Hedging: A Challange for Pragmatics and Discourse Analysis

RAIJA MARKKANEN and HARTMUT SCHRÖDER

 

"If you hedge against something unpleasant or unwanted that might affect you, you do something which will protect you from it. If you hedge or hedge a problem or question you avoid answering the question or committing yourself to a particular action or decision." (Collins 1987) This description of the everyday meaning of the verb hedge suggests that as a linguistic term it might also refer to the choice of a certain kind of communicative strategy. This is actually the case in most uses of the term hedge in linguistic literature: it is possible to detect at least some affinity between its everyday meaning and the linguistic expressions referred to when using the term. However, there have been linguists who, like Skelton (1988), deplore the use of the term precisely because it reflects the connotations of the everyday usage of the word.

Dictionaries of linguistics do not normally mention either the concept or the term 'hedge'. Exceptional in this respect are the Dictionary of Stylistics and two German dictionaries of linguistics. Both German dictionaries use the term 'Heckenausdruck' and list it under the headings of semantics and lexicography. Their definitions for the term are the following:

"Adjektivische oder adverbiale Wendung, durch die angegeben werden kann, in welchem Maße in einer gegebenen Sprache und dem dazugehörenden Kulturraum kategorisierende Aussagen im Sinne einer Repräsentativitätsskala möglich bzw. sinnvoll sind." (Bußmann 1990) and "Bezeichnung für Ausdrücke, die andeuten, in welchem Sinne bestimmte Exemplare einer bestimmten Kategorie zugeordnet werden. Aus der Tatsache, daß (jeweils relativ zu einem spezifischen kulturellen Hintergrund) manche Exemplare als bessere/typischere Beispiele einer Kategorie angesehen werden (...) ergibt sich ein Bedürfnis für solche Hecken." (Gippert 1993)

These definitions indicate that the formulation of concepts in everyday communication requires the use of hedges because concepts (e.g. 'bird') trigger prototypical images in people's minds, which makes it necessary to somehow mark their less prototypical representatives. Thus, if we mark a concept with a hedge, we do not refer to a prototypical representative of the class, but to a non-prototypical one. Therefore a sentence like A penguin is sort of a bird is acceptable, but A raven is sort of a bird is absurd.

Compared with the two German dictionaries, the Dictionary of Stylistics gives a somewhat different view. This dictionary, too, refers to the semantic origin of the concepts 'hedge' and 'hedging', but it classifies them as belonging to the fields of discourse analysis and speech act theory and defines them as "qualification and toning-down of utterances and statements (...) in order to reduce the riskiness of what one says". The motivation for their use is given as "mitigation of what may otherwise seem too forceful" and "politeness or respect to strangers and superiors".

The use of hedge as a linguistic term goes back at least to the early 1970s, when G. Lakoff (1972) published his article Hedges: A Study in Meaning Criteria and the Logic of Fuzzy Concepts. Lakoff was not interested in the communicative value of the use of hedges but was concerned with the logical properties of words and phrases like rather, largely, in a manner of speaking, very, in their ability "to make things fuzzier or less fuzzy" (Lakoff 1972, 195). Interestingly, however, Lakoff also briefly points out the possibility that hedges may "interact with felicity conditions for utterances and with rules of conversation" (Lakoff 1972, 213). In accordance with Lakoff's main concern, however, the term hedge has later been defined, for example by Brown/Levinson (1987, 145) as "a particle, word or phrase that modifies the degree of membership of a predicate or a noun phrase in a set; it says of that membership that it is partial or true only in certain respects, or that it is more true and complete than perhaps might be expected". This definition is interesting in that it includes in hedges both detensifiers and intensifiers, which was how Lakoff also saw hedges. However, many users of the term limit it only to expressions which show that "the match between a piece of knowledge and a category is less than perfect" (Chafe 1986, 270).

