Вы находитесь на странице: 1из 15


Johan Nordlander

2 3

The English Present Perfect: A discourse marker of intra-sentential cohesion.1


5 6 7 8 9 10 11 12 13 14 15 16 17 18

This article will propose a somewhat different analysis of the meaning of the perfect. Introductory books and articles touching on the subject, such as Comries Aspect (1976), Andersons 1982) The Perfect as a Universal and as a Language-Particular Category, and Schwenters Hot News and the grammaticalization of Perfects (1994), among others, have all tried to approach the problem of what the general, or the most typical, meaning of the present perfect actually is. What these authors, and others before them, have arrived at is a typology of different perfects, given labels such as the perfect of result and the perfect of recent past, etc. Other authors 2 , such as Klein (Time in Language, 1994), Klein and Vater (The Perfect in English and German, 1998) and Elsness (The Perfect and the Preterite in Contemporary and Earlier English, 1997) have assumed that the general meaning of the present perfect is possible to establish in terms of temporal relations only. Thus Elsness claims that a present perfect verb form will express strong deixis, since it does establish the temporal reference by itself. (Elsness, 1997:10) What the above-mentioned authors have not addressed in their analyses is the possiblility of a discourse function which would turn out to be common for all these perfects. 3

19 20 21

22 23 24

1 2 3

I am very grateful to Anders Steinvall and Jrgen Svensson for reading and commenting on earlier versions of this article.

version date: November 5, 1998 1


The Problem

2 3 4 5

Instead, they have sought to describe the nature of the state obtaining or process developing at the identified temporal point of reference. The problem is that sometimes we cannot find any explicit state or process at that point of reference. In a sentence such as (1) below, for example He has lived in Singapore.


7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

no state as such can be said to obtain at its temporal point of reference (here this point is presumably the moment of speech). Instead, under normal contextual circumstances, we will infer the opposite: that the state of living in Singapore does not obtain at the chosen reference point. This means that in order to make sense of (1) we have to (a) assume an implicit temporal point of reference, and (b) put the explicit statement in example (1) in some sort of relation to this implicit point of reference. That is, only by using our pragmatic knowledge of the world which says that utterances do make sense under normal circumstances (cf. Grices Cooperative Principle, 1989:2627) can we properly understand example (1). In a sense the hearer cooperates with the speaker in creating the message of example (1). However, in order to discuss the meaning of the present perfect, it is important to bear in mind that in the present article this category is conceived of as a semantico-grammatical category, rather than a morphological one. Thus we should not confuse the present perfect in the discussion below with the way some languages employ, what we here might call, perfect morphology to indicate the preterite, for example the French pass compos (cf. Comrie 1976: 53). 1.2 Terminological Preliminaries



Here definitions of TMA, Verbal constituency, dynamicity etc. will be found.

version date: November 5, 1998 2



2 3 4 5 6 7

In view of the above, the fact the interpretation of English present perfect constructions depends to a large extent on pragmatic considerations, such as that of identifying the temporal point of reference, the aim of this article is to show that if discourse factors in general and that of cohesion in particular are taken into account in our investigation of the perfect, we will in fact find a common characteristic linking all English present perfects to each other. 2 The Perfect vs. the Perfective Aspect

9 10

However, any discussion regarding the perfect and its meaning(s) will have to begin by noting, as does Elsness, that [] there is no immediate connection between the present perfect in present-day English and the fundamental aspectual concept of perfectivity [] . (Elsness, 1997:18) That this difference is not as self-evident as we might assume, can be observed in how Quirk et al. for example, begin their discussion of aspect in English: [], the aspect constructions of English, the perfective and the progressive [], can be seen as realizing a basic contrast of aspect between the action viewed as complete (perfective), and the action viewed as incomplete, ie in progress (imperfective or progressive). (Quirk et al., 1985:188189) Here, the label perfective refers to what in this present article is called the (present) perfect. This kind of confused terminology could of course be nothing more than trivial but at the bottom of page 189 Quirk et al. (1985) contrast the simple past: John lived in Paris for ten years. with, what they call, the present perfective: John has lived in Paris for ten years., in which case this confusion cannot be said to be of a trivial nature any more. The reason is that what Quirk et al. define as present perfective is not perfective at all. In fact, it is quite the opposite, namely imperfective, because it is unbounded (cf. Bybee and Dahl 1989:55), that is, it is not possible
version date: November 5, 1998 3

11 12 13 14

15 16

17 18 19 20 21

22 23 24 25 26 27 28 29 30

1 2 3 4 5 6 7

to identify any definite point of completion. We just do not get to know whether John still lives in Paris or not. The fact that there appear to exist identifiable differences between the perfect and the perfective means that we cannot regard the perfect as taking the perfective aspect by definition. Quite on the contrary, sentences indicating the perfect of imperfective situations are very common in English, for example He has been doing this for hours.


