that times and temporal expressions behave quite differently than individuals and noun phrases do according to generally accepted principles of topic-focus - or: information - structure.
A general assumptions about noun phrases and their status with regard to topic-focus structure is that among the noun phrases in a clause, the speaker can decide in principle freely whether they are topical or not - regardless of their grammatical function. Thus it seems that, given the material provided by a clause, topics can be chosen freely. Whether a certain item is topical or not can be signalled or supported by several linguistic means, of course. In German, for instance, scrambling and intonation patterns as well as question/answer contexts strongly support certain topic-focus structures. Thus, being scrambled signals being topical (4- 18); raising accent signals focal status while falling accent signals topical status (4- 19); and being asked for in a wh-question implies being focal (4-20).
(4-18) a. UNMARKED: Maria hat gestern den Schlüssel dem Studenten geliehen.
Maria has yesterday the key the student borrowed b. Maria hat [T den Schlüssel] gestern dem Studenten geliehen.
Maria has the key yesterday the student borrowed
c. Maria hat [Tdem Studenten] gestern den Schlüssel geliehen.
Maria has the student yesterday the key borrowed
(4-19) a. UNMARKED: Maria hat gestern den Schlüssel dem Studenten geliehen.
Maria has yesterday the key the student borrowed \ /
b. Maria hat gestern [T den SCHLÜSsel] [F dem StuDENten] geliehen.
Maria has yesterday the key the student borrowed / \
c. Maria hat gestern [F den SCHLÜSsel] [T dem StuDENten] geliehen.
Maria has yesterday the key the student borrowed
(4-20) a. Wer hat gestern den Schlüssel dem Studenten geliehen? - [F Maria] hat Who has yesterday the key the student borrowed? - Maria has
gestern den Schlüssel dem Studenten geliehen.
yesterday the key the student borrowed
b. Wem hat Maria gestern den Schlüssel geliehen? - Maria hat gestern
To-whom has Maria yesterday the key borrowed? - Maria has yesterday den Schlüssel [Fdem Studenten] geliehen.
the key the student borrowed
c. Was hat Maria gestern dem Studenten geliehen? - Maria hat gestern What has Maria yesterday the student borrowed? - Maria has yesterday [ F den Schlüssel] dem Studenten geliehen.
the key the student borrowed
Given the flexible behavior of noun phrases with respect to topic-focus structure, why should times and temporal expressions behave so differently? Why should the time located by a tense be obligatorily topical? Why should it be impossible to assign topical or non-topical status to times as freely as to noun phrases? I cannot see any convincing reason why that should be so.
More crucially, there are also empirical reasons to reconsider the concept of topic times: their alleged role in temporal semantics contradicts intuitions about the interpretation of specific utterances. We have seen in chapter II that positional temporal adverbials in perfect clauses can function as TT-specifiers or as TS- specifiers. This is illustrated again with a past perfect construction in (4-21).
(4-21) Er war um zehn weggegangen.
he was/had at ten left
a. TS-SPECIFICATION: = The leaving took place at ten.
b. TT-SPECIFICATION: = At ten, he was gone.
When the adverbial is understood as a TT-specifier, we indeed have the intuition that the speaker asserts something about the tense time - i.e. the time located by the tense of the clause. When the adverbial is understood as a TS-specifier, however, this seems counterintuitive. Rather, it seems that the speaker asserts something about the situation time of the VP. Taking these intuitions seriously, we cannot but conclude that both the tense time and the situation time of the VP can appear as TOPICAL TIMES.22, 23 Analogous considerations apply to present perfect clauses and future perfect clauses.24
22 For details see Klein (1994:165). I tend to attribute a stronger role to temporal adverbials in the assignment of values for topic times than Klein (1992). I do not find his view in this respect very convincing - e.g. in connection with historical present tense, it leads to quite unintuitive consequences, for example that Jörg Bieberstein wird 1956 geboren. does not have 1956 as its topic time, but at least the whole time interval from 1956 up to the present.
23 A similar account is pursued by Waugh (1987) for French and by Dentler (1997) for the historical development of the present perfect in German. Moreover, also Marillier (1997:
section 3.1.1.) hints at a non-arbitrary connection between topical status of times and the presence and reference of temporal adverbials. According to him, a temporal adverbial shows
"welche von beiden Prädikationen (i.e. the one of the VP-situation or the one of the resultant- state, R.M.) informativ die wichtigere ist."
24 In principle, Klein (1992) seems to be perfectly aware of the fact that in (German) perfect constructions, the interpretation of the overt morphological tenses does not correspond to the generalization that tenses locate topic times: "Das deutsche System der Tempusmarkierung ist einigermaßen verwickelt. (...) Für TT< [TT vor TU] gibt es das Präteritum (schlief) wie auch das Perfekt (hat geschlafen), die beide besondere Gebrauchsbedingungen haben. Wie dies im