Since the early 1970s the concept of hedge has moved far from its origins, particularly since it has been adopted by pragmatists and discourse analysts. The term is no longer used only for expressions that modify the category membership of a predicate or noun phrase. As a matter of fact, Lakoff himself had already mentioned in his 1972 article Robin Lakoff's observation that certain verbs and syntactic constructions convey hedged performatives (I suppose/guess/think that Harry is coming; Won't you open the door?). The idea of hedged performatives became then one way of widening the concept of hedges. Fraser (1975) considered the effect that modals and semi-modals have on the illocutionary act denoted by a performative verb in performative sentences like I must advise you to remain quiet. Thus, for example, the modal must gets the speaker off the hook, relieves him/her from some of the responsibility. Fraser calls these cases 'hedged performatives', without naming the modals 'hedges'. In a later article (Fraser 1980), in which he discusses the mitigation of the harshness or hostility of the force of one's actions, Fraser seems to have adopted Lakoff's view of hedges and, accordingly, limits them to expressions like kind of, sort of. Other researchers, like House/Kasper (1981) and Blum-Kulka/Ohlstein (1984), have also discussed hedges as a means of modifying certain types of speech-acts, notably requests and apologies.

In addition to the idea of hedged performatives, the concept was also widened in another way when hedges were taken to be modifiers of the speaker's commitment to the truth-value of a whole proposition, not just the category membership of a part of it. Thus, for example, Vande Kopple (1985) in his categorization of metadiscourse types considers the use of hedges as showing a lack of full commitment to the propositional content of an utterance. In other words, hedges (e.g. perhaps, seem, might, to a certain extent) are by him seen as modifying the truth-value of the whole proposition, not as making individual elements inside it more imprecise.

This widening of the concept of hedge to contain the modification of commitment to the truth of propositions has led some researchers to think it necessary to distinguish between two types of hedges. This has been done for example by Prince/Frader/Bosk (1982) in their discussion of hedging in physician-physician discourse. They start from Lakoff's definition of hedges as devices that make things fuzzy, but add that there are at least two kinds of fuzziness. One is fuzziness within the propositional content, the other fuzziness "in the relationship between the propositional content and the speaker, that is the speaker's commitment to the truth of the proposition conveyed" (Prince/Frader/Bosk 1982, 85). Accordingly, there are two types of hedges: those that affect the truth-conditions of propositions, which Prince/Frad-er/Bosk call approximators (e.g. His feet were sort of blue), and shields, which do not affect the truth-conditions but reflect the degree of the speaker's commitment to the truth-value of the whole proposition (e.g. I think his feet were blue). A similar distinction is drawn by Hübler (1983), who distinguishes between what he calls understatements and hedges, although both are devices used for expressing 'indetermination'. For example, a sentence like It's a bit cold in here is indeterminate. However, according to Hübler, there are two kinds of indetermination: phrastic and neustic. Phrastic indetermination concerns the propositional content of a sentence, whereas the neustic type is connected with the claim to validity of the proposition a speaker makes. This distinction Hübler then carries over to distinguish between understatements, i.e. expressions of phrastic determination, and hedges, i.e. expressions of neustic indetermination. Thus, a sentence like It is a bit cold in here contains an understatement, while It is cold in Alaska, I suppose contains a hedge. Hübler's division thus greatly resembles that by Prince/Frader/Bosk, whose approximators correspond to Hübler's understatements and shields to his hedges. In both cases, it can be asked what end this division serves, what is gained by it besides making it easier for someone carrying out concrete analysis to limit his/her data. Hübler himself admits that both understatements and hedges perform the same function of expressing indetermination, of making sentences more acceptable to the hearer and thus increasing their chances of ratification. It is thus easy to agree with Skelton's (1988, 38) criticism of Prince/Frader/Bosk's division: it "seems to be sustainable only in the abstract: it looks more like a description of a property of text sentences than of language use".

Furthest away from the original concept of hedge are those approaches in which hedges are treated as realizations of an interactional/communicative strategy called hedging. Thus, Markkanen/Schröder (1989; 1992), who discuss the role of hedges in scientific texts, see them as modifiers of the writer's responsibility for the truth value of the propositions expressed or as modifiers of the weightiness of the information given, or the attitude of the writer to the information. According to them, hedges can even be used to hide the writer's attitude. Markkanen/Schröder also suggest that hedges offer a possibility for textual manipulation in the sense that the reader is left in the dark as to who is reponsible for the truth value of what is being expressed (Markkanen/Schröder 1992). When this kind of purely functional starting point is adopted, there is no limit to the linguistic expressions that can be considered as hedges. Thus, Markkanen/Schröder consider as important devices for hedging - at least in scientific writing - the use of certain pronouns and avoidance of others, the use of impersonal expressions, the passive and other agentless constructions, in addition to the use of modal verbs, adverbs and particles, which are usually included in hedges. They also suggest that certain rhetorical and stylistic devices could be included. The difficulty with these functional definitions is that almost any linguistic item or expression can be interpreted as a hedge. Thus, for example, Darian (1995) claims that hedges can belong to any part of speech and lists as examples nouns, verbs, adjectives, adverbs, and even articles ('one solution is...'). In this context it should be emphasized that no linguistic items are inherently hedgy but can acquire this quality depending on the communicative context or the co-text. This also means that no clear-cut lists of hedging expressions are possible. However, the advantage of functional definitions is that they make it possible to draw attention to an important aspect of communicative behaviour.