9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

in which we have a non-anterior, realis imperfective situation which is marked for present time relevance (cf. Bybee & Dahl 1989).4 Note in this respect that all TMA values (see section 1.2) are already present in this syntagm before has is added: been (anterior, realis, perfective), doing (nonanterior, realis, imperfective). What has does is to cancel explicitly the point of termination encoded in been but leave the rest of its anteriority unchanged. That is, the situation began in the past, but its non-termination and hence its simultaneity with its reference time is explicitly asserted. So, in this article we will assume that the English present perfect marks a verbal situation5 for cancelled termination6 ; but that the perfective is an aspectual value indicating that the verbal constituency7 should be seen as, unbounded, as an indivisible unit, with no internal structure (see Nordlander, 1997a: Section 5.2). To see why an equation of the perfect and the perfective is incorrect even in principle, we must pin down what it is in have that gives present relevance/cancelled termination to the perfect as opposed to the indifference

5 6

Briefly speaking, in Nordlander (1997a) the verbal constituency is conceived of as consisting of three binary semantic distinctions: tense (anterior vs. non-anterior), mood (realis vs. irrealis) and aspect (perfective vs. imperfective), plus the verbal nucleus (the semantic core of the main verb). For more in-depth definitions of tense, mood, aspect and verbal nucleus, see Nordlander 1997a, Chapter Four (verbal nucleus) and Chapter Five (tense, mood and aspect). For a definition of verbal situation, see Nordlander (1997:3336). Comrie (1976: 5ff) labels this present relevance. According to him the present relevance may be either absoluterelated to the moment of speechor relativerelated to some other point of reference. For a discussion on the verbal constituency, see Nordlander (1997: Chapter 5). version date: November 5, 1998 4

1 2

to the point of reference which we find in the perfective. This can be illustrated as in (3): He has (just) done it.


4 5 6 7 8 9 10 11

Here, just as in (2) above, all TMA values are present in the VC (the nucleus do plus anterior tense, realis mood and perfective aspect). What is provided by the auxiliary have is in effect the link or mediator (Nordlander 1997a:92) between the two incompatible tense values anterior (the beforeness of done) and non-anterior (the simultaneity with the moment of speech). Thus we may conclude that whatever function we ascribe the the present perfect this function is not to mark verbal situations for perfective aspectuality. 3 The Stative Nature of the Perfect


13 14 15 16 17 18

In view of the above we see that what the perfect appears to do is to indicate a certain state: the state of having something (done). In fact, the conclusion that the English periphrastic have-perfect construction indicates a type of state, is also borne out by its history. According to Visser (1973: 21892192) this construction started off as a possessive have plus object complement construction. Originally have in colligation with a past participle was a notional verb denoting possession, while the past participle was a complement or attribute to the object and had a good deal of adjective force, [] (Visser, 1973:2189) So, if it is correct that the present perfect construction indicates some kind of stative or state-like situation, we then have to determine what this means to the present discussion of the present perfect as a cohesive device. The crucial characteristic of states is that they are internally infinite (Nordlander 1997a: Sections 2.7.1 and 2.8) and therefore always extend to the point when their termination is imposed (rather than reached). That is, since states do not need

19 20 21 22 23

24 25 26 27 28 29

version date: November 5, 1998 5

1 2 3 4 5 6 7

input of energy to go on, they go on until they are stopped in one way or the other. There are no points of termination encoded in the states themselves. Therefore, through the stativizing of the situation by periphrastic have, its duration is depicted as explicitly devoid of termination at one particular point in time: the chosen temporal point of reference. The internal procedural development of a perfect situation can thus be described as a first, punctual phase of entry-into-a-state, followed by a second, stative phase. 4 The Role of Cohesion

9 10

In section 1.1, discussing example (1), it is claimed that in order to understand sentences such as (1), here repeated as (4) He has lived in Singapore.