As there are so many different ways that the concept hedge has been defined, some narrower than others, it is natural that there are other terms and concepts that come close to it, sometimes covering partly the same area of language use. The most important concept that cuts across the area of hedges is that of modality. Like definitions of hedge, those of modality also vary a lot in scope. According to Palmer (1986, 2), the notion of modality is vague and leaves open a number of possible definitions. A lot of the discussion on modality has concentrated on the modal auxiliaries (e.g. Palmer 1986) but modality has also been treated as "a central organizing principle in language" (Stubbs 1986, 4).

If we consider Lyons' (1977, 797) definition of epistemic modality, "Any utterance in which the speaker explicitly qualifies his commitment to the truth of the proposition expressed by the sentence he utters, whether the qualification is made explicit in the verbal component (...) in the prosodic or paralinguistic component, is an epistemically modal or modalized utterance", we can see its affinity to many of the definitions of hedge, for example to Hübler's view of hedges as expressions of neustic indetermination, or to Prince/Frader/Bosk's shields. But it is not possible to include in Lyons' conception of epistemic modality hedges as defined by Lakoff and others who see them as modying parts of the proposition. However, even these hedges can be included within the realm of epistemic modality if we accept Stubbs' (1986, 5) view that "it is possible to indicate degrees of commitment to just three kinds of linguistic item: not only (1) to propositions but also (2) to illocutionary forces and (3) to individual lexical items". The last mentioned type, lexical commitment, is realized through items like so called, so to speak, quote unquote, which makes it possible to include also the Lakoffian type of hedges in modal expressions. Stubbs' view of modality allows even the inclusion of linguistic items and structures like logical and pragmatic connectors, past tense when used hypothetically, and passivization, which allows the deletion of the agent and therefore the avoidance of commitment. This view of modality comes close to the functional, pragmatic definition of hedges referred to above (e.g. Markkanen/Schröder 1989; 1992).

The concepts of modality and hedge thus overlap to a lesser or greater extent depending on their respective definitions. This connection is very clear in the case of modal verbs with epistemic meanings. When hedges are taken to be modifications of the commitment to the truth-value of propositions, for example the English modal auxiliary may is always listed as a typical example. Thus, in It may be true it is a hedge but also an expression of epistemic modality. Sometimes also the deontic meanings of modals allow interpretation as hedges. For example, in English the hypothetical would could be seen as a hedge because it makes an utterance non-categorical. Preisler (1986, 92) actually points out that "even when modal forms convey speaker-external meanings, these are often given interpersonal significance by the particular context in which they appear, usually as part of a tentativeness strategy". It seems possible to see the relationship between modality - mostly of the epistemic type - and hedges in two ways: either modality is the wider concept and includes hedges or the other way round, hedging is the umbrella term and epistemic modality a part of it.

Another concept that cuts across the area of hedges - and epistemic modality - is evidentiality, again depending on how broadly hedge is understood. Chafe (1986, 271) defines evidentiality as "any linguistic expression of attitudes toward knowledge", i.e. assessment of its reliability. Knowledge, according to Chafe, has various modes: belief, induction, hearsay, and deduction, each of which is based on a different source. Most of the expressions that Chafe gives as examples of the realization of these different modes are expressions that have also been included in hedges by other linguists. Chafe himself uses the term hedge only of expressions that denote that "the match between a piece of knowledge and a category may be less than perfect" (Chafe 1986, 271), thus agreeing with Lakoff's original idea of hedges.

Vagueness is another concept close to hedging as it refers, among other things, to the use of expressions like about, sort of, i.e. expressions that denote the impreciseness of quantity, quality, or identity, which is very much like Lakoff's "fuzziness" (cf. Channell 1990; 1994; Zuck/Zuck 1985). Similarly, von Hahn (1983, 99) uses the term 'Vagheit' (vagueness) when referring to the false assumption that the language in scientific texts is exact.

Linguists who, like Chafe, use the term hedge in a more limited way need other concepts/terms to cover the rest of the phenomena included in hedges in the more extensive definitions. A further example of this is House/Kasper's (1981) use of the term 'hedge' only of adverbials with which the speaker avoids precise propositional specification (e.g. kind of, sort of, somehow, rather). They then distinguish several other types of what they call modality markers, like understaters, which are used to underrepresent the state of affairs denoted by the proposition (e.g. a little bit, not very much) and downtoners (e.g. possibly, perhaps), which are used to modulate the impact an utterance is likely to have.

As to the motivation for the use of hedges, a lot of the discussion has concentrated on their use in spoken discourse, and the most frequently mentioned motivating factor is politeness, as defined by Brown/Levinson (1987). In their view, hedges are mainly used for negative politeness in face-saving, in which they are put to elaborate use. In positive politeness they figure only in expressions of extremes, like marvellous and appalling, which are typical of this form of politeness, 'safely vague' because they leave it to the addressee to figure out how to interpret them. In negative politeness strategies hedges are used, according to Brown/Levinson, to hedge on the illocutionary force of an utterance or on any of the four Gricean Maxims. In both cases the motivation for their use is the desire to save face, either the hearer's or the speaker's. Although Brown/Levinson (1987, 67) say that it is possible to distinguish between acts that primarily threaten the hearer's face and those that threaten the speaker's own face, they admit that the latter acts are also potential threats to the hearer. Thus, in their discussion of politeness and the ways to express it, it is the hearer's face-wants that get emphasized.

It is, however, possible to turn up the other side of the coin and emphasize the importance of hedges for the speaker's own face. Their use may be motivated, for example, by the fear of being proved wrong later on. Being imprecise or mitigating one's commitment to the truth-value of a proposition or a claim makes it possible to say, if proved wrong, that the claim was only tentative or an approximation. This explanation is supported by Hübler's (1983) view that the reason for using hedges is to make sentences more acceptable to the hearer and thus increase their chances of ratification, which in turn is made necessary by the inherent negatability of sentences. According to Hübler, the function of hedges is to reduce the risk of negation. Thus, it can be claimed that, in all communication, while showing deference to the addressee, the speaker or writer also tries to protect him/herself from potential anger, contempt or other humiliation on the part of the addressee. Both desires are present in all communicative situations, but one may be stronger than the other in a particular case. House/Kasper (1981, 157) seem to share this view when they say that "both these functions - one defensive and ego-oriented, the other protective or alter-oriented are fulfilled by politeness". Thus, in some situations the desire to protect oneself from the potential denial of one's claims may be greater than the desire to show deference to the addressee. The surer a speaker feels about his/her own position vis-a-vis the interlocutor, the less need there is for hedging for the purposes of self-protection.

Related to the above issue are the claims made by some linguists that gender has an influence on the use of hedges, meaning that women hedge more than men. Thus, R. Lakoff (1975) lists hedges as one of the features typical of women's speech, as one way of sounding feminine and thus reflecting their position in society. Preisler's (1986) study of linguistic sex roles also shows that linguistic tentativeness signals are correlates of both sex and interactional role, which is either task-oriented or socio-emotional. It is the latter role that correlates with tentativeness. As can be expected, Preisler's data show that women lead in the use of hedges in general, whereas men lead in linguistic assertiveness. Preisler therefore concludes that sex is a more general determinant of speech than the interactional role, which means that "men and women have developed sex-specific speech patterns" (Preisler 1986, 288). However, there is no general agreement on this; other linguists (e.g. O'Barr/Atkins 1980) have suggested that, rather than being a feature of women's speech only, the use of hedges is a marker of 'powerless language' in general, i.e. of the way language is used by people who are in an inferior position in society. Holmes' (1990) findings in her study of gender differences in the use of hedges and boosters disagree even more strongly with Lakoff's and Preisler's claims: Holme's data show that hedges are actually frequently used by women "as positive politeness devices signalling solidarity with the addressee, rather than as devices for expressing uncertainty" (Holmes 1990, 202). Also, in a study on the use of metadiscourse by American and Finnish university students in argumentative writing Crismore/Markkanen/Steffensen (1993) found that Finns used more hedges than Americans, i.e. a cultural difference, and that Finnish men used them more frequently than women. The issue is thus more complex than has been assumed.