12 13 14 15 16

we have to have an implicit temporal point of reference, and put the explicit statement (4) in some kind of relation to the implicit temporal point of reference in order to make sense of it. This section will show that it is possible to define this relation in terms of what Halliday and Hasan8 term cohesion, the concept of which they claim [] is a semantic one; [and] it refers to relations of meaning that exist within the text, and that define it as a text. (Halliday and Hasan, 1976:4) Moreover, they also stress the importance of cohesion in general for an overall understanding of utterances, thus [c]ohesion occurs where the INTERPRETATION of some element in the discourse is dependent on that of another. The one PRESUPPOSES the other, in the sense that it cannot be effectivel decoded except by recourse to it. (Halliday and Hasan, 1976:4)

17 18 19

20 21

22 23 24 25 26

version date: November 5, 1998 6

1 2 3 4 5 6

However, since we have no reason to believe that cohesion is restricted only to multi-sentential texts, the assumption in this paper is that the notion of cohesion is also applicable to texts that are of a much shorter and less complex constitution, texts such as linked verbal situations containing a present perfect construction. In fact, the definition of text offered by Halliday and Hasan easily lends itself to such a conclusion saying that text refers to [] any passage, spoken or written, of whatever length, that does form a unified whole. (Halliday and Hasan, 1976:4)

7 8 9

10 11

In view of the above I will therefore use cohesion in Hallidays and Hasans sense as it is elegantly summed up by Crystal in which it refers to [] those surface-structure features of an utterance or text which link different parts of sentences or larger units of discourse, [] (Crystal, 1991: 61)

12 13 14

15 16 17 18 19 20 21 22 23 24 25

Following the above definitions of cohesion and text, we may then say that the present perfect situation is a sub-part of a small text, indicating a special kind of cohesive relation which holds between the verbal constituency, with its TMA and nucleic values (the present perfect situation), and some other point of reference. Therfore, consider the following sentence pairs, in which all eight sentences have the present moment of speech as their chosen temporal point of reference. Here we see that in order to refer to a relevant resultant state, perfect marking by means of periphrastic have is required for (6a), telic nucleus, (7a), processive nucleus, and (8a), eventive nucleus. No such marking is needed for (5a) which has a stative nucleus. (a) (b) (a) (b) He knows the truth. He has known the truth. He has made a new chair. He has been making a new chair.





version date: November 5, 1998 7


(a) (b) (a) (b)

He has run his own business. He has been running his own business. He has killed his black sheep. He has been killing his black sheep.


3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

As indicated above, the a-sentences have one important feature in common which is lacking in the b-ones: they all refer unambiguously to the direct, unterminated, existence of whatever state is indicated by these situations. That is, (5a) asserts the existence of knowledge of the truth; (6a) asserts the existence of a new chair, which is the state resulting from completing the making of a chair; (7a) asserts the existence of having run a business, which is the state resulting from terminating to run a business,9 and finally (8a) asserts the existence of (all) black sheep being dead, which is the state resulting from completing the killing of the black sheep10 . The important point is that in all of the above a-sentences the nontermination assertions concern the existence of some kind of relevance of these states, at their respective temporal points of reference, rather than about the states themselves. In effect, this means that they stand in a cohesive relation to their temporal points of reference since they link, in order to keep the construction comprehensible, the existence of relevance, of say (5a), to the temporal refence point at which this existence is supposed to be relevant. That the perfect may be a question of relevance (at some chosen point of reference) of the existence of a particular state or (process), rather than the relevance of this state (or process) itself, is also implicit in Comries discussion (1976: 5661) of the four major types of the perfect he identifies: the perfect of result, the experiential perfect, the perfect of persistent situation and the perfect of recent past.11



However, the insertion of the adverbial since in a perfect sentence like this would normally give the situation a direct, on-going relevance. There is no termination implied in He has run his own business since .... . As pointed out to me by Professor Gunnar Persson (personal communication) all is a cancellable inference here since it is possible to add phrases such as except one, etc. However, the effect of this insertion is not on a par with the extension of (7a), accounted for in footnote (9), it is still an assertion of a state; the state of having killed all sheep except one. The difference between (7a) and (8a) is that in the extended version of (7a) we get a switch from the state of to the ongoing process of , whereas we have no such switch in (8a). Schwenters (1994) account of the hot news perfect also lends itself to such an interpretation.