There has also been a growing interest lately in hedging and the motivation for its use in academic/scientific writing. That hedges are actually used in scientific discourse, which is supposed to be above all rational and neutral, is connected with the fact that scientific discourse obeys the same mechanisms as ordinary everyday communication does, although it tries to hide this, more or less successfully, by using a code of its own. We assume here that science is not only content; that is, scientific texts are not only content-oriented and informative but also aim at convincing and influencing their audience. According to classical rhetoric, formulation of a scientific text is not merely built on pragma and on docere (instructing, informing) but also on delectare (entertaining) and movere (moving, enchanting). Thus, in addition to 'going into the subject matter' (pragma), a text should also emphasize the reliability of the author (ethos) and also move the reader emotionally (pathos). The last two, ethos and pathos, are closely connected with the expressions used in the text, including hedging. The rhetorical style of a scientific text is not merely a decorative addition to an otherwise informative text; rather form and content are inseparable, as Stolze (1992, 232) suggests in the following quotation:

Hermeneutisch gesehen sind die rhetorisch-stilistischen Mittel, von denen eine leserbeeinflussende Wirkung ausgeht, nur ein Teil der Form, die mit dem lediglich analytisch abzutrennenden Inhalt eine unlösbare Einheit bildet. Texte lassen sich zwar didaktisch in diese beiden Elemente aufteilen, aber weder der Inhalt noch die pure Form hätten für sich eine Substanz, welche jene Wirkung hervorbringen könnte.

It should also be emphasized here that the stylistic devices used, including hedges, acquire their meanings through the author-reader interaction, on the basis of the text and the commucation situation.

We can thus come to the conclusion that a hedge is 'a textual phenomenon' and 'a virtual quality' of a text. Hedges are of course present at various levels of textualization but are not inherent characteristics of texts. As noted above, a text does not contain hedges per se but gets them through the author-reader interaction. This apparent subjectivity is, however, to some extent controlled by culture, since people who belong to a particular language community normally share socially determined aesthetic ideals through their shared educational background (cf. Spillner 1974, 67). But in an intercultural communication situation, cultural differences may cause misunderstanding and communication breakdown, because the author and the reader do not share the same norms and expectations.

As in other types of communication, in scientific writing politeness has been seen as the motivating factor for hedging. Myers (1989) claims that even in this type of discourse hedging is used for the sake of negative politeness, more specificially, to mark a claim "as being provisional, pending acceptance in the literature, acceptance by the community" (Myers 1989, 13). He even suggests that an unhedged claim, or what looks like a claim, is "probably not a statement of new knowledge" (Myers 1989, 13). Furthermore, according to Myers, hedges reflect a relation between the writer and the readers, rather than the degree of the probability of the statement. Although Myers (1989, 4) suggests that everyone, regardless of their position, must appear as humble servants of the discipline, it can still be assumed that the amount of hedging writers employ depends on such factors as their position within the scientific community, the potential readership, and even the writers' personalities, i.e. how sure or unsure they feel about their own position within the field (cf. Markkanen/Schröder 1989; 1992).

Furthermore, it could be assumed that in academic writing the use of hedges varies according to the field the writer represents, i.e. that there are scientific fields in which hedging is more frequent than in others. It could be expected that the texts of fields like linguistics and philosophy, for example, would contain more hedging than the texts of natural sciences and technology because of the different bases of argumentation in these fields. Argumentation in philosophy is not based on bringing in experimental data and concrete evidence, as in natural sciences and technology. As Spillner (1983, 35) points out, in texts in which the use of experimental data and logical deduction are not so important, the style of writing becomes an essential element in achieving credibility. The convincingness of an argument in such texts depends on the use of linguistic devices, including hedges (Markkanen/Schröder 1989).

However, the use of hedges is not alien to writers in science and technology either. Dubois (1987, 15) discusses the myth of perfect objectivity in science noting that "it has been shown that the scientist can unobtrusively inject his personal views into his communication" by using linguistic items that express uncertainty or impreciseness. Similarly, Butler (1990) shows the importance of modals in English biological and physical texts, in which they "serve the weighing of evidence and the careful drawing of conclusions from data, in the making of claims from evidence, and more particularly in making generalizations about what is possible in the behaviour of the universe, on the basis of observation of what actually happens" (Butler 1990, 139). In his article dealing with hypotheses in introductory science texts, Darian (1995, 101) also comes to the conclusion that "hedges are probably the clearest indicators of hypotheses" and gives an extreme example from a text in geology: (Although the steps in the creation of oil are still very poorly known,) the following (1) simplified theory is (2) rather (3) widely held and is supported by enough facts to be (4) at least (5) somewhat (6) near the truth. (Darian 1995, 102)

These findings suggest that the differences in the use of hedges between texts in different fields are not so great as has been often assumed. It could even be claimed that hedging in academic fields, regardless of the scientific paradigm, actually serves the struggle for objectivity, which the scientific community often considers a necessary characteristic of scientific writing. Prince/Fraser/Bosk (1982), who have analysed spoken discourse among physicians, suggest that their frequent use of hedges demonstrates a scholarly orderliness in the representation of knowledge.

It was suggested above that the amount of hedging may depend on the individual language user's position vis-a-vis the audience and even on his/her personality. There is thus room for individual style in the use of hedges even in academic writing. A skillful writer may also use them for his/her own specific purposes. In his analysis of F.R. Leavis' literary-critical text "The Great Tradition", Simpson (1990) shows that, contrary to expectations, Leavis modalizes (i.e. hedges in our terminology) information that is not "risky" or controversial and leaves controversial information unmodalized. The effect of this is, as Simpson (1990, 91) points out, that "important questions concerning the canon of English literature can be glibly passed over and, at the same time, attention can be deflected toward issues that are, by contrast, less significant and more peripheral". The pattern of hedging (or modalization in Simpson's terminology) used by an individual writer becomes really significant when we think of the effect that it may have on readers, of how it directs the readers to evaluate the information. Thus, in the case of an influential writer like Leavis, it might even lead students "to challenge all sorts of commonly held assumptions about the canon of English literature" (Simpson 1990, 92). This also supports the view proposed above that hedging is not an inherent characteristic of a text but a product of writer-reader communication, i.e. the linguistic expressions used in hedging get their meaning through the response they produce in the readers.

An interesting new challenge for research on hedging is provided by the new postmodern concept of science and its influence on discourse and, along with it, on hedging as a device of textualisation. Geist (1992) criticizes the research on language used in academic writing for its still continuing use of key words like economy, precision and explicitness, although words like complexity, uncertainty and ambiguity are currently in use elsewhere. Geist sees the situation as paradoxical: a lot of the research on scientific writing is still firmly based on the idea of specialization, which was problematized in scientific thinking already in the 1980s (Geist 1992, 234). According to Geist, the problem is that our understanding of reality and with it our concept of what constitutes a scientific approach have changed profoundly and found a completely new orientation. These new tendencies have also had an effect on scientific thinking, which is now characterized, among other things, by the problematization of the commonly accepted.

The 'new discourse' that arises from the above tendencies seems to be in contrast with what has until now been considered as self-evident, particularly in the Anglo-American text conventions. In his analysis of the 'new discourse' in the humanities, Laermann (1991) comes to the conclusion that things like problem, argument and evidence are not valid anymore:

Bis über die Schmerzgrenzen der Sinnlosigkeit entfernen sie sich von zwei elementaren Voraussetzungen traditioneller Wissenschaft: von Themen und von Argumenten. Über lange Passagen hinweg verzichten sie zum einen darauf, sich nur einem Thema zu widmen. Statt dessen wenden sie sich eher Problemkomplexen zu, die kaum eindeutig zu umreißen sind. (...) Wenn der herrschende Betrieb (oft fragwürdig genug) Prämien für Interdisziplinarität verteilt, dann sind die neuen Unklaren darüber jeweils schon hinaus; denn sie bewegen sich in den Sphären einer Transdisziplinarität, die keinerlei Grenzen anerkennt. (...) Die innere Organisation undeutlicher Texte ist zum anderen gekennzeichnet durch einen weitgehenden Verzicht auf Argumentationen.

In the same vein Lepenies (1993, 18) writes of a fusion and mixture of genres. It would be interesting to find out whether the use of hedges and the attitude to hedging have changed as a result of these developments in postmodern science. It could even be assumed that hedging might play an essential role in the new discourse.