version date: November 5, 1998 8

1 2 3 4 5 6 7 8 9

If we investigate what Comrie says about his typology of perfects it is clear that all of the states (or processes) that he discusses, in some way or the other are relevant at their individual temporal points of reference. That is, it is important to the coherence of the text (to the discourse coherence) that the relevance of these states (or processes) are marked or made clear. Or more precisely, that the existence of this relevance is marked. These observations are also corroborated by Klein and Vater (1998) in their discussion of Andersons (1982) five different usages of the present perfect12 : Clearly, in all cases [Andersons five usages] the event itself, that is TSit, precedes the time of utterance. Moreover, it is also intuitively clear that something is said about what is presently the case, that is TT includes TSit [] (Klein & Vater, 1998: 228)13 The fact that the existence of a state in many cases coincides with the period of time during which this existence is relevant for the current discourse focus, has no bearing upon the definition of the perfect. In order to be relevant at a certain point of reference, the state in question does not necessarily have to obtain. If we examine the b-sentences we see that the VCs of (6b), (7b) and (8b) turn out not to be states but processes, and only (5b) retains its stativity. The fact that the former three VCs are processes would contradict the assumption made above that the present perfect is a stativizer and instead indicate that here the perfect is employed to mark these situations for some kind of additional imperfectivity. However, again it is not the processes themselves which are non-terminated and thus stand in a relevance relation to the moment of speech. The assertions are made, as indicated in the paragraphs above, about the non-terminated existence of the processes in (6b), (7b) and (8b). It is the existences (which are states) of these processes that stand in relevance relations to the reference points, not the processes themselves. And since

10 11 12 13 14

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

12 13

In Kleins (& Vaters) terminology TT is topic time (roughly corresponding to Reichenbachian point of reference, R); TSit is the time of the situation (Reichenbachian the event, E); and TU is the time of utterance (Reichenbachian speech time, S) (Klein & Vater, 1998; Klein, 1994).

version date: November 5, 1998 9

1 2 3 4 5 6 7 8 9 10

existence by definition is stative, this would then lend support to the above assumption that the perfect is a stativizer rather than an imperfectivizer. It must be pointed out that perfect marking of sentences such as those found in the b-examples above, in addition to its basic stativizing effect, also has a tendency to impose the need for further qualification in the shape of additional linguistic or extra-linguistic context. As for (5b), the most likely interpretation is that the state created by the perfect is not the state of knowledge of the truth, but rather the opposite, the state of non-knowledge/lost knowledge of the truth. This can easily be reversed by, for example, an adverbial phrase such as for ages, etc., giving He has known the truth for ages.



12 13 14 15 16 17 18 19 20 21 22 23 24

in which knowledge of the truth is asserted.14 Because of this need for further qualification, we might say that in a sense the meaning of has known the truth is open to interpretation or even elaboration in a way which is not possible for its non-perfect version in (5a). That is, unless the speaker provides an unambiguous framework for sentences like (5b), the hearer will have to disambiguate them in some way in order to understand the intentions of the speaker. This, that the hearer participates actively in the creation of the message (by way of supplying himself with additional information) was pointed out by Roman Jakobson. In his terms, this process was one of elliptical perception, the goal of which would be to facilitate the understanding of vague or ambiguous utterances. Or in his own words (Jakobson & Pomorska 1990b: 172): A technique which today receives even less consideration is that of elliptical perception, by which the listener fills in (again on all linguistic levels) whatever has been omitted by him as listener. We have also failed to appreciate properly the subjectivism of the hearer, who fills in the elliptic gaps creatively. Here lies the heart of the issue of disambiguation.

25 26 27 28 29 30

It is possible to negate, yet again, the assertion of knowledge of the truth in (9) by adding but now he doesnt anymore. The point is that perfect marking of imperfective VCs gives these situations a certain amount of open-endedness.

version date: November 5, 1998 10

1 2 3 4 5 6 7 8 9 10 11 12

It appears, however, that this active participation in the creation of the meaning is not required for the understanding of (5a). In the case of (6b), the assertion is made about the existence of the process of making a new chair, rather than of the process of making a new chair. What the perfect does is to assert that this existence of the process is relevant at the chosen point of reference. We might say that this on-going (that is, non-terminated) relevance is co-extensive with the moment of speech in both extensions of (6b) presented below, (10a) and (10b), with the difference that in example (10a) our understanding is that the process itself is also depicted as co-extensive with the moment of speech, whereas in example (10b) the interpretation is that the process did terminate at some point in time, prior to the present moment of speech. (a) (b) He has been making a new chair, but it is not finished yet. He has been making a new chair, but he is not at it any more.