As the above discussion suggests, hedging is an important interactional strategy both in spoken and written communication. Thus, to be effective communicators we should acquire a feel for its appropriate use in different communication situations. It is therefore no wonder that - in addition to the analysis of hedges and hedging for their own sake - there have been studies of a more applied kind. These studies aim at finding ways of making communication more effective through providing communicators in different areas of life with both knowledge about hedges and the opportunity to develop their skill in using them. This practical aim is evident in studies of hedging at least in the following areas:

(i) contrastive/cross-language analysis and comparative stylistics;

(ii) translation studies: for example, studies concerned with target-language oriented translation, studies on the equivalence of effect between source and target texts, problems connected with text types in different cultures, and 'untranslatable' linguistic and cultural phenomena;

(iii) studies on teaching writing in the mother tongue and the readability of texts;

(iv) studies on foreign language teaching, particularly on teaching writing in a foreign language, which also includes contrastive rhetoric, i.e. the comparison of culture-bound text conventions;

(v) intercultural communication studies, e.g. studies on the reception of texts written in a foreign language.

As the above list indicates, studies on the use of hedges are particularly relevant for language teaching, both in the mother tongue and foreign languages. The difficulties experienced in the appropriate use of hedges may partly be due to the lack of teaching of this area of language use. Students are not taught to modulate their propositions, not even sensitized to the issue. The use of hedges in writing may even be discouraged, perhaps because many of the words and phrases used as hedges are seen as empty fillers. Judging by the guidebooks for good writing, these items may be commented on in passing but not systematically taught. Teaching the appropriate use of hedges, like other pragmatic phenomena can be very problematic for several reasons. One reason is that, as suggested above, hedges get their meaning through the contexts in which they occur. Another reason is that their use is often connected with the speakers/writers' values and beliefs, even their personalities, which makes teaching them a delicate matter. Therefore, some kind of awareness-raising may be the only possible method of teaching in this area of language use.

As using hedging expressions appropriately may cause problems in some communicative situations - like academic writing - even in the mother tongue, it is no wonder that it is problematic in a foreign language. This is because the rules of appropriateness vary across cultures. Consequently, to become effective communicators in a foreign language, learners have to acquire these rules, particularly since transfer from the mother tongue is possible also in this area. The skillful use of hedges, which requires subtlety and sophistication even in the mother tongue (Skelton 1988), is clearly part of a language user's pragmatic competence, lack of which may lead in foreign language use to mistakes that are more serious than, for example, grammatical errors. This is because pragmatic errors are not so 'obviously erroneous' as faulty syntax; they only make the foreign language user sound, in the case of hedging, more impolite or agressive, more tentative or assertive than he/she intends to be, which then may even lead to a communicative failure. (cf. Thomas 1983)

There are some interlanguage studies that suggest the kind of problems foreign language learners might have in the use of hedges. In a study on the interlanguage of German learners of English, Kasper (1979) suggests that a kind of modality reduction takes place in the foreign language learners' speech. This means that modality is present at an early stage of the learner's speech act planning but does not for some reason occur in its surface realization. This reduction, she claims, is "a consequence of low awareness of modality as a pragmatic category" (Kasper 1979, 274). Other researchers have found another, more concrete, type of reduction in foreign language learners' expression of modality: their repertoire of expressions is more limited when compared with that of the native speakers of the language. For example, Kärkkäinen (1990) found that Finnish learners of English use fewer expressions of epistemic modality than do native speakers of English. They have less variation in the expressions and stick to a few favourite ones like I think. Foreign language learners also seem to prefer the more explicit kind of modification to the implicit ones. According to Färch and Kasper (1989), this preference is due to the fact that the explicit expressions (e.g. I think ) are longer and have their own propositional content and illocution.

The above mentioned interlanguage studies are all concerned with spoken discourse, but of course problems with hedges occur also when writing in a foreign language, and they seem to be similar. Thus, according to the findings of Ventola/Mauranen (1990), Finns writing in English also showed the tendency to stick to a few 'safe' expressions of epistemic modality, had less variation in the expressions than did native speakers of English, i.e. they did not behave in a native-like manner in this respect. Clyne's (1991) study of German scholars writing in English confirms the assumption mentioned above that there may be transfer from the mother tongue and one's native culture in the use of hedges. According to Clyne, who has looked at the amount of hedging in academic texts, German writers hedge more both in their native language and in English than do native speakers of English. They even use double and triple hedging. In other words, they follow the norms of their native culture when writing in a foreign language. To sound native-like in a foreign language, a speaker or writer should, then, have a rich repertoire of hedging expressions at his/her disposal and use an appropriate amount of them.