14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

It is worth noting that in (10a) the co-ordinated but-clause is focused on the resultant state, or rather the absence of such a state, whereas in (10b) the focus of the co-ordinated clause is on the (absence of the) process itself. That is, in (10a) it is the existence of a new chair that is negated, whereas in (10b) the negation is of the chair-making process. In neither case, however, is the relevance of the existence of a new chair or of the process of making a new chair negated for the current line of discourse. When it comes to (7) and (8) the same relation described above as holding between (5a) and (5b), and (6a) and (6b), respectively, can be said to hold between the (a) and (b) variants of these examples. The difference between (8a) and (8b) is that in the former the assertion is about the existence of a state: (all) his black sheep are killed15 , but in the latter the assertion is about the existence of a process: the killing of his sheep16 . Note that for (8b) we cannot tell whether the process itself (killing sheep) is terminated or not.

15 16

In line with the reasoning above in the discussion of examples (5) and (6), the assertion is about the on-going relevance of the existence of a state such as (all) his sheep are killed/are dead. Or rather, the assertion is about the on-going relevance of the existence of a process such that sheep are being killed.

version date: November 5, 1998 11

1 2 3 4 5 6 7 8 9 10

The only thing we can infer is that the existence of such a process is relevant to the continuing line of discourse. Thus (11) below shows that perfect marking of imperfective VCs with eventive nuclei is neutral, in the absence of further qualification, to on-going relevance at the present moment of speech in Comries (1976) sense. In (11a) no process of killing sheep is going on any more, but the very existence of such a process is felt to be relevant anyway. As for (11b) this process is still in progress. This again points to the conclusion that it is not the propositional content itself of the VC marked for the present perfect which is affected, but the existence of such a proposition. (a) (b) He has been killing his black sheep and now theyre all dead. He has been killing his black sheep and hes still at it.



12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

The same state/process distinction manifested in the difference between (8a) and (8b) is also found between (7a) and (7b). In (7a) the existence of the state resulting from having run a business is at stake, but in (7b) the existence of the process of business-running itself is indicated as being relevant. As can be inferred from the discussions on the b-examples, and their possible expansions, the common denominator is their greater degree of indeterminacy; in order to understand the intentions of the speaker properly, the hearer needs some additional clarifying means to avoid having to resort to mere guess-work. This clarifying means is provided by the present perfect construction in the form of a cohesive link between the proposition expressed in the present perfect clause and one particular contextual implicit or explicit feature which will help the hearer to understand the sentence in a coherent way, namely the temporal point of reference. That is, the present perfect urges the hearer to look for something, no matter what, which will explain why the existence of this proposition is relevant at the chosen temporal point of reference. Thus we may define the semantico-grammatical category of the present perfect as: A cohesive relation holding between the relevance of the existence of the result of the event or the process, referred to by a VC, and a chosen temporal reference point, in the sense that this relevance is
version date: November 5, 1998 12

30 31 32

1 2

explicitly said not to have expired at the chosen temporal reference point. Therefore the only conclusion to be drawn here is that the present perfect is a discourse marker which operates upon the entire verbal constituency in order to focus or highlight the existence of some propositional content indicated by this VC and in turn to put this focused existence in relation to the chosen point of reference. We are being told how the clauses of the sentence (be they explicit or only implicit) are supposed to be connected and related to each other an intra-sentential kind of cohesion in Hallidays ( 1976 & 1985) terms. However, there are other reasons, purely contingent on the definitions of tense and aspect given in Nordlander (1997a: Chapter 5), for the analysis of the perfect as a discourse marker of intra-sentential cohesion. Since the perfect indicates a situation which can be said to stand in a relevance relation to another situation/point in time, it follows from this that it (the perfect) relates two situations temporarily to each other. And as I argue (Nordlander 1997a:110) this is precisely what aspect does not do but rather is the task of tense. Thus the perfect cannot be an aspect, even though it indicates the presence of an internal procedural structure. Turning the argument around, it might be tempting to analyse the perfect as a tense since, as mentioned above, it relates verbal situations to each other and thus cannot be an aspect. But this analysis will fail as well because tense has nothing to do with the internal temporal constituency of a verbal situation, whereas this is precisely part of the meaning of the perfect. The only possible conclusion must be that the perfect is neither tense nor aspect although it exhibits some, but not all, characteristic features of both tense (it relates situations to each other) and aspect (characterization of the internal procedural development), respectively. In the framework presented and outlined in Nordlander (1997a) there is thus no room for the perfect within the verbal constituency; it must constitute part of the situational periphery, being a modifier on a par with other peripheral modifiers. It is, however, given a place within the range of aspectual peripheral modifiers since it is focused on one single phase of the situational procedural development, namely the final, terminating one.