Another area in which the differences in the cultural conventions concerning hedging cause problems is translation. If the requirement is accepted that a good translation must not only preserve the meaning of the original but also produce in a reader an equivalent response to that produced by the original text, adjustments in hedging may be necessary. If the cultural norms have caused the writer of the original text to use a lot of hedging and the translator has not reduced their amount to suit the norms of the target culture, the result may be an irritatingly tentative, uncertain text. The opposite can of course also happen: the translated text may sound much too assertive to people used to more 'hedgy' texts. This is a real problem in the case of scientific texts, which must often be translated into a foreign language, most frequently into English these days. To be accepted into an international journal using English, the text should conform to the 'Anglo-American' cultural norms. The core of the translation problem is, however, that the translator should know the intentions of the original writer, i.e. how assertive or tentative s/he intended to be. This knowledge may be impossible to acquire, and the translator does not dare to make adjustments in hedging.

In the light of the problems hedges cause for learning to write effectively, for foreign language learning and for translation and revision of texts, it is obvious that much more research is needed. What seems to be required are more studies on cross-cultural differences, on the effects of hedges on readers, more comparisons of hedging phenomena in texts of different academic fields, of different genres, and their treatment and role in translation.

From the methodological point of view, we see the following as possible points of departure for future systematic research on hedges - we are thinking here in the first place of hedging in written communication, particularly in the academic/scientific field. Firstly, the starting point could be text as product, i.e. the focus could be on hedges as text-internal phenomena. Secondly, the starting point could be text as process, where the focus would be on the text-external factors that have an effect on text-production, including taking into consideration the potential readership. A third possibility would be to treat text as interaction, which would lead to studies on the reception of texts and their influence on readers. Naturally, it would also be interesting to combine all the three perspectives in one and the same analysis, which would give a more complete view of hedging.

From the more practical point of view, i.e. for the purposes of intercultural communication, language teaching and translation, we see as relevant at least the following types of studies, in addition to the self-evident contrastive/cross-linguistic studies:

(i) comparisons of codified stylistic conventions and criteria for 'good style', particularly in the area of academic/scientific communication;

(ii) examination of non-codified criteria for 'good style' and quality criteria for scientific texts;

(iii) analysis of texts written in a foreign language and of translated texts;

(iv) description and comparison of how texts are received by readers who have the same cultural background as the author and those who do not.

The relatively new area of research on writing in a foreign language uses the methods of text analysis but also tries to analyse the writing process itself. In the latter approach, the influence of the conventions used in writing in the mother tongue are of great interest. It would be a relatively simple task to compare the codified stylistic conventions for example in guidebooks for writing and advice given by the editors of scientific journals in different languages. This approach is actually very close to contrastive rhetoric, the basic assumption of which is described by Martin (1992) as follows: "the rhetorical organization of L2 written texts is the result of the transfer of L1 rhetorical forms for any language". This strong assumption is later moderated (Martin 1992, 3): "L1 rhetorical norms and strategies, while undoubtedly influential for the shape of L2 written texts, are but one factor influencing the text shape". Purves (1988, 13) seems to agree with this latter view when he suggests that, although the rhetorical devices used in textualisation are culture dependent, they are not necessarily dependent on the language of the writer. The conventions of the discipline in question may also have their role to play in scientific writing.

What we hope to have accomplished with the above introductory discussion is to show the importance of the phenomenon that we have called 'hedge/ hedging' for different areas of language use. We readily admit the difficulty connected with the concept. As was shown above, its use originates in logic and semantics, but has lately been developed further in pragmatics and discourse analysis so far that it now extends to areas like metacommunication and to communication strategies like mitigation and politeness. Through this extension the concept has lost some of its clarity and sometimes seems to have reached a state of definitional chaos, as it overlaps with several other concepts. This problem concerns many other linguistic concepts and their definitions, beginning with the concept of 'language' itself. However, the concept 'hedge' has the advantage of being usable for phenomena for which no other suitable concept exists. All the contributions in this volume adhere to this extended sense of the concept, some extending it more than others, but none of them restrict its use to the narrow meaning given to it by its 'founding fathers'. The articles also demonstrate the various perspectives - some more theoretical, some more practical - from which hedges and hedging can be considered, thus suggesting directions for future research.

References