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

version date: November 5, 1998 13


2 3 4 5 6 7 8

Therefore, summing up the above we may say that the English present perfect is a discourse marker of intra-sentential cohesion because it helps the hearer/reader keep track of a more general line of reasoning than that which is conveyed and exposed by purely linguistic means. That is, the occurrence of an English present perfect construction urges the hearer/reader to look for some kind of implicit connection between two different verbal situations, one of which might not be explicitly expressed in the sentence in question.

version date: November 5, 1998 14

References Anderson, L. 1982. The Perfect as a Universal and as a Language-Particular Category. In Hoper, P.J. (ed) 1990. pp.227264. Bybee, J. & . Dahl 1989. The Creation of Tense and Asspect Systems in the Languages of the World. In Studies in Language. Volume 13. No 1, pp. 51103. Comrie, B. 1976. Aspect. Introduction to the Study of Verbal Aspect nad Related Problems. Cambridge: Cambridge University Press. Crystal, D. 1987. The Cambridge Encyclopedia of Language. Cambridge: Cambridge University Press. . 1991. A Dictionary of Linguistics and Phonetics. Oxford: Blackwell. Elsness, J. 1997. The Perfect and the Preterite in Contemporary and Earlier English. Berlin: Mouton de Gruyter. Grice, P. 1989. Studies in the Way of Words. Cambridge, Massachusetts: Harvard University Press. Halliday, M. A. K. 1985. An Introduction to Functional Grammar.17 London: Edward Arnold. Halliday, M .A. K. & R. Hasan 1976. Cohesion in English. London: Longman. Hoper, P.J. (ed) 1982. Tense-Aspect: Between Semantics & Pragmatics. Amsterdam/Philadelphia: John Benjamins. Jakobson, R.O. & K. Pomorska 1990. The Time Factor in Language. In Waugh, L.R. & M. Monville-Burston (eds) 1990. Klein, W. 1994. Time in Language. London and New York: Routledge. Klein, W. & H. Vater 1998. The Perfect in English and German. In Kulikov, L. & H. Vater (eds) 1998. Kulikov, L. & H. Vater (eds) 1998. Typology of Verbal Categories. Papers presented to Vladimir Nedjalkov on the occasion of his 70th birthday. Tbingen: Max Niemeyer Verlag. Moens, M. 1987. Tense, Aspect and Temporal Reference.18 Thesis. University of Edinburgh. Nordlander, J. 1997a. Towards a Semantics of Linguistic Time. Uppsala: Swedish Science Press. . 1997b. On Verbal Dynamicity, Markedness and Temporal Location. In Rydn, M., Kardela, H., Nordlander, J. & B. Odenstedt (eds) 1997. Quirk, R., Greenbaum, S., Leech, G.& Svartvik, J. 1985. A Comprehensive Grammar of the English Language, London: Longman. Rydn, M., Kardela, H., Nordlander, J. & B. Odenstedt (eds) 1997. From Runes to Romance. Uppsala: Swedish Science Press. Schiffrin, D. 1994. Approaches to Discourse.19 Oxford: Blackwell. Schwenter, S. 1994. Hot News and the Grammaticalization of Perfects. Visser, F.Th. 1973. An Historical Syntax of the English Language.Leiden: E.J.Brill. Waugh, L.R. & M. Monville-Burston (eds) 1990. On Language. Roman Jakobson. Cambridge, Massachusetts: Harvard University Press.

17 18 19

printout date: November 5, 1998 15